Liver SBRT

Liver stereotactic body radiotherapy (SBRT) is image-guided, high-precision external-beam radiation that delivers a tumor-ablative dose in a small number of fractions, usually five or fewer. ^[3] The technique relies on conformal dose shaping, respiratory-motion management, and frequent imaging to place a steep dose gradient around the target; this permits treatment of a discrete hepatic lesion while limiting irradiation of the uninvolved liver and nearby organs. ^[5] SBRT is therefore a treatment strategy rather than a single prescription: fraction number, dose, image guidance, motion control, and organ-at-risk constraints vary according to tumor geometry, liver function, prior treatment, and institutional practice.

The term needs separation from several neighboring modalities. Conventional external-beam radiotherapy spreads treatment over many daily fractions and generally uses a less ablative dose per fraction; SBRT concentrates the dose into fewer fractions and demands substantially tighter geometric accuracy. ^[7] Stereotactic radiosurgery traditionally describes stereotactic treatment delivered in a single fraction, most often for intracranial targets, whereas liver SBRT usually denotes stereotactic treatment of hepatic tumors delivered in one or a limited number of fractions; single-fraction liver treatment is an available but less common form of stereotactic ablative radiotherapy. ^[14] Hypofractionated chemoradiation combines radiation with concurrent chemotherapy and is defined by the combined treatment program, not simply by a small number of fractions or by stereotactic dose conformality. SBRT itself is local radiation; systemic therapy may be given before, after, or alongside it, but that sequencing does not change the radiation technique's definition. Intra-arterial radioembolization , also called transarterial radioembolization (TARE), delivers radioactive material through the hepatic arterial circulation, whereas SBRT delivers external-beam radiation from outside the body; the two can be sequenced or combined, but they are not interchangeable forms of radiation. ^[9]

The clinical problem is a balance between tumor ablation and preservation of functioning liver. A high dose can sterilize a focal target, but the usable hepatic reserve may already be reduced by cirrhosis , prior liver-directed therapy, vascular disease, or diffuse tumor involvement. Treatment selection must therefore consider not only whether the tumor can be covered, but also how much functional liver and which adjacent organs will receive consequential radiation. This balance is especially consequential in HCC, which commonly arises in cirrhotic liver; prognosis reflects both tumor progression and the risk of hepatic decompensation. ^[10] Pretreatment residual liver function may determine tolerance more strongly than individual dosimetric details, particularly during repeat irradiation. ^[7]

SBRT can occupy several positions in a multidisciplinary treatment plan. It may serve as definitive local therapy when resection, transplantation, or thermal ablation is not feasible, and it may provide a noninvasive alternative for lesions that are difficult to reach safely percutaneously or that lie near vessels, bile ducts, or hollow organs. ^[5][8] “Definitive” describes the intent to eradicate the treated lesion; it does not imply control of occult intrahepatic or extrahepatic disease, and it does not make SBRT equivalent to surgery or transplantation when those options are appropriate.

For metastatic disease, SBRT is most often a metastasis-directed treatment for selected patients with limited hepatic burden, controlled or controllable extrahepatic disease, and sufficient life expectancy to benefit from durable local control. It may be used after systemic therapy as consolidation, to treat oligoprogression while the effective systemic regimen continues, or as salvage for a previously treated or recurrent lesion. A prospective phase 2 study in oligoprogressive HCC evaluated SBRT to all progressing sites while continuing first-line PD-1 inhibitor-based therapy, illustrating the distinction between treating resistant sites and abandoning systemic therapy. ^[17] Evidence for salvage and reirradiation remains less secure: published abdominal reirradiation data are dominated by retrospective studies and show marked heterogeneity in selection, dose, fractionation, and toxicity reporting. ^[7]

In HCC, SBRT may also function as a bridge to transplantation by maintaining local control during the waiting period, or as downstaging therapy when tumor burden initially exceeds transplant criteria. A systematic review and meta-analysis of 19 predominantly nonrandomized studies reported favorable radiologic and pathologic responses in transplant candidates, but the design and selection of those studies limit direct comparison with other locoregional treatments. ^[4] SBRT may similarly be incorporated into conversion or multimodality strategies intended to make subsequent resection possible, although such use remains investigational and depends on preservation or recovery of adequate future liver function.

Palliation is a further, narrower role. In a patient with a symptomatic or threatening hepatic lesion, SBRT may relieve tumor-related morbidity or delay local progression when the expected benefit justifies the treatment burden and the risk to hepatic reserve. Palliative use should not be inferred from the word “stereotactic”: the ablative dose and technical demands of SBRT are justified only when local control is clinically valuable and the patient can tolerate treatment.

The radiobiology of hepatic stereotactic ablation begins with the linear-quadratic model , which estimates cell killing as the sum of a dose-proportional component and a dose-squared component: (E=\alpha D+\beta D^2). For a course of (n) fractions of size (d), the corresponding biologically effective dose is (BED=nd[1+d/(\alpha/\beta)]). The ratio (\alpha/\beta) describes how strongly tissue response depends on fraction size: a low ratio predicts greater sensitivity to hypofractionation, whereas a high ratio predicts less fraction-size dependence. The model is useful for comparing fractionation schedules, but it is not a direct measure of tumor control or normal-tissue toxicity. Liver SBRT is commonly described using an ablative BED10, and contemporary reviews identify BED10 values of at least 100 Gy as a frequent radiobiologic benchmark for tumor treatment ^[21].

(BED_{10}) uses (\alpha/\beta=10) Gy as a practical approximation for tumor and early-responding tissue. It is favored because it provides a common language for comparing heterogeneous SBRT prescriptions and because a higher BED10 generally represents greater predicted tumor cell kill when other determinants, tumor size, hypoxia, repopulation, and treatment accuracy, are comparable. (BED_{3}), using (\alpha/\beta=3) Gy, is often used as a conservative approximation for late-responding normal tissues, including bile ducts and other slowly repairing structures. The lower denominator magnifies the calculated effect of large dose per fraction; therefore, BED3 is particularly useful for recognizing the late-toxicity penalty of hypofractionation. These are conventions, not tissue-specific constants. A BED10 or BED3 value should never be interpreted in isolation from irradiated volume, dose heterogeneity, baseline organ function, or the spatial relationship between the target and an organ at risk.

At ablative doses, the linear-quadratic model becomes an extrapolation rather than a validated biological law. It assumes that the same fraction-size relationship continues into dose ranges in which clustered DNA damage, vascular injury, stromal remodeling, immune effects, and depletion of clonogenic cells may become increasingly important. The model also treats tissue response as if it were determined by dose and fractionation alone, whereas hepatic injury depends on the volume and functional distribution of irradiated parenchyma. The optimal dose and fractionation for patients with impaired liver function remain unknown, and the literature continues to describe fractionation as an area of debate ^[22]. Use BED to compare plans; do not use it as a substitute for clinical judgment about functional liver reserve or adjacent-organ anatomy.

The dominant antitumor mechanism remains irreparable DNA damage . Ionizing radiation produces direct DNA ionization and indirect injury through reactive oxygen species, generating base damage, single-strand breaks, and, most consequentially, clustered double-strand breaks. Tumor cells that cannot repair this damage undergo mitotic catastrophe, apoptosis, senescence, or loss of reproductive capacity. Hypoxia attenuates the indirect component because oxygen stabilizes radiation-induced free-radical damage; this helps explain why poorly perfused tumor regions may be relatively radioresistant. Fractionation can permit partial reoxygenation between treatments, allowing previously hypoxic cells to become more radiosensitive, while also permitting repair in normal tissues. These mechanisms support the biologic rationale for high-dose, highly conformal treatment, but they do not establish that any single BED threshold guarantees ablation.

Radiation also affects the tumor beyond the clonogenic cell. Endothelial injury, microvascular collapse, altered perfusion, and subsequent stromal inflammation can deprive surviving tumor cells of oxygen and nutrients. In parallel, radiation may remodel the tumor microenvironment by changing antigen presentation, cytokine signaling, and lymphocyte trafficking. The immune consequence is bidirectional: SBRT can promote immunogenic cell death and an antitumor response, yet treatment can also produce substantial systemic lymphopenia, with loss of CD4-positive and CD8-positive T cells, B cells, and natural-killer cells ^[30]. Reviews describe a potential interaction between ablative radiation and immune-checkpoint blockade, but the clinical abscopal effect, regression of untreated disease outside the irradiated field, remains inconsistent and should be regarded as a biologic possibility rather than a dependable treatment endpoint ^[25]. Dose, fractionation, irradiated blood volume, tumor immune phenotype, and concomitant therapy probably determine whether immune stimulation or immunosuppression predominates; a recent review specifically describes a dose-dependent immunomodulatory window rather than a uniformly increasing immune effect with dose ^[24].

The liver behaves predominantly as a parallel organ . Functional subunits are distributed throughout the parenchyma, so focal irradiation can be tolerated when enough uninvolved, functioning liver remains. Toxicity therefore reflects not only the maximum dose but also the volume, location, and functional quality of spared liver. This differs from a serial organ , in which injury to a short critical segment can interrupt the function of the entire structure. The stomach, duodenum, bowel, and central bile ducts consequently require strict attention to small-volume high-dose exposure: a focal ulcer, perforation, obstruction, or biliary stricture may occur even when most of the organ receives little radiation. In clinical imaging after liver SBRT, higher dose exposure to the central hepatobiliary tract and proximity of a lesion to that tract have been associated with bile-duct dilatation and biochemical hepatobiliary changes ^[27].

Radiation-induced liver disease (RILD) is best understood as a spectrum rather than a single syndrome. Classical RILD reflects radiation injury to central veins and sinusoids, followed by venous outflow obstruction, retrograde congestion, hepatocyte necrosis, and hepatomegaly; the clinical picture may include anicteric hepatomegaly, ascites, and deterioration in liver function. Nonclassic RILD is more characteristic of patients with pre-existing liver disease and is expressed primarily as hepatocellular loss or dysfunction with sinusoidal endothelial injury. In practice, nonclassic injury may present as worsening bilirubin or albumin, coagulopathy, ascites, encephalopathy, or a clinically meaningful rise in the Child-Pugh score . The distinction is mechanistic, but overlap with tumor progression, infection, portal-vein thrombosis, medication toxicity, and natural cirrhotic decompensation is common. Classical RILD has been described as central-vein occlusion and obliteration with retrograde congestion and hepatocyte necrosis, whereas nonclassical RILD is associated with hepatocellular loss or dysfunction and sinusoidal endothelial damage ^[22].

Baseline cirrhosis , portal hypertension , and limited hepatic reserve lower the margin for injury because the liver begins with fewer effective functional units and less regenerative capacity. Albumin, bilirubin, coagulation, ascites, encephalopathy, platelet count, portal-vein patency, and the albumin-bilirubin score therefore provide biologic context that a geometric dose-volume histogram cannot supply. Post-treatment decline in liver function may be gradual and multifactorial: in one retrospective series, ALBI deterioration after SBRT correlated with lesion size, planning target volume, mean liver dose, and multifocality, while coexisting cirrhosis contributed independently ^[26]. A patient with preserved reserve may tolerate focal parenchymal injury without clinical failure; the same dose distribution may precipitate decompensation in a patient with portal hypertension, marginal albumin, active ascites, or substantial prior liver-directed treatment. The relevant therapeutic objective is thus not simply tumor ablation, but tumor ablation while preserving enough functioning parallel liver to sustain portal and synthetic function.

For liver SBRT, the indication is defined less by a fixed tumor size or number than by whether focal ablation can meaningfully change the patient’s disease course without exhausting functional liver. The decision should be made in a multidisciplinary conference that includes hepatology, transplant surgery, liver surgery, interventional radiology, medical oncology, diagnostic radiology, and radiation oncology. SBRT is a local treatment: it can sterilize visible disease, but it does not substitute for systemic therapy when occult or extrahepatic disease is likely. ^[3]

Hepatocellular carcinoma

Localized HCC unsuitable for surgery or thermal ablation. Resection and transplantation remain the principal curative treatments for HCC, but SBRT is a reasonable definitive option when resection would remove excessive functioning liver, transplantation is unavailable or inappropriate, or percutaneous ablation is unsafe or technically ineffective. Typical examples include a lesion near major vessels, the hepatic dome, the liver capsule, the hilum, or an adjacent stomach or bowel loop, and a lesion poorly visualized or inaccessible for probe placement. ^[33] The rationale is anatomical: SBRT does not require needle access and is not limited by the thermal-sink effect that reduces the efficacy of radiofrequency ablation near flowing blood. ^[8]

For a solitary small HCC, SBRT should be viewed as an alternative rather than an automatic replacement for resection, transplantation, or thermal ablation. A prospective phase 2 study of previously untreated solitary HCC in patients unsuitable for or refusing surgery and radiofrequency ablation used 40 Gy in 5 fractions; the median tumor size was 2.3 cm, 3-year local control was 93%, and 3-year overall survival was 82%, although enrollment closed early and the study was underpowered. ^[40] In recurrent solitary HCC measuring 5 cm or less, a randomized trial found better 2-year local progression-free survival with SBRT than radiofrequency ablation, 92.7% versus 75.8%, while progression-free survival, overall survival, and safety were comparable. ^[38] These results support SBRT particularly when recurrence is difficult to ablate, but they do not establish superiority over resection or transplantation.

Unresectable or medically inoperable HCC. SBRT is most defensible as curative-intent local therapy for limited intrahepatic disease in a patient who cannot undergo resection, transplantation, or ablation because of comorbidity, poor operative reserve, tumor location, or prior treatment. The treatment may be definitive for the treated lesion, consolidative after an incomplete response to another liver-directed therapy, or salvage for recurrence. In the phase 3 NRG/RTOG 1112 trial, patients with locally advanced HCC unsuitable for or refractory to standard local-regional therapy were randomized to sorafenib alone or SBRT followed by sorafenib; median overall survival was 12.3 versus 15.8 months, and median progression-free survival was 5.5 versus 9.2 months, respectively. ^[36] The adjusted overall-survival analysis favored SBRT, but the trial accrued slowly and closed early after systemic-treatment standards changed; therefore, SBRT should complement rather than displace contemporary first-line systemic therapy.

Compared with transarterial chemoembolization, SBRT offers a nonarterial option when arterial supply is unfavorable, embolization is unsafe, or prior embolotherapy has compromised treatment feasibility. A meta-analysis of randomized and propensity-score-adjusted studies found comparable overall survival between SBRT and transarterial chemoembolization but better local control with SBRT; substantial heterogeneity limits the certainty of that comparison. ^[37] Transarterial chemoembolization remains attractive for multifocal, liver-confined HCC when arterial treatment is feasible, whereas SBRT is generally better suited to a small number of discrete targets. Transarterial radioembolization may be preferable when disease is more diffuse or when selective arterial microsphere therapy can treat a larger intrahepatic territory; SBRT is preferable when the clinical problem is a geometrically definable lesion and preservation of uninvolved liver is paramount. ^[33]

Macrovascular invasion. Portal-vein or hepatic-vein tumor thrombus does not make SBRT curative by itself, but it can be a useful local component of treatment when the thrombus is symptomatic, threatens portal flow, or is the dominant site of progression. Radiation may reduce tumor thrombus and restore portal venous patency, potentially allowing subsequent liver-directed or systemic treatment. The evidence is less mature than for conventional localized HCC: patients with macrovascular invasion are generally classified as advanced HCC and are commonly offered systemic therapy, while SBRT is selected for carefully chosen patients or used in combination strategies. ^[16]

A propensity-score-matched comparison of surgery and SBRT in HCC with macrovascular invasion found similar median overall survival of 16 months after matching; surgery had lower 1-year local failure, whereas distant failure was more frequent after surgery, and in-hospital mortality was 9% after surgery while 14% of SBRT patients experienced post-treatment liver impairment. ^[35] These nonrandomized data do not prove equivalence. They support SBRT when surgery carries prohibitive morbidity or when the treatment objective is local control within a broader systemic strategy. Combination treatment with transarterial chemoembolization and a tyrosine-kinase inhibitor improved overall survival compared with transarterial chemoembolization and tyrosine-kinase inhibitor therapy alone in one randomized study of HCC with portal-vein thrombosis, but this approach remains dependent on patient selection and institutional expertise. ^[16]

Bridge to transplantation and downstaging. Use SBRT as a bridge when a transplant candidate needs durable control during the waiting period, and consider it for downstaging when the treated disease can plausibly be brought within transplant criteria. The objective is not merely radiographic shrinkage; viable tumor, vascular invasion, tumor biology, wait-list dropout risk, and the feasibility of transplantation must all be reassessed. A 2025 systematic review and meta-analysis of 19 predominantly nonrandomized studies involving 476 patients reported a pooled radiologic objective response of 61.2%, pathologic response of 83.8%, 5-year overall survival of 76.8%, and grade 3 or higher toxicity in 1.2%. ^[4] These results are encouraging but vulnerable to selection bias, because patients reaching transplantation are a selected subgroup; SBRT should therefore be considered one locoregional option alongside transarterial chemoembolization, transarterial radioembolization, and ablation rather than a proven superior bridge.

Oligoprogression. In a patient receiving an otherwise effective systemic regimen, irradiate a limited number of progressing HCC sites when progression is confined to those sites and continuation of systemic therapy is clinically valuable. This strategy can delay a switch from an effective regimen, preserve systemic disease control, and avoid exposing the patient to a new systemic toxicity. Prospective phase 2 data enrolled patients with no more than five progressing lesions in no more than three organs, treated all progressing sites with SBRT while continuing first-line PD-1 inhibitor-based therapy plus lenvatinib; median progression-free survival was 11.3 months, and grade 3-4 toxicity occurred in 8.5%. ^[17] The study was single-arm, so the result supports feasibility and a treatment hypothesis rather than a universal standard.

Intrahepatic cholangiocarcinoma and other primary hepatic malignancies

For intrahepatic cholangiocarcinoma, SBRT is best regarded as selected local therapy rather than an established substitute for surgery. Resection remains the preferred potentially curative treatment for anatomically resectable disease, while systemic therapy is central for unresectable, multifocal, or metastatic disease. SBRT may be considered for a solitary or limited number of unresectable lesions, residual disease after systemic therapy, local recurrence, or a lesion in which surgery would cause unacceptable loss of liver or abutment of critical structures. Modern radiation therapy has become an important modality for intrahepatic and extrahepatic biliary cancers, but the evidence base is less definitive than that supporting SBRT for selected HCC and liver metastases. ^[1]

For other primary hepatic tumors, including selected rare primary liver malignancies, SBRT should generally be discussed in a specialist multidisciplinary setting or clinical trial. The indication is strongest when disease is localized, surgery is not feasible, and a safe high-dose plan can be produced; diffuse intrahepatic spread, uncontrolled extrahepatic disease, and poor hepatic reserve favor systemic or supportive approaches. Retrospective comparisons of photon SBRT with carbon-ion therapy remain hypothesis-generating rather than practice-defining, because no standardized carbon-ion protocol has been established and prospective validation is lacking. ^[44]

Liver metastases

SBRT for metastatic disease is a metastasis-directed treatment, not a histology-independent cure. Offer it when the liver burden is limited enough to treat safely, the lesions are technically targetable, and the primary cancer is controlled or has a realistic systemic treatment plan. The most favorable pattern is one to three, sometimes up to five, lesions with limited total volume, no rapidly progressive extrahepatic disease, and adequate uninvolved liver. More numerous lesions, large target volumes, bilobar distribution, diffuse replacement of liver, or uncontrolled disease outside the liver reduce the likelihood that local treatment will alter survival, even if each treated lesion can be controlled. Across heterogeneous liver-metastasis series, an international meta-analysis reported 1-, 2-, and 3-year local-control rates of 85%, 75%, and 68%, respectively; grade 3 or higher toxicity occurred in fewer than 10% of cases and in fewer than 5% of more recent series. ^[8]

Resection remains the preferred local treatment when complete clearance is feasible with adequate future liver remnant and acceptable operative risk. Thermal ablation is particularly competitive for lesions smaller than 3 cm that are accessible and distant from major vessels, bile ducts, and bowel. SBRT becomes more attractive for lesions near vessels, beneath the diaphragm, subcapsular lesions, lesions poorly seen on ultrasound, or patients unable to tolerate anesthesia or an invasive procedure. No randomized trial has directly compared SBRT with resection for liver metastases, and retrospective comparisons are confounded because SBRT is often used for larger or anatomically difficult tumors. ^[8] SBRT and ablation should therefore be regarded as complementary, and different lesions in the same patient may appropriately receive different local treatments.

Colorectal cancer. For colorectal liver metastases, resection remains the reference treatment when an R0 resection is feasible, and conversion chemotherapy may convert initially unresectable disease to resectability. SBRT is appropriate for unresectable or medically inoperable lesions, residual lesions after surgery, lesions in unfavorable locations, or oligoprogression during an otherwise effective systemic regimen. ASCO guidance has supported SBRT for colorectal patients with liver oligometastases who are not candidates for resection, while European guidance lists both SBRT and thermal ablation as options. ^[8] In a retrospective strategy combining surgery with planned SBRT for residual unresectable and unablatable lesions up to 5 cm, the 2-year cumulative incidence of local failure was 9.6%, but the study was nonrandomized and intrahepatic recurrence remained the dominant first progression pattern. ^[42] Hepatic arterial infusion, systemic chemotherapy, and selected transarterial approaches remain important alternatives because SBRT treats visible targets and does not control microscopic or rapidly emerging disease.

Neuroendocrine, breast, lung, renal, and melanoma metastases. Treat SBRT as part of an oligometastatic or oligoprogressive strategy when the primary tumor and systemic disease are controlled or controllable, the patient has limited liver burden, and the untreated disease is not expected to dominate prognosis. Breast and lung metastases may be particularly suitable for consolidation in patients with durable systemic control; renal-cell and melanoma metastases require careful attention to dose, lesion size, and the possibility of relative radioresistance. These are selection principles rather than evidence of equal SBRT efficacy across histologies: mixed-primary studies include colorectal, lung, breast, and other cancers, but no randomized evidence establishes SBRT as superior to resection, ablation, or continued systemic therapy for each of these cancers. ^[8][34]

Observation and palliation. Observation is reasonable for indolent, asymptomatic lesions when systemic therapy is controlling disease, when treatment would not alter management, or when the risks of liver injury and treatment burden outweigh local benefit. Conversely, SBRT may be reasonable for painful or threatening focal disease when durable local relief is expected and the patient’s hepatic reserve permits treatment; palliative radiation has a different objective from ablative SBRT and should not be labeled curative. For every metastatic indication, the practical question is whether local control will preserve an effective systemic regimen, prevent a specific complication, enable a subsequent curative-intent strategy, or improve symptoms. If none of these outcomes is plausible, observation or systemic therapy is usually more appropriate than liver SBRT. ^[33][34]

Selection begins with a multidisciplinary judgment that focal tumor treatment is likely to improve the patient’s course without exhausting hepatic reserve. Bring the case to a tumor board with hepatology, radiology, surgery, interventional radiology, medical oncology, and radiation oncology; treatment decisions require both oncologic assessment and technical feasibility of local therapy ^[47]. Favor SBRT when the disease is sufficiently limited to permit safe focal treatment, the patient can tolerate simulation and image-guided treatment, and the expected benefit exceeds the competing risk of hepatic failure, extrahepatic progression, or death from comorbidity.

Record performance status , frailty, major comorbidities, cognition, ability to lie supine, and expected life expectancy. A patient with poor performance status or a short life expectancy from disseminated cancer, progressive cirrhosis, or noncancer illness is unlikely to benefit from a burdensome ablative course unless treatment is clearly palliative. In a national quality-project cohort, better performance status was associated with better overall survival, while smaller planning target volumes were associated with better local control and overall survival ^[5]. Assess the tumor burden in practical terms: maximum diameter, number of lesions, distribution across segments, proximity to the hilum and central bile ducts, contact with stomach or bowel, subcapsular or dome location, vascular invasion, and the volume of uninvolved liver that can be preserved. Size and location do not create universal exclusion thresholds; they determine whether a plan can adequately cover all disease while respecting the remaining liver and adjacent organs. Caudate, perivascular, hilar, and otherwise inaccessible lesions may favor SBRT over percutaneous ablation, but the final decision depends on anatomy and the competing treatment options ^[53].

Define the treatment intent before simulation. For localized primary liver cancer, intent may be definitive; for transplant candidates, bridging or downstaging; and for metastatic disease, consolidation, oligoprogression treatment, or salvage. Count every intrahepatic and extrahepatic lesion and establish whether extrahepatic disease is absent, limited and controlled, or progressing despite systemic therapy. SBRT is most defensible when systemic disease is absent or when effective systemic treatment is expected to control untreated disease; focal control alone cannot compensate for rapidly progressive disseminated cancer. Review prior resection, ablation, transarterial chemoembolization, transarterial radioembolization, surgery, and systemic therapy, including response and timing. Prior liver-directed therapy can reduce the remaining functional reserve, and prior radiotherapy requires composite dose review; abdominal reirradiation evidence remains predominantly retrospective and heterogeneous, with reported severe toxicity usually 5-15% but occasional lethal radiation-induced liver disease ^[7].

Assess hepatic reserve with both a clinical score and objective laboratory data. Calculate Child-Pugh score and document class A, B, or C, rather than reporting only “cirrhosis.” Record bilirubin, albumin, prothrombin time or INR , and the presence and control of ascites and encephalopathy. Calculate ALBI grade , which uses albumin and bilirubin and reduces the subjectivity inherent in clinical Child-Pugh components; ALBI grade 2 and Child-Pugh class B provided complementary prognostic information in a retrospective SBRT cohort ^[48]. Also obtain platelet count, AST, ALT, alkaline phosphatase, creatinine, and a current complete blood count. Platelet reduction may signal clinically relevant portal hypertension, while AST and ALT identify active hepatocellular injury and creatinine affects contrast selection, procedural safety, and overall treatment fitness.

Child-Pugh B and C require individualized review rather than automatic exclusion. Ask whether decompensation is stable or reversible, identify the cause of ascites, treat infection or gastrointestinal bleeding, and reassess after correction of obstruction, volume disturbance, or alcohol-related injury when appropriate. The available evidence is small and selected: a retrospective series of 38 patients with Child-Pugh B/C function included only one Child-Pugh C patient, so it could support cautious treatment of selected Child-Pugh B patients but could not establish safety for Child-Pugh C disease ^[22]. A prospective pilot study enrolled nine transplant candidates with Child-Pugh B8 or worse treated to one lesion and found that six remained eligible for or underwent transplantation at 1 year, while one developed grade 4 hepatic toxicity ^[12]. These data justify individualized consideration, not routine treatment of decompensated patients.

Obtain high-quality multiphasic liver imaging before committing to SBRT. Use contrast-enhanced multiphasic CT with arterial, portal-venous, and delayed venous phases, or contrast-enhanced MRI when it better characterizes lesions, vascular anatomy, biliary anatomy, or the relationship to adjacent organs. Liver MRI generally offers better soft-tissue characterization and can detect additional lesions; diffusion-weighted imaging improves lesion detection and characterization ^[47]. Acquire chest imaging and disease-appropriate metastatic staging; PET/CT is not a mandatory routine test for HCC, but it may be useful when the suspected tumor type or clinical question warrants it ^[47]. Repeat staging when disease evolution could change intent or eligibility; the EORTC/ESGAR consensus considered repeat liver imaging necessary when it is more than 1 month old and repeat thoracic imaging when it is more than 3 months old ^[47].

Use LI-RADS or the disease-specific diagnostic framework when HCC is suspected in an at-risk liver. Obtain AFP and, when relevant, other disease-specific markers such as PIVKA-II for HCC, CA 19-9 for suspected cholangiocarcinoma, or the marker used for the known primary malignancy. A biopsy is appropriate when imaging is indeterminate, when the patient is not in a validated at-risk population, when a mixed tumor or metastasis is possible, or when histology will change systemic or local treatment. Even when imaging is typical, biopsy is particularly useful if there has been no prior histologic diagnosis and is recommended for additional hypovascular hepatobiliary-phase hypointense lesions larger than 1 cm ^[47].

Complete an etiologic and transplant assessment. Test for active or prior hepatitis B and hepatitis C , quantify alcohol exposure, assess metabolic-associated steatotic liver disease and other causes of chronic liver disease, and arrange antiviral or addiction-management input when indicated. Evaluate transplant candidacy before treating a potentially transplantable HCC: document tumor burden against accepted criteria, vascular invasion, extrahepatic spread, psychosocial and substance-use factors, cardiopulmonary fitness, and center-specific listing requirements. SBRT can be used as bridge or downstaging therapy, but the transplant pathway, not local response alone, determines whether treatment advances the patient toward transplantation; pooled transplant-candidate data are derived largely from nonrandomized studies ^[4].

Look specifically for portal-hypertension and biliary hazards. Examine for ascites, splenomegaly, abdominal-wall collaterals, and varices; review prior endoscopy and obtain endoscopic assessment when clinically indicated. Confirm portal-vein patency or characterize tumor thrombus. Identify biliary obstruction, cholangitis, ductal dilatation, biliary stents, and lesions abutting the hilum or central ducts before planning. Correct clinically significant obstruction or infection first when possible, because these conditions can worsen liver function independently of SBRT and increase the consequences of treatment-related biliary injury.

Finally, confirm that the patient can undergo contrast imaging, immobilization, respiratory motion assessment, and repeated image guidance. Check prior iodinated or gadolinium contrast reactions, renal function, pregnancy potential when relevant, pain, orthopnea, claustrophobia, and the ability to maintain a reproducible supine position. Ascites, encephalopathy, severe cough, inability to follow breath-hold instructions, or painful musculoskeletal disease may require optimization or an alternative motion-management strategy. The technical plan must preserve adequate uninvolved liver, but the decision to proceed belongs only after the clinical, oncologic, hepatic, and logistical assessments converge.

Accurate target delineation begins with a single question: which tissue contains viable tumor, and which tissue represents motion, treatment effect, or normal liver? Answer it by reviewing all available imaging together rather than allowing one acquisition to dictate the contour. The core dataset is a simulation-quality four-dimensional computed tomography (4DCT), supplemented by contrast-enhanced multiphasic CT and diagnostic liver MRI . In selected patients, add prior treatment imaging, angiographic or cone-beam CT, and PET/CT when the clinical question is occult metastatic disease or metabolically active cholangiocarcinoma. The diagnostic study establishes lesion identity; the simulation study establishes geometry, motion, and the relationship to organs at risk.

Simulation CT and motion-inclusive geometry

Acquire the planning CT in the treatment position with a reproducible scan range that includes the entire liver and adjacent organs at risk. A 4DCT should contain a sufficient number of respiratory phase bins to show the full excursion of the lesion, liver dome, stomach, duodenum, and bowel. Contour the tumor on each usable phase rather than assuming that the liver moves as a rigid body. A recent prospective planning workflow used an eight-phase 4DCT to assess tumor and organ motion, while conventional definitions commonly use up to 10 phases ^[59][58].

The gross tumor volume (GTV) is the macroscopic tumor visible on the selected image set, including the enhancing or otherwise convincingly viable component. It is not automatically the entire region of post-treatment hypoattenuation, edema, perfusion alteration, or scar. For a lesion whose position changes over respiration, the internal target volume (ITV) is the union of the GTV positions over the explicitly assessed respiratory cycle. This definition matters: an ITV is a motion-derived volume, not a generic expansion of the GTV ^[58]. If the treatment workflow explicitly models motion without constructing a geometric union, document that method and do not label the resulting dose-robustness margin an ITV.

The clinical target volume (CTV) represents suspected microscopic or infiltrative extension beyond the visible GTV when the disease and institutional protocol justify a margin. Infiltrative HCC, an ill-defined cholangiocarcinoma interface, and tumor extending along a vascular pedicle require a different clinical judgment from a sharply circumscribed metastasis. The planning target volume (PTV) is the CTV or ITV plus the geometric allowance for residual motion, setup error, image-registration uncertainty, and other delivery inaccuracies. It is therefore a treatment-geometry construct, not a statement that every voxel contains tumor. The International Commission on Radiation Units and Measurements definition emphasizes that the PTV margin incorporates the possible geometric variations and inaccuracies affecting dose delivery ^[58].

MRI integration and viable-tumor definition

Use multiparametric liver MRI when CT does not show the tumor boundary confidently, when the lesion is close to the diaphragm or bowel, when there is suspected infiltrative disease, or when prior treatment has distorted enhancement. A practical protocol includes T2-weighted imaging, precontrast and dynamic arterial/portal-venous/delayed T1-weighted imaging, and diffusion-weighted imaging with apparent diffusion coefficient maps. A representative liver-MRI protocol included T2-weighted, Dixon T1-weighted, diffusion-weighted sequences with multiple b-values, and dynamic contrast-enhanced T1-weighted imaging ^[59].

Contour the enhancing viable component by integrating enhancement pattern, diffusion behavior, T2 signal, and the pre-treatment appearance. Do not equate every arterialized focus with residual tumor: arterioportal shunting, transient perfusion changes, hemorrhage, and post-ablation or post-embolization changes can enhance without viable malignancy. Conversely, infiltrative tumor may be poorly marginated and less conspicuous than a discrete mass; in that circumstance, include the abnormal tissue supported by the combined imaging pattern and discuss uncertainty at multidisciplinary review. Gadoxetate disodium can add hepatobiliary-phase information because functioning hepatocytes take up and retain the agent, whereas many tumors remain relatively hypointense; dynamic enhancement remains essential because hepatobiliary hypointensity alone is not a viability test. Gadoxetate-enhanced dynamic MRI has also been used to generate voxelized surrogate maps of liver function for investigational functional-risk modeling ^[64].

After transarterial chemoembolization, lipiodol may obscure or alter CT attenuation, and the treated region may not correspond to the residual viable tumor. Compare pre-treatment imaging, post-treatment multiphasic CT or MRI, and the simulation dataset. A study of liver SBRT after lipiodol retention found no significant difference between free-breathing and averaged CT plans in several evaluated dose metrics, but that dosimetric result does not remove the need to identify residual enhancing tumor anatomically ^[61]. After ablation, contour the viable peripheral enhancement separately from the ablation cavity; after prior radiation, interpret enhancement against the prior field and dose distribution rather than using enhancement alone.

Vascular invasion, difficult locations, and registration

Portal-vein or hepatic-vein tumor thrombus is tumor, not merely bland thrombus or an adjacent organ-at-risk structure. Delineate the enhancing intravascular component on multiphasic CT or MRI, trace its cranial and caudal extent through the involved branch, and distinguish it from nonenhancing bland thrombus. Include contiguous parenchymal tumor and the involved vascular segment when the treatment intent is to control the dominant tumor-thrombus complex. Document portal-flow compromise, cavernous transformation, and the relationship to the hilum because these findings affect both the clinical target and the safety of treating surrounding functional liver.

Subcapsular and dome lesions deserve deliberate review of the pleura, diaphragm, ribs, stomach, colon, and small bowel. The liver capsule is not a reliable surrogate for tumor extent, and a subcapsular lesion may be inseparable from adjacent bowel on one phase or one modality. Register MRI to the simulation CT using rigid alignment followed by carefully checked deformable registration only when the deformation is anatomically credible. Verify registration at the tumor, liver contour, diaphragm, portal and hepatic veins, vessels, and nearby bowel; do not accept a visually plausible overlay if these landmarks disagree. Contrast CT and corresponding anatomic landmarks improved deformable-registration accuracy in a liver reirradiation study, which found mean target-registration errors of 1.6-1.7 mm with structure-guided registration ^[62]. Those values are study-specific, not a universal uncertainty allowance.

Prior treatment imaging and cone-beam CT can show interval deformation and day-of-treatment anatomy, but cone-beam CT has poorer soft-tissue contrast than diagnostic MRI and may not identify a small lesion reliably. In a CT-guided adaptive series, the distance between the PTV and nearest organ at risk changed during treatment, and PTV-organ overlap developed in patients who had separation at simulation ^[63]. Registration and contour review must therefore account for changing bowel filling, stomach distension, liver deformation, and lesion conspicuity rather than treating the simulation anatomy as invariant.

Before finalizing the plan, label three separate uncertainties on the image review: the boundary of viable tumor, the extent of infiltrative or vascular disease, and the geometric uncertainty created by motion and registration. Expand only for the uncertainty that actually exists. A large undifferentiated volume can protect against ambiguity at the cost of irradiating functional liver and adjacent bowel; a narrowly drawn volume can spare normal tissue but miss infiltrative or moving disease. The safest contour is the one whose assumptions are explicit, multimodality-supported, and reproducible on the treatment image set.

Respiratory motion management is a geometric intervention, not a cosmetic planning preference: diaphragmatic excursion can enlarge the treated volume, blur cone-beam CT (CBCT), and bring bowel or stomach into a high-dose region. The German Society of Radiation Oncology (DEGRO) therefore frames liver SBRT planning around explicit requirements for imaging, planning, treatment delivery, and motion management, although no universal minimum technical standard has been established. ^[69][70]

Motion assessment before choosing a technique

Begin with the patient supine in the intended treatment position, using the final indexed immobilization system and arm position. A flat tabletop, vacuum cushion or indexed body cradle, and reproducible upper-abdominal support reduce translational and rotational setup variation; the device must not compromise diaphragmatic excursion, venous return, or the patient’s ability to breathe or perform a breath-hold. Rehearse the complete workflow before simulation: coaching, contrast injection, respiratory monitoring, image acquisition, CBCT matching, and beam-on/beam-off behavior. The selected technique is only valid if the patient can reproduce it when fatigued and on every treatment fraction.

For free-breathing simulation, acquire respiratory-correlated four-dimensional CT (4DCT) covering the liver, adjacent bowel, stomach, kidneys, and the lower thorax. Record the respiratory trace and inspect every reconstructed phase rather than accepting the automatic phase sorting without review. Measure tumor or surrogate displacement in superior-inferior, anterior-posterior, and left-right directions; document the maximum excursion, the end-inspiratory and end-expiratory positions, and whether the trajectory differs during inspiration and expiration. That difference is respiratory hysteresis: the tumor may occupy different positions at the same nominal respiratory amplitude depending on the direction of travel. Hysteresis, phase lag between the external signal and internal liver motion, and deformation of the liver relative to the diaphragm can make a single breathing trace an imperfect surrogate for tumor position.

Assess the breathing trace for irregular cycle length, variable amplitude, cough, pauses, double peaks, and baseline drift. Irregular breathing can produce misassigned or duplicated anatomy in 4DCT; severe artifacts in liver or lung metastasis 4DCT were associated with worse local control in a retrospective cohort, although the association does not prove that the artifacts caused recurrence. ^[78] Registration errors also increase with respiratory irregularity: in one study, the correlation coefficient between irregularity and registration error was 0.65 for diaphragm-guided and 0.54 for fiducial-guided registration. ^[76] If the trace is unstable, repeat coaching and acquisition, use a more robust motion strategy, or abandon an ITV derived from a technically unreliable 4DCT.

Compare the internal anatomy, not merely the external respiratory waveform, between simulation and treatment. Record the patient’s baseline end-expiratory or end-inspiratory level, breathing amplitude, cycle length, and the relationship of the liver dome and tumor to the vertebral bodies. A change in baseline position can move the tumor outside the expected envelope even when the breathing amplitude appears acceptable. Contrast itself may alter the measured liver position: in expiratory breath-hold, the mean cranio-caudal interfraction variation was 2.5 ± 2.6 mm on noncontrast CT but 6.4 ± 6.4 mm when contrast-enhanced CT was compared with CBCT. ^[77] Use the same coached respiratory state and, where feasible, the same contrast phase for simulation and image registration; if they differ, verify the internal liver and target relationship before accepting the match.

Selecting the motion strategy

Free-breathing 4DCT remains the most broadly deployable approach. It is appropriate when the patient cannot reliably gate or hold the breath, when treatment time must remain efficient, or when the residual motion is measurable and an ITV-based plan remains dosimetrically safe. Use the phase union only when the respiratory trace and reconstructed images are credible. The meta-analysis of liver SBRT found weighted mean ITV margins of 13.5 ± 4.9 mm superior-inferior, 7.3 ± 7.9 mm left-right, and 6.3 ± 7.6 mm anterior-posterior for free breathing; these are population estimates, not a substitute for patient-specific motion measurement. ^[71]

Abdominal compression is useful when it reduces motion without producing irregular respiration. Reassess the trace after compression because the device can change respiratory mechanics rather than simply reduce amplitude. The same feasibility study found that 73 of 78 patients tolerated compression, but residual liver-dome motion still varied widely; therefore, confirm the postcompression amplitude on imaging and do not use a fixed “compression margin.” ^[67]

Gating is reasonable when a stable respiratory phase places the tumor away from bowel or stomach and the patient’s duty cycle is acceptable. Define the window from internal anatomy whenever possible, and verify that the tumor or liver dome occupies the expected position during the entire acquisition. External abdominal motion is a surrogate, not the target itself. If baseline drift moves the liver outside the window, pause treatment, reacquire the baseline, and repeat the internal match; do not widen the window merely to preserve beam-on time.

Breath-hold is preferred when the patient can reproduce a stable position and the reduction in motion meaningfully improves the target-organ-at-risk geometry. The simulation must include repeated breath-hold CT acquisitions, because one successful breath-hold does not establish reproducibility. A practical protocol may acquire arterial and venous contrast-enhanced breath-hold CT plus two or three scans during the beginning, middle, and end of a breath-hold to estimate both inter-breath-hold and intra-breath-hold variation. ^[74] In a retrospective comparison, breath-hold reduced planning volumes and normal-liver dose relative to free breathing, but the study was small and nonrandomized; its clinical outcomes cannot establish superiority. ^[72] Surface guidance can monitor the chest or abdominal surrogate, but it does not eliminate the need to confirm internal anatomy with CBCT or another validated image-guidance method.

Simulation, contrast, immobilization, and image guidance

Use indexed positioning so that the simulation reference can be recreated on the treatment couch. Record the table coordinates, arm position, head support, knee support, compression setting, and any respiratory-monitoring hardware. Perform the breathing or breath-hold training in the CT room before acquiring the planning scan. Patients selected for an intensive breath-hold workflow should understand the instructions and demonstrate stable breath-holds; one clinical protocol required at least 20 seconds and used a 2-mm gating window, whereas other repeated-breath-hold workflows used holds of 20 seconds or less. ^[74][72] The exact threshold is platform-dependent; the operational requirement is that each beam segment and each verification image can be completed without loss of the prescribed internal position.

Administer intravenous contrast when renal function, allergy history, and the clinical question permit. A published liver SBRT breath-hold protocol used 150 mL of iohexol 300 mg iodine/mL at 3 mL/s followed by 30 mL saline, with arterial and venous phase acquisitions. ^[74] Use multiphasic imaging to delineate viable tumor and vessels, but do not assume that the contrast-enhanced simulation anatomy will match a noncontrast treatment CBCT. Register on stable liver and vertebral landmarks, then inspect the target, diaphragm, portal/hepatic veins, bowel, and stomach manually. If enhancement washout makes the lesion less conspicuous on later breath-hold scans, register subsequent scans to the first contrast-enhanced acquisition rather than interpreting a smaller apparent tumor as biologic change. ^[74]

CBCT is often limited by respiratory blur, low soft-tissue contrast, metallic artifact, and poor visualization of small or hypovascular lesions. When the lesion is poorly visualized, use a validated surrogate such as the liver contour, diaphragm, vascular landmarks, or a fiducial marker only after confirming its relationship to the tumor. Diaphragm-guided registration can have millimeter-scale errors and becomes less reliable as breathing irregularity increases. ^[76] Acquire the CBCT in the same respiratory state used for simulation whenever possible; breath-hold CBCT can remove motion blur and improve image quality, whereas free-breathing CBCT may obscure the target. ^[74]

Place fiducial markers selectively, not reflexively. They are most useful when the tumor is not reliably visible on CBCT and when tracking or marker-based registration will change treatment feasibility. Discuss percutaneous bleeding, infection, marker migration, and the possibility that a marker will not remain geometrically coupled to a deforming tumor. In a recent series, technical placement success was 96%, clinical completion of SBRT without displacement was 88%, and migration into the right atrium prevented treatment in two cases. ^[82] Avoid placing markers near major hepatic veins or other locations where migration risk is increased. ^[82]

Target construction and uncertainty accounting

Choose the target strategy from the measured motion, not from the availability of a device. For free breathing, use a phase-based ITV only after reviewing all usable phases and correcting or excluding artifact-laden reconstructions. For gating, define the target from anatomy within the validated gate and add the measured residual motion at the gate edges. For breath-hold, combine repeated breath-hold target positions and account for breath-hold-to-breath-hold variation. For tracking, model the tracking error, marker-to-tumor relationship, phase lag, and beam-delivery latency; a moving beam does not remove uncertainty, it changes its source.

Add residual setup error, image-registration error, intrafraction drift, and delivery uncertainty after the motion strategy has been defined. Do not double-count motion by first creating a full free-breathing ITV and then adding a second respiratory margin for the same displacement. Conversely, do not use a small breath-hold or gated target without including the measured variation between repeated states. In tracking plans, calculate dose on the moving anatomy when feasible: a four-dimensional planning study showed that ignoring beam motion can allow an organ at risk to exceed its dose limit, whereas aperture sorting improved organ-at-risk sparing while preserving target coverage in a small planning analysis. ^[79]

Verify the treatment breathing pattern at every fraction. Match the patient’s indexed setup, inspect the respiratory trace or breath-hold level, acquire CBCT or orthogonal kV images as appropriate, and compare the internal anatomy with the simulation reference. If the baseline, diaphragm position, tumor-marker relationship, or bowel position differs materially, stop and correct the setup, repeat imaging, or replan. A clinically implemented breath-hold workflow required confirmation that expiratory-phase CBCT agreed with the planning CT before irradiation. ^[72] This verification is the practical safeguard against a technically perfect plan being delivered to the wrong respiratory state.

Pearl: Measure motion in the patient’s treatment position, select the simplest strategy that produces a reproducible internal state, and treat any mismatch between simulation and delivery as a geometric error requiring correction, not as a margin problem to be ignored.

Prescription is not universal. Choose dose and fractionation from the disease being treated, tumor size and geometry, uninvolved liver volume and function, proximity of serial organs, prior radiation, and the treatment protocol. Common schedules include 40-50 Gy in 5 fractions and 45-54 Gy in 3 fractions; published practice spans a wider range, including 36-60 Gy in 3 fractions, 45 or 50 Gy in 5 fractions, and other hypofractionated regimens. Tumors abutting stomach, duodenum, or bowel often require a risk-adapted increase to 5-8 fractions with a lower per-fraction dose rather than forcing a three- or five-fraction plan to meet an ablative prescription. ^[85] A multicenter prospective study of selected hepatocellular carcinoma (HCC) treated 60 Gy in 3 fractions reported 5-year local control of 90.7%, with no grade ≥3 gastrointestinal toxicity and one case of nonclassic radiation-induced liver disease; those results support the regimen in appropriate patients, not its use as a universal standard. ^[87]

Report the prescription in three complementary ways: total dose and fractions, the dose-volume prescription point or isodose surface, and tumor BED10 (biologically effective dose calculated with an α/β ratio of 10 Gy). BED10 permits comparison across fractionation schedules, but it does not erase differences in irradiated volume, dose heterogeneity, liver reserve, or organ-at-risk exposure. Higher BED10 was associated with better local control in a national quality-project analysis, but the study was observational and cannot define a single ablative threshold. ^[5]

The prescription isodose must be stated because “50 Gy in 5 fractions” does not describe the spatial dose distribution. A plan prescribed to the 80% isodose gives a central maximum near 125% of prescription, whereas a plan prescribed nearer the 100% isodose is more homogeneous; both can deliver the same nominal prescription but expose normal tissue differently. Prescribe to a clinically justified isodose surface, then report the target minimum, near-minimum, median, and maximum doses. Do not substitute a central hot spot for peripheral coverage: the dose at the edge of the planning target volume (PTV) governs whether the entire target receives an ablative dose, whereas a central hot spot may reflect intentional inhomogeneity within tumor. In a retrospective five-fraction HCC series, the minimum dose covering 95% of the PTV (PTV D95%) predicted local control, and no local recurrences occurred when PTV D95% was at least 40 Gy; the finding requires cautious interpretation because the study was single-center, retrospective, and included dose reductions near organs at risk. ^[86] Cold spots in the PTV can matter even when the nominal prescription and BED10 appear adequate; a retrospective colorectal liver-metastasis analysis found PTV minimum dose, rather than prescription dose alone, to be an independent predictor of freedom from local progression. ^[88]

For normal liver, evaluate both the mean liver dose (MLD) and spared-liver volumes. Contour total liver minus gross tumor volume, and when appropriate report liver minus the PTV or minus all treated targets; the chosen definition must be explicit because each produces a different dose-volume histogram. MLD is a useful global measure of integrated dose, while the volume receiving less than a specified dose indicates how much functional liver remains below clinically relevant exposure. Commonly used planning principles include maintaining at least 700 cm³ of uninvolved liver and, in some protocols, keeping the uninvolved-liver volume receiving less than 15 Gy above 700-1000 cm³; these are practice thresholds, not universally validated tolerances. ^[85][52] MLD and spared-volume metrics must be interpreted alongside Child-Pugh or albumin-bilirubin grade, portal hypertension, ascites, thrombocytopenia, multifocality, prior liver-directed treatment, and the location of spared functional parenchyma. ALBI deterioration after SBRT has been associated with lesion size, PTV size, MLD, and multifocality. ^[26]

Use the following constraints as planning domains rather than immutable limits. Numeric values vary by protocol, fractionation, organ contouring, dose-calculation method, and prior treatment. Reirradiation requires summation of prior and current dose to the liver and each nearby organ, with attention to dose distribution rather than simply adding prescription doses. The abdominal reirradiation literature is predominantly retrospective, reports wide variation in dose, fractionation, and constraints, and found severe toxicity in approximately 5-15% of patients; one study reported fatal radiation-induced liver disease in 25% of liver-reirradiation patients. ^[7]

The liver is a parallel organ, so a small high-dose region can be tolerated when sufficient functioning liver remains outside the treated volume; the stomach, duodenum, bowel, and central bile ducts behave more like serial critical structures, for which a small high-dose focus may dominate risk. This distinction explains why an apparently acceptable MLD does not rescue a plan with an excessive duodenal or bile-duct hot spot, and why a plan that meets gastrointestinal limits may still be unsafe in a patient with marginal hepatic reserve. ^[85][27]

When normal-tissue limits compete with target coverage, prioritize the clinical objective explicitly. First preserve the uninvolved functional liver and avoid catastrophic serial-organ injury; then maintain adequate peripheral target dose, accepting a central hot spot when the protocol permits it. If the plan cannot satisfy both requirements, reduce dose per fraction, expand the number of fractions to five through eight, alter the treated volume only when oncologically justified, or choose another local therapy. Do not label a plan “ablative” from BED10 alone: individualized modeling in HCC 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 hepatic reserve changes the efficacy-toxicity trade-off. ^[50]

Pearl: Report the prescription isodose, PTV near-minimum dose, central maximum dose, BED10, MLD, spared uninvolved-liver volumes, and every relevant serial-organ metric; a nominal prescription without those data is not a reproducible liver SBRT prescription.

CT-based inverse planning should begin with a simulation dataset that represents the patient’s treatment position, respiratory state, immobilization, and intended motion-management strategy. Use the delineated gross tumor volume and internal target volume only after reviewing all usable respiratory phases; do not compensate for an unmeasured or poorly characterized motion pattern by applying a generic margin. The planning target volume must incorporate the residual motion, setup, registration, and delivery uncertainties that remain after motion management, without counting the same uncertainty twice. DEGRO recommends that liver SBRT planning explicitly integrate imaging, target definition, motion assessment, dose calculation, delivery, and quality assurance rather than treating them as separate technical steps. ^[70]

Photon planning

For volumetric-modulated arc therapy (VMAT), use CT-based inverse planning to optimize multileaf-collimator apertures, dose rate, and gantry speed across one or more partial arcs. Select arc ranges according to the target’s beam-eye-view relationship with uninvolved liver, stomach, duodenum, bowel, kidneys, chest wall, and spinal cord. Exclude or down-weight sectors that traverse excessive normal liver or create an unfavorable entrance path through serial organs; an automated gantry-angle study found that minimizing the beam path through normal liver reproduced expert VMAT choices without compromising reported dosimetric parameters. ^[94] For multiple lesions, separate arc groups or spatially sequenced apertures can reduce inter-target leakage and low-dose exposure to uninvolved liver, although a small retrospective planning study found that the low-dose-sparing strategy had slightly inferior conformity and gradient index than conventional single-isocenter VMAT. ^[91]

Use intensity-modulated radiation therapy (IMRT) when fixed-field geometry gives better avoidance of a critical organ, when couch or gantry clearance restricts arc delivery, or when a small number of carefully selected beam directions provides a more robust plan than continuous arcs. Three-dimensional conformal radiotherapy (3D-CRT) remains reasonable for a small, peripherally located target when several noncoplanar or coplanar beams provide adequate coverage and rapid dose fall-off without excessive liver dose; its principal advantage is simple, transparent delivery, whereas IMRT and VMAT generally provide more degrees of freedom for conformal shaping around bowel, stomach, and central bile ducts. The choice of platform should therefore follow the achievable dose distribution and delivery reliability, not the nominal sophistication of the machine. ^[70]

Optimize to cover the target while preserving functional liver and protecting serial organs. Review at least the prescription isodose, PTV near-minimum and median dose, GTV median and near-maximum dose, conformity index, gradient index, and dose to every relevant organ at risk; the DEGRO/DGMP consensus identifies these as minimum elements of stereotactic dose documentation. ^[93] A highly conformal plan is not automatically a safe plan: a steep gradient may improve bowel or stomach sparing but become vulnerable to small geometric errors, whereas a broader low-dose bath may preserve the high-dose geometry yet irradiate more uninvolved liver. When target coverage and an organ-at-risk constraint conflict, protect functional liver and prevent catastrophic injury to stomach, duodenum, bowel, or central bile ducts before pursuing an idealized ablative dose. ^[93]

Use heterogeneity-corrected dose calculation appropriate for small fields and abdominal tissue interfaces. Factor-based type-A algorithms are not adequate for abdominal SBRT; at minimum, use a type-B algorithm, and use a type-C or Monte Carlo-class algorithm when tissue heterogeneity, steep gradients, very small targets, or complex beam paths could alter dose-volume metrics. A calculation grid of 2 mm or less is recommended, with 1-1.5 mm considered for targets smaller than 2 cm³. ^[93] Recalculate or independently verify plans when metal, contrast, embolic material, or other artifacts may distort electron-density assignment; segment residual artifacts and assign the most defensible tissue density rather than allowing the treatment-planning system to extrapolate silently. ^[93]

Assess low-dose exposure as deliberately as high-dose conformity. Report mean dose to uninvolved liver and relevant low-dose volumes, defining clearly whether “uninvolved liver” excludes the gross tumor, the PTV, or all treated targets. Common planning objectives include a mean uninvolved-liver dose of approximately 10-15 Gy and preservation of more than 700 cm³ of uninvolved liver, but these are planning principles rather than universal tolerances and must be individualized to hepatic reserve and prior treatment. ^[85] Functional-avoidance planning using quantitative gadoxetic-acid MRI has reduced mean dose to a high-function liver compartment in a retrospective planning study, supporting selective use when functional liver maps are available; it is not yet a substitute for conventional liver-volume and dose assessment. ^[13]

Proton planning

Consider proton therapy when a clinically meaningful reduction in integral dose or distal dose to uninvolved liver, stomach, bowel, kidneys, or other organs is needed and the proton plan remains robust to range, setup, and motion uncertainty. The Bragg peak can place most of the dose near the target with little exit dose, but the benefit is anatomy-dependent: dosimetric comparisons found the largest organ sparing for peripheral tumors, whereas photon stereotactic plans were comparable for small target volumes. ^{[85] [97]} Proton therapy should therefore be selected because it solves a specific dosimetric problem, often limited functional liver, a large target, reirradiation, or an adjacent serial organ, not because protons are intrinsically superior for every hepatic SBRT case. Comparative clinical evidence remains incomplete; the NRG GI003 phase III trial is assessing proton versus photon therapy for HCC. ^[85]

For pencil-beam scanning, optimize over multiple fields or arcs with explicit robustness evaluation for patient setup, respiratory motion, and CT Hounsfield-unit-to-stopping-power conversion. Place the distal edge within the target only when the range uncertainty and anatomy are sufficiently stable; avoid placing a distal edge immediately against the stomach, duodenum, or bowel when a small range error could move the high-dose fall-off into that organ. If this cannot be achieved, change beam angles, use a different field arrangement, increase the margin only when justified, or select photons if their plan is more reliable. The NRG proton survey found that most centers assessed setup perturbations and stopping-power uncertainty, commonly using ±5 mm setup scenarios and ±3.5% conversion uncertainty, but these are reported practice patterns rather than universal standards. ^[85]

Motion is particularly consequential for pencil-beam scanning because sequential spot delivery can interact with respiratory motion. This interplay can produce dose heterogeneity or target underdosage even when the static plan appears satisfactory. Prefer reproducible breath hold or gating when feasible; otherwise use robust optimization, rescanning or repainting, and motion scenarios that reflect the patient’s measured breathing pattern. ^[85] Proton robustness should be judged on target coverage and organ-at-risk dose across perturbed scenarios, not on the nominal dose distribution alone. A photon-like, centrally concentrated proton dose distribution has been proposed for ablative planning, but the supporting study was a 10-patient dosimetric and deliverability analysis, so it should be regarded as investigational rather than a default prescription strategy. ^[96]

Motion management and delivery geometry

Choose abdominal compression , respiratory gating, or breath hold only after testing tolerance, reproducibility, residual motion, and the resulting dose distribution in the final treatment position. Compression does not guarantee a fixed residual motion: in one feasibility cohort, 73 of 78 patients tolerated compression, but liver-dome motion still ranged from 0.5 to 18.3 mm, with a median of 5.3 mm. ^[67] Breath hold can reduce the ITV and improve bowel or stomach sparing when the patient can reproduce the same position for approximately 15-20 seconds, but repeated rehearsal and verification are required; a small planning study found better target coverage and lower organ-at-risk doses with breath hold than with compression or free breathing, without establishing clinical superiority. ^[68]

Gating requires a validated respiratory signal, a defined internal gating window, and a plan that remains deliverable during the portion of the cycle actually treated. Free-breathing ITV planning remains appropriate when motion is measurable, the reconstructed phases are interpretable, and the resulting liver dose is acceptable. Reject or reacquire a simulation study when irregular breathing, phase sorting, baseline drift, or hysteresis creates substantial artifacts; severe 4DCT artifacts were associated with worse local control in a retrospective liver and lung metastasis cohort, although the study did not prove causation. ^[78]

Before treatment, confirm indexed immobilization, arm position, respiratory equipment, treatment couch, and patient clearance for every intended gantry and couch angle. Perform collision checks with the couch, immobilization devices, abdominal-compression apparatus, imaging arms, gantry, and patient contour; a plan that meets dose objectives but cannot be delivered without collision or unplanned repositioning is not an acceptable plan. For noncoplanar beams or arcs, verify that the treatment couch does not obstruct the beam and that the selected geometry remains compatible with the available imaging and respiratory-monitoring system. These checks are part of the physical and technical requirements of safe liver SBRT delivery. ^[70]

For dynamic tracking or other free-breathing delivery, quantify tracking error and confirm that the tracking model remains valid throughout beam delivery. Early clinical implementation of gimbal-mounted tracking-VMAT reported a mean three-dimensional tracking error of 2.4 ± 1.8 mm, but this is platform-specific evidence and should not be adopted as a universal tolerance. ^[99] For proton scanning, evaluate motion interplay with the actual spot-scanning sequence and use rescanning or repainting when the dose calculation demonstrates clinically relevant degradation. ^[85]

Every plan requires independent dose calculation or an equivalent patient-specific verification before treatment. Use a secondary calculation engine, measurement-based patient-specific quality assurance, delivery-log analysis, or a validated combination appropriate to the platform; scrutinize small targets, steep gradients, heterogeneity interfaces, and proton range-sensitive plans particularly carefully. The NRG proton survey describes independent calculation and patient-specific measurement as complementary safeguards rather than interchangeable administrative steps. ^[85] A radiation oncologist, medical physicist, and dosimetrist should review the final target coverage, conformity, gradient, liver dose, serial-organ constraints, motion strategy, robustness analysis, and deliverability. Require multidisciplinary review for high-risk anatomy, including tumors abutting stomach, bowel, central bile ducts, major vessels, or previously irradiated tissue. ^{[70] [93]}

Pearl: The best hepatic SBRT plan is the one that maintains clinically adequate target dose across realistic motion, setup, range, and delivery uncertainties while preserving functional liver and avoiding catastrophic serial-organ injury, not the plan with the most complex delivery technology.

Daily image guidance is part of the treatment, not a final check appended to it. High-dose hypofractionation leaves little tolerance for an unrecognized change in tumor position, respiratory phase, or stomach and bowel geometry. Interfraction motion from gastric filling and intestinal gas can move gastrointestinal organs toward the high-dose region and produce unanticipated injury. ^[104] Use the same immobilization, arm position, respiratory instruction, contrast strategy when applicable, and indexed reference marks at every fraction; otherwise, the registration may be precise but precise to the wrong anatomy. ^[104]

Daily image acquisition and registration

Acquire a daily volumetric image before each fraction. Standard cone-beam CT (CBCT) is suitable when the tumor, liver contour, diaphragm, vessels, or other reproducible surrogate is visible. Use a breath-hold or gated CBCT when respiratory motion would blur the target; free-breathing CBCT is more vulnerable to motion and gas-related artifacts. ^[104] Kilovoltage (kV) CBCT generally provides the principal volumetric dataset, whereas megavoltage (MV) orthogonal images are useful for confirming bony anatomy, couch shifts, and the absence of an unintended patient displacement after the CBCT registration. A diagnostic-quality CT-on-rails system can improve soft-tissue and organ-at-risk visualization when conventional CBCT is inadequate, although its workflow is resource-intensive. ^[104]

Use four-dimensional cone-beam CT (4D-CBCT) when the target or diaphragm is poorly represented by a single respiratory phase, when breathing is irregular, or when the adequacy of an ITV derived from simulation is uncertain. Phase-resolved CBCT can reveal baseline drift and the relationship between tumor, diaphragm, and bowel during the treatment session; it does not replace intrafraction monitoring because it samples the anatomy before beam delivery. CBCT has been shown to underestimate the ITV in some abdominal workflows, and a simulation-time 4DCT cannot capture every change in complex intrafraction motion. ^[105]

Register in a deliberate sequence rather than allowing an automated fusion to determine the clinical priority. First verify the external setup and bony anatomy, then assess the liver contour and diaphragm, and finally align the tumor or its most reliable internal surrogate. The tumor receives priority for target coverage, but not at the price of an avoidable high dose to a serial organ. If perfect tumor matching would place the stomach, duodenum, colon, or small bowel inside a high-dose region, shift within the institution’s approved action level to protect the bowel and recalculate or use a verification plan when the shift is not dosimetrically demonstrable. A CT-on-rails workflow illustrates this principle: after vertebral and tumor alignment, the final shift was modified when transferred isodose lines overlapped a gastrointestinal organ; larger overlaps prompted treatment interruption and possible replanning. ^[104]

The registration hierarchy should reflect the anatomy that determines both target coverage and catastrophic toxicity:

• Tumor: Match visible viable tumor whenever the daily image provides sufficient contrast. Do not accept a registration based only on the liver edge when the tumor has deformably moved relative to it. If the tumor is not visible, match a validated surrogate, such as a nearby vessel, fissure, cyst, lipiodol deposit, or diaphragm, and document why that surrogate is expected to remain coupled to the lesion. Liver-contour matching can approximate fiducial-based positioning better than vertebral-bone matching in selected patients, but it remains vulnerable to liver deformation. ^[108]
• Liver and diaphragm: Use the liver contour and diaphragm to detect gross deformation, respiratory-phase mismatch, and baseline drift. The diaphragm is a useful motion surrogate in selected markerless workflows, but the tumor-to-diaphragm relationship must have been established at simulation and checked on the daily image; a moving diaphragm is not automatically an equivalent tumor position. ^[110]
• Biliary tree: When the target is hilar or near the central bile ducts, prioritize registration that preserves the spatial relationship between the tumor, ductal confluence, and adjacent vessels. A registration that improves tumor overlap but brings the biliary tree into an unplanned hot region is not acceptable; obtain better imaging, alter the isocenter within the verified tolerance, adapt, or stop treatment.
• Stomach and bowel: Treat their daily position and filling as a safety constraint, not as passive anatomy. Evaluate gas, distention, and the distance from the PTV to the nearest luminal organ on every CBCT. If the organ approaches or overlaps a transferred high-dose isodose contour, do not proceed on the assumption that the simulation anatomy persists. Daily CT-guided adaptive treatment demonstrated that PTV-to-OAR distance changed during treatment and that new PTV-OAR overlap could arise despite separation at simulation. ^[63]

When registrations disagree, do not average them. Identify which structure moved, whether the discrepancy is translational or deformational, and whether the respiratory state is comparable. Repeat the scan after correcting the breathing instruction or bladder, stomach, and bowel preparation when the mismatch may be transient. If tumor and diaphragm agree but bone does not, investigate patient rotation, couch coordinates, and deformation before accepting the shift. If bone and diaphragm agree but the tumor does not, suspect altered tumor-liver geometry or an incorrect respiratory phase; obtain 4D imaging, diagnostic-quality CT, or marker imaging rather than forcing the tumor onto the planned position. Record the competing registrations, the selected match, the shift, the reason for selection, and the physician’s decision.

Verification methods and failure actions

Fiducials are appropriate when the lesion is not reliably visible on noncontrast daily imaging and no validated surrogate is available. Implant markers near the lesion, generally under image guidance, and verify their position before treatment. Marker placement is not risk-free: a recent series achieved technical facilitation of SBRT in 96% of tumors and treatment completion without displacement in 88%, while marker migration into the right atrium prevented treatment in two cases. ^[82] Use the fewest markers that provide robust localization, avoid major hepatic veins when possible, and reassess the marker-tumor relationship after implantation and on every treatment image. Residual lipiodol from prior transarterial therapy can serve as a radiopaque surrogate in selected patients, avoiding another invasive procedure. ^[105]

Respiratory and intrafraction verification

Verify the respiratory condition used for planning before beam-on. For breath hold , confirm the patient reaches the same monitored level and can sustain it for the imaging and delivery interval; surface motion alone does not prove that the liver or tumor has reproduced its planned position. For respiratory gating , confirm the amplitude window, phase, breathing regularity, and absence of baseline drift with fluoroscopy, kV imaging, an external surrogate, or an internal marker. In a clinical liver workflow, fluoroscopy was used before CBCT to verify that fiducials remained within the expected respiratory window because 4DCT may underestimate the motion actually occurring during treatment. ^[105]

Continue monitoring during beam delivery when the margin is small, the target moves substantially, the treatment is prolonged, or the lesion lies beside a serial organ. With fiducial-based kV monitoring, acquire images at predefined intervals or at each gated beam-on cycle and define an intervention threshold before treatment. One liver SBRT workflow used a 3-mm fiducial displacement threshold and paused treatment when three consecutive images showed a marker outside that tolerance. ^[105] A pause corrects the respiratory window or baseline drift; a shift relocalizes the patient or target. Resume only after the internal anatomy has returned to the accepted position and the delivered segment has been documented.

Interrupt treatment for persistent marker displacement, an unverifiable respiratory state, loss of target or OAR visibility, a new PTV-OAR overlap, an unexpected couch or registration discrepancy, or a change large enough that the planned dose distribution cannot be trusted. A single transient image artifact may justify another image; repeated discordant images require a clinical decision, not silent acceptance. If four CT-on-rails images still demonstrated inconsistent positioning in one published workflow, treatment was aborted for the day and replanning was considered. ^[104]

Record the delivered dose rather than documenting only that the fraction was completed. The treatment record should contain the accepted image set, respiratory condition, registration method and shifts, any intrafraction pauses or corrections, beam-on interruption, delivered monitor units, and the fraction dose actually delivered. If treatment was shifted away from perfect tumor alignment to protect bowel, document the resulting target coverage and OAR doses using a recalculated or verification plan. Preserve cumulative liver, stomach, bowel, duodenal, and biliary-tree dose for subsequent fractions, particularly during reirradiation.

Adaptive liver SBRT

Adapt when anatomy changes invalidate the assumptions that made the scheduled plan safe. Triggers include meaningful tumor shrinkage or progression, liver deformation, new or increased ascites, altered diaphragm excursion, weight loss that changes immobilization, gastric distention, bowel gas, persistent stomach or bowel displacement, marker migration, and a reproducible change in respiratory motion. Do not adapt merely because a contour looks different; adapt when the change alters target coverage, the dose to functional liver, or the dose to a serial organ.

Online adaptation uses the daily image during the treatment visit. Recontour or propagate the target and OARs, review the deformation, recalculate dose on the anatomy of the day, and select a plan that meets the approved target and normal-tissue objectives before beam delivery. This approach is most useful when a stomach or bowel loop has moved into the high-dose region or when liver deformation changes the PTV-OAR relationship. In a retrospective CT-guided series, 77% of 65 fractions were adapted; adapted plans achieved PTV V100% of at least 95% in 92% of fractions versus 64% with scheduled plans, and reduced maximum dose to the nearest OAR when PTV-OAR overlap was present. ^[63] Markerless CBCT online adaptation is feasible, but one feasibility analysis found a mean propagation error of 4.3 ± 2.1 mm and required a larger PTV; adaptation nevertheless reduced priority dose-constraint violations from 4.2% to 0.9% and improved target coverage. ^[83]

Offline adaptation uses the accumulated evidence from one or more fractions to generate a revised plan for subsequent treatment. Use it when the anatomy has changed consistently, for example, sustained weight loss, persistent ascites, repeated bowel displacement, or a stable change in tumor position, but the patient can be safely managed while the revised plan is prepared. Recalculate dose on the new anatomy, assess cumulative liver and serial-organ exposure, obtain physician and physics approval, and define the exact fraction at which the new plan begins. An offline plan is unsafe if the anatomy continues to change unpredictably; in that circumstance, online verification or treatment interruption is preferable. Offline adaptation can delay treatment and may become obsolete as the anatomy changes again. ^[104]

For either strategy, deformable registration is an aid to dose accumulation, not a substitute for physician review. Check propagated contours against the daily anatomy, inspect the tumor-liver interface and every nearby luminal organ, and review the accumulated dose in the functional liver and previously irradiated structures. Reirradiation magnifies the value of this discipline: a DIR-based adaptive re-SABR workflow predicted higher cumulative mean liver dose and D700cc and smaller uninvolved-liver volumes than a non-DIR workflow, demonstrating how registration methodology can change the apparent safety of retreatment. ^[112] When target coverage and OAR protection cannot both be maintained, prioritize avoidance of catastrophic bowel, stomach, or biliary injury and preservation of functioning liver; reduce dose per fraction, increase the number of fractions, defer treatment, or select another local therapy rather than deliver an unverified plan.

Outcomes after liver SBRT must be read on two separate axes. Local control asks whether the irradiated target remains controlled; it does not establish durable cure, because SBRT does not sterilize occult intrahepatic or extrahepatic disease. In a nationwide multicentre registry, local control and overall survival differed substantially: among all treated lesions, local control was 69.9% at 1 year, 52.2% at 2 years, and 32.4% at 5 years, whereas patient-level overall survival was 74.3%, 49.7%, and 19.4%, respectively. ^[5] The apparent divergence reflects both competing mortality and progression elsewhere in the liver or outside it. Interpret a favorable treated-lesion response alongside new intrahepatic lesions, extrahepatic progression, systemic-therapy durability, liver function, and patient-reported quality of life rather than labeling radiographic ablation a cure.

Hepatocellular carcinoma

For selected hepatocellular carcinoma (HCC), SBRT can provide durable control of a focal lesion, particularly when resection, transplantation, or thermal ablation is unsuitable. In the Belgian registry, HCC lesions had 1-, 2-, and 5-year local-control rates of 82.8%, 76.2%, and 43.6%, respectively, and corresponding overall-survival probabilities of 75.4%, 49.1%, and 23.8%. ^[5] These figures are registry estimates, not the expected outcome for every HCC patient: the cohort combined curative, salvage, and advanced-disease indications, and local control was assessed retrospectively.

Tumor burden and hepatic reserve are the dominant clinical modifiers. Larger tumors require a larger high-dose volume, leave less uninvolved liver, and are more vulnerable to incomplete peripheral coverage; in a meta-analysis of HCC measuring at least 5 cm, pooled local control was 81% at 1 year and 69% at 2 years, while overall survival was 60% and 37%, respectively. ^[113] Multifocality worsens progression-free survival because each treated lesion may be controlled while untreated or subsequently emerging lesions continue to appear. In patients with Child-Pugh B disease, baseline Child-Pugh score was associated with overall survival and baseline ALBI grade with local control; the reported 1- and 2-year local-control rates were 96% and 50%, with median overall and progression-free survival of 14 and 9 months. ^[28]

Child-Pugh class and ALBI grade should therefore be interpreted together. Child-Pugh incorporates clinically consequential decompensation, including ascites and encephalopathy, whereas ALBI uses albumin and bilirubin to provide a more objective biochemical measure. In a retrospective HCC cohort treated with SBRT, ALBI grade 2 and Child-Pugh class B independently predicted poorer overall survival; their combined model discriminated outcomes better than either classification alone. ^[48] A falling ALBI score after treatment may represent treatment-related functional loss, cirrhosis progression, tumor progression, or intercurrent illness rather than SBRT toxicity alone. In one comparative cohort, ALBI deterioration after SBRT was associated with lesion size, planning target volume, mean liver dose, and multifocality. ^[26]

Ablative dose helps only when the entire target receives it. In the Belgian registry, BED10 of at least 100 Gy and smaller PTV volume independently correlated with better local control and overall survival, but this association is not proof that BED10 alone determines outcome; higher dose is often selected for smaller, technically favorable tumors in patients with better liver function. ^[5] For colorectal liver metastases treated with at least 70 Gy BED10, the minimum PTV dose, not nominal prescription dose, was the strongest dosimetric predictor of freedom from local progression, and polymetastatic disease independently worsened it. ^[88] A cold peripheral region or marginal miss can therefore produce local failure despite a high central dose. Report target coverage and the minimum or near-minimum PTV dose, not prescription BED10 alone.

Portal-vein invasion identifies a different prognostic group. SBRT may reduce dominant tumor thrombus and preserve threatened portal flow, but vascular invasion usually coexists with higher tumor burden and systemic relapse risk. In a retrospective series restricted to selected patients with portal-vein tumor thrombosis, the response rate was 83.3%, median overall survival was 13 months, and median progression-free survival was 10.2 months. ^[16] These results support local treatment of a threatening vascular component; they do not equate vascular control with cure.

Biomarker kinetics can refine interpretation but should not replace imaging and clinical assessment. AFP, and when relevant PIVKA-II, may fall after effective treatment, while persistently elevated or rising markers can indicate viable tumor or disease elsewhere; their prognostic value is strongest when interpreted with tumor burden, liver function, and imaging. The biological context also matters: prior systemic therapy was associated with inferior local control in the Belgian registry, a finding compatible with selection of radioresistant clones but vulnerable to confounding by more advanced disease. ^[5] Conversely, local SBRT may prolong benefit from an otherwise effective systemic regimen in oligoprogressive HCC. In a prospective phase 2 study of patients with no more than five progressing lesions in no more than three organs, SBRT delivered to all progressing sites while PD-1 inhibitor-based therapy and lenvatinib continued produced a median progression-free survival of 11.3 months and a 2-year overall survival of 84.9%; there was no randomized comparator. ^[17]

Intrahepatic cholangiocarcinoma

For intrahepatic cholangiocarcinoma , the available SBRT evidence is less mature and less homogeneous than the HCC literature. Patients are commonly selected for limited, technically targetable disease, adequate uninvolved liver, and controlled or controllable extrahepatic disease; diffuse intrahepatic involvement, progressive metastatic disease, biliary obstruction, or poor hepatic reserve shifts the expected benefit toward systemic or supportive treatment. Modern radiation therapy is used across primary liver and biliary tract cancers, but the supplied outcome literature does not establish a reliable disease-specific range for local control, intrahepatic control, progression-free survival, overall survival, freedom from liver failure, or quality of life in intrahepatic cholangiocarcinoma. ^[1] Do not import HCC control rates into cholangiocarcinoma; primary tumor biology, biliary anatomy, systemic-therapy sensitivity, and the extent of microscopic spread may produce different failure patterns.

Liver metastases

For liver metastasis , SBRT is best understood as metastasis-directed therapy. Its value depends less on controlling one visible lesion than on whether local ablation changes the course of a limited metastatic state while the primary tumor and systemic disease remain controlled. In the Belgian registry, metastatic lesions had local-control rates of 67.6%, 48.2%, and 31.0% at 1, 2, and 5 years, respectively, with corresponding overall-survival probabilities of 75.4%, 49.8%, and 18.6%. ^[5] Modern highly selected series can report higher target control: for colorectal liver metastases treated with MR-guided SBRT, local control was 93.8% at 12 months and 86.6% at 24 months, but distant hepatic control was only 52.5% and 49.2%, and progression-free survival was 31.5% and 25.5%. ^[116] The contrast is clinically decisive: an excellent treated-lesion result may coexist with new liver lesions or extrahepatic progression.

Primary tumor biology modifies both local response and survival. Colorectal metastases, particularly after several systemic-therapy lines or with elevated carcinoembryonic antigen, may be less radiosensitive than metastases from some other primaries; in a colorectal cohort, PTV minimum dose and polymetastatic disease independently predicted freedom from local progression. ^[88] The number of lesions, bilobar distribution, total treated volume, and the ability to cover every target without excessive normal-liver irradiation determine whether SBRT is a focal intervention or an unsafe attempt to treat diffuse disease. Systemic therapy must remain active or controllable: SBRT can consolidate an oligometastatic response or treat oligoprogression, but it cannot compensate for rapidly progressive polymetastatic disease.

Patterns of failure and interpretation of endpoints

Classify failure anatomically. In-field or marginal failure suggests residual viable tumor, inadequate target coverage, motion or registration error, or a cold PTV periphery; new intrahepatic failure indicates untreated microscopic or macroscopic liver disease and is captured more accurately by distant hepatic control than by local control; extrahepatic progression reflects systemic tumor biology and may dominate progression-free and overall survival even when every irradiated lesion remains controlled. The MR-guided colorectal series illustrates this separation: target local control remained high while distant hepatic control and progression-free survival were substantially lower. ^[116]

Freedom from liver failure is a distinct endpoint from tumor control. A patient may have controlled HCC or metastases but lose hepatic reserve through cirrhosis, portal hypertension, tumor replacement, biliary obstruction, or treatment-related injury. Conversely, a patient may retain liver function while dying from extrahepatic progression. Track bilirubin, albumin, INR, ascites, encephalopathy, Child-Pugh class, and ALBI grade alongside oncologic endpoints. Severe radiation-induced lymphopenia is another adverse prognostic signal: in an HCC cohort it occurred in 13.5% and was associated with shorter median progression-free survival and overall survival; portal hypertension and larger low-dose liver volumes increased its likelihood. ^[115]

Quality of life is often the patient-relevant endpoint when SBRT is used for oligoprogression or palliation, but outcome studies frequently emphasize radiographic control and survival rather than validated longitudinal quality-of-life measures. The Belgian registry specifically identified the need for future studies addressing patient selection, dosing, systemic combinations, long-term survival, and quality of life. ^[5] A stable treated lesion should therefore justify continued surveillance or systemic therapy only when symptoms, liver function, treatment burden, and patient goals remain favorable.

Pearl: A high local-control rate proves that the irradiated target was treated effectively; it does not prove eradication of the patient’s cancer. Judge SBRT by the combined trajectory of treated-lesion control, new intrahepatic and extrahepatic disease, hepatic reserve, systemic-therapy control, survival, and quality of life.

Acute toxicity usually occurs during treatment or within 90 days of completion; late toxicity is defined as occurring after 90 days. CTCAE version 5.0 should be used for grading, but the grade must be interpreted alongside symptoms, laboratory trends, and hepatic function rather than treated as a surrogate for liver failure ^[22].

The common acute syndrome is usually mild and self-limited: fatigue, anorexia, nausea, vague or localized right-upper-quadrant abdominal pain, and transient elevation of AST, ALT, or alkaline phosphatase. Ask about oral intake, vomiting, pain, fever, jaundice, bleeding, and stool color at each early post-treatment assessment. Give an antiemetic such as ondansetron when nausea limits intake, use acetaminophen cautiously for pain within the patient’s hepatic and alcohol-use context, and avoid NSAIDs when thrombocytopenia, portal hypertension, renal dysfunction, anticoagulation, or gastrointestinal mucosal injury increases bleeding risk. A short-lived transaminase rise is not equivalent to radiation-induced liver disease (RILD): assess bilirubin, albumin, INR, creatinine, platelet count, AST, ALT, alkaline phosphatase, and the presence of ascites or encephalopathy before attributing deterioration to radiation ^[22].

Cytopenias may reflect marrow irradiation, hypersplenism from portal hypertension, systemic therapy, infection, or bleeding rather than direct hepatic injury. Radiation-induced lymphopenia deserves separate attention: severe lymphopenia was defined as an absolute lymphocyte count below 0.5 × 10⁹/L, occurred in 13.5% of one HCC cohort, and was associated with larger low-dose liver volumes and portal hypertension ^[115]. Obtain a CBC when symptoms, baseline cytopenia, portal hypertension, concurrent systemic treatment, or a large low-dose bath raises concern; investigate fever or neutropenia urgently and manage clinically significant anemia or thrombocytopenia according to cause and procedural needs.

RILD is best separated into classical and nonclassic forms. Classical RILD reflects central-vein and sinusoidal injury with hepatic congestion, hepatocyte necrosis, hepatomegaly, and ascites. Nonclassic RILD is more relevant to patients with cirrhosis and is expressed as worsening hepatocellular function, rising bilirubin, falling albumin, coagulopathy, ascites, encephalopathy, or deterioration in Child-Pugh class ^[22]. A laboratory abnormality alone may be CTCAE grade 1 or 2 and may recover; clinically significant liver failure requires functional assessment and urgent management of the precipitating complication. Track both Child-Pugh and ALBI, where ALBI is an objective score derived from albumin and bilirubin. The measures are complementary: ALBI grade worsening at 6 months identified more cases of nonclassic RILD than a two-point Child-Pugh increase, and most cases were nonoverlapping ^[48].

Risk rises when little functional liver is spared, the mean liver dose is high, the treated volume is large or multifocal, or baseline reserve is poor. In a retrospective SBRT cohort, change in ALBI at 6 months correlated with lesion size, PTV size, mean liver dose, and multifocality ^[26]. Cirrhosis, portal hypertension, ascites, thrombocytopenia, impaired portal-vein patency, low albumin, Child-Pugh B or C disease, and prior liver-directed treatment narrow the margin for additional injury. Child-Pugh B7-B9 can be treated selectively, but the evidence remains retrospective and highly selected; a 38-patient series contained only one Child-Pugh C patient, so routine treatment of Child-Pugh C cannot be inferred ^[22]. Protect the largest possible volume of uninvolved liver, minimize low-dose exposure as well as high-dose exposure, correct reversible ascites or biliary obstruction, and involve hepatology early when reserve is marginal.

Central hilar tumors require special caution because the central bile ducts and portal structures behave as serial organs. Prior biliary instrumentation, stents, drains, cholangitis, or obstruction may make a post-treatment bilirubin rise difficult to interpret and may increase the consequences of a duct injury. Obtain liver tests and, when clinically indicated, ultrasound, multiphasic CT, or MRCP to distinguish radiation-related hepatocellular dysfunction from obstruction, stent occlusion, portal-vein thrombosis, infection, or tumor progression. A new or worsening biliary stricture may require endoscopic or percutaneous decompression and stenting; treat cholangitis with urgent cultures, antibiotics, and drainage when obstruction is present.

Prior radiation is a major late-risk modifier because cumulative dose, rather than the second plan alone, determines the tolerance of liver and adjacent serial organs. Reirradiation series are small and heterogeneous, but cumulative dose assessment with deformable registration predicted higher mean liver dose and smaller uninvolved-liver volumes than a nonaccumulated analysis ^[112]. Recalculate cumulative exposure to liver, stomach, duodenum, bowel, central bile ducts, chest wall, and ribs before retreatment. Anticoagulation does not itself cause radiation injury, but it can magnify the clinical consequences of a radiation-associated ulcer, biliary intervention, hemobilia, or gastrointestinal bleeding; coordinate interruption and resumption with the prescribing team and proceduralist.

Late gastrointestinal injury may present months after treatment with epigastric or right-upper-quadrant pain, nausea, anorexia, vomiting, dyspepsia, occult or overt bleeding, ulceration, stricture, obstruction, fistula, or perforation. The mechanism is progressive microvascular injury, ischemia, inflammation, and fibrosis; a focal hot spot against the stomach, duodenum, or bowel can therefore produce severe injury even when the mean dose to the organ is modest ^[32]. Investigate persistent pain, vomiting, anemia, melena, hematemesis, fever, or peritoneal signs with CBC, metabolic and liver panels, lipase when appropriate, contrast CT, and endoscopy for suspected upper gastrointestinal mucosal injury. Start a proton-pump inhibitor for suspected or confirmed ulceration, involve gastroenterology for endoscopic hemostasis or assessment of healing, and provide nutritional support when pain or obstruction limits intake. A refractory late ulcer may be considered for hyperbaric oxygen only after multidisciplinary review; evidence after liver SBRT consists principally of case-level experience rather than a validated treatment protocol ^[32].

Rib fracture and chest-wall injury are uncommon but become relevant for dome or subcapsular lesions. Ask about focal pleuritic or positional pain and examine for localized tenderness; use CT to evaluate a suspected fracture and exclude tumor progression or infection. Treat uncomplicated pain with a liver-conscious analgesic plan, while a displaced fracture, enlarging mass, neurologic deficit, or uncontrolled pain merits specialist evaluation.

Portal-hypertension complications may emerge or worsen through loss of hepatic reserve: ascites, variceal bleeding, splenic sequestration with thrombocytopenia, and hepatic encephalopathy. Manage these with hepatology-directed therapy, including sodium restriction and diuretics for appropriate ascites, lactulose with or without rifaximin for encephalopathy, and urgent endoscopic or interventional management for variceal hemorrhage. Do not label ascites or encephalopathy as “radiation toxicity” without evaluating infection, portal-vein obstruction, tumor progression, medication effects, gastrointestinal bleeding, and alcohol-related or metabolic decompensation.

Pearl: A falling albumin or rising bilirubin is a warning signal, not a diagnosis. Diagnose clinically meaningful hepatic failure only after integrating Child-Pugh or ALBI trajectory, INR, ascites, encephalopathy, portal and biliary imaging, infection, bleeding, tumor progression, systemic therapy, and the cumulative radiation plan ^[22][48].

Surveillance after liver SBRT must answer two questions in parallel: is the treated focus controlled, and is the remaining liver tolerating treatment? Use multiphasic contrast-enhanced liver MRI or CT , chest imaging, laboratory testing, and clinical assessment. The interval is not uniform across patients; tailor it to tumor biology, primary tumor type, liver reserve, ongoing systemic therapy, and the intent of treatment, particularly whether the patient is being bridged to transplantation or evaluated for salvage. No universally accepted post-SBRT imaging schedule exists, although retrospective HCC series commonly obtained liver MRI at 1, 3, 6, 9, and 12 months and then every 3-6 months. ^[121][124]

At each visit, assess symptoms, performance status, weight and nutritional intake, abdominal pain, nausea, fever, jaundice, bleeding, ascites, edema, and cognition. Obtain a complete blood count, creatinine, bilirubin, albumin, alkaline phosphatase, AST, ALT, and INR; calculate Child-Pugh and ALBI scores when liver reserve or treatment tolerance is in question. Follow disease-specific biomarkers when informative, such as AFP for HCC and PIVKA-II or CA 19-9 when relevant to the tumor type. A falling marker supports response only when it agrees with imaging and the clinical course; a normal marker does not exclude residual tumor. Post-treatment response assessment is intrinsically multiparametric and commonly integrates contrast enhancement, nonenhancing tissue, diffusion, and tumor-marker behavior. ^[122]

Compare every examination with the pretreatment diagnostic study and the planning CT, not merely with the immediately preceding scan. The treated region may develop a geographic or wedge-shaped radiation-induced focal liver reaction, with early arterial hyperemia or rim-like perilesional enhancement, T2-weighted hyperintensity, and hepatobiliary-phase hypointensity, followed by delayed enhancement, focal volume loss, and capsular retraction. These findings can extend beyond the original tumor and persist for months; their geographic distribution within the radiation field and stability or gradual involution favor benign radiation change over viable tumor. In one pathology- and biomarker-correlated series, successfully treated HCCs showed early hyperemia followed by capsular retraction and delayed enhancement, while persistent enhancement did not imply treatment failure in the absence of growth. ^[122][126]

Do not call residual HCC solely because arterial phase hyperenhancement (APHE) or washout persists early after SBRT. Radiation-related perfusion and inflammatory changes can preserve both appearances despite tumor control; persistent APHE was present in 58% of treated HCCs in a longer-term MRI cohort, and successful treated lesions could retain central APHE and washout through 12 months. ^[125][126] Favor viable tumor when enhancement becomes new, nodular, mass-like, or progressively enlarging, particularly when it is accompanied by washout, increasing T2 or diffusion-weighted signal, failure of the apparent diffusion coefficient to rise, interval growth, or a concordant biomarker increase. Serial change is often more informative than a single early scan: gradual loss of APHE and slow size reduction are expected after SBRT, whereas increasing mass size or new nodular APHE suggests local progression. ^[127]

Use the LI-RADS treatment-response framework for treated HCC, selecting the radiation-specific pathway when available. The 2024 radiation algorithm emphasizes mass-like enhancement and interval behavior and was designed to reduce false-positive viable calls after radiation; lesions categorized as nonprogressing warrant close short-interval imaging rather than immediate salvage on the basis of equivocal early enhancement. ^[128] Record the actual imaging observations, enhancement morphology and phase, washout, diffusion, subtraction behavior, hepatobiliary-phase signal, size, and relation to the treated field, in addition to the categorical LI-RADS assessment. Subtraction images are particularly useful when intrinsic T1 hyperintensity, hemorrhage, proteinaceous material, or post-treatment signal obscures enhancement; gadoxetate-enhanced MRI can show hepatobiliary-phase defects, but irradiated liver itself may be hypointense, so hepatobiliary hypointensity alone does not establish viable tumor. ^[122][130]

RECIST 1.1, which relies predominantly on change in lesion diameter, is an incomplete response measure after ablative therapy. Radiation-treated lesions may remain enlarged, shrink slowly, or show transient enhancement without viable tumor; studies of liver metastases therefore found that size-based assessment alone can misclassify response. ^[127][137] For metastases, inspect the treated margin for new or enlarging lobulated or nodular enhancement and interpret it with the original dose distribution and serial morphology. Lobulated enhancement preceded size-based progression in half of progressive lesions in one retrospective metastasis series, making it a useful warning sign rather than a stand-alone diagnosis. ^[136]

Evaluate toxicity alongside tumor response. A transient AST or ALT rise may reflect treatment-related injury, but progressive bilirubin elevation, falling albumin, rising INR or creatinine, new or worsening ascites, encephalopathy, jaundice, fever, or right-upper-quadrant pain requires assessment for nonclassic radiation-induced liver disease and competing causes such as infection, biliary obstruction, portal or hepatic-vein thrombosis, bleeding, drug toxicity, and tumor progression. New vomiting, melena, hematemesis, anemia, or peritoneal signs should prompt urgent evaluation for gastrointestinal injury. Interpret laboratory trends against baseline liver reserve and systemic therapy rather than assigning causality from a single abnormal result.

Reserve PET/CT for a defined clinical question, such as suspected extrahepatic progression, an FDG-avid metastatic or cholangiocarcinoma phenotype, or discordance between anatomic imaging and clinical findings; it is not a routine substitute for multiphasic liver MRI or CT in HCC. After liver-metastasis SBRT, FDG uptake may remain moderately elevated during tissue repair and then decline; one study found a controlled-lesion SUVmax plateau near 3.1 at approximately 5 months, whereas SUVmax ≥6 was proposed as suspicious for local failure. ^[138] These thresholds are disease- and protocol-dependent and should not be transferred uncritically to HCC or to different scanners and acquisition schedules.

Perform biopsy when imaging remains indeterminate and histology would change management, for example, when mixed tumor, metastasis, treatment-related change, or a second primary is plausible; when salvage resection, ablation, transplantation, or a change in systemic therapy is being considered; or when serial imaging and biomarkers remain discordant. Do not biopsy a typical stable focal liver reaction simply to prove necrosis. Refer persistent equivocal findings, suspected recurrence near the hilum or major vessels, possible transplant-exclusion disease, and suspected severe toxicity to multidisciplinary review involving abdominal radiology, hepatology, transplant surgery when relevant, medical oncology, interventional radiology, and radiation oncology. The purpose is not to force an early label, but to integrate imaging morphology, dose distribution, liver function, competing disease, and the feasibility of salvage.

Pearl: After liver SBRT, viable tumor is a pattern of progressive mass-like change, not a synonym for persistent enhancement; interpret the treated focus, the radiation-shaped liver reaction, liver function, biomarkers, and extrahepatic disease as one longitudinal assessment. ^{[125][127][128]}

Multidisciplinary review should decide whether SBRT is replacing, complementing, or preserving another curative option. The hepatobiliary surgeon should first determine whether resection can achieve adequate margins without precipitating liver failure; SBRT is not an equivalent default for an operable patient because comparative evidence remains largely retrospective, although a post-hoc analysis of two randomized cohorts in recurrent small HCC found similar 3-year local progression-free survival and overall survival after SBRT and resection. ^[39] The KLCA 2026 expert consensus frames SBRT primarily for HCC when resection, transplantation, or local ablation is not feasible, while the international-guideline review reports that surgery remains the preferred radical treatment for operable early HCC and SBRT is generally an alternative or salvage treatment. ^[3][141]

For a lesion technically amenable to thermal ablation, compare radiofrequency ablation , microwave ablation , and SBRT by tumor size, visibility, access route, vessel proximity, dome or subphrenic location, and the volume of functioning liver that each option would sacrifice. Thermal ablation remains attractive for small accessible tumors, whereas SBRT is particularly useful when heat-sink from a major vessel, poor percutaneous access, or an unfavorable diaphragm, cardiac, hilar, or bowel relationship makes ablation unreliable or unsafe. ^[141] Do not present retrospective comparisons as proof of superiority: a 2025 meta-analysis of HCC within Milan criteria found no significant difference in overall or progression-free survival between SBRT and radiofrequency ablation, whereas a broader 2026 meta-analysis found lower local recurrence with external-beam radiotherapy but no overall-survival advantage; both analyses were dominated by nonrandomized studies. ^[149][150]

Prior ablation or resection does not preclude SBRT, but it changes the target and the risk calculation. Fuse pretreatment imaging with the planning dataset to distinguish viable peripheral tumor from cavity, scar, lipiodol, altered perfusion, or postoperative anatomy; then reconstruct prior dose and reassess the remaining functional liver and nearby stomach, bowel, duodenum, and bile ducts. A recurrent focus after resection may be treated with SBRT when repeat surgery would require excessive parenchymal loss, while SBRT after incomplete ablation is reasonable only when residual viable disease can be identified and safely encompassed. ^[141]

Transarterial chemoembolization (TACE) and transarterial radioembolization (TARE) are complementary rather than interchangeable treatments. Use TACE when the expected benefit comes from arterial devascularization and embolic or cytotoxic effect; use SBRT for a persistent dominant focus, an anatomically difficult target, or localized progression after incomplete TACE response. Retrospective comparisons suggest better in-field control with SBRT than TACE, but selection bias and heterogeneous treatment intent prevent a general superiority claim. ^[141] Combining TACE and SBRT may be considered for selected bulky or incompletely controlled HCC, but the evidence is mainly retrospective; a propensity-matched analysis of recurrent HCC found no statistically significant survival improvement from adding TACE after SBRT, despite no acute grade ≥3 toxicity in either group. ^[148]

TARE followed by or preceding SBRT demands particular caution because both treatments irradiate liver, although their dose distributions differ. Reconstruct the microsphere-treated territory when possible, estimate the cumulative exposure to uninvolved liver and serial organs, and avoid assuming that a normal bilirubin or stable ALBI grade eliminates delayed risk. A small 12-patient retrospective series of TARE plus SBRT included only Child-Pugh A patients, had a median interval of 6.5 months between treatments, and reported no toxicity above grade 3; this supports feasibility in carefully selected patients, not a standard sequence. ^[9]

The sequence of local therapy and systemic therapy should follow disease biology and protocol safety rather than a presumed radiosensitizing effect. Continue effective systemic treatment during SBRT only when the protocol permits it, the patient has adequate hepatic and marrow reserve, and the expected gain from uninterrupted control outweighs the risk of overlapping toxicity. Tyrosine-kinase inhibitors, cytotoxic chemotherapy, and antiangiogenic agents can increase concern for hepatic injury, thrombocytopenia, impaired mucosal repair, hypertension, bleeding, gastrointestinal ulceration, or perforation; antiangiogenic therapy is especially problematic when a high-dose region abuts stomach or bowel. Specify drug holds, restart criteria, laboratory monitoring, and management of hypertension, anticoagulation, thrombocytopenia, and active ulcer disease in the treatment protocol rather than applying a universal interval. Evidence supporting combined SBRT and pharmacologic therapy remains heterogeneous: a 2025 meta-analysis included only nine studies, comprising randomized and retrospective designs, and concluded that larger prospective trials are needed despite favorable pooled efficacy estimates. ^[142]

Immune checkpoint inhibitors (ICIs) may be continued, started, or resumed around SBRT only after explicit multidisciplinary agreement. Radiation can increase antigen release and inflammatory signaling, providing a biologic rationale for combination treatment, but the optimal timing and dose are unsettled; a 2026 systematic review found highly heterogeneous regimens and no consistent increase in severe toxicity, while emphasizing that concurrent and sequential strategies cannot yet be considered equivalent. ^[25][143] The prospective PEMRAD phase II study started SBRT on day 2 after pembrolizumab and continued pembrolizumab every 21 days, but it enrolled only 18 patients, reported grade ≥3 treatment-related toxicity in 22%, and included one treatment-related death from myocarditis; these data support feasibility under protocolized monitoring, not routine concurrent treatment. ^[144]

For oligoprogression, preserve a systemic regimen that is controlling most disease and irradiate only the progressing sites when all sites can be safely treated. The prospective phase II study of first-line programmed-death-1 inhibitor therapy plus lenvatinib treated up to five progressing lesions in no more than three organs with SBRT while continuing the original regimen; median progression-free survival was 11.3 months and grade 3-4 toxicity occurred in 8.5%, without a randomized comparator. ^[17] This strategy is most defensible when progression is limited, performance status and hepatic reserve remain adequate, and there is a credible systemic option to continue; diffuse progression, worsening liver function, or rapidly changing drug toxicity favors systemic-treatment modification rather than serial focal irradiation.

Hepatic arterial infusion chemotherapy (HAIC) may be paired with SBRT for selected HCC with portal-vein tumor thrombus or other high-risk intrahepatic disease, but the available evidence is retrospective and does not establish the sequence, chemotherapy regimen, or treatment-hold interval. A propensity-matched multicenter study reported longer survival with HAIC plus SBRT than HAIC alone, but its nonrandomized design leaves confounding by patient selection and treatment intensity. ^[147] Treat portal-vein tumor thrombus as tumor when defining the radiation target, while separately assessing portal flow, biliary drainage, collateral vessels, platelet count, and the risk of hepatic decompensation.

SBRT can serve as a bridge to transplantation when it controls a transplant-eligible tumor during the waiting period, and as a downstaging treatment only when the transplant center confirms that the expected post-treatment disease state can enter its accepted criteria. Before SBRT, document the transplant endpoint, current and maximum tumor burden, vascular invasion, AFP or other relevant biomarkers, extrahepatic staging, wait-list dropout risk, and the effect of prior TACE, TARE, ablation, resection, or radiation on future transplantation. After treatment, persistent enhancement alone should not trigger transplant exclusion; the transplant team should integrate serial multiphasic imaging, biomarkers, tumor biology, and any biopsy or explant findings. A 2025 systematic review of 19 predominantly nonrandomized studies reported pooled radiologic response of 61.2%, pathologic response of 83.8%, and grade ≥3 toxicity of 1.2%, but these estimates are vulnerable to selection and publication bias. ^[4]

Transplant-center review is mandatory for patients with marginal hepatic reserve, prior liver-directed therapy, portal hypertension, or disease near the hilum. In a prospective pilot of nine transplant candidates with Child-Pugh B8 or worse cirrhosis treated with 40 Gy in 5 fractions, six remained transplant eligible or underwent transplantation at 1 year, while one patient developed grade 4 acidosis, acute encephalopathy, and hepatic failure; the study was too small to support routine treatment of decompensated cirrhosis. ^[12] The transplant program should also define whether prior SBRT could complicate explant dissection, biliary reconstruction, vascular management, or interpretation of viable tumor, because local control does not by itself establish transplant candidacy.

Repeat liver SBRT and reirradiation require a new cumulative-risk assessment, not simply repetition of the original plan. Reconstruct prior dose on the current anatomy, account for interval liver regeneration or atrophy, identify overlap with stomach, duodenum, bowel, central bile ducts, kidneys, and spinal cord, and use stricter target-volume and organ-at-risk priorities than for an unirradiated liver. Favor reirradiation only when the recurrence is focal, the prior dose distribution is recoverable, hepatic reserve is stable, and no safer local or systemic alternative offers comparable benefit. If target coverage conflicts with preservation of functioning liver or avoidance of catastrophic serial-organ injury, reduce dose per fraction, increase the number of fractions, defer treatment, or choose another modality; cumulative BED alone cannot determine safety. ^[141]

Extrahepatic oligometastases may be treated with SBRT when the primary tumor and other systemic disease are controlled or controllable, the number and geometry of lesions permit complete local treatment, and treatment of the liver will not compromise systemic therapy. The decision should be made jointly by disease-site surgery, interventional oncology, medical oncology, and radiation oncology; local treatment of one compartment is not a substitute for systemic therapy when new lesions are appearing elsewhere. ^[6]