A question from one of our readers.

Could you explain why doctors keep putting pediatric patients on Gleevec when it doesn’t work?

The answer is simple, although unsatisfactory. We do not know if Imatinib (Gleevec, Glivec) is, or is not, effective in patients with pediatric or wildtype GIST. There are two main reasons why we do not know the answer to this question.

First, there are very few patients with pediatric GIST, which makes it extremely difficult to perform a clinical trial to determine the effectiveness of Imatinib in this setting.

We asked a NCI statistician to design a hypothetical trial for us. “A randomized trial comparing Imatinib with placebo, to determine if treatment with Imatinib prolongs the time to recurrence, in pediatric patients following complete removal of GIST nodules.” Since this is a hypothetical trial, we did not concern ourselves with important questions such as which dose to give, the length of therapy and differences in distribution of the drug in the body.

In any trial, all of the investigators, sponsors and patients are hoping for a positive result. However, it is also very important to know that an experimental agent did not work. Although disappointing, it would still provide a definitive answer. The worst-case scenario is that after a large trial is completed, that we do not get either a positive or a negative result. One way to avoid this is to guarantee that enough patients participate in the study to arrive at either a positive or a negative answer to the question being asked.

Our hypothetical question is whether Imatinib is effective in prolonging the time to recurrence of tumor. To answer this question, we must have an idea of what the time to recurrence is without any therapy. In the inaugural Pediatric GIST clinic, there were nine pediatric patients. Three patients never had complete removal of the primary tumor. Two patients had complete removal and after two years are free of tumor. Four patients recurred following full resection after 22, 24, 38 and 41 months, an average of 31 months. So based on the average of six patients, we are estimating that the average time to recurrence of tumor in pediatric patients after initial complete removal is 31 months. Again, because this is a hypothetical question, we are not concerned that only a very small number of patients was used to arrive at this estimate.

Now, since the question was whether Imatinib could prolong the time to recurrence, we had to ask how long a time period we wished to measure? So we asked to see the number of patients required, to detect a difference between 31 and 36 months, 31 and 48 months, and 31 and 60 months. At the bottom of this page are the exact calculations that the statistician returned to us.

Assuming that if patients recur, they will recur at 31 months following the initial surgery, in order to determine if Imatinib prolongs the time to tumor recurrence, we would have to enroll the following numbers of patients

          An increase from 31 to 36 months requires 2,666 patients
          An increase from 31 to 48 months requires 352 patients
          An increase from 31 to 60 months requires 174 patients

For common cancers, such as lung cancer, it is possible that several cancer centers could register this number of patients in one year alone. The six patients from the inaugural clinic who met the criteria to enroll in this hypothetical study were diagnosed over a six-year period, which averages to one person per year. Assuming that this is the actual rate, it would take 174 years to finish this study. To give the benefit of a doubt, we will multiply the number of patients by ten and assume that every year ten patients would be eligible to participate in this study. Even then it would take 18 years to complete this study.

This example gives any idea of why it has not been possible to even consider performing a trial that would give a definitive answer to the question.

The second reason that we cannot definitively say that Imatinib has no role in treatment for pediatric GIST is the following.

If we treated several pediatric patients with unresectable tumors and the tumor shrank, that would be evidence that pediatric GIST is responsive to Imatinib. Unfortunately, we have not observed this, and some people therefore conclude that pediatric GIST is not responsive to Imatinib. This is one possibility and it may be correct. However, there are other possibilities that must be considered.

The table below lists tumor sizes of two hypothetical patients, measuring the size of a single tumor nodule in one dimension, every 6 months, for 2 years.

Patient A has an unresectable GIST lesion discovered in Feb 2005 at 2.0 cm and chooses not to receive any therapy. In Aug 2005 and Feb 2006, scans reveal that the tumor continues to grow. Patient A begins therapy with Imatinib in Feb 2006 when the tumor is 4.5 cm. Despite therapy, the lesion continues to grow to 9.0 cm in Feb 2007. At that point, Patient A stops taking Imatinib and enrolls in a different trial. Not many people would argue that this is a case in which Imatinib was not effective.

Patient B also has an unresectable GIST lesion discovered in Feb 2005 at 2.0 cm and chooses not to receive any therapy. In Aug 2005 and Feb 2006, scans reveal that the tumor continues to grow, though slightly slower than in Patient A. Patient B begins therapy with Imatinib in Feb 2006 when the tumor is 2.8 cm. There are no side effects and in Feb 2007, the tumor has increased only slightly in size from 2.8 cm to 3.2 cm over the span of one year while on Imatinib. Patient B asks to stop taking Imatinib, because the tumor has continued to increase in size. Imagine you are the oncologist for Patient B, what would you advise?

We will provide some additional information if you have not yet decided how you would advise the patient. Had Patient B not taken Imatinib, the tumor size in Feb 2007 would have been 6.0 cm and not 3.2 cm. Clearly in this case, Imatinib has slowed the progression of the tumor and you would advise Patient B to continue taking Imatinib.

On the other hand, what if we revealed that had Patient B not taken Imatinib, the tumor size in Feb 2007 would still have been 3.2 cm, the same size as when Patient B took Imatinib. Clearly, in this case, Imatinib did nothing to change the tumor size and you would advise Patient B to stop taking Imatinib.

If we knew the natural history (what happens to the tumor if we do nothing at all) of pediatric GIST, we would not have to guess which of the two possibilities above was occurring and treatment plans and recommendations would be much easier. Unfortunately, we do not know the natural history of pediatric or wildtype GIST very well. This is one of the main objectives of the clinic, to compile information on how GIST tumors behave and the impact that different treatments have in modulating tumor progression.

Although the scenario presented above was hypothetical, there were patients in the inaugural Pediatric GIST clinic who had similar questions. Our advice was very specific, based on each individual’s case. But listed below are some our general guidelines.

For patients with fully resected GIST with no evidence of tumor:
We would advise no additional therapy.

The rationale:
Risks? There is a high risk of side effects.
Benefits? There is no evidence that Imatinib is effective in this setting.

For patients who have unresectable GIST and scans reveal stable disease or very slow progression, and who are currently tolerating Imatinib:
We would advise to continue taking Imatinib.

The rationale:
Risks? With no current side effects, there is less risk in continuing.
Benefits? It is possible that Imatinib is slowing the rate of tumor growth.

Following is the the statistician’s reply with the details of our hypothetical trial.

Thanks for passing the question along to me. What follows are approximate sample sizes for a trial comparing placebo to Gleevec based on a log-rank test to compare EFS curves, made using an approach that does not require knowledge of the accrual rate, but rather makes an assumption that you will be able to follow patients until at least a certain total number of events (on both arms combined) have taken place. I agree with your approximate estimate of 31 months, although since it is only based on 6 patients, it may not be very reliable.

I have created two sets of examples, one based on a strict phase III design with a 0.05 alpha level two-tailed log-rank test and the other based on the phase 2.5 design using a one-tailed 0.10 alpha level test. Keep in mind that using a phase 2.5 design is really intended for the case in which the trial could at least theoretically be followed by a definitive phase III trial in the event that the phase 2.5 shows a potentially promising result. Each provides 80% power.