Is cap still useful for colon adenoma detection rate improvement in the artificial intelligence era?

Adenoma detection rate (ADR) is firmly established as one of the strongest quality indicators in colonoscopy. A lower ADR is consistently linked to a significantly increased risk of interval colorectal cancer or post-colonoscopy cancer.1 For instance, an ADR <16% doubles the risk of post-colonoscopy colorectal cancer compared with ADR ≥30% (hazard ratio, 2.01; 95% confidence interval [CI], 1.35–3.0).2 Moreover, every 10% increase in ADR is associated with a 30% to 40% reduction in colorectal cancer incidence.3 Given its profound clinical impact, ongoing efforts to enhance ADR remain essential.

Artificial intelligence (AI) has emerged as a key optical enhancement technology designed to reduce missed lesions by providing real-time visual alerts and facilitating pattern recognition. A recent meta-analysis of 44 randomized controlled trials (RCTs) demonstrated that AI-associated colonoscopy (AIC) significantly increases ADR compared with standard colonoscopy (44.7% vs. 36.7%; relative risk, 1.21; 95% CI, 1.15–1.28).4 In contrast, mechanical enhancement devices, such as a transparent cap or EndoCuff, represent another technology to increase ADR by physically expanding mucosal exposure. A recent international RCT in Asia showed that cap-associated colonoscopy (CAC) improved ADR compared with standard colonoscopy (55.1% vs. 50.0%, p<0.01).5 EndoCuff Vision, a second generation of cuff device, has also shown similar benefits, with a meta-analysis of eight RCTs reporting significantly higher ADR compared with standard high-definition colonoscopy (49.8% vs. 45.6%, p=0.02).6 Other devices, such as WingCap, have likewise shown superior ADR relative to standard colonoscopy.7

Tan et al.8 provide valuable real-world data directly comparing CAC and AIC using a propensity score-matched design. Notably, the study used a simple and cost-effective transparent cap, which is highly relevant in resource-limited environments. The study found no significant difference in ADR between CAC and AIC (47% vs. 51%, p=0.69), nor in polyp detection rate (80% vs. 71%, p=0.35).8 These findings are particularly encouraging, as they suggest that simple and low-cost cap-assisted colonoscopy alone may yield ADR improvements comparable to those achieved with AIC. Given the paucity of direct comparisons between CAC and AIC so far, this study provides meaningful insights into device performance in real-world practice. The results also align with a recent RCT comparing computer-aided detection (CADe) and EndoCuff Vision, which similarly reported no major difference in overall ADR.9

Even though technological adjuncts can effectively enhance ADR, the magnitude of benefit from AI and mechanical devices appears most significant among endoscopists with lower baseline ADR. One meta-analysis of EndoCuff Vision showed no significant improvement when the baseline ADR exceeded 50%.6 Similar results were observed in the WingCap RCT.7 Another trial found no differences between EndoCuff Vision, CAC, and high-definition colonoscopy among high-performing endoscopists.9 The benefit of AI for ADR improvement also diminishes as the operator’s baseline ADR increases,10 although some studies showed benefit across various endoscopist experience levels. However, emerging studies suggest that combining CADe with mucosal exposure devices may provide additive benefits, potentially narrowing performance variability among endoscopists with different baseline ADRs.9 Whether such combinations can ultimately overcome the limitations of each modality—their tendency to offer greater benefit mainly for diminutive polyps or among endoscopists with lower baseline ADR—remains an important area for future investigation.

While the development of AI offers opportunities in the field of endoscopy, it also introduces new concerns. A recent observational study reported a paradoxical decline in ADR for non-AI colonoscopies after endoscopists were exposed to AI, with ADR decreasing from 28.4% to 22.4%.11 While this observation requires further validation regarding its long-term impact, the potential ‘deskilling’ phenomenon nevertheless highlights an important lesson: as technology continues to evolve, the central challenge is not whether AI or mechanical enhancement devices can surpass the endoscopist, but how these tools can be responsibly integrated to support high-quality, meticulous, and consistent colonoscopy performance.

True progress in ADR will come not from devices alone, but from a balanced combination of technological innovation and unwavering expertise among endoscopists. The cap remains a highly useful and relevant tool in the era of AI. Its distinct mechanism of improving mucosal exposure, combined with its cost-effectiveness, ensures its continued value, either alongside or synergistically in combination with AIC.