In 2007 there was much excitement about the publication of a study by the researchers behind the landmark IALT adjuvant chemotherapy trial, which suggested that patients with early stage NSCLC could be divided into those who benefited greatly from cisplatin-based adjuvant chemotherapy and those who did not. The marker that defined these patients was called excision repair cross-complementation group 1 (ERCC1), a critical protein in the repair of DNA damage from platinum, and patients with low levels who did not receive chemotherapy after surgery had a much worse prognosis than those with high levels.

More importantly, the patients with high levels of ERCC1 did not seem to get any benefit from chemotherapy, with a survival numerically (but not significantly) worse than patients who did not receive chemotherapy. Patients with low ERCC1 levels, whose tumors presumably could not repair DNA damage as well, had a markedly better survival after platinum-based adjuvant chemotherapy compared to the group who didn’t get chemo. Dr. West covered this in a post in 2007, and I think accurately captured the state of excitement at the time on this potentially very useful test.

IALT ERCC1 Survival Curves

(Click on image to enlarge)

Of course, 2 years later we still do not routinely test for ERCC1, and presumably many patients get adjuvant chemotherapy that will be unlikely to benefit from it (or do not get it when they probably should). I’m not sure why this hasn’t been tested more aggressively, but it may be because of the underwhelming results to date from ERCC1-guided treatment of patients with metastatic NSCLC. In these patients, low ERCC1 levels seem to predict for increased response rate and time to progression, but don’t seem to have much impact on survival. This may be because of the heterogeneity of metastatic cancer, such that a biopsy to test ERCC1 in one lesion does not accurately reflect ERCC1 levels everywhere. Alternatively, ERCC1 may not be the whole answer…

At the 2009 ASCO meeting in Orlando, the IALT investigators were back with a vengeance. This time they presented data, again on patients from the adjuvant IALT trial, looking at expression of another DNA repair protein called Mut S Homolog 2 (MSH2) in the patients. MSH2 is also critical to the repair of platinum DNA damage, and while there is some overlap with ERCC1 levels, not all patients overlap with similar levels of the 2 enzymes.

What the investigators found was that levels of MSH2 had a very similar pattern of significance to ERCC1. Namely, patients with low MSH2 levels had markedly better survival with adjuvant chemotherapy, 16 months longer on average than the group not receiving chemotherapy. Those with high levels seemed to get no benefit, and in fact had a median survival 9 months shorter than those who did not get chemo, although this was not statistically significant.

The most interesting part of this was when they looked at both ERCC1 and MSH2 in the same patients, which they were able to do in 658 patients. What they found was essentially that both ERCC1 and MSH2 had to be low in order for there to be a benefit from adjuvant chemotherapy, and if both were low the median survival was 21 months longer in the chemotherapy-treated group. Patients with low ERCC1 or low MSH2 but high levels of the other enzyme (or if both were high) seemed to get no survival benefit from platinum chemotherapy!

Low ERCC1 Survival Curves on IALT

High ERCC1 Survival Curves on IALT

So should we be doing this now? This study would suggest that perhaps a third of the patients getting adjuvant chemotherapy currently actually get a significant benefit from it, and we may be able to tell who they are ahead of time. This probably isn’t quite ready for prime time (OK, I know I use that expression a lot: is anyone else getting sick of it yet?).

For one thing, this is only one study that has yet to be replicated by anyone else. Also, what are the alternatives to platinum chemotherapy for the patients with high levels of these enzymes? Should they get chemotherapy without platinum, such as gemcitabine and Taxotere? And where is the best cutoff of levels of these proteins? The authors used an arbitrary cutoff for saying tumors were “low” or “high” expressing, but can we truly say that there is no benefit to any of the patients on the borderline?

So the question REAL question is this: will this now prove to be enough evidence to stimulate some prospective trials using these markers? I know that trials looking at ERCC1 were planned, but have no idea where they are in development. This study would suggest that they should probably be amended to look at MSH2 also. Personalized medicine is not always about sexy new markers like EGFR and ALK. Sometimes it is about finding the right way to use what you already have!