A phase I trial of stereotactic body radiation therapy (SBRT) for liver metastases

Schefter. IJROBP 2005;62:1371

Purpose: To determine the maximum tolerated dose (MTD) of stereotactic body radiation therapy (SBRT) for liver metastases. A multicenter Phase I clinical trial was conducted. Eligible patients had one to three liver metastases, tumor diameter <6 cm, and adequate liver function. The first cohort received 36 Gy to the planning target volume (PTV) in three fractions (F). Subsequent cohorts received higher doses up to a chosen maximum of 60 Gy/3F. At least 700 mL of normal liver had to receive a total dose <15 Gy. Dose-limiting toxicity (DLT) included acute Grade 3 liver or intestinal toxicity or any acute Grade 4 toxicity. The MTD was exceeded if 2/6 patients in a cohort experienced DLT.

Results: median aggregate gross tumor volume, 18 ml (range, 3–98 ml). Four patients had multiple tumors. No patient experienced a DLT, and dose was escalated to 60 Gy/3F without reaching MTD.

Conclusions: Biologically potent doses of SBRT are well tolerated in patients with limited liver metastases. Results of this study form the basis for an ongoing Phase II SBRT study of 60 Gy over three fractions for liver metastases.

Stereotactic body radiation therapy (SBRT) is an emerging treatment paradigm defined in recently published American Society for Therapeutic Radiology and Oncology (ASTRO) and American College of Radiology guidelines as a “treatment method to deliver a high dose of radiation to the target, utilizing either a single dose or a small number of fractions with a high degree of precision within the body”. SBRT has previously been labeled with an assortment of monikers such as extracranial radiosurgery and extracranial stereotactic radioablation. Potential indications for SBRT include a broad spectrum of tumor types and locations.

Liver metastases provide a logical opportunity for investigating the application of SBRT. Surgical resection of limited metastatic disease in the liver is known to be associated with favorable outcome in well-selected patients suggesting a clinical benefit from metastectomy in some patients. SBRT, if demonstrated safe and similarly effective, would be advantageous over surgery, being entirely noninvasive and also deliverable on an outpatient basis, with no requirement for anesthesia.

The few reported studies of SBRT for liver tumors have included both single-fraction and multiple fraction regimens. Herfarth and colleagues at Heidelberg University applied single-fraction SBRT to primary and metastatic liver lesions and safely escalated the dose from 14 to 26 Gy. Blomgren and colleagues at the Karolinska Institute administered 20–45 Gy in two to four fractions to a group of 17 patients with liver metastases, observing only one instance of serious toxicity possibly attributable to SBRT. Wulf and colleagues at the University of Wurzburg used fractionated SBRT, typically 30 Gy in three fractions, in 23 patients with solitary liver lesions and observed no Grade 3 or higher acute or late toxicity from treatment.

Our protocol design involves the use of multiple fractions rather than a single-fraction regimen and is similar to the strategy employed in the Indiana University Phase I trial of SBRT for medically inoperable lung cancer. The rationale for a multiple fraction regimen is based on several factors including the well-described putative advantages of fractionation including reoxygenation of otherwise radioresistant hypoxic tumor and redistribution of tumor clonogens into sensitive phase. This Phase I study was designed to establish the maximum tolerated dose (MTD) of a three-fraction regimen of liver SBRT for patients with limited liver metastases from solid tumors in anticipation of a subsequent Phase II study to evaluate the efficacy of SBRT in that setting.

Normal tissue dose constraints

To establish specified limitations on the dose to normal liver, we applied a strategy concordant with the so-called “critical volume model” for normal tissue injury originally proposed by Yaes and Kalend. Published reports have indicated that up to 80% of the liver can be safely removed in a patient with adequate liver function. To exercise an even greater level of caution in terms of the volume of normal liver tissue that must be preserved, estimating that a typical normal liver volume is approximately 2000 mL, we specified that a minimum volume of 700 mL or 35% of normal liver should remain uninjured by SBRT. The tolerance dose when the entire liver is irradiated has been demonstrated to be at least 33 Gy in 22 fractions. For a generic alpha/beta ratio estimate of 3 Gy and the assumption of no tumor proliferation during treatment, the biologically equivalent dose of this schedule of 33 Gy in 22 fractions is 49.5 Gy3 according to standard linear-quadratic modeling, ignoring any proliferation correction. An SBRT regimen of 15 Gy in three fractions would have a normal tissue biologically equivalent dose of 40 Gy3, safely below the maximum tolerable level. We therefore mandated that at least 700 mL of normal liver (entire liver minus cumulative GTV) had to receive at total dose less than 15 Gy.

Furthermore, at least 67% of the right kidney had to receive a total dose of less than 15 Gy in three fractions (5 Gy per fraction). The percent of total kidney volume (defined as the sum of the left and right kidney volumes) receiving 15 Gy total in three fractions (5 Gy per fraction) was required to be less than 35% of the total kidney volume. The maximum dose to any point within the spinal cord could not exceed 18 Gy total in three fractions (6 Gy per fraction), and the maximum point dose to the stomach or small intestine could not exceed 30 Gy total in three fractions (10 Gy per fraction).


Tolerance of liver SBRT

Results of the present Phase I study indicate that a stereotactic body radiation therapy (SBRT) dose of at least 60 Gy in three fractions may be safely administered to patients with one to three discrete liver metastases, as long as an adequate volume of normal liver tissue is spared from the high-dose region. Doses were escalated according to standard Phase I design, from 36 Gy to 60 Gy (in three fractions) in increments of 6 Gy per cohort. There was no Grade 3 or 4 toxicity; therefore, the MTD was not reached up to the predefined upper limit of 60 Gy.

In the liver SBRT experience reported from the Karolinska Institute, the single instance of serious toxicity occurred in a patient with a history of gastritis who experienced an exacerbation after SBRT that might have been prompted by the treatment. We did not observe any Grade 2 or higher bowel toxicity in the present study, but it should be emphasized that strict normal tissue dose limits were applied. It is conceivable that for some lesions on the medial surface of the liver, it might not be possible to restrict the maximum point dose to the stomach or intestinal tissue to 10 Gy or less. We would urge extreme caution in this setting, and we would not extend the observations from the present study to imply an understanding of the stomach or intestinal MTD with SBRT, because few patients had lesions that closely approached the gastrointestinal tract.

The critical volume model applied to the normal liver dose constraint in the present study of liver SBRT differs fundamentally from the Lyman normal tissue complication probability model that has been employed by others in dose selection for partial liver irradiation using conventional fractionation. Observations from the University of Michigan group have demonstrated that the Lyman normal tissue complication probability model is a good predictor of radiation-induced liver disease. Among a large series of patients treated with chemotherapy and partial liver irradiation, 19 of 203 (9%) suffered Grade 3 or higher radiation-induced liver disease (RILD), consistent with the Lyman model’s predictions. The mean dose to the liver was a robust predictor of RILD: no patient who received a mean total liver dose below 31 Gy suffered RILD, supporting a concept of liver tissue architecture as a parallel arrangement of functional subunits.

The lack of any RILD in the present series does not contradict these prior predictions and observations from series of conventionally fractionated liver irradiation in any way. We would attribute the lack of RILD after SBRT with a generally low mean dose to the uninvolved liver. The mean dose to uninvolved liver remained acceptably low even in the highest dose group (60 Gy in three fractions to the PTV), ranging from 3.3 to 23.9 Gy (median, 15.3 Gy)—all below the threshold expected to induce RILD.

Normal tissue effects: radiobiologic and clinical implications

Stereotactic body radiation therapy induced characteristic hypodense radiographic changes in regions of uninvolved liver, as originally described by Herfarth after single-fraction SBRT. The radiographic change did not correlate with any clinical endpoint, and there was no distortion of normal liver anatomy otherwise seen (i.e., no swelling of the liver and no displacement of liver vessels or bile ducts). Nevertheless, it is interesting to consider the effect in terms of the radiobiologic features, which might give a hint of the underlying etiology. In their analysis, Herfarth and colleagues identified a median threshold dose for the observed normal tissue reaction of 14 Gy at an interval of 2–3 months after SBRT. The approximate isoeffective dose in three fractions from our observations was 30 Gy in the cases analyzed in the present study.The preselected maximum dose level to be tested, 60 Gy in three fractions, was chosen because it is predicted to provide a very high rate of tumor control. Fowler and colleagues considered the question of how much dose is necessary to achieve a very high tumor control probability (TCP) with SBRT, primarily analyzing the use of SBRT for lung cancer, and calculated that a total dose of 60–69 Gy in three fractions (20–23 Gy per fraction) is necessary if there is even a small amount of hypoxia present. Predictive models are, of course, only as good as the baseline assumptions applied. For example, in the present series the mean total equivalent uniform dose given to the GTV in the 60 Gy-dose cohort was actually 72 Gy in three fractions, or 24 Gy per fraction, and the median GTV was 17 mL. If we apply the same assumptions for estimating TCP that we have previously employed, namely an estimated clonogen density of 107/mL and a Gaussian distribution of radiosensitivity across the tumor cell population, varying the estimated surviving fraction after 2 Gy from 0.4 to 0.5 will alter the predicted TCP for a 17-mL tumor from greater than 99% down to approximately 75%. It will only be through careful future observations of the actual control rates achieved with aggressive SBRT that we will be able to refine our estimates of the true radiosensitivity parameters in vivo and possibly also understand the influence of tumor hypoxia in this setting.