• colorectal;
  • liver;
  • stereotactic body radiotherapy;
  • outcomes


  1. Top of page
  2. Abstract
  6. Acknowledgements


This study was undertaken to determine outcomes of stereotactic body radiotherapy for colorectal liver metastases in a pooled patient cohort.


Patients with colorectal liver metastases from 3 institutions were included if they had 1 to 4 lesions, received 1 to 6 fractions of stereotactic body radiotherapy, and had radiologic imaging ≥3 months post-treatment. Sixty-five patients with 102 lesions treated from August 2003 to May 2009 were retrospectively analyzed. A tumor control probability (TCP) model was used to estimate the 3-fraction dose required for >90% local control after converting the schedule into biologically equivalent dose (BED), single-fraction equivalent dose, or linear quadratic model-based single-fraction dose.


Forty-seven (72%) patients had ≥1 chemotherapy regimen before stereotactic body radiotherapy, and 27 (42%) patients had ≥2 regimens. The median follow-up was 1.2 years (range, 0.3-5.2 years). The median dose was 42 gray (Gy; range, 22-60 Gy). When evaluated separately by multivariate analysis, total dose (P = .0015), dose/fraction (P = .003), and BED (P = .004) all correlated with local control by lesion. On multivariate analysis, nonactive extrahepatic disease was associated with overall survival (OS; P = .046), and sustained local control was closely correlated (P = .06). By using single-fraction equivalent dose, BED, or linear quadratic model-based single-fraction dose in the TCP model, the estimated dose range needed for 1-year local control >90% is 46 to 52 Gy in 3 fractions.


Liver stereotactic body radiotherapy is well tolerated and effective for colorectal liver metastases. The strong correlation between local control and OS supports controlling hepatic disease even for heavily pretreated patients. For a 3-fraction regimen of stereotactic body radiotherapy, a prescription dose of ≥48 Gy should be considered, if normal tissue constraints allow. Cancer 2011. © 2011 American Cancer Society.

Colorectal cancer is the third most common cancer and is also the third leading cause of cancer death.1 Metastatic colon cancer was once considered an incurable disease with poor prognoses. However, the advancements in systemic therapy and improved chemotherapy combinations have resulted in dramatically improved survival of approximately 2 years.2-4 Consequently, there has been a shift in the treatment paradigm for this cancer leading to more aggressive local therapy for patients with a limited number of metastases. For these patients, it is possible that addressing oligometastases could result in prolonged disease-free periods with a possibility of cure.

Stereotactic body radiotherapy represents an emerging new technology that has shown efficacy in ablating liver tumors.5-10 However, series are difficult to compare because of heterogeneity in the primary site of disease, the number of liver tumors, and radiation doses used across studies. Our goal was to determine the outcomes after stereotactic body radiotherapy for colorectal liver metastases in a large pooled cohort of patients with long-term follow-up from 3 institutions: Princess Margaret Hospital (Toronto, Ontario), University of Colorado Denver (Aurora, Colo), and Stanford University (Stanford, Calif).


  1. Top of page
  2. Abstract
  6. Acknowledgements

This study was approved by the institutional review boards of each of the participating centers. The inclusion criteria were 1 to 4 lesions, stereotactic body radiotherapy treatments delivered in 1 to 6 fractions, and imaging at least 3 months after stereotactic body radiotherapy. A total of 65 patients with 102 colorectal metastases treated between August 2003 and May 2009 were identified and included for analysis. Of these, 55 (85%) patients were enrolled in prospective trials evaluating the efficacy of stereotactic body radiotherapy for liver tumors as part of broader trials that included all histologies. Patients enrolled prospectively at Stanford University and Princess Margaret Hospital were required to have unresectable disease or be medically inoperable, whereas those enrolled in the University of Colorado Denver prospective trial were not. Fifty-seven patients (90 lesions) received stereotactic body radiotherapy from a conventional linear accelerator, whereas 8 patients (12 lesions) were treated with CyberKnife (Accuray, Sunnyvale, Calif). Treatment planning and delivery technique varied by institution and have been described in depth previously.8, 9, 11

Briefly, patients treated at Stanford University had gold fiducials implanted within or near the tumor. The gross tumor volume (GTV) was outlined using contrast-enhanced computed tomography (CT) scans and positron-emission tomography (PET) with a 3 mm expansion for the planning target volume (PTV). The CyberKnife treatment machine is a frameless image-guided system consisting of a 6 MV linear accelerator mounted on a robotic arm. Daily setup position was verified using 2 orthogonal oblique x-ray images that align the fiducials. Respiratory tracking was performed using Synchrony (Accuray).12

At University of Colorado Denver, patients underwent simulation with either abdominal compression or active breathing control. The GTV was defined by CT, sometimes augmented by magnetic resonance imaging (MRI) or PET. Expansions of 5 to 7 mm radially and 10 to 15 mm craniocaudally were used for the PTV, depending on the setup technique. Stereotactic body radiotherapy was delivered via dynamic conformal arcs or multiple noncoplanar static beams using 6 to 15 MV of photons. Daily image guidance, either by orthogonal x-rays or onboard CT imaging, verified the target position before each treatment delivery. Stereotactic body radiotherapy was administered in a 3-fraction course to be completed in no more than 14 elapsed days, but typically within 1 week.9

At Princess Margaret Hospital, patients similarly had either abdominal compression or active breathing control to account for respiratory motion. The GTV was defined using CT and MRI, and an 8 mm expansion was added for the clinical target volume. An additional 5 mm or greater margin formed the PTV expansion. Daily image guidance was performed using the dome of the diaphragm or the liver as a surrogate for the liver tumor position. Two-dimensional orthogonal MV imaging was used early on for setup verification, but eventually, 3-dimensional kV cone-beam CT combined with 2-dimensional kV fluoroscopy became standard. Repositioning was performed for ≥3 mm misalignments.8

The prescription dose varied by treating institution. The Stanford University trial was a single-fraction dose-escalation trial beginning at 18 gray (Gy) and increasing by 6-Gy increments to 30 Gy. Patients enrolled at University of Colorado Denver received a 3-fraction regimen beginning at 36 Gy and escalating to 60 Gy. At Princess Margaret Hospital, the dose was administered in a 6-fraction schedule. The total dose was chosen based on the risk of liver toxicity using the Lyman-Kutcher-Burman normal-tissue complication probability model.13, 14

The median age of patients was 67 years (range, 39-87 years). The median total dose prescribed was 41.7 Gy (range, 22-60 Gy) in 6 fractions (range, 1-6 fractions). Sixty-seven (67%) lesions received 6 fractions, 21 (21%) lesions received 3 fractions, 8 (8%) lesions received 1 fraction, and 6 (6%) lesions received 5 fractions. Forty-seven (72%) patients had at least 1 chemotherapy regimen after diagnosis of liver disease, and 27 (42%) patients had ≥2 prior regimens. Table 1 shows the clinical and treatment characteristics for all study patients.

Table 1. Patient and Lesion Characteristics
CharacteristicPatients, n = 65Lesions, n = 102
  1. SBRT indicates stereotactic body radiotherapy; GTV, gross tumor volume; Gy, grays; BED, biologically effective dose.

Age, y67 (39-90)
 Caucasian57 (88%)
 Other8 (12%)
 Princess Margaret41 (63%)68 (67%)
 University of Colorado16 (25%)22 (22%)
 Stanford University8 (12%)12 (12%)
No. of prior chemotherapy regimens
 018 (27%)
 120 (31%)
 ≥227 (42%)
No. of post-SBRT chemotherapy regimens
 040 (61%)
 116 (25%)
 ≥29 (14%)
No. of liver lesions
 138 (58%)
 214 (22%)
 312 (18%)
 41 (2%)
GTV, mL30.1 (0.6-3088)
Total dose, Gy41.7 (22-60)
Dose/fraction, Gy8 (5-30)
BED, Gy75 (40.5-180)
No. of fractions6 (1-6)
 15 (8%)8 (8%)
 316 (25%)21 (21%)
 54 (6%)6 (6%)
 640 (62%)67 (66%)
Active nonhepatic disease
 Yes22 (34%)
 No43 (66%)

Patients were monitored for in-field local recurrences, out-of-field hepatic recurrences, and extrahepatic recurrences using CT, MRI, and, if available, fluorodeoxyglucose-PET imaging. All patients were required to have an imaging study at least 3 months after stereotactic body radiotherapy, and repeat imaging was generally done at 3-month intervals during the first year followed by 3- to 6-month intervals thereafter.


At Stanford University and University of Colorado Denver, local control was defined as the lack of progression of the treated lesion on imaging, either as increase in size or as metabolic activity. At the third institution (Princess Margaret Hospital), local control was defined by Response Evaluation Criteria in Solid Tumors criteria. Local control was analyzed by patient and by lesion. Out-of-field hepatic and extrahepatic recurrences were also monitored. Patients were not censored for local control if they developed recurrences elsewhere in the liver or distally. In addition, patients were only censored for local control at the time of their most recent imaging study that included the liver.

SAS and JMP software were used for all statistical computations (SAS Institute, Cary, NC). The Kaplan-Meier product limit method provided estimates of local control and overall survival (OS).15 The log-rank test statistic was used for univariate analysis to assess the level of statistical significance between strata of selected prognostic factors. Cox regression provided a multivariate analysis of these endpoints with selected prognostic factors; backward selection provided the most parsimonious final model.

The relationship between radiation dose and tumor control was explored with a tumor control probability (TCP) model. The patient tumor control data were fit to the following logistic TCP model16:

  • equation image

where D refers to radiation dose analyzed, TCD50 refers to the dose that achieves 50% tumor control, and the coefficient k describes the slope of the sigmoid curve. The tumor control endpoint was defined as local control by lesion at the 12-month time point. Lesions that failed after 12 months were scored as locally controlled at 12 months, and lesions controlled but not followed for at least 12 months were excluded from this analysis. The curve was fitted by nonlinear regression using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, Calif). The input dose value D was considered as single-fraction equivalent dose calculated according to the method of Park and colleagues,17 the biologically equivalent dose (BED) calculated using the linear-quadratic model with an α/β ratio of 10, or a linear quadratic model-based estimate of the equivalent single-fraction dose using the same α/β ratio. Once the curve was fitted, the 3-fraction stereotactic body radiotherapy dose required to achieve 90% local control was then derived from the relevant value of single-fraction equivalent dose, BED, or linear quadratic model-based single-fraction dose.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The median follow-up was 1.2 years (range, 0.3-5.2), and among living patients it was 1.4 years (range, 0.3-4.6). At the time of analysis, 38 (58%) patients had died and 27 (42%) patients were alive. In total, 30 (29%) local in-field recurrences were observed in 24 (37%) patients, and 44 (68%) patients had progression outside of the liver. Forty-four (68%) patients had out-of-field liver failures, 32 (73%) of whom also progressed outside of the liver. Of the 27 patients with local recurrences, 9 (33%) had no other sites of failure.

When analyzed by lesion, the 12-month, 18-month, and 24-month local control rates were 67%, 65%, and 55%, respectively. On univariate analysis by lesion, age (P = .045), BED (P = .00001), dose per fraction (P = .0019), total dose (P = .0007), and maximum lesion size (0.04) were significant for local control (Table 2). The 12-month, 18-month, and 24-month local control rates by lesion were 84%, 84%, and 66%, respectively, for doses ≥42 Gy, and they were 48%, 43%, and 43%, respectively, for doses <42 Gy (Fig. 1). The 12-month, 18-month, and 24-month local control rates by lesion were 86%, 80%, and 71%, respectively, for BED ≥79 Gy, and they were 42%, 31%, and 31%, respectively, for BED <75 Gy (Fig. 2). Total dose, dose per fraction, and BED were strongly correlated with each other (data not shown) and, therefore, were included separately in the multivariate analysis (Table 3). All 3 factors correlated significantly with local control.

thumbnail image

Figure 1. Actuarial local control by lesion stratified by total dose delivered is shown. Gy = gray.

Download figure to PowerPoint

thumbnail image

Figure 2. Actuarial local control by lesion stratified by biologically effective dose (BED) delivered is shown.

Download figure to PowerPoint

Table 2. Univariate Analysis
FactorLocal Control by PatientLocal Control by LesionOverall Survival
  • GTV indicates gross tumor volume; BED, biologically effective dose.

  • a

    Total GTV in the patient.

  • b

    Sum of maximum tumor size for each tumor.

No. of prior chemotherapy regimens.99.87.28
Maximum lesion dimension.99b.04.019b
Total dose.028.0007.21
Active nonliver disease.54.7.04
No. of lesions.39.35
Table 3. Multivariate Analysis
FactorLocal ControlFactorOverall Survival
By LesionBy Patient
  • SBRT indicates stereotactic body radiotherapy; GTV, gross tumor volume; BED, biologically effective dose.

  • a

    Total GTV in the patient.

Total dose.0015.034Active nonliver disease.046
No. of prior chemotherapy regimens.63.84Local failure.06
Age.13.42Total No. of chemotherapy regimens.64
No. of days of SBRT.75.88No. of lesions (1 vs 2-4).5
Dose per fraction.003.18 
No. of prior chemotherapy regimens.6.81
No. of days of SBRT.11.37
No. of prior chemotherapy regimens.42.58
No. of days of SBRT.2.5

When analyzed by patient, the 12-month, 18-month, and 24-month local control rates were 62%, 55%, and 45%, respectively. When local control was analyzed by patient on univariate analysis (Table 2), total dose (P = .028) and BED (P = .05) were significant, whereas dose per fraction was not (P = 0.17). On multivariate analysis, when analyzing total dose, dose per fraction, and BED separately (Table 3), only total dose (≥42 Gy vs <42 Gy) significantly correlated with local control by patient (P = .034).

The 12-month, 18-month, and 24-month OS rates were 72%, 55%, and 38%, respectively. On univariate analysis, BED (P = .015), maximum lesion size (P = .04), and active extrahepatic disease (P = .017) were significantly associated with OS (Table 2). The 12-month, 18-month, and 24-month OS rates were 77%, 60%, and 45%, respectively, for patients without active extrahepatic disease, and 63%, 45%, and 18%, respectively, for patients with extrahepatic disease (Fig. 3). To determine the effect of a local failure on survival, local control was analyzed as a prognostic factor by stratifying patients into 3 categories: no local failure, local failure occurring before the median failure time (0.7 years), and local failure occurring after the median failure time. On univariate analysis, local control was not correlated with improved OS (P = .09). On multivariate analysis for OS, only the absence of active extrahepatic disease was associated with improved OS (P = .046). However, local control was borderline significant for OS (P = .06).

thumbnail image

Figure 3. Actuarial overall survival stratified by the presence or absence of active extrahepatic disease is shown.

Download figure to PowerPoint

Figure 4 shows the fitted TCP curves of dose versus 1-year local control, considering dose expressed as single-fraction equivalent dose, BED, or linear quadratic model-based single-fraction dose. The R2 value is highest for the single-fraction equivalent dose curve, although the fit is qualitatively similar for all 3 dose indices. The estimated doses required for a 90% local control at 1 year are 48, 117, and 33 Gy when expressed as for single-fraction equivalent dose, BED, and linear quadratic model-based single-fraction dose, respectively. Converting these values back into a 3-fraction stereotactic body radiotherapy regimen yields an estimate of the required stereotactic body radiotherapy dose to achieve 90% local control of 46 to 52 Gy in 3 fractions.

thumbnail image

Figure 4. Dose versus tumor control probability curves using (Top) single-fraction equivalent dose (SFED) based on the universal survival curve, (Middle) biologically effective dose (BED), and (Bottom) single-fraction equivalent dose using the standard linear quadratic model are shown. The latter 2 are both based on the standard linear quadratic model (α/β = 10) . LC indicates local control; Gy, gray; TCD50, dose that achieves 50% tumor control; k, slope of the sigmoid curve; LQ, linear quadratic.

Download figure to PowerPoint


Toxicities occurring within 3 months of stereotactic body radiotherapy were defined as acute, and those occurring after 3 months were considered late. Overall, there were 11 (17%) patients with grade ≥2 acute gastrointestinal (GI) toxicity. Two (3%) patients had grade ≥3 elevated liver enzymes, but none had symptomatic liver toxicity.

Six patients experienced late toxicities. Four (6%) patients had late GI toxicities grade ≥2: 2 with grade 3 gastritis and 2 with grade 2 small bowel ulcers. Two patients had grade 3 elevated liver enzymes. Finally, 2 patients had persistent chest wall pain. One patient had gastritis and chest wall pain, whereas another patient had both gastritis and elevated liver enzymes. No rib fractures were observed.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Results of the present study indicate that stereotactic body radiotherapy is a safe and effective means of eradicating liver metastases in patients with colorectal cancer. Even in the population of heavily pretreated patients analyzed in the present study, the control of hepatic metastases appeared to be an important clinical objective, given the correlation between sustained local control and improved survival.

Overall, the outcomes of patients with metastatic colorectal cancer have improved significantly in the past few years with improvements in systemic therapy, achieving median survivals approaching 2 years.2-4 For the subgroup of patients with limited metastatic disease in the liver, nonrandomized studies have shown that patients who can undergo resection of liver metastases have 5-year survivals of 30% to 40%.18-20 As a result, the integration of aggressive local therapy along with systemic treatment has been playing an increasingly important role in treatment for this disease. Even in the setting of complete radiographic responses after chemotherapy, local therapy is indicated. In a study by Benoist et al., there was an 83% rate of local failure or persistent disease in sites that had shown complete radiographic response to chemotherapy by CT.21

Surgery has historically been the standard of care and, when feasible, is considered potentially curative therapy for properly selected patients. However, because of tumor size and location, 80% to 90% of patients present with unresectable disease,22 and neoadjuvant chemotherapy only renders 10% to 30% of these patients resectable.23 Therefore, nonsurgical options have emerged, such as radiofrequency ablation (RFA),24 arterial infusion,25 radioembolization,26 and cryoablation.27, 28

Stereotactic body radiotherapy applies the principles of stereotactic radiosurgery, long used for brain tumors,29 to discrete tumors outside the brain. Advances in image-guided technologies and treatment delivery have safely allowed treatment of tumors in the lung and liver with hypofractionation. Data for stereotactic body radiotherapy for early stage lung tumors have shown extremely impressive results, with 3-year local control rates ≥90%,30, 31 which is a dramatic improvement over historical local control rates of 30% to 50% using conventionally fractionated radiotherapy.32, 33 Multiple studies of liver stereotactic body radiotherapy have been reported showing local control rates ranging from 65% to 95% at 1 to 2 years.5-10

Indeed, problems arise when comparing stereotactic body radiotherapy studies because of patient and treatment heterogeneity. Referral patterns and selection criteria are different at each institution and, because these early studies included numerous tumor histologies, the results are difficult to apply to management of colorectal metastases. In addition, various dose-fractionation schedules have been used in these studies, ranging from 1 to 10 fractions, with different fraction sizes.

This study was undertaken to obtain outcome data on a large cohort of patients with colorectal liver metastases, and patients were pooled from multiple institutions to increase its statistical power. Although this method increased heterogeneity of clinical parameters because of differences in practice patterns and treatment techniques, variability in some factors such as dose fractionation and tumor size could be exploited for exploratory statistical analysis.

This study demonstrates that local control for colorectal metastases is dose-dependent, with a 18-month local control by lesion of 84% for total doses ≥42 Gy versus 43% for total doses <42 Gy. Overall, total dose (P = .0007), dose per fraction (P = .0019), and BED (P = .0001) were significant on univariate analysis. Separately analyzing each of these dose factors on multivariate analysis, all 3 factors significantly correlated with local control (Table 3), which is expected given their strong correlation with each other. By using BED to control for differences in dose fractionation schedules used in this study, the 18-month local control was found to be higher for schedules using ≥75 Gy compared with those using <75 Gy (80% vs 31%, P = .00001).

In addition, using tumor control modeling, we were able to demonstrate a dose-response relation between dose and local control by lesion at 12 months, as shown in Figure 4. On the basis of the best-fit curve, a BED of 117 Gy would be needed for a 90% local control rate. By comparison, Onishi et al. have showed that a BED ≥100 Gy provided a 92% local control rate for medically inoperable early stage lung cancer in a series with nearly twice the median follow-up (2 years compared with 1.2 years in the present series).34 The mechanism for this increased radioresistance is not well understood. It is possibly a difference in mean tumor volume between the 2 studies, but it also suggests a more aggressive nature of liver metastases because of selection of more aggressive clonogens capable of metastasizing and forming viable liver tumors. Nevertheless, this juxtaposition highlights the need for continued dose refinement to improve upon the results of liver stereotactic body radiotherapy to match its great success in lung malignancies while minimizing morbidity. Based on this analysis, a dose schedule of at least 48 Gy in 3 fractions or the equivalent should be considered.

The limitations of this analysis should be noted. First, there were differences among the participating institutions on how local control was defined. Second, lesion size likely played an important role in deciding the dose–fractionation schedules used in these patients. Patients and lesions that were treated with lower BED schedules had larger lesions, and GTV was significantly correlated with BED (Table 4). Certainly, this factor is an important confounding variable, because large lesions likely have higher baseline risks of local failure. However, the univariate analysis and multivariate analysis did not show that GTV size correlated with local control. This finding is in contrast to earlier publications from 2 of the participating institutions that suggested that local control is worse for larger lesions,8, 9 although they included many different tumor histologies in addition to colorectal adenocarcinomas. A study by Kang et al. looked at stereotactic body radiotherapy for metastatic colorectal cancer and found tumor volume to be the only factor associated with local control.35 However, this study included all sites of metastases, with most being in lymph nodes and only 17% of lesions being hepatic.

Table 4. Baseline Characteristics of Each Lesion Stratified by Dose Received
>42 Gy, n=51<42 Gy, n=51P>75 Gy BED, n=51<75 Gy BED, n=51P
  1. Gy indicates grays; BED, biologically effective dose; SBRT, stereotactic body radiotherapy; GTV, gross tumor volume.

Age, y6269.0036269.004
No. of prior chemotherapy regimens
 014 (27%)12 (23.5%).107113 (25.5%)13 (25%).0008
 120 (39%)12 (23.5%)24 (47%)8 (16%)
 ≥217 (33%)27 (53%)14 (27.5%)30 (59%)
No. of days of SBRT11 (3-17)12 (1-19).0186 (1-19)12 (1-19)<.0001
GTV, mL12 (0.6-236)90 (0.8-3088)<.000111 (0.6-212)90 (1.8-3088)<.0001

Tumor location is a crucial factor in determining how much liver, bowel, and chest wall can be spared and, depending on the situation, the fraction size may be modulated to minimize late toxicity. This study showed that BED ≥75 Gy was also correlated with improved local control, and this parameter may be a practical guideline that allows flexibility in tailoring the number of fractions and fraction size to avoid damaging critical structures. Certainly, as more data are reported on this topic, these dose recommendations can be further refined.

A more important concern to address is the impact of stereotactic body radiotherapy on survival. The ability to downsize unresectable disease to allow resection has been shown to improve survival,36 but it is not yet known whether stereotactic body radiotherapy can produce similar results. Neoadjuvant chemotherapy has become a standard of care to downsize disease for local therapy, address micrometastatic disease, and test tumor sensitivity to particular systemic agents. In the current study, the patient population was heavily pretreated systemically, with most patients (73%) having had at least 1 line of chemotherapy after the diagnosis of liver disease and almost half (42%) having had at least 2 lines. Therefore, we believe that this study cohort is an accurate reflection of the current treatment paradigm. The most intriguing finding is that maintenance of local control nearly significantly correlated with improved survival (P = .06), strongly suggesting that stereotactic body radiotherapy benefited these patients. Interestingly, the previously mentioned study by Kang et al. reported that a dose of >41 Gy, generally given over 3 fractions, was borderline associated with improved OS (P = .06), suggesting a dose response of stereotactic body radiotherapy on survival.35 Taken together, we conclude that stereotactic body radiotherapy should be considered as a locally ablative modality for liver metastases if surgical resection is not offered after chemotherapy.

Proper selection of patients for stereotactic body radiotherapy is not well defined, given the lack of data, but results from surgical data can offer guidance. Active extrahepatic disease has been shown in multiple surgical series to correlate with worse survival in patients who undergo liver resection.19, 37-39 In the current multivariate analysis, the results also showed that active extrahepatic disease was the only factor that correlated with survival (P = .04). The number of liver lesions has also been shown to be prognostic in some series19, 40-42 but not others.43 The current study limited patients to a maximum of 4 liver lesions, and having a single lesion versus 2 to 4 was not associated with survival. However, no conclusions can be drawn from this study regarding the maximum number of lesions suitable for stereotactic body radiotherapy.

Overall, we conclude that stereotactic body radiotherapy is an effective treatment option for colorectal liver metastases. Its efficacy is favorable when compared with RFA, a major nonsurgical ablative modality. A recent study by Otto et al looking at 28 colorectal cancer patients who underwent RFA of liver metastases showed a cumulative 32% local failure rate, a 12-month local control rate of 58%, and a 12-month rate of repeat hepatic treatment of 33%.24 A series by Solbiati et al observed local failures in 70 (39%) of 179 patients, with an 18-month local control rate of 56%.44 In another series reported by the same group, the investigators reported a local failure rate of 40%.45

Consistent with other stereotactic body radiotherapy series, treatment was generally well tolerated. Although late toxicity was infrequent, liver tolerance using hypofractionation is poorly understood, and investigators have erred conservatively by choosing fractionation schedules and dose volume constraints believed to minimize liver toxicity. In the current study, larger tumors were generally treated with longer courses of stereotactic body radiotherapy with lower BED out of necessity given the amount of normal liver that would be irradiated. These results taken with other stereotactic body radiotherapy studies showing few cases of symptomatic hepatotoxicity suggest there is likely room for dose escalation, particularly for larger tumors, which likely carry a worse local control. More studies on liver tolerance are needed.

Second, toxicities of adjacent nonliver organs, particularly the bowel, stomach, and chest wall, account for most late toxicities. Therefore, careful attention must be paid to these organs. The radiation tolerance of these organs remains unknown, and studies are beginning to address these issues.46-48


When treating colorectal liver metastases, liver stereotactic body radiotherapy is an effective ablative therapy with minimal toxicity. For a 3-fraction regimen, a dose of approximately 48 Gy is recommended when allowed by dose constraints from adjacent normal tissues. Patients without active extrahepatic disease have better OS than patients with active extrahepatic disease. The correlation between sustained local control and improved OS supports the clinical value of controlling hepatic disease even for heavily pretreated patients.


  1. Top of page
  2. Abstract
  6. Acknowledgements

We thank Christopher Morris, MS, for statistical support in this study and Jessica Kirwan for editing the manuscript.


  1. Top of page
  2. Abstract
  6. Acknowledgements