Since the earliest clinical trials of EGFR inhibitors in NSCLC, certain clinically defined patient subsets became identified as more likely to show a benefit than others. Such studies suggested that women, patients with adenocarcinomas rather than squamous cell carcinomas, Asian patients, and never-smokers compared with current or former smokers were the patients who would do well with EGFR tyrosine kinase inhibitors like gefitinib (Iressa) or erlotinib (Tarceva). Since that time, we have been debating whether these targeted therapies should be used primarily or exclusively in selected populations or whether they should be used in general lung cancer populations. The correlation of rash with better outcomes is another interesting issue that is a big enough topic I’ll discuss it separately in the near future. Since it’s not something you know about before you start treatment, unlike patient sex race or subtype of lung cancer, it’s a different issue.

The populations that have been identified as particularly likely to do well with EGFR inhibitors emerged very early in the development of these agents, such as in the pair of phase II studies known as the IDEAL (Iressa Dose Evaluation in Advanced Lung Cancer) trials. Each of these enrolled over 200 patients, with one done in Japan and Europe (IDEAL-1 abstract here) and the other done in the US (IDEAL-2 abstract here) (see figure below for trial design for both).

(Click to enlarge)

IDEAL-1 showed that the response rate to Iressa was 27% in Japanese patients compared with only 10% in non-Japanese patients, but that this difference was not as significant as the differences in response rate favoring women over men, patients with adenocarcinomas over squamous carcinomas, and healthier patients over less healthy ones (“performance status”, or the ability to perform the activities of daily living without assistance or significant symptoms). The US-based IDEAL-2 trial also showed women receiving iressa to be more likely to have symptomatic improvement (50% vs. 31%) and significant tumor shrinkage (19% vs. 3%) compared with men, and also that patients with adenocarcinomas were more likely to have symptomatic improvement (43% vs. 30%) and tumor shrinkage (13% vs. 4%) compared with those who had squamous tumors. Meanwhile, a retrospective review from Memorial Sloan Kettering Cancer Center in New York (abstract here) showed that in reviewing how 139 patients with advanced NSCLC over a five-year period responded on single-agent iressa, the factors of never-smoking, being female, and having adenocarcinoma and especially the bronchioloalveolar carcinoma (BAC) subtype of adenocarcinoma were significantly more likely to respond well to Iressa than other patients.
So the results with regard to tumor shrinkage are pretty clear. However, it gets less clear when you look at survival. Here, very interesting results come from the randomized BR.21 trial I mentioned in a prior post, where previously treated patients with advanced NSCLC were randomized to receive tarceva (two thirds of patients) or a placebo (one third of patients). As described in my last post, overall there was an approximately two month survival benefit for the general population, although some got no benefit or were even harmed by the active drug, while some other patients got much more of a benefit. When looking at different patient groups, you see that there are big differences in likelihood of tumor shrinkage, but generally NOT differences in survival benefit. Looking at the curves for women vs. men, for instance, as shown below, you can see that while the women had a likelihood of response to tarceva (abbreviated as ORR, for objective response rate, on the figure) that was more than twice that of the men (14% vs. 6%), the curves separate the same amount, showing that the survival benefit is the same between women and men. On the figure, you can also see something called a hazard ratio, abbreviated as HR on the figure. This is a number that reflects the improvement or worsening of survival from a treatment, and decimals lower than one means that there is a benefit to the treatment, and the lower that decimal, the better. A HR of 0.70 means that there is a 30% improvement in survival overall during the period of follow-up if patients received tarceva vs. placebo. So what? So you can see that the HR for men is actually slightly lower than for women; this isn’t a real difference, but it shows that men got absolutely as much of a survival benefit as women did, even though they had a much lower likelihood of the cancer shrinking.

(click to enlarge)

The same exact situation occurs when looking at adenocarcinomas vs. squamous cell carcinomas. As shown in the figure below, you can see that the response rate is more than three times higher for adenocarcinomas than squamous carcinomas (14% vs. 4%), but the survival benefit is essentially the same (numerically slightly better, but not a real difference) for patients with squamous tumors.

One exception is with never-smokers, and the differences are shown in the figure below. On the BR.21 trial, never-smokers were WAY more likely to have significant tumor shrinkage (25% vs. 4% for current or former smokers), and they also had a more than doubling of survival compared with never-smokers who received placebo only (HR of 0.42 means a 58% improvement in survival over the course of the trial). In contrast, current or former smokers had a modest benefit compared with placebo, reflected by the tarceva curve a bit to the right of the placebo curve, but the HR of 0.87 is much closer to 1.0 (no difference in survival between the two arms) than the 0.42 seen in never-smokers.

Importantly, to underscore that the tarceva benefit is really not just in Asian women never-smokers with BAC, let’s look at what is pretty close to the opposite of that group. This last figure shows the difference between tarceva and placebo on the BR.21 trial in men with squamous cell carcinomas who smoke or previously smoked. If you think tarceva only works for narrow subgroups, these people should be very unlikely to benefit. And in fact they have a remarkably low likelihood of major tumor shrinkage, about 2%. However, you can see from the curve that there is a very impressive survival benefit in this supposedly “unfavorable” group, as much as that seen for the overall trial in general (and numerically, the HR in male smokers with squamous cell cancers is a bit lower than for the trial in general, so that suggests a slightly better benefit in this group).

I’ll have more to say about never-smokers, who get such a profound and consistent benefit from EGFR inhibitors that it’s arguable they should be treated differently than other folks with lung cancer. But my key take home points are these:
1) certain clinical subgroups are consistently more likely to have significant tumor shrinkage than others when EGFR tyrosine kinase inhibitors are given
2) patients in groups that are very unlikely to show tumor shrinkage from these drugs can still get the same degree of improvement in survival
3) this is presumably because keeping the cancer from growing, leading to stable disease for several months (and sometimes beyond a year), can translate to a significant survival benefit even without shrinking the cancer
4) this is not a situation of the overall benefit in the general population being propelled by a few patients who did amazingly well while the rest got no benefit. The survival benefit is much broader than that.

There’s a lot of information here, so please feel free to ask questions, raise objections, or give any comments about what I’m saying here. There are many people who are not giving EGFR inhibitors to people who don’t fit their idea of the kind of patient who responds to them, and I would contend that this is a mistake.