Liver Stereotactic Body Radiation Therapy

Liver stereotactic body radiation therapy (SBRT) is highly conformal, image-guided external-beam radiation therapy that deposits a high, spatially focused dose in a small number of ablative fractions, typically five or fewer, while limiting irradiation of uninvolved liver and adjacent organs. ^[3] The treatment is not defined by fraction number alone: its defining features are stereotactic precision, steep dose fall-off outside the target, and reproducible control of respiratory motion and intrafraction position. ^[1][4] Four-dimensional imaging, breath-hold, abdominal compression, respiratory gating, or continuous image guidance may be used because liver position changes during breathing and the target may be difficult to distinguish from surrounding soft tissue. ^[7]

The clinical label describes both the tumor being treated and the treatment context. Primary liver SBRT targets a primary hepatic malignancy, most often hepatocellular carcinoma or intrahepatic cholangiocarcinoma . Metastasis-directed SBRT treats one or more hepatic metastases as part of an oligometastatic or oligoprogressive strategy; it seeks durable control of selected lesions while systemic therapy addresses disease elsewhere. Liver metastases are frequently unresectable because of lesion number, location, comorbidity, or uncontrolled extrahepatic disease, making noninvasive local treatment an alternative to resection in selected patients. ^[4][5] Postoperative or salvage SBRT refers to treatment of residual, recurrent, or previously treated disease when further surgery or percutaneous ablation is unsuitable. Prior liver-directed treatment can reduce the remaining functional reserve, so the purpose is not simply to retreat the radiographic lesion but to achieve local control without exhausting hepatic function. ^[4][9] Reirradiation is a new course after previous radiation; it may overlap the former high-dose region or involve a nonoverlapping region, and its feasibility depends on the prior dose to uninvolved liver, the location of the new target, and proximity to bile ducts and luminal organs. ^[6]

The central therapeutic trade-off is between tumor ablation and preservation of the liver that remains. A higher dose and tighter conformality can increase the probability of sterilizing the target, but cirrhosis, prior liver-directed treatment, multifocal disease, and larger treatment volumes reduce the liver’s functional reserve and increase the consequences of incidental irradiation. ^[1][9][18] SBRT therefore does not pursue the largest possible tumor dose in isolation. It seeks the highest biologically effective tumor dose that can be delivered while retaining enough functioning parenchyma to prevent radiation-induced liver disease and subsequent hepatic decompensation. Functional avoidance planning illustrates this principle by reducing dose to functioning rather than merely anatomical liver tissue when standard plans threaten hepatic reserve. ^[9]

The objective may be curative, consolidative, conversion-oriented, or palliative. Local ablation means sterilizing the treated focus; durable local control means preventing regrowth there long enough to affect the patient’s disease course. In selected primary tumors, SBRT can substitute for surgery, transplantation, or thermal ablation when those approaches are infeasible, although comparative evidence remains dominated by retrospective and nonrandomized studies. ^[3][14] In metastatic disease, the aim may be to delay intrahepatic progression, postpone a change in systemic therapy, or contribute to prolonged disease control; survival benefit cannot be inferred from local control alone because prognosis also depends on extrahepatic disease and tumor biology. ^[4]

SBRT can also serve as a bridge to transplantation or other definitive treatment when local response makes previously unresectable disease amenable to curative intervention. This conversion strategy remains selective and multimodality rather than a routine consequence of irradiation. ^[13] Palliation is a different objective: treatment is chosen to relieve symptoms or prevent an imminent local complication, and the acceptable balance between target dose, treatment burden, and hepatic risk may differ from that required for durable ablation. Across all objectives, preservation of liver function is not a secondary endpoint; it is the condition that allows a local treatment to benefit the patient rather than replace tumor progression with liver failure. ^[1][18]

The radiobiologic premise of liver SBRT is not simply that a large physical dose is delivered in few fractions. It is that a high dose can be confined to the entire tumor while the steep dose gradient preserves enough uninvolved, functioning liver to avoid radiation-induced liver disease and hepatic decompensation. Modern SBRT for hepatocellular carcinoma (HCC) commonly aims for a tumor BED10 of at least 100 Gy, where BED10 assumes an α/β ratio of 10 Gy for tumor tissue, although the clinically optimal dose depends on liver function, tumor volume, competing mortality, and the relative value assigned to tumor control versus toxicity.^[1][20]

Linear-quadratic modeling: useful shorthand, imperfect biology

The linear-quadratic model describes cell killing as the sum of a linear component, αD, and a quadratic component, βD², where D is dose. For n fractions of size d, the model gives:

[ BED = nd\left(1 + \frac{d}{\alpha/\beta}\right) ]

and the equivalent dose in 2-Gy fractions is:

[ EQD2 = \frac{BED}{1 + \frac{2}{\alpha/\beta}}. ]

BED expresses the biologic effect relative to dose without specifying a conventional fractionation schedule; EQD2 expresses that effect as the dose that would produce it using 2-Gy fractions. The α/β ratio is a fitted measure of how strongly fraction size influences tissue response: a higher ratio implies less sensitivity to fraction size, whereas a lower ratio implies greater sensitivity. BED10 is therefore a convenient common language for comparing ablative plans, but it is not a direct measurement of tumor sterilization or liver injury.^[1][32]

At the large fraction sizes used in SBRT, the quadratic term can dominate the calculated effect, so small changes in assumed α/β ratio produce large changes in BED. The model also extrapolates beyond the fractionation range from which many parameters were derived and does not fully represent sublethal-damage repair during prolonged delivery, intratumoral hypoxia, repopulation, spatially heterogeneous dose, or the effects of vascular and immune injury. Clinical dose selection therefore should not rely on BED alone: a recent utility model estimated optimal tumor BED10 values ranging from 112 Gy for a patient with Child-Pugh A cirrhosis to 26 Gy for a patient with Child-Pugh C cirrhosis, illustrating how baseline liver reserve can outweigh a nominal tumor-dose target.^[20] EQD2 and BED are especially useful for cumulative-dose assessment during reirradiation, but deformable dose accumulation and uncertainty in tissue geometry remain necessary because simple physical dose summation can overestimate dose to moving organs.^[32]

What an ablative dose does

The first and most established mechanism is direct tumor-cell killing. Ionizing radiation produces clustered DNA damage, including irreparable double-strand breaks; when the dose is sufficiently high throughout the viable tumor, clonogenic cells lose the capacity to reproduce. In liver SBRT, the relevant endpoint is therefore not merely the prescription dose at the tumor center but the dose delivered to every region that may contain viable tumor, including microscopic extension, vascular invasion, and the biologically active margin. In colorectal liver metastases, the minimum dose received by the planning target volume predicted freedom from local progression more reliably than the nominal prescription dose, and low minimum PTV BED10 was associated with inferior control.^[24]

The second mechanism is injury to the tumor microenvironment. Radiation can damage endothelial cells, alter perfusion, increase vascular permeability, and remodel stromal and immune compartments; these effects can amplify clonogenic death but also vary with oxygenation and baseline vascular architecture. Hypoxia reduces radiation effectiveness by limiting reactive oxygen species, while strategies that normalize abnormal tumor vessels may improve oxygen delivery and radiosensitivity.^[1] These mechanisms explain why a high central dose does not compensate for an underdosed hypoxic or perivascular subregion.

Radiation also modifies antitumor immunity. Immunogenic cell death can release tumor antigens and danger signals, promote antigen presentation, and increase cytotoxic T-cell recruitment; the proposed systemic consequence is an abscopal response, meaning regression of a nonirradiated lesion after irradiation of another lesion.^[29] The evidence remains mechanistically persuasive but clinically heterogeneous: a narrative review found that differences in dose, fractionation, sequencing, and patient selection complicate interpretation, while a mouse metastatic-HCC study linked SBRT plus anti-PD-1 therapy to increased CD8-positive effector-cell infiltration and better control of irradiated and nonirradiated tumors.^[28][29] Liver metastases are particularly variable because their immune phenotype may be immunosuppressive and “cold,” and both intertumoral and intratumoral heterogeneity can alter response.^[19]

These mechanisms converge on a practical rule: cover the entire biologically relevant target with an ablative dose, not merely the gross-visible center. A cold spot at the tumor edge, near a vessel, or within a mobile portion of the target can preserve clonogenic cells despite a high prescription dose. Target coverage must therefore be judged with dose-volume metrics that reveal the minimum and near-minimum dose, together with motion and setup robustness, rather than by prescription dose alone.^[24]

Why the same physical dose does not have the same clinical effect

Dose per fraction is the principal lever governing both tumor effect and normal-tissue risk. Increasing fraction size increases the BED predicted by the linear-quadratic model, but it also magnifies injury to late-responding normal tissues and makes model error more consequential. The clinical balance is therefore not “higher is better”; it is the highest tumor dose that leaves adequate functional liver and acceptable exposure of the central biliary tree and adjacent luminal organs. Patients with cirrhosis have diminished regenerative capacity, and the biologically relevant normal-tissue volume may be much smaller than the anatomic liver volume.^[1][9]

Treatment volume matters because SBRT creates a steep dose gradient, but the gradient cannot protect liver that lies within or immediately around a large target. As target volume increases, more uninvolved parenchyma receives moderate and low doses, the remaining functional reserve decreases, and the opportunity to spare a substantial liver subvolume narrows. A retrospective HCC series found that tumors up to 10 cm could retain high in-field control, but treatment failure commonly occurred elsewhere in the liver; this distinction separates local tumor ablation from preservation of whole-organ function.^[23]

Partial-volume effects explain why mean liver dose alone is an incomplete description of toxicity. A small volume receiving a high dose may be tolerated if sufficient liver remains untreated, whereas a broad low-to-moderate dose bath can injure a larger fraction of the organ. The relevant volume is also functional rather than purely anatomic: in a retrospective DCE-MRI study, dose to high-function liver volumes predicted early liver-function deterioration better than dose to low-function or total anatomic liver volumes, although the study was small and retrospective.^[21] Functional-avoidance planning is therefore a rational refinement of conventional DVH analysis, not a substitute for clinical judgment; in a planning study, reoptimization reduced the median mean dose to functional liver by 13%, but the authors did not establish a universal functional-volume constraint.^[9]

Clinical toxicity is also confounded by the natural history of cirrhosis. In a longitudinal analysis of 260 patients with cirrhosis and HCC treated with SBRT, an estimated 14.2% of ALBI worsening at 6 months and 24.9% at 12 months was attributable to cirrhosis progression rather than radiation toxicity.^[22] Post-treatment laboratory deterioration should therefore be interpreted alongside baseline liver function, tumor response, infection, biliary obstruction, systemic therapy, and the spatial distribution of delivered dose.

Histology and expected radiobiologic behavior

HCC, intrahepatic cholangiocarcinoma (iCCA), and liver metastases should not be assigned the same biologic expectations. HCC is often treated with a high BED because durable ablation is feasible in selected tumors, but its response is constrained by cirrhosis, hypoxia, vascular invasion, and the amount of functioning liver that must be sacrificed to cover the target.^[1] iCCA generally arises in a more infiltrative, fibrotic, and anatomically challenging environment; evidence for dose escalation is derived largely from nonrandomized or registry experience rather than direct randomized comparisons with HCC. In a prospective multicenter proton registry, higher prescribed dose and greater separation from the gastrointestinal tract were associated with better outcomes in selected unresectable cholangiocarcinoma patients, but this cannot be interpreted as proof that iCCA has a universal dose requirement.^[26]

Metastases are biologically heterogeneous because radiosensitivity reflects the primary tumor, prior systemic therapy, clonogenic burden, hypoxia, and immune phenotype. Colorectal liver metastases provide a clear example: even when all lesions received at least 70 Gy BED10, local control was associated with the minimum dose throughout the PTV rather than the nominal prescription dose, and polymetastatic disease independently predicted poorer freedom from local progression.^[24] A radiosensitive metastasis may be sterilized at a lower BED than a hypoxic or relatively radioresistant lesion, but the available evidence does not support a single histology-independent BED threshold. Dose selection should therefore integrate histology, target volume, prior treatment, expected survival, liver reserve, and the feasibility of delivering ablative dose without compromising functional parenchyma.

Pearl: In liver SBRT, the biologically effective dose is a planning language, not the treatment objective itself; the objective is complete, robust ablation of the biologically relevant target while preserving enough functioning liver to sustain the patient after treatment.

Liver stereotactic body radiation therapy (SBRT) is best positioned as a local, potentially definitive treatment when resection, transplantation, or thermal ablation cannot be offered, and as a consolidative or salvage treatment when a limited number of liver-dominant sites remain after systemic or transarterial therapy. The indication is not created by technical treatability alone: the expected benefit must justify irradiating functioning liver, and the multidisciplinary decision should account for tumor distribution, liver reserve, competing extrahepatic disease, and the availability of more established curative treatments. ^[34]

Hepatocellular carcinoma

For hepatocellular carcinoma (HCC), SBRT is a reasonable definitive option for a solitary or limited number of tumors that are unresectable or medically inoperable, particularly when the lesion is poorly suited to percutaneous ablation because of size, location, adjacency to major vessels, or risk to an adjacent organ. The rationale is strongest when the disease remains liver-confined and the patient has sufficient hepatic reserve for local therapy. Comparative evidence is most mature for early-stage or recurrent HCC: randomized trials and pooled analyses support SBRT as at least comparable to radiofrequency ablation for selected small tumors, although surgery, transplantation, or thermal ablation remain preferred when they are feasible and appropriate. ^[34]

SBRT is also useful for residual or recurrent HCC after transarterial chemoembolization (TACE), particularly when enhancement persists in a discrete target, repeated embolotherapy is unlikely to eradicate the lesion, or arterial access and tumor vascularity limit TACE. In intermediate-stage disease, this is a selective rather than a universal substitute for transarterial treatment: a large solitary tumor or a small number of lesions may be appropriate, whereas diffuse multifocal disease generally favors an intra-arterial or systemic strategy because treating numerous targets increases the volume of irradiated liver. ^[34]

Macrovascular invasion, especially portal vein tumor thrombosis (PVTT), is a further indication in carefully selected patients. SBRT may target the primary tumor and tumor thrombus when vascular invasion threatens portal flow, causes progressive portal-hypertensive deterioration, or remains the dominant site of disease despite systemic therapy. The evidence supports selection rather than routine use: a randomized trial established benefit for SBRT followed by sorafenib over sorafenib alone in an advanced-HCC population treated before current immunotherapy combinations, while contemporary studies combining SBRT with immune-checkpoint or targeted therapy remain predominantly single-arm or retrospective. ^[34] Treating PVTT as local disease is most defensible when extrahepatic disease is absent or limited and when response could preserve liver function, enable resection, or permit transplantation; extensive disseminated disease usually shifts the purpose toward symptom control or prevention of imminent hepatic deterioration. ^[34]

SBRT can serve as a bridge to liver transplantation or as a downstaging treatment when tumor burden is potentially convertible to transplant or resection. The endpoint is not simply radiographic shrinkage: sustained control while the patient remains eligible for transplantation, biological response of vascular invasion, and pathological response at explant or resection determine whether the strategy has achieved its purpose. A 2025 systematic review found favorable radiological and pathological responses in transplant candidates, but the included evidence was largely nonrandomized and therefore vulnerable to selection bias. ^[36] SBRT may also be used as definitive treatment when conversion does not occur, or as salvage for recurrence after prior locoregional therapy, provided cumulative liver and gastrointestinal-organ exposure remains acceptable. ^[42]

Palliative SBRT has a narrower role than definitive SBRT in HCC. Use it for a symptomatic or threatening focal lesion, such as painful capsular disease, bleeding risk, or tumor-related vascular compromise, when local treatment can relieve a specific problem and systemic therapy alone is insufficient. Do not present local control as a surrogate for survival in advanced HCC: HCC staging systems do not accept local control as an established survival endpoint, and the relationship between treated-lesion control and overall survival remains imperfect. ^[34]

Intrahepatic cholangiocarcinoma and other primary hepatic malignancies

For intrahepatic cholangiocarcinoma (iCCA), SBRT is considered mainly for unresectable, liver-limited or liver-dominant disease when resection is not possible and a discrete target can be treated without sacrificing excessive functioning liver. It is particularly attractive for a centrally located or perihilar-adjacent tumor that is unsuitable for thermal ablation, or for persistent or recurrent disease after systemic therapy when further local control is clinically meaningful. The evidence base is more selective than for HCC: current literature describes SBRT within multimodality care, but prospective evidence establishing when it should replace surgery, chemotherapy, transarterial radioembolization , or other locoregional treatment remains limited. ^[40]

For iCCA, use SBRT as definitive local therapy in highly selected patients, as consolidation after a favorable response or stable disease on systemic therapy, and as salvage for an isolated intrahepatic recurrence. Downstaging to resection is possible in principle but should be regarded as an investigational or case-selected strategy rather than a routine expectation. Prospective registry data for dose-escalated proton therapy, not SBRT specifically, illustrate the selective nature of the evidence: outcomes were reported in unresectable cholangiocarcinoma without distant metastasis, and liver function and tumor proximity to the gastrointestinal tract strongly influenced outcome. ^[26]

Other primary hepatic malignancies, including selected biliary tract tumors, may be treated with SBRT when disease is focal, unresectable, and not adequately addressed by surgery, systemic therapy, or transarterial treatment. Evidence is generally extrapolated from iCCA or mixed liver-tumor series rather than derived from histology-specific randomized trials; treatment should therefore be individualized in a hepatobiliary multidisciplinary conference. ^[2]

Liver metastases

For liver metastases, SBRT is a metastasis-directed treatment rather than a histology-independent standard. The most defensible candidates have oligometastatic disease, meaning a limited number of active sites that can all be treated or otherwise controlled, or oligoprogressive disease, meaning a small number of liver lesions progressing while the remainder of the cancer remains controlled on systemic therapy. Selection favors a small total target volume, limited lesion number, absence of rapidly progressive extrahepatic disease, and a systemic-therapy plan that can control untreated microscopic disease. Surgery remains preferred when complete hepatic resection is feasible and compatible with oncologic strategy; thermal ablation remains attractive for accessible small lesions. SBRT becomes particularly useful when lesions are deep, subphrenic, adjacent to major vessels, difficult to reach percutaneously, or too large or numerous for safe ablation. ^[5]

Colorectal cancer liver metastases have the clearest disease-specific rationale among common metastatic histologies after surgical resection. SBRT can be definitive for unresectable oligometastases, consolidative after a response to chemotherapy, or salvage for a limited hepatic recurrence. Lesion number and diameter matter because increasing both worsens local control and often signals more disseminated biology; in one contemporary series, three or more metastases and diameter of at least 2.7 cm were adverse local-control factors. These findings support treating a small number of smaller lesions when feasible, while recognizing that they are observational selection factors rather than absolute exclusion thresholds. ^[41] In oligoprogression, continue or adjust systemic therapy according to the extrahepatic disease pattern; SBRT is intended to eradicate resistant liver clones or delay a change in otherwise effective systemic treatment, not to replace systemic control. ^[10]

For neuroendocrine tumor and breast cancer liver metastases, SBRT is reasonable when the hepatic disease is limited, technically unresectable or medically inoperable, and systemic therapy has produced durable control or the progressing lesions are few enough for metastasis-directed treatment. The evidence is less mature and less uniform than for colorectal metastases, so histology, expected survival, receptor-directed or endocrine treatment options, extrahepatic burden, and the consequences of interrupting systemic therapy should determine intent. Mixed-primary liver SBRT series include breast and other metastases, but the evidence remains largely retrospective and does not establish a histology-specific survival benefit. ^[4][5]

Intent should be stated explicitly. Definitive SBRT seeks durable eradication of all clinically active disease in the treated liver targets when surgery or ablation is unsuitable. Consolidative SBRT follows systemic or transarterial therapy when residual liver disease remains limited and controlled elsewhere. Salvage SBRT treats an isolated recurrence or resistant lesion after prior local treatment. Palliative SBRT treats a focal symptom or imminent complication when cure or durable disease control is unrealistic. Across metastatic cancers, local control alone does not prove a survival benefit; prospective randomized data are still needed to define the contribution of dose escalation and to distinguish patients who benefit from metastasis-directed therapy from those whose apparent oligometastatic state reflects short-lived systemic control. ^[10]

Refer for liver stereotactic body radiation therapy only after documenting the treatment objective and confirming that the expected local benefit justifies the risk of hepatic decompensation. Record ECOG performance status , frailty, nutritional status, life expectancy, symptom burden, and patient priorities. Define whether the intent is definitive local therapy, bridge or downstaging to liver transplantation , salvage after recurrence, consolidation of liver-dominant disease, or palliation; the choice of local treatment depends on treatment intent and underlying clinical factors ^[48]. Assess cardiovascular, pulmonary, renal, metabolic, and neurologic comorbidity, as well as alcohol use, obesity, diabetes, and sarcopenia. In HCC, treatment selection must integrate tumor burden, performance status, liver function, extrahepatic disease, and comorbidities ^[60].

Document every prior liver-directed intervention: resection, radiofrequency ablation , microwave ablation, transarterial chemoembolization , transarterial radioembolization, hepatic arterial infusion, prior radiation, and prior systemic therapy. Obtain the prior radiation plan and dose distribution whenever abdominal or liver irradiation has been given; feasibility depends on the dose previously received by uninvolved liver, overlap with the proposed treatment region, and proximity to bile ducts and luminal gastrointestinal organs. Record the date, indication, complications, and residual radiographic changes after each procedure. Prior treatment may reduce functional reserve even when the remaining liver appears anatomically adequate. In patients with recurrent HCC after transplantation, management should include the transplant surgeon, physician, oncologist, and radiologist ^[53]. Confirm graft function, rejection history, immunosuppressive regimen, and transplant-center follow-up before referral.

Pretreatment staging and imaging

Obtain a high-quality multiphasic contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI) of the liver. The examination should establish the number, size, and distribution of lesions; arterial enhancement and washout; relationship to portal and hepatic veins; biliary involvement; extrahepatic extension; and the amount and morphology of uninvolved liver. MRI is particularly useful when CT findings are equivocal, when lesions are small, or when prior treatment has altered enhancement. For colorectal liver metastases, CT and MRI simulation may both be obtained when feasible ^[51], and the diagnostic liver MRI should not be omitted simply because a CT has already been performed.

Stage the chest with CT, or with another appropriate chest study when CT is not clinically suitable, because extrahepatic disease changes the purpose and expected benefit of liver-directed treatment. Add positron emission tomography /CT when the primary tumor is uncertain, when conventional imaging leaves suspected extrahepatic disease unresolved, or when whole-body staging would alter the decision between local and systemic treatment. PET/CT is not a substitute for multiphasic liver CT or MRI: small or well-differentiated HCC may be inconspicuous on fluorodeoxyglucose imaging, whereas MRI or multiphasic CT defines vascular and hepatobiliary anatomy more reliably. In post-transplant recurrence, comprehensive staging may use dual-tracer PET/CT or contrast CT combined with bone scanning ^[53].

For HCC, assign Barcelona Clinic Liver Cancer (BCLC) stage because this system combines liver function, performance status, and tumor extent to guide treatment strategy ^[47]. For metastases, record the primary histology, systemic disease status, number and size of hepatic lesions, extrahepatic sites, prior response to systemic therapy, and whether the disease is oligometastatic or oligoprogressive. SBRT is most defensible when the liver lesions are technically treatable and systemic disease is absent, controlled, or being actively treated; in colorectal metastases, it may complement surgery or other ablation, serve patients unsuitable for resection, or be used after failure of previous local therapy ^[51].

Laboratory and functional assessment

Obtain a current complete blood count and chemistry and hepatic panel before referral. At minimum, measure total bilirubin, albumin, prothrombin time or international normalized ratio (INR), aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase, platelet count, and creatinine. Test for hepatitis B and hepatitis C infection, document prior antiviral therapy, and determine whether active viral replication requires treatment or hepatology input. Interpret abnormal aminotransferases in context: they may reflect tumor, viral hepatitis, alcohol, steatohepatitis, biliary obstruction, infection, or recent liver-directed therapy rather than radiation risk alone.

Calculate the Child-Pugh score from bilirubin, albumin, INR, ascites, and encephalopathy. Class A generally indicates the greatest hepatic reserve, class B intermediate reserve, and class C severely impaired reserve; the class is a selection variable, not merely a prognostic label. SBRT series have largely involved Child-Pugh A patients, and patients with Child-Pugh B8 or higher carry greater toxicity risk and are generally unsuitable for SBRT ^[47]. Selected patients with Child-Pugh B7 cirrhosis may be treated by an experienced multidisciplinary team, but preservation of liver function must dominate dose and target decisions ^[57]. Do not treat a rising Child-Pugh score as an isolated laboratory abnormality: new ascites, encephalopathy, or worsening synthetic function may represent clinically meaningful decompensation.

Calculate the albumin-bilirubin (ALBI) grade from serum albumin and bilirubin to provide a more objective estimate of hepatic reserve, particularly when ascites or encephalopathy makes Child-Pugh scoring subjective. A lower ALBI grade indicates better reserve; a worsening grade should prompt review of active hepatitis, portal hypertension, infection, biliary obstruction, tumor progression, and recent treatment before referral. Calculate the Model for End-Stage Liver Disease (MELD) score when cirrhosis is advanced, transplant candidacy is being considered, renal function is impaired, or short-term mortality risk may compete with local tumor control. MELD is complementary to Child-Pugh and ALBI: it is most useful for systemic severity and transplant urgency, whereas Child-Pugh and ALBI help describe the liver’s ability to tolerate focal treatment. Prognosis in HCC reflects both tumor progression and hepatic decompensation, requiring treatment decisions that preserve liver function and control portal hypertension ^[56].

Assess portal hypertension directly and indirectly. Look for splenomegaly, thrombocytopenia, gastroesophageal varices, portal-vein collateralization, recanalized umbilical veins, hepatofugal flow, and prior variceal bleeding. Record the presence, volume, and control of ascites , including whether it is diuretic-responsive or requires repeated paracentesis. Evaluate hepatic encephalopathy, jaundice, muscle wasting, and recent hospitalization for decompensation. Clinically significant portal hypertension, uncontrolled ascites, active variceal bleeding, or recent encephalopathy should trigger hepatology assessment before radiation and may make SBRT unsafe even when the tumor is technically focal. Cirrhosis from viral hepatitis, alcohol, or metabolic dysfunction-associated steatotic liver disease reduces the margin for functional loss; steatohepatitis associated with obesity or diabetes may coexist with apparently preserved laboratory values but still warrants careful assessment of reserve and competing cardiovascular risk ^[47].

Tissue diagnosis and multidisciplinary review

Do not require biopsy when the clinical context and multiphasic imaging establish HCC by accepted diagnostic criteria and tissue would not alter management. Obtain a biopsy when imaging is not diagnostic, when the liver is noncirrhotic and the diagnosis is uncertain, when cholangiocarcinoma or metastasis is plausible, when tumor biology would change systemic therapy, or when a solitary lesion may represent a benign or second primary tumor. Use image-guided core biopsy when feasible and review the bleeding risk in patients with portal hypertension, thrombocytopenia, or coagulopathy. Pathology should include immunohistochemistry and molecular testing when needed to distinguish HCC, intrahepatic cholangiocarcinoma, combined tumors, and metastatic disease.

Present every candidate at a multidisciplinary conference including hepatology, liver surgery and transplant surgery when relevant, interventional radiology, medical oncology, diagnostic radiology, and radiation oncology. The conference should decide whether resection, transplantation, thermal ablation, transarterial therapy, systemic therapy, or SBRT offers the best balance between tumor control and preservation of liver function. Multidisciplinary review is required because the optimal locoregional treatment depends on resectability, liver function, performance status, previous treatment, tumor location and size, and vascular anatomy ^[57]. The referral is complete only when the team has documented the diagnosis, stage, liver reserve, prior treatment and radiation exposure, transplant implications, treatment intent, and a plan for managing unfavorable findings.

Simulation for liver stereotactic body radiation therapy begins by reproducing the treatment position rather than by acquiring images immediately. Use a flat tabletop with a customized vacuum cushion; add a thermoplastic abdominal or pelvic mask when it improves day-to-day stability without compromising comfort or liver excursion. A published liver SBRT workflow used a 3.2-mm thermoplastic mask with a split-leg vacuum cushion, and a separate HCC workflow used a Vac-Lok system in the treatment position. ^[62][63] Keep arm position, torso rotation, table indexing, and external reference marks identical at simulation and treatment. The purpose is geometric reproducibility: SBRT has a steep dose gradient, so a small change in patient position can displace the high-dose region away from the intended target or toward uninvolved liver and adjacent bowel.

Acquire a planning CT with intravenous contrast when renal function and contrast safety permit. A noncontrast or low-dose dataset remains useful for electron-density assignment, device and embolic-material identification, and registration, but a single contrast phase is inadequate for many liver tumors. Use a multiphasic liver protocol that includes arterial, portal-venous, and delayed acquisitions; select slice thickness and reconstruction parameters that preserve small-lesion detail. Current abdominal SBRT guidance supports triple-phase CT supplemented by all available diagnostic imaging, because liver tumors may have poor contrast against normal parenchyma on routine CT. ^[62] The arterial phase is particularly useful for hepatocellular carcinoma (HCC), whereas portal-venous and delayed phases often clarify hypovascular metastases, intrahepatic cholangiocarcinoma , capsular extension, and washout or persistent enhancement.

Fuse the simulation CT with the most informative diagnostic magnetic resonance imaging (MRI), preferably using a rigid registration refined around the liver and target rather than relying only on bony anatomy. Review T1-weighted dynamic arterial, portal-venous, and delayed images; T2-weighted images; diffusion-weighted images (DWI); and, when obtained, hepatobiliary-phase images after a hepatocyte-specific contrast agent. MRI supplies information that CT may miss when tumor and liver have similar attenuation, while hepatobiliary agents improve lesion conspicuity and multiparametric assessment after locoregional treatment. ^[76] Treat deformable registration as an aid to interpretation, not as an unquestioned contour-transfer tool: abdominal deformation, different breath-hold levels, bowel filling, and interval treatment response can create false agreement. A study of MRI-to-CT registration found improved liver and portal-vein overlap with deformable rather than rigid registration, but this does not remove the need for visual review of the target boundary and adjacent vessels. ^[70]

Add PET/CT when it can alter staging, identify unsuspected extrahepatic disease, distinguish recurrent tumor from treatment change, or define metabolically active disease that is poorly characterized on CT and MRI. Do not use PET/CT as a substitute for multiphasic liver imaging. For prior treatment, import the pre-treatment diagnostic study, the treatment-planning images, and the most recent response examination. Register the images to stable hepatic landmarks, portal or hepatic veins, surgical clips, and the treated segment; then inspect the registration through the entire target region. After transarterial chemoembolization , dense lipiodol may obscure or mimic enhancing tumor and may alter CT attenuation, so compare the current arterial and delayed enhancement pattern with the pre-embolization lesion and, when available, post-treatment MRI. Post-SIRT PET/CT can also identify tumor subregions predicted to be underdosed and may help define a selective salvage target. ^[75][77]

Gross tumor volume

Contour the gross tumor volume (GTV) as the macroscopic viable tumor visible on the composite dataset, not as the largest contour on any single phase. For HCC, include arterial phase enhancement and corresponding delayed washout when these findings define the lesion; use portal-venous images to assess infiltrative spread, portal-vein tumor thrombus (PVTT), and venous continuity. DWI supports detection of viable tumor but should not be used in isolation because susceptibility, T2 shine-through, and post-treatment inflammation can produce false-positive signal. Hepatobiliary-phase hypointensity is sensitive for many HCCs and metastases, but it represents altered hepatocyte function rather than viable tumor by itself; require anatomic and dynamic-imaging correlation.

A synchronized contrast-enhanced 4DCT study found that adding contrast substantially improved lesion visibility through the breathing cycle, although two HCCs remained poorly conspicuous because of minimal enhancement. ^[62] When the lesion is not confidently visible on the planning CT, use MRI-to-CT fusion and, where available, contrast-enhanced phase-resolved CT rather than expanding the GTV to compensate for diagnostic uncertainty. In HCC, a small retrospective technical study found that contouring all respiratory phases remained the reference approach and that single projection datasets could underrepresent the target, particularly for heterogeneous tumors. ^[63]

Infiltrative HCC requires a different question: identify the full contiguous region of tumor rather than forcing an irregular infiltrative process into a discrete nodule. Combine arterial enhancement, washout, portal-venous expansion of the involved segment, DWI restriction, venous invasion, and interval growth. Include contiguous PVTT when the treatment intent is tumor control or preservation of portal flow; contour the intravascular component separately when this improves plan evaluation and image guidance. Do not include bland thrombus solely because it lies adjacent to PVTT. If the boundary remains uncertain, review the case jointly with abdominal radiology and compare all phases and prior examinations before choosing a margin.

After SBRT or other locoregional treatment, persistent enhancement does not automatically indicate viable tumor. Radiation-treated HCC may evolve from arterial enhancement and washout to delayed enhancement or nonenhancement, with signal and enhancement patterns stabilizing only after several months. ^[79] Contour residual or recurrent tumor when there is a convincing combination of interval growth, nodular arterial enhancement with washout, restricted diffusion, or a corresponding viable focus on prior imaging. For a new lesion arising at the edge of an ablation cavity, define the GTV around the viable focus and use the cavity as an anatomic reference; do not automatically treat the entire cavity as tumor. If the intent is salvage of an incompletely ablated lesion and viable disease cannot be separated from the ablation zone, document the uncertainty and individualize the target to the cavity margin most likely to contain residual disease.

Clinical, internal, and planning target volumes

The clinical target volume (CTV) represents suspected microscopic disease beyond the visible GTV. It is not synonymous with the motion envelope or the setup margin. For a discrete HCC or liver metastasis treated with definitive ablation, bridging, consolidation, or oligoprogressive intent, many practices use no elective microscopic expansion or a small individualized CTV because the target is radiographically defined and uninvolved liver is dose-limiting. The CTV should nevertheless include contiguous macroscopic extension, such as PVTT, bile-duct extension, or a recurrent focus along a resection or ablation margin. For iCCA, infiltrative HCC, postoperative recurrence, or a target with ill-defined longitudinal or ductal spread, use a deliberately chosen CTV that covers the suspected microscopic pathway; its size should reflect imaging uncertainty, histology, prior treatment, and liver reserve rather than a fixed automatic expansion. A larger elective volume is rarely justified when it would irradiate substantial functioning liver without a clear oncologic target.

For a motion-encompassing plan, the internal target volume (ITV) is the union of the GTV or CTV across the relevant internal positions. It accounts for internal motion and deformation; it is not a synonym for the GTV and should not be generated from one static diagnostic image. A technical study comparing phase-based methods found that specialty image projections differed from the all-phase reference, with single projections underrepresenting heterogeneous HCC. ^[63] The planning target volume (PTV) is then formed by adding the institution's setup and residual-registration margin to the ITV. The PTV margin covers geometric uncertainty after immobilization and image guidance; it does not compensate for an incompletely defined GTV, missed infiltrative disease, or an inadequately characterized post-treatment cavity. A published workflow used a 5-mm margin around the ITV, whereas guideline-based descriptions allow an additional 0-5 mm after ITV definition; the appropriate value remains institution- and patient-specific. ^[62][63]

Use consensus review for difficult targets. MRI-based delineation of liver metastases produced an average interobserver variation of 1.3-1.6 mm after exclusion or inclusion of reviewed outliers, with greater variation for irregular lesions. ^[64] That uncertainty is clinically meaningful when the lesion abuts the capsule, diaphragm, stomach, duodenum, colon, or chest wall. Preserve the true tumor boundary on MRI and CT, verify the diaphragmatic interface in multiple planes, and do not trim the GTV simply because the lesion is close to an organ at risk. If the interface is obscured, obtain radiology review or repeat diagnostic imaging rather than converting uncertainty into an unrecorded margin reduction.

Respiratory motion management is a geometric decision, not merely an imaging preference. The liver moves in all three translational axes and may deform or rotate during respiration; in a 10-patient HCC study, mean motion was 1.2 cm superoinferiorly, 0.5 cm anteroposteriorly, and 0.4 cm laterally. ^[63] Assess motion before choosing a technique, because a small margin may be adequate with reproducible breath-hold or tracking but unsafe when breathing is irregular, the baseline drifts, or the tumor-diaphragm relationship changes during treatment. The DEGRO Working Group’s practical guideline treats motion management as part of the physical requirements for safe liver SBRT. ^[80]

Selecting the motion-management strategy

Deriving the ITV from motion information

For free-breathing motion-encompassing treatment, contour the GTV or CTV on each usable respiratory phase and define the ITV as the geometric union of those phase-specific volumes. This preserves the trajectory rather than replacing it with a scalar superior-inferior amplitude. Review the phase cine or image sequence for hysteresis, deformation, and baseline shifts; a target that occupies different positions during inspiration and expiration cannot be represented faithfully by a simple symmetric expansion. ^[63]

Use all-phase contouring when tumor conspicuity changes with contrast phase or respiratory position, particularly for HCC. In a 10-patient technical study, an ITV made from only two phases had a Dice similarity index of 0.87 against the all-phase reference, whereas single-projection approaches performed worse; the authors concluded that all-phase ITV remained the reference approach for HCC. ^[63] Maximum- or minimum-intensity projections may assist with homogeneous lesions, but they should not be accepted automatically as the ITV: heterogeneous, arterially enhancing HCC can be underrepresented when only one projection is used. ^[63]

For repeated breath-holds, do not contour one breath-hold scan and assume that every subsequent breath-hold will reproduce it. Acquire multiple breath-hold datasets after coaching, register the internal liver and tumor anatomy, and take the union of the observed target positions; include residual motion during each hold and the measured inter-hold displacement. In a clinical breath-hold workflow using nasal high-flow therapy, the breath-hold ITV was formed from 4-5 breath-hold CT scans to account for inter- and intrabreath-hold variation. ^[83] When breath-hold reproducibility is poor, convert to a free-breathing ITV strategy rather than retaining a falsely narrow breath-hold volume. ^[7]

Compression should be validated with repeat imaging under the exact compression position and force intended for treatment. A practical threshold is to investigate respiratory control when measured tumor or diaphragm motion exceeds 5 mm, but the decision should use the residual motion, target location, and adjacent-organ geometry rather than the threshold alone. ^[82] Compression is most useful when it converts a large, reproducible excursion into a smaller one; it does not correct irregular breathing or guarantee stable internal anatomy.

Uncertainty that motion amplitude alone misses

Irregular breathing can corrupt 4DCT phase sorting and make the apparent tumor trajectory depend on the respiratory cycle selected for reconstruction. Baseline drift is a separate error: the patient may breathe with a stable amplitude while the end-expiration or end-inspiration position gradually shifts. Hysteresis means that the tumor may occupy different positions at the same external respiratory amplitude depending on whether the patient is inhaling or exhaling. Treat these findings as evidence that the ITV or gating model is incomplete, and repeat motion assessment after coaching or change the technique. ^[63][87]

External surrogates, abdominal belts, pressure sensors, spirometry, surface imaging, or chest-wall markers, measure breathing, not the tumor itself. Their correlation with internal liver motion may change with baseline drift, abdominal deformation, altered breathing pattern, or tumor location. In the prospective KIM implementation, adding internal fiducial localization to the external surrogate avoided clinically relevant dose error that would have occurred with the external surrogate alone in 2 of 15 patients. ^[89] Use an external surrogate for coaching and temporal correlation, but validate its relationship to internal anatomy and do not treat it as a substitute for internal verification when geometric precision depends on that assumption.

Fiducials, coils, and surrogate structures

Use an implanted fiducial marker when the lesion is poorly visible on treatment imaging, when free-breathing tracking is required, or when the planned gating signal cannot be trusted to represent tumor position. Gold markers placed with image-fused ultrasound, CT, or MRI guidance enabled adequate treatment in 24 of 25 tumors in one series, but marker migration into the right atrium prevented SBRT in two cases. ^[90] Migration risk is particularly relevant near hepatic veins and subcapsular surfaces, so placement must be weighed against the benefit of tracking.

Radiopaque treatment deposits from prior transarterial chemoembolization may serve as a noninvasive internal surrogate when they are spatially stable and sufficiently visible, avoiding a new puncture; this remains an observational feasibility approach rather than a universally validated replacement for fiducials. ^[91] Implanted coils or other radiopaque markers should be used only when their relationship to the tumor has been demonstrated on multiphasic CT or MRI and remains stable through the respiratory cycle. If the marker is separated from the tumor by deformable liver, tracking the marker alone can create a systematic geographic error.

When the tumor itself is visible on cine MRI, fluoroscopy, or repeated CT, a visible internal structure such as the tumor edge, diaphragm dome, biliary stent, embolic deposit, or implanted marker can function as a surrogate. For a subdiaphragmatic lesion, measure both the lesion and the right hemidiaphragm when possible; diaphragm excursion is an informative but imperfect surrogate because local liver deformation and sliding at the diaphragm can decouple the dome from a deeper tumor. A corset study quantified diaphragm amplitude directly and showed that compression reduced median motion from 1.13 cm to 0.55 cm, illustrating why patient-specific measurement is preferable to assuming a uniform compression effect. ^[82]

Patients unable to cooperate with breath-hold

Start with coaching, visual feedback, and a short test sequence. Select free-breathing 4DCT with an ITV when the patient cannot sustain or reproduce breath-hold; this is a valid motion-encompassing strategy rather than a reason to omit SBRT if the resulting ITV and organ-at-risk geometry remain acceptable. ^[7] Abdominal compression is a reasonable alternative when tolerated, especially for large baseline diaphragm excursions, but confirm residual motion under compression. ^[15]

For selected patients with poor voluntary breath-hold but adequate cardiopulmonary reserve, mechanically assisted noninvasive ventilation or nasal high-flow therapy can prolong and stabilize breath-hold. These approaches require dedicated anesthesia or respiratory support, screening for gastric insufflation and other contraindications, and direct assessment of liver position: a prospective MANIV study reported a liver case in which gastric insufflation displaced the liver cranially by more than 1 cm, and occasional caudal drift exceeding 1 cm led the investigators to recommend pretreatment stability testing for liver lesions. ^[85] MANIV improved motion-related image quality and registration consistency in another study, but its resource requirements limit routine use. ^[84]

Gating or internal tracking is preferable to an enlarged ITV when the patient breathes irregularly but can maintain a measurable respiratory pattern or when the target lies close to a dose-sensitive luminal organ. Expect longer fractions with gating because beam delivery is interrupted outside the accepted window. Tracking is most compelling when internal fiducials or a stable visible structure are available; otherwise, the apparent precision may simply reflect a precise measurement of the wrong surrogate. ^[86][89]

Pearl: Measure internal tumor or diaphragm motion under the actual treatment condition, form the ITV from the full observed internal excursion when using motion-encompassing treatment, and abandon a narrow breath-hold or surrogate-based margin when reproducibility has not been demonstrated.

Dose selection in liver stereotactic body radiation therapy (SBRT) is an isotoxic decision: prescribe the highest biologically effective tumor dose that preserves functioning liver and respects the tolerance of the central bile ducts, stomach, duodenum, bowel, and chest wall. Tumor size, histology, location, prior irradiation, liver reserve, and the anticipated survival benefit all alter that balance. Larger lesions expose more uninvolved liver to low- and intermediate-dose radiation, whereas lesions abutting hollow viscera may require dose reduction or additional fractions even when the tumor itself could tolerate escalation. ^[96][104]

Calculate the biologically effective dose using BED₁₀ = nd [1 + d/10], where n is the number of fractions and d is the dose per fraction in Gy. BED₁₀ is useful for comparing regimens whose tumor effect is modeled with an α/β ratio of 10 Gy, but it is not a direct measure of tumor sterilization or hepatic tolerance; dose heterogeneity, motion, hypoxia, repair, and the different radiobiology of normal tissues limit its interpretation. ^[94] A tumor BED₁₀ of approximately 100 Gy is commonly pursued when safe, and retrospective HCC data associate BED₁₀ ≥100 Gy with better local control than lower BED₁₀, although dose selection is confounded by tumor size, stage, and liver function. ^[99][98]

Dose constraints in liver stereotactic body radiation therapy are risk-adapted planning objectives, not interchangeable numbers. A constraint derived for 50 Gy in 5 fractions cannot be transferred automatically to 30 Gy in 5 fractions, 45 Gy in 3 fractions, or a single fraction, because both fraction size and the volume receiving dose alter normal-tissue injury. Protocols also differ in contouring: some define the liver as the entire organ, some report uninvolved liver after subtraction of the target, and some use functional liver imaging. The AAPM TG-101 framework, RTOG/NRG liver protocols, and institutional protocols should therefore be treated as protocol families rather than a single universal table. The prescribing physician must use the constraints belonging to the selected fractionation, contouring atlas, and treatment platform. Benchmarking across 35 institutions showed that a sufficiently detailed set of clinical goals was required to harmonize plans, and that prescription method itself changed the resulting target and organ-at-risk doses ^[106].

How to evaluate the liver. Contour the total liver and report the uninvolved liver separately, preferably as total liver minus the gross tumor and the motion-inclusive target according to the protocol in use. The total-liver mean dose describes the average irradiation of the organ; the uninvolved-liver mean dose better reflects the dose to parenchyma available to sustain hepatic function. Neither metric is sufficient alone. A high mean dose may result from irradiating a small volume intensely or a large volume moderately, whereas the same mean dose can have different consequences in a normal liver and a cirrhotic liver. In a retrospective series of patients with hepatocellular carcinoma, the uninvolved-liver mean dose was associated with intrahepatic out-of-field recurrence, but the authors explicitly described the threshold as exploratory and requiring external validation ^[107]. In Child-Pugh B patients, higher mean liver dose, higher dose to one-third of normal liver, and larger low-dose volumes were associated with grade III/IV liver toxicity; the corresponding critical constraint was not identified in Child-Pugh A patients ^[108].

Report both mean dose and spared-liver volumes. The clinically useful question is not simply “What is the liver mean dose?” but “How much functional, uninvolved liver remains below a tolerable dose?” A commonly used functional-planning objective is to keep at least 30% of the predicted functional liver volume at a BED3 of 40 Gy or less; in one example of 50 Gy in 5 fractions, this corresponded to at least 700 cc receiving no more than 18 Gy and a functional-liver mean dose of 16 Gy or less ^[113]. That objective came from a prospective institutional approach using sulfur-colloid SPECT and should not be represented as a validated universal threshold. In advanced cirrhosis, anatomic liver volume may substantially overestimate functioning parenchyma; functional imaging can therefore change which voxels are prioritized for avoidance ^[113].

Use low-dose, intermediate-dose, and high-dose metrics together. V20, V25, V30, and V50 describe the percentage or absolute volume receiving at least the stated dose, whereas Dmax and D0.1 cc or D1 cc describe near-maximum exposure. Liver injury generally reflects distributed parenchymal loss, so mean dose and spared-volume metrics are more informative than a single hottest voxel. By contrast, the stomach, duodenum, small bowel, colon, esophagus, and central biliary tree behave more like serial or functionally serial structures: a small length or focal region receiving a high dose can cause major injury even when the organ mean dose is low. For these organs, prioritize Dmax and near-maximum doses, then add a small-volume metric such as D0.1 cc or D1 cc. A point dose should never be interpreted without the dose-calculation grid, contour quality, and the volume over which it is measured.

A planning constraint is a prospective optimization objective. A hard stopping rule is a limit beyond which the plan should not be delivered without a documented multidisciplinary decision, altered fractionation, adaptive replanning, or abandonment of the treatment plan. The distinction matters because small-volume dose uncertainties, daily organ displacement, and dose-calculation differences can make a nominal constraint appear exceeded without equivalent biological exposure. Conversely, a plan that meets the simulation constraints can exceed them on treatment: daily imaging studies found clinically relevant dose excesses mainly in organs close to the target, particularly the stomach, heart, and esophagus ^[111]. Daily correction improves target coverage, but does not eliminate dose variation in organs at risk ^[105].

Adjacent organs. Contour the stomach, duodenum, small bowel, and colon as separate structures when they may approach the target; include the esophagus for dome, hilar, and upper-abdominal targets; and contour the spinal cord, kidneys, heart, pericardium, chest wall, ribs, and skin when they enter the low-dose fall-off or beam path. Contour the central biliary tree when the lesion is hilar, perihilar, central, adjacent to the common hepatic duct, or associated with biliary obstruction. For bowel and stomach, evaluate both the planning scan and each treatment-day image when available, because gas, filling, and craniocaudal organ shifts can change the target-organ relationship. Online adaptive planning has reduced stomach and duodenal maximum dose when treatment-day overlap emerged, while improving target coverage compared with delivering the scheduled plan unchanged ^[122].

The central biliary tree deserves separate attention from the remainder of the liver. Its tolerance is not captured by the uninvolved-liver mean dose, and biliary obstruction, stents, infection, and prior biliary intervention can reduce the margin for additional irradiation. A deep-learning analysis of liver SBRT plans identified the hepatobiliary tract as a higher-risk region for hepatobiliary toxicity than individual liver segments, supporting explicit biliary contouring when anatomically relevant ^[121]. For a central tumor, do not solve a target-duct conflict by silently trimming gross disease; instead, document the competing objectives, consider more fractions, reduce the dose per fraction, use adaptive delivery, or select another local therapy.

The following values are commonly used starting points or study-specific examples. They are not universal stopping rules, and the fractionation column identifies where the cited value was reported rather than implying that the same number applies to every regimen.

Modifying the constraints. For Child-Pugh B cirrhosis, do not merely apply a Child-Pugh A plan with a lower prescription. Reduce the tolerated mean liver dose and low-dose exposure, preserve the best-functioning parenchyma, and consider functional imaging. Child-Pugh B patients in a phase 1-2 series had more grade III/IV liver toxicity than Child-Pugh A patients, and toxicity correlated with mean liver dose, one-third normal-liver dose, and low-dose volumes ^[108]. A functional-planning series in Child-Pugh B/C patients illustrates the rationale for basing constraints on functional rather than anatomic liver volume, but its results remain institutional and nonrandomized ^[113]. Midtreatment reassessment can provide an additional safety margin: change in ALBI, the albumin-bilirubin score derived from routine laboratory values, improved prediction of treatment-related toxicity beyond baseline assessment ^[115].

Prior liver radiation, transarterial radioembolization, repeated embolization, ablation, or surgery requires cumulative review of dose and functional reserve. Recontour the previously irradiated liver and adjacent organs, deformably register prior and current plans only after visual validation, and identify overlap between the new high-dose region and the prior high-dose region. Segmental radioembolization followed by SBRT was reported as tolerable in a retrospective comparison with TACE, but the authors called for longer follow-up and larger studies ^[119]. A single case of repeat SBRT showed marked treated-lobe contraction with compensatory contralateral hypertrophy, which supports feasibility in selected patients but cannot establish a general reirradiation constraint ^[118]. Reirradiation should therefore use more conservative uninvolved-liver and biliary constraints, avoid cumulative high dose to luminal organs, and favor adaptive or staged treatment when geometry permits.

Biliary obstruction, central tumors, and gastric or bowel invasion convert an ordinary liver plan into a competing-objective plan. A visible invasion interface must remain part of the target when tumor control requires it; the solution is not to remove disease from the contour. Instead, use fractionation that lowers dose per fraction, prioritize the serial-organ Dmax and near-maximum limits, consider staged treatment or nonradiation local therapy, and obtain hepatobiliary and gastrointestinal input when the planned high-dose region directly involves a duct or hollow viscus. Daily anatomy matters because organ-at-risk violations of 1-6 Gy in a single fraction were simulated with setup correction alone and reduced below 1 Gy after adaptive reoptimization in selected cases ^[112].

Finally, treat protocol limits as decision-support tools rather than permission to ignore clinical context. A plan that exceeds a planning objective may remain deliverable after expert review if the excess is small, the dose is uncertain, the organ is not functionally critical, and no safer plan preserves meaningful target coverage. A hard stopping rule is different: do not proceed when the relevant serial-organ limit is exceeded in a reproducible way, when cumulative dose cannot be reconstructed, or when insufficient functional liver remains to support the patient’s reserve. The decision should be documented in terms of total and uninvolved-liver dose, functional spared volume, serial-organ Dmax or near-maximum dose, daily anatomy, prior exposure, and the patient’s hepatic and gastrointestinal vulnerability.

Inverse planning for liver stereotactic body radiotherapy (SBRT) should begin with a clinical hierarchy rather than an unconstrained search for the most conformal dose distribution. First secure biologically meaningful coverage of the gross disease and clinical target; then protect functioning uninvolved liver, central biliary structures, stomach, duodenum, small bowel, colon, and chest wall; finally reduce intermediate- and low-dose spill. The liver behaves as a parallel organ, so integrated dose to uninvolved parenchyma is a major determinant of radiation-induced liver injury, whereas small-volume gastrointestinal structures can suffer focal injury from a high maximum dose ^[126]. The optimization objective is therefore isotoxic: preserve target coverage and dose intensity while minimizing the volume of functioning liver and bowel exposed to unnecessary dose.

Use inverse-planned intensity-modulated radiation therapy (IMRT) or volumetric-modulated arc therapy (VMAT) when beam arrangement and multileaf-collimator modulation can improve conformality without creating a broad low-dose bath. Optimize the high-dose region to encompass the PTV with a steep, spatially appropriate gradient, but inspect the dose outside the PTV rather than accepting conformity at the expense of excessive V5, V10, or V15 to uninvolved liver. A lower prescription isodose can create a steeper fall-off and a higher central tumor dose, but the trade-off must be tested against motion and organ-at-risk exposure; a recent phantom-based study found that lower isodose plans tolerated greater simulated motion and reduced peripheral normal-tissue dose, although this approach remains investigational rather than a universal prescription method ^[72].

Plan evaluation should include the dose-volume histogram (DVH), axial and three-dimensional isodose review, and quantitative conformity and gradient indices. The conformity index asks how closely the prescription isodose conforms to the target; the gradient index assesses how rapidly dose decreases outside that volume. Review target D95 and near-minimum dose, not only the nominal prescription or maximum dose. This distinction matters because a retrospective study of colorectal liver metastases found that PTV minimum dose, rather than nominal prescription dose alone, independently predicted freedom from local progression; cold regions at the tumor edge or near bowel therefore require deliberate review rather than automatic acceptance ^[24]. A plan with an attractive conformity index but a cold PTV rim is not an ablative plan.

When the target abuts stomach, duodenum, or bowel, do not solve an organ-at-risk conflict by trimming the true target. Instead, identify the undercovered region, determine whether the limitation is geometric, motion-related, or dose-tolerance-related, and choose among altered beam angles, a different prescription, more fractions, adaptive delivery, or a multidisciplinary decision to accept undercoverage. Report the undercovered target subvolume and its location. The clinical question is whether the cold region lies within viable tumor, microscopic-risk CTV, or only an intentionally lower-priority margin. The 2024 International Stereotactic Radiosurgery Society (ISRS) systematic review and practice guideline supports personalized planning based on lesion and histologic factors, but its pooled outcomes do not establish a single universal dose or planning metric for every liver metastasis ^[127].

For multiple lesions, optimize the composite plan rather than each lesion independently. Evaluate the summed dose to uninvolved liver and every shared organ at risk, including overlap of intermediate-dose regions between lesions. Closely spaced targets may require separate isocenters, dedicated arcs, altered collimator sequencing, or staged treatment. In the NRG-BR001 technical framework, lesions within 5 cm created specific composite-planning trade-offs, and benchmark plans showed substantial variability in conformity despite meeting protocol requirements ^[128]. A 2026 retrospective planning study similarly found that conventional single-isocenter VMAT had greater normal-liver dose and greater sensitivity to rotational setup error than a spatially grouped sequencing strategy, although this newer approach requires clinical validation before routine adoption ^[31].

Robotic delivery with CyberKnife may be useful when continuous respiratory tracking and many noncoplanar beam directions improve target conformity or permit bowel sparing, but the tracking chain must be validated for marker visibility, migration, latency, and correlation between external respiration and internal tumor position. The Chinese expert consensus describes robotic HCC treatment as combining three-dimensional multi-angle irradiation, fiducial localization, and synchronous respiratory tracking to reduce respiratory off-target dose ^[125]. This is a platform-specific capability, not evidence that robotic delivery is superior to a well-commissioned linac-based SBRT plan.

Proton therapy , particularly pencil-beam scanning, can reduce exit dose and integral exposure to normal liver distal to the target because of the Bragg peak. Its advantage is most relevant when photon plans cannot meet liver or bowel objectives, but range uncertainty, density changes, respiratory motion, and interplay between moving anatomy and sequential spot delivery must be included in planning and verification ^[126]. The NRG Oncology 2025 survey found that 17 of 18 responding proton centers used a robustness strategy; common perturbations were ±5 mm setup error and ±3.5% relative stopping-power uncertainty, while 14 of 18 optimized on ITV or CTV and 14 of 18 used a robust target-coverage criterion of D95 ≥95% of prescription dose ^[126]. These values describe contemporary practice patterns, not a universal society-mandated constraint.

Evaluate robustness against the uncertainties that can change the delivered dose: setup translation and rotation, residual respiratory motion, baseline drift, contouring and registration error, CT-density or proton stopping-power uncertainty, and, when relevant, interplay. For photon plans, perturb the patient geometry and recalculate or recalculate-and-accumulate dose across respiratory phases when motion could alter target coverage or bowel dose. For proton plans, evaluate range and setup perturbations together rather than separately. Review the worst clinically plausible scenario, not only the nominal plan. In the NRG survey, nine of 18 centers based decisions on the worst-case scenario, and 17 of 18 performed robust-plan evaluation ^[126].

Quality assurance must verify the entire chain from image registration and contour transfer to dose calculation and delivery. The radiation oncologist owns target and OAR definition and clinical trade-offs; the dosimetrist owns inverse-planning execution and DVH reporting; the medical physicist owns commissioning, dose calculation validation, machine QA, patient-specific QA, and end-to-end testing; and the radiation therapist verifies treatment setup and delivery records under the department’s SBRT policy. CARO’s 2012 SBRT scope guideline recommends a dedicated SBRT program, defined team responsibilities, and SBRT-specific quality assurance ^[129].

Use a validated treatment-planning algorithm appropriate to small, highly modulated fields and heterogeneous abdominal anatomy. Compare the clinical calculation with an independent dose calculation, preferably using a second algorithm such as Monte Carlo when the primary algorithm is analytical. The NRG proton survey reported independent secondary calculations at only 10 of 18 centers, illustrating why this check should be an explicit local requirement rather than an assumed feature of the TPS ^[126]. Validate small-field output factors, MLC or aperture modeling, dose-rate response, monitor-unit linearity, and detector corrections during commissioning and after relevant hardware or software changes. A stereotactic detector study found that upgraded array response variation remained below ±0.4% across the tested clinical dose-rate range and that field-size agreement with the TPS was below 0.95% for fields ≥1.5 × 1.5 cm²; such results are equipment-specific and cannot replace local measurement ^[133].

Patient-specific QA should combine an appropriate measurement with an independent calculation and, when available, delivery-log analysis. Use high-spatial-resolution detectors or small-volume ion chambers when the relevant penumbra or 50%-50% dose region is small. The NRG survey reported patient-specific QA at 16 of 18 proton centers; commonly reported passing criteria were gamma 3%/3 mm with a 90% passing rate, although some centers used 2%/2 mm or a 95% threshold ^[126]. Gamma passing is a screening tool, not proof of clinical adequacy: investigate failures by separating delivery error, registration error, detector limitation, dose-calculation error, and respiratory-motion effects.

End-to-end testing should reproduce the clinical workflow in an anthropomorphic or motion phantom, including simulation, image registration, contour transfer, planning, motion management, delivery, and dosimetry. Independent audit data show that advanced-therapy phantom tests commonly use acceptance limits of 5-7% and 3-4 mm, but the moving multi-target liver phantom had the lowest reported pass rate among the reviewed phantom types, at 73% ^[136]. For moving liver targets, compare motion-coupled measurements with a motion-resolved dose calculation when possible; a 2026 audit methodology found that conventional moving-film analysis was overly sensitive and that a motion-corrected approach improved specificity while permitting tighter 4%/2 mm criteria ^[134].

Adaptive replanning is indicated when the anatomy used for optimization no longer represents the anatomy at treatment. Trigger review when there is substantial change in liver volume, ascites, tumor position, body contour, bowel filling, stomach position, or the spatial relationship between target and an OAR. Repeat CT or MRI, re-register the target and OARs with visual quality control, recalculate the delivered and remaining composite dose, and re-optimize when constraints or target robustness are no longer satisfied. Proton range is particularly sensitive to anatomic change; the NRG survey article describes repeat CT, forward dose calculation, and dose-metric review as methods for detecting clinically meaningful changes before or during treatment ^[126]. Adaptive workflows have also reduced stomach and duodenal dose when treatment-day overlap emerges in abdominal SBRT ^[137]. Do not regard setup correction alone as an adaptive solution when bowel filling, ascites, or liver deformation changes the dose distribution.

Pearl: A liver SBRT plan is acceptable only when target coverage, normal-liver preservation, OAR safety, and delivery robustness remain satisfactory after the clinically plausible errors, not merely on the nominal DVH.

Image-guided radiation therapy is a treatment-day safety process, not a single matching step. The CARO SBRT scope guideline recommends a dedicated SBRT program with defined radiation oncologist, physicist, and dosimetrist responsibilities, together with site-specific quality assurance ^[129]. For liver SBRT, the workflow must verify three linked conditions before beam-on: the patient is in the planned position, the liver target occupies the planned respiratory geometry, and the stomach, bowel, central biliary tree, and other serial organs remain within the assumptions used for planning.

Treatment-day imaging and registration

Acquire a pretreatment cone-beam CT (CBCT) whenever the platform permits. Register the bony anatomy for reproducible patient setup, then register the liver, tumor, fiducials, clips, stents, or other high-contrast surrogates used in the plan. Volumetric registration is preferable to a single projection because it exposes changes in liver shape, stomach or bowel position, ascites, body contour, and target-to-organ relationships. In the NRG Oncology RTOG 1112 credentialing analysis, the median difference between clinical and expert registrations was 3 mm and 95% of fractions were within 5 mm; soft-tissue registrations exceeded 10 mm in 3% of fractions, whereas no high-contrast registration exceeded 5 mm ^[116].

Use four-dimensional cone-beam CT when respiratory phase, baseline drift, or target motion cannot be judged reliably on a static CBCT. A phase-resolved dataset can show whether the treatment-day motion envelope remains compatible with the planning internal target volume (ITV), but it is only useful when the respiratory signal is sufficiently regular. If phase sorting produces severe artifacts or incomplete respiratory states, do not accept the apparent ITV as evidence of geometric stability; repeat the acquisition after coaching or convert to a robust free-breathing strategy.

Use orthogonal kilovoltage imaging when implanted markers or clips provide a reliable internal surrogate. Fiducial tracking measures the target surrogate directly and can detect intrafraction translation and rotation; one fiducial-tracking study found the largest translational variability in the superior-inferior direction and recommended margins of 2 mm left-right, 3 mm anterior-posterior, and 7 mm superior-inferior for its treatment system ^[146]. These values are not transferable margins: calculate the local margin from the measured residual motion, imaging frequency, latency, rotational uncertainty, and dose gradient.

Use MRI-guided radiotherapy when the lesion, liver interface, or adjacent luminal organ is poorly visible on CT, or when online adaptation is likely to change the therapeutic ratio. MRI-guided platforms combine improved hepatic visualization with adaptive planning and advanced motion management ^[139]. Before treatment, verify the magnetic-resonance sequence, contrast phase, patient position, respiratory state, and coil-related setup against simulation. During online adaptation, recontour the target and every nearby serial organ that can influence the plan; do not regard the baseline plan as a surrogate for the dose delivered by an adapted plan, because adaptive treatment can increase near-maximum organ doses ^[143].

Verification tasks and required response

A planned match is not automatically a safe match. Interrupt the fraction when the target or implanted marker leaves the validated gating or tracking window, when a respiratory baseline shift changes target position beyond the residual-motion allowance, or when new stomach or bowel overlap invalidates the approved dose distribution. A 5-mm investigation threshold is reasonable for many workflows because respiratory-control guidance recommends investigating residual motion above 5 mm, but the operational threshold must come from the plan-specific uncertainty budget ^[146]. For systems using respiratory tracking, respiratory phase shifts can produce clinically relevant dose loss; in one dosimetric analysis, the 90th-percentile phase shifts reduced target D90 by as much as 6.6% ^[148].

Repeat the fraction only after physics and physician review of the delivered dose. Do not repeat solely because a post-treatment image looks different if the delivered dose remains within the approved uncertainty envelope; conversely, do not declare a fraction adequate merely because the nominal monitor units were delivered. If dose reconstruction cannot demonstrate acceptable target coverage after a significant geometric failure, either deliver a physician-approved supplemental fraction with cumulative-dose review or replan the remaining course. The rationale, estimated delivered dose, and added normal-tissue exposure must be documented.

Registration uncertainty and adaptive interpretation

A poorly visualized lesion demands a composite registration rather than confidence in one image. Fuse diagnostic MRI with multiphasic CT and CBCT, use vascular, capsular, diaphragmatic, and biliary landmarks, and have the radiation oncologist and radiologist review the result when the lesion boundary is uncertain. Use implanted gold markers only after confirming that they remain spatially related to the tumor. Marker migration, marker-to-tumor deformation, and a marker outside the imaging or tracking field can create a precise registration of the wrong structure.

Treat deformable image registration (DIR) as an aid to dose accumulation and image interpretation, not as ground truth. A liver DIR study found that structure-guided registration using anatomical landmarks reduced mean target-registration error to 1.6-1.7 mm in its test cohort, but the authors still evaluated accuracy against landmarks and dose-gradient effects ^[140]. In MR-guided liver SBRT, seven dose-warping algorithms produced highly patient- and fraction-dependent variation; the reported variation in accumulated PTV D95 reached 24.9 percentage points between algorithms ^[141]. Therefore, inspect the deformation field, Jacobian behavior, landmark correspondence, and dose gradient; compare at least two accumulation methods when a clinical decision depends on a narrow margin. If the deformed anatomy is implausible or landmarks disagree, use a conservative dose-volume estimate rather than a single automated accumulation.

Stomach and bowel position can change enough during a course to alter both target coverage and serial-organ dose. Liver-volume reductions during treatment were associated with larger increases in accumulated normal-tissue dose than stable liver volumes, and deviations greater than 5% of prescription occurred in both groups ^[142]. Repeat CBCT or MRI when the body contour, ascites, liver volume, stomach filling, bowel position, or tumor-to-organ distance changes. Setup correction alone does not correct an organ that has moved relative to the liver; perform online adaptation or replan the remaining fractions when the treatment-day anatomy fails the approved target and OAR objectives.

Liver reirradiation: cumulative-dose assessment

There is no validated universal liver-reirradiation constraint set. A 2025 systematic review found that abdominal reirradiation studies were predominantly retrospective, used heterogeneous prescriptions and intervals, and lacked consensus on cumulative dose metrics or limit values ^[42]. Reirradiation therefore requires a patient-specific multidisciplinary decision rather than a nominal “safe interval” or an isolated mean-liver-dose cutoff.

Reconstruct the prior course before approving a new plan. Obtain the original CT or MRI simulation, structure set, treatment plan, dose grid, delivered dose if available, fractionation, beam or proton parameters, image guidance records, adaptive plans, and treatment interruptions. Confirm whether the prior course was external-beam radiation, and separately record prior radioembolization , transarterial chemoembolization , ablation, surgery, or biliary intervention because these treatments may alter liver reserve without providing a directly summable external-beam dose. The ReCOG review found that prior dose, reirradiation dose, interval, and delivered-versus-planned dose were often incompletely reported, which is itself a reason to widen uncertainty assumptions ^[42].

Accumulate dose only after reviewing registration quality. Rigid registration may be adequate for stable bony anatomy but is insufficient when the liver has changed shape, volume, or position. Use landmark- or structure-guided DIR with visual quality control, then compare the accumulated dose with a conservative DVH-sum approach. In adaptive treatment, accumulate the delivered adapted plans rather than the original baseline plan, because the baseline plan can underestimate actual OAR exposure ^[143]. Report the registration method, target-registration error, dose-grid resolution, deformation quality checks, and the uncertainty range for each critical organ.

Express tumor dose and serial-organ dose in separate radiobiological languages. Calculate tumor BED using α/β = 10 Gy and organ BED using the relevant low-α/β assumption, commonly α/β = 3 Gy for late-responding serial organs; report the actual fractionation beside every BED or EQD2 value. Sum prior and current biologically equivalent doses only after converting each course with its own fraction size, and do not imply that BED addition accounts for spatial dose distribution, repopulation, repair, hypoxia, or uncertainty. For the liver, evaluate cumulative mean dose, low- and intermediate-dose volumes, spared uninvolved liver, and the best-functioning parenchyma rather than relying on a single cumulative BED. For bowel, stomach, duodenum, and central bile ducts, prioritize cumulative near-maximum and small-volume doses because a focal hot spot can dominate injury.

Use the interval since prior treatment as a modifier, not as proof of recovery. Published abdominal reirradiation practice has used assumptions ranging from no recovery to a fixed 15% tissue recovery per year, and the ReCOG review found that most studies did not report how recovery was assumed or calculated ^[96][42]. If recovery is modeled, state the tissue, α/β, time interval, recovery fraction, and whether recovery is applied to the entire prior dose or only to the dose relevant to the overlapping organ volume. Present a no-recovery calculation alongside any recovered-dose estimate when the decision depends on the assumption.

Make current liver reserve the dominant clinical safeguard. Reassess Child-Pugh score , ALBI score , portal hypertension, ascites, bilirubin, albumin, INR, platelets, renal function, sarcopenia, and recent decompensation immediately before reirradiation. Preserve the best-functioning liver and avoid cumulative overlap with central bile ducts and luminal organs whenever possible. A favorable interval cannot compensate for poor current reserve, extensive prior uninvolved-liver irradiation, or a new target abutting a previously irradiated stomach, bowel, duodenum, or bile duct.

Classify the retreatment as non-overlapping, partially overlapping, or overlapping with the prior high-dose target and OAR regions. For non-overlapping treatment, prior uninvolved-liver dose and residual functional volume still govern feasibility. For overlapping treatment, reconstruct cumulative dose to the overlapping liver and serial organs, calculate current and cumulative BED/EQD2, and define the uncertainty interval before selecting dose and fractionation. If cumulative constraints cannot be established because the prior plan or dose grid is unavailable, treat the missing information as additional dose uncertainty, not as zero dose, and consider functional imaging, more fractions, reduced dose, staged treatment, or an alternative local therapy.

Pearl: A liver SBRT fraction is acceptable only when the delivered geometry, not the prescription alone, remains within the validated target, respiratory, and serial-organ uncertainty envelope; in reirradiation, the same rule applies to the reconstructed cumulative dose.

Multidisciplinary sequencing determines whether liver SBRT functions as definitive local treatment, a bridge to curative therapy, or a means of maintaining systemic control. The decision belongs in a liver tumor board that includes hepatobiliary surgery, transplant hepatology and surgery, interventional radiology, medical oncology, radiation oncology, radiology, and pathology. The team should define the endpoint before treatment, resection, transplantation, durable local control, downstaging, salvage, or preservation of a functioning liver, and should reassess that endpoint after every interval scan. The COLLISION multidisciplinary consensus emphasizes that local treatment selection should reflect anatomy, treatment goal, treatment-related morbidity, and patient factors rather than a fixed hierarchy of procedures ^[154].

Resection remains the preferred curative local treatment when an R0 resection can be achieved while preserving adequate functional liver, vascular inflow and outflow, and biliary drainage. SBRT should not replace a technically feasible operation solely because it is less invasive. It becomes useful when resection would require excessive parenchymal sacrifice, when comorbidity makes surgery unsafe, when disease is initially unresectable but potentially convertible, or when recurrence lies in a previously operated field. For colorectal liver metastases, the 2020 COLLISION consensus classifies SBRT as an option for selected unresectable lesions that are neither resectable nor thermally ablatable, but assigns this recommendation low evidence and strong or moderate consensus depending on the clinical subgroup ^[154]. Comparative evidence remains nonrandomized for most primary and metastatic liver tumors; a propensity-weighted 2026 analysis in small BCLC stage 0 HCC found comparable adjusted outcomes between resection and SBRT, but this does not establish interchangeability across tumor size, liver reserve, or anatomy ^[161].

After SBRT, surgery may remain feasible, but the surgeon should expect altered tissue planes, radiation fibrosis, adhesions, and possible vascular or capsular reaction. Review the radiation plan before operating, including the high-dose region, dose to the hepatic hilum and major vessels, and the location of the planned transection. Allow sufficient time for acute inflammation and edema to settle, but do not delay resection when progression would remove the curative window. In a small transplant series, the median interval from SBRT to transplantation was 141 days, visible radiation reaction or fibrosis was present in 44% of patients, and one patient required temporary inferior vena cava clamping because of the treatment reaction ^[160]. These findings support coordinated timing rather than a universal waiting interval.

Thermal ablation and SBRT are complementary rather than competing treatments. RFA or MWA is attractive for small, accessible lesions in patients who can tolerate an invasive procedure, whereas SBRT is often favored for deep, subphrenic, perivascular, or poorly visualized lesions, or when anesthesia and needle access pose substantial risk. Different lesions in the same liver can receive different treatments, provided that the composite plan protects functional parenchyma and adjacent organs. The 2026 practice review reports no randomized prospective trial directly comparing SBRT with thermal ablation for liver metastases; ASCO suggested SBRT for selected colorectal liver oligometastases not suitable for resection, whereas ESMO lists both SBRT and thermal ablation as options, leaving the choice to anatomy and multidisciplinary judgment ^[5].

Do not schedule SBRT and thermal ablation as though their toxicities were independent. Both can injure bile ducts, vessels, and functioning liver; their combination may enlarge the effective treated volume and complicate interpretation of post-treatment enhancement. When both modalities are needed, treat the lesion that offers the greatest immediate oncologic or technical benefit first, document the ablation cavity and radiation dose distribution, and reassess liver function before the second intervention. A staged approach is preferable when the composite low- and intermediate-dose exposure would threaten the future liver remnant.

Transarterial therapy may precede, follow, or occasionally be interposed between SBRT and systemic treatment. For persistent or recurrent HCC after TACE, SBRT can sterilize residual viable tumor that remains discrete and targetable; after SBRT, TACE can treat new or marginal intrahepatic disease when arterial anatomy and liver reserve remain favorable. A propensity-matched 2026 analysis of recurrent HCC found that sequential TACE after SBRT was feasible without observed acute grade 3 or higher toxicity, but the survival advantage was not statistically significant, so this remains hypothesis-generating rather than standard sequencing ^[155]. TARE or Y-90 radioembolization may be useful for lobar or multifocal liver-dominant disease, but prior embolic or radiation exposure must be incorporated into cumulative liver and biliary risk. The COLLISION consensus regarded transarterial Y-90 and TACE-based downstaging sequences for colorectal metastases as evolving treatments under investigation rather than established components of a universal algorithm ^[154].

For portal-vein tumor thrombus, the treatment objective is not only tumor ablation but also preservation or restoration of portal flow and prevention of hepatic decompensation. Contour viable thrombus as part of the treatment target when control of the thrombus is intended, and coordinate SBRT with hepatology and interventional radiology because portal hypertension, collateral flow, and biliary drainage may change during treatment. HAIC may be paired with SBRT in selected HCC with PVTT, but the available evidence is retrospective: a 2026 propensity-matched multicenter study reported better outcomes with HAIC plus SBRT than HAIC alone, without a statistically significant difference in grade 3-4 adverse events ^[157]. This supports clinical-trial or expert-center use, not routine substitution for established systemic therapy.

Systemic therapy should generally provide disease control before and after SBRT when extrahepatic or multifocal disease remains possible. In metastatic disease, use SBRT to consolidate a response, eradicate oligoprogressive lesions while maintaining an effective systemic regimen, or delay a systemic switch when the progression is anatomically limited. The prospective phase 2 study of oligoprogressive HCC continued the existing PD-1-inhibitor-based regimen during SBRT and reported reversible grade 3-4 toxicity in 8.5% of patients; this is encouraging but does not define a universal concurrent-treatment standard ^[159]. For HCC with PVTT, retrospective data also support combinations of SBRT with targeted and immune therapy, but the reported grade 3 or higher toxicity rate of 50% in one dual-center cohort illustrates why these combinations require careful liver and hematologic surveillance and prospective validation ^[158].

Avoid routine concurrent chemotherapy or immunotherapy solely to radiosensitize liver SBRT. The biologic rationale is plausible, radiation can release tumor antigens and alter the immune microenvironment, but clinical schedules, fractionation, systemic agents, and patient selection remain heterogeneous ^[29]. When systemic treatment is continued, distinguish evidence-based disease control from experimental radiosensitization, document the rationale, and obtain medical-oncology agreement on interruption and restart dates. A practical approach is to withhold agents with uncertain radiosensitizing, bleeding, gastrointestinal, or hepatic effects around SBRT and to individualize the break according to pharmacokinetics, treatment intent, blood counts, liver function, and the proximity of bowel or stomach. The EORTC-ESTRO OligoCare consensus cited in the 2026 review permits some antibody treatments on the day of SBRT but advises interrupting other targeted therapies and immune checkpoint inhibitors for 1-2 weeks before and after treatment ^[5]. No single interval applies to every agent.

Monitor CBC, bilirubin, albumin, INR, aminotransferases, alkaline phosphatase, creatinine, and clinical evidence of ascites or encephalopathy before each fractionated course and during the post-treatment interval. Marrow toxicity becomes more plausible when systemic therapy has caused cytopenias or when large-volume liver irradiation, prior chemotherapy, or pelvic irradiation reduces reserve; hepatic toxicity is amplified by cirrhosis, prior TACE/TARE, ablation, surgery, reirradiation, and concurrent hepatotoxic treatment. If counts or liver tests deteriorate, hold the systemic agent, investigate infection, bleeding, progression, biliary obstruction, and drug-induced injury, and do not attribute every abnormality to radiation.

Transplantation requires a separate bridge strategy. SBRT can maintain or restore transplant eligibility when the lesion is unsuitable for ablation or when TACE is ineffective, contraindicated, or insufficient for durable control. The 2025 systematic review of SBRT in transplant candidates included randomized and nonrandomized studies and concluded that radiologic and pathologic responses were favorable, but the evidence base remained heterogeneous and largely nonrandomized ^[36]. Before SBRT, the transplant team should document the applicable listing and downstaging criteria, anticipated wait time, vascular invasion status, and whether radiation could compromise future reconstruction. After SBRT, use serial multiphasic MRI or CT and tumor markers according to the transplant program’s protocol; persistent enhancement early after radiation does not by itself prove viable tumor.

After transplantation, prior SBRT may produce fibrosis or adhesions in the explant bed and may alter vascular dissection, but available clinical experience does not support excluding transplantation on that basis alone. Review the prior plan with the transplant surgeon, anticipate difficult hilar or caval dissection, and coordinate immunosuppression, infection prophylaxis, and graft function. The transplant team should also decide whether post-transplant immunosuppression requires modification, since the oncologic and graft-safety consequences cannot be separated.

Pearl: Sequence SBRT around the treatment that offers the clearest route to cure, but recalculate the price of every additional modality in functional liver, not merely in tumor volume.

Acute toxicity after liver stereotactic body radiation therapy is usually mild and self-limited. Fatigue, nausea, anorexia, and vague right-upper-quadrant or epigastric discomfort commonly occur during treatment or within the first several weeks; exclude dehydration, infection, uncontrolled pain, medication effects, and disease-related symptoms before attributing them to radiation. Antiemetics, oral hydration, small frequent meals, nutritional review, and short-term analgesia are usually sufficient. Avoid routine corticosteroids unless there is a specific inflammatory or compressive indication. In a prospective study of advanced cirrhosis treated with 40 Gy in 5 fractions, one of nine patients developed grade 4 acidosis, acute hepatic encephalopathy, and hepatic failure within 1 week to 3 months, illustrating that an apparently modest treatment volume can still precipitate decompensation in a marginal liver ^[33].

Transient increases in AST, ALT, alkaline phosphatase, or bilirubin may occur during the first days to 3 months. Repeat the liver panel, albumin, INR, creatinine, and CBC; assess symptoms, volume status, and baseline Child-Pugh and ALBI scores rather than labeling an isolated enzyme rise as radiation injury. The differential includes tumor progression, biliary obstruction or cholangitis, bacterial infection, viral-hepatitis flare, portal-vein thrombosis, ischemia, alcohol or acetaminophen exposure, and systemic-treatment toxicity. Obtain blood cultures and abdominal imaging when fever, hypotension, jaundice, or focal pain is present; use Doppler ultrasound or contrast CT/MRI when portal-vein thrombosis or biliary obstruction is suspected. Hold or modify hepatotoxic systemic treatment in collaboration with the treating medical oncologist, and treat the identified cause rather than escalating liver-directed therapy empirically.

Radiation-induced liver disease (RILD) has two clinical patterns. Classical RILD reflects central-vein injury with hepatic congestion and hepatocyte necrosis; it classically presents several weeks after treatment with anicteric hepatomegaly, right-upper-quadrant discomfort, weight gain, and new ascites, usually without a marked bilirubin rise. Nonclassic RILD reflects hepatocellular loss or dysfunction and sinusoidal endothelial injury and is more likely in patients with cirrhosis, limited spared-liver volume, large treatment volumes, multifocal irradiation, prior liver-directed therapy, or combined liver treatments ^[8]. It may present with worsening transaminases or alkaline phosphatase, bilirubin elevation, a rising INR, falling albumin, ascites, encephalopathy, or a Child-Pugh increase. A practical research definition is an increase of at least 2 Child-Pugh points from baseline within 90 days; the 90-day boundary helps separate treatment effect from the natural course of cirrhosis, although the two can overlap ^[8].

Risk is determined less by the prescription dose alone than by the amount and function of liver spared. Cirrhosis, Child-Pugh B or worse disease, portal hypertension, low residual functional-liver volume, a large PTV, multifocal treatment, reirradiation, and recent or planned TACE, TARE, ablation, or systemic therapy lower the margin for additional injury. A gradual ALBI decline after SBRT may reflect both treatment and cirrhosis: in one retrospective cohort, larger lesion and PTV size, mean liver dose, and multifocality correlated with ALBI change, while coexisting cirrhosis contributed to deterioration ^[18]. Use the lowest-risk feasible dose distribution, preserve the best-functioning liver, and review cumulative exposure before retreatment.

Management of suspected RILD is supportive and multidisciplinary. Admit patients with progressive jaundice, coagulopathy, renal dysfunction, tense ascites, gastrointestinal bleeding, fever, or encephalopathy. Stop alcohol and unnecessary hepatotoxic drugs; correct hypovolemia, infection, electrolyte abnormalities, and renal precipitants; treat ascites with sodium restriction and diuretics when appropriate; perform therapeutic paracentesis for tense or respiratory-compromising ascites; and use lactulose for overt hepatic encephalopathy. Give antibiotics promptly when spontaneous bacterial peritonitis or cholangitis is suspected. Involve hepatology early and consider transplant evaluation for progressive hepatic failure. Do not assume that corticosteroids reverse RILD, and do not rechallenge with additional liver-directed therapy until the cause and residual hepatic reserve are clear.

Late toxicity reflects the dose to adjacent serial organs and central structures. Gastric or duodenal ulceration can present with epigastric pain, nausea, anemia, melena, or hematemesis months after treatment; ulcer perforation and bleeding are uncommon but potentially fatal. New pain, vomiting, gastrointestinal bleeding, or free air requires urgent CT and gastroenterology or surgical assessment; manage ulceration with proton-pump inhibition, correct coagulopathy, transfuse when indicated, and obtain endoscopic hemostasis for active bleeding. Perforation requires immediate surgical consultation. Risk rises when the stomach or duodenum overlaps the high-dose region, when treatment volume is large, when daily filling differs from simulation, and when anti-angiogenic or other radiosensitizing treatment is given. A prospective SBRT series that enforced organ-specific constraints reported no grade 2 or higher toxicity, whereas published esophageal fistula cases in the planning literature received point doses of approximately 51.5-52 Gy and 1-cc doses of 48.1-50 Gy, underscoring the danger of focal high dose to hollow organs ^[11].

Central or hilar treatment may injure the biliary tree. Late biliary stricture can cause cholestatic enzyme elevation, jaundice, recurrent cholangitis, or secondary hepatic decompensation; the risk is greater with pre-existing obstruction, biliary stents or instrumentation, infection, and reirradiation. Evaluate with liver tests, blood cultures when febrile, multiphasic CT or MRI, and MRCP; perform ERCP or percutaneous biliary drainage for clinically significant obstruction or cholangitis. In a small MR-guided series of HCC with bile-duct tumor thrombus, the maximum radiation-attributed event was a transient grade 3 bilirubin elevation that resolved without biliary intervention, but the limited sample cannot establish safety for central bile-duct irradiation ^[166]. Prefer more fractions, lower dose per fraction, adaptive planning, and explicit central-bile-duct constraints when tumor control permits.

Chest-wall pain, rib fracture, and localized skin injury occur when a peripheral or subcapsular target abuts the ribs, intercostal muscles, or skin. Pain may begin months after treatment and can persist; a fracture may present as focal tenderness or an incidental imaging finding. Examine the skin and chest wall, obtain CT or targeted radiographs for persistent focal pain, and treat with acetaminophen when safe, topical measures for dermatitis, and short-course analgesia; use opioids or pain-specialist referral for severe pain. Manage fracture conservatively unless instability or another cause is suspected. Dose reduction, a larger number of fractions, and chest-wall or rib sparing reduce risk, but target coverage must not be sacrificed without multidisciplinary agreement.

Vascular injury is uncommon but clinically consequential. Radiation-associated portal-vein or hepatic-vein stenosis, thrombosis, or hemorrhage can mimic tumor progression and may cause new ascites, abdominal pain, portal-hypertension complications, or hepatic dysfunction. Compare multiphasic imaging with baseline studies and use Doppler ultrasound or CT/MR angiography; distinguish bland thrombus from tumor thrombus by enhancement, expansion, continuity with viable tumor, and interval behavior. Treat infection, obstruction, or thrombosis according to the cause and bleeding risk; involve hepatology, interventional radiology, and vascular specialists before anticoagulation or intervention. Prior vascular invasion, portal hypertension, thrombocytopenia, vascular instrumentation, and reirradiation increase concern.

Spinal-cord injury and cardiac or pericardial injury are rare because most liver SBRT plans avoid these structures, but risk becomes relevant for posterior, superior, left-lobe, or reirradiation targets. New weakness, sensory change, gait disturbance, or bowel/bladder dysfunction requires urgent spinal MRI and corticosteroid treatment for suspected cord compression or inflammatory myelopathy, with neurosurgical and radiation-oncology review. New chest pain, dyspnea, syncope, arrhythmia, or unexplained troponin elevation warrants ECG, troponin testing, echocardiography, and cardiology assessment. Do not attribute these symptoms to benign post-treatment fatigue until cord compression, pulmonary embolism, infection, ischemia, and tumor progression have been excluded.

Use CTCAE v5.0 for every toxicity, recording the symptom, laboratory abnormality, attribution, onset, maximum grade, intervention, and resolution. For clinical communication, grade 1 is mild or asymptomatic, grade 2 requires medical intervention or limits instrumental activities, grade 3 is severe or limits self-care, grade 4 is life-threatening, and grade 5 is death. In the toxicity literature, acute toxicity is defined as occurring within 90 days after treatment, late toxicity as occurring after 90 days, and grade 3 or higher as high-grade toxicity ^[8]. Grade the event that occurred, not merely the worst laboratory value: a bilirubin rise with cholangitis, obstruction, or hepatic failure should be recorded with the associated clinical syndrome and its intervention.

When post-SBRT symptoms or laboratory changes are discordant with the expected course, reassess rather than assuming radiation injury. Tumor progression may cause enlarging mass effect, vascular invasion, biliary obstruction, or constitutional decline; infection may cause fever, leukocytosis, hypotension, or cholangitis; viral-hepatitis flare may produce marked aminotransferase elevation; portal-vein thrombosis may cause new portal-hypertension findings; and systemic treatment may produce overlapping hepatic, gastrointestinal, or constitutional toxicity. The safest attribution uses timing, baseline reserve, dose distribution, serial laboratory trends, microbiology, vascular and biliary imaging, medication review, and multidisciplinary examination together.

Pearl: The clinically meaningful endpoint is not a normal post-treatment scan or an isolated abnormal laboratory value; it is preservation of functioning liver and adjacent-organ integrity while promptly identifying the small subset of patients whose apparently minor toxicity is evolving into hepatic, gastrointestinal, biliary, vascular, or neurologic decompensation.

Response assessment after liver stereotactic body radiation therapy (SBRT) must separate delayed radiation response from viable tumor. Size alone is inadequate because irradiated lesions may regress slowly, remain temporarily enlarged, or retain arterial enhancement despite effective treatment ^{[169][175][176]}. Establish a post-treatment baseline with multiphasic contrast-enhanced magnetic resonance imaging (MRI) or computed tomography (CT) at approximately 4-6 weeks when feasible, then image about every 3 months during the first year; if findings remain reassuring, extend the interval to every 4-6 months according to liver cancer type, recurrence risk, transplant or salvage plans, and systemic disease status ^[170]. Perform clinical review and liver-function testing at the same visits, including bilirubin, albumin, INR, AST, ALT, alkaline phosphatase, and a clinically appropriate assessment of ascites, encephalopathy, jaundice, pain, weight, and performance status. A worsening laboratory profile is not synonymous with tumor progression: evaluate obstruction or cholangitis, infection, portal-vein thrombosis, viral hepatitis, ischemia, alcohol or drug injury, and systemic-therapy toxicity before attributing the change to radiation ^[170].

For hepatocellular carcinoma (HCC), measure alpha-fetoprotein (AFP) when it was elevated before treatment and interpret the trajectory alongside imaging rather than as a stand-alone response test. AFP often falls most rapidly during the first 3 months, but a normal value does not exclude residual tumor and a rising value warrants earlier imaging or additional staging ^[169][170]. Continue surveillance of the untreated liver for new HCC using the standard LI-RADS diagnostic approach, because successful treatment of the index lesion does not prevent new intrahepatic tumors ^[170]. In intrahepatic cholangiocarcinoma and liver metastases, use the same liver protocol but add contrast-enhanced chest and systemic imaging according to the primary tumor, prior metastatic sites, systemic-treatment plan, symptoms, and suspected pattern of failure. PET/CT is not a substitute for multiphasic liver imaging; reserve it for FDG-avid metastases or cholangiocarcinoma when metabolic response, occult extrahepatic disease, or an equivocal treated focus will change management ^[172][174].

How to judge the treated lesion

Use RECIST 1.1 to report overall anatomic response when required for a clinical trial or systemic-therapy assessment, but do not use it alone to declare local SBRT failure. RECIST 1.1 measures the longest diameter and can label a slowly involuting, successfully irradiated lesion as stable disease. mRECIST measures the arterial-enhancing component, which is closer to viable HCC than total lesion size, yet early persistent arterial phase hyperenhancement (APHE) can produce a false impression of viable tumor after SBRT ^[176][177]. In a successfully treated HCC series, persistent central APHE occurred from 3 through 12 months and persistent washout was common despite pathological or biochemical response ^[175].

For HCC, report the LI-RADS Treatment Response Assessment (TRA) category and specify that the radiation-treatment algorithm was used. The 2024 radiation TRA classifies absent enhancement as LR-TR Nonviable, stable or decreasing mass-like enhancement as LR-TR Nonprogressing, and new or enlarging mass-like enhancement as LR-TR Viable; it deliberately avoids treating stable post-radiation enhancement as residual cancer ^[170]. The evidence base remains less mature than the terminology: the update is supported mainly by retrospective data, and the minimum duration of stability required to exclude microscopic viable tumor is not established ^[170]. A late-term assessment is generally more discriminating than a 3-6-month assessment, so a reassuring early scan should be confirmed longitudinally rather than used to stop surveillance ^[171].

MRI is preferred when available because it combines multiphasic enhancement with T2-weighted imaging, diffusion-weighted imaging (DWI), apparent diffusion coefficient (ADC) mapping, and, when gadoxetate is used, hepatobiliary-phase imaging ^[170]. Benign post-SBRT change tends to become less T2-bright and less DWI-bright, with rising ADC, while viable tumor more often shows increasing T2 or DWI signal, persistent low ADC, new or enlarging nodular enhancement, and washout ^{[169][175][177]}. Do not diagnose recurrence from DWI signal alone. Confirm true restriction on the ADC map and correlate it with a discrete enhancing focus, because motion, susceptibility, and T2 shine-through can mimic restriction; radiation edema and inflammation can also produce T2 hyperintensity ^[170]. Hepatobiliary-phase hypointensity in the irradiated parenchyma is expected and reflects radiation-related hepatocyte dysfunction rather than tumor by itself ^[169][180].

When a lesion is LR-TR Nonprogressing or otherwise equivocal, compare the current study with the planning or pretreatment examination and every intervening scan. Repeat multiphasic MRI or CT in approximately 3 months rather than retreating solely for stable enhancement ^[170]. Escalate sooner when a focus enlarges, becomes nodular or mass-like, develops washout, shows convincing restriction with low ADC, or is accompanied by rising AFP or clinical deterioration. Ask for expert abdominal-radiology review and multidisciplinary discussion when registration, phase timing, motion, hemorrhage, lipiodol, or biliary intervention obscures interpretation. Obtain image-guided core biopsy when imaging and biomarker or clinical findings remain discordant after short-interval reassessment and histology would change management, for example, before high-risk salvage therapy, transplant exclusion, or a change in systemic treatment. Biopsy is not a routine requirement for every treated lesion because most response studies rely on serial imaging, clinical course, AFP, or explant pathology rather than percutaneous sampling ^[169][175].

For liver metastases, combine RECIST 1.1 size change with enhancement morphology and, when the primary tumor is FDG-avid, metabolic response on positron emission tomography (PET)/CT. A new lobulated enhancing component is particularly suspicious: in a retrospective metastasis cohort, lobulated enhancement preceded size-based progression in half of progressive lesions ^[181]. PET uptake may remain transiently elevated during early radiation inflammation; in one study, controlled lesions declined toward a plateau over the first months, whereas persistent high uptake was more concerning for failure ^[182]. Interpret PET with the same longitudinal discipline as MRI or CT and avoid applying a single SUV cutoff across scanners, tumors, or time points.

Expected post-SBRT findings and their interpretation

The treated zone commonly evolves from early hyperemia and geographic arterial enhancement to delayed enhancement, increasing T1 signal, decreasing T2 signal, focal low attenuation on CT, capsular retraction, and focal volume loss. These findings reflect inflammation followed by fibrosis and remodeling, not necessarily residual tumor ^[169][175]. A geographic or rim-like enhancement pattern that conforms to the irradiated volume and remains stable or decreases is more consistent with focal liver reaction (FLR) than viable disease. By contrast, a new or enlarging nodular or mass-like focus, especially with washout, increasing T2/DWI signal, or a corresponding rise in tumor marker, should be treated as suspicious ^[169][177].

Recognizing failure

Classify recurrence by location because each pattern changes the therapeutic question. Local failure is progressive viable tissue within or immediately along the treated volume, usually seen as interval enlargement or a new/enlarging nodular or lobulated enhancing component with washout, increasing T2/DWI signal, or persistent metabolic activity. Intrahepatic failure consists of new lesions elsewhere in the liver or progression of untreated lesions; continue whole-liver surveillance even when the SBRT target is controlled. Extrahepatic failure includes new or progressive disease in the lung, nodes, bone, peritoneum, or other systemic sites and should be sought with chest and disease-directed systemic imaging when the clinical course or primary tumor warrants it ^{[48][174][181]}.

A single suspicious feature should trigger structured reassessment, not reflexive retreatment. Confirm that the finding is real on equivalent phases and planes, compare it with the prior examination and treatment geometry, review AFP and liver function, and present the case at multidisciplinary review. Conversely, a stable or slowly involuting lesion with geographic enhancement, capsular retraction, volume loss, declining T2/DWI signal, and no biochemical or clinical progression is usually radiation effect; continued surveillance is safer than declaring local failure from persistent enhancement alone ^{[169][170][175][177]}.

Local control is the most direct measure of what liver stereotactic body radiation therapy (SBRT) accomplishes. Overall survival is harder to interpret because it reflects liver reserve, extrahepatic disease, performance status, subsequent therapy, and competing cirrhosis-related mortality. A large Belgian multicentre registry illustrates this distinction: among 354 evaluable lesions, local control at 1, 2, and 5 years was 69.9%, 52.2%, and 32.4%, whereas overall survival for the treated population was 74.3%, 49.7%, and 19.4% at the same time points. ^[4] These outcomes should not be transferred across series without checking lesion size, histology, dose, imaging criteria, follow-up duration, and the proportion of patients with extrahepatic disease.

Hepatocellular carcinoma

For hepatocellular carcinoma (HCC), SBRT produces durable control in carefully selected patients with preserved hepatic function, including patients who are unsuitable for resection or ablation, have residual disease after transarterial therapy, or require salvage treatment. In the Belgian registry, HCC lesions had higher local control than metastases: 82.8% at 1 year, 76.2% at 2 years, and 43.6% at 5 years. ^[4] A 2026 systematic review of tumors at least 5 cm reported pooled 1- and 2-year local-control rates of 81% and 69%, respectively, with corresponding overall-survival rates of 60% and 37%; the analysis supports feasibility for large tumors but remains dependent on retrospective cohorts and heterogeneous regimens. ^[187]

HCC local control is not synonymous with eradication of the patient's cancer. New intrahepatic tumors, portal-vein tumor thrombosis, extrahepatic spread, and cirrhosis may determine progression-free and overall survival despite control of the irradiated lesion. In selected patients with portal-vein tumor thrombosis, adding SBRT to lenvatinib in a retrospective propensity-matched cohort was associated with longer median overall survival and progression-free survival than lenvatinib alone, and the combination had a higher conversion rate to salvage hepatectomy; the signal was confined largely to Cheng type I-II thrombosis, so it cannot establish a universal benefit for advanced portal invasion. ^[46]

Downstaging and transplantation are higher-order endpoints. SBRT may be judged successful when it produces sustained radiologic or metabolic response that permits resection or transplantation, but conversion cohorts are highly selected and do not establish the probability of conversion for an unselected HCC population. Reports of complete necrosis in explanted specimens are biologically persuasive, yet pathological response is not a substitute for randomized survival evidence. ^[1]

Intrahepatic cholangiocarcinoma

For intrahepatic cholangiocarcinoma (iCCA), the evidence base is smaller and less mature than for HCC. SBRT is most defensible for a focal, unresectable or recurrent lesion when adequate liver function can be preserved and surgery, ablation, or systemic therapy alone is insufficient. The principal rationale is anatomical: centrally located or perihilar tumors may be inaccessible to ablation, while a conformal external-beam treatment can cover the target without needle traversal. A contemporary multidisciplinary review describes SBRT as a non-invasive option for centrally located or perihilar iCCA not amenable to ablation, while emphasizing that prospective, modality-stratified trials remain necessary. ^[40]

Local control, progression-free survival, and overall survival should therefore be reported separately in iCCA. A controlled local lesion may delay intrahepatic progression or relieve tumor-related symptoms, but available evidence does not establish that SBRT routinely improves overall survival over resection, systemic therapy, TACE , or TARE . TARE has a more developed role for selected infiltrative, unilobar, or portal-vein-compromised disease, whereas SBRT remains an individualized focal treatment within multimodality care. ^[40] Symptom relief is a reasonable palliative endpoint when pain, capsular discomfort, or focal tumor-related obstruction is present, but symptom response has not been consistently measured across the available SBRT literature and should not be inferred from radiographic control alone.

Metastatic disease

In liver metastases, SBRT is a metastasis-directed treatment rather than a replacement for systemic disease control. The strongest rationale is a limited number of technically treatable lesions in patients with absent, controlled, or concurrently treated extrahepatic disease. Resection remains preferred when complete resection is feasible, but only a minority of patients with liver metastases are surgical candidates; SBRT is particularly useful for deep, subcapsular, perihilar, vessel-adjacent, or otherwise needle-inaccessible lesions. ^[4]

Histology and dose matter. In the Belgian registry, metastatic lesions had 1-, 2-, and 5-year local-control rates of 67.6%, 48.2%, and 31%. ^[4] A prospective study of one to three colorectal metastases treated with 50-60 Gy in five fractions reported 2-year local control of 95.0%, progression-free survival of 61.3%, and overall survival of 88.1%, but patients were inoperable or declined surgery and the cohort was small. ^[11] For colorectal metastases, the delivered minimum or near-minimum PTV dose may be more informative than the nominal prescription: a retrospective analysis found PTV minimum BED10 and polymetastatic disease to be independent predictors of freedom from local progression, with a 100-Gy BED10 cut point associated with better 1-year control. ^[24]

Local control may delay intrahepatic progression, preserve an effective systemic regimen, or postpone a change in systemic therapy; these are clinically meaningful benefits even when overall survival is unchanged. They must be separated from survival claims because patients selected for SBRT often have fewer metastases, better performance status, longer expected survival, and more responsive systemic disease. Evidence for breast, neuroendocrine, and other metastatic histologies remains predominantly retrospective, and the optimal approach to multifocal disease is unresolved.

Comparative evidence

Comparative evidence is strongest for selected HCC and remains vulnerable to selection bias. A propensity-score-weighted analysis of treatment-naive BCLC stage 0 HCC found that SBRT and resection had comparable adjusted 3-year overall survival, disease-free survival, and recurrence outcomes, although patients receiving SBRT were older and had more comorbidity and poorer liver function before adjustment. ^[161] A separate post-hoc analysis of two randomized trials in recurrent solitary HCC no larger than 5 cm found no significant difference between SBRT and resection in 3-year local progression-free survival, progression-free survival, or overall survival; because treatment assignment was not a single randomized SBRT-versus-resection trial, the result is supportive rather than definitive. ^[162]

For small HCC, a 2026 meta-analysis of 12 comparative studies found lower local recurrence and better local progression-free survival with external-beam radiotherapy than with radiofrequency or microwave ablation, but no significant overall-survival difference. ^[14] This pattern is clinically plausible: SBRT avoids the heat-sink effect and reaches lesions that are difficult to puncture, whereas ablation remains attractive for small, accessible tumors. The comparison does not prove superiority because the underlying studies included retrospective and prospective designs with different patient selection, tumor locations, and liver function.

For colorectal liver metastases, surgery remains the standard when complete resection is technically and physiologically feasible. A small retrospective comparison reported a nonsignificant overall-survival difference between resection and SBRT, with median survival of 51 versus 32 months; the sample size and nonrandomized allocation prevent a treatment-effect conclusion. ^[190] No randomized prospective SBRT-versus-resection trial has established equivalence across metastatic histologies. Similarly, SBRT should not be declared superior to TACE , TARE , thermal ablation, or systemic therapy solely because it provides high local control in selected lesions. Comparisons with observation are particularly confounded: observation is generally chosen for patients with limited life expectancy, uncontrolled systemic disease, poor hepatic reserve, or low expected benefit from local therapy.

Prognostic factors

The most reproducible adverse factors are large target volume, multiple lesions, inadequate biologic dose or target cold spots, poor baseline liver function, portal hypertension, elevated tumor burden markers, extrahepatic disease, poor performance status, and prior treatment that either selects resistant clones or reduces hepatic reserve. These factors operate through different mechanisms and should not be collapsed into a single “SBRT risk” label.

Tumor size and number. Increasing size increases the volume of liver receiving low and intermediate doses, makes complete target coverage more difficult, and often signals more infiltrative or hypoxic biology. In the national registry, PTV volume independently predicted both local control and overall survival; lesions with PTV volumes above 130 cc fared worse than those below 45 cc. ^[4] In colorectal metastases, polymetastatic disease independently predicted local failure, while clinical series identify three or more lesions as an adverse selection feature rather than an absolute contraindication. ^[24]

Histology and tumor biology. HCC generally demonstrates more favorable local control than colorectal metastases in registry data, but histology was not independently prognostic after adjustment in the Belgian analysis. ^[4] Colorectal metastases may require particular attention to robust minimum PTV dose, because dose heterogeneity and resistant subvolumes can dominate failure despite an adequate prescription BED. ^[24] iCCA requires individualized consideration of infiltrative spread, ductal extension, and early systemic dissemination; a technically controlled lesion may not translate into prolonged survival. ^[40]

Biologic dose and delivered coverage. BED10 of at least 100 Gy is commonly associated with better local control in retrospective liver-SBRT cohorts, but dose is confounded by lesion size, location, liver function, and treatment intent. In the Belgian registry, BED10 at least 100 Gy remained independently associated with better local control and overall survival. ^[4] The more clinically useful question is whether the entire biologically relevant target received adequate dose: in colorectal metastases, PTV minimum BED10, not nominal prescription alone, was the strongest dosimetric predictor of local control. ^[24] Dose escalation therefore remains a hypothesis to test, not a justification for exceeding normal-liver or luminal-organ tolerance.

Baseline liver function and portal hypertension. Child-Pugh class, ALBI grade, albumin, bilirubin, ascites, and portal hypertension influence both treatment tolerance and non-cancer mortality. A retrospective HCC study found that ALBI grade 2 and Child-Pugh class B were independently associated with poorer overall survival, supporting their complementary use rather than reliance on either score alone. ^[188] Selected Child-Pugh B patients can achieve local control with reduced-dose regimens, but a 31-patient series reported median overall survival of 14 months and progression-free survival of 9 months, underscoring the competing risk of hepatic dysfunction. ^[17] Portal hypertension should therefore be treated as a prognostic and safety variable, not merely a technical planning detail.

Alpha-fetoprotein and other biomarkers. Baseline and post-treatment alpha-fetoprotein (AFP) can refine prognostic assessment in HCC when elevated at diagnosis, but normal AFP does not exclude viable tumor and AFP kinetics should not replace multiphasic imaging. Evidence for early response biomarkers is emerging: a prospective study found that change in diffusion-derived ADC correlated with tumor-size change and PIVKA-II change but not with AFP change. ^[186] AFP is therefore supportive rather than definitive, particularly in AFP-nonsecreting tumors.

Extrahepatic disease and performance status. Extrahepatic disease reduces the chance that liver local control will improve overall survival, although SBRT may still be appropriate for oligoprogression or liver-dominant disease when systemic therapy remains effective. Performance status is both a treatment-selection variable and an independent survival determinant; in the Belgian registry, poorer performance status and larger PTV were associated with worse overall survival. ^[4] A local intervention should be offered only when the expected time to benefit exceeds the patient's competing risk of systemic progression or hepatic decompensation.

Prior treatment. Prior systemic therapy was associated with inferior local control in the Belgian registry, possibly reflecting more advanced disease or selection of radioresistant clones rather than a direct adverse effect of the drugs. ^[4] Prior resection, ablation, TACE, TARE, or radiation may reduce functional liver reserve, distort target anatomy, or create dose-overlap uncertainty. Conversely, prior therapy can identify patients with demonstrated disease control who may benefit from consolidation or salvage SBRT. The prognostic meaning of treatment history is therefore conditional on response, residual liver function, and the reason SBRT is being considered.

Evidence standards and unresolved questions

The evidence hierarchy matters. Prospective single-arm cohorts establish feasibility and provide estimates of local control, toxicity, and patient-reported outcomes but cannot prove superiority. Retrospective comparisons can generate clinically useful hypotheses after propensity adjustment, yet residual confounding remains substantial because SBRT is preferentially offered to patients with difficult anatomy, poorer liver function, or contraindications to surgery. Registry studies improve generalizability but often use nonuniform imaging and local-control definitions. Randomized evidence remains limited: the strongest comparative HCC data include post-hoc analyses of separate randomized cohorts, while randomized dose-escalation evidence in liver metastases is still being tested rather than reported. ^[10][162]

Future studies should prespecify lesion-level local control, intrahepatic progression-free survival, overall survival, symptom relief, liver-function change, quality of life, and conversion to resection or transplantation. They should stratify by histology, lesion size and number, baseline liver function, portal hypertension, AFP, extrahepatic disease, performance status, prior treatment, and delivered minimum target dose. Without these strata, a higher local-control rate may simply reflect a more favorable patient population.

Dose escalation remains unresolved because increasing BED may improve sterilization while increasing hepatic, biliary, or gastrointestinal injury. The phase III ULTRAS protocol is testing MR-guided single-fraction 27 Gy versus 38 Gy for liver oligometastases or oligoprogression, with local control as the primary endpoint and survival, toxicity, quality of life, and intrahepatic and extrahepatic progression as secondary endpoints. ^[10] Its results will address a clinically relevant question that retrospective BED analyses cannot answer.

Combining SBRT with immunotherapy or targeted therapy is biologically attractive but clinically unsettled. Timing, radiosensitization, patient selection, and hepatic toxicity remain heterogeneous, and apparent survival gains from retrospective combination cohorts may reflect selection and systemic-treatment responsiveness. Proton therapy may reduce low-dose liver exposure in selected geometries, but no evidence establishes universal superiority over modern photon SBRT. MRI-guided adaptation can improve target visualization and respond to day-of-treatment anatomy, yet whether it improves local control, survival, or quality of life remains unproven. Multifocal treatment likewise requires prospective study because composite liver dose, shared-organ constraints, and competing systemic progression become limiting before the physics of individual-lesion ablation does.

Pearl: Report local control, intrahepatic progression, extrahepatic progression, progression-free survival, overall survival, symptom response, and liver-function preservation as separate outcomes; a high local-control rate is clinically meaningful only when the treated patient has sufficient hepatic reserve and a disease pattern in which local control can change the next therapeutic decision.