FIRST LINE IMMUNOTHERAPY FOR NSCLC

Clearly, these newer agents have activity. Randomized trials of pembrolizumab^[1] show that patients with high programmed death ligand 1 (PD-L1) expression have an improved progression-free and overall survival. I think it is very, very important that whenever we have a better drug, we start with that. Nothing is more important than inducing remission and maintain it for as long as possible up front.

How do we choose at the current time? The first thing, of course, is to talk to the patient and see what they're thinking about their illness and the approach they would like to take. We all learn in our training that, first and foremost, we have to be very clear in giving patients the facts they need to make decisions and then to clearly listen to them.

Many of us are impressed by the fact that when PD-L1 drugs work they can induce very long remissions, and that overall the side-effect profile is quite different from that of standard chemotherapies. However, when these side effects occur, they can be every bit as serious as a bout of neutropenic sepsis or a pneumonitis caused by erlotinib or crizotinib.

All of these drugs have serious toxicities, of different types and intensities. In general, when you look at the totality of toxicity, as has been shown in the second-line trials comparing chemotherapy with nivolumab and pembrolizumab, the side effects are generally perceived to be fewer in overall frequency and severity. This being the case, I think it is very reasonable to offer these agents up front. There are more and more randomized data to help us support that decision.

Three Predictors of Benefit 

There are some caveats to consider. There are things that suggest that these drugs would be more effective in an individual, one of which is the degree of PD-L1 expression. Nobody knows the exact cutoff point or is saying that they can absolutely select patients based on PD-L1. You're instead making a value judgment by putting all of the data together. It is clear that the more expression of PD-L1 there is, the more likely there is going to be benefit. The potential benefit is clearer for people with PD-L1 levels of 0 or 100; it's the people in between for whom it's more difficult to establish.

The second important thing to note comes from data originally published by my former colleague, Naiyer Rizvi, and my current colleague, Matthew Hellmann,^[2] which showed that the number of nonsynonymous mutations (known as the mutational burden) is also associated with benefit from these agents. The higher the number of these mutations, the greater the chance of benefit. Those patients with high PD-L1 expression and a large number of mutations are the most likely to benefit. Conversely, patients with fewer mutations, generally such as in those with driver oncogenes like anaplastic lymphoma kinase (ALK)-positive or epidermal growth factor receptor (EGFR) mutation lung cancers, would therefore be less likely to benefit.

A history of cigarette smoking is the last surrogate for mutations, one that, with more research, may prove to be an independent predictor in and of itself. The more you smoke, the more DNA damage occurs. You clearly can make some additional discrimination over who should receive these drugs as initial therapy by the degree of smoking. A history of prolonged or even current smoking should be assessed together with having a high degree of PD-L1 expression and a high number of synonymous mutations, both of which we can now more readily test for.

If you have all three of those factors, it predicts a greater likelihood of benefit. Patients with those characteristics who also wish to move ahead with this kind of immune therapy would be the best candidates for it. Conversely, patients who do not have those characteristics, most obviously being patients with EGFR mutations or ALK rearrangements, would be the least likely to recommend this strategy to first.

I want to remind everyone that the presence of an EGFR mutation does not mean that a patient will not respond to one of these agents. When weighing the choice of initial therapy and the likelihood of benefit in these patients, I think a tyrosine-kinase inhibitor would be more favorable as initial therapy. However, it does not mean that at some point in the patient's illness, using a checkpoint inhibitor will not be the best therapy for them.

In summary, just like in every other treatment decision, we are constantly balancing benefits and risks. We are very careful to try to do that in concert with our individual patients' wishes. We need to make a decision and try as best as we can to choose the best treatment for initial therapy. If that initial therapy is not successful, we can then present options again in the order that we think is likely of benefit.

The time is here to start thinking about using these drugs as initial therapy. We cannot do it in every single patient, of course. We have to be very careful about individual patient differences, characteristics, and wishes. Having these drugs gives us another way to fight their cancer. It is a real step forward for those patients who experience major benefit, because not only can they have good remissions, but they are long and, in some patients, can lead to disease-free status.