Aims
Examine the comparative diagnostic performance of forceps biopsy (FB) and JNET classification for identifying dysplasia and carcinoma in large laterally spreading lesions (LSLs), correlating these findings with definitive histology.
Methods
The study included 110 patients with laterally spreading lesions (LSLs) measuring ≥20 mm in diameter, excluding those with evidence of invasive carcinoma [1, 3]. Group I consisted of 70 patients (63.6%) presenting with the granular subtype (LSL-G), whereas Group II comprised 40 patients (36.4%) with the non-granular variant (LSL-NG) [4]. Optical assessment [2, 3] was performed by an expert endoscopist using Indigocarmine chromoendoscopy and NBI, after which forceps biopsy and endoscopic resection (EMR, pEMR, ESD or hybrid ESD) were carried out. All diagnostic findings were subsequently compared with histopathological results from the resected specimens. Statistical analyses included Chi-square testing, Wilson confidence intervals, descriptive metrics, and Fisher’s exact test, performed using Statistica 13.
Results
Mean lesion size was 38.8 ± 16.9 mm in LSL-G and 23.8 ± 6.4 mm in LSL-NG group. In the first group, 15.7% of lesions were benign, while 15.7% demonstrated low-grade dysplasia; high-grade dysplasia was identified in 28.6%, and carcinoma in situ in 40%. In the second group, 50% of lesions were benign, with 12.5% exhibiting low-grade dysplasia; 25% showing high-grade dysplasia, and 12.5% containing foci of carcinoma in situ.
|
Diagnostic method |
Sensitivity, % (95% СІ) |
Specificity, % (95% СІ) |
Positive prognostic value, % (95% СІ) |
Negative prognostic value, % (95% СІ) |
Diagnostic accuracy, % (95% СІ) |
p-value |
|
LSL-G |
||||||
|
JNET |
70.83 (55.89–83.12) |
90.91 (70.84–98.88) |
94.44 (81.34–99.32) |
58.82 (41.29–74.91) |
75 (62.09–85.27) |
0.000003 |
|
Forceps biopsy |
54.17 (39.18–68.53) |
95.45 (77.16–99.88) |
96.3 (81.03–99.91) |
48.84 (34.32–63.53) |
66.67 (53.31–78.31) |
0.00002
|
|
LSL-NG |
||||||
|
JNET |
73.33 (44.90–92.21) |
100.00 (86.28–100.00) |
100.00 (71.51–100.00) |
86.21 (68.34–96.11) |
90.48 (76.87–97.28) |
0.0001 |
|
Forceps biopsy |
64.29 (38.62–83.67) |
96.15 (80.36–99.90) |
90 (55.50–99.75) |
83.33 (65.28–94.36) |
85.71 (70.61–94.46) |
0.0007 |
Conclusions
The JNET classification demonstrates superior diagnostic performance compared with forceps biopsy across most evaluated metrics and serve as a more reliable method for prognosing histology changes in both granular and non-granular large LSLs.