Over the past several years, probably the biggest development in the field of NSCLC has been the recognition of the importance of molecularly-defined subgroups that help define the clinical patterns of how patients are likely to do with various treatments. We've seen this clearly illustrated with EGFR mutations vs. EGFR wild type (no mutation), and more recently with the very uncommon but clinically important ALK rearrangements.

One newly defined clinical setting that has emerged has been the group of patients who experience a very good response to an oral EGFR tyrosine kinase inhibitor (TKI) like Tarceva (erlotinib) or Iressa (gefitinib), generally but not necessarily those with an activating EGFR mutation (most typically on exon 19 or 21, as described in this great summary by Dr. Nate Pennell) who respond for a period of months or years and then develop resistance. Why does it happen, and what are the leading options for managing this situation?

Early work on this problem of acquired resistance (as opposed to primary resistance, which is the situation in which a person starts out by not being very responsive to an EGFR TKI), largely out of the Harvard hospitals in Boston and Memorial Sloan-Kettering Cancer Center in New York City, have shown that about half or slightly more develop a resistance mutation called T790M on exon 20 (and a minority of people with an activating EGFR mutation also have this from the beginning, likely helping to explain why the response rate with one is not 100% but more like 70-75%). Another 10-20% or so are found to have over-expression of c-MET (targeted by agents like ARQ-197), reflecting a pathway that can bypass the EGFR signaling cascade as a mechanism for resistance.

A very recent publication by a group of investigators out of Memorial Sloan-Kettering offers another important piece of this puzzle, demonstrating that the clinical setting of "acquired resistance following a response to an EGFR TKI" is heterogeneous as a function of whether patients have a demonstrated T790M mutation or not, with different clinical behavior for these two groups. They, like the folks at Dana Farber and Massachusetts General Hospital, have begun to routinely re-biopsy tumors of patients who demonstrate progression after responding well to an EGFR TKI, and this paper reports on the clinical patterns of outcomes for 93 such patients biopsied and followed clinically over several years. This work is really the first that carefully describes outcomes for such a cohort of patients with acquired resistance.

The authors found a median overall survival (OS) in the entire group of 16 months, and that the 58 patients (62%) who had a T790M mutation had a significantly longer median overall survival after progression (median 19 vs. 12 months, p = 0.036).

(click on image to enlarge)

In addition, there were differences in the characteristics of the patients who were found to have a T790M mutation at the time of progression: those without a T790M mutation were significantly more likely to have progression in a new organ system vs. progression of previously identified lesions alone(p = 0.010), and they were more likely to have a worse performance status than those with a T790M mutation (p = 0.007).

The authors also checked whether there was a difference in the duration of an EGFR TKI-induced response in patients who ultimately developed a T790M mutation vs. another cause of acquired resistance, which would suggest that development of these resistance mechanisms may develop before progression is manifested. However, the two groups had a pretty similar duration of response on an EGFR inhibitor of 11 and 14 months for this without and with a T790M mutation, respectively (p = 0.10, non-significant). But because there is often a vague period of mild progression following a good response to an EGFR TKI that makes it very hard in many cases to firmly define when resistance has developed, the authors also looked at the harder endpoint of survival from the time of the start of EGFR TKI therapy and saw significant difference that favored those with a T790M mutation (39 vs. 26 months, p = 0.007).

The Memorial investigators also provided several important caveats. First, they noted that there has been some variability in the frequency with which T790M mutations and other molecular markers have been detected, probably largely as a function of the differences in techniques for detection. In fact, they noted that their own technologies for mutation testing have evolved over the past several years.

A second point they made is that their practice has been to continue the EGFR TKI after patients have demonstrated progression on an EGFR TKI, a pattern that 87% of the patients in this series followed. However, they acknowledged that this isn't the prevalent practice at most institutions. In fact, as I've noted in several replies on the forums section over the past several months, this question of whether to continue the EGFR inhibitor even after progression remains without good evidence and has been an evolving debate. I had previously been more concerned that there may be an antagonism, a detrimental interaction, between EGFR TKI and concurrent chemo, but some emerging study evidence over the past year or two have suggested that there may be little interaction. Some thoughtful oncologists may favor continuing the EGFR TKI, while others may favor stopping it in such situations.

What this series most clearly illustrates is that there isn't a single uniform population of patients with acquired resistance to EGFR TKIs and that it makes sense to recognize this in our development of new trials that focus on acquired resistance. It may well be that some patients, perhaps those with a T790M mutation, are well served by continuing an EGFR inhibitor, while those who don't have a T790M mutation may derive no benefit from continuing one. We might also well imagine that the patients who have c-MET over-expression or some other non-T790M mutation-related mechanism of resistance are more likely to benefit from a c-MET inhibitor like ARQ-197, but that pooling together patients with very different mechanisms of resistance might make it easy to miss a real effect on a narrower population.

Of course, this paper also provides a potential answer to the question of why it might be worth doing a repeat biopsy on a patient with progression after they have developed progression on an EGFR TKI. Though I have historically been skeptical of doing tests that won't lead to a change in our management, this work elegantly highlights, similar to the EGFR mutation-based breakdown of the results from the IPASS trial, that molecular marker-based work is continuing to inform and refine our understanding of how to group patients with NSCLC and identify the best treatments based on narrower subpopulations. I believe this paper is a major step, though admittedly an early one, in our understanding of how we might move forward with strategies for treating patients who respond well to an EGFR TKI but then develop acquired resistance.