Please Note: New Treatments Have Emerged Since this Original Post

The question of which patients to test for an EGFR mutation, and when, has become one of the most timely and rapidly changing ones in lung cancer. The IPASS trial was a landmark study that definitively illustrated that molecular testing overshadows clinical selection for EGFR tyrosine kinase inhibitor (TKI)-based therapy, demonstrating that the 60% of Asian never-smoker or light former smoking patients with advanced lung adenocarcinoma on this trial that had an activating EGFR mutation demonstrated a significantly higher response rate and progression-free survival (PFS) with the EGFR tyrosine kinase inhibitor (TKI) Iressa (gefitinib) than with standard chemotherapy (carboplatin/Taxol (paclitaxel) in this particular study). In contrast, the 40% of patients with the same clinical characteristics on this trial but without an EGFR mutation had a significantly better response rate and PFS with standard chemotherapy. Overall survival (OS) results were preliminary and revealed the same mutation-dependent trends, though they were not significantly different in the approximately one third of patients who had tumor tissue available for mutation testing. My take on this work was that there was now compelling evidence that clinical selection of patients to receive chemotherapy vs. EGFR-based first line therapy is most definitely inferior to molecular selection, but this was just one study and a retrospective analysis of mutation status in only a limited subset of patients, and perhaps the trend of an OS difference could be reduced if we work to ensure that patients all get access to the same agents over time. Essentially, the question to me was that even if some patients are better served by receiving one treatment vs. another, and this can largely be predicted by the presence or absence of an EGFR mutation, does it really matter whether the optimal treatment for that particular patient is given first line or second line? Does order of treatment matter if everyone gets the opportunity to receive the same treatments over time? A few very recently presented studies corroborate the findings from IPASS in a population identified by prospective EGFR mutation testing and lead me to be newly convinced that, at least for patients with an adenocarcinoma or those with little or no smoking history, order of treatment can make a meaningful difference, so it's ideal to do molecular testing before starting systemic therapy.

First, the TORCH study randomized 760 previously untreated patients with advanced NSCLC who were not clinically or molecularly selected to either chemotherapy with cisplatin/gemcitabine followed by planned Tarceva (erlotinib) at progression, versus Tarceva as first line therapy with planned progression to cisplatin/gemcitabine at progression. (click on image to enlarge) This cross-over design isolates the question of whether order of therapy matters, because the same population is being assigned to the same combination of therapies over time. In fact, the trial, designed for non-inferiority with regard to PFS , was halted early by the Data Safety Monitoring Committee when emerging results demonstrated that the arm assigned to initial erlotinib followed by chemotherapy was associated with significantly inferior PFS (2.2 vs. 5.7 months) and OS (8.5 vs. 12.0 months, HR 1.36, p = 0.002). This is a clinically and molecularly unselected population, in which we can estimate that approximately 90% of the patients are EGFR wild type, which would be consistent with a more favorable response to initial chemotherapy. But there was a significant penalty in terms of overall survival even when patients were assigned to transition to chemotherapy after progression on first line Tarceva, perhaps explained by the finding that 49% of patients with progression did not ultimately receive chemotherapy as planned. Even among those patients who did receive cisplatin/gemcitabine after first line Tarceva, the response rate was only 9%, compared with 27% for the same regimen when administered as a first line therapy. The cause of this difference is unclear, but it may be related to such patients having a worse performance status and diminished ability to tolerate effective doses of this chemotherapy regimen (just speculation). Regardless of the cause(s), this study provides convincing evidence that unselected patients are harmed by foregoing first line chemotherapy, even if there is a plan to have them receive the same treatment opportunities over time. It reinforces the results we saw in the IPASS trial. The just published Japanese study by Maemondo and colleagues in the New England Journal of Medicine provides a similar argument in favor of initial EGFR TKI-based therapy for patients who we might prospectively detect as having an EGFR mutation. Like the TORCH trial, this study was terminated early, after an interim analysis demonstrated a highly significant prolongation of PFS (10.8 vs. 5.4 months, HR 0.36, P < 0.001), which was the primary endpoint of the study, for patients who received first line Iressa, compared with EGFR mutation-positive recipients of first-line chemotherapy with carboplatin/paclitaxel. Patients receiving Iressa also demonstrated a significantly higher response rate (73.7% vs. 30.7%, P <0.001), largely recapitulating the results of the IPASS study. Although the difference in OS did not reach statistical significance (30.5 vs. 23.6 months), the numeric difference of seven months despite the fact that 95% of patients assigned to initial chemotherapy received subsequent Iressa is suggestive to me that there is a real advantage to receiving the optimal therapy, as defined by EGFR mutation status, at the earliest opportunity. Statistically significant difference or not, I know which curve I'd rather be on. Together, this pair of trials provides evidence that there are adverse consequences to receiving first line Tarceva instead of chemo if you are EGFR wild type, and conversely it's potentially detrimental to delay EGFR TKI therapy until second line or later if you have an activating EGFR mutation. The reality is that this needs to be weighed against practicalities such as accessibility of tumor tissue and associated risks of additional invasive procedures, delays in instituting treatment, and other factors, but I believe that these challenges will be partly overcome by more widespread availability of EGFR mutation testing within 5-7 days, and a gradual shift to core biopsies over fine needle aspirates (tiny needles, but tiny tissue samples) as molecular testing becomes more critical. Also very important is the fact that the investigational agent crizotinib (PF-02341066), an inhibitor of anaplastic lymphoma kinase (ALK) activity, has now been shown to be associated with response rates in the range of 60%, with many very prolonged responses lasting months or a year or longer, in patients with an ALK rearrangement. Because crizotinib represents a potentially very active agent, but thus far only for those who have molecular testing that demonstrates an ALK rearrangement (which is seen in patients with similar demographics as those with an EGFR mutation but is almost completely mutually exclusive with EGFR and KRAS mutations), we now have another important reason to collect tissue for molecularly directed treatment opportunities. I expect that these opportunities will only increase as more and more of our targeted therapy trials assess for particular activity in biomarker-defined subsets and therefore require tissue for submission. Still, we need to also recognize that EGFR mutations and ALK rearrangements are far more likely to be detected in enriched populations, such as those with an adenocarcinoma and little or no smoking history. Such patients are the ones in whom molecular testing as early as feasible is a high priority to me. Molecular testing is still desirable, but to me not as imperative, in patients with an adenocarcinoma and a more significant smoking history, or those with a minimal or no smoking history and a non-adenocarcinoma histology. For patients with a non-adenocarcinoma histology and also a significant smoking history, the probability of either an EGFR mutation or an ALK rearrangement are low enough that I would say that molecular marker testing is not at this point of critical importance, especially if these patients have the opportunity to receive Tarceva in later lines of therapy, as would be appropriate based on its FDA approval and demonstrated survival benefit in a broad clinical population. Patients vary greatly in their pre-test probability of having an EGFR mutation or ALK rearrangement, so I believe it is appropriate to not have a blanket approach of requiring immediate testing of all patients, including those with a vanishingly low probability of finding a result that would change management.