I just recently wrote a post (here) that describes how I became convinced that under certain circumstances there could be a genuine value in determining whether a particular lung cancer patient has a tumor with an EGFR activating mutation. While these have seemed to predict that these patients are quite likely to respond with dramatic tumor shrinkage, correct about 70% of the time, I had previously been impressed by how correlated EGFR mutation results were with some other clinical factors, such as being Asian and especially a history of never-smoking. But recent results that I described in my last post convince me that 1) mutations trump clinical factors 2) patients who receive EGFR inhibitors as first line therapy but don't carry an EGFR mutation don't seem to do as well as those who received chemo. I also said that this doesn't mean that patients without mutations don't benefit from an oral EGFR inhibitor like tarceva (erlotinib) or iressa (gefitinib) in the second or third line setting (the evidence is far stronger with tarceva, though, showing a survival benefit vs. placebo that hasn't been shown with iressa). Member sunnyside then asked about the evidence with regard to EGFR mutations and second/third line treatment. Here's what we know.

The original publication of the pivotal BR.21 trial that directly compared tarceva to placebo in previously treated patients with advanced NSCLC (2/3 receiving the active drug) (abstract here) was published in the New England Journal of Medicine along with a companion article about molecular and clinical correlates for this trial (abstract here). They tested a total of 325 tumors of the 731 total on the trial, but only 177 had tissue available for mutation testing (again, one of the major limitations of molecular testing is a lack of available tissue in the majority of patients with advanced NSCLC, unless we do an extra biopsy procedure just to get more tissue). The end result was rather puzzling, in that it showed that patinets without mutations had a significant benefit from tarceva, but there wasn't a suggestion of a difference from patients who had a tumor with an EGFR mutation: (click to enlarge) This was exactly the opposite of what you'd expect, and part of the reason why it's been challenging to build any consensus about the role of EGFR mutations. In fact, in their paper, none of the methods of molecular testing (EGFR immunohistochemistry to detect protein expression, FISH testing to assess gene expression, or mutations to determine problems with the gene itself) was found to be a significant factor in a multivariate analysis of survival (an analysis that teases apart the importance of multiple variables that are often correlated with each other). In fact, though, there were some questions raised about the quality control in this analysis, since several patients with reported EGFR mutations had new ones that had never been seen before, leading people to wonder whether these analyses were done properly and, even if so, whether these previously unrecognized mutations should be pooled with the ones established as predicting a high likelihood of clinical benefit from EGFR inhibitors. In particular, two EGFR mutations have been commonly reported and associated with favorable results: a deletion of a portion of "exon 19", and a substitution of an amino acid (arginine) in "exon 21". Exons are portions of the gene that are expressed in protein synthesis. The same group out of Toronto just published a revised analysis of their molecular testing in the BR.21 trial (abstract here). This time, they refined several aspects of their initial report, collecting tissue to test a total of 206 patients and running two different tests to detect EGFR mutations. Finally, they only included the well recognized exon 19 and 21 mutations in their category of EGFR mutations and pooled the few otherwise unrecognized ones in with the "wild type" (normal) EGFR group. Interestingly, one patient previously noted to have a mutation couldn't get that result replicated despite several attempts, and 11 of 148 patients previously reported as without mutations were found to have an exon 19 deletion. This highlights another shortcoming of molecular testing: you'd really like it to be so reliable that repeat testing doesn't give different results, but the reality is that different parts of the same tumor, and certainly some tumor areas and not others may have different genetic findings. After the repeat analysis, it was again quite clear that the patients without EGFR mutations received a benefit vs. placebo, while those with mutations at least looked better, but it didn't seem that the benefit was overwhelming. If anything, the benefit with tarceva still wasn't as clear as was seen in the "wild type" patients without EGFR mutations: In light of how different these findings are from what you'd expect, where the EGFR mutation carriers should receive a major benefit, I still don't know exactly what to conclude from this. You can say, though, that it definitely doesn't appear that you need to have an EGFR mutation to benefit from tarceva as a second or third line therapy. The fact that male smokers with squamous tumors got the same overall survival benefit as the overall trial population (described further in one of my earliest posts) also highlights that the benefits aren't restricted to the few patients with EGFR mutations. I suppose you could ask whether something happens that makes a patient with an EGFR mutation highly likely to benefit if an EGFR inhibitor is given first line but not second line or later. That would explain the very discordant and disappointing results for EGFR mutation patients on the BR21 trial, but I really doubt it. Another trial that was presented from Spanish investigators (described in a recent prior post) showed that patients selected for tarceva after testing and being found to have an EGFR mutation actually had as dramatic a survival benefit if they received it as second or third line therapy as if they received it first line. Those of you reading my post in the link from the paragraph above will see that my acknowledgment of the potential value of EGFR mutation testing in some settings represents a real change from my prior view. But I think it's important to avoid being so mired in a mindset that you can't be swayed when new information emerges that justifies a change in your conclusions. So there you go. But the puzzling results of the patients with EGFR mutations on the BR.21 trial still raises some flags about the limitations of EGFR mutation testing. I think we still haven't ironed out the reliability issues, and the lack of available tumor tissue will remain a problem until we change our standards for how much tissue we need to obtain in our initial workup of patients with lung cancer. If we are potentially going to move to an approach of using the specific tissue findings to guide our treatment algorithm, we're going to need more than just a few cancer cells.