In my last post I wrote about the prognostic value of molecular markers like EGFR and K-Ras that have generally been studied in patients with advanced NSCLC and treated with EGFR inhibitors, but these studies looked at prognosis in patients with early stage NSCLC who underwent surgery. These studies also provided some interesting results on the prognostic value of some clinical variables as well.

The Japanese surgical series of 397 patients with resected adenocarcinomas (abstract here) reported several associations of survival with clinical variables, as shown in the figures below:

(Click to enlarge)

Some of these findings are very intuitive. In the top right, we see that patients with stage I adenocarcinomas have a far superior survival to patients with higher stage cancers. Since staging is designed as a method to predict prognosis, these results corroborate what we'd expect. I've also written a prior post about more poorly differentiated cancers being associated with a worse prognosis than better differentiated cancers, as is shown in the bottom right panel. And there has been a growing collection of evidence that, as a population, women with lung cancer have a more favorable prognosis, stage for stage, than men (see prior post), as shown in the panel on the lower left.

In this retrospective review, the patients who had never smoked had a significantly better survival, as shown in the upper left (and, just to reiterate a striking result, 48% of the patients in this Japanese series were never-smokers). This report excluded patients who were known to have started an EGFR inhibitor, so this improved survival presumably represents a more favorable underlying cancer biology, and/or a lower risk of dying from some other smoking-related health risks. And the US-based trial from Memorial-Sloan Kettering in NYC showed the same association of better survival in never-smoking patients with resected NSCLC:

Both the Japanese and the Memorial studies also showed, as I described in my prior post, that the survival of patients with a tumor that has an EGFR mutation was significantly better than the survival in patients without an EGFR mutation, and this was correlated with smoking history. In fact, the Japanese study showed that the more significant a patient's history of smoking, the lower the chance of them having an EGFR mutation and the higher the chance of them having a tumor with a K-Ras and/or p53 mutations, which are both associated with a worse prognosis:

Interestingly, the association of a gradually decreasing likelihood of an EGFR mutation with increasing smoking status was first reported from -- you guessed it -- Memorial Sloan Kettering. This goes back a few years, and it focused on patients with advanced NSCLC, but it showed that while never-smokers were the group with the highest frequency of EGFR mutations, the smoking population wasn't all the same, but rather had a clear association of greatest smoking history with lowest likelihood of an EGFR mutation:

Overall, I think it's important to recognize that the favorable results among never-smokers may be primarily driven by the association of never-smoker lung cancer with the more favorable biology of an EGFR mutation. But it's also important to recognize that separating patients into ever-smokers and never-smokers is an oversimplification.

As is the distinction between clinical predictors and molecular predictors, since they are clearly associated with each other. Multiple different factors go into the mix of trying to predict the biology of a cancer, and we use all the information we can get in trying to individualize our recommendations.