Two Approaches To PD-1 Blocker Pediatric Studies Could Help Clarify Biomarker’s Role

Merck plans to enrich its Phase II study for MK-3475 with patients whose tumors express PD-L1; Bristol would take all comers in its Phase I/II trial of nivolumab.

A Nov. 5 pediatric advisory subcommittee highlighted a difficulty sponsors face as they develop PD-1 checkpoint inhibitors to treat cancer – uncertainty about whether PD-L1 expression should be used as a biomarker to identify patients for treatment.

More data is needed to determine how suited PD-L1 expression is as a predictive marker for efficacy, noted Gregory Reaman, associate director for oncology sciences in FDA’s Office of Hematology and Oncologic Drugs during the subcommittee meeting of the Oncologic Drugs Advisory Committee. “There are multiple unknowns, many unknowns. Everything is unknown,” he said, and “we have to seize every opportunity” to find the answers.

One such opportunity is pediatric development plans for the two PD-1 checkpoint inhibitors – Merck & Co. Inc.’s MK-3475 and Bristol-Myers Squibb Co.’s nivolumab. MK-3475 – that were the subject of the meeting (Also see "FDA’s Approach To Combination Pediatric Therapies: Monotherapy Comes First" - Pink Sheet, 6 Nov, 2013.).

Checkpoint blockers aim to defeat a tumor’s resistance to the T-lymphocytes produced by a patient’s immune system to fight intruders. PD-1 cells serve as a regulatory switch on the T-cells to prevent them from attacking a patient’s healthy tissue; tumors can trip the same “off” switch by expressing PD-L1 and PD-L2 ligands.

The companies differ in their approach to utilizing the potentially predictive biomarker of PD-L1 expression in their pediatric programs. Merck plans to enrich its Phase II with PD-L1 expressers, while Bristol would take on all-comers in Phase I/II.

Bristol reported the predictive potential of PD-L1 expression when it unveiled Phase I results in solid tumors (Also see "Bristol’s New Anti-PD-1 Immunotherapy An ASCO High Point" - Pink Sheet, 11 Jun, 2012.). During the 2013 American Association of Cancer Research annual meeting, Johns Hopkins’ oncologist Suzanne Topalian, who had reported the PD-L1 expression findings in 2012, urged caution urged caution in using the marker because of the relatively limited data on it (Also see "Immune Checkpoint Blockade: The New Big Idea In Cancer" - Pink Sheet, 29 Apr, 2013.).

The two different approaches taken by Merck and Bristol may offer the opportunity to learn “whether or not there may be predictive information that we can learn from PD-L1,” Reaman told the pediatric subcommittee.

Bristol and Merck are not alone in their quest for a PD-1/L1 checkpoint inhibitor and biomarker. Roche/Genentech Inc. is developing an immunohistochemistry assay to screen for PD-L1 expression alongside its PD-L1 inhibitor MPDL3280A (Also see "Roche’s House Advantage: Does Having It All Under One Roof Make A Difference?" - Pink Sheet, 10 Jun, 2013.).

Enhanced Treatment Effect

Merck is developing an IHC assay that can detect PD-L1 in formalin-fixed paraffin-embedded human tumor samples. The company plans to evaluate the assay retrospectively in a Phase I dose-escalation study and use it prospectively to select an enriched patient population for a Phase II adaptive indication-finding study. Depending on the data from the two early trials, the assay could be used to enrich or stratify patients in a Phase III safety and efficacy study.

Proposed MK-3475 Pediatric Plan

• Phase I: a dose escalation study that will evaluate two to three dose levels in children between six months and 18 years of age with advanced melanoma and advanced relapsed/refractory solid tumors and lymphoma; may adjust dose levels based on emerging PK data
• Phase II: an adaptive indication-finding study that will test the recommended Phase II dose in children with tumor types prioritized based on PD-L1 expression data from banked tumors and efficacy signals in Phase I; expected to exclude patients with no detectable PD-L1 expression
• Phase III: the study will select one indication, based on Phase II results, to evaluate in a randomized comparison to standard-of-care; PD-L1 expression could be used to enrich or stratify participants

MK-3475 may work on a broad range of pediatric patients, “but in the beginning we might miss a very reactive drug if we’re not looking potentially for a PD-L1 frame,” Eric Rubin, Merck’s therapeutic area head for oncology clinical development, told the panel.

He cited Pfizer Inc.’s Xalkori (crizotinib), which received accelerated approval for late-stage non-small cell lung cancer in 2011, as an example of a drug that might have failed if the company had not conducted its trial in patients whose tumor expresses the abnormal anaplastic lymphoma kinase gene (Also see "Pfizer’s Crizotinib Eases Past FDA With Targeted Population" - Pink Sheet, 29 Aug, 2011.).

Uncertainty About Predictiveness

Bristol, which has the first checkpoint blockade therapy with its CTLA-4 inhibitor Yervoy (ipilimumab) for melanoma, is more cautious about the PD-L1 biomarker. Published data suggest that PD-L1 negative tumors may be less likely to respond to nivolumab, but the underlying data were uncontrolled and collected retrospectively and are insufficient to conclude the PD-L1 expression is predictive, the company explained in briefing materials for the subcommittee. Bristol currently is studying the biomarker potential of PD-L1 expression in a Phase III adult trial in non-small cell lung cancer, melanoma and renal cell carcinoma.

There is also a question of the frequency of PD-L1 expression on pediatric tumors. So instead of bringing the biomarker into the pediatric clinical program early on, the company is conducting a nonclinical study on multiple archived or tissue samples from pediatric tumors to identify the types that may more often express PD-L1. The firm has an IHC assay developed with Dako AS, which was acquired by Agilent Technologies Inc. in 2012 (Also see "Agilent Builds Cancer Diagnostics Biz Via Dako Acquisition For $2.2 Billion" - Medtech Insight, 21 May, 2012.).

Bristol will use the biomarker information to prioritize pediatric tumor types for clinical study. The company also plans to look at archived biopsy samples for the type and quantity of tumor-infiltrating lymphocytes (TIL) to identify tumor types that “may have a higher propensity to engage the immune system and respond to nivolumab.”

Proposed Nivolumab Pediatric Phase I/II Plan

• Dose finding: Initiation at the adult dose of 3 mg/kg every two weeks in two groups of six, those two to 22 years of age and those 12 to 18 years of age, with select pediatric tumors with unmet medical need
• Trial expansion: expand the size of the trial by adding patients, including young adults, with Ewing sarcoma, osteosarcoma, rhabdomyosarcoma, neuroblastoma and other tumors if identified in the initial portion of the study
• Combination component: nivolumab/ipilimumab at a dose of 1/1 to 1/3 mg/kg every three weeks in the same two age groups

More activity is expected “with the high PD-L1 expresser,” but the company also is interested in gathering data on how nivolumab performs in patients without the marker, Bristol said.

FDA stressed the importance of gaining information about marker-negative patients in a draft guidance the agency issued in 2012 on enrichment strategies for clinical trials. The draft discusses strategies that use markers predictive of response, such as PD-L1 expression, prognostic markers that identify patients likely to have a disease-related endpoint event, and those aimed at decreasing the heterogeneity of patients.

While enrolling patients more likely to respond to a therapy can make the treatment effect more evident, marker-negative data are reassuring about whether the off-target population could benefit as well, the guidance notes (Also see "Enriched Trials Not Just About Marker-Positive Patients, FDA Says" - Pink Sheet, 24 Dec, 2012.).

A New Development Paradigm, Or Not?

Lynne Yao, associate director of the pediatric and maternal health staff in the Office of New Drugs, suggested immunotherapies and associated biomarkers are changing the way drugs are developed and regulators might need new models.

“It seems that there’s a little bit more tension between who might actually benefit and who might not, and that creates this change in paradigm between a Phase I traditional [trial] and a Phase II,” she noted.

Reaman countered that a new paradigm has yet to be defined, but conceded that drug development is becoming a bit more unsettled as sponsors move beyond compounds with known “mechanisms of action that were very basic and crossed many tumor types.”

One of the issues developers and regulators confront is whether to “enrich so that you can provide the greatest benefit for the population of patients for whom you think you have the ability to predict they will benefit. But we also know from experience that some of these new agents are also active in the population of patients that are negative for that predictive biomarker. So we don’t really know how to do this yet.”

In the case of the pediatric studies, he noted, “anything that we do has to really address that very issue. We need to prospectively evaluate biomarkers and we have to retrospectively evaluate biomarkers for their predictive ability and I think we’ll have the opportunity to do both.”