Over the past few years, the role of post-operative, also known as adjuvant, chemo has become increasingly accepted as a standard of care. Several trials have shown an improvement in survival at about 5 years that is in the 5-15% range for recipients of chemo. However, the evidence indicates that the people at higher risk receive more benefit, as you'd expect: the risks of chemo are the same no matter what stage cancer someone has, but if the chemo reduces the recurrence rate by a similar fraction for everyone, the person with a 60% risk of recurrent cancer is going to benefit far more than the person with a 25% risk of recurrence. And many of our trials have failed to show a benefit for patients with resected stage I NSCLC, at least for those with cancers smaller than about 4 cm.

I think many oncologists and patients think of adjuvant chemotherapy as either helping significantly or not helping significantly. But there is some limited evidence that demonstrates that some people really may be harmed by chemo, either during the time of chemo, or perhaps more distantly in the future. For instance, I described early work (prior post here) on a marker called ERCC-1 that is correlated with resistance to cisplatin (low expression is associated with sensitivity to cisplatin-based chemo and better outcomes after adjuvant chemo, while high expression is associated with resistance to cisplatin-based chemo and worse outcomes). In fact, that large subset analysis showed that patients who showed high ERCC-1 expression -- a pattern of resistance -- actually did modestly worse than patients who were observed and received no chemo. So it may not just be that we need to understand better who is going to benefit more and who will benefit less, but rather that we need to predict better because the ones we aren't helping we may actually be harming. We can overtreat as well as undertreat.

In fact, one of the more interesting presentations at ASCO on the subject of lung cancer was a report of the eight year follow-up of patients on the International Adjuvant Lung Trial, or IALT. This was presented initially at the plenary session at ASCO 5 years ago (abstract here), where we saw the first convincing evidence of a survival benefit from post-operative chemo, and then it was published in the New England Journal of Medicine (abstract here) as a new standard of care. It enrolled 1867 patients with resected stage I to stage IIIA NSCLC to receive post-operative observation alone or any of 4 cisplatin-based chemotherapy regimens:

At the time of the initial presentation and publication, the survival curves separated convincingly, with chemotherapy superior, and the five-year overall survival was 4.5% better for chemo recipients than for patients who were observed without any further treatment after surgery:

The "hazard ratio" of 0.86 correspondes to a 14% improvement in survival on the chemo arm compared with the control (no chemo) arm.

So Dr. Thierry Le Chevalier, a physician from France, presented the updated data from the same trial, now with eight year follow up (abstract here). The big news was that the curves came back toward each other, and the difference between the two curves was not statistically significant anymore:

So you can see from the figure above that there were fewer deaths on the chemo arm before 5 years, but a higher number after 5 years.

Why did this happen. He presented data that showed that patients who received chemo were significantly less likely to develop distant metastases (338 vs 378 cases for chemo vs. control arm), as you'd expect chemo to work. There were no differences in second cancers, so it doesn't appear that, at least in the time frame of up to about 8 years from treatment, chemo was causing more cancers later on (this may be an issue with longer follow-up). But there was a borderline significant (p = 0.06) increase of about 34% in the risk of non-cancer deaths in the chemo arm, although Dr. Le Chevalier didn't have more specific information to speak to the underlying causes.

In fact, there is a little more information that corroborates this point. A publication in press now in the Journal of Clinical Oncology (abstract here) performed a "meta-analysis", pooling the data from over 4500 patients on 5 recent large adjuvant chemotherapy trials and revealed that while cisplatin-based chemo was associated with an overall survival benefit, beyond 5 years there are more deaths among the chemotherapy recipients than those just observed -- this amounted to an increased risk of 1.4%, attributable to a range of non-cancer causes, such as more pulmonary emboli (blood clots in the lungs), heart attacks, infections, and a few odd individual events.

So what is the take home message? It's still appropriate to say that high risk patients can have their risk of recurrence reduced after surgery by receiving chemotherapy. But the benefit may be reduced over the long term because of problems caused by chemotherapy. That's still probably acceptable -- you need to survive beyond five years (when the threat of recurrent cancer is all but gone) to reach the point of higher long-term risk from chemo. But it's something to factor into the decision, especially if an individual patient is really at the lower limit for having enough risk to justify pursuing chemo. As these trials mature, we'll only get more long-term data, which may show that the lower risk patients are actually likely to be harmed by chemotherapy, even as the high risk patients were helped.

Next, I'll turn to a study on pre-operative chemotherapy that also highlights a difference between lower risk and higher risk surgical NSCLC patients.