As we watch the clinical development of PD-1 pathway inhibitors we are struck by the ability of this approach to produce clinical efficacy in diverse cancers. Here we will briefly run through the rest of the landscape, starting with non-small cell lung cancer (NSCLC), then touching on metastatic renal cell carcinoma (RCC), glioblastoma (GBM), bladder cancer (UB), ovarian carcinoma and others. In many cases we are beginning to see the use of combination therapy, a setting that generally requires a careful look at toxicity. We’ll also look at the contentious issue of biomarker development for PD-1 and PD-L1 antagonists. The data are broken down into easy to understand bits, otherwise the whole thing is overwhelming.
The Emerging Role of PD-1 Pathway Inhibition in Lung Cancer
In the context of the often brutally aggressive tumors classified as the Non-Small-Cell Lung Cancers (NSCLC) progress is difficult and even incremental improvement in care is cause for celebration. Even so it was difficult to gather a consensus view of the clinical data being generated by the PD-1 pathway antibodies in NSCLC. In the melanoma setting (parts 1 and 2) our enthusiasm for PD-1 pathway inhibitors is driven by really terrific responses in some patients. The goal there is to improve the response rates while controlling toxicity. The clinicians argue, certainly with merit, that the responses seen justify the occasionally difficult toxicity. After all, all of their available therapies are limited by toxicity concerns, not only chemotherapy but also the small molecule targeted therapeutics. In this context, PD-1 directed therapies are well positioned.
It is not yet clear how these issues will play out in NSCLC. On the positive side, a subset of patients respond remarkably well to PD-1 pathway inhibition, and we may develop an understanding of how identify such patients. However, the overall response rates remain low, and impact on PFS and OS is small (see part 1 for a list of abbreviations). Complicating our understanding of the benefit of this class of therapeutics in NSCLC are three observations. The first observation is that tumor responses in this disease setting can be anomalous and may not be appropriately captured by standard RECIST tumor response criteria. The second observation is that severe toxicity can truly derail patients, and may even sensitize some patients to chronic toxicity that prevents application of other types of therapies. The third observation is that some targeted therapeutics for molecularly defined subsets of NSCLC patients are an attractive option to PD-1 directed therapeutics, and we don’t know yet if these can be combined.
The amount of data presented at ASCO14 was huge, and won’t try to cover it all.
There was a pretty dramatic response to the clinical trial data describing the utility of PD-1 pathway inhibitors in NSCLC. Bristol-Myers Squibb’s (NASDAQ: BMY) stock price dropped more than 6% from the start of ASCO on May 30th through the following week, and this on top of a long slide that started in March. Merck & Co (NYSE: MRK) stock jumped more than 2.5% at the same time. What was going on here? Let’s look at the estimated market sizes for three critical indications:
It is clear from the table that NSCLC is the largest patient population by more than 10-fold. It is also the case that NSCLC lacks the range of treatment options available to advanced melanoma and RCC patients. From the investor perspective then, NSCLC is a very big deal. Lets start there, and see how the emerging PD-1 pathway therapeutic class did in this setting.
Nivolumab anti-PD-1 antibody was studied in a number of clinical settings (ASCO14 abstracts #8024 and #8113). As monotherapy, nivo was generally well tolerated, and very effective for some patients. The ORR = 22% – 36% and a subset of patients had a durable response. Combining nivo with dual-platinum based chemotherapy increased the response rate but dramatically increased the SAEs, and in this early data did not appear to impact 1 year survival rates significantly. This table sums up some of the available data.
Some patients responded very well – these were typically patients that had a non-squamous cell phenotype and whose tumors expressed PD-L1 and were therefore actively shutting down T cell responses by binding PD-1 on T cells. This figure is from a poster presented by Scott Gettinger (ASCO14 Abstract #8024).
Tumor size is given on the Y-axis and duration is shown across the X-Axis, so for some responding patients the outcome is very good. In an effort to boost the effectiveness of immunotherapy for NSCLC, BMY ran a combination trial of nivo + ipi, as they had done previously in melanoma. They enrolled chemo-naive NSCLC patients, stratified by cell type into squamous or non-squamous groups, then gave these patients an induction regimen of nivo + ipi for 84 days (4 x 21 day cycles). The doses were either 1 + 3 mpk IV Q3W or 3 + 1 mpk IV Q3W, referring to the dose or nivo + ipi, respectively. Then the patients went onto a nivo maintenance schedule. The combination worked rather well, seeming to wipe out the disparity between the two cell types and overcoming some of the resistance seen in PD-1 tumors. Here are some illustrations of the data from Scott Antonio’s poster (ASCO14 Abstract #8023):
As you can see, a fair number of patients have a sustained decrease in tumor volume or a stable disease course (no change in tumor volume over time). That’s a very nice result. Now the bad news. This combination therapy was nasty, a toxic brew. Discontinuation rates averaged 35% across the treatment arms with AEs including pneumonitis, liver damage, colitis, autoimmune nephritis (kidney inflammation & fibrosis), pulmonary hemorrhage, endocrinopathy, neuropathy, etc. Six patients (12%) died. We note that only 3 fatalities were directly attributed to the study drugs, but no one felt comfortable with these results. BMY’s stock price promptly dropped. We won’t know if the reaction was justified until other dosing and combo regimens are tried, but investors found another home, and that was with the competing drugs from Merck and Astra Zeneca (NYSE: AZN).
Immediately striking were the results from an NSCLC trial with the anti-PD-1 antibody pembrolizumab as monotherapy. Patients were selected based on positive PD-L1 staining on > 1% of tumor cells and given 2 or 10 mpg pembro Q3W or 10 mpk Q2W (ASCO Abstract #8077). The response rate was 26% or more, and the SAE rate was low (4%) although 18% dropped out due to AEs of any grade. Responses were durable with more than half of responders still on treatment at the time of data lock. A couple of things to note: these were treatment-naive patients, so presenting early in disease course. Second this figure (from the presentation by Naiyer Rizvi, ASCO #8007) shows maximum responses (ie. best response) and would not capture rebounding tumor size.
As side note: the authors introduced an “immune response” criteria response rate because some patients would respond after having a new lesion appear (which would trigger the progressive disease (PD) score). Using these new criteria responses were even higher. We’ll see if these become more widely accepted.
AZN/Medimmune presented early expansion data on their anti-PD-L1 antibody, MEDI4736 (ASCO14 Abstract #3002 presented by Neil Segal) in advanced solid tumors. 84 patients with NSCLC were enrolled. The reported ORR was low but this was an accident of sampling as most patients had not gotten to their second screen yet and could not be scored as responders, per protocol. More impressively the vast majority of patients, across diverse tumor classes, remained on therapy.
While it will be critical to see this data updated, the early read is very encouraging. Not waiting, AZN has initiated a pivotal trial in NSCLC and also a combo trial with their own anti-CTLA4 antibody, tremelimumab. We should be cautiously optimistic that the combination of anti-PD-L1 antibody with anti-CTLA4 antibody will have fewer tox issues than the anti-PD-1 combination, as mild tox appears to be a common feature of targeting the ligand rather than the receptor.
At this point of development, nivo has run into some problems in the combination setting, pembro looks promising, and MEDI4736 also looks promising. It will be very interesting to follow these story lines as they mature.
OK, RCC and other tumor types next, stay tuned.