Genotype-specific efficacy in the Imatinib mesylate treatment of Gastrointestinal Stromal Tumors (GISTs)

A report detailing recent progress in genotype-targeted GIST therapy

by Micah J. Benson

Genetics 144

Click to return to oncogenomics home

This report covers the following topics. First, the biology of imatinib mesylate (gleevec) is introduced, with a focus on the mechanism of its specific inhibition of the BCR-ABL oncoprotein. Two further targets of gleevec are covered: c-Kit and PDGF-Rα. Gastrointestinal stromal tumor biology is explained, and recent progress pertinent to the theme of clinomics is presented through a discussion of two clinical trials which show that mutational analysis of kit and pdgf-rα can predict the clinical response to imatinib.

 

Introduction

The recent success in the development of genotype-specific molecularly targeted cancer therapeutics has given rise to a new paradigm in drug development. Indeed, when all else is the same, a drug may fail in treating the cancer of one person while succeed in treating their neighbor, all due to the specific genotype of the cancer. While initial clinical trials gave gefitinib (Iressa) a bleak outlook for future use in treating non-small cancer lung cancer (NSCLC), the finding that this drug is highly efficacious against patients harboring certain mutations in the epidermal growth factor receptor (EGFR) gene has led to the striking realization that the exploration of therapeutics not only targeting specific molecules should occur, but that molecularly targeted therapeutics exist that are genotype specific, and should be tested as such (Lynch et al),(Paez et al). Another success story loyal to this paradigm involves the drug imatinib mesylate (Gleevec), which was initially found to potently inhibit the oncoprotein product of the Philadelphia chromosome translocation, BCR-ABL, as well as members of the split tyrosine kinase domain type III receptor family, including Kit, PDGF-Rα and FLT-3 (reviewed in Witte et al). As such, this drug successfully treats chronic myelogenous leukemia (CML), as well as other blood cancers and solid tumors whose transformed state is dependent on these proteins. Recent reports have described the efficacy of imatinib in treating gastrointestinal stromal tumors as dependent on the genotype of the tumor. Certainly, this will not be the last genotype specific cancer therapeutic.

Gleevec and BCR-ABL, c-Kit, and PDGF-Rα

The Philadelphia (Ph) chromosome is both the initiating genetic lesion and diagnostic clinical marker of chronic myelogenous leukemia (CML). A translocation occurs in a hematopoietic stem cell, and while void of influence on granulocytes and erythrocytes, the oncoprotein product of this translocation is a transformation-inducing factor of the myeloid lineage. The translocation occurs between chromosomes 9 and 22, fusing the bcr gene to the abl gene. Three different breakpoints occur within the bcr gene, each yielding a different sized protein product, with BCR-ABLp185 associating with acute lymphocytic leukemia (B-ALL) and CML, BCR-ABLp210 associating with CML, and BCR-ABLp230 associating with neutropenic CML and CML (Witte et al, Figure 2).
The ABL protein is a tyrosine kinase that associates with mitogenic and antiapoptotic proteins such as RAS and PI3K. The BCR-ABL translocation creates an oncoprotein with enhanced and constitutive tyrosine kinase activity, thus leading to enhanced RAS and PI3K activation, with the end result cellular transformation. The chimeric BCR-ABL oncoprotein has constitutive tyrosine kinase activity due to structural blockade of tyrosine kinase inhibitory factors that are normally present in wildtype ABL. Due to the gene fusion event, the ABL protein gains oligomerization motifs that are normally absent, with oligomerization amplifying the tyrosine kinase activity of BCR-ABL. A further mechanism for constitutive BCR-ABL tyrosine kinase activity is the elimination or sequestration of the inhibitory domains. The SH2 domain of ABL functions to keep tyrosine kinase activity in an inactive state. Upon chimerism, the A and B boxes of the BCR protein sequester the SH2 domain, and maintain BCR-ABL in an open and active conformation (Harrison, Figure 1). As a result of increased tyrosine kinase activity by BCR-ABL, increased amounts of GTP are loaded onto RAS through the adaptor protein sos-1, with constitutive activation of the RAS-MAPK, Jak/STAT and PI3K/AKT pathways. Imanitib mesylate inhibits the tyrosine kinase activity of BCR-ABL by displacing ATP and trapping BCR-ABL in an inactive conformation.

Imatinib mesylate was initially identified in a screen of phenylamino-pyrimidines in a PDGFR suppressor assay by investigators at Novartis ((Buchdunger et al). As well as potently inhibiting BCR-ABL, imatinib was also found to inhibit c-KIT and FLT-3, both members of the split tyrosine kinase domain type III receptor family, and to which family PDGFR belongs (Buchdunger et al). The sequence of the tyrosine kinase domain of BCR-ABL is highly divergent from the corresponding domains of the split tyrosine kinase domiain type III receptor family members, yet imatinib mesylate inhibits both.

The protein c-Kit is expressed on hematopoietic stem cells and progenitor cells, as well as on mast cells, melanocytes, germ cells and gastrointestinal interstitial cells of Cajal (ICCs). A tyrosine kinase transmembrane receptor, the ligand for c-Kit is homodimerized stem cell factor (SCF). The extracellular region of c-Kit contains five immunoglobulin domain repeats, three of which are involved in ligand binding. The binding of ligand to c-KIT leads to its homodimerization and activation of its intrinsic intracellular tyrosine kinase enzymatic activity, with subsequent autophosphorylation. The intracellular region of c-Kit contains two tyrosine kinase domains, TK I and TK II, as well as a juxtamembrane domain negatively regulating the TK I and TK II domains.

The physiological roles of c-Kit are readily apparent upon examination of loss-of-function, decrease of function, and gain-of-function mutations within the c-kit gene in both controlled murine experiments and from human cancer cell lines. It is apparent that c-Kit mediates cell proliferation, with cells either c-kit proliferation-dependent or independent. Mature cells that endogenously express c-kit, such as mast cells and ICCs, undergo c-kit-dependent proliferation. A decrease in c-kit signaling such as that observed in the Wv/Wv mouse decreases the prevalence of the aforementioned c-Kit expressing cell compartments. Anemia due to a lack of erythrocytes occurs, as well as white coat color in mice due to insufficient numbers of melanocytes. Sterility occurs as a result of the impact of impaired c-Kit signaling on germ cells, as well as the depletion of mast cells and ICCs. Gain-of-function mutations observed in both transgenic mice and in human tumors lead to mast cell leukemias, germ cell tumors, and gastrointestinal stromal tumors.
The signaling pathways activated by c-Kit functions are similar to those activated by BCR-ABL. In cells whose survival depends on c-Kit signaling, apoptosis is inhibited through the PI3K/AKT pathway. Activated c-Kit also recruits SH-2 containing proteins such as Grb2, She and SHp2 to associate to its intracellular domain. Grb2 associates with sos-1, and sos-1 colocalizes and activates Ras, which becomes activated. The downstream components of the Ras/MAPK pathway become phosphorylated, and proliferation is induced. Imatinib mesylate functions to inhibit c-Kit tyrosine autophosphorylation, thus inhibiting c-Kit-mediated proliferation.

Similar to c-Kit signaling, PDGF-Rα signaling occurs through receptor dimerization, autophosphorylation of the intracellular tyrosines, and activation of the Ras/Raf and PI3K/AKT pathways, with the end result cellular proliferation and survival. As previously mentioned, the structures of PDGF-Rα and c-Kit are homologous.
Interestingly, gain-of-function mutations in the c-Kit gene tend to occur at different sites of the Kit molecule in different neoplastic disorders, suggesting that the pathologic phenotype elicited as a result of c-kit mutation is determined by both the domain of kit affflicted by the mutation, and the cell type in which the mutation occurs. Exons 1-9 encompass the extracellular domains and exons 11-17 comprise the intracellular domains. Exon 10 constitutes the transmembrane domain, exon 11 is the intracellular juxtamembrane domain, and exons 13 and 17 are the tyrosine kinase domains. Mutations in exon 2 of the extracellular portion have been described in myeloproliferative disorders, exon 8 in acute myeloid leukemia, and in exon 9 in gastrointestinal stromal tumors. Mutations in exon 11, which is the juxtamembrane domain of kit which negatively regulates the tyrosine kinase domains of exons 13 and 17, have been described in GISTs. This is the most common site of mutation in GISTs. Mutations in exons 13 and 17, which code for the tyrosine kinase domains of c-Kit, are detected frequently in systemic mastocytosis, core factor binding leukemias, and seminomas. Mutations in the tyrosine kinase domains affect the ATP binding ability of c-Kit, and can yield gain-of-function or loss-of-function tyrosine kinase activity.

GIST Biology and Imatinib mesylate treatment of GISTs

The prevalence of GISTs, historically, is underrepresented, due to morphological similarities between gastrointestinal tumors of two different cellular origins. GISTs are neoplasms from the interstitial cells of Cajal (ICCs), which are spindle shaped cells lining the gastrointestinal walls, and are mesynchymal in origin. Gastrointestinal cancers of smooth muscle origin look morphologically similar to GISTs under H&E staining. These diagnoses are currently differentiated by immunohistochemical staining, as the phenotypic markers describing these neoplasms are distinct. The discovery of CD34 as a diagnostic marker allowed clinicians to begin correctly diagnosing gastrointestinal stromal tumors as neoplasms of ICCs. It is now known that 70% of GISTs express CD34. As Kit signaling is essential for the development and maintenance of ICCs, these cells constitutively express c-Kit and 95% of GISTs are positive for Kit (also known as CD117). GIST biology is reviewed in more detail in Sawaki et al.

In a seminal report, Hirota et al described the gain-of-function mutations that occur in c-Kit leading to GISTs, as well as documenting the exact mutations found in this neoplastic state. It was found that the majority of GISTs analyzed were mutated in exon 11, or the juxtamembrane domain in c-Kit. This region functions to inhibit receptor dimerization in the absence of SCF. Since this discovery, there have been multiple reports of exon 11 mutations leading to gain-of-function Kit oncoproteins. From 322 GIST tissue samples, 66% contained activating mutations in exon 11, with a high frequency of mutations between codons 552-661 (Corless et al). Other common gain-of-function mutations within c-Kit that have been described in GISTs include mutations in exon, the extracellular domain, which occur at a frequency of 19% in GISTs, and which almost always results in an alanine-tyrosine duplication or insertion. These are very prevalent in GISTs of the small intestine. In exon 13, the tyrosine kinase I domain, rare mutations occur which leads to the ligand-independent activation of c-Kit. In exon 17, the tyrosine kinase II domain, mutations have been observed that also lead to GOF mutations.

A small percentage of GISTs were observed to contain wild-type c-kit, and it was hypothesized that a gain-of-function mutation had afflicted a tyrosine kinase other then c-Kit. The receptor platelet-derived growth factor receptor alpha (PDGF-Rα) was found to be the culprit, with activating mutations identified in the juxtamembrane region and in the activation loop, similar to the activating mutations identified in c-Kit. The mutated PDGF-Rα protein was found to be constitutively phosphorylated in the the absence of ligand, and was found to result in morphologically distinct GISTs with an epithelioid character. It has been found that 7% of GISTs result from PDGF-Rα mutations. The constitutive activation of PDGF-Rα and c-Kit appear to be mutually exclusive of each other, as tumors are not found harboring both mutated genes.

Research into the efficacy of imatinib mesylate towards the treatment of GISTs was initiated upon the observation that imatinib inhibits the tyrosine kinase activity of both wildtype and mutant Kit in vitro. Imatinib was then shown to inhibit a GIST cell line containing a c-Kit activating mutation, and a patient with GIST was granted compassionate use of imatinib (Heinrich et al). The patient's tumor size decreased dramatically within weeks, and the favorable response continued for months (Joensuu et al). This quickly led to a multicenter trial, where the success of imatinib in the treatment of GISTs was apparent, with 54% having partial responses, 28% having stable disease, and progression occurring in 14% of patients treated (Demetri et al).

In order to determine whether the efficacy of imatinib relies on specific c-kit and pdgf-rα mutations intrinsic to the tumor, research groups from both europe and the United States correlated the mutational status of c-Kit and PDGFR to clinical response to imatinib. The european study, presented in the European Journal of Cancer in 2004, is titled Use of c-Kit/PDGFRA mutational analysis to predict the clinical response to imatinib in patients with advanced gastrointestinal stromal tumours entered on phase I and II studies of the EORTC Soft Tissue and Bone Sarcoma Group.

GIST samples used in this study were obtained from adults with histologically-confirmed unresectable or metastatic GIST that were c-Kit+ by immunohistochemical staining. Upon GIST diagnosis, patients were assigned to receive one of three different doses of imatinib, either 400mg daily, 300 mg twice daily, or 400 mg twice daily. GIST samples were fixed in paraffin, and immunohistochemical studies were performed using anti-c-Kit antibodies. For mutational analysis of tumor specimens, genomic DNA was extracted and c-Kit exons 9, 11, 13 and 17 and PDGF-Rα exons 12 and 18 were amplified by PCR. PCR products were screened for mutations using D-HPLC, and samples showing an aberrant elution profile were re-amplified and sequenced in both directions.

Of the 37 patients enrolled in the trial, 62% showed a partial response to treatment, 30% remained stable and three had progressive disease. Based on the Kaplan-meier analysis, the overall survival for all patients at 730 days was 78.3%, as shown in Figure 1a.

In the tumors harboring c-Kit mutations, 83% had mutations in exon 11, 14% had mutations in exon 9, and 3% had mutations in exon 13. The overall survival with patients harboring a c-Kit mutation was much higher at 730 days then patients with GISTs not containing c-kit mutations (Figure 1b). This illustrates the efficacy of imatinib against GISTs harboring c-kit mutations.

The exon 11 mutations included in-frame deletions either alone or associated with missense mutations or insertions, with single base substitutions not found. The most prevalent mutation was a WK 557-558 deletion, with these codons involved in other mutations as well. The activating mutations in exons 9 and 13 were the same as those previously described in GISTs, with an ala-tyr duplication. In 8 tumors with wildtype c-Kit, 6 tumors had PDGF-Rα mutations. Figure 2 describes the c-Kit exon 11 mutations in further detail.

The response data for diverse tumor genotypes show that 72% of GISTs carrying c-kit mutations showed a positive response, with 24% exhibiting stable disease. Patients with GISTs with kit exon 11 mutations were more likely to achieve a PR on imatinib mesylate, with 83% of these patients with exon 11 mutations exhibiting a partial response. One patient had progressive disease upon treatment with imatinib, and treatment was thus stopped. Disease remained stable after the patient stopped receiving imatinib. Patients with PDGF-Rα mutations exhibited progressive disease, even with imatinib treatment.

The clinical response data show that 72% of all GISTs carrying c-kit mutations showed a positive response, with 24% exhibiting stable disease. Patients with GISTs with c-kit exon 11 mutations were more likely to achieve a PR on imatinib mesylate then all other patients, with 83% of these patients exhibiting a partial response, versus 23% for patients carrying other mutations. One patient had progressive disease upon treatment with imatinib, and treatment was thus stopped with disease remaining stable. The two patients with PDGF-Rα mutations exhibited stable disease. This data is presented in Table 2. The duration of response data closely compared the clinical response data, as shown in Table 3. Again, patients with mutations in exon 11 of c-Kit exhibited a much more favorable progression-free duration then all other patients.

The conclusions that can be made from this study is that the likelihood of a clinical response to imatinib correlates with c-kit mutational status, as patients with exon 11 mutations are more likely to exhibit a positive response on imatinib therapy (83%) then all others (23%). Thus, information derived from clinomics is useful to predict the clinical outcome to imatinib therapy, however, patients with GISTs containing no known c-Kit or PDGF-Rα mutations may have benefited from imatinib treatment (2 patients out of 37) in this study, so imatinib treatment cannot be confined solely to patients with c-Kit exon 11 mutations.

A parallel study that occured in the United States exhibited similar results (Heinrich et al). In a phase II study examining 127 GIST patients for c-Kit and PDGF-Rα mutations, and seeking to correlate mutational status to imatinib response and clinical prognosis, the results depicted c-Kit exon 11 mutations as a favorable prognosis, duplicating the results presented by the european study. In patients with a c-Kit exon 11 mutation, 83% had a partial response to imatinib. In patients with a c-Kit exon 9 mutation, 47.8% had a partial response to imatinib. Conversely, patients with c-Kit TK II mutations do not respond to imatinib, while PDGF-Rα mutations in the juxtamembrane domain response to imatinib.

The idea that imatinib will only be given to GIST patients harboring certain c-Kit and PDGF-Rα mutations is certainly not out of the realm of possibility. The obvious boon of diagnosing imatinib-sensitive and resistant patients include minimizing the cost of care, as well as relying on secondary kinase inhibitors known work even in the presence of imatinib resistance such as indolinone, which inhibits both the juxtamembrane and TK II domains of c-Kit. The risk in this approach is to potentially deprive a patient thought to be resistant to a drug, when in fact this patient may be an exception to previous work, and other undescribed genetic lesions or secondary modifications may yield their tumor imatinib sensitive. Much care and research must be taken when applying these results to the clinic.