In my last post, I described the somewhat disappointing results for tarceva compared with chemotherapy in a trial of unselected advanced NSCLC patients with a marginal performance status. However, EGFR tyrosine kinase inhibitors like iressa and tarceva were developed as targeted therapies, so perhaps they might prove to be more effective if used selectively, in a targeted population. That targeting might be based on clinical characteristics like using it in never-smokers or bronchioloalveolar carcinoma (BAC), or it might be based on molecular markers like mutations in the EGFR gene or overexpression of the number of copies of the EGFR gene, as determined by fluorescence in situ hybridization, also known as EGFR FISH testing. All of these methods have been employed in early but very promising studies of iressa or tarceva in selected populations.

I have already reviewed (in a post here) the activity of both iressa and tarceva in advanced BAC, where EGFR inhibitor therapy is generally used early in treatment, either before chemo or after one line of chemo. In the main trials of advanced BAC, iressa and tarceva have a response rate in the 15-25% range, and many of these are very prolonged responses. Another trial (Lee and colleagues paper here), clinically selecting based on never-smoking status, gave iressa to 37 previously untreated patients in Korea. In keeping with the usual findings for the never-smoking population with lung cancer, the patients on the trial were primarily women (92%, in fact), and all of the patients had adenocarcinomas or adeno/BAC mix. The response rate in this group was 69% (!), and the median time until progression was about 7.5 months, with an estimated one year survival of 73%. Interestingly, 7 of the 10 patients with brain metastases had shrinkage while on iressa. Also, while we have often thought of BAC as being particularly responsive to EGFR inhibitors, the response rate was actually higher among the patients with adenocarcinomas that were not mixed with BAC (76% vs. 43%). Another trial with iressa was done on the other side of the world, in Italy. Cappuzzo and colleagues (abstract here) selected patients for gefitinib (38% as first-therapy, the others after 1-2 prior lines of chemo) on the basis of either never-smoking OR being positive for EGFR by FISH or having a downstream marker for EGFR activity called AKT. Thirty-six of 42 patients were never-smokers, and about two thirds also had molecular markers that predicted doing well. On this trial, the response rate was 48%, the median time to disease progression was over 6 months, and the one year survival was 69% (particularly notable since 2/3 of the patients were getting this as 2nd or 3rd line treatment). On this trial, it appeared that molecular features trumped smoking status, since the six smokers who had EGFR FISH positive tumors did just as well as the never-smokers, and the response rate among never-smokers was remarkably better if their tumors were FISH positive (71%) vs. FISH negative (12%). At our big national US-based oncology conference this past year (ASCO), another rather small trial by Paz-Ares and colleagues generated big interest. This study came out of Spain and started by reviewing the records of over 1000 patients with advanced NSCLC, from which they were able to obtain tumor tissue from nearly 300 patients. Of those, 37 had EGFR mutations, and those patients were treated with first-line tarceva at 150 mg per day. As in the other trials, these patients shared many characteristics: 70% were never-smokers, 65% were women, and 88% had adenocarcinomas or BAC. The efficacy of tarceva in this limited group was truly remarkable, with 82% of patients responding with >50% tumor shrinkage(95% response rate for patients with exon 19 mutations, compared with a 70% response rate for those with exon 21 mutations), a median time of disease progression of over 13 months, and a one-year survival of 82%. Never-smokers were more likely to respond than current or prior smokers (90% vs. 70%, respectively). Given the challenges we currently face in dealing with advanced lung cancer, it is hard to imagine offering a treatment for advanced lung cancer that you could expect 4 out of 5 patients to respond to, and in whom you might expect to have them still doing well on it a year later. But that seems to be possible if you identify a well-selected population and provide a targeted therapy right for them. The problem? The Spanish trial started with over 1000 patients, among whom only 30% had tumor tissue to work with, ending with only 37 patients with the EGFR mutations who were left as candidates for the trial. The Spanish Lung Cancer Group plans to follow-up this trial with a larger one randomizing advanced NSCLC patients with an EGFR-mutation-positive tumor to tarceva or standard platinum-based doublet chemo as first-line treatment. But to get the hundreds of patients needed for that trial, they'll need to start with several thousand patients, given all of the hurdles. I'm afraid this type of trial will be very hard to complete. However, given the groundbreaking results seen when we match the most well-suited patients for a targeted therapy, we need to keep working to define the patients most likely to be great beneficiaries of targeted therapies.