Targeted Therapies spoke with Suzanne Topalian, MD, professor of Surgery and Oncology at Johns Hopkins University School of Medicine, about the progress and promise of PD-1 targeted cancer therapy.
Suzanne Topalian, MD
In 2012, Suzanne Topalian, MD, professor of Surgery and Oncology at Johns Hopkins University School of Medicine, and colleagues published an article inThe New England Journal of Medicine(NEJM) that introduced readers to the biological foundation for development of anti-PD-1 antibody therapy.1The study presented early evidence for the safety and efficacy of PD-1 blockade with nivolumab for the treatment of several forms of cancer in a phase I study.
One year later, several phase III trials of anti-PD-1 and anti-PD-L1 agents are now recruiting participants after impressive evidence for rapid and durable progression-free (PFS) and overall survival (OS) emerged from a number of presentations at the 2013 American Society of Clinical Oncology (ASCO) Annual Meeting and the 2013 European Cancer Congress. Along with advanced safety and efficacy studies are investigations into reliable biomarkers of response to PD-1 pathway blockade.
Targeted Therapiesspoke with Dr. Topalian about the progress and promise of PD-1 targeted cancer therapy.
TheNEJMarticle1reported results from a phase I study that enrolled patients with advanced melanoma, non-small cell lung cancer (NSCLC), castrationresistant prostate cancer, renal cell, or colorectal cancer. Could you briefly describe the findings?
Dr. Topalian:The study that was published and reported at ASCO  was a phase I, dose-seeking trial. The primary objectives were to characterize the safety and tolerability of nivolumab, the anti-PD-1 drug. A secondary objective was preliminary assessment of efficacy. We reported that there were significant rates of objective response in three different kinds of cancer, among the five kinds we studiedNSCLC, melanoma, and kidney cancer.
Since then we have published some updated results from the same study in abstract form at the ASCO meeting in 2013.2What we presented at ASCO this year was a different kind of analysis, now with longer follow-up, looking at PFS and OS of those patients. We hypothesized, based on objective response rates, and also based upon the fact that there were additional patients whose cancer stabilized for 6 months or longer, that those clinical outcomes might actually be reflected in durable OS.
What we reported at ASCO were survival rates at 1 and 2 years, and median OS curves that we considered to be quite favorable in the context of published literature in similar groups of patients with advanced treatment-refractory lung cancer, melanoma, and kidney cancer.2The original study was expanded to include over 300 patients by adding cohorts to address new questions that arose during the study, to gain more experience with the drug, and to explore in a preliminary way what the response rates might be. But, we cannot really know if there is a survival benefit without information from phase III trials, which are now in progress in NSCLC, melanoma, and kidney cancer with nivolumab.
Your investigations included looking into PD-L1 expression as a candidate molecular marker, or biomarker?
We published preliminary findings in a subset of patients on the nivolumab trial inNEJMin 2012, showing that there was a statistically significant correlation between tumor cell expression of PD-L1 in pretreatment tumor biopsies (assessed by immunohistochemical staining techniques developed in our labs) and objective responses to nivolumab by RECIST.1Of course, there are still many questions about this marker, and new investigations in the past year and a half using different drugs or using the same drug to block the PD-1 pathway have explored some of these issues.
At ASCO this year, there were a few abstracts reporting on the search for biomarkers related to PD-1 and PD-L1 blockade.
Yes, at ASCO there was a report from the Bristol-Myers Squibb team, presented by Joseph Grosso.3They looked at a larger series of biopsies from patients who were included in the study reported inNEJM, but they had developed a new PD-L1 biomarker assay. They used a different antibody for staining the tissue sections. They had also developed an automated technique. (What we published was a rather painstaking manual technique that we did in our own laboratories.)
Also, in the Grosso study, they looked at only one tumor biopsy from each patient, whereas inNEJM, if multiple biopsies were available from the patients, we analyzed every biopsy for PD-L1 expression.
Also at ASCO, Roy Herbst from Yale and John Powderly from Carolina BioOncology reported on biomarker studies that were done in conjunction with studies on Roche’s anti-PDL1 drug, MPDL3280A.4In that study, they used yet a different antibody for staining biopsies, a different automated staining technique, and a different definition of what they were calling “PD-L1 positive”a more liberal method of scoring.
In the Powderly et al study, as in the Grosso study, the results were in line with what we reported inNEJM, which is that patients whose tumors expressed PD-L1 were more likely to respond to PD-1 or PD-L1 blockade. So, while the results are not exactly the same, I think it is encouraging that the results are very similar, given that the assay techniques are different, and in one case even the definition of “PD-L1 positive tumor” is different.
What are next steps in the development and expansion of this type of immunotherapy?
The first step is going to be conducting the phase III studies to explore potential survival benefit. If this is something that can be documented, then one or more of these drugs would be expected to be approved by the regulatory agencies as a standard of care in the three diseases mentioned [NSCLC, melanoma, and kidney cancer]. That makes it much easier to go to the second step, which we think is the futurethe development of combination therapies based on PD-1 or PD-L1 blockade. We can lump them into what we call “PD-1 pathway blockade.”
Where do you think researchers are in the process of biomarker development?
In terms of the biomarkers, that is a really complicated area. One of the difficulties is that the more recent studies with the automated assays have found that some patients whose tumors do not stain positively for the molecular marker nevertheless can respond to the drug.3
This [evidence] calls into question [whether], if we only select patients whose tumors are defined as “PD-L1 positive” for treatment, we are going to be denying treatment to a group of patients who might be able to respond to the treatment but are assay “negative.” This is a real concern.
That may be a glitch in the momentum of biomarker development?
While all of the studies showing that patients whose tumors are PD-L1 positive using these various assays and definitionsthe biomarker-positive patients—are more likely to respond to PD-1 pathway blockade, the newer studies are also finding a small group of patients, at around the 10% to 15% level, who are so-called “marker-negative” but who still are showing response to treatment. This really now raises concerns regarding how these various tests might be used to select patients. [However], we think that this kind of test might be used to identify different kinds of cancers that should be tested for susceptibility to PD-1 blockade.
Are you saying the thinking is to use PD-L1 immunostaining to identify types of cancers other than those for which the current PD-1/PD-L1 blockers are being tested, to see whether they are also responsive PD-1 pathway blockade?
Yes. Right now, only a limited number of kinds of cancers have been tested in the clinic for response to PD-1 pathway blockade. There are hundreds of different kinds of cancer, and we are now faced with a really daunting task of prioritizing what the next studies should focus on, and what different kinds of cancers might be susceptible to treatment [with PD-1 pathway blockade].
In a follow-up to a large phase Ib study of a cohort of patients with a variety of advanced solid tumors who were treated with nivolumab, the PD-1 pathway inhibitor was associated with durable overall survival (OS), high response rates, and an acceptable safety profile in pretreated patients with advanced melanoma.1
Nivolumab was praised for its ability to “break the ceiling” of durable tumor response in patients with advanced cancer when the phase I trial results in advanced lung cancer, melanoma, and renal cancer were first reported at the 2012 American Society of Clinical Oncology (ASCO) annual meeting and published simultaneously in The New England Journal of Medicine.2Previously, durable tumor response had not often exceeded 15%.
“Median duration of response was nearly 2 years, also a very high number,” said trial investigator Mario Sznol, MD, of the Yale Cancer Center, New Haven, Connecticut. “The responses occurred rapidly in most patients, and patients could continue to respond even after the drug was stopped.”
For the long-term follow-up analysis, a total of 107 patients with advanced melanoma participating in the phase I study2had received the PD-1 pathway inhibitor as of July 2012. Of this number, 103 patients were ECOG performance status ≤1. Approximately 25% of patients had undergone three or more prior therapies. Median OS for the 3 mg/kg dosage level chosen by investigators for subsequent phase III study was 20.3 months, and 40% of patients were alive at 3 years (Table).
Of 29 responding patients who had initiated treatment at 1 year or more prior to data analysis, 16 had responses lasting for longer than 1 year.
Drug-related adverse events (AEs) of any grade occurred in 82% of patients: 21% were grade 3-4. The most common of these included lymphopenia (3%), fatigue, and lipase increases (2%). Grade 3-4 AEs included diarrhea (2%), endocrine disorders (2%), and hepatitis (1%). No grade 3 or greater drug-related pneumonitis was seen in melanoma patients participating in the trial.
ORR, n (%)
Median OS, months (95% CI)
16.8 (12.5 NR)
NR (8.2 NR)
20.3 (8.2 NR)
10.5 (7.2 NR)
% (95% CI)
Patients at risk, n
aResponse-evaluable patients on study for 6 months or longer;b1 complete response;cOS estimates after 1 year reflect heavy censoring and shorter follow-up for patients enrolling later in the study. NR indicates not reached; ORR, overall response rate; OS, overall survival.
“The basic message here is that we are reporting survival for the first time in patients with previously treated melanoma who received this drug,” Sznol said. “The median survival was 16.8 months, which compares favorably to other agents. The objective response rate in 103 patients was 31%a high number for previously treated patients.”
We feel that PD-L1 expression by tumors may provide a way to sort this outto create a priority list—of different kinds of cancers that should be looked at in the clinic. Our group has published [evidence] within the past year that head and neck cancers express PD-L1,5and that Merkel cell carcinoma, a rare and aggressive kind of skin cancer, expresses PD-L1.6These are just two examples of cancers that might be suitable targets for clinical study.