This was a retrospective cohort study comprising patients with a preoperative diagnosis of UC. All patients underwent proctocolectomy with IPAA between 2005 and 2022 at the University of Tsukuba Hospital. All patients underwent endoscopic monitoring of the pouch. This study was approved by the ethics committee of the University of Tsukuba Hospital (R03-248) and conducted in accordance with the Declaration of Helsinki. The use of an opt-out consent approach was approved by the ethics committee of the University of Tsukuba Hospital (R03-248). Informed consent was waived.

As our standard operating protocol, we perform an annual pouchoscopy for UC patients with IPAA after ileostomy takedown. Pouchoscopy is also indicated when patients developed clinical symptoms suggestive of pouchitis. All available images from endoscopic examinations of the pouch after ileostomy takedown were retrospectively assessed by S.A. and S.M. Endoscopic findings were assessed at each anatomical location of the J pouch; the afferent limb (AL), inlet (IL), tip of the J, proximal pouch, distal pouch, and rectal cuff (Supplementary Fig. 1). Based on the Chicago classification [11], endoscopic examination findings were classified into the following 7 phenotypes: (1) normal; (2) AL involvement; (3) IL involvement; (4) diffuse inflammation of the pouch body; (5) focal inflammation of the pouch body; (6) cuffitis; and (7) pouch fistulas (Supplementary Fig. 2). Inflammatory findings on endoscopy based on the Pouchitis Disease Activity Index (PDAI) included erythema/edema, erosions/friability, ulceration, stenosis, granularity, and loss of vascular pattern [23,24]. Endoscopic examinations with no evidence of pouch inflammation were recorded as “normal pouch.” Endoscopic examinations with any evidence of inflammation in the AL, IL, or rectal cuff were recorded as “AL involvement,” “IL involvement,” or “cuffitis,” respectively. “Pouchitis” was defined as one or more inflammatory findings in the tip, proximal, or distal areas of the pouch. Two or more endoscopic findings in all anatomical locations of the pouch body were defined as “diffuse inflammation of the pouch body.” Cases of pouchitis which did not meet the criteria for diffuse inflammation were recorded as “focal inflammation of the pouch body.” [11] “Pouch fistula” was defined as any type of fistula noted on endoscopy or other imaging studies ≥ 6 months after ileostomy takedown to ensure the exclusion of fistula as a result of surgery-related complications (Supplementary Fig. 2) [25].

All available endoscopic images were evaluated according to the Chicago classification [11] to determine the overall endoscopic pouch phenotype. If all endoscopic examinations in an individual patient were reported as normal, the patient was categorized into the “normal” phenotype. If an inflammatory phenotype was identified on at least one endoscopic examination, the patient was included in the analysis for the respective phenotypic category. The finding of focal inflammation and diffuse inflammation of the pouch body on separate endoscopic examinations was recorded as “diffuse inflammation of the pouch body” rather than focal inflammation of the pouch body [11].

Data regarding the following categories were retrospectively collected: age at diagnosis of inflammatory bowel disease, age at colectomy, sex, body mass index, race, smoking status, family history of inflammatory bowel disease, disease duration until colectomy, indication for colectomy, preoperative Clostridioides difficile infection, UC disease extent based on the Montreal classification (E1, proctitis; E2, left-sided disease; E3, extensive disease) [26], number of IPAA stages, IPAA technique, presence of primary sclerosing cholangitis, postoperative complications (anastomotic leak, pelvic sepsis, abdominal abscess requiring drainage, fistula formation before ileostomy takedown, and ileus), and medication history. Clinical PDAI subscores (range, 0–6) [23,24] were also assessed at the clinical visit closest to the date of initial endoscopy at the study center.

Our primary outcome was pouch failure. Pouch failure was defined as a pouch condition requiring DLI or pouch excision and was considered a poor outcome. Secondary outcome was chronic pouchitis. Chronic pouchitis was defined as pouch conditions requiring antibiotics (e.g., ciprofloxacin and/or metronidazole) for more than 4 weeks as well as DLI [2].

In the histological analysis, we included patients with available biopsy samples obtained by endoscopy. All biopsies were undertaken at the discretion of endoscopists in clinical settings. Biopsies were formalin-fixed, paraffin-embedded, and then sectioned at 3 μm thickness. Alcian blue (pH 2.5) and HID staining were performed as previously described [16,27]. Colon and ileum samples from UC patients were used as positive and negative controls, respectively (Supplementary Fig. 3). Biopsy samples from the rectal cuff or polyps were excluded from the study analysis. For example, we excluded samples with endoscopic reports or images indicating that biopsies were obtained from the rectal cuff with almost 100% HID+ GCs. Numbers of HID-positive (HID+) and HID-negative (HID–) GCs were independently counted in each biopsy specimen by 2 investigators (S.A. and T.O.). The proportion of HID+ GCs in all GCs (HID+ GCs plus HID– GCs) was then calculated. Consistency between the 2 datasets was confirmed. Areas of disagreement or uncertainty were resolved by consensus among the authors including pathologists.

Categorical data were compared using the Fisher exact test while continuous data were compared using the Mann-Whitney U test. Using Kaplan-Meier curves, pouch survival from the date of ileostomy takedown to the date of pouch failure was estimated and compared using the log-rank test. P-values less than 0.05 were considered statistically significant. EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan) [28], which is a graphical user interface for R (The R Foundation for Statistical Computing, version 4.0.2, Vienna, Austria), was used for all data analyses.

The endoscopic phenotypes of 50 patients with a preoperative diagnosis of UC treated by proctocolectomy with IPAA were assessed. All patients were Japanese. The median age at UC diagnosis and colectomy was 32 years (range, 9–57 years) and 42 years (range, 14–66 years), respectively. The median disease duration until surgery was 4.9 years (range, 0.08–34.8 years). Extensive colitis (E3) was predominant, affecting 88% of patients. The majority of patients (70%) required colectomy due to disease refractory to medical therapy, while 9 patients (18%) had dysplasia or colorectal cancer. The majority of patients were receiving oral mesalamine or sulfasalazine (94%) and prednisone (94%) preoperatively, with 22% of patients receiving infliximab. Regarding surgical procedures, 2-stage IPAA and 1-stage IPAA were performed in 70% and 18% of patients, respectively. Stapled anastomosis was performed in the majority of cases (84%). Among the 22 patients (44%) who experienced postoperative complications, the most common postoperative complication was ileus (77.3%). Data regarding clinical characteristics, surgery, and preoperative therapy are summarized in Table 1 and Supplementary Tables 1, 2, respectively.

The present study evaluated 390 endoscopic examinations. The frequencies of each endoscopic phenotype are summarized in Supplementary Table 3. The median duration from ileostomy takedown to initial endoscopy at the study center was 8.1 months (range, 0.43–180 months). The median PDAI clinical score at the clinical visit closest to the initial endoscopy was 1 (range, 0–4) (Table 2). The median interval between these dates was 18 days (range, 0–183 days). The most common phenotype on initial endoscopy was focal inflammation (82%) (Table 2).

The findings of serial endoscopic examinations were evaluated for each patient, with an overall endoscopic phenotype assigned accordingly. The median number of endoscopic examinations per patient was 7 (range, 1–18), with a median follow-up time of 8.9 years (range, 0.6–16.9 years). The median number of endoscopic examinations per year was 0.94 (range, 0.07–3.90) (Table 2). The most frequent overall phenotype was focal inflammation (78%), followed by cuffitis (72%), IL involvement (34%), AL involvement (22%), diffuse inflammation (22%), pouch fistulas (14%), and the normal phenotype (0%) (Table 2). The order of endoscopic phenotypic frequencies in the present study was the same as for patients preoperatively diagnosed with UC in the original University of Chicago study (Supplementary Fig. 3) [29]. Of the 41 patients with focal inflammation on initial endoscopic examination, 6 patients (15%) subsequently developed diffuse inflammation.

Data on postoperative treatments demonstrated that 78% of patients were receiving loperamide. Regarding therapies for pouchitis, 52% of patients were receiving metronidazole whereas 18% of patients were receiving both metronidazole and ciprofloxacin. The proportions of patients on oral mesalamine, topical mesalamine, and topical steroids were 34%, 18%, and 18%, respectively (Supplementary Table 4). Of the 7 patients with pouch fistula, 5 patients (71%) had perianal or anorectal fistula. Five patients (71%) required seton drainage, and 2 patients (29%) were treated by incision and drainage. One patient (14%) was treated with infliximab and azathioprine (Supplementary Table 5).

Pouch failure occurred in 5 patients (10%). While no patients required pouch excision, DLI was performed for all patients with pouch failure (Table 1). The 10-year pouch survival rate was 92% (95% confidence interval [CI], 76%–97%) (Fig. 1). Patients with diffuse inflammation, pouch fistula were at significantly increased risk of pouch failure (P= 0.005 and P< 0.001, respectively) (Fig. 1). Notably, the 10-year pouch survival rates in patients with diffuse inflammation and pouch fistula were 72% (95% CI, 35%–90%) and 50% (95% CI, 11%–80%), respectively. Patients with focal inflammation had favorable pouch outcomes, with a 10-year pouch survival of 100% (Fig. 1). The rates of diffuse inflammation and pouch fistula were significantly higher in patients with pouch failure compared to patients without pouch failure (P= 0.006 and P< 0.001, respectively) (Supplementary Table 6). Twenty-four patients developed chronic pouchitis and the rates of diffuse inflammation and/or pouch fistulas were significantly higher in the patients with chronic pouchitis compared with those without chronic pouchitis (Supplementary Table 7).

To evaluate colonic metaplasia of GCs in the J pouch, we performed histological assessments of pouch biopsies with HID staining. We identified a total of 113 available biopsy specimens from 64 endoscopic examinations in 28 patients. Twenty-three specimens obtained from the rectal cuff or polyps were excluded. Of the remaining 90 biopsy specimens, 87 specimens were obtained from the pouch body (97%) and 3 were obtained from the AL (3.2%). Eight samples did not include GCs (7 from the pouch body and one from the AL).

The final cohort included 82 samples obtained from 52 endoscopic examinations in 23 patients. In almost all endoscopic examinations (96%), biopsies were performed to assess the degree of mucosal inflammation. Samples of the colon and small intestine obtained from patients with UC were used as positive and negative controls, respectively. The proportions of HID+ GCs in the colon and small intestine were 99% and 1.9%, respectively (Supplementary Fig. 4). The median proportion of HID+ GCs in biopsies obtained from the J pouch was 9.9% (range, 0%–93%) (Fig. 2A).

We divided the cohort into patients with greater than 10% HID+ GCs in at least one biopsy (high HID group, n = 13) and patients with no biopsies with greater than 10% HID+ GCs (low HID group, n = 10). We then compared demographic data and endoscopic phenotypes according to the Chicago classification between the high and low HID groups. Age at diagnosis and at colectomy were lower in the high HID group (21 [9.0–47] years and 30 [16–51] years, respectively) compared to the low HID group (29.5 [13–43] years and 38 [14–46] years, respectively). Interestingly, the high HID group had significant longer disease duration until colectomy (7.1 [0.25–14] years) compared with the low HID group (1.1 [0.08–15] years). No significant difference in the follow-up time from the date of ileostomy takedown to the last clinical visit was observed between the 2 groups (high HID group, 12 [2.4–16] years; low HID group, 12 [8.5–16] years). The rate of 2-stage IPAA in the low HID group was higher than that in the high HID group. The high HID group had numerically higher risk of chronic pouchitis compared with the low HID group (69% vs. 50%), although not statistically significant. Otherwise, no significant differences in preoperative disease extent, infliximab use, anastomosis type, postoperative antibiotics, use of mesalamine, use of topical mesalamine or steroids, or DLI were observed between the 2 groups (Tables 3, 4).

Regarding endoscopic phenotypes in this patient cohort, the most frequent overall phenotype was cuffitis (74%), followed by focal inflammation (70%), IL involvement (43%), AL involvement (35%), diffuse inflammation (30%), and pouch fistula (13%). The number of endoscopic examinations per patient was similar between the 2 groups (high HID group, 11 [1–18] examinations; low HID group, 11.5 [9–14] examinations). The proportions of patients with diffuse inflammation or pouch fistula in the high HID group (46%) was 2-fold higher than in the low HID group (20%). The frequencies of focal inflammation or AL involvement were lower in the high HID group (62% and 30%, respectively) than in the low HID group (80% and 50%, respectively) (Table 4).

To evaluate the association between a higher proportion of sulfomucin-producing GCs on pouch biopsy and endoscopic pouch phenotypes with poor outcomes, we compared HID positivity in GCs between biopsies obtained from patients with diffuse inflammation or fistula (42 specimens from 26 endoscopic examinations in 8 patients) and patients with other endoscopic phenotypes (40 specimens from 26 endoscopic examinations in 15 patients). The median proportions of HID+GCs were 25.9% (range, 2.3%–91.5%, 25 samples in 5 patients) in diffuse inflammation without fistula, 10.3% (range, 1.1%–40.7%, 17 samples in 3 patients) (Supplementary Fig. 5) in pouch fistula ( ± diffuse inflammation), and 6.0% (range, 0%–93%) in the other endoscopic phenotypes. The median proportion of HID+ GCs was significantly higher in diffuse inflammation or pouch fistula compared to other endoscopic phenotypes (14% [range 1.1%–92%] vs. 6.0% [range 0%–93%]; P= 0.011) (Figs. 2B, 3).