Nivolumab yields superior OS in NSCLC compared with docetaxel

Ben Creelan, MD

The median difference in OS of 3 months seen in the trial presented by Paz-Ares and colleagues is outstanding. Since docetaxel was approved in 1999, there has been essentially nothing, either alone or in combination, which has been able to beat it. Now ramucirumab (Cyramza, Eli Lilly) was recently approved, but in combination with docetaxel. This really does represent a big leap forward. Those of us who participated on these clinical trials like CheckMate 057 have patients who are still in remission 3 years later. They’re doing regular activities, and their lives are no longer defined by the label of lung cancer. They do not look like patients with lung cancer walking in the door. The cancer is really a minor component of their lives. They’re living. They’re not just surviving — they’re living.

Importantly, these 057 results were positive in the overall population, so, my expectation is that the drug would be FDA approved for the overall population, regardless of PD-L1 status. It is true that in this non-squamous population, we see a dramatic difference in outcomes with the PD-L1–positive patients. Even just using a 1% cutoff — which is the one of the lowest thresholds currently used for most companies — demonstrated benefit with nivolumab. In contrast, for  squamous lung cancer, PD-L1 expression, as evidenced by The New England Journal of Medicine publication by Brahmer and colleagues in CheckMate 017 — also presented by Spigel and colleagues at the ASCO Annual Meeting — was completely irrelevant. So, the survival benefit was seen in both PD-L1–negative and PD-L1–positive squamous patients regardless of the cutoff point.

So regarding 057 and the PD-L1 question; let’s look into the purported clinical utility of that biomarker in more detail. What are the important components of a good biomarker? One: you want a biomarker if it will spare patients treatments that are not going to work, when other effective treatment is available. Does this biomarker do that? No.  There are PD-L1–negative patients who still respond. Even with a less than 1% cutoff, there are some patients who respond. Two: you want a biomarker if it’s also going to spare patients toxicity from a drug that might not work. In this case, what do I have them to offer that is so much better than nivolumab in the second-line setting? Frankly, the outcomes with docetaxel have been pretty dismal for long-term durable responses. We know that the OS for patients in the squamous population with docetaxel was 22% at 1 year, and that might be a little higher in the non-squamous population, but you’re still talking about a median survival of less than half a year. If I have a drug that can push that up any more than that or anything around that with less toxicity, I’m going to use it.

Would I not treat a patient who is PD-L1 negative? I would still have a discussion with them and I’d still give the drug if the patient wants it, and most patients do. I find patients are usually the ones pushing us to do those treatments. If I’m a patient and I know that there’s a 10% to 12% response rate with this drug since I’m PD-L1 negative, I’m guessing a lot of patients are going to go for it. Most people would rather go for a smaller chance of a long-term durable remission with fewer side effects and better quality of life than going with our traditional options. Let’s also remember that a PD-L1 result must always be viewed with skepticism, due to the intrinsic nature of lung cancer:  1) it gives a score, not a yes/no answer, 2) PD-L1 is an inducible biomarker which changes over time, and 3) there is high variability about where in the tumor you test PD-L1.

One interesting aspect of these trials — a current unanswered question — is that both trials continued treatment with nivolumab every 2 weeks indefinitely for as long as it was working. What we don’t know is whether stopping the treatment and allowing the immune system to continue its vigilance would work. We do have a clinical phase 3 trial, CheckMate 153 comparing 1 year of treatment vs. continuation in patients, but frankly, that trial may end up being somewhat underpowered to answer that question. In the long-run, it may be handled by the physician’s discretion. Cost aside and as long as the patients are in agreement, I would keep giving the drug as long as it’s working because there have been patients from the original phase 1 trial, — the CheckMate 003 trial —where the drug was stopped at 2 years, who have progressed, were retreated and have had a response again. That suggests a benefit to continuing the drug, at least anecdotally.

References:

Brahmer J, et al. N Engl J Med. 2015;doi:10.1056/NEJMoa1504627.

Paz-Ares LG, et al. Abstract LBA109. Presented at: ASCO Annual Meeting. May 29-June 2, 2015; Chicago.

Spigel DL, et al. Abstract 8009. Presented at: ASCO Annual Meeting; May 29-June 2, 2015; Chicago.

Roy S. Herbst, MD, PhD

Lung cancer is the number one cause of cancer death. There are 1.6 million cases in the world, the majority of which are non-squamous. This randomized phase 3 trial of almost 600 patients demonstrated a 27% improvement in outcome with nivolumab at 1 year. The results are perhaps even better than that for some because there is clearly a tail to the curve. It is still early in follow-up, but we know that at 1 year, the results are not going to be any worse than what we saw at ASCO. There is definitely improvement in survival and less toxicity. The grade 3 to grade 4 toxicities were five times less in the patients who received nivolumab vs. docetaxel.
This is now the second positive phase 3 trial in lung cancer in previously treated patients, the other being in squamous disease. The interesting thing about this trial was the biomarker analyses. There are only data at a few cutoff levels, although it does look like using the biomarker can improve the outcome, at least above the 1% level. However, in the group that is completely negative by the biomarker, although the benefit is diminished, it does not seem like the drug is harmful or any worse than docetaxel (with the potential still for a long-term benefit in a few with reduced toxicity). If we can learn to pick out a PD-L1–positive patient population that really seems to benefit, this will be the next great leap forward.  Although again, this will leave out a lot of patients who would benefit who you then might not be treating. Thus, where you pick your cutoff point for these biomarkers is very critical.
The problem with the assay is that there is so much variability and heterogeneity within the specimen. There can be heterogeneity in the tumor when you actually look at the biomarker. In some cases, the biomarker is evaluated at the time the patient is treated, but some of these tissue samples can be 1 year old or more. Interferon-gamma stimulates PD-L1, so we might be looking at an older specimen and it might not reflect what is happening in real time. There are a lot of variables that can impact how this biomarker works, unlike with a mutation, which you either have or you don’t.
We have to remember these drugs have side effects, too. Immunotherapies have fewer side effects, perhaps, than chemotherapy, but they do have unique side effects related to inflammation. They include pneumonitis, inflammation in the lung, hepatitis, thyroiditis, rash on the skin, colitis, etc. We should try to identify who treatment is going to benefit the most, because to give someone toxicity without any chance of benefit is never good.
The trial is called CheckMate, so I would say, “Is this a checkmate, or are we still in the endgame?” I think it’s the latter. This is important progress for patients with lung cancer; however, we are very far from where we need to be. Clearly, this is wonderful for the 20% or 30% of patients who are benefitting; however, we have to think about how to identify the 70% who are not as early in their care as possible. Who might benefit from a combination of therapies? Who might benefit from a combination with chemotherapy and a combination of immunotherapy? There is a lot of work that still needs to be done. The biomarker is interesting, but it needs to be better developed. This is a great first step, and now we want to move these agents into even earlier stages of lung cancer. This is a paradigm-shifting event in lung cancer, but we have to make the next move carefully.