What is the Japanese NBI Expert Team (JNET) classification for colorectal lesions and how should each type be managed?

JNET Classification for Colorectal Lesions

The JNET (Japanese NBI Expert Team) classification is a narrow-band imaging magnifying endoscopy system that categorizes colorectal lesions into four types based on vessel and surface patterns, with Type 1 lesions (hyperplastic/sessile serrated polyps) requiring no resection if in the rectosigmoid, Type 2A (low-grade dysplasia) requiring standard polypectomy, Type 2B (high-grade dysplasia/superficial submucosal cancer) requiring en bloc resection when feasible, and Type 3 (deep submucosal invasive cancer) requiring surgical referral. 1, 2

Classification System Structure

The JNET classification evaluates two key endoscopic features using NBI with magnification:

Type 1 (Hyperplastic Polyp/Sessile Serrated Polyp)

• Vessel pattern: Invisible vessels
• Surface pattern: Regular dark or white spots
• Management: Leave in place if located in rectosigmoid colon; resect if >5mm or proximal to sigmoid 1, 2
• Diagnostic accuracy: Sensitivity 73-87%, specificity 95-100%, accuracy 93-99% 3, 4

Type 2A (Low-Grade Intramucosal Neoplasia)

• Vessel pattern: Regular caliber vessels
• Surface pattern: Regular tubular, branched, or papillary structures similar to surrounding mucosa
• Management: Standard polypectomy or endoscopic mucosal resection (EMR) 1, 2
• Diagnostic accuracy: Sensitivity 74-88%, specificity 72-93%, accuracy 77-84% 3, 5, 4

Type 2B (High-Grade Dysplasia/Superficial Submucosal Invasive Cancer)

• Vessel pattern: Variable caliber, irregular vessels
• Surface pattern: Irregular or obscure surface pattern
• Management: En bloc resection (endoscopic submucosal dissection preferred over piecemeal EMR) to allow accurate histologic assessment of invasion depth 1, 3
• Diagnostic accuracy: Sensitivity 44-62%, specificity 68-97%, accuracy 55-78% 3, 5, 4
• Critical caveat: This type has the lowest positive predictive value (51-55%) among all categories, so additional assessment with chromoendoscopy pit pattern may be needed for confirmation 4

Type 3 (Deep Submucosal Invasive Cancer)

• Vessel pattern: Loose vessel areas with interruption of thick vessels
• Surface pattern: Amorphous areas
• Management: Surgical referral - do not attempt endoscopic resection 1, 2
• Diagnostic accuracy: Sensitivity 38-55%, specificity 94-100%, positive predictive value 93-95% 3, 4
• Key strength: Extremely high specificity (>99%) makes it reliable for identifying lesions requiring surgery, though sensitivity is limited 4, 6

Clinical Application Algorithm

Step 1: Examine lesion with high-definition white light to assess morphology (Paris classification) and size

Step 2: Switch to NBI with magnification and focus on the most suspicious area (dominant nodule in granular LSTs, depressed areas, irregular surface)

Step 3: Assess vessel pattern first, then surface pattern to assign JNET type

Step 4: Apply management based on type:

• Type 1 in rectosigmoid: Leave in place (resect-and-discard strategy not recommended here - just leave it)
• Type 1 proximal or >5mm: Resect
• Type 2A: Standard polypectomy/EMR
• Type 2B: En bloc resection (ESD if >20mm or if piecemeal resection would compromise histologic assessment)
• Type 3: Biopsy and surgical referral

Important Clinical Nuances

Confidence level matters: The JNET classification performs best when endoscopists make high-confidence assessments. When confidence is low, default to resection and histologic evaluation 1, 7, 3

Type 2B is the weak link: With only 51-62% sensitivity and 55% positive predictive value, Type 2B lesions frequently contain either less advanced (Type 2A) or more advanced (Type 3) pathology than predicted 3, 4. However, because Type 2B carries risk of superficial submucosal invasion, always resect en bloc when feasible - the cost of under-staging is higher than over-treatment 1, 8

Morphology provides additional context:

• Nongranular LSTs have higher malignancy risk than granular LSTs (11.7% vs 5.9% submucosal invasion) 1, 7
• Depressed morphology (Paris 0-IIc) strongly suggests invasive cancer even in small lesions 1
• LST-G with mixed-sized nodules: focus JNET assessment on the largest nodule where invasion is most likely 1

Common Pitfalls to Avoid

Don't apply JNET without magnification: The classification was designed for magnifying endoscopy. While some studies suggest non-magnified NBI may work, diagnostic accuracy drops significantly 1, 3

Don't rely on JNET alone for Type 2B: Given its lower accuracy, consider supplementing with Kudo pit pattern classification using chromoendoscopy for Type 2B lesions when treatment planning requires certainty 4

Don't assume small size equals benign: JNET type trumps size for management decisions. A 6mm Type 3 lesion requires surgery, while a 30mm Type 2A can be resected endoscopically 1

Training and competency are mandatory: Studies show variable performance among non-expert endoscopists 3. Implement credentialing requirements and periodic audits before adopting JNET-based management strategies in clinical practice 9

Practical Limitations

The JNET classification is used mainly in Asian countries and has limited adoption in the United States, where the NICE classification (without magnification) predominates 1. However, for centers with magnifying endoscopy capability, JNET provides equivalent diagnostic performance to the established Kudo pit pattern classification while being simpler to apply 5, 2.