A Pilot Study of Pembrolizumab in Combination With Y90 Radioembolization in Subjects With Poor Prognosis Hepatocellular Carcinoma

Abstract Background Combination checkpoint inhibition therapy with yttrium-90 (Y90) radioembolization represents an emerging area of interest in the treatment of advanced hepatocellular carcinoma (HCC). HCRN GI15-225 is an open-label, single-arm multicenter, pilot study (NCT03099564). Methods Eligible patients had poor prognosis, localized HCC defined as having portal vein thrombus, multifocal disease, and/or diffuse disease that were not eligible for liver transplant or surgical resection. Patients received pembrolizumab 200 mg intravenously every 3 weeks in conjunction with glass yttrium-90 (Y90) radioembolization TheraSphere. Primary endpoint was 6-month progression-free survival (PFS6) per RECIST 1.1. Secondary endpoints included time to progression (TTP), objective response rate (ORR), overall survival (OS), and safety/tolerability. Results Between October 23, 2017 and November 24, 2020, 29 patients were enrolled: 2 were excluded per protocol. Fifteen of the remaining 27 patients were free of progression at 6 months (55.6%; 95% CI, 35.3-74.5) with median PFS 9.95 months (95% CI, 4.14-15.24) and OS 27.30 months (95% CI, 10.15-39.52). One patient was not evaluable for response due to death; among the remaining 26 patients, ORR was 30.8% (95% CI, 14.3-51.8) and DCR was 84.6% (95% CI, 65.1-95.6). Conclusion In patients with localized, poor prognosis HCC, pembrolizumab in addition to glass Y90 radioembolization demonstrated promising efficacy and safety consistent with prior observations (ClinicalTrials.gov Identifier: NCT03099564; IRB Approved: 16-3255 approved July 12, 2016).


Discussion
Liver cancer is the sixth most commonly diagnosed cancer and the third leading cause of cancer death worldwide; hepatocellular carcinoma (HCC) accounts for 75%-85% of such diagnoses. 1Early-stage HCC is amenable to locoregional treatments including surgical resection, radiofrequency ablation, transarterial chemoembolization (TACE), Y90 radioembolization, and liver transplantation. 2,3raditionally, locoregional therapies such as Y90 radioembolization have been limited to intermediate stage HCC. 4 In a randomized controlled trial involving 45 patients with early to intermediate stage HCC, Salem et al demonstrated that patients randomized to Y90 radioembolization had significantly improved median time to progression (TTP) but similar median overall survival (OS) compared with conventional TACE.6][7] High-risk features of localized poor prognosis HCC include portal vein thrombosis (PVT), multifocal disease, or macrovascular invasion (MVI).Patients with localized high-risk HCC treated with Y90 therapy had a median OS of 13.3 months for Child-Pugh A (CP-A) and 6.9 months for Child-Pugh B7 (CP-B7) patients. 6,7ore recently, developments in immunotherapy have introduced nivolumab, 8,9 nivolumab plus ipilimumab, 10 pembrolizumab, [11][12][13][14] atezolizumab plus bevacizumab, 15 and sintilimab plus a bevacizumab biosimilar 16 in the management of advanced HCC.This pilot study explores whether using immunotherapy drugs in combination with locoregional therapies for patients with non-metastatic HCC is safe and beneficial, using pembrolizumab every 3 weeks starting 1 week before initial Y90 radioembolization, as depicted in Fig. 1.Author disclosures and references available online.

Treatment
After enrollment, patients underwent mandatory pretreatment biopsy, followed by pembrolizumab 200 mg intravenous every 3 weeks.Glass Y90 radioembolization occurred 7-10 days after the first dose of pembrolizumab.In patients requiring a second dose of Y90 radioembolization for bilobar disease, a second dose was administered within 4 weeks of the initial procedure, after cycle 2 and at least 1 week prior to cycle 3 of pembrolizumab.Dosimetry of Y90 was performed for each individual patient at the discretion of the authorized user.Treatment ranged from a single segment to whole liver.Based on patient anatomy and disease burden, the number of sessions and number of separate radioembolization sites per session were determined.Dose was optimized to not exceed approved standards nor exceed acceptable doses.For single lobe treatments, a dose of greater than 130 Gy was often targeted.Patients received pembrolizumab every 3 weeks and one or 2 doses of Y90 radioembolization as described above for up to 24 months or until disease progression, unacceptable toxicity, study withdrawal, or death.Imaging was conducted every 9 weeks to assess for tumor response and disease progression.

Endpoints and Assessments
Primary criteria used for duration of response was Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1.As secondary criteria, disease was also evaluated using modified RECIST (mRECIST) for HCC.Adverse events were assessed using NCI Common Terminology Criteria for Adverse Events (NCI CTCAE) v5 during treatment, at 30 days after completion of treatment, and at 3-month intervals thereafter for up to 2 years.The primary endpoint was the proportion of patients alive and progression-free 6 months after treatment initiation (6-month progression-free survival, or PFS6), assessed according to RECIST 1.1 by imaging review.This study has not been powered for formal statistical testing.However, the precision of our study's 6-month PFS rate has been estimated in terms of the length of the half-width of the one-sided exact 95% confidence interval.We note that the distance from each proportion (p) to each lower bound is ~0.17 (with rounding).Published studies show that current treatment regimens lead to a median PFS of approximately 6 months (or p = 0.5 at 6 months).Clinically relevant improvements in median PFS would be an increase to 8, 9, or 10 months.In terms of 6-month PFS rates, this represents p = 0.59, 0.63, or 0.66, respectively (assuming an exponential distribution).We planned to accrue a total of 30 evaluable subjects at the rate of 2-3 per month for approximately 1 year.In particular, based on a one-sided one-arm log-rank test (Table 1), we can see the resulting powers at N = 30 and "Follow-up=24 months," for testing PFS of 6 months vs. PFS of 8, 9, 10 months, respectively, are all exceeded by N = 29 and "Follow-up=36 months."Secondary endpoints included time to progression (TTP; time from initiation of treatment to first documented disease progression); objective response rate (ORR; proportion of patients achieving confirmed complete response or partial response), assessed according to RECIST 1.1; overall survival (OS; time from initiation of treatment to death from any cause); and safety/tolerability.

Drug Information
Generic  3 and 4, Fig. 3).RECIST response data are included in Table 3 and subgroup analyses are similar across patient groups (Fig. 4).

Safety
In all treated patients, median treatment duration was 5.5 months (range, 20 days to 38 months).Treatment-emergent adverse effects were observed in 26 of 27 patients (96.3%); of these, 13 patients (48.1%) had adverse events grade 3 or higher by CTCAE v5.0 including lower lymphocyte counts (19%), elevated bilirubin (11%), and elevated liver function tests (7%).Common events were mostly of grade 1 or 2 in severity.Grade 5 hepatic failure was observed in one patient and deemed related to Y90.Details regarding treatmentemergent adverse effects are presented in Table 5.These reports further support our findings that the addition of pembrolizumab to Y90 radioembolization is a promising and reasonable therapeutic option for patients with poor prognosis HCC.Prior studies evaluating Y90 radioembolization alone in patients with poor prognosis HCC demonstrated a median OS of 13.3 or 6.9 months, 6,7 while our findings demonstrate that the median OS with the addition of pembrolizumab to be 27.3 months in our treated population.Other historical controls have demonstrated median OS 13.9-14.6months for pembrolizumab monotherapy 12,17 and 18.1 months for atezolizumab plus bevacizumab. 18While these studies cannot be compared, the large difference in median OS warrants further evaluation in a larger, randomized, 2-arm study.These findings are more striking when evaluating the intent to treat (ITT) population which demonstrated an OS of 30 months and similar rates of response, TTP, and PFS (Tables 6 and 7, Figs.5-9).These atypical findings can be explained by both patients (1024 and 1026) included in ITT but untreated by our regimen to be "long survivors" with 1024 censored at 24.9 months and 1026 at 22.3 month.Treatment regimens past progression were not collected in this study but given the established benefit checkpoint blockade in later settings, further evaluation in a larger study is warranted to determine if earlier treatment with checkpoint blockade similar to our study is beneficial to patient survival.Previous studies have shown that Y90 radioembolization has shown benefit in patients with early intermediate stage HCC, 19 but not in patients with localized, poor prognosis HCC. 6,7More recently, other studies have shown that immune checkpoint inhibition as monotherapy or as combination therapy can provide a clinical benefit for patients with advanced HCC.CheckMate-040 8,10 and CheckMate-459 9 demonstrated benefit of nivolumab or nivolumab plus ipilimumab in the treatment of advanced HCC.KEYNOTE-224 (11,13), KEYNOTE-240 (12), and KEYNOTE-394 ( 14) demonstrated efficacy of pembrolizumab monotherapy in advanced HCC.IMbrave150 demonstrated benefit of atezolizumab plus bevacizumab in locally advanced metastatic or unresectable HCC 15 ; ORIENT-32 examined sintilimab plus a bevacizumab biosimilar (IBI305) as first-line treatment for unresectable HBV-associated hepatocellular carcinoma. 16Systemic therapies in combination with Y90 radioembolization have been explored in patients with advanced HCC as well.In another recent single-center prospective study also involving patients with advanced HCC and Child-Pugh A cirrhosis, 36 patients were given Y90 radioembolization followed by nivolumab with 30.6% objective response rate, 20 similar to the 30.8% observed in our patients.Other historical controls demonstrate ORR 12.7%-18.3%for pembrolizumab monotherapy, 17 ORR 20% for nivolumab monotherapy, 8 and ORR 30% for atezolizumab plus bevacizumab. 18However, a multicenter study involving pembrolizumab in combination with Y90 radioembolization has not previously been performed.

Assessment, Analysis, and Discussion
The synergistic effect between radiotherapy and immunotherapy has been explored in recent preclinical and clinical models.The ability of stereotactic radiotherapy to induce endogenous antigen-specific immune responses significantly increased when combined with anti-PD-1 therapy in murine models of melanoma and breast carcinoma. 21In murine models of HCC, radiation upregulated PD-L1 expression in tumor models via IFN-γ/STAT3 signaling, resulting in significantly suppressed tumor growth and improved survival in mice receiving combination anti-PD-L1 and radiation therapy. 22stemic immune activation was described in analysis of the immune landscape of tumor-infiltrating lymphocytes from patients before and after receiving Y90 radioembolization. 23his was further elucidated in patients with intermediateadvanced HCC receiving Y90 radioembolization, in which tumor irradiation induced an altered adaptive and innate response that peaked 1 month after treatment and diminished significantly at 3 and 6 months.Moreover, many of the induced CD4+ and CD8+ T cells expressed high levels of inhibitory checkpoint markers PD-1 and LAG3, from which the investigators speculate that immune checkpoints administered after Y90 radioembolization would enhance both the immunological and clinical efficacy of irradiation. 24ur results suggest that pembrolizumab administration in combination with glass Y90 radioembolization was safe and tolerable.This combination also provided durable antitumor activity, promising PFS and OS in patients with advanced, poor prognosis HCC who had not previously received anti-PD-1, anti-PD-L1, or anti-CTLA-4 antibody therapy.These data demonstrating tolerability and improved OS suggest that patients with poor prognosis HCC may have similar risk stratification as metastatic patients warrant further evaluation.Median OS is 30 months with 95% CI (12.9, ---).The upper bound was not calculable due to insufficient follow-up for survival.

e407Figure 3 .
Figure 3. Waterfall plots for target lesion size change from baseline based on RECIST (top) and mRECIST (bottom).PFS and mPFS times are listed for all patients.Patients with right censored PFS or mPFS data are indicated with a ">" in front of the time.

Figure 4 . 3 e409Figure 5 .
Figure 4. Subgroup analyses of percent of patients progression free at 6 months (top) and objective response as a percent (bottom).

Figure 7 .
Figure 7.ITT OS curve: based on 29 patients in the ITT population.Median OS is 30 months with 95% CI (12.9, ---).The upper bound was not calculable due to insufficient follow-up for survival.

Figure 8 .
Figure 8. Waterfall plot based on RECIST in the ITT population.

Figure 9 .
Figure 9. Waterfall plot based on mRECIST in the ITT population.
mRECIST based on 24 patients: patients 1024 and 1026 are excluded, patients 1006 and 1031 have missing mRECIST, and patient 0007 has NE for mRECIST.Best mRECIST based on 25 patients: patient 0007 has PR for best mRECIST.

Table 1 .
Numeric results for the one-sided, one-sample Logrank test.

Table 6 .
ITT RECIST 1.1.Response is based on 27 patients because patients 1024 and 1026 were missing response data.