Back when I first started doing this, one of my earliest posts (see here) was on the question of whether EGFR tyrosine kinase inhibitors (TKIs) (see introduction to this work in prior post). My point, which is still true, is that EGFR TKIs aren't only effective in a narrow population of patients, whether identified based on molecular or clinical variables. But in the more limited number of patients who have the "activating mutation" in the actual receptor, who might be expected to do especially well with an EGFR TKI, how do they do? How does a targeted therapy do in a precisely targeted population?

There are several studies that have now tested this by identifying patients who have an EGFR mutation and received an EGFR TKI. As shown in the table below, the results are very clear and consistent.

(Click to enlarge)

The vast majority of these were done with iressa (gefitinib), and most were done in Asia, where patients with EGFR mutations account for about 1/3 of patients with NSCLC and not only about 10%, as is the case in North America and Europe, where the few other studies were done. But they all tell the same story. If patients are screened and selected for an EGFR inhibitor on the basis of having an EGFR mutation, the group typically has a response rate in the range of 70-80%, a median progression-free survival (PFS) of 9-13 months. Median overall survival isn't reported in most of these trials because it's so long that it hasn't been reached in many of them -- patients clearly did unusually well.

There are several important conclusions to make from this work. First, not all of the patients with EGFR mutuations responded. The response rate of 70-80% is far higher than we expect to see in patients receiving chemo (generally in the 30% range), but it's not 100% either. We are still in the early stages of learning about the different particular mutations, and some are associated with better results than others -- it may be that this is a reason that the response rate isn't 100%.

At the same time, these responses also last for longer than most responses to chemotherapy agents, but most patients will show resistance to EGFR TKI therapy after a year or two. There are a minority of more fortunate patients who show no progression for many years, but they represent a small proportion of responders. And we are still struggling with ways of getting the genie back in the bottle and restoring responsiveness for people who previously did very well but subsequently progress on an EGFR TKI.

Still, it's important to highlight that EGFR mutations are most definitely not the whole story. As I noted in a prior post and will reproduce here, a table of great responders to iressa from one of the original publications that identified the EGFR mutations shows that one of these nine didn't have an EGFR mutation, and this was actually the patient who cotinued to respond for the longest of any of these patients:

Finally, as shown in the "waterfall plot" from a trial out of Memorial Sloan Kettering, in which patients with BAC or adeno/BAC who received tarceva (erlotinib), the best responders (with the furthest downward bar showing tumor size decreasing after treatment) were patients with EGFR mutations (gray bars), but many very good responders didn't have an EGFR mutation or gene amplification (EGFR FISH positive, here done slightly differently, with a technique abbreviated CISH):

So here are my take home messages:

1) Patients who are known to have EGFR mutations have a very high probability of having a very good response to an EGFR TKI, which typically lasts for many months or even 1-2 years, sometimes longer

2) Nevertheless, an EGFR mutation is neither necessary (some patients do very well with an EGFR TKI despite not having a mutation) nor sufficient (a minority of patients with an EGFR mutation don't have a good response to an EGFR TKI).

One other limitation is that in most recent trials that have tried to collect information on molecular variables, only perhaps 1/3 have tissue available. Outside of research-oriented centers, that proportion is likely to be even lower. In addition, mutation testing still requires a few weeks, and many patients with advanced NSCLC are reluctant to wait three weeks to develop a treatment plan.

Nevertheless, this work provides a clear indication that we can produce results like we've never seen in advanced lung cancer if we can deliver targeted therapies to a targeted, selected population most likely to do well with these treatments.