CALGB 80802 (Alliance): Impact of Sorafenib with and without Doxorubicin on Hepatitis C Infection in Patients with Advanced Hepatocellular Carcinoma

Sorafenib efficiently blocks nonstructural protein 5A (NS5A)-recruited c-Raf–mediated hepatitis C virus (HCV) replication and viral gene expression (1). Sorafenib a highly specific c-Raf inhibitor and mediated antiviral effects were demonstrated to negate HCV replication and viral gene expression (2). In HCV-replicating cells, sorafenib also decreased the hyperphosphorylated form of NS5A that led to the formation of added hypophosphorylated forms. In comparison, inhibition of targets downstream of c-Raf or inhibition of other tyrosine kinase inhibitors like sunitinib did not affect HCV replication. HCV infection of Huh-7 cells has been shown to induce VEGF expression that depolarizes hepatocytes and disrupts tight junctions, thereby making cells more susceptible to further HCV infection in an autocrine fashion (3). Sorafenib treatment inhibited this cycle of events. These data highlight a potential role for VEGF antagonists to treat HCV infection in addition to their intrinsic properties of regulating hepatocellular carcinoma (HCC) growth and development.

Release of Raf-1-Ask-1 dimer and inhibition of Raf-1 via sorafenib putatively differ in the presence or absence of doxorubicin. Doxorubicin-induced cytotoxicity is mediated by the proapoptotic ASK-1 pathway. Basic FGF activity has been associated with expression of Raf-1, which binds to and neutralizes ASK1, preventing the cell from undergoing apoptosis. Inhibition of Raf-1 by sorafenib would be expected to restore chemosensitivity to doxorubicin (4). Potential effects of Raf onto the sensitivity of cells to doxorubicin and in regulation of the multidrug resistance-1 (mdr-1) gene led to the consideration of overcoming doxorubicin resistance by combining it with anti-Raf (5, 6). This led to a phase I study evaluating the addition of doxorubicin to sorafenib. Four patients with HCC achieved prolonged stability of disease and continued therapy for more than 1 year. Despite a 21% AUC increase of doxorubicin when administered concomitantly with sorafenib, no substantial worsening of toxicity over what would be expected from either compound administered individually (7). Collectively, the above led to a positive randomized phase II clinical trial of doxorubicin plus sorafenib versus doxorubicin (8). The finding of greater median time to progression, overall survival (OS), and progression-free survival (PFS) with sorafenib plus doxorubicin compared with doxorubicin monotherapy led to Cancer and Leukemia Group B (CALGB) 80802, a randomized phase III trial that evaluated doxorubicin and sorafenib combination versus sorafenib in patients with advanced HCC (9). CALGB is now part of the Alliance for Clinical Trials in Oncology (Alliance). In addition, an analysis of prognostic factors of another two randomized phase III studies of sorafenib alone suggested better survival in patients with HCV albeit no correlative analyses were performed (10). Patients were accrued from 2010 to 2015, which predated the availability of anti-HCV therapies.

Despite the pretrial hypothesis of a favorable drug–drug interaction, CALGB 80802 did not show an improvement in median OS with the addition of doxorubicin to sorafenib compared with sorafenib alone. Median OS was 9.3 months [95% confidence interval (CI): 7.3–10.8 months] in the doxorubicin plus sorafenib arm and 9.4 months (95% CI: 7.3–12.9 months) in the sorafenib alone arm (HR, 1.05; 95% CI: 0.83–1.31). Anticipating the importance of HCV in the outcome of patients with advanced HCC, preplanned study objectives included an analysis of the possible antiviral effect of therapy on HCV, perhaps above and beyond effects on the cancer itself. The main objective of the correlative study reported herein was to evaluate HCV titers and change with treatment and to examine the correlation with patient outcome.

Study investigators obtained written informed consent from all subjects. The studies were conducted in accordance with the Declaration of Helsinki, and they were approved by the Institutional Review Board (IRB) at all sites and/or the NCI central IRB and was registered (ClinicalTrials.gov NCT01015833). This unblinded randomized phase III clinical trial was led by Alliance in collaboration with Canadian Cancer Trials Group, Eastern Cooperative Oncology Group-American College of Radiology Imaging Network, and Southwest Oncology Group. It was launched in February 2010 and completed in May 2015; data were also analyzed during this time frame. Patients with histologically proven advanced HCC, no prior systemic therapy, Child-Pugh grade A score, Eastern Cooperative Oncology Group performance status of 0 to 2 (later amended to 0–1), and adequate hematologic, hepatic, renal, and cardiac function were eligible. The OS primary endpoint had a final analysis planned with 364 events observed among 480 total patients with 90% power to detect a 37% increase in median OS.

To determine the endpoint of HCV on outcomes with sorafenib treatment in each arm, preplanned study objectives were used to determine the proportion of patients with undetectable viral load (<50 copies/mL), and to assess and compare the HCV antiviral effect of sorafenib or sorafenib plus doxorubicin. That was performed by comparing patients’ rate of undetectable viral load or 2-log decline in viral load, and association with OS, PFS, and response rate for the whole study population, and on each treatment arm. The rate of virologic failure, defined as reversing from undetectable (<50 copies per mL) to detectable viral load, or increase in 2 logs above nadir, at any point in time was also to be assessed.

Peripheral venous blood was collected at baseline and on-treatment from patients who provided written informed consent. Samples were processed on-site within 2 hours; serum was aliquoted, frozen, and shipped on dry ice to the Alliance Pathology Coordinating Office for centralized storage. Aliquots of frozen serum were shipped to the Memorial Sloan Kettering lab for all HCV analyses. HCV RNA levels were measured by the Roche cobas AmpliPrep/cobas TaqMan HCV Test as described previously (11). Specimens were evaluated by PCR and genotyped at baseline on day 1 of cycle 1, and HCV titer levels were evaluated on day 1 of cycles 2, 3, and 4 as well as at time of disease progression and end of active treatment. Detectable HCV titers were defined as HCV RNA ≥50 copies/mL. Quasispecies changes for resistant mutations were investigated for patients who exhibited “virologic failure” (a reversal from undetectable <50 copies/mL to detectable ≥50 copies/mL viral load or an increase of 2 logs above nadir) during the study. Samples were stored and analyzed at the end of the follow-up period. Laboratory investigators were blinded to patient identity and outcomes. Influence of the HCV titer levels was evaluated in relation to the study objectives OS and PFS.

The prevalence of HCV with or without presence of hepatitis B virus (HBV) was evaluated across all patients. Those with HCV at baseline who had stored serum had HCV RNA and titer levels assessed. Summary statistics were calculated across all patients with HCV at baseline. We further summarized and compared participants who had serum versus those who did not have serum available for these analyses.

The number of copies/mL within each treatment group and across all patients was summarized at each timepoint. Differences over time and between treatments were assessed graphically, and fold changes were calculated in relation to baseline. Viral loads were evaluated as a continuous measure as well as categorized as undetectable versus detectable (<50 copies/mL; ≥50 copies/mL). A subset of patients did not have titer data available at baseline; given the observed consistency of titer levels over time, we also evaluated the influence of HCV from the first sample, where we imputed missing baseline measures using cycle 2 day 1 measures. Analyses were conducted using only those with baseline samples and using all patients with baseline values imputed.

The influence of HCV titer levels was evaluated in relation to PFS and OS. PFS was defined as the time from randomization to the progression and/or death due to any cause, where patients who were progression-free at their last evaluation were censored at that time. OS was defined as the time from randomization to death due to any cause, where patients alive at last follow-up were censored at that time. Kaplan–Meier methods were used to evaluate differences among those with detectable versus HCV undetectable titer levels. Cox proportional hazards models were used to evaluate the influence of continuous HCV titers and detectable versus undetectable HCV status in relation to PFS and OS, stratifying by treatment arm and by exploring potential effect modification of treatment arm (sorafenib with vs. without doxorubicin).

Data collection and statistical analyses were conducted by the Alliance Statistics and Data Management Center. Data quality was ensured by review of data by the Alliance Statistics and Data Management Center and by the study chairperson following Alliance policies. All analyses were based on the clinical data used for the primary clinical article for this trial (6).

Deidentified patient data may be requested from Alliance for Clinical Trials in Oncology via [email protected] if data are not publicly available. A formal review process includes verifying the availability of data, conducting a review of any existing agreements that may have implications for the project, and ensuring that any transfer is in compliance with the IRB. The investigator will be required to sign a data release form prior to transfer.

Of 356 patients enrolled on CALGB 80802, 83 patients were HCV-positive at baseline. Of note, the HCV infection rate was significantly higher in Black/African American patients (25/50 = 50%) compared with White patients (54/239 = 23%) or other race groups (4/67 = 6%; P < 0.0001; Table 1).

Among the 83 HCV-positive patients at baseline, serum and thus HCV titer data were available on 53 patients. Per arm data are featured in Supplementary Fig. S1. Among these 83 HCV-positive patients, the only baseline clinical factor that was significantly different was that those in whom serum was available tended to have a higher percentage of Eastern Cooperative Oncology Group performance score (ECOG PS) >0 relative those who did not have serum available (68.6% vs. 41.3%, respectively; P = 0.042).

Of the 53 patients with HCV titer data, 12 patients (doxorubicin + sorafenib: 5, and sorafenib: 7) had undetectable HCV titer at their first measurement. The HCV titer levels did not significantly differ between treatment arms and the values remained very consistent over time (Supplementary Fig. S2). One patient in each arm went from detectable to HCV-UN and in so doing had a >2 log reduction in their titer levels from baseline. On treatment, 39 patients had HCV- detectable while 14 were HCV-UN (sorafenib + doxorubicin: 8, sorafenib: 6 patients).

Viral load over time per treatment arm is illustrated in Fig. 1A for doxorubicin plus sorafenib (A), and in Fig. 1B for Sorafenib alone. In view of the consistency of the titer levels, we imputed the cycle 2 day 1 HCV titer status (detectable vs. not) for patients who were missing baseline samples, albeit this presumption could not be validated. For the 2 patients who had a significant drop in HCV titer levels, this occurred by the cycle 3 day 1 timepoint. Among the 12 patients who had undetectable HCV at their first sample, 8 had measurements taken from pretreatment cycle 1 day 1 baseline sample and 4 were from a cycle 2 day 1 sample.

The study was halted after accrual of 356 patients with a futility boundary crossed at a planned interim analysis. In the overall trial, median OS was 9.3 months (95% CI: 7.3–10.8 months) in the doxorubicin plus sorafenib arm and 9.4 months (95% CI: 7.3–12.9 months) in the sorafenib alone arm (HR, 1.05; 95% CI: 0.83–1.31). The median PFS was 4.0 months (95% CI: 3.4–4.9 months) in the doxorubicin plus sorafenib arm and 3.7 months (95% CI: 2.9–4.5 months) in the sorafenib alone arm (HR, 0.93; 95% CI: 0.75–1.16).

In this subset of 53 HCV-positive patients with HCV titer data available, the median follow-up was 36.1 months (95% CI: 34.7 to not reached) and 50 patients had a PFS event (progression and/or death) and 47 patients died. Median PFS for all patients across both treatment arms was 5.5 versus 2.8 months for HCV-detectable versus undetectable viral load, respectively (Fig. 2).

PFS was significantly worse in those with undetectable HCV titer levels at their first sample timepoint compared with those with detectable HCV (HR = 3.51, 95% CI: 1.58–7.78; P = 0.002). This result held even when the analysis was restricted only to those individuals with baseline samples (HR = 3.47, 95% CI: 1.47–8.17; P = 0.004; Table 2). Differences in PFS and OS between HCV-detectable and HCV-UN viral load both at baseline and postbaseline are delineated in Table 2. These differences did not translate into significant differences in OS (HR = 1.14, 95% CI: 0.58–2.23; P = 0.70; Fig. 3). These results were consistent even when looking at results by treatment arm (Supplementary Figs. S3 and S4).

In addition to the HCV titer data, we also assessed the corresponding genotype data for these patients. Of the 53 patients with measurable samples, 12 were classified as “undetected” for the genotype. Of the remaining 41 patients, the majority (n = 23) were 1a, with 8 patients as 1b, and 2 additional patients as 1 who were not able to be further differentiated as 1a versus 1b. The remaining 7 patients were classified as 2 (n = 3) and 3 (n = 4). When looking at PFS in relation to genotype, we found that there were no significant differences based on genotypes; however, those classified as “undetected” for genotype had significantly worse prognosis compared with all others (HR = 3.41, 95% CI: 1.66–6.99; P = 0.0008; Table 3).

This multigroup study of the addition of doxorubicin to sorafenib therapy did not show improvement of OS or PFS in patients with HCC. Although this study, like almost every other contemporary study in HCC, was restricted to patients who could have no worse than Child-Pugh grade A cirrhosis, HCC is a very heterogeneous disease, and similar patient populations can have very different outcomes (e.g., the SHARP North American/European and the Asia-Pacific sorafenib phase III trials; ref. 9).

Contrary to the preclinical hypothesis that sorafenib could efficiently block HCV replication and viral gene expression (2), we observed that sorafenib did not influence HCV titer levels in patients with HCC in this study. Only 2 patients showed conversion from detectable to undetectable HCV titer, one on each treatment arm. Patients treated with sorafenib, independent of the addition of doxorubicin, and who had HCV-UN had a poorer PFS compared with patients with higher titers of HCV titers at baseline; there was no difference in survival outcomes based on HCV genotype in patients with HCV infection. Our results are like the collaborative effort of two prior phase III clinical trials of sorafenib in Chinese patients with advanced HCC (12). It is important to note though that patients with HCV-UN had more T4 tumors thus more metastatic disease. Even if not statistically significant this may have influenced the suggested worse PFS. In all three studies, patients with HCV had better OS compared with patients with HBV infection (9, 10). The data presented herein, however, suggest that these outcomes are independent of any coincidental anti-HCC viral effect of the anticancer treatment.

Losses and gains of chromosome regions may differ in HCC caused by HBV as compared with HCV infection, which may partly explain the findings presented herein (10–13). The therapeutic field has evolved with the advent of checkpoint blockade as an effective treatment for HCC. The etiology of HCC is tightly connected to the activation of immune cells, in that CD8⁺ T cells help induce nonalcoholic steatohepatitis-HCC, rather than the anticipated immune surveillance (14). The impact of the tumor immune microenvironment extends beyond exhausted CD8⁺ T cells. An elevation in Ki67+ subset of CD8⁺ T cells leads to further improvement in outcome, independent of etiology, that is still more enhanced with the addition of anti-CTLA4 therapy (15).

Limitations of this correlative study include the potential invalidity of the hypothesis because novel data suggest a considerably more complex tumor immune microenvironment than originally thought, as solely impacted by the viral status etiology and extent (16, 17). In patients with chronic HCV, severe CD4⁺ and CD8⁺ T-cell dysfunctions have been reported. This suggests that HCV could employ certain mechanisms to counteract or suppress the host T-cell responses (18). An understanding of such implied molecular mechanisms has already started. Novel molecules that are carcinogenic to hepatitis viruses and activate the PI3K-Akt-mTOR pathway have been identified (19). These include five hub genes, POLR2A, POLR2B, RPL5, RPS6, and RPL23A, add to M2-type macrophages. These would be expected to serve as molecular markers and targets for diagnosis and treatment of HCC.

The small number of patients with HCV included and the small number of patient samples analyzed is also noted. When CALGB 80802 was initiated, tissue and biospecimen acquisition was optional out of concern that mandatory eligibility criteria might inhibit patient enrolment. Even though biospecimens were banked in only a subset of these patients, the correlative results provide meaningful data despite the disappointing clinical results.

Future investigation is needed on the outcomes of therapies based on the underlying etiology of HCC, particularly given the notable emergence of new therapies for HCC over the last half decade (20, 21).

To conclude, despite a strong preclinical rationale, we observed no difference in outcome for patients with HCC treated with sorafenib and observed an unexplained improvement in PFS in a small patient subset.

G.K. Abou-Alfa reports grants from Agenus, Arcus, AstraZeneca, BioNtech, BMS, Elicio, Genentech/Roche, Helsinn, Parker Institute, Pertzye, Puma, QED, Yiviva and personal fees from Astellas, AstraZeneca, Autem, Berry Genomics, BioNtech, Boehringer Ingelheim, BMS, Eisai, Exelixis, Fibriogen, Genentech/Roche, Incyte, Ipsen, Merck, Merus, Neogene, Novartis, Servier, Tempus, Thetis, Vector, Yiviva outside the submitted work; in addition, G.K. Abou-Alfa has a patent to PCT/US2014/031545 filed on March 24, 2014, and priority application Serial No.: 61/804,907; Filed: March 25, 2013 licensed. F. Innocenti reports other from AbbVie outside the submitted work. Q. Shi reports personal fees from Yiviva Inc, Regeneron Pharmaceuticals, Inc., Hoosier Cancer Research Network, Kronos Bio, Mirati Therapeutics Inc; grants from Janssen, BMS, Genetech, and Novartis outside the submitted work. P. Kumthekar reports consulting fee (e.g., advisory board): Enclear Therapies, Affinia Therapeutics, Biocept, Janssen), Bioclinica, Novocure, Mirati, Servier, Roche; Data Safety Monitoring Community: BPGBio; speaker's bureau: Seagen; Contracted Research: Genentech, Novocure, DNAtrix, Orbus Therapeutics, Servier, Plus Therapeutics. L.H. Schwartz reports other from Merck, BMS, and Regeneron outside the submitted work. A.B. El-Khoueiry reports grants and personal fees from AstraZeneca; grants from Auransa, Fulgent; personal fees from BMS, Roche/Genentech, Exelixis, Merck, Qurient, Senti Biosciences, Tallac, and Agenus outside the submitted work. J.J. Knox reports grants from Merck, Ibsen, Roche; personal fees from AstraZeneca, Incyte, and Ibsen outside the submitted work. M.M. Bertagnolli reports grants from NCI during the conduct of the study. J.A. Meyerhardt reports personal fees from Merck Pharmaceutical outside the submitted work. E.M. O'Reilly reports grants from Genentech/Roche, BioNTech, Arcus, Elicio Therapeutics, Parker Institute, Pertzye, NIH/NCI; grants and other from AstraZeneca; personal fees from Boehringher Ingelheim, BioNTech, Novartis, AstraZeneca, Astellas, Autem, BMS, Fibrogen, Merus, Genentech-Roche, and Servier; personal fees and other from Ipsen outside the submitted work. No disclosures were reported by the other authors.

G.K. Abou-Alfa: Conceptualization, data curation, formal analysis, supervision, validation, investigation, methodology, writing-original draft, project administration, writing-review and editing. S.M. Geyer: Formal analysis, supervision, validation, investigation, writing-original draft, writing-review and editing. A.B. Nixon: Data curation, supervision, investigation, writing-original draft, writing-review and editing. F. Innocenti: Formal analysis, supervision, investigation, writing-review and editing. Q. Shi: Conceptualization, data curation, formal analysis, supervision, validation, investigation, methodology, writing-review and editing. P. Kumthekar: Data curation, writing-review and editing. S. Jacobson: Formal analysis. I. El Dika: Data curation, investigation, writing-original draft, writing-review and editing. A. Yaqubie: Resources, data curation, formal analysis, writing-review and editing. J. Lopez: Data curation, formal analysis, writing-review and editing. B. Huang: Data curation, formal analysis, writing-review and editing. Y.-W. Tang: Data curation, formal analysis, writing-review and editing. Y. Wen: Resources, supervision, project administration, writing-review and editing. L.H. Schwartz: Conceptualization, data curation, visualization, writing-review and editing. A.B. El-Khoueiry: Data curation, writing-review and editing. J.J. Knox: Data curation, supervision, writing-review and editing. L. Rajdev: Data curation, writing-review and editing. M.M. Bertagnolli: Supervision, project administration, writing-review and editing. J.A. Meyerhardt: Conceptualization, supervision, writing-review and editing. E.M. O'Reilly: Conceptualization, data curation, formal analysis, supervision, validation, visualization, methodology, writing-original draft, project administration, writing-review and editing. A.P. Venook: Conceptualization, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing-review and editing.

Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Numbers U10CA180821, U10CA180882, and U24CA196171 (to the Alliance for Clinical Trials in Oncology), U10CA180820, U10CA180863, U10CA180888, U10CA189859, UG1CA233180, UG1CA233253, UG1CA233290, UG1CA233327, and UG1CA233373. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. https://acknowledgments.alliancefound.org. The National Cancer Institute of the National Institutes of Health reviewed and approved the design and conduct of the study. Bristol-Myers-Squibb and Sanofi had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the abstract; and decision to submit the abstract for publication. Bayer, Bristol-Myers-Squibb, and Sanofi are thanked for their partial support.