Prevalence of Extensive and Limited Gastric Intestinal Metaplasia and Progression to Dysplasia and Gastric Cancer

Abstract

Background and Aims

Guidelines cite extensive gastric intestinal metaplasia (GIM) as a bigger risk factor for gastric cancer (GC) than limited GIM and an indication for endoscopic surveillance. Data on progression of extensive GIM to GC in the USA are limited. This study aimed to estimate the prevalence and progression rates of extensive GIM in a US cohort.

Methods

This retrospective study assessed the prevalence of extensive GIM between 1/1/1990 and 8/1/2019 at a large academic medical center. Multivariable regression was used to identify predictors of extensive GIM. Incidence of GC on follow-up was calculated as number of new diagnoses divided by person-years of follow-up. Presence of GIM on subsequent follow-up endoscopy was assessed.

Results

Of 1256 individuals with GIM, 352 (28%) had extensive GIM and 904 (72%) had limited GIM. On multivariable analysis, older age (OR 1.01, 95% CI 1.00–1.02) and Hispanic ethnicity (OR 1.55, 95% CI 1.11–2.16) were predictive of extensive GIM. The annual incidence of GC for GIM overall was 0.09%. There was no difference in progression to GC between extensive or limited GIM (IRR 0, 95% CI 0–2.6), or to advanced lesions overall (IRR 0.37, 95% CI 0.04–1.62). 70% of individuals had persistent GIM on follow-up biopsy, and 22% with limited GIM had extensive GIM on follow-up biopsy.

Conclusions

28% of individuals with GIM have the extensive subtype, and are more likely to be older and of Hispanic ethnicity. There was no difference in progression to GC between extensive and limited GIM. Further research is needed to better assess risk of GIM in the US context.

Introduction

Gastric cancer is an important public health concern, accounting for over 26,000 new cases in the United States (US) annually [1]. Outcomes are poor, with 5-year survival of 31%, but improve when lesions are detected at early stages amenable to curative therapy [2, 3]. Gastric intestinal metaplasia (GIM) is an intermediate pre-cancerous lesion that increases risk of gastric adenocarcinoma, or gastric cancer (GC). Current estimates suggest that there are approximately 12.1 million adults with GIM in the USA [4, 5]

GIM provides a potential target for early intervention, and screening programs have decreased mortality in high-prevalence countries like Japan, South Korea, and China [6]. Data from Western countries, however, have been lacking. Experts have emphasized the importance of standardized endoscopic approaches, including systematic inspection and photo-documentation of the entire gastric surface, examination duration over 7 min, as well as routine use of biopsy to document extent and histologic subtype of GIM [7-12]. In practice, adherence to these recommendations is inconsistent, and documentation of extent and histologic subtype remains underutilized in the USA largely due to paucity of data regarding the impact of these findings in populations with a low prevalence of GC [7-9].

Prior studies have looked at prevalence and progression of GIM overall in Western countries, with GIM prevalence estimates ranging from 5 to 19% overall and 12–50% in high risk groups such as Hispanics, Blacks, and Asians [13-17]. Moreover, the extensive subtype of GIM (involving the corpus and antrum, rather than antrum alone) has specifically been recognized as an important risk factor for GC, and some suggest that differentiating between extensive and limited GIM aids in risk stratification [18-20]. Recent American Gastroenterological Association (AGA) guidelines and prior Management of Epithelial Precancerous Conditions in the Stomach (MAPS) II guidelines from Europe cite extensive GIM as a potential indication for surveillance [4, 19]. Nonetheless, recent reviews on GIM management underscored an absence of rigorous, well-defined studies in the USA and identified critical knowledge gaps [18]. To date, one small study assessed progression rates of the extensive GIM subtype to GC in a US cohort and a second study addressed it as a risk factor when looking at composite outcomes of advanced neoplasia, with both finding positive associations [4, 21, 22]. On the other hand, a large study from California did not find a significant association between extensive GIM and progression to GC, though the study did not describe how extensive GIM was defined [23]. Two additional studies from one VA hospital compared risk factors for developing extensive GIM, though these did not assess progression to GC and were conducted in a predominantly non-immigrant, male cohort thus limiting generalizability [24, 25].

In response to these evidence gaps, this study aimed to utilize pathology data from a large, urban medical center in the USA to estimate the prevalence of extensive and limited GIM, identify risk factors for developing extensive GIM, and determine progression rates from extensive GIM to GC. We also aimed to assess presence of persistent GIM on subsequent follow-up biopsy in individuals with GIM on initial assessment.

Materials and Methods

Study Design

In this retrospective cohort study, we identified individuals age ≥ 18 years with GIM diagnosed on upper endoscopy between 1/1/1990 and 8/1/2019 at Columbia University Irving Medical Center. Only samples from the first available endoscopy with biopsy specimens available from both the distal stomach (antrum/pre-pylorus/pylorus) and proximal stomach (body/fundus) were included to allow for accurate diagnosis of extensive and limited GIM. Individuals with GIM in the cardia alone and those with high-grade dysplasia or esophageal or gastric cancer at the time of initial biopsy were excluded (including intestinal or diffuse adenocarcinoma, neuroendocrine tumor, lymphoma, gastrointestinal stromal tumor, and squamous cell carcinoma). Individuals with a personal history of esophageal or gastric cancer, history of inflammatory bowel disease, or a known history of genetic cancer syndromes prior to the index endoscopy were also excluded. The exposure of interest was extensive GIM, with the reference group being limited GIM. The primary outcome was development of non-cardia gastric adenocarcinoma, or gastric cancer (GC). Data on age, gender, race, ethnicity, and indication for endoscopy were obtained from the electronic medical record. Participants were identified by computerized key word search of pathology reports to confirm availability of gastric biopsies, and individual pathology reports were reviewed to confirm classification of extensive vs. limited GIM, low grade dysplasia (LGD), high grade dysplasia (HGD), or GC. Biopsy location was confirmed on endoscopy reports as needed.

Procedures were characterized by indication as (1) diagnostic procedures for symptomatic evaluation (routine indications for evaluation of abdominal pain, dyspepsia, reflux, anemia, diarrhea, nausea/vomiting, bloating, variceal screening, or follow up of atrophic gastritis, celiac disease, or prior ulcers; or urgent indications for weight loss, melena, hematochezia, hematemesis, coffee ground emesis, dysphagia, odynophagia, elevated gastrin levels, or abnormal imaging), (2) surveillance procedures (done for follow up of Helicobacter pylori (H. pylori), family history of gastric cancer, follow up of Barrett’s esophagus, follow up of prior intestinal metaplasia or LGD that did not have biopsies from both proximal and distal stomach, or follow up of other mucosal nodules or gastric polyps), or (3) other diagnostic procedures (when no specific indication was available in the records).

Furthermore, we assessed whether individuals with extensive and limited GIM on initial endoscopy had persistent GIM on follow up biopsy. Those with follow up biopsies available from the proximal and distal stomach on a subsequent endoscopy were included in this analysis. Progression to LGD, HGD, and GC in each group was reported. The presence of limited GIM or extensive GIM on follow up biopsy was also assessed.

Pathology

Specimens were characterized as limited (GIM found only in the distal stomach, defined as the antrum, pre-pylorus, or pylorus) or extensive (GIM in both the proximal and distal stomach, or proximal stomach alone, defined as the body or fundus). Specimens with GIM in the cardia alone were excluded. Presence or absence of Helicobacter pylori infection on initial biopsy was also noted. The overall prevalence of extensive and limited GIM was calculated. The number of individuals developing LGD, HGD or GC among individuals with extensive vs. limited GIM was calculated. Individual outcomes were counted according to the most advanced pathology detected on biopsy. Given that evaluation and progression of intestinal metaplasia may be confounded by the presence of autoimmune atrophic gastritis, a sub-analysis was conducted after identifying and excluding individuals with autoimmune atrophic gastritis.

Statistical Analysis

Continuous variables were expressed as means and standard deviations, and differences in distributions were compared using Student t tests. Categorical variables were summarized as counts and percentages, and Chi-square tests or Fisher’s exact tests were used for comparison. Univariable and multivariable regression analyses were used to compare predictors of having the extensive subtype of GIM on initial endoscopy, having persistent GIM on a subsequent follow up endoscopy, and of having extensive GIM on follow up endoscopy in individuals with limited GIM on initial endoscopy. The number of advanced lesions (LGD, HGD, and GC) detected on follow-up was recorded. Incidence of GC was calculated as the number of new diagnoses divided by person-years of follow-up. The incidence rate ratio comparing rates of GC in individuals with extensive and limited GIM was calculated, along with associated 95% confidence intervals. Patients were censored at death (if documented in the medical record) or last documented visit. Sensitivity analysis was also done censoring at the time of last available follow up endoscopy in those with available endoscopic follow-up. An alpha of 0.05 was considered statistically significant. One prior study from Japan provides annual progression rates from limited GIM to GC to be 0% and extensive GIM to GC to be 5.6% [26]. Our sample of 1,256 individuals, with 28% extensive GIM and 72% limited GIM, would have > 90% power to detect a statistically significant difference with this effect size at a Type I error of 5%. Statistical calculations were performed in STATA 16 (Stata Corp, College Station, TX) and SAS version 9.4 (SAS Institute, Cary, NC). This study was approved by the Institutional Review Board of Columbia University Irving Medical Center.

Results

Initial review of pathology specimens identified 3,400 individuals with gastric intestinal metaplasia seen on biopsy from upper endoscopy during the study period. Of these, 1,334 individuals had biopsy specimens available from both the distal stomach (antrum/pre-pylorus/pylorus) and proximal stomach (body/fundus) to allow characterization of extensive and limited GIM. Seventy-eight of these individuals were excluded from the analysis for the following reasons: presence of esophageal or gastric cancer or dysplasia on initial EGD (n = 10), history of esophageal or gastric cancer or high grade dysplasia before initial EGD (n = 59), history of Crohn’s disease or ulcerative colitis (n = 6), and history of hereditary non-polyposis colorectal cancer syndrome (n = 3).

Of the remaining 1256 endoscopies, 352 (28%) had extensive GIM and 904 (72%) had limited GIM. Patients with extensive GIM were older than those with limited GIM at the time of initial biopsy (67.1 years vs. 64.4 years, respectively, p < 0.01; Table 1). There was no difference in the proportion of females with extensive vs. limited GIM (59% vs. 60%, respectively, p = 0.67). Those with extensive GIM were more likely to be Hispanic than those with limited GIM (32% vs. 25%, respectively, p = 0.04), though there was no difference in race between the two groups (p = 0.48). There was also no difference in rates of Helicobacter pylori infection detected on initial biopsy (p = 0.25).

Table 1.

Characteristics of individuals with extensive and limited intestinal metaplasia on gastric biopsies

Characteristic	Overall intestinal metaplasia	Extensive intestinal metaplasia	Limited intestinal metaplasia	P value
Count	1256	352 (28%)	904 (72%)
Age (mean in years)	63.6 (SD 14.1)	67.1 (SD 12.4)	64.4 (SD 14.6)	< 0.01
Sex
Female	747 (59%)	206 (59%)	541 (60%)	0.67
Male	509 (41%)	146 (41%)	363 (40%)
Race
White	512 (41%)	133 (37%)	379 (42%)	0.48
Black	134 (11%)	37 (11%)	97 (11%)
Asian/Pacific Islander	105 (8%)	34 (10%)	71 (8%)
Unknown/other	505 (40%)	148 (42%)	357 (39%)
Ethnicity
Non-Hispanic	493 (39%)	123 (35%)	370 (41%)	0.04
Hispanic	336 (27%)	111 (32%)	225 (25%)
Unknown/other	427 (34%)	118 (33%)	309 (34%)
Helicobacter pylori on biopsy
Not infected	946 (75%)	273 (78%)	673 (74%)	0.25
Infected	310 (25%)	79 (22%)	231 (26%)
Procedure indication
Routine symptoms	840 (66.8%)	229 (65%)	609 (67%)	0.01
Urgent symptoms	250 (19.9%)	63 (18%)	187 (21%)
Helicobacter pylori follow up	40 (3.1%)	17 (5%)	23 (2.5%)
Family history of GC	11 (0.9%)	2 (0.5%)	9 (1%)
Barrett’s follow up	25 (2%)	4 (1%)	21 (2%)
IM/LGD follow up	33 (2.6%)	16 (4.5%)	17 (2%)
Nodule/gastric polyp F/U	25 (2%)	11 (3%)	14 (1.5%)
No indication listed	34 (2.7%)	10 (3%)	24 (3%)

Overall, 840 (66.8%) of the procedures were done for routine indications and 250 (19.9%) were done for urgent indications (see Methods for specific indications). An additional 40 (3.1%) were done for follow up of H. pylori, 11 (0.9%) for family history of GC, 25 (2%) for follow up of Barrett’s esophagus, 33 (2.6%) for follow up of GIM or LGD, 25 (2%) for follow up of mucosal nodules or gastric polyps, and 34 (2.7%) were listed as diagnostic procedures with no specific indication.

On multivariable analysis, older age (OR 1.01, 95% CI 1.00–1.02) and Hispanic ethnicity (vs. non-Hispanic, OR 1.55, 95% CI 1.11–2.16) were associated with increased odds of extensive GIM (Table 2). As we would expect, extensive GIM was more likely to be found on a surveillance procedure than one done for symptomatic evaluation (OR 1.61, 95% CI 1.09–2.37; Table 2). Specifically, the surveillance indications associated with detection of extensive GIM were follow-up of H. pylori infection (OR 2.05, 95% CI 1.06–3.95) and follow-up of prior GIM or LGD (OR 2.27, 95% CI 1.11–4.64; Supplemental Table 1). Of note, those with prior GIM or LGD were only included in this analysis if the prior evaluation did not have biopsies of both the proximal and distal stomach for characterizing extent of GIM.

Table 2.

Multivariable analysis of factors associated with presence of extensive versus limited intestinal metaplasia among all patients with intestinal metaplasia on initial biopsy

Characteristic	Multivariable analysis
Count	Odds ratio (95% CI)	P value
Age	1.01 (1.00, 1.02)	< 0.01
Male sex	1.02 (0.79, 1.31)	0.91
Race
White	Ref	Ref
Black	1.07 (0.70, 1.65)	0.77
Asian/Pacific Islander	1.35 (0.84, 2.16)	0.22
Unknown/other	1.18 (0.83, 1.69)	0.36
Ethnicity
Non-Hispanic	Ref	Ref
Hispanic	1.55 (1.11, 2.16)	0.01
Unknown/other	1.10 (0.74, 1.62)	0.65
Helicobacter pylori infection on biopsy	0.84 (0.62, 1.13)	0.24
Procedure indication
Symptomatic evaluation*	Ref
Surveillance**	1.61 (1.09–2.37)	0.02
Unknown	1.27 (0.59–2.71)	0.54

^*Symptomatic evaluation includes all endoscopies done for routine indications (abdominal pain, dyspepsia, reflux, anemia, diarrhea, nausea/vomiting, bloating, variceal screening, or follow up of atrophic gastritis, celiac disease, or prior ulcers) and urgent indications (weight loss, melena, hematochezia, hematemesis, coffee ground emesis, dysphagia, odynophagia, elevated gastrin levels, or follow-up of abnormal imaging)

^**Surveillance includes follow up of prior H. pylori infection, family history of gastric cancer, follow up of Barrett’s esophagus, follow up of gastric intestinal metaplasia or low-grade dysplasia, or follow up of prior nodule/gastric polyp

Next, we were interested in assessing how many individuals with GIM on initial biopsy had persistent GIM on follow-up biopsy. A total of 278 individuals from the original study population had a second follow-up endoscopy with biopsies of both the proximal and distal stomach. One hundred and ninety-five (70%) of these individuals had persistent GIM on follow-up biopsy, and in 83 (30%) GIM was not detected on follow-up (Supplemental Table 2). Those with GIM on follow-up were older than those without GIM (65.5 vs. 60.7 years, respectively, p < 0.01), and were less likely to be white (38% vs. 52%, respectively) and more likely to be Asian/Pacific Islander (16% vs. 4%, respectively, p < 0.01). They were also more likely to have extensive GIM than limited GIM on initial biopsy (39% vs. 19%, respectively, p = 0.01). Of note, no advanced lesions (LGD, HGD, or GC) were found among those with no GIM on follow-up, while those with persistent GIM also developed LGD (n = 7), HGD (n = 2) and GC (n = 4). Notably, the group with persistent GIM had longer follow-up than the group without, which may account for the higher subsequent lesion detection. On multivariable analysis, age (OR 1.03, 95% CI 1.01–1.06), Asian/Pacific Islander race (compared to white, OR 6.4, 95% CI 1.76–23.22), H. pylori infection on prior biopsy (OR 1.99, 95% CI 1.01–3.91), and extensive GIM on prior biopsy (OR 2.56, 95% CI 1.34–4.91) were all predictors of persistent GIM on follow up biopsy (Table 3).

Table 3.

Multivariable analysis of factors associated with presence of intestinal metaplasia on a follow up EGD with both antrum and body biopsies available among individuals with intestinal metaplasia on initial biopsy

Characteristic	Presence of intestinal metaplasia on follow-up
Count	Odds ratio (95% CI)	P value
Age	1.04 (1.01, 1.06)	< 0.01
Male sex	1.29 (0.72, 2.31)	0.39
Race
White	Ref	Ref
Black	1.51 (0.59, 3.90)	0.39
Asian/Pacific Islander	6.40 (1.76, 23.23)	< 0.01
Unknown/other	1.20 (0.49, 2.92)	0.67
Ethnicity
Non-Hispanic	Ref	Ref
Hispanic	1.61 (0.74, 3.53)	0.23
Unknown/other	1.27 (0.49, 3.30)	0.63
Helicobacter pylori infection on biopsy	1.99 (1.01, 3.91)	0.05
Extensive intestinal metaplasia on initial biopsy	2.57 (1.34–4.92)	< 0.01

Given that evaluation of the presence and progression of intestinal metaplasia may be confounded by the presence of autoimmune atrophic gastritis, a sub-analysis was conducted after identifying and excluding individuals with autoimmune atrophic gastritis (n = 70). Of the remaining 1186 endoscopies, 283 (24%) had extensive GIM and 903 (76%) had limited GIM. On multivariable analysis, increased odds of extensive GIM were again associated with older age (OR 1.04, 95% CI 1.01–1.06; Supplemental Table 3) and undergoing a surveillance procedure (OR 1.89, 95% CI 1.27–2.82), though Hispanic ethnicity was no longer significant. Two hundred and fifty-eight of these individuals had a repeat EGD, of which 177 (69%) had persistent intestinal metaplasia. On multivariable analysis results were similar to the overall analysis, with age (OR 1.03, 95% CI 1.01–1.06), Asian/Pacific Islander race (compared to white, OR 6.77, 95% CI 1.84–24.87), H. pylori infection on prior biopsy (OR 2.27, 95% CI 1.14–4.55), and extensive GIM on prior biopsy (OR 2.23, 95% CI 1.11–4.47) being predictors of persistent GIM on follow up biopsy (Supplemental Table 4).

It has not been well established whether limited and extensive GIM remain in these distributions over time, or whether it is possible to “progress” from the limited to the extensive type of GIM. Among the 904 individuals with limited GIM, 185 had subsequent follow-up endoscopies with biopsies of the proximal and distal stomach. None of these individuals had evidence of autoimmune atrophic gastritis. One hundred and eighteen individuals had persistent GIM, among which 26 (22%) had extensive GIM on follow-up biopsy and 92 (78%) had limited GIM detected (Table 4). Of note, 16 of individuals with extensive GIM on initial biopsy “regressed” to limited GIM on follow up EGD. The overall annual incidence rate of GC was higher among individuals with extensive GIM on follow-up at 1.3%, compared to 0.44% for those with persistent limited GIM, though this difference was not statistically significant (Incidence Rate Ratio 3.04, 95% CI 0.22–41.97, Table 4). This was likely limited by a modest total duration of follow-up. On multivariable analysis, the only factor associated with detection of extensive GIM on follow-up in individuals with limited GIM on initial biopsy was male sex (OR 3.18, 95% CI 1.17–8.59, Table 5).

Table 4.

Univariable analysis of individuals with limited intestinal metaplasia on initial biopsy comparing presence of extensive and limited intestinal metaplasia on a follow up EGD with both antrum and body biopsies

Characteristics	Intestinal metaplasia on follow-up biopsy
	Overall n = 118	Extensive intestinal metaplasia n = 26 (22%)	Limited intestinal metaplasia n = 92 (78%)	P value
Median time to follow up EGD (person-years)	2.7	3.5	3.9	0.62
Age (mean in years)	64.7	64.1	64.8	0.38
Sex
Female	72 (61%)	12 (46%)	60 (65%)	0.08
Male	46 (3(%)	14 (54%)	32 (35%)
Race
White	45 (38%)	10 (38%)	35 (38%)	0.91
Black	14 (12%)	3 (12%)	11 (12%)
Asian/Pacific Islander	22 (19%)	6 (23%)	16 (17%)
Unknown/other	37 (31%)	7 (27%)	30 (33%)
Ethnicity
Non-Hispanic	49 (42%)	10 (38%)	39 (42%)	0.30
Hispanic	32 (27%)	10 (38%)	22 (24%)
Unknown/other	37 (31%)	6 (24%)	31 (34%)
Helicobacter pylori on initial biopsy
Not infected	78 (66%)	15 (58%)	63 (68%)	0.30
Infected	40 (34%)	11 (42%)	29 (32%)
Also progressed to advanced lesions (n)
Low-grade dysplasia	6	2	4	Incidence rate ratio for gastric cancer(95% confidence interval)
High-grade dysplasia	2	2	0
Gastric cancer	4	2	2
Total follow up to last EGD (person-years)	603.9	149.4	454.5	3.04 (0.22–41.97)

Table 5.

Multivariable analysis of individuals with limited intestinal metaplasia on initial biopsy comparing presence of extensive and limited intestinal metaplasia on a follow up EGD with both antrum and body biopsies

Characteristic	Presence of extensive intestinal metaplasia on follow-up
Count	Odds ratio (95% CI)	P value
Age	0.99 (0.94, 1.03)	0.53
Male sex	3.18 (1.17, 8.59)	0.02
Race
White	Ref	Ref
Black	0.61 (0.13, 2.93)	0.53
Asian/Pacific Islander	1.61 (0.44, 5.81)	0.47
Unknown/other	0.81 (0.19, 3.45)	0.77
Ethnicity
Non-Hispanic	Ref	Ref
Hispanic	2.22 (0.68, 7.27)	0.19
Unknown/other	0.80 (0.19, 3.41)	0.77
Helicobacter pylori infection on biopsy	1.58 (0.59, 4.21)	0.36

Next, we evaluated the overall progression of the extensive and limited subtypes of GIM to LGD, HGD, and GC. This study included a total of 5846.3 person-years of follow up when individuals were censored at the time of last visit or death. Over this time period, there were 5 cases of GC, 2 cases of HGD, and 8 cases of LGD in this cohort (Table 6). All cancers were detected at least 9 months after initial endoscopy. The annual incidence rate of GC for intestinal metaplasia overall was 0.09% (Table 6). There was no significant difference in the overall annual incidence rate of GC among individuals with extensive or limited GIM (Incidence Rate Ratio 0, 95% CI 0–2.6, Table 6). There was also no significant difference in incidence of advanced lesions overall (LGD, HGD, or GC) between individuals with extensive and limited GIM (Incidence Rate Ratio 0.37, 95% CI 0.04–1.62).

Table 6.

Incidence of low-grade dysplasia (LGD), high-grade dysplasia (HGD), and gastric cancer (GC) among patients with extensive and limited intestinal metaplasia

Characteristic	Overall intestinalon metaplasia	Extensive intestinal metaplasia	Limited intestinal metaplasia	Incidence rate ratio (95% confidence interval)
Total follow up to last visit (person-years)	5846.3	1724.3	4122.0
Advanced lesions (n)
LGD	8	2	6
HGD	2	0	2
GC	5	0	5
Annual incidence rate (%)
GC	0.09%	0%	0.12%	0 (0–2.6)
LGD, HGD, or GC	0.26%	0.12%	0.32%	0.37 (0.04–1.62)

Censoring at the time of last visit may result in underdetection of cancer. Therefore, a subgroup analysis was done among individuals with follow up endoscopies to confirm presence or absence of malignancy at time of censoring (n = 499). There were a total of 2280.5 person-years of follow-up in this cohort. Again, there was no significant difference in the overall annual incidence rate of GC among individuals with extensive or limited GIM (Incidence Rate Ratio 0, 95% CI 0–2.41, Supplemental Table 5), or in incidence of advanced lesions overall between individuals with extensive and limited GIM (Incidence Rate Ratio 0.17, 95% CI 0.004–1.13).

Discussion

In this study, we found that the prevalence of the extensive subtype of GIM among patients with GIM and biopsies available from both the proximal and distal stomach was 28%. These individuals were older and more likely to be of Hispanic ethnicity. 70% of individuals had persistent GIM on follow-up biopsy, and persistent GIM was associated with older age, Asian/Pacific Islander race, and presence of extensive GIM and H. pylori infection on initial biopsy. Furthermore, 22% of individuals with limited GIM on initial biopsy had extensive GIM detected on follow-up biopsy. The annual incidence rate of GC was not significantly different between individuals with extensive and limited GIM on initial biopsy. There was a trend toward increased rates of GC when extensive GIM was detected on follow-up, but was not significant.

Prior studies have looked at prevalence and progression of GIM overall in Western countries, with GIM prevalence estimates ranging from 5 to 19% overall and 12–50% in high risk groups such as Hispanics, Blacks, and Asians [13-17]. This study provides valuable insights into our understanding of gastric pre-neoplastic lesions in the context of the US population in particular. Two prior studies from the Veterans Affairs Medical Center in Houston reported the prevalence of extensive GIM among patients with GIM at around 27%, which is similar to our findings [24, 25]. These studies were well designed, with a prospective cohort of around 400 patients with GIM. However, they were limited to a predominantly non-immigrant, male population which may not be generalizable to high-risk, multi-ethnic urban populations such as the one at our institution. Furthermore, they did not report progression rates to GC. Similar to our study, these studies reported older age and Hispanic ethnicity to be more common among individuals with extensive GIM. They also found African American race to be a predictor of extensive GIM, which we did not observe in our study. This may be due to geographic variation, or differences in follow-up time and absolute numbers of individuals with extensive GIM, which were not reported by race.

To our knowledge, this is the largest study to assess the progression rate of extensive GIM (when defined as GIM extending to the proximal stomach) to GC in a US context. One prior nested case–control study by Reddy et al. (8 cases of GC, 40 controls with GIM) found a 9.4-fold higher risk in patients with extensive GIM compared to limited GIM [21]. While valuable, there were some limitations in the size and design of this study [21]. Most guidelines define extensive GIM as present in both the gastric antrum and body, while limited GIM is found solely in the antrum; however, this study defined extensive GIM as present in at least 2 gastric locations (which could have included the cardia) or GIM noted in at least 2 biopsy specimens (possibly from the same gastric anatomic location, rather than proximal and distal stomach) [18]. Another study by Dhingra et al. found extensive GIM to be a risk factor for progression from GIM to a composite outcome of LGD, HGD, or GC, though details regarding the number of cases with extensive GIM and their progression rates were not provided [22]. On the other hand, a large study by Li et al. did not find a significant association between extensive GIM and progression to GC; however, this study also did not describe how extensive GIM was defined [23]. The AGA technical review on GIM included the Reddy study above and a second Japanese study by Shichijo et al. and found a twofold increased risk of progression to GC in extensive compared to limited GIM, but notably this risk was not statistically significant [27]. Our study similarly did not find a significant difference in progression to GC between extensive or limited GIM, which is in line with the findings by both Li et al. and the AGA technical review.

Furthermore, this is the first US study to assess the natural history of extensive and limited GIM on follow-up biopsies. We found that 70% of individuals had persistent GIM on follow-up biopsy, which was associated with older age, Asian/Pacific Islander race, and presence of extensive GIM and H. pylori infection on initial biopsy. Furthermore, 22% of individuals with limited GIM on initial biopsy had extensive GIM on follow-up biopsy. While the annual incidence rate of GC was not statistically different between individuals with extensive and limited GIM on initial biopsy, there was a trend toward increased rates of GC in individuals who progress from limited to extensive GIM. Given the limited follow-up, we were likely underpowered to detect this difference. Nonetheless, it is possible that extensive GIM may be a considerable risk factor and larger studies are needed to better assess this question.

Targeting preneoplastic lesions like GIM as a route to improve the prevention and early detection of gastric cancer has been an effective tactic in eastern countries. Screening and surveillance programs in regions with high GC burden like japan, South Korea, and China have demonstrated a 40% reduction in GC mortality and have been shown to be cost effective [6, 28-34]. While similar mass screening and surveillance programs are unlikely to be effective in low-prevalence regions like the USA and other western countries, risk-stratification for targeted surveillance of such lesions in high risk groups has garnered a lot of attention in recent years. Recent American Gastroenterological Association (AGA) guidelines recommend against routine surveillance in patients with GIM, but advocate surveillance should be considered in high-risk groups, including individuals with extensive GIM [18]. Furthermore, guidelines from the UK (also a low-prevalence region) and the Management of Epithelial Precancerous Conditions in the Stomach (MAPS) II guidelines from Europe advocate screening and surveillance for individuals over age 50 with risk factors including extensive GIM [19, 20]. Nonetheless, such guidelines acknowledge the paucity of conclusive data on progression from GIM to GC from countries with low incidence of gastric cancer and highlight this as an important focus of future research.

This study has many strengths. First, our institution serves a diverse, urban population, which is likely representative of many high-risk regions across the USA. Our population was 59% female, providing greater diversity by sex than prior studies. Despite the retrospective design, we had access to a large collection of pathology specimens obtained to preserve anatomic location, with samples from both the proximal and distal stomach.

This study also had limitations. Given the retrospective, observational design, it is possible that there could be bias in what patients were selected to have biopsies of both the proximal and distal stomach, leading to higher rates of pathology. Biopsies from both regions of the stomach are necessary to define the variable of interest, extensive intestinal metaplasia, and accumulating a prospective cohort of the size analyzed in this study would be logistically challenging. It is reassuring that our rates of extensive and limited GIM were similar to those in the Tan et al. and Nguyen et al. studies, given their prospective design [24, 25]. The relatively small number of individuals progressing to GC in the cohort limited our ability to assess differences in risk between these groups. Furthermore, our study provides valuable insight into the natural history of extensive and limited GIM on follow-up biopsies. Nonetheless, given that GIM can be patchy, it is possible that GIM can simply be missed despite obtaining proximal and distal biopsies, which could have biased our results. Large, prospective studies are needed to further address these important questions.

In conclusion, we found that in a cohort from a large, urban medical center 28% of individuals with intestinal metaplasia have the extensive subtype, and are more likely to be older and of Hispanic ethnicity than individuals with limited GIM. While there was no statistically significant difference in incidence of GC among individuals with limited and extensive GIM on initial biopsy in this study, presence of extensive GIM on follow up biopsies may be a potential risk factor. Additional large studies are needed to better understand the risk associated with extensive GIM in the US population.

Abbreviations

US

United States

GIM

Gastric intestinal metaplasia

GC

Gastric cancer

LGD

Low grade dysplasia

HGD

High grade dysplasia

AGA

American Gastroenterological Association

MAPS

Management of Epithelial Precancerous Conditions in the Stomach