Endoscopy 2022; 54(S 01): S58
DOI: 10.1055/s-0042-1744683
Abstracts | ESGE Days 2022
ESGE Days 2022 Oral presentations
10:00–11:00 Friday, 29 April 2022 Club E. Early colorectal cancer: diagnosis and treatment

PREDICTED ABSOLUTE RISK OF LYMPH NODE METASTASIS IN T1 COLORECTAL CANCER IN THE SOLE PRESENCE OF TUMOUR BUDDING, LYMPHOVASCULAR INVASION, OR POOR DIFFERENTIATION: A META-ANALYSIS

S.R.B. Lamme
1   University Medical Center Utrecht, Gastroenterology & Hepatology, Utrecht, Netherlands
,
L. van der Schee
1   University Medical Center Utrecht, Gastroenterology & Hepatology, Utrecht, Netherlands
,
K.M. Gijsbers
1   University Medical Center Utrecht, Gastroenterology & Hepatology, Utrecht, Netherlands
,
K.J.C. Haasnoot
1   University Medical Center Utrecht, Gastroenterology & Hepatology, Utrecht, Netherlands
,
P. Didden
1   University Medical Center Utrecht, Gastroenterology & Hepatology, Utrecht, Netherlands
,
M.M. Lacle
2   University Medical Center Utrecht, Pathology, Utrecht, Netherlands
,
S.G. Elias
3   University Medical Center Utrecht, Julius Centrum, Utrecht, Netherlands
,
L.M.G. Moons
1   University Medical Center Utrecht, Gastroenterology & Hepatology, Utrecht, Netherlands
› Author Affiliations
› Further Information
Also available at
 

Aims High-grade (Bd2/Bd3) tumour budding (TB), poor differentiation (PD) and lymphovascular invasion (LVI) are regarded as the strongest predictors for lymph node metastasis (LNM), and their presence considered an indication for completion surgery. However, these risk factors are strongly intercorrelated and their individual predictive strength is unclear. This study therefore aimed to investigate the absolute risk of LNM in the presence of only one of these risk factors.

Methods Studies were eligible for this meta-analysis if a multivariable analysis on the risk of LNM in T1CRC was performed with at least LVI, PD and TB as risk factors. The adjusted odds ratios were pooled in a random-effects model. To convert the pooled adjusted odds ratios (pORs) to absolute risks, a multivariable logistic regression model including the pORs but with an undefined intercept was fitted on a retrospective multicentre cohort of 628 Dutch T1CRC patients (12% LNM, 92% deep submucosal invasion, 93% non-pedunculated, 32% rectum). Predicted probabilities of LNM with their corresponding 95% confidence intervals (95%CI) were calculated in the presence of one risk factor.

Results A total of 14 studies (4628 patients) were included in the meta-analysis. LVI was the strongest predictor (pOR: 4.89 [95%CI: 2.89-8.27]), followed by PD (pOR: 3.10 [95%CI: 2.02-4.75]) and TB (pOR 2.34 [95%CI: 1.69-3.25]). The results of the fitted model (AUC 0.72 [95%CI: 0.66-0.77]) are summarized in [Table 1]

Absence of risk factors

TB+, PD-, LVI-

TB-, PD+, LVI-

TB-, PD-, LVI+

Predicted probability (95%CI)

3.1% (2.3% –3.9%)

6.9% (5.2% −8.6%)

8.9% (6.8% –11.1%)

13.4% (10.3%−16.5%)

Conclusions The absolute risk of LNM in the presence of a single histopathological risk factor varies between 6.9% and 13.4%, and is strongest for LVI.



Publication History

Article published online:
14 April 2022

© 2022. European Society of Gastrointestinal Endoscopy. All rights reserved.

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany