As I mentioned in my last post on the recent results on pre-operative (neoadjuvant) chemotherapy, the results of this work failed to achieve statistical significance but did appear to be associated with a degree of benefit comparable to the magnitude of benefit seen with post-operative (adjuvant) chemotherapy, but the neoadjuvant trials were smaller and therefore underpowered. At the same time, there are differences in the patients who receive neoadjuvant vs. adjuvant chemotherapy. Specifically, the patient population getting pre-operative treatment are a broader population that hasn't had the patients with the most aggressive disease or marginal performance status drop out before getting to the post-operative therapy part. The patients enrolling on adjuvant therapy trials had to have gone through surgery, tolerated it OK, been found to not have more advanced disease, and still feel up to pursuing more treatment: they've already cleared several hurdles.

The best way to directly compare a pre-operative to a post-operative chemotherapy strategy is to directly compare the same patients randomized in the same trial. This was the premise of the Neoadjuvant vs. Adjuvant Taxol/Carbo Hope (NATCH) trial, which randomized 624 patients to either surgery alone, carbo/taxol for three cycles pre-operatively, or the same chemo post-operatively. Enrolled patients could have anywhere from stage IA (with at least a 2 cm tumor) to stage IIIA N2 NSCLC.

As expected, the trial demonstrated that far more patients intended to get both surgery and chemo received their planned treatment if they started with chemo: 90.4% received all three cycles, and another 4.5% received two cycles. In stark contrast, only 60.9% of the patients assigned to post-operative chemo received the planned three cycles, with another 3.8% receiving two cycles. That's a very significant drop-off that corroborates our presumption that one of the major benefits of pre-operative chemotherapy is that you can reliably deliver it.

Nevertheless, the trial didn't show any significant differences among the three arms of the trial. The disease-free survival (DFS) curves showed a trend toward more favorable results with chemotherapy, especially if delivered as neoadjuvant therapy:

(click on image to enlarge)

In numeric terms, the 5-year DFS was 34.1% for surgery alone, 38.1% for neoadjuvant therapy, and 36.6% for adjuvant therapy. Overall survival results also only showed minor trends in the same direction, again with neoadjuvant therapy slightly superior to adjuvant chemo.

What went wrong? Not only did pre-operative chemotherapy fail to demonstrate a benefit, but so did post-operative chemotherapy, which has previously been established as conferring a survival benefit. Various media reports highlighted that the trial was negative and that these approaches apparently don't work, but they provided no insight or explanation for why these results would contrast with other studies.

While it's possible that this failure was due to selection of carbo/taxol as the chemotherapy rather than cisplatin-based chemotherapy, I believe the most likely explanation by far is that the overwhelming majority of the patients enrolled were not appropriate patients for a trial to test the value of chemotherapy before or after surgery. We know that the patients who are the clear and consistent beneficiaries of adjuvant chemotherapy are those with stage II and IIIA NSCLC, who have a high enough risk of recurrence that chemotherapy can have a significant impact (see excellent review of chemotherapy in early stage NSCLC for a good summary). In contrast, those with stage I NSCLC are much less consistently shown to benefit, and in general the benefit has been in patients with larger stage IB tumors of 4 cm or greater, who have the highest risk of recurrence among patients with stage I NSCLC. There is no evidence that patients with stage IA patients benefit, and some evidence that they are actually harmed by receiving post-operative chemotherapy.

But unfortunately, 75% of the patients enrolled on NATCH had clinical stage I NSCLC (meaning, their imaging studies prior to surgery indicated they had stage I disease). Meanwhile, 25.9% of patients who had initial surgery (either alone or followed by chemo) was found at the time of surgery to have more advanced NSCLC (higher than stage IIIA N2), reduced to 19.3% with pre-operative chemo. These patients have disease that is too advanced to be an appropriate candidate for a curative approach of surgery combined with chemotherapy. So a huge proportion of the patients on this trial were inappropriate for the question, having either too high or too low a risk of recurrence to be the target population for modulating risk of recurrence with chemo compared with surgery alone. In fact, when only patients with stage II or IIIA patients were included, the DFS difference was far more impressive and in line with what we might expect (except a little low overall, I'd say): 25% for the surgery alone arm, 36.6% for neoadjuvant chemo (HR 0.81, corresponding to a 19% relative improvement), and 31% for adjuvant chemo (HR 0.87, or a 13% relative improvement).

Though a trial of 624 patients sounds big, when they're divided among three groups and the majority are inappropriate and not as likely to benefit as patients who are the best candidates for pre- or post-operative chemotherapy, you end up with a trial that doesn't answer the question at all because you can only detect trends that provide a glimpse of what the right trial would potentially show, diluted by patients who are either too likely to do well without more treatment, or too high-risk and likely to recur even with chemo.

Interestingly, there was no aspect of this trial in which post-operative chemotherapy looked better than pre-operative chemo. If you interpret an element of truth to the study, it's hard to escape the idea that neoadjuvant chemotherapy is at least as good as adjuvant chemotherapy, and potentially better if a properly powered trial with the right patient population had been conducted. This may simply be because 1/3 more patients actually received the intended treatment.

But for now, this is what we're left with. My interpretation is that while post-operative chemotherapy is the standard of care, pre-operative chemotherapy is an appropriate alternative in certain circumstances that I believe is every bit as good, if not better.