Taking It To The Front Lines: PD-1 Inhibitors Moving Earlier In Lung Cancer

With data on Merck’s pembrolizumab and Bristol’s nivolumab in lung cancer, the question has become whether the response rates are good enough to consider first-line therapy. It is also becoming a proving ground for PD-L1 expression as a biomarker for the class.

Data for competing anti-PD-1 antibodies from Merck & Co. Inc. and Bristol-Myers Squibb Co. in first-line lung cancer raised questions at the recent American Society of Clinical Oncology annual meeting about how good response rates need to be in this important indication and whether PD-L1 is the right biomarker to select patients.

Lung cancer is one of many indications where PD-1 and PD-L1 inhibitors could play a role (see related story, (Also see "Checkpoint Immunotherapy’s Growing Pains Show At ASCO" - Pink Sheet, 16 Jun, 2014.)). Due to the large patient population, it’s been a top priority for big pharmas in the immuno-oncology space, with many companies moving quickly after the initial tests were positive in melanoma (Also see "Big Pharma Explores New Frontier Of Lung Cancer Immunotherapy" - Pink Sheet, 26 Aug, 2013.).

Merck presented data from the Phase I KEYNOTE-001 study of pembrolizumab in 73 patients at ASCO, 78% of whom were positive for PD-L1 (defined as at least 1% of tumor cells stained). Of those, they evaluated response based on RECIST criteria (42 patients) and immune-related response criteria (investigator-reviewed, 45 patients).

The objective response rate using RECIST was 26% and 45% with immune-related response criteria (irRC). The 26% ORR is comparable to chemotherapy in other first-line trials, said Johns Hopkins’ lung cancer specialist Julie Brahmer, discussing the results at the meeting on June 2.

Median progression-free survival was 6.75 months by central review and 9.25 months using irRC criteria. Brahmer described the 6.75 month figure as “very intriguing,” because it’s longer than the 4.5 months to 5.3 months reported for doublet chemotherapy in first-line NSCLC trials.

During Merck’s June 2 investor event on the sidelines of the ASCO meeting, the company noted the difference between response by RECIST versus irRC remains a challenge. “We do have to really get our heads around how we really reconcile these two forms of response assessment,” said Roy Baynes, senior VP of global clinical development.

Bristol’s competing nivolumab demonstrated a 30% ORR, working in both squamous and non-squamous disease types, as a first-line monotherapy in NSCLC in the Phase I CheckMate-012 study of 20 patients. The effects were durable and there were three complete responses. The ORR was 50% in PD-L1+ patients and 0 for those who were PD-L1-.

“Is PD-L1 positivity the right biomarker to use and what is the best cut-off to use?” Brahmer said in discussing the results. “I could spend the rest of the hour discussing this point.”

The role of PD-L1 expression levels as a potential biomarker remains a controversy with regard to the relevance in different settings and the different methods for testing.

The companies are using different definitions of PD-L1 expression and different diagnostic tests. Merck data recently released at the American Association for Cancer Research meeting in April had indicated that the PD-L1 biomarker was looking useful in lung cancer, but much less so in melanoma (Also see "Merck Data Could Support Use Of PD-L1 Biomarker In Lung Cancer" - Pink Sheet, 22 Apr, 2014.). However, the lung cancer data suggested that the right threshold for PD-L1 positivity in this tumor type might be ≥50% of tumor cells stained.

There was some thought that the Merck data should have been better, given the levels of expression in the trial. BMO Capital Markets analyst Alex Arfai said in a June 2 note to investors that the 26% rate for pembrolizumab “seems rather low given that 78% of the evaluable patients in this group were determined to be PD-L1 positive based on Merck’s own assay.”

Bristol’s response rate in PD-L1 positive patients was more than 50% by comparison, granted the study was small, and the assay and definition of PD-L1 were different, the analyst said. Data so far suggest that the anti-PD-1 antibodies have very similar clinical efficacy, he added. During its June 2 ASCO investor event, Bristol execs expressed confidence in their assay, which they described as “very sensitive” and “robust to detect PD-L1 in different tumor types.”

In an interview after the meeting, Brahmer said a response rate on par with chemo (26% with pembrolizumab) was not a bad result, particularly if the treatment is easier to tolerate. Of course, oncologists would love to see a response rate that is higher in PD-L1+ patients, but the threshold in the trial for this marker was lower than what is being tested in other trials, she said.

Nevertheless, as with other tumor types, clinicians are questioning the use of PD-L1 expression as a biomarker in lung cancer, based on limited early data.

“Is PD-L1 positivity the right biomarker to use and what is the best cut-off to use? And in fact I could spend the rest of the hour discussing this point, but I won’t bore you,” Brahmer said, during her discussion of pembrolizumab data at the meeting.

She also questioned the requirements for approval in first-line lung cancer. According to ASCO’s own criteria for a meaningful improvement in outcomes in this setting, median overall survival should improve by from 2.5 months to 4 months with a target hazard ratio of 0.8 or less, one-year survival should be up by about 10% improvement. A PFS improvement of 3-4 months is meaningful, according to the criteria, which were published in the Journal of Clinical Oncology in March 2014.

However, Brahmer asked whether these criteria would take into account the raising of the tail of the survival curve, “which is where the power of immunotherapy theoretically lies.” Growing experience with immunotherapy suggests that rather than immediate responses, the true benefit of therapy lies with long-term effects.

Approval requirements in later lines of therapy will be tested with the review of Bristol’s nivolumab, which was recently filed as a rolling BLA for the indication of third-line squamous NSCLC (Also see "Cancer Checkpoint Roundup: Big Pharmas Advance Ahead of ASCO" - Pink Sheet, 5 May, 2014.).

In addition to the first-line results, Bristol reported additional long-term data from a Phase I study (Study 003) of 129 heavily pretreated NSCLC patients. In that trial, the two-year survival rate for all doses tested was 24%. The two-year survival rate was 45% when the drug was given at a 3 mg/kg dose, which is being tested in Phase III.

“These are third-, fourth-line lung cancer patients,” Fouad Namouni, development lead for nivolumab, said during the company’s investor event. “Reporting this type of survival in our belief is remarkable and really gives us an indication about the potential of our large Phase III program in lung cancer.”

Bristol is running three Phase III studies of nivolumab in NSCLC and one Phase III study of the CTLA-4 inhibitor Yervoy (ipilimumab) in this tumor type.