We've come along way over the past decade. In the first half of the 1990s, the value of treating metastatic NSCLC was debated and not clear. A "meta-analysis" that pooled the results from 11 chemotherapy trials, 8 with cisplatin, and nearly 1200 patients demonstrated a modest but convincing improvement in survival compared to supportive care alone (article here). The figure summarizing the improvement by adding chemo is shown here:

(click to enlarge)

Although the difference now seems pretty convincing to me, and probably to you, at a time when treatment for advanced NSCLC was otherwise not felt to be beneficial, these results didn't take the world by storm. Fortunately, several new drugs emerged for lung cancer that were often well tolerated and had clear activity in lung cancer as single agents:

In oncology, if we see a benefit from cisplatin, and a suggestion of a benefit from one of these newer agents, our next step is to check whether the combination is better than one drug alone. So several trials after that were conducted that randomized people with advanced NSCLC to either cisplatin alone or cisplatin with another chemo drug, and these trials showed a clear benefit in response rate and survival for the doublets:

So we had several platinum-based doublets, several companies with new drugs that were all convinced their drug was the best to add to a platinum. At the same time, in addition to cisplatin, which was the old standard chemo drug for lung cancer, a newer cousin of cisplatin, known as carboplatin, also became available as a potentially suitable alternative to cisplatin, but with different and generally fewer side effects. It was time for a showdown!

The Eastern Cooperative Oncology Group (ECOG) developed a trial that would directly test four different chemo regimens against each other. This trial, known as ECOG 1594 (abstract here), compared cisplatin and paclitaxel (taxol) to cisplatin with docetaxel (taxotere) or gemcitabine (Gemzar), and a fourth arm with carboplatin replacing cisplatin in combination with taxol:

So this trial, with just over 300 patients on each arm and 1207 patients in total, would give us the answer to the question of what is the best two drug combination for advanced NSCLC. The results were clear:

Unfortunately, the answer was that all of the four combinations gave literally superimposable results. There were no meaningful differences among the groups across the main efficacy endpoints:

The only endpoint that met statistical significance was the progression-free survival of cisplatin-gemcitabine arm being about a month longer than the other arms. One issue, though, is that the cisplatin/gemcitabine treatment cycle was for four weeks, while the others were for three weeks, which meant that the follow-up scans after two cycles were two weeks later with that arm, so this makes the importance of this difference pretty questionable. The other issue is that the importance of a difference of a few weeks in progression-free survival is probably not a clinically significant difference compared to living longer or having fewer side effects. And there were only mild differences in side effects, overall favoring the carboplatin/paclitaxel arm, as we might have expected.

Over the past several years, many other trials have compared one doublet combination to another, and while there have been a few that showed minor differences, overall these trials have shown that they all perform remarkably similarly. There is no regimen of choice, but rather a choice of regimens, to be based on issues like particular side effects to avoid or schedule (some administer chemo once every three weeks, some on a weekly basis), and a physician's experience and comfort with a regimen (it's generally helpful to have an oncologist who is quite familiar and comfortable with the regimen you're getting). And they're all an appropriate option.

The one monkey wrench is the that Avastin was added to the carboplatin/paclitaxel regimen in the 4599 trial and shown to have a survival benefit in certain patients (see post reviewing this trial here). While I suspect that the survival benefit from Avastin will eventually be shown to exist when it is given with other chemo combinations, the US FDA approval is with carboplatin/paclitaxel. Because serious and even fatal toxicities can occur with Avastin even when used by the book, and we don't have as much experience with different chemo combinations with Avastin, I choose to use carboplatin/paclitaxel with avastin for patients who are eligible and want to pursue this approach. For people who aren't receiving avastin, I often give carboplatin and gemcitabine, which I find to be especially well tolerated by a majority of patients, but there's no one best choice when they all achieve very similar results.