Among the many variables that can potentially be helpful in predicting outcomes after surgery are some imaging results. One of these is cavitation, or hollowing out of the inside of some part of the tumor. Although most clinicians think of this as a feature of squamous cancers, it can also be seen with adenocarcinomas and other histologies less frequently. On a CT, this appears as a black area of air in the middle of a nodule or mass, and this generally occurs from a cancer growing to outstrip its blood supply, so that the inside of a tumor no longer gets nutrients and dies, from the inside out. In fact, a study from MD Anderson of stage I cancers with or without cavitation (abstract here) indicated that those with cavitation are associated with a significantly worse survival:
(Click to enlarge image)
Although this is a small trial, with just 72 patients included, it does suggest that tumor cavitation may be a helpful tiebreaker in cases where people are on the fence about pursuing post-operative therapy.
Another potentially relevant issue is the standard uptake value (SUV), a measure of the metabolic activity of a tumor on a PET scan, which just about everyone now gets before surgery to clarify staging. SUV is a measure of labeled glucose (sugar) uptake by a tumor, which is increased by higher metabolic activity like rapidly dividing cells, which leads to faster tumor growth. At the World Conference on Lung Cancer in Korea, Dr. Goodgame and colleagues evaluated variables that predicted recurrence among 136 patients with stage I NSCLC tumors resected at Washington University in St. Louis (abstract here):
Higher risk of recurrence was associated with several factors, including higher T stage (not surprising, that’s what staging is for), and also more likely with adenocarcinomas vs. squamous carcinomas (I described this in another post). It wasn’t associated with age. PET SUV of 5.5 or more was associated with a significantly greater risk of recurrence than that seen in patients with an SUV below that. And when they did a multivariate analysis where they determined which variables were more or less important in predicting what would happen (many correlating with each other), the SUV was the strongest independent predictor of outcome. And pretty much all of the little evidence we have about predicting outcomes based on PET results for early stage disease corroborates this same conclusion. While the break point is a little different in one study or another (often in the 5-7 range), the work thus far has uniformly suggested that tumors with a higher SUV are more worrisome and more likely to recur than the early stage cancers with a lower SUV.
posted by Dr. West @ 10:16 pm link to this post








October 10th, 2007 at 10:39 am
Dr West:
Thank you for the interesting postings on prognostic factors for early stage NSCLC. I was wondering about the multivariate analysis of prognostic actors. In particular, is there (somewhere) a list of the “independent” prognostic factors for early NSCLC with they relative “strength”? Without this ranking we are left with a list of factors but no way of deciding which ones are the most important ones.
I have been looking at the literature for independent prognostic factors for fully resected stage IIIA but nobody had collected these with their “strength” in a single document. I believe doing so will be useful for stage IIIA as for early NSCLC since it may give us a more refined prognosis that is not only dependent on the stage (which if I understand correctly is the strongest independent prognostic factor). Even with the caveat that most of these prognostic factors come from relatively small studies they may still inform our outlook on the disease.
Carlos
October 10th, 2007 at 10:27 pm
Carlos,
I don’t know of any source that includes this kind of analysis. It would actually be a pretty significant undertaking, basically requiring a meta-analysis of many early stage trials and then a multivariate regression analysis of the factors. It very well might be done sometime, but I haven’t seen it yet.
For now, we clinicians also need to view these results and mentally adjust by saying “the tumor size was very small, but it was a poorly differentiated tumor”. Perhaps eventually we’ll have nomogram-type software (for instance, at www.nomograms.org — and yes, I’ll do a post on their nomogram for predicting a smoker’s risk for developing lung cancer in the next 10 years) that allows us to enter multiple variables and have the program calculate a more exact set probability of recurrence. But not just yet.
-Dr. West