May 15, 2007
| One of the most obvious trends in modern drug discovery is the explosion of new imaging approaches. In particular, these are being put to great use in oncology research. Just a couple of years ago, Pfizer researchers demonstrated a startling example of such use in studies looking at the effects of a new drug called Sutent (sunitinib malate) in a small human trial.
What was so surprising about these studies wasn't just that they quickly and efficiently proved the drug was working. Rather, it was that the studies demonstrated an unconventional marker — tumor metabolism rate instead of the typical marker of tumor size. At the time, one of the questions dogging Sutent was that the drug did not seem to shrink tumors in patients. However, animal studies had suggested it had great promise. To reassure themselves the drug was working in people as it had in animals, Pfizer researchers looked very closely at its effects in small “proof of concept” trial that included patients with gastrointestinal stromal tumor (GIST). This particular tumor is one of the ideal targets for Sutent, a multikinase inhibitor.
Patients in the small trial received fluorodeoxyglucose (FDG)-PET scans during drug treatment, and then after therapy was suspended. Soon, the investigators' questions were put to rest. The scans showed that just days after starting Sutent therapy, glucose metabolism plummeted in the tumors, despite the fact that they remained about the same size. The tumors were indeed being attacked by the drug, even if they weren't shrinking. The drug went on to become the first oncology product to be simultaneously approved by FDA for two indications — GIST and advanced kidney cancer. (FDG)-PET scans are being increasingly used to understand better how cancer drugs work, and the FDA is very interested in seeing data about how these new imaging approaches compare to standard techniques for judging cancer drugs' efficacy.
With this experience in mind, it was very interesting to hear a talk recently by Dominic Spinella, senior director in translational medicine, global oncology lead, at Pfizer. Speaking at CHI’s recent Clinical Biomarkers* meeting, Spinella described how the company aims to integrate new imaging approaches with traditional molecular biomarker research to accelerate research and increase the success rate of oncology projects.
The group starts by looking for proof of mechanism. Selecting the patients for these studies is as important as selecting the markers. Certain types of tumors are clearly driven largely by specific molecular drivers, but finding those isn’t always easy. “Sometimes you can analyze archival biopsies, but you have to wonder if the biopsy reflects what’s going on in the patient’s tumor now,” Spinella said. “If you are looking at a causative mutation, it is very rarely back mutated. But wild type p53 in an archival biopsy doesn’t say anything about the tumor’s p53 status today.”
Whenever possible, Pfizer scientists like to start with molecular assays that act as surrogates for particular imaging tests. For example, a GLUT-1 test can be used to test the appropriateness of FDG-PET, or a thymidine-kinase assay for FLT-PET. As the lead compound nears nomination, the group starts using preclinical imaging.
“Our goal is to understand which imaging modality is most appropriate for which drug, and when is the optimal time to acquire images,” said Spinella.
In the Clinic
Once in the clinic, the oncology group will do a dose escalation trial in a standard “all-comers” Phase I design, to obtain safety and tolerability data. They are unlikely to do imaging studies at this point. But, “Once we have the maximum tolerated dose worked out, we will do an expanded cohort study with about 15 to 20 patients to look at the molecularly relevant markers,” he said.
If they can’t see an effect on the target marker (or markers), “We terminate the project,” said Spinella. “If there is a significant effect, we employ a de-escalation dosing strategy with imaging, to establish the exposure/response relationship.”
One of the advantages of using imaging is that “If the lower dose doesn’t work, you can tell right away.” If you can get the same effect below the maximally tolerated dose, “Then let’s do what we do in other disease,” he said. “Let’s take several doses into trials so we can use something less than the maximally tolerated dose in patients.”
The group looks for a statistically significant change in the imaging end-point across the entire treated cohort, but uses a somewhat less stringent p-value in these small, mechanism-oriented studies. “I’m looking for a trend,” Spinella said, “Such as a 20% reduction in Ktrans.” While the effect may seem muted and slight, “In the right patient population, even if you have a small responder subset, the whole cohort mean changes enough that you can see the effect.”
While the Pfizer group is leaning hard on biomarkers, and has dropped some projects due to negative biomarker data, “Objective tumor response always trumps negative biomarker data,” said Spinella.
* Clinical Biomarkers Summit, CHI, Coronado, Calif., March 19-21, 2007.
Subscribe to Bio-IT World magazine.