Second-Line Pembrolizumab Confers Improved OS, PFS in Patients With Advanced HCC

The final analysis of KEYNOTE-240, a phase III study of pembrolizumab in advanced hepatocellular cancer, demonstrated favorable benefit in overall survival and progression-free survival, according to Richard S. Finn, MD, during his presentation at the 13th Annual Conference of the International Liver Cancer Association in Chicago.

Richard S. Finn, MD

The final analysis of KEYNOTE-240 (NCT02702401), a phase III study of pembrolizumab (Keytruda) in advanced hepatocellular cancer (HCC), demonstrated favorable benefit in overall survival (OS) and progression-free survival (PFS), according to Richard S. Finn, MD, during his presentation at the 13th Annual Conference of the International Liver Cancer Association in Chicago.

Sometimes not reaching prespecified statistical significance isn’t the end of the line for second-line therapies for advanced HCC. Clinical significance should be considered when it comes to the results of the KEYNOTE-240 study for this patient population, Finn argued.

OS was higher in the pembrolizumab arm, at 13.9 months versus 10.6 months on placebo (HR, 0.781; 95% CI, 0.611-0.998;P= 0.0238). In both the primary and final analysis, PFS in the pembrolizumab arm was only slightly longer than placebo, 3 months versus 2.8 months. Both had favorable HRs of less than 1. The primary analysis had a HR of 0.775 and aP-value of 0.0186 (95% CI, 0.609-0.987). The final analysis had a HR of 0.718 and aPvalue of 0.0022 (95% CI, 0.570-0.904), again barely missing the efficacy endpoint of 0.0020, noted Finn.

The positive data generated over the last 2 years has seen new agents in liver cancer. Lenvatinib (Lenvima) has been added as a frontline therapy, as well as several agents in the second lines, including regorafenib (Stivarga), cabozantinib (Cabometyx), and the monoclonal antibody ramucirumab (Cyramza).

“However, still this is an incurable disease and we know that HCC does develop into an inflammatory milieu and therefore it’s been of interest to use drugs that target the immune system,” said Finn, assistant professor of medicine, Department of Medicine, David Geffen School of Medicine, University of California Los Angeles. Early phase II studies have demonstrated encouraging antitumor activity and manageable safety, including the CheckMate-040 (NCT01658878) study of nivolumab and the KEYNOTE-224 (NCT02702414) study of pembrolizumab. Both of these therapies have now reached accelerated approval as a result of these data sets in the United States.

“KEYNOTE-240 was the definitive study to show that pembrolizumab would improve survival versus best supportive care (BCS) in the second-line population. Arguably this is the last placebo-controlled phase III study in the second-line that we’ll see,” said Finn.

As with other second-line studies, the design of the trial focused on patients who had a progression or intolerance on first-line therapy sorafenib. Patients with main portal vein invasion were excluded. A study population of 413 patients was randomized 2:1, with patients in the pembrolizumab arm getting 200 mg every 3 weeks plus BCS and the other group getting saline-placebo every three weeks plus BCS.

The study had dual primary endpoints: OS and PFS assessed using conventional RECIST v1.1 criteria. “It’s very important we understand the statistical analysis for this study,” noted Finn. “There were dual endpoints and therefore the alpha (α) was split.” Starting with a Type 1 error of (α)=0.025 that was then allocated between progression-free survival, where PFS α=0.002 and overall survival, where OS α=0.023.

“Over the course of the study there were a number of interim analyses which required the OS α to be redistributed,” said Finn. Therefore, to be considered positive at the end of the study, efficacy boundaries were set atP= .0174 for OS andP=.0020 for PFS.

The pembrolizumab arm had 278 randomized patients, and the placebo arm had 135. At the time of Finn’s analysis, 5 patients completed and 28 patients were still being treated on the pembrolizumab arm, versus 1 patient completing treatment and 4 ongoing on the placebo arm. For both arms, the majority of patients, 245 in pembrolizumab and 130 in placebo, discontinued treatment, most due to progressive disease, 173 and 100 patients, respectively.

Overall, 40% of patients came from the EU and North America and 40% came from Asia and Japan, well-balanced in both arms. Twenty percent of patients had Barcelona Clinic Liver Cancer B stage, 80% had C. In addition, 40% had viral hepatitis as an etiology, and of those, a majority had hepatitis B (25.9% in pembrolizumab and 21.5% in the placebo arm).

Prior sorafenib use was similar across both arms, with discontinuation due to intolerance at approximately 13% and discontinuation due to progressive disease at approximately 87%. Duration of sorafenib was approximately 5 months and time from stopping sorafenib to starting KEYNOTE-240 was a little over a month, similar for both arms. Macrovascular invasion was 12.9 in the pembrolizumab arm. “That’s lower than we’ve seen in other phase III studies in the second-line,” noted Finn.

The objective response rate (ORR) at final analysis was encouraging, said Finn. A total of 18.3% patients in the pembrolizumab arm had a partial or complete response (2.2%) with the median duration lasting 13.8 months, compared to placebo with only 4.4% of patients showing only a partial response with no duration of response reached.

“Pembrolizumab, essentially, has shifted the number of patients who are non-responders to responders and also decreased the number of patients who had progressive disease,” said Finn.

The adverse events related to pembrolizumab were not surprising, said Finn though the number of Grade 3/4 events were increased. The number of immune-mediated adverse events (IMAEs) were also higher in the pembrolizumab patients, at 18.3% versus 8.2% versus placebo. However, the number that led to discontinuation of treatment for IMAEs was only 3.6%, meaning that IMAEs are something that can be managed, said Finn.

The most common adverse events regardless of etiology were an increase in AST (22.6%), bilirubin (18.6%), fatigue (18.6%), and pruritus (18.3%). There was also a frequency of these adverse events that occurs with placebo. The most common pembrolizumab treatment-related adverse events were mostly low grade events of pruritus (13.3%), fatigue(10%), diarrhea (8.2%) and rash (8.2%). A greater percentage of AST increase were Grade 3/4 (5.4%).

Immune-mediated events were increased with pembrolizumab but there were no new immune-mediated events compared to other tumor types, said Finn. “These were generally dominated by hypo- and hyperthyroidism.”

“Using 3 different analyses of post-treatment impact on OS, we’re seeing hazard ratios in the 0.6 range. This is closer to what was anticipated in the study design when taking into account subsequent treatments,” said Finn.

In summing up the phase III study, Finn noted: Pembrolizumab reduced the risk of death by 22%, with a hazard ratio of 0.78. It also improved PFS over placebo in patients with advanced HCC. However, these P-values did not reach statistical significance as outlined by the study. The response rate was higher with pembrolizumab, confirming the findings of phase II study. The safety profile wasn’t different for this class of drug across other tumor types.

“The benefit here, we would say, as captured by the OS and PFS response rate is favorable, and, I would argue supports the FDA approval for pembrolizumab in the second-line settings based on accelerated approval mechanisms,” said Finn. In the meantime, he’s awaiting the results of an ongoing phase III study from China, the KEYNOTE-3994 in the Asian Pacific region.

Finn also shared previously unpublished data on biomarkers. While the KEYNOTE-40 trial didn't require tissue sample, PDL-staining was performed on 76 patients in the pembrolizumab arm. Combined positive (CPS) and tumor proportion (TPS) were not significantly associated with clinical response to pembrolizumab, contradicting previously observed data in KEYNOTE-224 that found a clinical response association. Finn urged extreme caution in interpreting the results, as the number of patients studied was small.


Finn DS, Ryoo BY, Merle P, et al. Results of a phase 3 study of pembrolizumab versus best supportive care for second-line therapy in advanced hepatocellular carcinoma: KEYNOTE-240. Presented at: 13th Annual Conferences of the International Liver Cancer Association; September 20-22, 2019; Chicago, IL. Abstract O-01.