Is the Quantification of Interstitial Cells of Cajal in Gastric Biopsies in Patients with Gastroparesis Ready for Prime Time?

Gastroparesis and functional dyspepsia (FD) are two common sensorimotor disorders in the gastroduodenal region, and the distinction between both has been a matter of ongoing debate (1). The overlap in symptoms between gastroparesis and FD, especially postprandial distress syndrome, is well established (2). In addition, patients with gastroparesis frequently present with symptoms of avoidant/restrictive food intake disorder (3), and patients with eating disorders may have objectively delayed gastric emptying (4). The correlation between retardation of gastric emptying and symptoms is imperfect, either at baseline, during the gastric emptying measurement, or in response to prokinetic therapy (2,5–9), and there is significant intra-individual variation in the measured gastric emptying of solids, whether measured using scintigraphy or stable isotope measurements (10–13). This variation was also documented in 41% of 249 patients with upper GI symptoms studied in the NIH/NIDDK gastroparesis consortium follow-up cohort on two occasions 48 weeks apart (14). These were 79 of 189 patients with delayed gastric emptying at baseline which normalized (<10% retained at 4 hours) at 48 weeks, and 22 out of 60 with normal emptying at baseline which was delayed (>10% retained at 4hours) at 48 weeks (14).

In addition, the same cohort follow-up study presented highly innovative quantification of interstitial cells of Cajal (ICCs) in full-thickness biopsy specimens obtained in the gastric body from 9 patients with idiopathic gastroparesis, 9 non-diabetic patients with FD (similar symptoms but normal gastric emptying rate) who were undergoing implantation of a gastric electrical stimulator, as well as 9 controls without diabetes or gastroparesis symptoms undergoing obesity surgery (15). There were 8 females and 1 male in each of the 3 subgroups. Tissue collection followed standardized established protocols for the acquisition, and subsequent processing and analysis by a single histology core. Based on that study, it was concluded that the numbers of ICCs per high-power field in the circular muscle layer were no different between FD and gastroparesis, and hence that the pathological basis for these two conditions was similar. This observation was used to support the notion proposed by the authors that FD and idiopathic gastroparesis in tertiary care are interchangeable syndromes with common clinical and pathological features.

We have conducted further analysis of the data included in that paper (14), and we have appraised the literature regarding quantitation of ICCs using diverse methodologies and in several different species. This analysis was conducted to address the question whether quantification of the numbers of ICCs in human gastric biopsies provides conclusive insights regarding the pathogenesis, categorization of diseases presenting with common upper gastrointestinal symptoms, and potentially inform the optimal approach to treatment. Among the 9 patients in each of the gastroparesis and FD groups, there were also differences such as 2-fold numbers of overweight or obese patients and an average 20% higher BMI in the FD compared to the gastroparesis group, and twice as many gastroparesis patients were using narcotics (89% vs 44%) among the gastroparesis compared to the FD patients (14). However, it is unknown whether such potential confounders would impact the histological findings in the study.

In the cited study (14), the median and interquartile range (IQR) of ICCs in the circular muscle layer were: 5 (IQR 5,7.75) in obese controls, 3.3 (3.0, 4.1) in FD, and 1.5 (1.0, 4.1) in idiopathic gastroparesis. Statistical analysis based on analysis of variance on ranks showed significant differences (P<0.001) among the groups, and subsequent pairwise comparisons using Tukey test showed that each disease group was significantly different from controls, but there was no significant difference between the FD and gastroparesis groups, confirming the report in the article (14). However, the latter comparison was associated with insufficient power (0.184) of the performed test with α=0.050. A similar analysis was performed regarding the numbers of CD206 macrophages in the myenteric plexus, which showed 6.2 (5.0, 7.75) in controls, 4 (3.65, 4.5) in FD, and 4 (2.6, 5.5) in gastroparesis. Similar results were obtained on statistical analysis, but, again, the comparison between FD and gastroparesis was associated with insufficient power (0.05) of the performed test, with α=0.050. Therefore, confirmation of the histopathological findings in a larger cohort of patients is needed to support the conclusion that there are common pathologic features in gastroparesis and FD.

This report also raises the question: “What is known about the density and distribution of ICCs in the stomach?”. This may be relevant since the precise location of biopsies may not be identical, particularly because the different regions of the stomach are quite extensive relative to the size of a high-power field. Thus, for example, is a biopsy taken from the greater curve section of the gastric body identical to the anterior or posterior wall or the lesser curve of the same region of the stomach? It is impossible to ascertain whether there is potential for such sampling differences in some studies; therefore, we need to address evidence from the literature regarding the density and distribution of ICCs in the literature based on animal and human stomach.

Earlier studies emanating from research conducted at the University of Nevada had provided important insights on differences between the corpus, fundus, and antrum of the murine stomach. Thus, Song et al. (15) showed that, in the corpus, intramuscular ICCs (ICC-IM) were found along the greater curvature near the fundus. ICC-IM decreased in density in the circumferential axis toward the lesser curvature and in the longitudinal axis toward the antrum, as also demonstrated by an earlier study (16). ICC-IM were absent from the longitudinal layer of the antrum, but they were found in the circular layer of each region (17). There were also functional differences, since cholinergic and nitrergic motor neurones formed close contacts with ICC-IM in the corpus, but not in the antrum. It is also important to note that prior gastric surgery, such as sleeve gastrectomy in mice, resulted in significant disruption in ICCs in the gastric antrum with a loss of pacemaker activity (18), suggesting that a prior gastric intervention may introduce confounding in the interpretation of ICCs in gastric biopsies.

In the canine stomach antrum (19), pacemaker capability is observed throughout the circular muscle, usually with a dominant pacemaker near the myenteric plexus that drives slow waves, and the latter actively propagates throughout the circular muscle layer. This pacemaker activity and the active propagation pathway may occur in networks of ICCs that are distributed in the region of the myenteric plexus and throughout the circular muscle layer. There were four populations of ICCs within the antral muscularis based on the anatomical locations: (i) ICCs within the region between the circular and longitudinal muscle layers (IC-MY); (ii) intramuscular ICCs within bundles of circular and longitudinal muscle cells (IC-IM); (iii) ICCs that lie within septa between bundles of smooth muscle cells (IC-SEP); and (iv) ICCs at the submucosal surface of the circular muscle layer (IC-SM). All these different classes of ICCs, which have also been identified in human stomach (20), are able to generate pacemaker activity. Given this distribution, it is critically important for comparisons to be made in the same layer of muscle taken during gastric biopsies.

More recent studies introduce more automation in the quantitation of ICCs and confirm the same principles regarding standardization of biopsy site. Mah et al. (17) analyzed regional variations of ICCs in the murine distal stomach using confocal imaging and machine learning methods, and they demonstrated an increase in ICC-MP (myenteric plexus) volume from proximal to distal antrum, but the percent of ICC volume was similar for ICC-LM (longitudinal muscle) and for ICC-CM (circular muscle) between proximal and distal antrum and was much lower than in the ICC-MP. This suggests that, in addition to precise anatomical location of the biopsies, it is essential that counts of ICCs be performed in a specific muscle layer for comparison. These observations also suggest that, if the quantification is performed in a specific layer of muscle, there does not appear to be much difference between the proximal and distal antrum of the murine stomach. The quantitation of ICCs was conducted in the circular muscle layer of the gastric body in the human study comparing FD and gastroparesis (14). It would be interesting to know whether the ICCs in the myenteric plexus were different in the latter studies.

In a separate study of murine stomach (21) that incorporated supervised machine learning techniques to extract the ICC networks from 3-dimensional confocal microscopy images, Mah et al. showed regional variation in ICC network density and thickness along the circumferential and longitudinal axis of the mouse antrum. An inverse relationship was observed in the distal and proximal antrum for density (proximal: 9.8±4.0% vs distal: 7.6±4.6%) and thickness (proximal: 15±3 μm vs distal: 24±10 μm). A more recent study adopted the validated Fast Random Forest classification method using segmentation of the networks of ICC applied to confocal microscopy images of a whole mount distal antrum of a mouse stomach (583 × 3,376 × 133 μm³, parcellated into 24 equal image stacks). The study documented that there was a pronounced decline of up to 80% in ICC-LM (from 3,705 μm³ to 716 μm³) over 279.3 μm in the distal antrum (22). These data suggest that quantification of regional ICCs necessitates precise sampling and documentation of the site relative to the circumferential and longitudinal axes of the stomach to facilitate comparisons between disease groups. The data also suggest that the quantitation of ICCs in the circular muscle layer or myenteric plexus may be more informative than the results obtained in longitudinal muscle layer.

In human gastric biopsies obtained in the muscular layer of the junction of distal antrum and pylorus after submucosal tunneling during a G-POEM procedure, Shah et al. (23) showed that the numbers of ICCs were higher in those who experienced clinical response to the procedure than in non-responders as defined by significant improvement in nausea and vomiting after the G-POEM. It was also previously shown that, in full-thickness biopsies from both the pylorus and the antrum from patients with gastroparesis, there was greater loss of ICCs in the pylorus than in the antrum (24).

In summary, in order not to end up in a dead-end street which equates FD and gastroparesis based on histopathological analysis of gastric biopsies (25), it is critically important for standardization of the biopsy location, depth, and analysis in appraising the pathology of the enteric nervous system and pacemaker cells in health and disease. Advances in endoscopic interventions provide opportunity to standardize biopsies by focusing on circular muscle layer or myenteric plexus for ICCs by using full-thickness endoscopic biopsies at a precise site such as 2 cm proximal to the pylorus or at the incisura of the stomach (26), as well as the pyloric muscle itself in patients undergoing G-POEM procedure, as shown by Shah et al. (23). If patients are not candidates for G-POEM procedure, an alternative approach is endoscopic full-thickness resection (EFTR) which is typically used with an over-the-scope-clip-assisted system (27) for complete resection of tumors, but it may also be applied for diagnostic purposes using the full-thickness resection device (FTRD^®). Such resections are achieved effectively and safely as documented in two recent systematic reviews and meta-analyses (28,29).

Consensus guidelines for histological techniques and reporting for the assessment and diagnosis of gastrointestinal neuromuscular disease have been recommended by the Gastro 2009 International Working Group (30). These recommendations may be further enhanced through the application of standard quantification methods using machine learning methods as recently introduced (17,21). These approaches are necessary to statistically appraise differences in the numbers of ICCs or CD206 macrophages, given the estimated sample size of 24 patients in each of the two groups to demonstrate a 50% difference in the number of ICCs per hpf (mean 1.57) based on the pooled standard deviation (1.913) in the numbers of ICCs observed in the study from the NIH gastroparesis consortium (14). The estimated sample size to detect a 25% difference in the numbers of ICCs per hpf would be 95 patients in each group. These calculations lead to the conclusion that the sample size examined to date (14) was insufficient to conclusively support the hypothesis that there are no differences in the numbers of ICCs in gastric biopsies between gastroparesis and FD in addition to the potential differences in the sites of the full-thickness biopsies obtained in the gastric body in light of the published literature showing the variation in numbers of ICCs based on site of biopsies in the longitudinal or circumferential axis of the stomach.

Data transparency statement:

All relevant data are included in the paper; as the data are from review of published literature, there are no new data presented.