Right biopsy protocol for microscopic colitis: Reduced workload and carbon footprint while maintaining diagnostic accuracy

Abstract

BACKGROUND

Histological findings of microscopic colitis (MC) are patchy, with potential for missed diagnoses. While guidelines recommend biopsies from the right and left sides of the colon, the optimal approach remains uncertain.

AIM

To interrogate diagnostic biopsies in a cohort of patients with confirmed MC with a view to improving biopsy protocols.

METHODS

We conducted a retrospective study of index colonoscopies in patients with MC over a 14-year period. We recorded demographics, symptoms, total number and location of biopsies, and whether they revealed features of lymphocytic or collagenous colitis.

RESULTS

A total of 228 index colonoscopies were included: 64% (145/228) lymphocytic colitis and 36% (83/228) collagenous colitis. The median age at diagnosis was 60 (interquartile range: 48-72) years, with 27% (61/228) aged < 50 years. Overall, 71% (162/228) of the patients were female, similar across subtypes. Males were older at diagnosis than females (62 vs 56, P = 0.03). Symptoms were known in 86% of patients (196/228), of whom 98% (192/196) had diarrhea. Among 150 cases with biopsies from the right colon, MC features were present in the right-colon specimens in 100%, with a patchy distribution in only 2% (3/150). In contrast, among 158 cases with biopsies from the left colon, MC features were present in the left-colon specimens in 92%. Among cases with paired right and left biopsies, 10% (13/136) had positive right-sided biopsies with negative left-sided biopsies; the converse was never observed. Regarding specimen containers, those from the left colon were more likely to be negative [9% (17/199) vs 2% (4/209), odds ratio: 4.79, P = 0.0025].

CONCLUSION

In the era of green endoscopy, our data support a protocol of taking biopsies from at least two sites in the right colon only, processed in a single container.

Key Words: Microscopic colitis; Colon biopsy; Biopsy protocol; Diagnostic yield; Green endoscopy; Chronic diarrhea

Core Tip: Current guidelines recommend biopsies from both the right and left colon for diagnosing microscopic colitis; however, the optimal strategy remains uncertain. In this retrospective study of 228 index colonoscopies in patients with microscopic colitis, right colon biopsies were diagnostic in all cases, while left colon biopsies had a significantly lower yield and provided no additional diagnostic value. These findings suggest that a simplified protocol of two right colon biopsies in a single container could preserve diagnostic accuracy while reducing workload, cost, and carbon footprint, consistent with green endoscopy principles.

INTRODUCTION

Microscopic colitis (MC) is a chronic inflammatory disorder of the colon. While the presentation can vary, chronic non-bloody diarrhea is a common symptom associated with significantly impaired quality of life[1]. The prevalence of MC is estimated to be 119 per 100000 persons (95% confidence interval [CI]: 73-166), with a prevalence of 50.1 per 100000 persons for collagenous colitis (CC) and 61.7 per 100000 persons for lymphocytic colitis (LC)[2]. Although endoscopic findings are increasingly recognized, the diagnosis is still made based on consistent histological findings from mostly normal-appearing colonic mucosa.

The 2018 British Society of Gastroenterology guidelines highlight the high diagnostic yield of colonoscopy for chronic diarrhea[3]. In one study of 168 patients with chronic diarrhea, colonoscopy and biopsy yielded a specific histological diagnosis in 52 patients (31%)[4]. These included inflammatory bowel disease (n = 16), MC (n = 13), ischemic colitis (n = 3), and infectious colitis (n = 6). In keeping with this, the majority of consensus guidelines agree that endoscopy, over any other investigation, is mandatory in most patients presenting with chronic diarrhea, because MC cannot be ruled out without colonic biopsies. The British Society of Gastroenterology guidelines further state that “both right- and left-sided colonic biopsies are necessary”[3], a recommendation echoed by most consensus guidelines. This recommendation reflects the recognized patchy distribution of histological findings in MC and the potential for missed diagnoses[5]. However, the optimal sampling approach remains uncertain. In a study by Shale et al[6] involving younger patients under the age of 50 years undergoing colonoscopy for diarrhea without high-risk features, 85% of inflammatory bowel disease or MC cases could have been detected by flexible sigmoidoscopy and biopsy alone. Similarly, Tanaka et al[7] reported a sensitivity of 82% for CC with samples taken from the left colon only. In addition to biopsy location, the number of biopsies needed to diagnose MC also remains debated. In a recent multicenter study from Europe by Kanstrup Fiehn et al[8], obtaining four or more biopsies was associated with greater diagnostic accuracy, as was sampling from the right colon. More recently, the same author reported findings from a larger multicenter study of histological changes in 255 patients with MC, suggesting that although histological changes were more pronounced in the proximal colon, they were seen throughout the colon with relative rectal sparing, thereby questioning the previously held belief that MC is a patchy disease[9].

In addition to considerations of diagnostic accuracy, biopsy protocols should also take environmental sustainability into account. It is now well recognized that gastroenterology, and endoscopy in particular, are major producers of both waste and greenhouse gases, and there is a growing consensus among international societies that our practices must evolve in line with the principles of reduce, reuse, and recycle to minimize environmental impact[10,11]. Random sampling and histological assessment add considerably to the carbon footprint of endoscopy procedures and should be rationalized. As such, and in keeping with the additional green endoscopy principles of rethinking and research, further studies are warranted to define the optimal biopsy protocol for accurate detection of MC while minimizing waste.

While previous studies have examined segmental histological patterns in MC[6-9], few have quantified diagnostic yield by biopsy location, assessed whether left-sided sampling adds information beyond right-sided biopsies, or examined the practical implications for sampling strategy. We therefore aimed to evaluate the diagnostic yield of colonic biopsies in MC and consider whether a simplified, sustainability-aligned sampling approach could maintain diagnostic accuracy.

MATERIALS AND METHODS

We conducted a retrospective cohort study of individuals diagnosed with MC over a 14-year and 7-month period, from January 2010 to August 2024. Colonoscopies were performed across two hospitals that share the same histopathology laboratory. All biopsies were reviewed by consultant histopathologists in the same department. Although absolute counts of intraepithelial lymphocytes and collagen band thickness were not always reported, diagnoses of MC were made using established diagnostic criteria. To identify cases, we searched all histopathology reports on the hospital’s laboratory reporting system using the keywords “LC” and “CC”. Search results were screened to exclude irrelevant returns and cases with features only suggestive of MC in which the diagnosis had been refuted following clinical correlation. We then excluded cases in which the diagnosis had been made previously, so that only index colonoscopies for MC were included in the final analysis. This ensured that histological findings reflected the baseline distribution of disease at diagnosis, unaffected by prior therapy. The process of case screening and inclusion is illustrated in the CONSORT flow diagram (Figure 1).

Figure 1  Case screening and inclusion flow diagram.

We recorded patient demographics (date of birth and gender) and bowel symptoms. Each histopathology report was reviewed to record the number of specimen containers submitted; the labelled anatomical location of each container; and whether features of LC, CC, or neither were identified in each specimen. We also recorded whether the features seen were considered definitively consistent with MC or merely suggestive. If histology was not definitive, clinical records were reviewed to determine whether the diagnosis was confirmed or refuted following clinical correlation. If a report described features of both LC and CC without concluding that one subtype was favored, the case was classified as CC, as CC can have the histological features required for a diagnosis of LC, whereas the converse is not true. We recorded the total number of individual biopsy fragments received, but did not record the number of fragments per container. Two researchers independently collected data from the electronic laboratory reporting system. To ensure accuracy, a random audit of the dataset was performed by one researcher, who cross-checked a subset of cases against the original reports to confirm consistency in data collection.

The primary outcome was the diagnostic yield of MC according to colonic location. This was assessed both at the level of biopsy sets from the right and left colon and at the level of individual specimen containers. We also included an analysis of cases in which paired samples from both sides were available.

To illustrate the potential time and carbon footprint implications of biopsy processing, an exploratory time–motion assessment was conducted in our histopathology laboratory. A laboratory scientist directly observed and timed the stages of specimen processing, including cut-up, embedding, microtomy, slide sorting and quality control, consultant reporting, and clerical typing. The activities were timed across multiple cases, and mean times were calculated for each stage.

Statistical analysis

As a retrospective, exploratory study, no power calculation was performed. Non-parametric continuous variables were compared using the Wilcoxon rank-sum test, and categorical variables were compared using the χ² test. This study was approved by the clinical audit manager of Tallaght University Hospital.

RESULTS

A total of 228 index colonoscopies diagnostic of MC were included in the final analysis. Of these, 145 (64%) were classified as LC and 83 (36%) as CC. Of the LC cases, 144 were diagnosed based on definitive histological features, and one had histological features suggestive but not definitive of LC, with the diagnosis confirmed on clinical correlation. Among the CC cases, 81 had definitive histological findings, while two had histological features suggestive but not definitive of CC and were subsequently confirmed clinically. In seven cases, histological features of both LC and CC were present; these were categorized as CC for analysis. As expected, there was a female predominance, 162 (71%) female and 66 (29%) male patients. The gender distribution was similar across subtypes. The median age at diagnosis was 60 (interquartile range: 48-72) years overall and was similar for each subtype. In total, 61 of the 228 patients (27%) were aged less than 50 years at the time of diagnosis. Notably, males were significantly older than females at diagnosis (62 years vs 56 years, P = 0.03), and females had nearly twice the odds of being diagnosed before the age of 50 years (odds ratio: 1.96). Bowel symptom data were available in 86% of patients. Of these, the vast majority (98%) had diarrhea, and only one patient did not (Table 1).

Table 1 Demographic characteristics and bowel symptoms of index colonoscopies, comparing lymphocytic and collagenous subtypes, n (%).

	Microscopic colitis	Lymphocytic colitis	Collagenous colitis
Total number	228	145	83
Age at diagnosis (yr)	60 (IQR: 48-72)	60 (IQR: 47-71)	60 (IQR: 50-74)
Male	66 (29)	47 (32)	19 (23)
Female	162 (71)	98 (68)	64 (77)
Bowel symptoms available	196 (86)	122 (84)	74 (89)
Diarrhea (n = 196)	192 (98)	119 (98)	73 (99)
Alternating motions (n = 196)	3 (1.5)	2 (1.6)	1 (1.4)
No diarrhea (n = 196)	1 (0.5)	1 (0.8)	0 (0)

IQR: Interquartile range.

The median number of individual colonic biopsies taken was 6 (interquartile range: 4-7). There was no significant difference between the mean number of biopsy containers taken from the right colon and the left colon (excluding the rectum). Table 2 describes the biopsy sampling patterns observed in the 228 index colonoscopies.

Table 2 Biopsy sampling patterns observed among 228 index colonoscopies.

Biopsy sampling pattern	n (%)
Biopsy container labelled “random colon” present	60 (26)
Only a container labelled “random colon” present	46 (20)
Full serial biopsies from all colonic segments	8 (4)
A container labelled “right colon” and one labelled “left colon”	84 (37)
Other combinations of containers	76 (33)
At least one container identified as originating from the right colon	150 (66)
At least one container identified as originating from the left colon (excluding the rectum)	158 (69)
At least one container identified from both right and left colon (excluding the rectum)	136 (60)
At least one container labelled “rectal”	117 (51)

Of note, in the eight cases in which a full set of serial colonic biopsies was provided (with a container from each colonic segment), five demonstrated features of MC in all containers, while three did not. Of the 150 cases that had at least one container labelled such that it could be identified as originating from the right colon (a total of 209 such biopsy containers), all 150 had a finding of MC in at least one of these containers (100%, 95%CI: 98-100). Only four out of the total of 209 containers labelled such that they could be identified as being from the right colon (2%, 95%CI: 1-5) were negative, representing three subjects, all males over 50 years of age. Thus, 3 out of 150 cases (2%) demonstrated a patchy distribution in the right colon.

Of the 158 cases that had at least one container labelled as originating from the left colon (excluding the rectum), 145 (92%, 95%CI: 86-95) had findings of MC in at least one of those biopsies. A total of 199 containers were identified as being from the left colon, of which 17 (9%, 95%CI: 5-13) did not demonstrate features of MC. Among cases with at least one biopsy container identifiable as being from the left colon, the likelihood of detecting MC in at least one biopsy was significantly lower than in cases with at least one right-colon container (145/158, 92%, 95%CI: 86-95 vs 150/150, 100%, 95%CI: 98-100; P = 0.0003). Furthermore, at the individual container level, left-colon containers were significantly more likely to be negative than right-colon containers (17/199, 9%, 95%CI: 5-13 vs 4/209, 2%, 95%CI: 1-5; odds ratio: 4.79, P = 0.0025). A container labelled as originating from the rectum was obtained in 117 of 228 colonoscopies (51%), of which 88 (75%, 95%CI: 67-82) demonstrated features of MC. Diagnostic yields by biopsy site among colonoscopies with samples from each location are shown in Figure 2 and Table 3. The container-level findings are detailed in Table 4.

Figure 2  Diagnostic yield of microscopic colitis by biopsy location.

Table 3 Diagnostic yield by location, calculated per set of biopsies from each site among colonoscopies with at least one container identified from that site.

Location of biopsy set	Diagnostic yield
Right colon (n = 150)	150/150 (100%, 95%CI: 98-100)
Left colon (n = 158)	145/158 (92%, 95%CI: 86-95)
Rectum (n = 117)	88/117 (75%, 95%CI: 67-82)

CI: Confidence interval.

Table 4 Diagnostic yield by location, calculated per individual biopsy container.

Location of containers	Diagnostic yield
Right colon (n = 209)	205/209 (98%, 95%CI: 95-99)
Left colon (n = 199)	182/199 (91%, 95%CI: 87-95)

CI: Confidence interval.

The statistically lower diagnostic yield for left-colon biopsies compared to right-colon biopsies remained evident in subgroup analyses for females but not for males. This difference also persisted when stratified by age (< 50 years vs ≥ 50 years). Subgroup analyses by histological subtype also demonstrated a significantly lower diagnostic yield from the left colon compared with the right for both LC and CC (P = 0.001 for both subtypes).

Among the 136 cases with both a container identifiable as originating from the right colon and another from the left colon (excluding the rectum), features of MC were seen in at least one right-sided container in all cases (100%, 95%CI: 97-100), compared with 123 cases (90%, 95%CI: 84-94) in which features of MC were seen in at least one left-sided container (P = 0.0002). In other words, while some cases had positive right-colon biopsies with negative left-colon biopsies (13/136, 10%, 95%CI: 6-16), the converse was not observed. Diagnostic yield for this paired subset is detailed in Table 5.

Table 5 Diagnostic yield by location, calculated per set of biopsies from each site, among colonoscopies with at least one container identified from the right colon and one container identified from the left colon (excluding the rectum) (n = 136).

Location of biopsy set	Diagnostic yield
Right colon	136/136 (100% 95%CI: 97-100)
Left colon	123/136 (90% 95%CI: 84-94)

CI: Confidence interval.

The time–motion assessment in our laboratory found the mean total staff time per biopsy container was 8 min and 37 s, comprising 4 min and 58 s of laboratory scientist time, 1 min and 38 s of consultant histopathologist reporting time, and 2 min of clerical typing.

DISCUSSION

We conducted a retrospective study of 228 index colonoscopies from an Irish MC cohort over a 14-year period, focusing on diagnostic yield by biopsy location. Although the median age at diagnosis was 60 years, almost a quarter of our cohort was under the age of 50 years. Virtually all patients with MC presented with diarrhea. LC was more frequent (64%) than CC, and there was a female predominance irrespective of subtype. Notably, males tended to be older at the time of diagnosis, and females had nearly twice the odds of being diagnosed before the age of 50 years.

In our cohort, all 150 colonoscopies in which biopsies were taken from the right colon had diagnostic features of MC identified in the right-colon samples, while only 4 of the 209 right-colon containers (2%) were negative. In contrast, among the 158 colonoscopies in which biopsies were taken from the left colon, 145 (92%) had diagnostic features of MC in the left-colon samples, and 182 of 199 left-colon containers (91%) were positive. These findings support a proximal predominance of pathology and a patchy distribution with rectal sparing, despite recent assertions to the contrary[9].

Critically, there was not a single case in which left-colon biopsies added diagnostic information beyond that provided by right-colon biopsies. Thus, routine left-colon sampling may represent unnecessary workload, cost, and waste. We believe that our findings could inform future sampling strategies. In the era of green endoscopy, our data support a protocol of taking biopsies from at least two sites in the right colon only, processed in a single specimen container, as opposed to the current recommendation for distinct random right- and left-colon biopsies.

In our cohort, the right-sided predominance and sampling efficacy were not affected by age; however, in terms of gender, this finding held true for females but not for males. This reflects the fact that the small number of negative right-colon containers (n = 4) came from three male patients. Right-sided predominance was consistent across both LC and CC, indicating that the observed distribution is not limited to a specific histological subtype. Despite the yield (92%) from the left colon in our study being better than that reported in previous studies (85% and 82%)[6,7], it remains insufficient for clinical purposes.

We excluded rectal biopsies and analyzed these separately from left-colon biopsies, which may account for these differences, as there is increasing evidence of rectal sparing in MC. Indeed, our findings reinforced this evidence, with rectal biopsies showing a substantially lower diagnostic yield (75%). Overall, one in 10 cases would have been overlooked if only left-colon samples had been analyzed. This supports the need for a full colonoscopy to allow proximal sampling in all patients, regardless of age, presenting with chronic diarrhea.

Our study builds on previous reports by describing biopsy sampling patterns and analyzing diagnostic yield in a large real-world cohort. It differs from earlier work by specifically quantifying diagnostic yield by biopsy location and assessing its implications for sampling practice, with efficiency and waste reduction in mind. We analyzed the yield of right- and left-sided samples both across the entire cohort and within colonoscopies in which paired samples from both sides were available. In addition, we performed a container-level analysis. To our knowledge, few prior studies have examined diagnostic yield at this level or assessed its practical implications for biopsy protocols.

Although the impact of treatment on biopsy yield is not known and could potentially alter the distribution of pathology and diagnostic accuracy, a strength of our cohort, despite its retrospective design, is the inclusion of only index cases for analysis, ensuring that all cases were treatment-naive.

Our study has several limitations. As a retrospective study, our analysis is subject to potential selection and information bias. However, several factors mitigate these effects. The cohort represents all consecutive cases diagnosed in a single laboratory during a 14-year period, and all biopsies were reviewed by consultant histopathologists using standardized diagnostic criteria. The primary outcome of histological confirmation of MC is objective. Multivariate analysis was not performed, as the study was primarily descriptive. Given the uniformly high positivity in right-sided biopsies and consistent findings across subgroup analyses, additional adjustment was unlikely to alter the conclusions.

A particular potential source of bias is the heterogeneity in biopsy sampling patterns among colonoscopies, with some cases having samples limited to one side of the colon only, as shown in Table 2. To address this, we performed a subgroup analysis limited to procedures in which biopsies were obtained from both the right and left colon (excluding the rectum). In this subset, the significantly higher diagnostic yield from the right colon remained consistent, suggesting that variability in biopsy patterns did not materially affect the principal finding of a superior yield from the right side.

Future prospective studies could further clarify the optimal biopsy strategy for MC by applying a standardized, predefined sampling protocol. Such a design would minimize selection and information bias and provide higher-quality evidence regarding segmental diagnostic yield. However, this approach would require multiple biopsies from all colonic segments in every patient, which would increase procedure time, cost, and patient burden, and would conflict with the sustainability goals of green endoscopy. For this reason, our study was designed as a pragmatic, real-world evaluation of current biopsy practices, providing findings that are directly applicable to clinical practice. Diagnostic yield may also depend on whether biopsies are obtained randomly or are endoscopically targeted.

There is an increasing recognition of endoscopic features associated with MC, including alterations in the vascular mucosal pattern, mucosal nodularity, mucosal erythema or edema, linear colonic mucosal defects, and mucosal scarring or “cat scratch” colon[12,13]. The focus of our study was the role of random biopsies in an otherwise macroscopically normal colon. Whether subtle mucosal changes were present that guided the endoscopist’s biopsy strategy is unknown and was not recorded. The impact of additional targeted biopsies on histological yield and diagnostic accuracy in MC warrants investigation but is beyond the scope of this study. Prospective studies that systematically document and compare targeted and random biopsies could clarify whether targeted sampling offers additional diagnostic benefit. Finally, as the study was conducted in two nearby hospitals in Ireland, the findings may not be fully generalizable, though both centers serve large catchment areas. To our knowledge, this is the largest Irish series to specifically assess the diagnostic yield of biopsies by colonic location. Nonetheless, future prospective studies should include multiple centers across Ireland to confirm the applicability of these findings to populations across the country.

Using data from our time–motion assessment, we explored the potential impact of a rationalized biopsy approach limited to right-colon sampling, processed in a single container. Based on a conservative assumption from published data that approximately 10 colonoscopies are required to diagnose one case of MC[4,14], and assuming one left-sided container per colonoscopy, the rationalized approach would have prevented the processing of approximately 2280 biopsy containers over the 14-year study period. Given the observed mean processing time per container and a 7-hour working day, this corresponds to an estimated saving of 27 full working days for laboratory scientists, 11 for clerical officers, and 9 for consultant histopathologists. As the processing of every three histology containers is estimated to produce carbon emissions equivalent to driving 2 miles in an average car[10], this equates to a reduction of approximately 1520 miles of car travel. These findings illustrate the potential environmental and efficiency benefits that may accompany evidence-based rationalization of biopsy protocols, as supported by our data.

CONCLUSION

This retrospective study of index colonoscopies in MC indicates that limiting evaluation to sigmoidoscopy would reduce diagnostic accuracy, underscoring the need for full colonoscopy in patients presenting with chronic diarrhea. Right-colon biopsies demonstrated superior yield, and left-colon sampling did not add diagnostic information, representing unnecessary workload, cost, and waste. These findings suggest that a simplified approach of obtaining right-colon biopsies processed in a single container would be sufficient to preserve diagnostic accuracy while reducing workload, cost, and carbon footprint, aligning with the principles of green endoscopy. However, this proposed protocol should be prospectively validated in larger, multicenter studies using a standardized biopsy protocol and clear documentation of whether biopsies are targeted or random, before it can be considered for adoption into standard practice.

ACKNOWLEDGEMENTS

We gratefully acknowledge the contributions from our colleague, Yvonne Connolly, from the Cellular Pathology Department of Tallaght University Hospital.