Historically, the main task of pathologists in lung cancer has been to divide them into small cell lung cancer and non-small cell lung cancer. Beyond that, there is now more of an emphasis than there used to be on trying to clarify whether a NSCLC tumor is a squamous cell carcinoma, adenocarcinoma, large cell, or another subtype, partly because we now have drugs approved just for certain subtypes (avastin for non-squamous tumors only) and others that are commonly observed to have a higher response rate in some histologies (lung cancer categories based on microscopic appearance) than others (sucha s EGFR inhibitors consistently showing a high response rate in adenocarcinomas/bronchioloalveolar carcinomas compared with other subtypes). Several members have asked questions similar to the one reported in a recent paper from the Mayo Clinic (abstract here): is the grade of the tumor, ranging from well differentiated to moderately differentiated to poorly differentiated and even undifferentiated, associated with a better or worse prognosis for a person?

What do we mean by differentiation? The degree of differentiation of a tumor is how much it looks like normal, non-mutant lung cells. Well differentiated lung cancers are pretty close to normal appearing, while poorly and undifferentiated lung cancers look completely wild, not like the cell they originated from. It's important to note that pathologists can readily and reproducibly say a well differentiated cancer is an adenocarcinoma, for instance, but a poorly differentiated tumor may be said to resemble an adenocarcinoma by one pathologist, while another pathologist may think that the same tumor is a poorly differentiated squamous cell carcinoma (and even give different results when the same pathologist reads a tumor at two different times). So there is a good concordance (agreement) of pathology findings only for the better differentiated tumors. In addition, there are not clear standards of what constitutes moderate differentiation and what consistutes poor differentiation, so this will differ somewhat from one pathologist to another.

But for prostate cancer, for instance, Gleason score for prostate cancer, which is a measure of differentiation, is one of the most important predictors of outcome and is used routinely in predicting outcome and altering treatment plans. And the Bloom-Richardson score for breast cancer tumors is also widely used in a similar capacity. But we've been slow to formally recognize the histologic grade as an important predictor in lung cancer, even though most of us who treat a lot of lung cancer realize that it is a useful factor. For instance, in my prior post on bronchioloalveolar carcinoma (BAC) being potentially over-treated by conventional standards for NSCLC, I likened some indolent BAC tumors (which should be well differentiated, by definition) to a similar situation as low-grade prostate cancer, which we know can be present in many older men but not be a real threat to their survival because a low grade prostate cancer can be so slow growing.

So this study from the Mayo Clinic looked at over 5000 NSCLC lung tumors that came through that institution from 1997 through 2003, plus another 700+ that were diagnosed in that area but outside of the Mayo Clinic, all of varying stages and receiving various treatments. They collected information on the grade and histologic subtype (adeno, squamous, etc.) of the tumor, as well as stage, age and gender of the patient, smoking history, treatment received, and every other variable they could identify. They found that well differentiated NSCLC tumors were more likely to be adenocarcinomas, to be found in women and never-smokers, and were also more likely to be diagnosed at an earlier stage, while poorly and undifferentiated cancers were more likely to be seen men, to be squamous or especailly large cell, to be found in recent or current smokers, and to be diagnosed at a later stage. Of course, there was a lot of variability in all of these, but those are the trends. Holding the other variables constant in a multi-variable analysis, they found that patients with well differentiated NSCLC tumors (a total of 668 of 5018 total tumors) had a significantly better survival than patients with moderately differentiated tumors (a total of 2111), who in turn had a better survival than patients with poorly differentiated (1410 of 5018) or undifferentiated tumors (the remaining 829). Poorly differentiated and undifferentiated tumors did not differ significantly, both being nasty and the most threatening. The survival differences held true no matter whether you were looking just within the same stage, at just women or just men, looking only at adenocarcinomas or squamous tumors, and no matter what treatment was given. Across all of these variables, people with well differentiated tumors did the best, and poorly/undifferentiated tumors did the worst, with moderately differentiated tumors in between. Specifically, patients with moderately differentiated tumors had a 40% worse survival than those with well-differentiated tumors, those with poorly differentiated tumors had a 70% worse survival, and the people with undifferentiated NSCLC tumors had an approximately 80% worse survival vs. well-differentiated tumors. Here's the figure showing the very different survival for the different tumor grades:

(click to enlarge)

All in all, grade was the third most important factor in predicting survival, behind stage (which is designed to predict outcome) and treatment given (largely determined by stage, and also affected by other issues like health of the patient). The analysis also showed that women had a 25% better survival than men, older patients didn't do as well as younger patients (65 as a cut-off), there was a trend toward better survival among never-smokers vs. current or former smokers that wasn't quite significant (only about 10% difference), and people with adenocarcinomas did a little better overall than patients with squamous cancers (about an 8% difference).

Because well differentiated tumors present at an earlier stage and have several other favorable characteristics associated with them, while poorly differentiated tumors have many other unfavorable characteristics commonly seen with them, it is important to separate out the overlapping variables. Once you do that, the differences aren't as dramatic, but they still show that the folks with well differentiated tumors are doing better than the ones with moderately or poorly or undifferentiated cancers:

The investigators found that the same trends applied to the tumors from other people in the county, seen and treated outside of the Mayo Clinic, so it wasn't just a phenomenon of something special about the patients who found their way to their institution.

Finally, they looked at 1302 tumors that were resected and noted that the recurrence rate for poorly/undifferentiated, moderately differentiated, and well differentiated tumors was 47.3%, 32.7%, and 20.8% respectively. Looked at another way, poorly and undifferentiated tumors were 2.1 times as likely to recur, and moderately differentiated tumors 1.4 times as likely to recur as well differentiated tumors after surgery.

This could have been analyzed a little differently to look at moderately differentiated tumors as the benchmark, which would have shown that well-differentiated tumors do better than average, while poorly/undifferentiated tumors do somewhat worse than average. You don't necessarily use the best cases as the standard. It doesn't matter much, since the data are the same, and the only thing that matters is what you measure against. But you can see from the total numbers that well differentiated tumors were a real minority of cases, at 13%, while more than 40% were moderately differentiated.

We've been looking at fancy gene signatures and other molecular markers for the use in potentially refining our treatment plans for individual patients, but it's worth remembering that tumor grade may be one of the more important factors that doesn't require sending tissue for cutting edge molecular analysis. While some older studies haven't shown a clear association of outcomes with tumor grade, several of these older trials were just too small to say anything definitive. It is worth noting that there are long-term survivors with poorly differentiated, high grade cancers, and there are patients with well differentiated low grade tumors that recur and can be very threatening. At the same time, we need to remember that pathologists don't all agree on which tumor is moderately differentated and which are well differentiated. The real world is a continuum, with a lot of gray areas between these neat categories. Nevertheless, this very large study convinced me that along with the fancier, newer technology approaches we have to help us refine our predictions of who might be better or less well served by adjuvant chemo, and which patients we might want to treat less aggressively and be optimistic that they might do particularly well. Grade isn't the most important factor, but I would say it's one that probably deserves to be factored in the overall equation.