Study approval was obtained from the research ethics board at Shinshu University (5359, 22 November 2021) and Tokyo Yamate Medical Center (J-155, 7 September 2022).

Twenty-five consecutive lesions of surgically resected CD-SBN from 15 patients between 2012 and 2021 were retrieved from the surgical pathology archives at Tokyo Yamate Medical Center. Hematoxylin and Eosin (H&E) sections were reviewed by three gastrointestinal pathologists (MI, HO, and RR). Neoplastic lesions were classified into dysplasia and adenocarcinoma; one case showed that dysplastic glands invaded only into the muscularis mucosae; however, no obvious submucosal invasion was identified; this case was classified into intramucosal carcinoma (pTis). A total 14 adenocarcinomas and 11 dysplasias were evaluated in this study.

At least one representative paraffin block of tumor was selected in each case for immunohistochemistry; if there was significant morphologic heterogeneity in a given case, multiple tumor blocks were selected as needed to adequately represent the entire tumor. Immunohistochemical staining was performed using commercially available antibodies with the immuno-enzyme polymer method (Histofine Simple Stain MAX PO Multi (Nichirei Biosciences, Tokyo, Japan) for MUC2, MUC5AC, MUC6, and SATB2, or Novolink Polymer Detection Systems (Leica, Wetzlar, Germany) for CDH17, CLDN18, and beta-catenin) with 3,3′-diaminobenzidine as the chromogen, or an automated slide preparation system (p53: VENTANA BenchMark ULTRA, Roche, Basel, Switzerland).

The following primary antibodies were used in accordance with the manufacturers’ instructions: CDH17 (clone: SP183; Cell Marque, Rocklin, CA, USA), CLDN18 (clone: EPR19203; Abcam, Cambridge, UK), MUC2 (clone CCP58, Agilent, Santa Clara, CA, USA), MUC5AC (clone: CLH2; Agilent), MUC6 (clone: CLH5; Novus Biologicals, Centennial, CO, USA), SATB2 (clone: EPNCIR130A; Abcam), beta-catenin (clone: EP35, Cell Marque) and p53 (clone: DO7, Agilent).Microsatellite-instability testing by immunohistochemistry for mismatch repair proteins (MMRs) (MLH1(clone: M1, Roche), MSH2 (clone: G219-1129, Roche), MSH6 (clone: SP93, Roche), and PMS2 (clone: A16-4, Roche) was conducted on an automated slide preparation system (VENTANA BenchMark ULTRA, Roche).

The extent of staining for CDH17, CLDN18, MUC2, MUC5AC, and MUC6 was scored semiquantitatively (no staining; < 10%; 10–25%; 26–50%; 51–75%; and 76–100%), and the maximum intensity was graded as negative, weak, moderate, or strong. For binary analyses, cases with 10% or more tumor cells showing moderate or strong intensity were considered positive (Supplemental Fig. 1). For CDH17 and CLDN18, cases with ≥ 40% [18] or 75% [19] tumor cells showing moderate or strong intensity were also noted which met the participation criteria of the CLDN18 clinical trials [18, 19]. To evaluate any a possible correlation with wnt pathway mutations with CDH17 or CLDN18 changes, beta-catenin staining was classified as membranous expression or nuclear expression. Expression of p53 was classified as wild type (variable weak to moderate staining) or mutant type (either diffuse strong staining or complete absence of staining). For any relationship with mismatch repair gene proteins MLH1, MSH2, MSH6, and PMS2, retained expression was defined as nuclear staining of any intensity within tumor cells, using infiltrating lymphocytes as a positive internal control. Deficient mismatch repair protein expression was defined as complete loss of expression of at least one of the 4 mismatch repair proteins. Two of the authors (MI and HO) reviewed the immunohistochemical stains and reached a consensus score.

Histology of Crohn’s disease-associated small bowel adenocarcinomas. Low-power view of infiltrating pattern (a) and high-power view of tubular morphology (b). Tubular and mucinous morphology (c) with focal signet ring cell differentiation (d)

Chi-squared test or Fisher’s exact tests were used to characterize the relationship between categorical variables. Differences were considered significant at P < 0.05. All statistical analyses were performed with EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria).