As endoscopic resection (ER) is routinely offered to patients with large non pedunculated colorectal polyps (LNPCP), diagnosing submucosal invasive cancer (SMIC) and differentiating resectable (adenomas and T1 SMIC) from unresectable LNPCP (≥T2 cancer) can be crucial. Crystal violet (CV) staining and magnifying endoscopy to delineate Kudo type V pit patterns may be useful for distinguishing such lesions but is almost never used in western practice and most endoscopists remain sceptical of its efficacy. We aimed to determine the value of magnifying endoscopy with CV staining compared to indigo carmine in diagnosing SMIC and determining resectability in patients with LNPCP referred to a tertiary centre.

All patients referred with LNPCP deemed suitable for ER but without confirmed invasive cancer from January 2018 to September 2025 were included. Patients with biopsy proven adenocarcinoma prior to referral were excluded. Patients were divided into a CV staining cohort after the introduction of this technique between September 2021 and September 2025 and compared to a cohort of patients referred between January 2018 and August 2021 when lesions were assessed with indigo carmine alone. Lesions were stained with crystal violet 0.05% for one minute or indigo carmine 0.2% and then examined in detail with magnifying endoscopy at 80-100x magnification. Sensitivity and specificity of CV staining and indigo carmine with magnifying endoscopy in diagnosing HGD or SMI on subsequent biopsy or examination of a resected specimen were calculated.

1448 lesions were referred for consideration of ER and assessed with magnifying endoscopy, 898 in the crystal violet staining cohort and 550 in the indigo carmine alone cohort. Patients in both cohorts were similar in terms of age, mean lesion size, sex, location, and incidence of SMIC. 82% of diagnoses of type Vi high grade (HG) pit pattern were made using CV and diagnosing any type V pit was more common in the CV cohort (p=0.02). Using CV, type Vn had a positive predictive value (PPV) for SMIC of 89% and type Vi HG pit had a PPV of 58% for SMIC. Vi HG had a PPV 83% for determining resectability (≤T1 cancer). Vn pit had a PPV of 57% for ≥ T2 cancer. Vi HG was rarely diagnosed in the indigo carmine cohort where Vn pit had a PPV of 91% and Vi undifferentiated pit had a PPV of 27% for SMIC. JNET 2b had a PPV of only 23% for SMIC.

This study suggests that the use of CV increases the likelihood of identifying any form of type V pit pattern and particularly type Vi pit pattern. Differentiating type Vi HG pit pattern is useful for diagnosing invasive cancer compared to JNET 2b vascular pattern and is highly predictive of lesions being resectable regardless of potential SMIC with a high NPV for ≥T2 cancer. Vn pit pattern is reasonably predictive of ≥T2 cancer. These findings support longstanding Japanese recommendations to use CV to further delineate surface characteristics when JNET 2b is identified.