In last month’s newsletter, the Life Raft Group published its study on the effects of imatinib dose upon the survival of metastatic GIST patients. On February 4, we also held a webcast, “The Link Between Dosage and Survival”. Since then we have had a number of questions about the study.
Some of the questions had to do with the design of the study. To answer these questions, it helps to remember that the original study was prepared in 2004 and at that time only looked at progression. One key question asked was, “ How accurately can patients report progression?”

To put this in perspective, and to understand the basis for our patient selection criteria, it might be helpful to review how this is done in clinical trials as well as with patient- reported data.

In clinical trials a number of “target” and “nontarget” lesions (tumors) are identified at the start of treatment and form a “baseline”. Target lesions are measured and if they increase by a specified amount, or if new tumors appear, the patient is deemed to have progressed.

This typically results in the patient being removed from the study, or crossingover to another treatment (such as a higher dose of Gleevec). One of the most popular methods to measure progression and response is the RECIST method. Using the RECIST method, the response of target lesions and non-target lesions, as well as the appearance of new lesions, are combined into an “overall response”.

A “response” to treatment is the flipside of progression. Instead of tumors growing by a certain amount, it is when tumors shrink by a certain amount. For RECIST criteria, to have a “partial response” the shrinkage must be at least 30 percent in the sum of the longest dimensions of target lesions.

So using a change in size of tumor, response to treatment can be described as:

Complete response – Tumors shrink until they are no longer visible on CT scans

Partial response – Shrinkage of more than 30 percent

• Stable disease – Ranges from shrinkage of less than 30 percent to growth of less than 20 percent
• Progression – Greater than 20 percent growth

Response (shrinkage) is often used in early studies as a “screen” to help evaluate whether a new drug or treatment is worth further studies. Response rate can also be used in larger randomized phase III trials, but it is seldom the only measurement criteria (end point). Other endpoints that are better indicators of clinical benefit include survival (the gold standard), progression-free survival (PFS) , and improvement in symptoms and/or quality of life. The most reliable and preferred endpoint is survival.

The two large phase III GIST trials both used RECIST criteria to measure tumor response. However, concerns about the use of RECIST as response criteria for GIST have been raised in the past few years. In particular, clinical benefit for GIST patients taking Gleevec does not appear to be strongly related to tumor shrinkage (response). Patients that only achieve “stable disease” as their best response have overall survival (OS) times that are just as good as those that have significant shrinkage.

One of the problems related to the use of RECIST in GIST is that tumor cell death does not always result in tumor shrinkage. In fact, a dying tumor can sometimes increase in size and this can be misinterpreted as progression. A related problem is that before starting Gleevec, existing tumors sometimes have a very similar appearance and density as surrounding tumors and are not visible (or easily missed) on CT scans. Upon successful Gleevec treatment, these tumors can start dying and become less dense. The less dense tumors now become visible on CT scans and look like new tumors.

So when considering the predictive ability of RECIST, it’s method of measuring response does not appear to be very good for GIST. However, with the exception of false progression, RECIST progression criteria do seem to be a pretty good predictor of benefit for GIST patients on Gleevec. The newer “Choi criteria”, which also takes tumor density into account, may prove to be a better predictor of benefit than RECIST.

The RECIST method uses (and requires) precise tumor measurements, with progression (tumor growth) being more predictive of benefit than response (tumor shrinkage).

Patient-reported progression/response data is not nearly as precise as that gathered in clinical trials. Patients may say their tumors grew, shrank, did not change in size, or they may say some tumors grew and some shrank or new tumors appeared. In general, they can not say things like, “The sum of my target lesions increased by 20 percent”.

So if we look at the limited usefulness of measuring tumor shrinkage using the RECIST criteria, the inability of patients to adequately quantify shrinkage in the LRG study does not seem to be a major concern. More importantly, can they tell you if they progressed? The answer comes from the 2007 update to the LRG study.

In the 2007 update, the LRG looked at the same 169 patients from the original study in 2004. One of the major reasons for not expanding the study to include more patients was so that we could go back and look at the original 169 patients and see what happened to them. As of December 2007, 81 percent of patients that told us their tumors had progressed by October 2004 died versus 11 percent for those that had not progressed by that date. This tells us that patient-reported progression was a very good surrogate for benefit (survival is the gold standard for benefit).
Additional support for the reliability of the patient-reported data can be found by comparing the survival curves of the LRG study and the MetaGIST project. The curves are different for the first year because patients in the LRG study had to have been on Gleevec for one year, but beyond the first year the data (both PFS and OS) is very comparable (see the March 2008 LRG newsletter for details) when starting dose is used for the analysis.
The following are some other common questions regarding the 2007 LRG study:

Question: Why were patients with stable disease excluded from the LRG study?

Answer: Because of the limitations of patient-reported data. At the time of the original study (2003/2004), there was some question in our mind about how well these patients were responding to Gleevec. Were they truly responding, or did they just have slowgrowing disease that mimicked a response? Because of the limitations that patients have in reporting very small changes in tumor size and because at the time of the original study we were only looking at PFS and not OS, we were concerned that we might not be able to adequately distinguish between those responding and those progressing. In 2008, due to the efforts of the GIST medical community, we know a lot more about stable disease. It would be interesting to expand the LRG dataset and look at more patients including those with stable disease.

Q : Why are the median PFS and median OS times so much longer than seen in the phase III trials?

A : About 15 percent of GIST patients do not respond to Gleevec. These patients will typically have progression fairly early, usually within the first six months or so. This is called “primary resistance”. Because our study design required patients to be on Gleevec for at least one year, it eliminated patients with primary resistance. By removing those that never respond to Gleevec, it restricts the study to those most likely to benefit from Gleevec (except for stable patients which are discussed separately). This increases both median PFS and median OS. All of the randomized trials include all patients, including those with primary resistance and stable disease.

Q : Which PFS and OS data is more likely to apply to me; the LRG data or the MetaGIST data?

A : All patients are individuals and will have unique characteristics that help define how well they will respond to therapy. When looking at how well someone is likely to respond, the most important variable is probably mutation type, but there are other factors. In contrast to the MetaGIST data, our data, which is preliminary, suggests dose may be a significant factor as well. The preliminary data presented by Dr. George Demetri at 2008 Gastrointestinal American Society for Clinical Oncologists (ASCO) meeting suggests that Gleevec drug levels may be a significant factor as well. Of course there will be a general relationship between dose and drug concentrations and it remains to be seen which is more important. I think an important point to consider is that when patients with primary resistance are included in an analysis, it “drags down” the average (or more accurately, the “median”). So when a patient that has been on Gleevec for one year looks at the MetaGIST “median PFS” data, they conclude that they can only expect another seven to 11 months of PFS. However, the benefit that this patient can “expect” is probably higher than this (on average) and probably more closely matches the benefit shown by the LRG data. To simplify, if a patient meets the criteria of the LRG study; that is, on Gleevec for one year with some initial shrinkage, then the median PFS and OS times shown by the LRG data may be a better predictor than those of the MetaGIST data. For patients just starting Gleevec, the MetaGIST data may more accurately reflect the median PFS and OS times, at least until they get beyond one year on Gleevec. Even more importantly, we must always remember that each patient is different and statistics apply to groups of patients, not individual patients. Interested readers might want to read an excellent article on applying statistics to their situation. It is called “The Median isn’t the Message” and was written by Stephen Jay Gould. Gould was a 20-year survivor of Mesothelioma, a type of cancer with a “median survival” of eight months at the time of Gould’s diagnosis.

Q : I am taking 400 mg of Gleevec to prevent a recurrence. Based on the LRG study, should I be taking more?

A : Neither the LRG study or any other study has looked at the question of higher doses for adjuvant (preventative) treatment. So we don’t know the answer to that question. The biggest concern here might be for those with exon 9 mutations. These seem to exhibit a “primary resistance” type of behavior at the standard 400 mg dose. So what happens to these GIST tumor cells if they are exposed to subtherapeutic drug levels for a year? I am concerned that this may be doing more harm than good for exon 9 patients.