The question of whether to use cisplatin or carboplatin in our "platinum-based chemotherapy doublets" that are the most common treatment for the first-line treatment of NSCLC has been a smoldering debate in lung cancer for more than a decade. Although at this point carboplatin is by far and away more commonly used than the generally less tolerable cisplatin, whether these are completely identical in their efficacy isn't entirely clear. Nobody questions that they're very close. And the reason most oncologists feel that carboplatin is the best choice is that if there is a difference, it's a slight one, felt to be more than offset by the toxicities and inconveniences of cisplatin compared to carboplatin, including considerably more nausea and vomiting, increased risk of kidney damage, as well as risk of hearing loss and peripheral neuropathy (nerve damage with numbness/tingling). Another factor that is important to many people is that cisplatin is administered over many hours, with lots of IV fluid support, so it usually requires either long days in the outpatient infusion center or a night or two in the hospital, while carboplatin is typically just a one-hour outpatient IV infusion.

Let's start with one figure that is perhaps the best summary of a bleak 5 year period of lung cancer research:

(Click to enlarge)

This figure of overall survival, from the ECOG 1594 trial led by Dr. Joan Schiller (abstract here), randomized over 1200 patients with previously untreated advanced NSCLC to any of 4 platinum-based chemo doublets, 3 with cisplatin and the 4th being carbo/taxol. As you can see from the figure, the survival curves for each of the four chemo combinations performed identically, with completely superimposable survival curves. This image is what oncologists think of when we consider the differences among the different platinum-based chemo doublets. They all get you to the same place.

But there have always been nagging questions about whether cisplatin may be just a shade more active. One well-publicized trial, known as TAX 326 (abstract here) and conducted by Aventis (now Sanofi-Aventis), the makers of taxotere, was very similar to ECOG 1594 in that it randomized 1219 previously untreated first-line patients with advanced NSCLC to cisplatin/navelbine, cisplatin/taxotere, or carboplatin/taxotere:

The news from this trial was that the overall survival with cisplatin/taxotere was significantly better than the "control arm" of cisplatin/navelbine. Having one doublet do significantly better than any other is pretty rare, so that's why some people would perhaps consider the cisplatin/taxotere doublet to be among the most active combinations for NSCLC, albeit by a small margin. The carboplatin/taxotere doublet was also compared to cisplatin/navelbine and wasn't significantly different, which isn't a great surprise. The issue, though, is that if you try to actually compare cisplatin/taxotere to carbo/taxotere, which the trial doesn't formally do because it was designed to just compare each of the taxotere-containing arms to cisplatin/navelbine, the carbo/taxotere arm appears notably worse: median survival of 11.3 vs. 9.4 months and one year survival 46% vs. 38%. However, this has never been highlighted.

Another trial that did directly compare cisplatin to carboplatin in a doublet with a mutual partner was conducted by Rafael Rosell and colleagues in Spain (full article here). In this trial, 618 first line NSCLC patients were randomized to cisplatin/taxol or carboplatin/taxol, and the results showed a significant survival benefit with cisplatin:

As noted in the slide on the right, there was more nausea/vomiting and kidney toxicity with cisplatin, and more of a problem with low blood counts with carboplatin. In this trial, they also measured quality of life and didn't see any significant differences.

So as we sometimes do when we have a complex mix of studies that show varying results, a meta-analysis was performed by Ardizzoni and colleagues (abstract here) in which the investigators pooled together the results from 9 different trials that included 2968 patients who were randomized to a cisplatin- or carboplatin-containing arm of the same trial. These meta-analyses are done to give us a summary sense of what's going on when you look across several studies that may be too small individually to detect real differences. Pooling the results of all of these cis vs. carbo studies together, the response rate with cisplatin combinations was higher (30% vs. 24%), and the overall survival benefit was 7% better with cisplatin, although that survival difference wasn't statistically significant. In the figure below, the size of the squares represents the size of each included trial, the horizontal lines to the sides of the boxes represents the variability in the data (so long lines means very inconsistent results), and the position of the square to the right or left of the vertical line represents whether each trial showed a survival benefit for cisplatin or carboplatin, respectively:

Looking at different subsets, though, the patients with non-squamous lung cancers and those treated with the most modern chemo agents as a partner for the platinum (like the taxanes, gemcitabine, or navelbine) had a survival with cisplatin that just reached statistical significance:

Maybe that really means something, but we always want to take subset analysis with a grain of salt, because those subset analyses are sometimes done when the orignical question didn't give you the answer you wanted, so you ask the question again 10 different ways. As the saying goes, "if you torture the data enough, eventually it'll tell you whatever you want." There may be something to these subsets, but the overall results are that the differences are right on the cusp of being statistically significant.

The real issue is whether the differences between cisplatin or carboplatin are clinically significant even if they are on the border of statistically significant. Some of the trials suggest a difference of perhaps a month, while others suggest less than that, in a setting where we know treatment isn't curative. As an increasing number of drugs become available and effective in improving survival in the second and third line setting, the small differences would be expected to be diminished further: later treatments after first line chemo should be an equalizer. Now that we have the opportunity for several lines of treatment for advanced NSCLC, I am partly thinking about preserving something for round 2 and round 3 as I make recommendations for first line treatment, rather than completely beat people down by making first line treatment more toxic than it needs to be. Finally, these trials have generally included very few older and/or sicker patients, so they tend to evaluate a population that is not always representative of a "real world" patient population that would likely tolerate cisplatin as well. But perhaps these considerations sound more like rationalizations to some people.

My perspective has been that I do recommend carboplatin-based chemo for the majority of my patients in the advanced disease setting, where cisplatin can't increase the cure rate, and the differences in efficacy are relatively small and, from my vantage point, offset by the often problematic toxicity of cisplatin, and that with the vast majority of my patients receiving later treatments, the differences in first-line treatment shoudl be smaller still. Finally, now that avastin is approved and well tested with carbo/taxol and shown to confer a significant survival benefit, this is the regimen I use for appropriate patients, and we don't have a hint of a cisplatin combination with avastin being superior. While the combination of cisplatin/gemcitabine with avastin has been shown to be feasible and associated with an improvement in progression-free survival compared to cisplatin/gemcitabine alone (abstract here), we don't have any data yet on overall survival with this combination, and definitely no hint it's better than carbo/taxol/avastin.

However, I do believe in my heart of hearts that cisplatin is marginally superior, and I think this actually matters much more in the setting of curative treatment, such as adjuvant chemo after surgery or as multi-modality treatment with radiation and/or surgery for stage III NSCLC. I think if a patient can tolerate it, pushing hard and accepting greater short term (and perhaps also long-term) side effects if it can translate to a cure rate that is even a few percent higher is likely worth it. It's possible that carboplatin-based chemo regimens are equally effective, and I'm sure they're close, but I tend to be conservative and offer the most aggressive option that makes sense in the curative setting.

I'm changing the polling question now to see whether people would favor a more aggressive, toxic treatment for marginally better activity in the curative setting, in any setting, or whether they would prefer to accept a better balance with potentially a little less efficacy but a considerably better side effect profile. I'd love to hear what people think of this issue, not only in their poll answers, but also with any comments people have. Oncologists have been debating these issues for many years, to the point that it's become like a "Coke vs. Pepsi" debate, but we haven't really heard from the patients and family members most directly affected.