Reports

This report is based on medical evidence presented at sanctioned medical congress, from peer reviewed literature or opinion provided by a qualified healthcare practitioner. The consumption of the information contained within this report is intended for qualified Canadian healthcare practitioners only.

FRONTLINE Based on presentations from the Eastern Ontario Gastrointestinal Stromal Tumours (GIST) Consensus Meeting

Ottawa, Ontario / November 29, 2007

Reviewed by

Shailendra Verma, MD, FRCPC

Chair, Ottawa Regional Cancer Centre Ottawa General Hospital Assistant Professor of Medicine University of Ottawa Ottawa, Ontario

Presenters

Dr. Timothy Asmis Department of Oncology The Ottawa Hospital

Dr. Robert (Robin) Fairfull-Smith Department of Surgery The Ottawa Hospital

Dr. Najla Fasih Department of Radiology University of Ottawa

Dr. Ralph George Head, Surgical Oncology Cancer Centre of Southeastern Ontario

Dr. Richard Gregg Head, Medical Oncology Cancer Centre of Southeastern Ontario

Dr. David Hurlbut Department of Pathology and Molecular Medicine Kingston General Hospital

Dr. Oliver Keller Medical Oncologist Ottawa Regional Cancer Centre

Dr. Blair Macdonald Department of Radiology The Ottawa Hospital

Dr. Celia Marginean Department of Pathology and Molecular Medicine The Ottawa Hospital

Background

In an initiative with substantial implications for improving cancer care, an eastern Ontario consensus conference was held to evaluate current concepts in detection, staging, and treatment of gastrointestinal stromal tumours (GIST) in the context of regional resources. The issues raised during deliberations in Ottawa are likely to address concerns relevant to other regions of Canada. In GIST, new diagnostic algorithms, advances in imaging and the introduction of the tyrosine kinase inhibitors have created a need to re-evaluate management strategies in relationship to current clinical practice. This attempt to specifically match strategies to regional resources represents a new approach that might be expected to be copied elsewhere. This document summarizes conclusions reached about how to approach diagnosis, imaging, prognostication and therapeutic options in GIST within the practical resources of community centres and tertiary referral hospitals.

Introduction

It is appropriate to consider standards for cancer care in the context of regional resources and needs. A consensus conference on gastrointestinal stromal tumours (GIST) was held in Ottawa, Ontario, in late 2007 in order to consider current diagnostic and therapeutic options for GIST in eastern Ontario. The content of these deliberations are likely to be relevant to other parts of Canada where variability in access to imaging, molecular testing, and reimbursement for maintenance treatments makes it helpful to develop a regional approach.

The organization of this consensus summary resembles the format of the meeting. A review of the epidemiology and natural history of GIST was followed by a critical assessment of the current options for diagnosis and treatment. Ample discussion of these options was led by individuals active in the management of GIST. The decision to rely on local expertise rather than external experts was intentional. The goal of the meeting was to focus on the practical issues of GIST management within the context of available regional resources. In several instances, the adequacy of resources for the management of GIST was challenged, which prompted consensus strategies to address unmet needs.

In this document, a brief background on the epidemiology and pathogenesis of GIST summarizes more complete data published elsewhere. The most important sections of the consensus document involve the recommendations for diagnosis and treatment, which were addressed within the context of the perceived needs and the available regional resources. This document is envisioned as a foundation for further deliberations. Whereas the advent of tyrosine kinase (TK) inhibitors has been an important milestone in both treatment of GISTs and understanding the pathophysiology of this malignancy in molecular terms, earlier diagnosis, monitoring response to therapy, and considering additional steps to obtain long-term remission or cure remain challenges to optimal care.

Epidemiology

GISTs are the most common mesenchymal tumours of the GI tract but account for only about 1% of GI malignancies. In Canada, the incidence is unlikely to exceed 500 cases per year based on extrapolations from other industrialized countries such as the US, where the incidence rate is an estimated 6.8 cases per million individuals (Tran et al. Am J Gastroenterol 2005;100:162-8). In some countries such as Sweden, incidence rates as high as 14.5 per million have been reported (Nilsson et al. Cancer 2005;103:821-9), but one review of the geographic variability in rates concluded that differences in diagnostic sensitivity may be more important than true differences in disease incidence (Reddy et al. J Clin Pharm Ther 2007;32:557-65). The Surveillance Epidemiology and End Results (SEER) database in the US suggests that the incidence of GIST has been rising for the past decade (Perez et al. J Am Coll Surg 2006;202:623-9), but it is possible that some of the recorded increases in GIST may at least in part be due to attention drawn to this malignancy from the recent advent of improved diagnostics.

GIST can occur at any age, but it may be appropriate to consider adult and pediatric GIST independently due to differences in the pattern of presentation and prominent characteristics (Miettinen et al. Am J Surg Pathol 2005;29:1373-81). The focus of this document will be on adult (> age 30) GIST for which there is a strong correlation between increasing risk and increasing age. The median age at the time of diagnosis of GIST in the stomach is 63 years old (Miettinen et al. Am J Surg Pathol 2005;29:52-68), which is likely to be similar to that of other sites in the GI tract. Although data from SEER suggest a slightly higher incidence of GIST in men than women (54% vs. 46%), demographic data remain limited and insufficient to detect meaningful differences, if any, in the influence of other variables, such as those exerted by ethnic or socioeconomic background. To date, most data on GIST have been collected in industrialized countries.

In Canada, as in many other parts of the world, a rise in the incidence of GIST is anticipated in conjunction with the growing proportion of patients older than age 60. The expected rise in incidence provides a rationale for increasing awareness about this malignancy and defining optimal strategies for detection and management. A more rigorous focus on this malignancy is also warranted by the recent introduction of effective therapies. While substantial strides have been made recently in the targeting of molecular pathways important to proliferation of malignant cells, the efficacy of TK inhibitors in GIST has been one of the most compelling examples of cancer control achieved by targeting specific molecular processes.

Presentation

GIST, a connective tissue sarcoma, typically develops in interstitial cells of Cajal, which provide pacemaker function for peristalsis, or in related stem cells. Although GISTs have been observed throughout the GI tract, approximately 60% occur in the stomach (Corless et al. J Clin Oncol 2004;22:3813-25). Other common sites include the duodenum and small intestine, which together account for about 30% of GISTs, as well as the esophagus, the rectum, and the colon, which each account for about 2% to 5% of GISTs (Figure 1). GISTs most commonly grow endophytically, or parallel with the GI lumen. Exophytic or endophytic growth may signify advanced disease. GISTs are typically identified when they are large enough to produce symptoms. This is dependent on the anatomic site but generally occurs in tumours of 2 to 4 cm. However, tumours first diagnosed as large as 40 cm have been reported. Although only about one in four GISTs are malignant, all have the potential for malignant transformation with risk closely related to tumour size. Both mitotic rate and growth are strong predictors of malignant potential, but GISTs <2 cm are considered to pose a relatively low risk of malignancy (Fletcher et al. Hum Pathol 2002;33:459-65).

Figure 1. GIST: Clinical Presentation

GISTs represent slightly less than 30% of soft tissue sarcomas in the GI tract, second only to leiomyomas, which account for approximately 35% (Nilsson et al. Cancer 2005;103:821-9). Most other GI sarcomas are either leiomyosarcomas or leiomyoblastomas. Prior to the 1990s, stromal tumours were not well defined. Identification of this subtype of sarcoma is now possible with histopathology and molecular analysis, which are useful for prognostication. However, even with these tools, a small but substantial number of sarcomas remain difficult to classify.

A palpable abdominal mass has been reported in up to 38% of some series (Nowain et al. J Gastro Hepatol 2005;20:818-24), but abdominal pain or discomfort near the site of tumour growth is a more common presenting complaint. Bleeding related to malignant growth in the form of hemorrhage, hematochezia, or anemia is also a frequent initial symptom. Weight loss, nausea, and complications of GI tract obstruction may also be initial manifestations. In about one-third of patients, GISTs are detected in asymptomatic patients incidental to other procedures, such as endoscopy. However, the significance of GISTs <2 cm remains unclear. In a consecutive autopsy series, the incidence of asymptomatic GIST in an elderly population was 22.5% (Agaimy et al. Am J Surg Pathol 2007;31:113-20). Almost half of a subset of tumours evaluated for c-KIT or other oncogenic mutations tested positively.

In patients with a suspected GIST, the differential diagnosis includes other GI stromal tumours, nerve sheath and melanocytic tumours as well as fibroblastic tumours. Most GISTs appear as a single well-circumscribed nodule with one of three histologic patterns. Spindle cells are the most common pattern, representing about half of GISTs. Most of the remaining tumours are epithelioid, although about 10% have a mix of spindle and epithelioid cells. Morphology alone is not generally useful for distinguishing GIST from other tumour types. Although a characteristic immunohistochemical profile has played a critical role in differentiating GISTs from other stromal tumours, heterogeneity in appearance, location, and features can complicate diagnosis.

Diagnosis

The consensus discussion on diagnosis, like that of treatment, addressed the need to customize current strategies to the resources currently available in eastern Ontario. The discussion was based in particular on practical considerations such as when or whether imaging or laboratory studies not necessarily available at all centres provide an opportunity to improve management and constitute a justification for co-operative services or referral.

The critical advance in facilitating recognition of GIST and, ultimately, in targeting therapy has been the appreciation of c-KIT expression. Approximately 80% of GISTs express c-KIT (CD117), a 145-kDa transmembrane glycoprotein receptor of the TK subclass III, which also includes platelet-derived growth factor (PDGF). Although the absence of CD117 expression does not rule out GIST, mutations in the KIT gene play an important role in the dysregulated growth that characterizes most GIST Hirota et al. Science 1998;279:577-80). Mutations in the PDGF-receptor-alpha (PDGFRA) gene appear to account for dysregulated cell growth in a substantial proportion of GIST tumours that are CD117-negative. In individuals negative for mutations in both KIT and PDGFRA genes, loss of tumour suppressor genes in chromosomes 14q and 22q appear to be the most important pathogenic mechanism. Such GISTs, which represent less than 5% of cases, often have other atypical features, including clinical indolence.

KIT expression has often been helpful in confirming GIST when a diagnosis is uncertain. Although several other tumours, such as granulocytic sarcoma and small cell carcinoma, also express KIT, these can usually be distinguished by other features. While it is now clear that a small but significant proportion of GIST tumours are negative for CD117 expression, the diagnosis of CD117-negative GIST is complicated by variability in intensity of KIT expression across tumours and the risk of overstaining, which yields false-positive identifications. This emphasizes the importance of documenting multiple pathologic characteristics both for diagnosis of GIST and for judging malignant potential and course. In addition to testing for CD117 expression, suspected GIST tumours should be evaluated for histologic type, biologic potential, tumour size and location, cellularity, mitotic count, necrosis, ulceration, invasion of the lymphovascular system and, if resected, margin status (Table 1).

Table 1.
rs for Consideration

<img1511|center>

The mutational analysis of GIST tumour samples is already useful for diagnosis and prognostication, but molecular characterization of tumours has the potential to further guide management. Kinase mutation status can be used to subclassify GISTs with distinct biological and clinical behaviours. For example, about two-thirds of the activated receptor tyrosine kinase (RTK) mutations on KIT involve exon 11 (Figure 2), which predicts good response to imatinib, the first of the TK inhibitor agents that have been made available for treatment of GIST. Exon 9 mutations, which are observed in about 10% of GIST patients with KIT RTK, have been somewhat less responsive to imatinib. Point mutation D842V on exon 18, the most common of the PDGFRA m
d with imatinib resistance.

Figure 2. RTK Mutation Frequencies

<img1512|center>

A variety of other markers are demonstrating promise both in the diagnosis and staging of GIST. These include DOG1, a surface protein of unknown function that can often be detected in tumours that are negative for KIT and PDGFRA (West et al. Am J Pathol 2004;165:107-13); neural markers such as nestin and NF68; and several smooth muscle markers that have been associated with a favourable prognosis. The added value of these markers in the management of GIST is now being evaluated.

Despite its promise for guiding management, the need to perform mutational analysis in all patients is not yet established from a practical perspective that would include the issues of benefit and cost. While mutational analysis can provide confirmation of a diagnosis in at least some GISTs that have not been identified by analysis of other characteristics, such as histology, it may be premature to require these studies as a standard of care. Conversely, denying tests with the potential to yield clinically useful information until cost efficacy has been proven is problematic in a malignancy that occurs with such infrequency.

Diagnosis: The Regional Perspective

• A collaborative regional quality assurance program for diagnosis of GIST is proposed in which tissue samples are submitted to a central facility for an evaluation that includes prospective capture of tumour characteristics, including site, size, mitotic rate, histology, and mutational status.

• While mandating mutational analysis in all patients cannot be supported by current data, the potential for such information to aid in characterizing GIST is sufficient to warrant prospective data collection on these and other markers with prognostic potential in GIST.

• A quality assurance panel should be engaged to review specimens for accuracy of the diagnosis and to gather data for outcome analysis and other clinical research projects. Specifications for tumour sample purity sufficient for DNA extraction and gene sequencing will facilitate standardization.

Imaging and Histology

Computed tomography (CT), positron emission tomography (PET), and magnetic resonance imaging (MRI) offer useful information in GIST management but are not interchangeable. Of these three modalities, CT is the most readily available and has been applied for both diagnosis and staging. Although PET is generally considered more sensitive than CT scanning for documentation of early response to therapy (Figure 3), it is not typically available outside of research centres. MRI may be the most accurate tool for detection and staging of metastases, particularly in the liver, but is not a substitute for CT or PET for documenting treatment response. Although a combination of CT and PET may provide better information on disease course than either modality alone, the relative advantage, if any, of employing both tools to measure outcome has not been quantified. CT scanning is effective for assessing tumour size and for guid
y, although the value of this form of tissue sampling has not been well established.

Figure 3. GIST: Imaging

<img1513|center>

In assessing treatment effects on CT, Response Evaluation Criteria in Solid Tumours (RECIST) has been widely employed but recently challenged for its relative insensitivity to the anti-tumour activity of TK inhibitors (Benjamin et al. J Clin Oncol 2007;25:1760-4). In a literature review by Dr. D. Blair Macdonald, Assistant Professor, Department of Diagnostic Imaging, University of Ottawa, response on RECIST criteria, unlike those proposed by Choi et al. (J Clin Oncol 2007;25:1753-9), were found to correlate poorly with outcome. These data corroborate other published assertions that RECIST is no longer appropriate for monitoring treatment response.

GIST does not generally spread through the lymphatic system, a characteristic that has favoured imaging as a means of staging disease in the absence of resection. However, cases of lymph node involvement have been reported in Canada and elsewhere. Moreover, evaluation of the therapeutic response can be challenging. For example, tumour necrosis, signifying a response, is often difficult to distinguish from tumour growth, particularly on CT scanning.

Histopathologic assessments are complicated by heterogeneity of GIST, while the prognostic value of many features remains unknown. Whereas about 50% of gastric GIST is classified as spindle cell, 40% as epithelioid, and 10% as mixed, there are important subclassifications in each. For example, the spindle cell characterization can be further stratified by sclerosing, pallisading, hypercellular, or sarcomatous spindle cell appearance. Similar stratifications are possible for epithelioid gastric GISTs. In contrast, intestinal GIST is primarily expressed as spindle cell with few subclassifications.

Despite the relatively low rate of lymphatic spread of GISTs, disease-free margins of resected tumour do not necessarily signify a low recurrence risk. Case reports of recurrence after well-documented complete excisions indicate microscopic seeding of the mesentery during surgical procedures.

Currently, all GISTs >2 cm in size should be considered malignant until proven otherwise. In GISTs >2 cm but <5 cm, the malignant potential remains low in tumours with a mitotic rate of 5 or less per 50 high-powered fields (HPF). The malignant potential begins to climb with greater size or greater mitotic rate (Figure 4). Although there is a correlation between tumour size and mitotic rate, the latter appears to provide additional evidence of malignant risk but is not currently a mandatory reporting featur
te biologic marker that distinguishes GISTs with malignant potential from those without.

Figure 4. Primary GIST: Risk Factors for Recurrence After Surgery

<img1514|center>

Imaging: The Regional Perspective

•The theoretical advantages of PET over CT for monitoring response to therapy are substantial, but they do not mandate referral of patients who are at centres where PET is not currently available.

•The Choi criteria should replace RECIST as the current standard for monitoring response to therapy on CT.

•Systematic data collection is essential to evaluate the influence of imaging technique on efficacy of management. Such data can address such questions as when or if PET is warranted. In particular, PET should be used if progression is in doubt and if documenting progression may lead to a change in therapy.

Treatment

Resection remains the mainstay of therapy for resectable primary GIST with a goal of complete excision of visible and microscopic disease (DeMatteo et al. Hum Pathol 2002;33:466-77). Although laparoscopic resection is an attractive approach for reducing the morbidity of surgery, its use in tumours >2 cm in size is controversial because of a perceived risk of tumour rupture or spillage. Increasingly, radiofrequency ablation and other ablative techniques are considered an attractive alternative to conventional resection, particularly for small tumours, because these methods of tumour eradication are less invasive, less expensive, and associated with faster recovery.

However, recurrence rates after surgery remain substantial as between 40% and 90% of all GIST surgical patients subsequently have post-operative recurrence or metastasis (Eisenberg BL, Judson I. Ann Surg Oncol 2004;11:465-75), generating interest in adjuvant TK inhibitors to consolidate response. Encouraging results were generated by a small pilot study in which there was only one recurrence (4%) of GIST in 23 highrisk patients followed for three years on adjuvant imatinib vs. 32 (67%) in a 48-patient historical controls group (Nilsson et al. Br J Cancer 2007;96:1656-8). In ACSOG Z9001, which had 644 evaluable patients randomized to imatinib or placebo as adjuvant therapy after resection, an interim analysis associated imatinib with a 67.5% reduction in the risk of a relapse event at the end of one year (HR 0.325; 95% CI, 0.198–0.534; P=0.0000014) (DeMatteo et al. ASCO 2007, Abs 10079). Relapses at one year were observed in only 3% of those randomized to imatinib vs. 17% of those on placebo (Figure 5). Accrual to the trial was halted by the trial’s data monitoring committee because of the
everal other studies are now underway which are also designed to prospectively test adjuvant imatinib as a strategy to reduce recurrence risk.

Figure 5. ACOSOG Z9001 Study: Relapse Events at One Year

<img1515|center>

There is also some evidence that neoadjuvant imatinib therapy can facilitate excision of borderline resectable tumours, particularly when the goal is to reduce damage of resection to adjacent organs or to preserve organ function (Bauer et al. Int J Cancer 2005;117:316-25). In unresectable primary GIST, imatinib is associated with high response rates and extended progression-free survival (PFS) relative to no treatment (Demetri et al. J Natl Compr Canc Netw 2007;5(suppl 2):S1-S29). Although about 5% of patients are intolerant to imatinib, toxicity in most individuals is manageable with grade 3 or 4 toxicities observed in a minority of patients. In the initial phase III study of imatinib, more than half of patients had a partial response (PR) (Demetri et al. N Engl J Med 2002;347:472-80). Although there were no complete responses (CRs), approximately 70% of those without a PR achieved stable disease.

Resistance to Primary Treatment

Subsequent clinical trials have indicated that approximately 15% of GIST cases demonstrate primary resistance to imatinib. Rates of secondary resistance, caused by different molecular mechanisms, have been as high as 40% on long-term treatment (Heinrich et al. J Clin Oncol 2006;24:4764-74). In patients resistant to imatinib, response rates can be improved by doubling the standard 400-mg imatinib dose to 800 mg (Verweij et al. Lancet 2004;364:1127-34).This has been observed in those with acquired or innate resistance such that conferred by an exon 9 mutation. Alternatively, other TK inhibitors can be considered. In a placebo-controlled trial of imatinib-resistant patients with advanced GIST, the TK inhibitor sunitinib provided a substantial and highly significant (P<0.0001) improvement in the time to progression (Demetri et al. Lancet 2006;368:1329-38). In patients with recurrent or metastatic GIST, imatinib has also demonstrated significant activity and should be considered a first-line approach. Similar to unresectable disease, progression on imatinib, signifying resistance, warrants a trial of sunitinib.

The efficacy of TK inhibitors in GIST is a compelling example of the ability of targeted therapies to control malignancy, but ongoing studies are likely to substantially refine how these agents can best be applied. Important clinical questions include whether molecular studies should be performed in advance of treatment to detect primary imatinib resistance; whether PFS can be improved with longer or even indefinite courses of TK inhibitors; and whether patients with unresectable disease who achieve a response on a TK inhibitor should undergo surgery to further reduce tumour burden. In addition to the studies of neoadjuvant or adjuvant TK inhibition in patients with resectable disease, novel strategies, such as combining TK inhibitors, are likely to be pursued in an effort to exploit the activity of these agents to improve outcome.

Treatment: The Regional Perspective

• Surgery is the first-line therapy for resectable GIST. Neoadjuvant TK inhibitors are promising strategiesto reduce the risk of recurrence. Further studies are warranted. Adjuvant imatinib improves RFS, with increased benefit evident in patients with larger tumour volumes or in patients with small high-grade GISTs.

• Imatinib is the first-line therapy for unresectable GIST, metastatic GIST, or recurrent GIST. In patients with primary resistance, secondary resistance, or intolerance to imatinib, sunitinib should be prescribed. In any case, therapy should be continued until there is evidence of progression.

• Molecular characterization of GIST to select therapies for which there is n
s a reasonable if unproven management strategy.

• There is a strong likelihood that results of ongoing and planned trials will substantially refine the current treatment algorithm, particularly in regard to the use of TK inhibitors (Figure 6).

<img1516|center>

Summary

The progress in the management of GIST, although welcome, has the potential to outpace protocol adjustments, particularly at regional centres where these tumours remain uncommon. While close attention to the medical literature can provide guidance about the direction of diagnostic and therapeutic advances, a collaborative regional consensus allows new data to be interpreted in practical terms relevant to available services. The document here is intended to augment rather than supercede other treatment guidelines through a critical assessment of information from a local perspective.