[681] Time-Dependent Analysis of the Lymph Node Ratio and Its Prognostic Impact in Advanced Colorectal Cancer Stratified by Mismatch Repair Status

Eva Karamitopoulou Diamantis, Alessandro Lugli, Inti Zlobec. University of Bern, Bern, Switzerland

Background: The aims of this study were (1) to evaluate the prognostic effect of the lymph node ratio (LNR: number of positive LN/number of LN harvested) in advanced colorectal cancer patients stratified by mismatch repair (MMR) status, (2) to compare the prognostic ability of this feature to the number of positive lymph nodes (LN) and (3) to identify a non-arbitrary cut-off score best discriminating high- and low-risk prognostic subgroups.
Design: The LNR was analyzed in 494 MMR-proficient and 86 MMR-deficient stage III-IV colorectal cancer patients. Cox regression and time-dependent receiver operating characteristic curve analysis (tROC) were carried out to analyze the prognostic abilities of each factor over time and for cut-point determination. Results were validated on a second cohort of 87 stage III patients.
Results: Relative risk of death with an elevated LNR was 6.91 (95%CI: 4.72-10.1) in MMR-proficient and 3.56 (95%CI: 1.1-11.53) in MMR-deficient cases. In multivariate analysis the LNR had a more unfavourable effect than the number of positive LNs (HR: 5.18 (95%CI: 3.5-7.6)) (p<0.001). The prognostic performance of the LNR was 10-15% greater than the number of positive LN regardless of MMR status. A threshold value of 0.231 was best for classifying patients into prognostic subgroups. This cut-off value was confirmed (p<0.001) using the validation cohort.


Conclusions: The LNR outperforms the number of positive LNs as a critical prognostic factor by 10-15% and is independent of MMR status. Patients with LNR >0.231 should be considered high-risk and may benefit from additional therapy. This cut-off score warrants validation in prospective studies.
Category: Gastrointestinal

Monday, March 19, 2012 8:30 AM

Platform Session: Section D, Monday Morning

 

Close Window