Evaluation of liver segmental dose threshold for hepatocyte regeneration following liver stereotactic body radiation therapy

Worldwide, hepatocellular cancer (HCC) is the fourth-ranking cancer-related mortality1. The age-adjusted incidence of HCC in India varies between 0.7 to 7.5 for men and 0.2 to 2.2 for women per 100,000 population per year. The age-standardized mortality rate for HCC is 6.8/100,000 and 5.1/100,000 for men and women, respectively in India2. Close to 80 per cent of HCCs occurring in India have a background of liver cirrhosis and about 60 per cent are hepatitis B carriers13. This association of chronic liver disease in almost all individuals with HCC presents a major challenge to treatment. Only a small fraction of these individuals are eligible for surgery, which thus far remains the ideal form of treatment4. The rest are treated with liver-directed therapies such as radiofrequency ablation (RFA), transarterial chemoembolization (TACE) and transarterialradioembolization5. Stereotactic body radiation therapy (SBRT) has made it possible to deliver conformal, high doses of radiation therapy to the tumour while sparing surrounding normal liver tissue. The use of SBRT upfront or as a consolidative treatment after TACE is associated with increased altogether survival6789. SBRT can also be used in a safe and effective way to treat large HCCs with or without prior treatments with good local control and survival10.

Radiation-induced liver disease (RILD) is a dose-limiting complication of radiotherapy (RT) to the liver, especially while treating large HCCs. Several studies have examined the risk factors of future RILD post-SBRT. Affected individuals with poor baseline Child–Pugh status, underlying hepatitis, cirrhosis and other adverse factors are more likely to do poorly post-RT111213. Applying strict dosimetric constraints to the normal, uninvolved liver helped decrease the risk of liver decompensation as shown in several studies, thus establishing a clear dose–response relationship with RILD891415. All the above studies showed damage of the uninvolved liver parenchyma due to chronic liver disease to be greater in the setting of HCC than in liver metastases.

In 2010, the Quantitative Analysis of Normal Tissue Effects in the Clinic (QUANTEC) recommended that the mean normal liver dose be kept <18 Gy and <6 Gy for a six-fraction SBRT in primary liver cancer with baseline Child A and B function, respectively, based on the SBRT experience from Toronto and Colorado1617. This was also the first time a cut-off value of 700 cc of normal liver receiving <15 Gy was conceptualized and proposed to reduce the incidence of non-classical RILD three months post-SBRT17. However, as individuals with HCC continue to live longer, there is a risk of developing long term hepatocyte dysfunction, especially in the setting of hepatitis and background liver dysfunction. Eovist-based functional magnetic resonance imaging (MRI) demonstrated that 20 Gy volume (delivered over three fractions) may be associated with loss of liver parenchymal function18. This may be associated with liver dysfunction and lack of hepatocyte regenerative potential in normal liver parenchyma receiving more than 20 Gy. There are presently limited data to suggest an impact of dose on changes in total liver or segmental volume following exposure to intermediate and high doses of radiation. Based on hepatocyte dysfunction, it is likely that after liver stereotaxy, some part of the liver will lose regenerative capacity in long term. This is particularly important as individuals continue to live longer with increasing need for salvage procedures for new lesions. Therefore, it is important to understand the dose thresholds that may allow for liver segment regeneration after liver stereotaxy. The present study was therefore designed to understand the impact of dose on overall and segmental liver volume after SBRT for large HCCs and liver metastases previously treated with SBRT10. The safety, efficacy and outcomes of SBRT in large HCCs have already been published by our group10. Our hypothesis was that there is a specific dose–volume response relationship at segmental level of the liver and that preferential sparing of these segments (similar to surgery) could facilitate hepatocyte regeneration.

Material & Methods

This study was carried out at the department of Radiation Oncology, a tertiary care centre after obtaining approval from the Institutional Ethics Committee.

Inclusion criteria: Sequential imaging and clinical datasets of individuals treated with liver SBRT between March 2015 and September 2018 were evaluated. The inclusion criteria for this study necessitated the use of SBRT for either primary or liver metastases and availability of at least one triphasic imaging dataset three months after SBRT treatment within the hospital digital imaging archival system.

Participant selection: For individuals eligible for the study, performance status [Eastern Cooperative Oncology Group (ECOG)], gender, Child–Pugh status at baseline and follow up, presence of portal venous thrombosis and previous treatment history (drug-eluting bead for TACE and sorafenib) were obtained at baseline and each follow up. The individual and tumour (dose and volume) characteristics were also obtained. The treatment plan was de-archived and sequential imaging datasets were imported into the Eclipse treatment planning system (TPS) (Varian Medical Systems version 13.5, Palo Alto, San Diego). The baseline images were then fused (liver to liver) with each serial image at follow up for better anatomical correlation in each case.

Liver volume delineation and dosimetry: Liver segmental contours were delineated slice by slice by two radiation oncologists on the baseline images and follow up scans using the contouring tools in the TPS. All phases (arterial, portal and venous) in each scan were cross-referenced while contouring liver segments to minimize errors. The liver segments were delineated as described by Couinaud19. This segmental classification described an eight-segment system, each having its own arterial and portal venous supply along with hepatic venous and biliary drainage. The liver is divided into left and right lobes by a line passing through the middle hepatic superiorly and the inferior venacava and gallbladder fossa inferiorly. These two lobes are divided into four sections each. The right lobe is divided into two anterior segments (V and VIII) and posterior sections (VI and VII). The left lobe is divided into two lateral sections (segment II and III) and two medial section (IVA and IVB). The vena cava and gallbladder were excluded from the segmental contours. The contours were reviewed independently by two other people: a radiation oncologist and an interventional radiologist and necessary modifications were incorporated. The liver and segmental contours at baseline and follow up encompassed the involved tumour (if applicable). The segmental contours were verified and finalized by a senior interventional hepatobiliary radiologist.

The serial volumetric measurements that included total liver volume, normal liver volume [total liver volume – gross tumour volume (GTV)] and volumes of each of the liver segments (I, II, III, IVa, IVb, V, VI, VII and VIII) were extracted from each of the image sets. An example of liver segmental delineation is depicted (Fig. 1;

Close

Fig. 1

(A-C) Delineation of liver segments at baseline. (D) The SBRT plan including the prescription dose (54 Gy/6#). (E) Loss in liver segmental volume between baseline (cyan) and follow up scan (pink). Also noted is peri-hepatic fluid accumulation secondary to liver decompensation on follow up. SBRT, stereotactic body radiation therapy; #, fractions.

Export to PPT

Segmental dose evaluation: All study participants were treated with SBRT and planned using rotational intensity-modulated radiation therapy or volumetric arc therapy on (i) Helical Tomotherapy (HI-ART II version 3.1, Tomotherapy Inc., Madison, WI, USA) or (ii) Linear Accelerator TrueBeam (Varian Medical Systems). Prescribed total dose, dose per fraction and total number of fractions were adopted for each patient based on tumour location, tumour size, amount of normal functional reserve volume and dose received by organs at risk (OARs). During planning, the mean dose of 700cc of normal liver was restricted to <15 Gy1617. Maximum dose to 0.5 cc of stomach, duodenum, small and large bowels and oesophagus was restricted to <30 Gy.

These delivered SBRT plans were then overlaid onto these segmental contours to compute the mean and median doses to total liver and normal liver volumes and mean doses to each individual segment at baseline.

Follow up assessment: Study participants were reviewed at six weeks post-SBRT for Child status assessment and thereafter every three months with a triphasic computed tomography (CT) scan liver and blood investigations for Child status. Toxicity was assessed using Common Terminology Criteria for Adverse Events (CTCAE version 5.0). Sorafenib was restarted 4-6 wk post-completion of SBRT when applicable. Response assessment was done using combined Response Evaluation Criteria in Solid Tumors (RECIST) criteria20.

Statistical analysis: Statistical analyses were performed using SPSS 19 (SPSS Inc., Chicago, IL, USA). Comparisons between baseline and follow up volume measurements on the same study participants were performed using the paired t-test. Pearson correlation was used to investigate the correlation of segmental liver dose with volume loss on follow up imaging. Furthermore, receiver operator’s curve (ROC) analysis was performed to find the segmental dose, i.e. predictive for liver volume loss; P<0.05 was considered as significant.

Results

Participants, tumour and treatment characteristics: In all, 21 study participants satisfied the inclusion criteria of the study. The majority of the participants were male (male:female ratio, 6:1) and the median age was 56 yr (range, 41-78 yr). Sixteen individuals had hepatocellular carcinoma and five had metastases with the following primary tumour: cervix (1), melanoma (2), neuroendocrine tumour (1) and intrahepatic cholangiocarcinoma (1). ECOG performances were less than two in all study participants. Among the HCC cases, 11 were Child–Pugh class A5, four were Child A6 and one was Child B7. All five liver metastasis cases were Child A5. Changes in underlying liver cirrhosis were found in most study participants with HCC; however, this was not quantified. None of the liver metastases individuals had any underlying cirrhosis. Five study participants had portal vein thrombosis. Seventeen participants received previous treatments including sorafenib, hepatic resection, RFA and TACE. The most common previous treatments were TACE (13/21) and sorafenib (12/21). The mean SBRT dose delivered was 40 Gy/6Fr (27.5 Gy-54 Gy/6 Fr determined by the V_eff irradiated and the risk of RILD)21. All study participants were treated with six fractions only. The median follow up in the cohort was seven months (4-35 months). Details about study participants, tumour and treatment characteristics are shown in Table. Fourteen (66.7%) of the cases were solitary liver lesions and seven were multicentric (33.3%). Among the multicentric lesions, 1 (4.8%) had two lesions, 1 (4.8%) had three lesions and 5 (23.8%) had more than three lesions. Median tumour size was 9.6 cm (5-21 cm).

Clinical outcomes: All study participants received the planned RT without interruption associated with treatment-related toxicity. The median follow up after SBRT was seven months (4-35 months). Two participants had complete response (CR), five showed partial response (PR), eight showed stable disease and six had progressive disease, respectively. Six study participants had disease progression, of which three were in-field progression and three were out-of-field progression. Grade 3 or greater treatment-related toxicity was reported in two participants on follow up. Three participants developed hepatic decompensation after receiving other salvage treatments like TACE, RFA, sorafenib and liver resection after SBRT. The scans obtained after progression were not included in the assessment of volume changes.

Liver volume changes: The number of follow up CT sets differed from individual to individual depending on the length of follow up. Sixteen of these participants had at least three sets of images for evaluation (one at baseline and two on FU). The mean interval between follow up scans was three months (1-22). The median baseline total liver volume was 1502.8 cc (855.2-2384.2 cc) and the median normal liver volume (total liver volume – GTV) was 1062.4 cc (675.2-1859.3 cc). The median tumour volume was 273.4 cc (23.1-630 cc). The median of the mean dose to normal liver was 13.5 Gy (12.5-15.5 Gy). Overall, 13 study participants showed loss of overall liver volume and eight showed gain of overall liver volume. The bar chart in Fig. 2 depicts the changes in overall and normal liver volume after SBRT. On follow up, 4/21 study participants had more than two-point change in Child score.

Close

Fig. 2

Change in (A) total and (B) normal liver volume (liver-GTV) on follow up scans after SBRT. GTV, gross tumour volume.

Export to PPT

A total of 182 liver segments were delineated in baseline and follow up scans. As tumour-containing segments and resultant segmental loss due to tumour shrinkage could impact the study results, only 140 non-tumour-containing segments were included in the final analysis. The median segmental volume at baseline was 127 cc (119-154 cc). After SBRT, 52 segments reported increase in liver volume and 82 segments reported loss in volume. No change in volume was noted in other segments. The median gain in segmental volume was 30 cc (0.6-375cc). This corresponded to an overall median gain in normal liver volume of 85 cc (7-213 cc) or 41 per cent (0.5-500%). The median (of the mean) dose in segments reporting an increase in volume was 9.1 Gy (7-36 Gy). The median volume reduction in segments that reported loss of segmental volume was −37 cc (−0.3 to −276 cc). This corresponded to −29 per cent loss in baseline volume (−1-−97%). Although there was a correlation between loss in volume with increasing segmental dose (Fig. 3A and B), the median (of the mean) dose in segments losing volume was 15.5 Gy (1-49 Gy). As the difference between the median dose received by segments with volume loss and volume gain is within narrow therapeutic window, we performed a bar plot and receiver operator characteristic analysis to understand the threshold value for volume loss. As seen in bar plot and ROC (Fig. 3C and D), 11 Gy segmental dose may represent the dose of no change in segmental liver volume, with higher doses resulting in loss of liver volume. As seen in bar plot, while planning liver SBRT, some segments should be preferentially spared to doses less than 9 Gy to allow regeneration. A higher absolute volume of segmental regeneration was seen with lower doses of radiation (1-6 Gy).

Close

Fig. 3

(A) A reduction in percentage segmental volume from baseline as a function of increasing segmental dose. (B) Change in absolute volume of liver segment as a function of dose. (C) Normal liver dose range for liver volume regeneration. As seen in the figure, no regeneration of segmental volume is seen beyond 10-11 Gy. Furthermore, higher doses are associated with volume loss as a function of increasing dose. (D) ROC analysis demonstrating sensitivity and specificity of various segmental dose levels in predicting loss in segmental liver volume. ROC, receiver operating characteristic curve.

Export to PPT

Univariate analysis: Univariate analysis was performed to see the impact of other factors on segmental volume. Baseline Child score (5 vs. 6), previous TACE (Yes vs. No), sorafenib use (Yes vs. No), portal vein thrombosis (Yes vs. No) and normal liver segmental dose (11 Gy or less than 11 Gy) were the imputed variables. Only liver segmental dose >11 Gy contributed to segmental volume loss (P<0.0001). No other factors were significant on univariate analysis and hence multivariate analysis was not performed.

Discussion

In the past few years, there has been growing evidence to show that hypofractionated RT (including SBRT) to the liver has contributed to morphological changes seen on follow up radiological scans2223. These findings have been correlated with microscopic veno-occlusive changes in segments close to the radiation field and, regeneration and hypertrophy in segments away from the field. These liver volume changes are similar to those that occur following hepatectomy, ablation, selective internal radiation therapy (SIRT), TACE and other focal therapies24. However, it must be noted that the pathogenesis across all these modalities is not identical and post-SBRT changes are primarily due to veno-occlusive disease2526. These changes are typically distributed around the high isodose regions of the isotropic non-coplanar dose distribution of SBRT27.

Many studies have shown that there is a greater risk of liver decompensation in indivudals with less volume of remnant liver after local therapies such as hepatectomy and SIRT28. Moreover, it has also been suggested that assessment of the functional status with a scoring system like Child–Pugh score in addition to volume measurements is important to predict future function28. An important drawback of our analysis is that degree of baseline cirrhosis was not quantified and only Child status was recorded in study participants. A cirrhotic liver most likely has a different regenerative pattern as compared to normal liver. Compensatory liver hypertrophy and degree of functioning liver post-SBRT would perhaps be greater in non-cirrhotic patients compared to those with underlying liver disease29. This was observed even in this study with a small sample size, with better volume gains in untreated, normal liver segments in the five metastasis participants compared to HCC (though not statistically significant). Future studies may have to incorporate pre-SBRT functional liver imaging in addition to quantifying uninvolved liver volume. In other local therapies, it is common practice to assess pre-treatment functional liver reserve (FLR) for participant selection, thus preventing post-procedure liver failure. Unlike surgery or TACE which essentially impacts the involved segments/lobe, intermediate to low dose spillage from SBRT to uninvolved liver segments may have impact on hepatocyte function which is not systematically investigated. Thus, it may be necessary to consider the liver segments as distinct entities that have both parallel (hepatocytes) and serial architecture (biliary drainage and vascular supply). Moreover, if the serial component of one or more segments is likely to be damaged, it would make sense to focus on optimally sparing the non-involved segments. Herfarth Type I-III reaction has been defined secondary to veno-occlusive effects and fibrosis in immediate vicinity of target volume for SBRT25. Radiological studies also report changes in CT density (7-10 HU) that corresponds to 30-35 Gy irradiated volume26. Radiological studies using MRI-specific contrast agents also demonstrate loss of hepatocyte regeneration with corresponding radiological changes in dose range of 19-20 Gy delivered over 3-6 fractions20. However, the impact of lower doses on hepatocyte regenerative capacity is not completely known. This may be of particular importance while treating large tumours or multiple intrahepatic tumours. While the critical volume approach suggests restricting mean dose of normal liver <15 Gy to <700 cc volume1718 this could be achieved by different dose volume histogram characteristics with different areas under the curve.

We observed close to 30 per cent liver segmental volume loss within the follow up imaging. This is similar to results from other liver SBRT studies. It is important to note that dose accumulation and volume studies were done based on liver-to-liver registration and no deformable registration software was used, and further validation studies with such a system in place may be needed in the future. This study had several limitations such as the small number of participants and the retrospective, exploratory design of the hypothesis. Second, there is a considerable difference in background liver architecture (and thus radiobiology) between HCC and liver metastases and this has not been accounted for in this study. Furthermore, we could not perform a univariate or multivariate analysis to compare outcomes in HCC and metastases as the number of participants with metastases were few. Another drawback of this study is that only 16 study participants had two follow up scans while five had only one follow up scan. Furthermore, the difference in time interval between scans for different study participants is also likely to have introduced a bias. There was also heterogeneity in the treatments before SBRT (e.g. TACE, sorafenib and RFA) which is a drawback for the study. However, the strength of the study lies in the rigorous SBRT protocol and follow up with diligent emphasis on serial volume mapping. While shorter median follow up and a lack of uniform imaging and scanning time point is a limitation, however, as a vast majority of the participants (76%) had large HCC with an expected median overall survival of about nine months and predisposition to early progression and hence image dataset exclusion, therefore the observations could be considered valid for this cohort30. Further expansion of this hypothesis in individuals with smaller tumours and a longer expected survival is warranted.

In the present study, we observed that the threshold of hepatocyte regeneration is <11 Gy and sparing few segments to even lesser doses (like <5-6 Gy) may result in liver regeneration (Fig. 3). This may be particularly of interest as survival with hepatic malignancies and oligometastatic disease increases, and there may be a need to offer salvage procedures or re-irradiation without placing individuals at risk of second procedure-related decompensation. The impact of lower doses of radiation like V5 Gy has also recently been found significant in a predictive model that imputes multiple parameters to investigate factors predicting for RILD30.

In addition to segmental dose, the study also investigated additional factors that may impact liver segmental regeneration such as Child status, previous procedures, use of sorafenib and vascular occlusion. However, none of these factors were identified to be significant. It is important to note here that sorafenib was stopped at least two weeks before the start of radiation therapy and was restarted four weeks after completion of SBRT where applicable to avoid any acute volume changes due to anti-VEGF effects of this drug as seen in some studies31. Although our observations are based on a relatively small sample size, we believe that the observations of the present study may have additional implications on normal liver (organ at risk) delineation and sparing. Based on the current work, it is proposed that while performing liver stereotaxy, especially for large tumours, liver segments should be individually delineated and 3-5 liver segments (equivalent to 400-650 cc or 30% liver volume) should be preferentially spared to <5-6 Gy while minimizing the mean dose to normal liver. This may have implications on treatment planning of SBRT and how a portion of liver rather than the entire normal liver may be considered a preferred avoidance volume. Similarly, based on the location of tumour, this goal may also be limited by vicinity to other organs at risk.

Overall, the present work demonstrates loss of segmental liver volume at segmental doses >11 Gy delivered during six-fraction SBRT. Lower segmental doses are associated with liver regeneration and efforts should be made to restrict limited segments to lower doses to facilitate liver volume regeneration after SBRT.