Stereotactic Body Radiation Therapy for Liver Metastases and Primary Hepatocellular Carcinoma: Normal Tissue Tolerances and Toxicity

Transcription

1 SBRT appears to be effective in treating hepatic metastasis, but further investigation is needed on its role for management of hepatocellular carcinoma. Wild Lettuce_2100. Photograph courtesy of Henry Domke, MD. Stereotactic Body Radiation Therapy for Liver Metastases and Primary Hepatocellular Carcinoma: Normal Tissue Tolerances and Toxicity Stephen M. Sawrie, MD, PhD, John B. Fiveash, MD, and Jimmy J. Caudell, MD, PhD Background: There is growing interest in stereotactic body radiation therapy (SBRT) as a noninvasive means of treating inoperable hepatic metastases and primary intrahepatic hepatobiliary carcinomas. While initial outcomes are encouraging, the safety of delivering such large, ablative doses is still being studied. Methods: We compiled all dose-volume constraints from seven prospective trials of liver SBRT and linked them to reported toxicities. Dose thresholds were made isoeffective, and grade 3 or higher toxicities for liver and adjacent normal tissues were correlated. Results: Four cases of grade 3-5 radiation-induced liver disease (RILD) were identified, including 1 treatmentrelated death, from all patients treated for metastasis. Three of these 4 cases were linked to excessive radiation doses in a large volume of liver. In 56 patients treated for hepatocellular carcinoma (HCC), 1 case of grade 5 RILD and 2 cases of grade 2 hepatic toxicity were reported. Additionally, a prominent retrospective series reported 3 cases of grade 5 RILD in 9 patients treated for HCC. Conclusions: SBRT appears to be safe for treatment of hepatic metastasis. The use of SBRT for HCC should be undertaken with caution or within the context of a clinical trial. Strict adherence to reported dose-volume constraints is advocated. Introduction Radiation hepatitis is a well-known complication of radiation therapy to the liver. First described in 1966, 1 the syndrome of radiation-induced liver disease (RILD) From the Department of Radiation Oncology at the University of Alabama Birmingham School of Medicine, Birmingham, Alabama (SMS, JBF), and the Department of Radiation Oncology at the University of Mississippi Medical Center, Jackson, Mississippi (JJC). Dr Sawrie is now with Gulf Coast Cancer Center, Daphne, Alabama. Submitted May 8, 2008; revised January 26, 2010; accepted January 28, Address correspondence to Stephen M. Sawrie, MD, PhD, Gulf Coast Cancer Center, Anchor Cross Boulevard, Daphne, AL ssawrie@gmail.com No significant relationship exists between the authors and the companies/organizations whose products or services may be referenced in this article. consists of rapid weight gain, anicteric ascites, hepatomegaly, an increase in abdominal girth, and a relative elevation in alkaline phosphatase. The pathophysiologic lesion is veno-occlusive disease caused by sinusoidal congestion in the region of the central lobules as well as obstruction of the smaller sublobular veins. Both are caused by the development of small intraluminal reticulin and collagen fibers as seen on electron microscopy. 2 Radiation tolerance to the liver has been well documented with conventional fractionation. Using a hyperfractionated regimen of 1.5 Gy twice daily, Russell et al 3 reported toxicity from cohorts receiving 27, 30, and 33 Gy to the whole liver. Whereas no patient receiving 27 Gy or 30 Gy developed significant liver toxicity, the actuarial rate of grade 3 or higher radiation April 2010, Vol. 17, No. 2 Cancer Control 111

2 hepatitis in those receiving 33 Gy was 10% at 6 months. In a compilation of the relevant literature at the time, Emami et al 4 reported in 1991 a 5% risk of liver failure within 5 years for 30 Gy to the whole liver in 2 Gy daily treatments. Partial irradiation tolerance doses are much higher. For instance, Dawson et al 5 treated 16 patients with liver metastases and 27 patients with primary intrahepatic hepatobiliary cancer using 3-D conformal techniques to a median dose of 58.5 Gy (28.5 Gy to 90 Gy) at 1.5 Gy per fraction twice daily. One incident of grade 3 RILD was reported, which was reversible, and no treatment-related deaths occurred. In a later study, Dawson et al 6 modeled normal tissue complication probabilities (NTCPs) for RILD within 4 months of conformal radiation therapy for treatment of liver metastases or intrahepatic hepatobiliary tumors. Their modeling revealed a strong volume effect, with mean dose to liver and NTCP significantly predicting the development of RILD on multivariate analysis. Other significant variables predicting a higher probability of RILD included primary liver disease (vs metastases) and male sex. It should be noted that these patients also received concurrent hepatic arterial chemotherapy, with bromo - deoxyuridine (vs fluorodeoxyuridine) also associated with a higher incidence of RILD. No cases of RILD occurred with mean liver doses less than 31 Gy. Over the past decade, interest has increased in capitalizing on advances in image guidance and computer optimization of treatment plans to deliver larger doses per fraction over a shorter time period in order to improve local control. These regimens invoke some component of stereotactic localization and immobilization, and they have been referred to either as extracranial stereotactic radiosurgery or, perhaps more aptly, as stereotactic body radiation therapy (SBRT). The first reported series of liver SBRT was a retrospective analysis of 31 consecutive patients treated at Karolinska Hospital in Sweden. 7 Twenty-three patients were treated for liver metastases (n = 14) or hepatocellular carcinoma (n = 9). Reference doses varied widely in this series, ranging from 8 Gy to 79 Gy in 1 to 4 fractions. Median follow-up was 10 months for patients with hepatocellular carcinoma (range, 1 to 38 months) and 9 months for patients with liver metastases (range, 1.5 to 23 months). Eleven of 17 metastases were controlled locally during this follow-up period. Of the 10 HCC lesions available for imaging surveillance, 5 remained unchanged, 3 diminished in size, and 2 demonstrated complete radiographic response. Results of prospective trials have also been impressive, with an 18-month actuarial local control rate as high as 93%. 8 Techniques for delivery of large fractions of radiation to the liver have been widely reported, including various means of body immobilization, stereotactic localization, image guidance, beam arrangements, and dosing techniques. 9 However, far less is known about tolerance of the liver or surrounding critical structures when using large fractions of radiation. The primary objective of this article is to provide the clinician with a resource to predict expected toxicity based on any SBRT dosing scheme of 1 to 5 large fractions to the liver. Methods The PubMed database was searched for all clinical trials in English that matched the terms stereotactic and liver. Trials had to report at least a dose-volume constraint for the liver and report subsequent acute or late liver toxicity. Reported constraints for any other neighboring structures were also compiled and organized by site. Primary endpoints were grade 3 or higher acute liver toxicity and grade 2 or higher late liver toxicity. If not specified by the authors, we ascribed toxicity grading according to the Radiation Therapy Oncology Group (RTOG) radiation morbidity scoring schema. 10 Secondary endpoints included similarly graded acute and delayed toxicities for all other reported neighboring organs. Constraints within each protocol were reported either as maximum point doses or as dose-volume constraints. The latter specifies volume of normal tissue to be relatively spared by constraining it to a certain threshold dose, and this can be reported several ways. We therefore converted the dose-volume constraints reported within each study to a standard reporting metric where the maximum dose threshold is set against a specified volume of normal tissue as follows: V Gy < volume (in cm 3 or as %) This metric can be read as the absolute or percentage volume of normal tissue that must be held within a specified dose threshold (in Gy). To place these dose thresholds in a common metric, we calculated the biologically equivalent dose (BED) for all constraints and single-fraction equivalent dose (SFED) for liver constraints. 11 The BED is extrapolated from the linear-quadratic (LQ) model of radiationinduced cell lethality, which assumes that radiation produces DNA double-strand breaks via a single radiation event. BED was calculated as follows: BED = nd (1 d/(α/β)) In this calculation, n equals the number of fractions and d equals the fraction size. The α component represents the linear portion of the cell survival curve, where a single radiation event (DNA double-strand break) causes cell death. The β component represents the quadratic portion of the cell survival curve, where cell death results from at least two double-strand breaks. The α/β ratio is the point where the α and β cell kill components are equal, with values ranging from 2 to 8, depending on the tissue (Tables 1 and 2). 12,13 BED has 112 Cancer Control April 2010, Vol. 17, No. 2

3 been used traditionally to compare various fractionation schemes for normal tissue and is reported units of Gy α/β to reflect the α/β ratio used in the calculation and to distinguish for actual dose. However, the formula for BED was derived from experimental models using more conventional fraction schemes. Recent reports suggest that the LQ model overestimates the radiation effect at large doses used in SBRT regimens. Park et al 14 reported an alternative cell survival curve referred to as the universal survival curve (USC), which hybridizes the low-dose region of the LQ model and the high-dose region of a competing model referred to as the multitarget model. When fitted to an experimental survival curve from a clonogenic assay of H460 NSCLC line, the sum of squares from the USC was substantially reduced relative to the LQ model fit. From the USC model, one can calculate a new parameter, the SFED, which may better represent cell survival from larger doses of radiation. It is calculated as follows: SFED = D (n 1) Dq In this instance, D equals total threshold dose and n equals the number of fractions. Dq is a parameter used to describe the shoulder of a cell survival curve in the multitarget model, and it essentially represents the relative radiosensitivity of a specific tissue. For this study we used the calculated Dq of 2.1 from the experimental data of Jirtle et al, 15,16 who estimated the values required for calculation of Dq in normal rat parenchymal hepatocytes via the method of least squares. 17 Results Table 3 presents all prospective clinical protocols meeting the criteria described above. All were phase I or II protocols. Sample sizes were modest, ranging from 17 to 39 for patients with liver metastases. Only four of the eight trials enrolled patients with HCC, with sample sizes ranging from 2 to 41 patients treated for HCC. Median follow-up ranged from 5.7 months to 4.3 years. Local control rates ranged from 67% to 93% at 18 months and from 61% to 79% at 24 months for liver metastases. Number of fractions per regimen ranged from 1 to 5. Méndez Romero et al 18 included several patients treated with 30 Gy in 10 fractions, but their study is not included in this report as it does not qualify as an accepted SBRT regimen delivered in 1 to 5 fractions. Single-fraction regimens ranged from 14 Gy to 26 Gy, whereas 3-fraction regimens ranged from total doses of 30 Gy to 60 Gy. One protocol treated several patients to 25 Gy in 5 fractions. Liver Table 1 sets forth all reported dose-volume constraints for liver. 8,18-24 Each constraint stipulated a dose and volume threshold. Each protocol provided a constraint to roughly one-third of normal liver tissue. Across all studies, threshold doses ranged from 7 Gy to 21 Gy for 30% to 33% of normal tissue. Studies by Kavanagh et al 8 and Schefter et al 20 used a critical volume constraint such that at least 700 cm 3 of normal liver had to be held under 15 Gy, assuming that an average liver would measure approximately 2,000 cm 3. Three studies also stipulated threshold doses to 50% of normal liver volume, with threshold doses ranging from 5 Gy to 15 Gy. However, the fractionation among these studies ranged from 1 to 5 fractions, making comparisons among studies difficult. Table 1 therefore includes these dose thresholds converted to common metrics. Using BED, threshold doses to roughly one-third of normal liver range from 12.4 Gy 3 to 70 Gy 3 (mean = 39.6 Gy 3 ) reflecting the biological influence of fractionation. Using the more intuitive parameter of SFED, the single-fraction equivalence Table 1. Summary of Dose-Volume Constraints for Liver With Conversion to Biologic Equivalent Dose (BED) and Single-Fraction Equivalent Dose (SFED) Study Dose-Volume Constraint Dose-Volume Constraint BED (Gy 3 ) SFED (as reported) (converted to V Gy ) Herfarth et al Gy to 30% V12 30% V60 30% V12 30% 7 Gy to 50% V7 50% V % V7 50% 1 fx 3 fx 1 fx 3 fx Wulf et al 21 D30 < 7 Gy V7 30% V % V % V7 30% V5 50% Wulf et al 23 D50 < 5 Gy V5 50% V % V7.8 50% V2.8 30% V0.8 50% Schefter et al cm 3 < 15 Gy V cm 3 V cm 3 V cm 3 Kavanagh et al 8 Hoyer et al Gy total < 30% V10 < 30% V21.1 < 30% V5.8 < 30% Méndez Romero et al 18 D33 < 21 Gy V21 33% V70 33% V % V % V % D50 < 15 Gy V15 50% V40 50% V30 50% V % V6.6 50% Tse et al 24 * mean dose < 22 Gy* N/A mean dose < 49.6 Gy 3 mean dose < 11.5 * This study determined liver dose constraint based on a previously reported normal tissue complication probability model described in the appendix. In this appendix it was noted that the constraint was usually mean dose < 22 Gy. April 2010, Vol. 17, No. 2 Cancer Control 113

4 Table 2. Summary of Dose-Volume Constraints for Organs at Risk With Conversion to Biologic Equivalent Dose (BED) Organs at Risk Study Dose-Volume Constraint Dose-Volume Constraint Biologic Equivalent Dose (as reported) (converted to V Gy ) Kidney Schefter et al 20 Right Kidney Right Kidney Right Kidney (α/β = 2) Kavanagh et al 8 V15 < 33% V15 < 33% V52.5 Gy 2 < 33% Total Kidney Total Kidney Total Kidney V15 < 35% V15 < 35% V52.5 Gy 2 < 35% Méndez Romero et al 18 D33 < 15 Gy V15 33% V52.5 Gy 2 33% V37.5 Gy 2 33% Spinal Cord Schefter et al Gy max 54 Gy 3 max (α/β = 3) Kavanagh et al 8 Hoyer et al Gy max 54 Gy 3 max Méndez Romero et al Gy max 40 Gy 3 max 30 Gy 3 max Tse et al 24 V27 < 0.5 cm 3 V27 < 0.5 cm 3 V67.5 Gy 3 < 0.5 cm 3 Stomach Herfarth et al Gy max 40.8 Gy 5 max (α/β = 5) 1 fx 3 fx Wulf et al 21 D100 < 7 Gy 16.8 Gy 5 max 10.3 Gy 5 max Wulf et al 23 Schefter et al Gy max 90 Gy 5 max Kavanagh et al 8 Méndez Romero et al 18 D5 cm 3 < 21 Gy V21 5 cm 3 V50.5 Gy 5 5 cm 3 V38.6 Gy 5 5 cm 3 Tse et al 24 V30 < 0.5 cm 3 V30 < 0.5 cm 3 V50 Gy 5 < 0.5 cm 3 Bowel Herfarth et al Gy max 30 Gy 8 max (α/β = 8) 1 fx 3 fx Wulf et al 21 D100 < 7 Gy 13.1 Gy 8 max 9 Gy 8 max Wulf et al 23 Schefter et al Gy max 67.5 Gy 8 max Kavanagh et al 8 Méndez Romero et al 18 D5 cm 3 < 21 Gy V21 5 cm 3 V39.4 Gy 8 5 cm 3 V32 Gy 8 5 cm 3 Tse et al 24 V30 < 0.5 cm 3 V30 < 0.5 cm 3 V48.8 Gy 8 < 0.5 cm 3 Esophagus Herfarth et al Gy max 63 Gy 4 max (α/β = 4) Méndez Romero et al 18 D5 cm 3 < 21 Gy V21 5 cm 3 V39.4 Gy 4 5 cm 3 V32 Gy 4 5 cm 3 Tse et al 24 V30 < 0.5 cm 3 V30 < 0.5 cm 3 V67.5 Gy 4 < 0.5 cm 3 1 fx 3 fx Duodenum Wulf et al 21 D100 < 7 Gy 13.1 Gy 8 max 9 Gy 8 max (α/β = 8) Wulf et al 23 Méndez Romero et al 18 D5 cm 3 < 21 Gy V21 5 cm 3 V39.4 Gy 8 5 cm 3 V32 Gy 8 5 cm 3 Tse et al 24 V30 < 0.5 cm 3 V30 < 0.5 cm 3 V48.8 Gy 8 < 0.5 cm 3 1 fx 3 fx Heart Wulf et al 21 D100 < 7 Gy 23.3 Gy 3 max 12.4 Gy 3 max (α/β = 3) Wulf et al 23 Tse et al 24 V40 < 0.5 cm 3 V40 < 0.5 cm 3 V130 Gy 3 < 0.5 cm Cancer Control April 2010, Vol. 17, No. 2

5 Table 3. Summary of Selected Liver SBRT Prospective Trials Study Sample Size Fractionation Prescription Method Follow-up Local Control Herfarth et al with mets Gy 1 Isocenter 5.7 mos ( ) 67% at 18 mos Wulf et al with mets 10 Gy 3 65% isodose line 9 mos (2 28) 61% at 24 mos Schefter et al 20 * 18 with mets Gy 3 80% 90% isodose line 7.1 mos ( ) NR 2 with HCC Wulf et al with mets 10 Gy 3 65% isodose line HCC: 15 mos (2 48) HCC: no local failure 5 with HCC 12.5 Gy 3 Mets: 15 mos (2 85) Mets: 66% at 24 mos 26 Gy 1 Kavanagh et al 8 * 21 with mets Gy 3 80% 90% isodose line 19 mos (6 29) 93% at 18 mos Hoyer et al with mets 15 Gy 3 Isocenter 4.3 yrs ( ) 79% at 24 mos Méndez Romero et al with mets 12.5 Gy 3 65% isodose line 12.9 mos (0.5 31) HCC: 40% at 24 mos 8 with HCC 5 Gy 5 Mets: 62% at 24 mos Tse et al with HCC/IHC 9 10 Gy 6 Unspecified 17.6 mos ( ) 65% at 12 mos * These two studies include many of the same patients. They are reported separately because Schefter et al 20 reported on 2 patients with HCC. IHC = intrahepatic cholangiocarcinoma, mets = metastasis, NR = not reported. of these same thresholds range from 2.8 Gy to 16.8 Gy (mean = 9.7 Gy). The study by Tse et al 24 reported a constraint in terms of mean dose, and its BED/SFED conversions are reported below. Despite these large fractions of radiation, acute and late liver toxicities were minimal. In fact, there were no significant acute or delayed liver toxicities in four of the studies. 8,19-21 Hoyer et al 22 reported one death from liver failure 7 weeks after SBRT. Despite their reported dose-volume constraint of V10 Gy < 30%, they note that 60% of the liver in this patient received greater than 10 Gy. Wulf et al 23 reported evidence of liver fibrosis, portal hypertension, ascites, and bleeding from esophageal varices at 28 and 41 months in 1 patient treated sequentially for 2 liver metastases close to the hilum. Three dosing schedules were used in this study, and it was not reported which was used to treat either of these lesions. Finally, Méndez Romero et al 18 reported grade 5 toxicity in 1 patient with HCC and associated cirrhosis and hepatitis B virus. Two other patients with HCC experienced grade 2 ascites with elevations in alkaline phosphatase, and they reportedly responded well to diuretic therapy. Two episodes of grade 3 toxicity with elevations in gamma-glutamyl transpeptidase (GGT) occurred among patients treated for liver metastasis in this study. In a study by Tse et al, 24 no grade 4-5 toxicities occurred in patients with HCC and intrahepatic cholangiocarcinoma (IHC). Stomach All but one study reported a constraint for stomach. Seven studies reported maximum doses ranging from 7 Gy to 30 Gy. BED ranged from 10.3 to 90 Gy 5. Méndez Romero et al 18 constrained 5 cm 3 of stomach to less than 21 Gy. The study by Hoyer et al 22 was the only one to report significant acute toxicities, with 2 patients experiencing grade 3 nausea. Twenty-one additional patients experienced grade 1 or 2 nausea. Maximum doses to stomach were not reported in these patients. Herfarth et al 19 reported 11 patients who experienced mild nausea or loss of appetite that resolved within 3 weeks. Wulf et al 21 stated that any acute nausea was treated symptomatically with anti - emetics or corticosteroids, but the authors did not report the number of patients. In the 2006 update, Wulf et al 23 reported simply that patients with targets in close proximity to the stomach were treated prophylactically with an H2 blocker or a proton pump inhibitor for 6 weeks (Table 2). Bowel and Duodenum Seven studies reported constraints for bowel, and four reported constraints for duodenum. The bowel constraints were identical to those reported for stomach but differed in terms of BED because of the α/β ratio of 8 for bowel. 8,18-21,23 Therefore, BED for bowel ranged from 9 Gy 8 to 48.4 Gy 8. For duodenum, Wulf et al 21,23 required that the entire volume be held below 7 Gy, whereas Méndez Romero et al 18 restricted 5 cm 3 to less than 21 Gy. Fractionation in these studies ranged from 1 to 5 fractions, with BED thresholds for duodenum ranging from 9 Gy 8 to 48.8 Gy 8. Hoyer et al 22 reported grade 3 diarrhea in 1 patient and grade 1 or 2 diarrhea in 14 additional patients. Colonic perforation that required surgery occurred in 1 patient, and 2 duodenal ulcerations were treated conservatively. Although this study did not stipulate a constraint for bowel or duodenum, a subsequent dosimetric analysis revealed that all 3 patients received total doses of 30 Gy (BED = 67.5 Gy 8 ) or higher to bowel or duodenum. April 2010, Vol. 17, No. 2 Cancer Control 115

6 Esophagus and Heart The esophageal constraint reported by Méndez Romero et al 18 included a dose-volume constraint of 5 cm 3 of esophagus to 21 Gy or less in 3 or 5 fraction regimens, whereas Herfarth et al 19 report only a maximum dose of 14 Gy in a single fraction. BED ranged from 32 Gy 4 to 63 Gy 4 depending on fractionation. One of the grade 5 toxicities reported by Méndez Romero et al 18 included bleeding esophageal varices, although this was likely due to liver dysfunction/toxicity because the patient had HCC with underlying cirrhosis (Child-Pugh B) and hepatitis B virus. No other significant esophageal toxicities were reported. The series by Wulf et al 21,23 stipulated a maximum dose of 7 Gy to the heart. No cardiac toxicities were reported in any study to date. The only other significant events included three instances of grade 3 skin toxicity. Kavanagh et al 8 reported that 1 patient in their study apparently began to develop skin erythema and pain in an area corresponding to a beam entrance approximately 6 months after treatment. This eventually required debridement and hyperbaric oxygen therapy. After a review of the dosimetry in this case, a 48-Gy hotspot was noted in the subcutaneous tissues underlying the skin breakdown. Hoyer et al 22 reported two instances of grade 3 skin toxicities with no associated dosimetric information and seven additional grade 1-2 skin toxicities. Finally, Schefter et al 20 reported two episodes of grade 1 dermatitis. There were also isolated instances of constitutional symptoms such as diffuse analgesias, fatigue, and fever that were either self-limiting or easily controlled medically. Finally, Herfarth et al 19 reported two episodes of singultus in patients treated for lesions near the diaphragm. Kidney and Spinal Cord Only three studies reported constraints for kidney. Schefter et al 20 and Kavanagh et al 8 reported dose-volume constraints for the right kidney as well as total kidney volume as presented in Table 1. Méndez Romero et al 18 constrained 33% of kidney volume to less than 15 Gy. Taken together, BED ranged from 37.5 Gy 2 to 52.2 Gy 2 to approximately one-third the volume of a single kidney. There were no reported kidney toxicities in these studies or in others that did not specify constraints. Four studies reported constraints for the spinal cord in the form of maximum dose. Volume of cord was not specified. Threshold doses ranged from 15 Gy to 18 Gy to the cord, which translated into 30 Gy 3 to 54 Gy 3 when accounting for the fact that regimens ranged from 1 to 5 fractions. There were no significant spinal cord toxicities reported in these studies. Discussion In earlier studies of radiation to liver malignancies using more conventional fractionation, dose to the entire liver has been limited to approximately 30 Gy due to subsequent hepatic toxicity. Local control using doses of such modest magnitude have been poor, leading to the conclusion that radiation therapy for liver metastases or primary liver malignancies was not effective. However, technologic advances leading to 3-D conformal radiation therapy led to a dose escalation trial of more limited liver volumes in patients with either hepatic metastases or primary intrahepatic hepatobiliary malignancy. 5 Treatment consisted of twicedaily treatment at 1.5 Gy per fraction (median dose = 58.5 Gy), and patients were also treated concurrently with intra-arterial chemotherapy. Nevertheless, significant liver toxicity was extremely limited, with only 1 case of reversible RILD. The response rate was 68%. Perhaps most importantly, the escalated radiation dose was independently and significantly associated with improved progression-free and overall survival; the median survival in the cohort receiving 70 Gy or more had not been reached at the time of publication. Results of this study renewed interest in utilizing radiation therapy in this setting. At the same time, technological advances in patient immobilization and imageguided radiation therapy (IGRT) have led to a growing interest in SBRT. The primary goal of SBRT is to noninvasively deliver larger, ablative tumor doses while minimizing dose to surrounding normal liver tissue and neighboring organs at risk. This is accomplished primarily through the use of a multiple noncoplanar beam arrangement with a tight margin on gross disease. These margins are typically on the order of millimeters and are accomplished via whole body immobilization, stereotactic localization, and image guidance. IGRT also allows realtime tumor tracking, and other modalities such as respiratory gating can control beam-on times in order to deliver radiation at specified times within the respiratory cycle, thereby further controlling for tumor motion. These technologies allow for additional reduction in the size of the radiation volume needed to effectively treat the tumor; this may further reduce toxicity to normal tissue. Logically, these considerations are critical when delivering such high doses to a tumor with surrounding critical organs. As noted above, Blomgren et al 7 and Lax et al 25 pioneered SBRT for liver tumors. Since then, several prospective phase I and II trials have used various SBRT dosing schemes to study the safety and efficacy of using ablative doses of radiation to treat liver malignancies. In this paper, we have examined the dose-volume constraints reported by each of these prospective trials and have attempted to link these constraints to reported toxicity. The field of radiation oncology has spent the past 8 decades studying the therapeutic window created by fractionation, gravitating toward fraction sizes in the 1.8 Gy to 2.0 Gy range in order to minimize normal tissue toxicity without significantly sacrificing tumor 116 Cancer Control April 2010, Vol. 17, No. 2

7 control. Therefore, little is known about normal tissue tolerance at the larger doses used in SBRT regimens. A compilation of dose-volume constraints and associated toxicities reported in liver SBRT protocols provides an initial database on which the clinician can draw to evaluate not only hepatic toxicities with these regimens, but also toxicity to nearby critical structures. Perhaps the most important conclusion from our review of these protocols is that SBRT regimens with their associated ablative fraction sizes are safe, at least in the treatment of liver metastasis. Of the patients with liver metastases from the seven studies evaluated in this paper (excluding overlap in related series), grade 3 or higher RILD (crude rate of 2.4%) was reported in only 4 patients, and 2 of these cases were reversible. One patient died within 7 weeks of treatment, and another case was within the context of progressive disease. One case of late grade 3 portal hypertension was reported. Two of the RILD cases were from one study using SFED thresholds as high as 16.8 Gy in a 3-fraction regimen, 18 and subsequent analysis of dose to normal liver in these patients revealed that only 638 cm 3 and 639 cm 3, respectively, were held to less than 15 Gy. One death occurred in a patient who received doses in excess of 10 Gy to 60% of the liver. 22 Hepatic toxicity in patients with HCC was perhaps more substantial. One case of grade 5 RILD occurred among 56 patients reported in four series. Although this patient had underlying liver disease (Child-Pugh B with hepatitis B virus and cirrhosis), the volume of liver that received over 30 Gy was only 6%, and the median liver dose was only 3.4 Gy. Two other patients with HCC experienced grade 2 ascites with elevations in alkaline phosphatase, although these patients responded well to diuretic therapy. Although results from the series by Blomgren et al 7 are not included in Table 3 as a prospective study, it is informative to note that there were 3 cases of grade 5 RILD among 9 patients treated for HCC in this study. The dose schedules used in these cases were as high as 48 Gy in a single fraction but as low as only 39 Gy in 3 fractions. Prior evidence suggests that patients with underlying liver disease from cirrhosis or hepatitis B virus have a higher chance of developing RILD. 26,27 Furthermore, primary intrahepatic disease (vs metastasis) was a significant predictor of RILD in the multivariate analysis by Dawson et al. 6 However, it should be noted that there were no grade 4-5 toxicities reported by Tse et al 24 using a 6-fraction regimen. Taken together, it would appear that the toxicity associated with SBRT for liver metastases is rather minimal. It is noteworthy that the 2 cases of reversible RILD held only 638 and 639 cm 3 of normal liver under 15 Gy, which is less than the 700 cm 3 critical volume threshold advocated by some investigators. 8,20 Strict adherence to any of the reported dose-volume constraints should be practiced, as our review highlights significant toxicity when these constraints are violated. On the other hand, SBRT for treatment of primary intrahepatic malignancy in the context of underlying liver disease should be undertaken with caution and should perhaps await further prospective clinical data. An active protocol has been specifically designed to address the maximum tolerated dose of a limited number of large fractions in patients with HCC. 28 Child- Pugh A patients are treated with a 3-fraction regimen, whereas Child-Pugh B patients are treated in 5 fractions. The study is scheduled to accrue 60 patients, with a scheduled completion date of December Perhaps the most significant nonhepatic toxicity reported in the studies we reviewed was bowel perforation and duodenal ulceration. However, subsequent dosimetric evaluation revealed maximum doses greater than 30 Gy in each of these cases. Several cases of grade 3 nausea were reported, reflecting the need to constrain dose to the stomach. Prophylactic antiemetics, H2 blockers, or proton pump inhibitors can also reduce the incidence and severity of this toxicity. Finally, it should be noted that there were 3 cases of grade 3 skin toxicity. While no dose-volume constraint to skin or subcutaneous tissue was reported in any of the seven studies reviewed here, we recommend reviewing these doses carefully, with specific attention to location of hotspots. The increasing use of multiple noncoplanar beam arrangements in SBRT plans heightens the possibility of increased tangentiality, particularly in peripheral lesions, and thereby increases the possibility of high-dose build-up in skin and subcutaneous tissue. However, by increasing the number of beams, usually above 10, hotspots may be reduced, as the dose is further spread among entry and exit sites of the beams. Although not a focus of this paper, it is worth noting that the reported 18- to 24-month actuarial local control data presented in Table 3 are encouraging, ranging from 67% to 93% at 18 months and 61% to 79% at 24 months for metastases. As a noninvasive technique, SBRT can treat lesions where other modalities cannot. For instance, SBRT can be utilized for lesions near vasculature that might serve as a heat sink for radiofrequency ablations (RFAs). Lesions near the hilum are often considered inoperable due to proximity to vasculature and are often difficult to access surgically for resection or RFA. Chemoembolization also has infrequent but significant complications such as tumor rupture and pancreatitis. 29,30 Therefore, the combination of nominal toxicity, expanded efficacy, and noninvasiveness without some of the technical and anatomic contraindications of other modalities warrants continued investigation into SBRT for liver malignancies. Clinical outcomes are already being reported anecdotally and can be appreciated by the following 2 cases treated at our institution. The first case involves a 42-year-old woman diagnosed originally with colon cancer who April 2010, Vol. 17, No. 2 Cancer Control 117

8 subsequently developed liver and lung metastases approximately 2 years after resection and adjuvant chemotherapy. She was started on a metastatic chemotherapy regimen and demonstrated stability or regression of disease in her pulmonary metastases. Unfortunately, her liver metastasis was progressive despite treatment, and the decision was made to treat this lesion with SBRT to a total dose of 45 Gy in 3 fractions. Fig 1A-C show contrasted computed tomography (CT) scans prior to SBRT, the dosimetry from the treatment plan used to treat the lesion, and a nearly complete radiographic response 6 months following SBRT. Fig 2A-C show similar CT scans from a 20-year-old man who developed HCC within the context of hepatitis B A A B B C Fig 1A-C. (A) Contrasted CT demonstrating lesion in the anterior segment of the right hepatic lobe. (B) Dosimetry from a 9-field treatment plan with 95% isodose line in red, 80% in blue, and 50% in green. (C) Follow-up CT 6 months after SBRT demonstrating near complete resolution of lesion as well as a hypodense area consistent with the 50% isodose coverage. C Fig 2A-C. (A) Contrasted CT demonstrating lesion in segment 8 adjacent to the middle hepatic vein. (B) Dosimetry from the 7-field treatment plan with 95% isodose line in red, 80% in blue, and 50% in green. (C) Follow-up CT 7 months after SBRT demonstrating complete resolution of lesion as well as a hypodense area consistent with the 50% isodose coverage. 118 Cancer Control April 2010, Vol. 17, No. 2

9 reportedly acquired from a blood transfusion at birth. At the time of diagnosis he had one lesion identified in segment V and a smaller lesion in segment VIII. The segment V lesion was resected, whereas the smaller segment VIII lesion was treated with alcohol ablation due to its proximity to vasculature. Serum α-fetoprotein level fell from a presurgical value of 4,132 ng/ml to 143 ng/ml postoperatively. Unfortunately the patient developed a recurrence at this site, with AFP rising to a high of 852 ng/ml (Fig 2A). At this point it was decided to pursue SBRT to treat this recurrence. A dose of 60 Gy in 3 fractions was prescribed (Fig 2B for dosimetry). Within 3 months of treatment the AFP had fallen to ng/ml, and within 7 months the lesion was not visible on CT (Fig 2C). The AFP has not exceeded 3 ng/ml over the past 14 months. It should be noted that the dose-volume constraints from Schefter et al 20 were met in both patients, and neither patient developed significant acute or late hepatic or nonhepatic toxicity. Conclusions Toxicity associated with SBRT for liver metastases was minimal in the eight prospective studies outlined in this article, whereas use of this technique in the treatment of primary intrahepatic hepatobiliary malignancy with significant underlying liver pathology should be exercised with caution. Further prospective data are being accrued in this latter group of patients and should further assist in informing dose-volume constraints and maximum tolerated dose. Ongoing attention also needs to be directed toward the most conservative dose-volume constraints for neighboring critical structures. Finally, it should be noted that Dawson et al 6 utilized dose-volume histogram information to develop an NTCP model that in turn could be used to individualize liver dose-volume constraints based on clinical factors specific to each patient. These efforts are required in order to more fully clarify safe dose-volume constraints in a potentially exciting new modality for patients with liver metastases and primary hepatobiliary malignancy. References 1. Reed GB Jr, Cox AJ Jr. The human liver after radiation injury: a form of veno-occlusive disease. Am J Pathol. 1966;48(4): Lawrence TS, Robertson JM, Anscher MS, et al. Hepatic toxicity resulting from cancer treatment. Int J Radiat Oncol Biol Phys. 1995;31(5): Russell AH, Clyde C, Wasserman TH, et al. Accelerated hyperfractionated hepatic irradiation in the management of patients with liver metastases: results of the RTOG dose escalating protocol. Int J Radiat Oncol Biol Phys. 1993;27(1): Emami B, Lyman J, Brown A, et al. Tolerance of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys. 1991;21(1): Dawson LA, McGinn CJ, Normolle D, et al. Escalated focal liver radiation and concurrent hepatic artery fluorodeoxyuridine for unresectable intrahepatic malignancies. J Clin Oncol. 2000;18(11): Dawson LA, Normolle D, Balter JM, et al. Analysis of radiationinduced liver disease using the Lyman NTCP model. Int J Radiat Oncol Biol Phys. 2002;53(4): Blomgren H, Lax I, Näslund I, et al. Stereotactic high dose fraction radiation therapy of extracranial tumors using an accelerator: clinical experience of the first thirty-one patients. Acta Oncol. 1995;34(6): Kavanagh BD, Schefter TE, Cardenes HR, et al. Interim analysis of a prospective phase I/II trial of SBRT for liver metastases. Acta Oncol. 2006;45(7): Potters L, Steinberg M, Rose C, et al. American Society for Therapeutic Radiology and Oncology and American College of Radiology practice guideline for the performance of stereotactic body radiation therapy. Int J Radiat Oncol Biol Phys. 2004;60(4): RTOG/EORTC Late Radiation Morbidity Scoring Schema. Radiation Therapy Oncology Group Web site. late.html). Accessed January 28, Brenner DJ. The linear-quadratic model is an appropriate methodology for determining isoeffective doses at large doses per fraction. Semin Radiat Oncol. 2008;18(4): Hall EJ. Radiobiology for the Radiologist. 5th ed. Philadelphia, PA: Lippincott Williams & Wilkins; Gay HA, Niemierko A. A free program for calculating EUD-based NTCP and TCP in external beam radiotherapy. Phys Med. 2007;23(3-4): Park C, Papiez L, Zhang S, Story M, et al. Universal survival curve and single fraction equivalent dose: useful tools in understanding potency of ablative radiotherapy. Int J Radiat Oncol Biol Phys. 2008;70(3): Jirtle RL, Michalopoulos G, McLain JR, et al. Transplantation system for determining the clonogenic survival of parenchymal hepatocytes exposed to ionizing radiation. Cancer Res. 1981;41(9 Pt 1): Jirtle RL, McLain JR, Strom SC, et al. Repair of radiation damage in noncycling parenchymal hepatocytes. Br J Radiol. 1982;55: Clifton KH, Draper NR. Survival-curves of solid transplantable tumour cells irradiated in vivo: a method of determination and statistical evaluation; comparison of cell-survival and 32-P-uptake into DNA. Int J Radiat Biol Relat Stud Phys Chem Med. 1963;7: Méndez Romero A, Wunderink W, Hussain SM, et al. Stereotactic body radiation therapy for primary and metastatic liver tumors: a single institution phase I-II study. Acta Oncol. 2006;45(7): Herfarth KK, Debus J, Lohr F, et al. Stereotactic single-dose radiation therapy of liver tumors: results of a phase I/II trial. J Clin Oncol. 2001;19(1): Schefter TE, Kavanagh BD, Timmerman RD, et al. A phase I trial of stereotactic body radiation therapy (SBRT) for liver metastases. Int J Radiat Oncol Biol Phys. 2005;62(5): Wulf J, Hädinger U, Oppitz U, et al. Stereotactic radiotherapy of targets in the lung and liver. Strahlenther Onkol. 2001;177(12): Hoyer M, Roed H, Traberg Hansen A, et al. Phase II study on stereotactic body radiotherapy of colorectal metastases. Acta Oncol. 2006;45(7): Wulf J, Guckenberger M, Haedinger U, et al. Stereotactic radiotherapy of primary liver cancer and hepatic metastases. Acta Oncol. 2006; 45(7): Tse RV, Hawkins M, Lockwood G, et al. Phase I study of individualized stereotactic body radiotherapy for hepatocellular carcinoma and intrahepatic cholangiocarcinoma. J Clin Oncol. 2008;26(4): Lax I, Blomgren H, Näslund I, et al. Stereotactic radiotherapy of malignancies in the abdomen. Methodological aspects. Acta Oncol. 1994; 733(6): Liang SX, Zhu XD, Xu ZY, et al. Radiation-induced liver disease in three-dimensional conformal radiation therapy for primary liver carcinoma: the risk factors and hepatic radiation tolerance. Int J Radiat Oncol Biol Phys. 2006;65(2): Cheng JC, Wu JK, Lee PC, et al. Biologic susceptibility of hepatocellular carcinoma patients treated with radiotherapy to radiation-induced liver disease. Int J Radiat Oncol Biol Phys. 2004;60(5): Dose Escalation Study for Primary Hepatocellular Carcinoma (SBF- HCC) ( identifier NCT ). Accessed January 28, Battula N, Srinivasan P, Madanur M, et al. Ruptured hepatocellular carcinoma following chemoembolization: a western experience. Hepatobiliary Pancreat Dis Int. 2007;6(1): López-Benítez R, Radeleff BA, Barragán-Campos HM, et al. Acute pancreatitis after embolization of liver tumors: frequency and associated risk factors. Pancreatology. 2007;7(1): April 2010, Vol. 17, No. 2 Cancer Control 119