The epidemiology of gastrointestinal stromal tumors in Taiwan, 1998–2008: a nation-wide cancer registry-based study

Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal neoplasms of the gastrointestinal system, characterized by an unique histological morphology and the expression of the KIT protein [1]. Previously, the majority of GISTs were diagnosed as smooth muscle tumors (e.g. leiomyoma and leiomyosarcoma) or as tumors of the nerve sheath origin (e.g. schwannoma and malignant nerve sheath tumors) [2, 3]. Because GISTs were previously difficult to define due to the lack of specific markers, few epidemiologic studies were published with no nation-wide cancer registry-based study of GISTs from Asia [4, 5]. The advancement of immunohistochemistry, molecular technology and the identification of KIT oncogene mutation in more than 80% of GISTs have accelerated our understanding of GISTs [6‐8]. In Taiwan, the diagnosis of GISTs by CD117 or KIT staining was established and widely adopted since 2002. Prior to 2002, the diagnosis of GISTs was based on histology and other immunohistochemical markers (CD34, vimentin, keratin, smooth muscle actin (SMA), and S100) [4, 9]. Using the Taiwan Cancer Registry (TCR) data from 1998 to 2008, our analysis elucidated the incidence and the distribution of GISTs before and after the implementation of CD117 or KIT staining for the definitive diagnosis of GISTs and compared them to those in the Western countries.

Complete surgical resection remains the only curative treatment of primary localized GISTs. The 5-yr survival rate after complete surgical resection was 50% before the era of molecular targeted therapy [10, 11]. The approval of imatinib mesylate (Gleevec®, Novartis Pharma, Basel, Switzerland), an oral inhibitor of KIT and platelet-derived growth factor receptor, alpha polypeptide (PDGFRA), to treat metastatic GIST by the USA FDA in 2002 has markedly changed the outcomes and treatment options for GISTs [12]. In Taiwan, imatinib was approved for reimbursement by the National Health Insurance Administration since 2004. Our analysis assessed the survival of GISTs by three time periods: 1) 1998–2001, before the approval imatinib to treat GISTs; 2) 2002–2004, after the approval of imatinib to treat GISTs and before the coverage of imatinib by the National Health Insurance of Taiwan; and 3) 2005–2008, after the coverage of imatinib by the National Health Insurance of Taiwan.

Results of this first Asian nation-wide cancer registry-based study of GISTs showed that the annual incidence of GISTs in Taiwan ranged from 1 to 2 cases per 100,000. In a hospital-based retrospective cohort study, Tzen et al. estimated that the incidence of GIST in Taiwan during 1998–2004 was 1.37 cases per 100,000, which was similar to our result [18]. In a hospital-based retrospective cohort study, the annual incidence of GISTs in Hong Kong was estimated to be 1.68-1.96 per 100,000 [4]. In a study based on pathology reports from 38 hospitals, Cho et al. reported that the incidence of GISTs in Korea was approximately 1.6-2.2 per 100,000 [19]. Studies from Europe and North America reported a GIST incidence of 1.45 per 100,000 in Sweden [20], 0.65-0.90 per 100,000 in Spain [21], 0.6-1.9 per 100,000 in Norway [22], 0.66 in Italy [23], 0.85-1.00 per 100,000 in France [24], 0.9 per 100,000 in Canada [25], 1.32 per 100,000 in United Kingdom [26], and 0.7 per 100,000 in USA [17]. Given the different study time periods and the lack of confirmation by KIT immunohistochemical staining in some studies, it is difficult to compare the incidence rates of GISTs across different countries; however, the published literature to date showed that the incidence rates of GISTs in different countries appeared to fall in a similar range.

Our analysis indicated that the incidence of GISTs in Taiwan increased during 1998–2008, with a more prominent rise since 2002 (Table 3 and Figure 1A). The possible reasons for the observed rise in the incidence of GIST include the improved quality of cancer registration, the advancement of diagnostic technology/method, and the increased awareness of GISTs by physicians which could be partly attributed to the emergence of effective targeted therapeutic agent, imatinib. Previously, GISTs might have been misclassified as leiomyosarcoma, leiomyoma or unspecified sarcoma. The exact diagnosis and tumor origin of GISTs were difficult to determine until the discovery of the gain-of-function mutation in the KIT oncogene. In Taiwan, the routine use of CD117 or KIT immunohistochemical staining to diagnose GIST began in 2002. Before 2002, the diagnosis of GIST was based on histology with variable use of staining markers. In addition, no unique code indicating “gastrointestinal stromal sarcoma” was available in the ICD-O-FT, which was used by the TCR before 2002. The rising incidence of GISTs in Taiwan might be attributed to the increased utilization of CD117 staining and the increased awareness of GIST by the physicians. Nevertheless, there was still a rising trend of GIST incidence from 2005 to 2008, during which the use of CD117 or KIT immunohistochemical staining had already been widely adopted for the diagnosis of GISTs. Further follow-up is necessary to clarify whether the incidence of GISTs is truly on the rise. In addition, there was a disproportional rise in the incidence of GIST arising from esophagus/others compared to those from colon/rectum, especially during the 2002–2008 period. (Table 3 and Figure 1B). The increase in the incidence of GIST from esophagus/others resulted mostly from the elevated incidence of GIST located in retroperitoneum and unspecific sites (separate data not shown). The increased awareness of physicians with a more active approach to tumors arising from non-gastrointestinal sites due to the progress in the diagnostic tools and the availability of targeted therapy may partially account for this finding.

In our study, there was a slight male predominance (M/F ratio = 1.0 ~ 1.3) in the incidence of GISTs, which was also observed by studies from Korea (M/F ratio = 1.1) [19], Norway (M/F ratio = 1.6) [22], and the United States (M/F ratio = 1.46) [17]. However, other studies reported a female excess in the number of GISTs [23‐26], while one study reported no difference by sex [20]. Taken together, it is not clear whether there is a sex difference in the incidence of GIST, and if existed, may be insignificant. The age distribution of GIST patients in our study is consistent with those reported in the literature, with the majority of GIST patients being diagnosed during the fifth to the seventh decade of life. GISTs are occasionally found in young adults, but rarely among those younger than 18 years of age. In our series, stomach was the most frequent site of involvement (47-59%) followed by small intestine (31-38%) and colon/rectum (6-9%). The site distribution of GISTs in our study is consistent with those published in the previous literature (stomach: 50-64%, small intestine: 17-44%, and colon/rectum: 2-19%) [17, 19‐26]. In our study, the percentage of GISTs originated from stomach increased with age, while the percentage of GISTs originated in the small intestine decreased with age. To our knowledge, our study is the first to report this interesting finding, which could partially be explained by the more aggressive clinical behavior of small intestine GIST [27]. The more aggressive clinical course and thus the earlier signs and symptoms of GIST from the small intestine as opposed to the more indolent behavior of GIST from other sites may lead to the diagnosis of small intestine GIST at a younger age. However, more investigations are needed to determine the causes for the differences in the percentages of GIST location with increasing age.

Surgery remains the optimal therapy for the curative treatment of GISTs, but unfortunately, more than 50% of patients will develop recurrence or metastasis. Single or combined cytotoxic chemotherapy have failed to yield a satisfactory response. Prior to the introduction of tyrosine kinase inhibitors, the outcome for patients with metastatic disease was poor with a median survival of < 2 years [28]. The prognosis of GIST improved dramatically after the introduction of imatinib, a tyrosine kinase inhibitor approved by the FDA in 2002 for treating KIT-positive GIST [29]. In Taiwan, imatinib became widely prescribed for recurrent or metastatic GISTs, after the approved coverage by the National Health Insurance Administration in 2004. In our analysis by the three time periods, the 5-yr observed OS rate of GISTs improved with the introduction of imatinib as a GIST treatment (1998–2001: 59% vs. 2002–2004: 67%), and further improved with the approved coverage of imatinib by the National Health Insurance Administration (as a proxy for a wider usage) (2005–2008: 70%). This is consistent with previous literature, with GISTs diagnosed in the pre-imatinib era having a 5-year survival ranging from 45% to 63% [4, 17, 21, 22] and GISTs occurring in the imatinib era having a better 5-year survival (79%) [30].

In our analysis, besides the year of diagnosis, female sex, younger age, and stomach location were independent favorable prognostic factors of survival. The impact of the anatomic sites of GIST on survival is equivocal in the literature. In some studies, GIST arising from the stomach was less aggressive than those from other sites while other studies showed no difference [10, 31, 32]. Our study showed that GIST arising from the stomach had a better survival rate than those affecting the small intestine. Notably, GIST from the colon/rectum exhibited the best 5-yr observed OS of 72.4%, although this survival advantage over GIST in the small intestine disappeared in the multivariable analysis, after adjusting for sex, age, and the year of diagnosis. The difference in survival between GISTs in the stomach and GISTs in the small intestine decreased with time (6.1% in 1998–2001; 5.2% in 2002–2004; 2.7% in 2005–2008), which could be attributed to the advancement in treatment, such as the use of imatinib. In our multivariable analysis, female sex was an independent favorable prognostic factor for survival (Table 5). The magnitude of survival advantage of women over men persisted (Additional file 1: Table S1, S2, S3) even during the era of imatinib treatment. Using SEER data, Tran et al. observed a survival advantage of women over men (women vs. men: 5-yr mortality risk HR = 0.83, 95% CI: 0.71-0.97) [17]. Similarly, in another cancer registry-based study of 46 c-KIT confirmed cases diagnosed in 1994–2001, women had a better 5-year survival than men (75% vs. 52%) [21]. In a cohort of 1,215 GISTs patients diagnosed between May, 2000 and October 2010, Call et al. also reported a better GIST survival in women compared to men (men vs. women: HR = 1.5, 95% CI: 1.2-1.8) [30]. It is unclear what contributes to the better survival of GISTs among women compared to men and further investigations are warranted.

This study has several strengths. This is the first nation-wide cancer registry-based study of GIST and one of the largest GIST studies from Asia. Because the GIST cases were identified from a nation-wide cancer registry, our results are population-based with a reduced probability of selection bias associated with identifying GISTs from a single or a few medical institutions. The other major strength is the long study period from 1998–2008, which spanned across the eras of pre-imatinib, transition, and imatinib, and allowed us to demonstrate the influence of change in treatment practice on the survival of GIST patients.

This study has several limitations. The TCR does not have complete information on the tumor size of GISTs and lacks data on mitotic index; therefore, risk stratification according to the Armed Forces Institute of Pathology (AFIP) criteria (also known as Miettinen’s criteria) to predict the prognosis of GISTs was not possible [33]. We used multiple ICD-O codes to represent GIST diagnosed in 1998–2001 due to the lack of an ICD-O code specific for GIST and the absence of confirmation by c-KIT staining. As a result, the incidence rates for 1998–2001 might have been overestimated due to the potential inclusion of other non-GIST mesenchymal tumors. However, studies suggested that that the majority of gastrointestinal tumors previously classified as tumors of smooth muscle, including leiomyosarcoma or nerve sheath tumors were GISTs [2, 34], which is consistent with our GISTs cases identified for the 1998–2001 period (83.5% was leiomyosarcoma, followed by 8.85% of sarcoma, not otherwise specified, and 3.36% of epithelioid leiomyosarcoma). In addition, compared to GISTs diagnosed during 2002–2008 after the establishment of c-KIT staining as part of the diagnostic protocol and identified by a single ICD-O-3 code (8936: gastrointestinal stromal sarcoma), GISTs from 1998–2001 showed similar distributions of sex, age, and primary sites (Table 1), supporting that the majority of our cases from 1998–2001 were likely GIST. Finally, although our analysis suggested that the introduction and the wider use of imatinib could contribute to the improved survival of GIST patients, it is possible that other factors may have enhanced the survival of GIST patients, including increased awareness of the disease, earlier diagnosis, improved treatment, and better overall population health.

The incidence of GISTs in Taiwan is comparable to those reported by the US and European studies. GIST is a rare cancer in Taiwan and its incidence has been increasing gradually, partially due to the advancement of diagnostic technology/method and the increased awareness of GISTs by physicians. The occurrence of GISTs is more common in men and the older population. The stomach is the most common primary site followed by the small intestine. Prognostic factors for a better survival of GIST include female sex, younger age, stomach location, and diagnostic years (likely as a proxy for change in treatment practice). Finally, our results suggest that the survival of patients with GIST has improved significantly by targeted therapy.