In my last post I described the results of the ChEST trial that showed a borderline statistically significant improvement in survival of patients who received cisplatin/gemcitabine chemotherapy for stage IB to IIIA NSCLC prior to surgery. This study was very similar to another neoadjuvant chemotherapy trial, known as SWOG 9900, which also randomized patients to upfront surgery or 3 cycles of pre-operative chemotherapy followed by surgery. In the SWOG trial, run at a few hundred sites throughout the US, the chemotherapy used was the commonly used carboplatin/taxol combination, for three cycles prior to surgery, compared to surgery alone:

Also just as with the ChEST trial, SWOG 9900 was closed to accrual quite early (2004), with only 354 of 600 planned patients enrolled, because it was felt to be unethical to continue enrolling patients onto a trial on which half of the patients would not receive chemo, in the wake of multiple positive trials that had recently been presented and showed a significant benefit from the addition of chemo.

Dr. Kathy Pisters from MD Anderson Cancer Center in Houston led the trial and presented the data at ASCO 2007 (abstract here). Importantly, 2/3 of the patients had stage IB/IIA NSCLC. This was a higher percentage than comprised the European adjuvant trials. This is because North American trials separate out the patients with N2 nodes involved and treat them separately (the stage IIIA patients on SWOG 9900 had T3N1 NSCLC), while many stage IIIA, N2 European patients receive surgery and approach them the same way as stage II patients.

Surprisingly, only 77% of patients received all three cycles of carbo/taxol before surgery: this isn’t much better than the ~70% rate of successful administration in many platinum-based trials in the adjuvant setting, and it’s notably lower than the 85% rate of delivering 4 cycles of the same carbo/taxol post-operatively in another trial (CALGB 9633, described in a prior post here). It’s also lower than the 85% rate of delivery of the three cycles of neoadjuvant cisplatin/gemcitabine given in the ChEST trial. But for whatever reason, their delivery of pre-op carbo/taxol was less than we might have hoped and expected.

At the same time, the response rate was 38%, very similar to that seen in the ChEST trial of cis/gem (35%). And several patients were bothered by problematic muscle and joint aches, as well as neuropathy. Most concerning, though, were the three deaths during pre-op chemo, which is really higher than you’d expect for treating 169 early stage patients, especially with carbo/taxol. There was also no decrease in the rate of pneumonectomies on this trial with chemo (a potential benefit of neoadjuvant chemo is that you might shrink cancers that need a pneumonectomy before chemo but only require a lobectomy after chemo); that rate was 17% in both arms of the study. There were also a few more peri-operative deaths on the chemo arm – specifically, 4 of the 24 patients (17%) who had a pneumonectomy after chemo died, while none of the 25 patients who had a pneumonectomy without prior chemo died in the weeks after surgery. There’s still some debate about whether certain types of chemo increase the rate of surgical complications later, but in truth the 0% death rate after pneumonectomy on the surgery alone arm was unusually good. The mortality rate after pneumonectomy is generally in the 4-10% range.

Despite these challenges, the progression-free survival (PFS) and overall survival (OS) rates were better for recipients of chemo compared to those who received surgery alone. Neither result was quite statistically significant, but the trends were clear and convincing (to me), and they appear to be of the same magnitude as the benefits seen with adjuvant chemo. The 5 year PFS benefit was 10%, and OS benefit was 7% with addition of chemo:

A potential issue that Dr. Pisters raised in her discussion of the SWOG 9900 trial at ASCO last year is that the trial may have been more positive if a cisplatin-based doublet had been used instead of carboplatin, since there is some evidence that cisplatin is a little more effective than carboplatin for NSCLC, which may be the difference of a few percent more patients being cured, even if it’s a more challenging approach in most settings.

My overall impression is that the results of both SWOG 9900 and the ChEST trials look very similar to those of recent adjuvant chemo trials, but the numbers here are too small, since these trials both closed early. At the same time, I expect that the SWOG 9900 trial was very limited in being able to show much of a difference because 2/3 of the patients were stage IB and IIA patients who probably didn’t get much of a benefit from chemo, if this trial is similar to the ChEST trial in that respect (see prior post). We’ve never seen a breakdown of efficacy outcomes as a function of a patient’s stage, but I’d love to see this analysis done. We might well find that the chemo has striking benefits in higher risk patients and nonexistent or even harmful effects in lower risk patients.