Over three and a half years ago, I wrote a post about about early work looking at the possibility that a PET scan done anywhere from three days to a month out from the start of a new systemic therapy (chemo or EGFR inhibitor) for advanced NSCLC could be predictive of a good outcome or not. Despite the increasing use of PET scans not only for staging but also for follow-up to assess response in patients with NSCLC, there hasn't been a lot of new information since then. However, a study presented at ASCO 2010, coming from several Australian hospitals in collaboration with a couple in southern California, compared PET scans done after two weeks and then two months of Tarceva (erlotinib) therapy with the more established standard measure of response assessment, CT scans after two months of therapy.

The study enrolled 74 patients with advanced NSCLC who had not received Tarceva in the preceding 14 days and hadn't previously received an EGFR inhibitor. This population was pretty typical for a non-Asian lun cancer population, with only about 10% never-smokers, 58% male. They all received three baseline scans in the 14 days leading up to starting Tarceva: a standard chest CT, an [18]- fluorodeoxyglucose (FDG)-PET scan that measures cellular metabolism (see this post or this podcast for general review of metabolic imaging with PET), and a newer and less commonly available form of PET scan with a different tracer called [18]-fluorodeoxythymidine [FLT] that measures cell proliferation. Then, in addition to standard CT assessments 56 days later (a "cycle" of Tarceva on a study is typically 28 days, so 56 days would be the usual interval for a follow up assessment after two cycles), the enrolled patients were supposed to undergo a repeat FDG-PET and FLT-PET at both 14 and 56 days into treatment: (click on image to enlarge) Of the 74 enrolled, 51 completed all of their scans as planned and were eligible to be analyzed. The study compared response rates as assessed by these methods and how well these modalities predicted progression-free survival (PFS) and overal survival (OS), including reviewing how they did in different clinical groups by molecular factors like EGFR and KRAS mutation status. While the criteria for response assessment by CT are pretty well established and based on measurements of the diameters of target lesions being followed (see this summary of the RECIST criteria), there aren't as well established for PET scans. In this study, the investigators defined a PET response as a 15% or greater reduction in the maximum standard uptake value (SUV), the pivotal measurement of PET imaging, in up to five identified target areas on PET. Below is the "waterfall plot" of responses as measured by a standard FDG-PET at 14 days, with the downward bars showing reduction of PET uptake, and the upward bars noting increasing PET uptake indicative of progression. Below the bars, and definitely only visible if you click to see the expanded version, are codes for the CT-defined response later seen at day 56, patient sex, tumor histology, and various molecular markers they collected: In the trial as a whole, only 4 of the 51 patients (8%) had a CT response, but 13 (26%) had a response as measured by an FDG-PET at 14 days. FLT-PET showed similar trends, though the response rates were a little lower. Progression numbers were pretty similar with all three imaging techniques. It's also interesting to note that PET responses were seen not only in three of the four patients with an EGFR mutation, but in several who didn't have an EGFR mutation, and some degree of decreased PET uptake was seen in some patients with a KRAS mutation. This strongly suggests that patients with a KRAS mutation may still benefit from an EGFR mutation, as some limited other evidence has also suggested. The authors also found that day 14 FDG-PET scan results predicted both PFS and OS, and that the FLT-PET didn't do a better job. Here's the curves for these endpoints, with groups separated by whether they demonstrated a an FDG-PET response at 14 days: Overall, this work suggests that PET scans within just a few weeks after starting Taceva can do a pretty good job discriminating who's likely to benefit. While there's little mystery about the minority of patients with an EGFR mutation who have a dramatic improvement within days and will show a response by just about any measure, this study is most interesting in demonstrating that there are subtle hints of who is showing some degree of benefit or not in the much larger population of patients who don't have an EGFR mutation and may even have a KRAS mutation. It also underscores that you don't need to have an EGFR mutation to actually derive benefit from Tarceva. One big question, though, is whether a 14 day assessment would actually lead people to change what they're doing. If you see a response or stable disease, that's reassuring, but it's not clear to me that someone who is showing increasing PET uptake two weeks into Tarceva will then decide to stop it and either move on to something else or abandon treatment. We generally follow the edict that a test is only worth obtaining if you'll use the information to change what you do. But if an early PET leads some people to continue it when they might otherwise drop it if they don't feel clearly better, or to stop it if the scan is moving in the wrong direction, then it may be a worthwhile approach to pursue.