Imatinib for Resected Gastrointestinal Stromal Tumors
Adjuvant treatment of GIST
Obligatory—This is not medical advice
A wise patient asked me: “if we can’t see my cancer, why am I taking medicine?” This is an entirely reasonable question. We are requesting that patients trust us when we tell them to take a toxic and sometimes expensive medication for years. They have every reason to wonder why and it’s our job to make sure that this information is clear. So, what is the basis for treatment with imatinib after resection of a GIST, anyways? How do we grade the limitations and incorporate them into the care of our patients? Let’s talk about what we do and do not know.
Mutations in Gastrointestinal Stromal Tumors
I will start with an apology. While my last post delved into how a technical diagnosis of GIST is made, I did not so much explore the details of the pathogenesis of this tumor, and why KIT is so important. I named a vital feature, but no justification for its prominence was given. Sometimes these things seem self-evident to us oncologists because we see them every day. I’ll work to ensure that doesn’t happen again. Let’s talk about KIT and other mutations that are the foundation of the development of GIST.
Principally, GIST are thought mostly to arise from the interstitial cells of Cajal. As might be noted from the below figure, molecular drivers can be contingent upon the location. There is some biological reasoning behind why this is the case, but it is beyond purview of this post. The fundamental premise here, and this is an oversimplification, is that at some point in the cellular life cycle, a mutation in one of these proteins promotes further cell growth, propagation of a clonal population and subsequent favorable selection. This is a poor summary of the process of neoplasia, but will suffice for now. These mutations can vary. Cells eventually become dependent on expression of this protein, which is usually a tyrosine kinase (the subfamily of proteins in which KIT exists).
60-70% of patients with GISTs will have mutations in KIT, while 10-15% will have them in PDGFRA.1 Historically, those that were negative for these mutations were labeled ‘wild-type’ GIST. This implies that they did not have a detectable mutation. As those who have read my prior article on GIST have noted, there was likewise a time when GISTs were indistinguishable from other smooth muscle tumors, and this, like much in medicine has appropriately changed. Wild-type has become somewhat of an antiquated term, as it includes a panoply of driver mutations in various proteins:
BRAF
NF1
SDH
RAS
NTRK
FGFR
Others
We continue to learn and move forward and our pace has picked up. Separating and qualifying the mutations for all of our patients with wild-type GIST will go a long way and I anticipate that, especially in instances where there is consideration of adjuvant or neoadjuvant therapy, it will become increasingly prominent.
Now, even amongst patients whose tumors have mutations in KIT, there is variation. Patients may have mutations in multiple exons (9, 11, 13, 14, 17, 18, etc). These have differing levels of sensitivity to imatinib and other medications in our armamentarium.
Adjuvant Therapy
Now we understand that GIST is a disease process driven by various mutations leading to overexpression or dependency on proteins. While it may seem obvious now, this was relatively new information just a few decades ago. Much has changed since the late 1990s when GIST was being separated out as a distinct clinical entity. 2
Also in the late 1990s, there were remarkable findings in the tyrosine kinase inhibitor world, with the discovery of imatinib. 3 Imatinib's activity against KIT led to its being studied in patients with advanced and metastatic GIST and the rest is history.4 5
The natural progression of FDA approval for medications is first for approval in patients with advanced or metastatic disease and subsequently at increasingly earlier points. This parallels the fact that it is often harder to measure a difference in populations of patients that have a higher chance of being disease free and are without visible disease.
This pattern was true for imatinib in patients with GIST. There are multiple notable trials, and likewise many details and/or peculiarities associated with them. I will try to hit the highlights. The first study we talk about is ACOSOG Z9001, which was a phase III trial that looked at relapse free survival for patients treated with 12 months of imatinib at a dose of 400mg daily.6 Patients must have a CD117 positive GIST that was at least 3 centimeters in size. 778 patients were recruited with 713 being analyzed in the intention to treat analysis. Crossover from placebo to imatinib was allowed. The corresponding figure is below.
The primary endpoint of relapse free survival was met with 8% of patients in the treatment group and 20% of patients in the placebo relapsing after a median of 20 months follow-up. This figure, however, indicates that the curative fraction perhaps was not altered, but rather, relapse may have been postponed. Relapse free survival is a composite endpoint that includes:
Recurrence of the cancer
Death
This was not the initial primary endpoint of this study, which had intended to assess for overall survival. The authors specify why this had changed, and certainly, this was early in the days of imatinib and so we might expect some fluctuations. Nonetheless, the results of this trial did not establish that there was a survival benefit to taking imatinib for one year after resection. It delays recurrence in a general population of patients with tumors which were greater than 3 centimeters.
Understandably, there remained real questions:
How do we better define who gets adjuvant treatment?
Tumor size
Mutations
Mitotic rate
Locations
Should we change how the medicine is given?
Longer duration
Different doses
Crossover considerations
Active controls
While there have been multiple trials, I will mostly discuss SSG XVIII/AIO. This was a Phase III clinical trial performed in Finland, Germany, Norway, and Sweden that randomized patients 1:1 to receive 12 months or 36 months of imatinib.7 The primary endpoint was relapse free survival, again, with a secondary endpoint of overall survival. Patients must have had a high estimated risk of recurrence (see prior post). In total, 397 patients were recruited. They were stratified by risk, age, and site of mutation. See the Kaplan-Meier curves below.
These patients were followed for on average 10 years, which is impressive. Investigators saw that at 5 years, the relapse free survival was 71.4% and 53% for the 36 and 12 month treatment groups, respectively. 5 year and 10 year overall survival was 92% and 79.0% in the 36 month group, and 85.5% and 65.3% in the 12 month group. The details of these findings are viewable in the figure.
How do I interpret these data? My feeling is that this justifies a longer adjuvant treatment with imatinib for high risk patients, particularly those whose tumors harbor KIT mutations. It is, however, important that we are honest about the results here. It does not appear that adjuvant imatinib is curative—there is no plateau to the 3 year treatment curve. As much as imatinib can treat this disease, it does not appear to cure it. I am very upfront about this. I portray what we know, and more importantly, what we don’t. I have had patients with KIT mutated disease decline therapy with imatinib based on their individual values and I definitely respect their perspective. Imatinib, as much as we might champion it, is not a sugar pill and can have real side effects.
Conclusions
Recommendations regarding adjuvant treatment for patients with resected GIST are not as simple as they may appear. The data are somewhat clearer for patients with high-risk tumors. If there are questions, these should be fielded by a subject matter expert and careful consideration of the risks and benefits of 3 years of adjuvant imatinib should be had. It is our duty to make certain patients are informed.
https://www.nature.com/articles/s41572-021-00254-5
https://pubmed.ncbi.nlm.nih.gov/10534170/
https://pubmed.ncbi.nlm.nih.gov/11705489/
https://www.nejm.org/doi/full/10.1056/nejmoa020461
https://jamanetwork.com/journals/jamaoncology/fullarticle/2766796