In this multicenter retrospective cohort study of patients with biopsy-proven GIM, the majority of our cohort was comprised of Hispanic and Asian patients, two ethnic minorities known to have an increased risk of GAC.2 In addition, a greater than average incidence of prior H. pylori infection was seen in our cohort, which is known to occur more frequently in ethnic minorities and is a recognized driver of GIM.11 Our cohort also had a high incidence (5%) of GAC at the time of index EGD with biopsy-proven GIM, which corroborates the need to explore the utility of GIM surveillance particularly in ethnic minorities.12 The recommended interval for repeat EGD for GIM surveillance varied greatly in our study population, with no recommendation for further GIM surveillance in a majority of patients, followed by 2–3 years as the most common time interval if surveillance was recommended. This variability in surveillance recommendations is likely driven by the lack of consensus in the US on the need for GIM surveillance and the optimal time intervals for surveillance when pursued.
Of the patients who underwent a repeat EGD with biopsy during the study period, none developed low- or high-grade dysplasia nor GAC. A small proportion (14%) had multifocal GIM detected at repeat EGD, though it is possible a portion of these patients already had multifocal GIM at index EGD that was not appreciated (e.g. due to limited biopsies). Our data suggest that the progression of GIM to low- or high-grade dysplasia and GAC is slow, and the same likely applies to progression to multifocal GIM. Thus, endoscopic surveillance following an index EGD demonstrating biopsy-proven GIM can likely be deferred at least 1–2 years, mirroring the median time interval between index and repeat EGD in our study.
In our subgroup analysis of patients with multifocal GIM found at repeat EGD, in comparison to patients without multifocal GIM at repeat EGD, there was a greater percentage of females with multifocal GIM, which is not an established risk factor for GIM or GAC. Similar to prior literature, there was a greater percentage of patients with a history of H. pylori amongst those who developed multifocal GIM.1,3 Smoking history and a family history of GAC, identified as risk factors for GAC and GIM progression in prior studies, were not reproduced as significantly more prevalent in patients with multifocal GIM at repeat EGD.13,14
It should be noted that of the patients who did have multifocal GIM at repeat EGD, the majority did not have biopsies taken in more than one location at index EGD. Moreover, there were patients without visualized abnormalities endoscopically who had multifocal GIM diagnosed on biopsy. This highlights the variability in the number of biopsies routinely obtained during diagnostic EGD when a diagnosis of GIM is not known.15,16 It also suggests that the true incidence of multifocal GIM may be higher than what is diagnosed based on current biopsy practices.
Our study had several limitations. Although we abstracted data from a five-year time period and achieved a cumulative follow-up of 119 patient-years, the median duration of follow-up from the time of index EGD to repeat EGD and last clinic encounter was less than two years. Additionally, the limited number of follow-up EGDs and few patients with documented progression to multifocal GIM precluded the use of multiple logistic regression to model risk factors for progression and time to progression of GIM that could better inform surveillance intervals in our population of interest. Furthermore, univariate analyses to compare characteristics between progressors and non-progressors were not performed given that the repeated use of hypothesis testing increases the chance of a Type I error and is not a suitable alternative when the number of outcomes is insufficient for multivariate analysis. These limitations highlight the challenges in abstracting robust data in safety-net settings where low follow-up rates are common and the importance of developing strategies to mitigate this.17 Furthermore, did not measure or model cost in our study; however, we believe that appropriately spacing out surveillance EGDs based on the natural history of GIM would not only provide healthcare cost benefits, but also decrease burden on patients.