Update of epidemiology, survival and initial treatment in patients with gastrointestinal stromal tumour in the USA: a retrospective study based on SEER database

Abstract

Objectives

An updated epidemiological analysis of gastrointestinal stromal tumour (GIST), the change of cancer-specific survival (CSS) and patterns of initial treatment are of interest.

Design

A retrospective study using data from the Surveillance, Epidemiology and End Results (SEER) database.

Setting and participants

A total of 5625 patients with GIST diagnosed between 2010 and 2019 were identified.

Primary outcome measures

Age-standardised incidence rate (ASIR) and annual prevalence rate were calculated. SEER combined stage, period CSS rate and initial treatment were summarised. All the data were calculated by SEER*Stat software.

Results

From 2010 to 2019, the ASIR of GIST increased from 0.79 to 1.02 per 100 000 person-years, with an increase of 2.4% annually. The increase was across age and sex subgroups. The prevalence trend was similar with the ASIR trend in each subgroup. The stage distributions were similar between different age groups, but varied among different primary tumour sites. More importantly, a stage shift from regional stage to localized stage at diagnosis was found, which may result in the improvement of CSS over years. Overall, the 5-year CSS rate of GIST was approximately 81.3%. Even for metastatic GIST, the rate exceeded 50%. Surgery was the most common treatment regimen for GIST, followed by surgery and systemic treatment. Whereas approximately 7.0% patients were undertreated, which was more pronounced among patients with distant and unknown stages.

Conclusions

The findings of this study suggest an improving early detection of GIST and an improving ability of accurate staging. Though most patients are effectively treated and perform good survivals, approximate 7.0% patients may be undertreated.

STRENGTHS AND LIMITATIONS OF THIS STUDY

• An updated epidemiological analysis of gastrointestinal stromal tumour (GIST) by using the large data from the Surveillance, Epidemiology and End Results database.

• The treatment pattern of GIST in the real world is summarised.

• The change of cancer-specific survival over years is calculated.

• The details of the systemic treatment are lacked.

• Limited to the follow-up time, the long-term survival of GIST is not estimated.

Introduction

Gastrointestinal stromal tumour (GIST) is the mostly common mesenchymal tumour that occurred throughout the gastrointestinal tract, mainly in the stomach and small intestine, and accounts for approximately 1%–3% of all gastrointestinal malignancies.^{1 2} GIST was demonstrated to originate from interstitial cells of Cajal or the stem cell precursors, with heterogeneity and variability in terms of size and malignant behaviour.³

By now, a few studies have reported the epidemiology of GIST in the US based on the Surveillance, Epidemiology and End Results (SEER) database, while the results varied a lot before and after the implementation of a GIST-specific histology code in 2001.^3–8 The incidence was underestimated before 2001 and a significant increase was found after 2001 compared with before 2001.^{4 5} For studies focused on epidemiology data of GIST after 2001, most of them only paid attention to a single site,^{3 6–8} the difference of incidence among different primary sites were not well addressed. In addition, most studies earlier focused on identifying the risk factors and developing nomogram for predict survival,^9–12 literatures are lack on the pattern of initial treatment for GIST, and survival trend in the last decade. Hence, an updated data regarding the epidemiology, survival and treatment of GIST are of interest.

This retrospective epidemiological analysis aimed to systematically character trends of the incidence and prevalence rates of GIST in the USA during the last decade, and describe the stage distribution, survival trend and patterns of initial treatment.

Methods

SEER study population

All the data of this study were obtained from SEER 17 registries, November 2021 submission, covered approximately 26.5% of the US population. The following criteria were used to identify GIST cases: patients age ≥20 years, one primary only, histologically confirmed, not reported by autopsy or death certificate only, GIST-specific histology code (International Classification of Diseases for Oncology, 3rd Edition, code 8936), and diagnosis between 2010 and 2019. Patients with unknown race were not included. The primary tumour site codes include stomach (C160-169), small intestine (C170-C179) and uncommon sites (C150-C159, C180-C189, C199, C209-C212, C218, C220-C221, C239-C260, C268-C269, C480-C482, C488).

Age was categorised as two groups, 20–49 years and ≥50 years. The race/ethnicity was categorised into four mutually exclusive groups: white, black, American Indians/Alaska Native (AI/AN) and Asian/Pacific Islander (API). The Combined Summary Stage (2004+) in the SEER database was used to describe the tumour stage of GIST, which divided the tumour stage into four categories, that is, localized, regional, distant and unknown stage.

Age-standardised incidence rate and annual prevalence

The age-standardised incidence rate (ASIR) (2010–2019) and annual prevalence rate (2010–2018) were calculated. Both rates were age-standardised to the 2000 US standard population and were reported for the overall population and by age (20–49 or ≥50 years), sex, race, primary tumour site and SEER combined stage at diagnosis, and were presented as per 100 000 person-years. The analysis of annual prevalence was performed from 2010 to 2018, because the prevalence date is limited to 1 January 2019. The period annual percentage change (APC) of ASIRs were quantified by the National Cancer Institute’s (NCI’s) Joinpoint Regression Programme (V.4.9.0.0; https://surveillance.cancer.gov/joinpoint/), which indicate the trends or the change in rates for a period or over time. The values were compared with zero, and two-sided p<0.05 was considered to indicate statistics significance.

Distribution of SEER combined stage

The distribution of SEER combined stage (year 2010–2019) was calculated and reported for overall, and by age and primary tumour sites. In addition, the shifts in stages over time were also estimated.

Cancer-specific survival rate

Cancer-specific survival (CSS) was defined as the follow-up time from diagnosis to death due to GIST. Cases died of other reasons or alive at last follow-up were regarded as censored. One-year, 2-year, 3-year, 4-year and 5-year CSS rates were calculated. To ensure at least a 5-year follow-up, this analysis was limited to year 2010–2014.

Pattern of initial treatments

First-course treatments to GIST (year 2010–2019) were summarised in terms of no treatment, single treatment of surgery, single treatment of systemic treatment, surgery and systemic treatment, and other treatments, and reported for overall and by age group. A small number of patients (n=14) who received surgery, systemic treatment and radiation were categorised into the pattern of surgery and systemic treatment. Other treatments included single treatment of radiation, systemic treatment and radiation, and surgery and radiation. The systemic treatment included conventional chemotherapy, target therapy and immunotherapy.

Statistical software

All data were obtained and analysed through the NCI’s SEER*Stat software (V.8.4.0; https://seer.cancer.gov/seerstat/), expect for the calculation of APC.

Patient and public involvement

This study is a retrospective analysis of public dataset, patients did not get involved in any part of this study, such as the recruitment, the research question, the outcome measures and the conduct of the study.

Results

Population overview

In SEER-17 registries 2010–2019, a total of 5625 patients with GIST were identified. Of these, approximately 51.1% were male, and 82.4% aged ≥50 years. Most patients with GIST were white (66.0%), followed by black (18.8%), API (14.9%) and AI/AN (0.3%). Among 5347 GIST (95.1%) with a known SEER stage, 3641 (64.7%) were localized, 599 (10.6%) were regional and 1107 (19.7%) were distant disease at diagnosis. In terms of primary tumour sites of GIST, stomach accounted for the most (61.9%), followed by small intestine (27.0%) and other uncommon sites (11.1%). The details were summarised in online supplemental table 1.

ASIR and trend

For the overall population, the ASIR increased from 0.79 per 100 000 person-years in year 2010 to 1.02 per 100 000 person-years in 2019 (a change of 29.1%, figure 1A), with an APC of 2.4% (95% CI 1.1% to 3.7%). Though the ASIR was relatively higher for male patients over time (0.91–1.05 per 100 000 person-years for male; 0.68–0.99 per 100 000 person-years for female; figure 1A), the increased trend was more pronounced in female patients (an APC of 3.1% (95% CI 1.0% to 5.2%) for female and 1.8% (95% CI 0% to 3.6%) for male). The ASIR for patients aged ≥50 years kept 5–6 fold higher than that for patients aged 20–49 years during the last decade (figure 1B). The ASIRs and trends varied among different races (figure 1C). The ASIRs for black (APC 3.3%, 95% CI 0.9% to 5.9%) and white (APC 1.9%, 95% CI 0.2% to 3.7%) increased significantly. While the ASIRs for AI/AN and API showed no significant changes.

Figure 1.

Age-standardised incidence rate of gastrointestinal stromal tumour, USA, 2010–2019. (A) Overall and by sex, (B) by age, (C) by race, (D) by primary tumour site, (E–G) by primary tumour site and SEER combined stage. SEER, Surveillance, Epidemiology and End Results.

For all stomach GIST, the ASIR increased from 0.43 in 2010 to 0.62 in 2019 (APC 3.3%, 95% CI 2.7% to 4.0%) (figure 1D), which was mainly due to the increase of localized and distant stages (figure 1E). For all small intestine GIST, no significant change in ASIR was observed, but an increased trend for localized stage and decreased trends for regional and unknown stages were found (figure 1D, F). For all GIST located in uncommon sites, only a decreased trend in ASIR for regional stage was observed (figure 1D, G). The details of ASIR and APC data were summarised in online supplemental tables 2 and 3.

Annual prevalence

From 2010 to 2018, the annual prevalence of all GIST increased from 0.74 to 0.86 per 100 000 person-years (a change of 16.2%) (online supplemental figure 1A). Further analysis regarding the annual prevalence stratified by sex, age, race and primary tumour site were performed. As shown in online supplemental figure 1, the change in trend of annual prevalence was similar with the trend of ASIR in each subgroup.

Stage distribution

Among all identified GIST patients, localized stage accounted for 64.7%, followed by distant stage (19.7%) and regional stage (10.6%). Approximately 4.9% patients had unknown stage (figure 2A). The frequency distributions of GIST stages between two aged group were similar. The proportion of localized stage was much higher among stomach GIST (Stomach vs small intestine vs uncommon sites, 72.0 vs 58.7 vs 38.7%), whereas the proportions of regional stage (8.1 vs 14.6 vs 15.3%) and distant stage (15.6 vs 22.2 vs 36.2%) were higher among GIST located at uncommon sites (figure 2A). Notably, among stomach GIST, a relative higher proportion of localized stage, but lower proportions of regional and distant stages were found in patients aged ≥50 years compared with patients aged 20–49 years.

Figure 2.

Stage distribution of gastrointestinal stromal tumour, USA, 2010–2019. (A) Overall and by age and/or primary tumour site, (B–D) the change of stage distribution over years of all and by age.

Over the past decade, the proportion of localized stage increased from 58.5% in 2010 to 67.9% in 2019, the proportion of distant stage increased from 15.2% to 23.3%. On the contrary, the proportions of regional stage and unknown stage decreased from 17.4% to 9.2% and 7.6% to 1.7%, respectively (figure 2B). The changes in different age groups were similar with that of the overall population (figure 2C, D).

Cancer specific survival

Estimated period CSS rates (year 2010–2014) based on SEER-17 were summarised in table 1. For all GIST, the 1-year, 2-year, 3-year, 4-year and 5-year CSS rate was 94.4%, 90.7%, 87.1%, 84.2% and 81.3%, respectively. The 5- year CSS rate was 92.2% for localized stage (96.8% for age 20–49 years; 91.1% for age ≥50 years), 79.7% for regional stage (85.2% for age 20–49 years; 78.1% for age ≥50 years) and 50.7% for distant stage (68.7% for age 20–49 years; 46.8% for age ≥50 years). In addition, female patients had better period CSS rates than male patients. Among all races, black population had the worst period CSS rates. Among different tumour sites, stomach and small intestine GIST had similar period CSS rates, which were better than that of uncommon sites GIST.

Table 1.

Cancer-specific survival rate of GIST, USA, 2010–2014

Subgroup	Survival rate, %
	1 year	2 years	3 years	4 years	5 years
All GISTs	94.4	90.7	87.1	84.2	81.3
Age, years
20–49	97.5	95.5	92.0	90.5	89.1
≥50	93.7	89.5	86.0	82.7	79.5
Sex
Male	93.3	89.2	85.0	81.8	77.8
Female	95.6	92.2	89.2	86.6	84.9
Race
White	94.4	91.0	87.6	84.4	81.8
Black	94.3	89.7	84.9	82.2	78.4
AI/AN	100	100	66.7	66.7	66.7
API	94.7	90.0	87.9	85.8	83.3
Primary tumour site
Stomach	95.7	92.7	89.6	87.4	84.7
Small intestine	96.2	92.8	88.4	84.8	82.2
Uncommon sites	84.0	75.6	71.7	66.6	62.1
SEER combined stage
localized	97.9	95.9	94.9	93.5	92.2
Regional	93.0	89.8	87.5	83.8	79.7
Distant	85.8	76.5	64.3	57.2	50.7
Unknown	89.7	84.7	81.0	77.2	73.3
Age 20–49 years
SEER combined stage
localized	99.0	97.6	97.6	96.8	96.8
Regional	94.1	89.7	88.2	86.7	85.2
Distant	94.2	91.8	76.2	73.8	68.7
Unknown	100	100	92.3	84.6	80.6
Age ≥50 years
SEER combined stage
Localized	97.7	95.5	94.2	92.7	91.1
Regional	92.7	89.9	87.3	83.0	78.1
Distant	84.0	73.2	61.8	53.6	46.8
Unknown	87.4	81.1	78.4	75.6	71.6

AI/AN, American Indians/Alaska Native; API, Asian/Pacific Islander; GIST, gastrointestinal stromal tumour; SEER, Surveillance, Epidemiology and End Results.

Furthermore, the trends of period CSS were estimated. Overall, the period CSS rates increased slightly over years (online supplemental figure 2A), which were mainly contributed by the improvement of period CSS in regional stage (online supplemental figure 2C). The period CSS rates in other stages showed no obvious change (online supplemental figure 2B and D, E).

Initial treatment

Overall, approximately 45.2% of patients received surgery only and 12.4% of patients received systemic treatment only, radiation was seldom administrated for GIST patients. Approximately one out of three (35.0%) GIST patients received both surgery and systemic treatment. Only 0.4% of patients received other treatments. Notably, approximately 7.0% of all GIST patients were categorised as receiving no treatment.

Among different stages, the initial treatment strategies varied (figure 3A). For patients with localized and regional stages, more than 90% of patients received surgery. Whereas, no more than 50% of patients with distant stage received surgery. For patients with localized stage, the most common treatment modality was surgery only (60.4%), followed by surgery and systemic treatment (30.9%). For patients with regional stage, the most common treatment modality was surgery and systemic treatment (60.1%), followed by surgery only (27.5%). Whereas for patients with distant stage, the most common treatment modalities were surgery and systemic treatment (38.8%), and systemic treatment only (38.3%). Patients with unknown stage (33.5%) and distant stage (11.1%) were more prone to receive no treatment.

Figure 3.

Summary of patterns of initial treatment. (A) overall, (B, C) by age. ^aA small number of these patients (n=14) received radiation. ^bOther treatments includes single treatment of radiation, systemic treatment and radiation, surgery and radiation.

Treatment patterns in the overall population were generally consistent for patients aged ≥ 50 years and patients aged 20–49 years (figure 3B, C). Note that patients aged ≥ 50 years had relatively higher proportions of receiving surgery only and systemic treatment only, whereas a lower proportion of receiving both surgery and systemic treatment. In addition to that, patients aged ≥50 years were more prone to receive no treatment (7.5% age ≥50 years, 4.8% for age 20–49 years). The details of patterns of initial treatment stratified by age and SEER stage were presented in online supplemental table 4.

Discussion

In this study, we updated the epidemiological data of GIST diagnosed between 2010 and 2019 in the USA. We found that the overall ASIR of GIST increased from 0.79 to 1.02 per 100 000 person-years, with a change of 2.4% annually. Combined with the result of an earlier study,⁶ the ASIR of GIST showed a persistently and steady increase trend in the USA from 2001 to 2019. Globally, the incidence of GIST varied from 0.4 to 15.6 per 100 000 person-years.^{13 14}

In last decade, the ASIR was still higher among males, blacks and patients aged ≥50 years, which were consistent with epidemiological data between 2001 and 2011.⁶ The ASIR for patients aged ≥50 years kept 5–6 fold higher than that for patients aged 20–49 years over years, and the ASIR peaked among 70–79 years old.

Consistent with previous studies, stomach and small intestine were the major sites of GIST occurrence in the USA,⁶ as well as among metastatic GIST patients.¹¹ In this study, we found an increasing trend of ASIRs for stomach GIST, but not for small intestine GIST or GIST located in uncommon sites. According to Giuliano’s report,⁸ the increase of stomach GIST was more sharply in patients with primary tumour size <5 cm. In addition, they also found that the incidence of small intestine GIST only increased in patients with primary tumour size <5 cm. The occurrences of GIST located in uncommon sites were rare. A previous study demonstrated an increase of GIST located in rectum and anus,⁷ while another study found no obvious change of GIST located in colon.¹⁵ The probable reasons for the inconsistent results could be the different tumour sites and time period selection.

In this study, we first charactered the epidemiological change of GIST by both primary tumour site and tumour stage. The results turned out that the increased trend of all GIST was mainly due to the increase of localized stage of stomach and small intestine GIST, whereas the ASIRs for regional stage of small intestine and uncommon sites GIST decreased. These were consistent with the changes of stage distribution for all GIST over years. Taken together, these may suggest a stage shift from regional stage to localized stage at diagnosis. This stage shift could be explained as the result of the increasing availability of non-invasive diagnostic techniques (such as increasing cancer screening programmes through endoscopy and endoscopic ultrasonography), the increasing medical surveillance, and the better access to healthcare services.^{13 16} Note that we also found an obviously decreased trend of unknown stage. This may suggest the improved ability of classifying patients with unknown stage to more specific categories. Probably most of these patients were categorised as distant stage.

In this study, we found that GIST patients performed a good GIST-specific survival, with approximately 80% patients survived over 5 years. Even for metastatic GIST, the rate exceeded 50%. More importantly, we contributed new evidences to the literature by showing the change of CSS with years. We observed a slight improvement on CSS over years. The possible explanations may be the following. First, during the imatinib eras, adjuvant imatinib combined with surgery as the initial treatments could significantly improve the efficacy and survival benefit of GIST.^{13 17} Next, a stage shift from regional stage to localized stage may contribute to a better long-term survival. Then, GIST patients could benefit from continuing therapy with a tyrosine kinase inhibitor even after disease progression.¹⁸ A few studies had analysed the risk factors affected survival and developed nomograms to predict survival.^19–23 In summary, the probable independent prognostic variables included age at diagnosis, sex, race, tumour grade, primary tumour site, tumour size, mitotic index, stage, surgery, marital status, liver metastasis, lymphadenectomy etc. A few studies evaluated the long-term survival of GIST patients treated with imatinib. Patrikidou et al found that the median progression-free survival (mPFS) and median overall survival (mOS) was 33 months and 99 months, respectively, and the 8-year OS rate reached 50%.²⁴ Among these patients, those with KIT exon 11 mutation had the best PFS and OS. Another study also demonstrated that GIST patients treated with imatinib could obtain a long-term survival, especially for those with KIT exon 11 mutations.²⁵

We summarised the treatment pattern of GIST in the real world by using a large data. Surgery was the predominant initial treatment for both localized and regional GIST. Besides, systemic treatment also played an important role among all GIST. For GIST patients with distant stage, no more than 50% of patients received surgery. The probable reason could be the high efficacy of imatinib in terms of survival for metastatic or advanced GIST.^{26 27} Be that as it may, resection of primary tumour could significantly improve the survival of metastatic GIST patients,11 17 28–32 especially combined with target therapy. Even for part of highly select unresectable GIST, receiving cytoreductive surgery combined with target therapy could benefit for survival.³³ Hence, a positive surgical treatment should be recommended for metastatic GIST patients if possible. While it should be noted that lymphadenectomy might be correlated with dismal survival in GIST patients.²²

Despite of the effective treatments for GIST, approximate 7.0% patients may be undertreated. This proportion was much higher in patients with distant stage (11.1%) and unknown stage (33.5%), or in patients aged ≥50 years. The specific reason is unclear. However, a positive treatment should be recommended for these patients because a potential good survival they could obtain.

There are some limitations in this analysis. First, the details of the systemic treatment are lacked. Second, as the retrospective nature of SEER database, it only covers a small portion of the US population and thus is not fully representative. Hence, the true incidence might be underestimated. Third, limited to the follow-up time, the long-term survival of GIST is not estimated.

Conclusion

This study provides a substantial reference for updated epidemiology of GIST in the last decade. In this study, the ASIR of GIST increased significantly from 2010 to 2019, and across age and sex subgroups. A stage shift from regional stage to localized stage at diagnosis may reflect the increased cancer screening programmes through endoscopy and endoscopic ultrasonography, and the stage shift may also be related to the improved period CSS rates. Though most patients are effectively treated and perform good survivals, approximate 7.0% patients may be undertreated. An effort to well treat this population should be performed. In addition, a long-term follow-up is warranted to better estimate the long-term survival of GIST.

Data availability statement

Data are available on reasonable request. The datasets for this study can be obtained from the corresponding author on any reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants but as this study is a retrospective analysis of public dataset, ethical approval for this study was not required. This study is a retrospective analysis of public dataset.