The Role of Alternative Lymph Node Classification Systems in Gastroenteropancreatic Neuroendocrine Neoplasms (GEP-NEN): Superiority of a LODDS Scheme Over N Category in Pancreatic NEN (pNEN)

Abstract

Lymph node (LN) involvement in gastroenteropancreatic neuroendocrine neoplasms (GEP-NEN) has been reported to have prognostic and therapeutic implications. Numerous novel LN classifications exist; however, no comparison of their prognostic performance for GEP-NEN has been done yet. Using a nationwide cohort from the German Neuroendocrine Tumor (NET) Registry, the prognostic and discriminatory power of different LN ratio (LNR) and log odds of metastatic LN (LODDS) classifications were investigated using multivariate Cox regression and C-statistics in 671 patients with resected GEP-NEN. An increase in positive LN (pLN), LNR, and LODDS was associated with advanced tumor stages, distant metastases, and hormonal functionality. However, none of the alternative LN classifications studied showed discriminatory superiority in predicting prognosis over the currently used N category. Interestingly, in a subgroup analysis, one LODDS classification was identified that might be most appropriate for patients with pancreatic NEN (pNEN). On this basis, a nomogram was constructed to estimate the prognosis of pNEN patients after surgery. In conclusion, a more accurate classification of LN status may allow a more precise prediction of overall survival and provide the basis for individualized strategies for postoperative treatment and surveillance especially for patients with pNEN.

Fig. 1

Association between LN parameters and clinicopathologic variables: Violin plots depicting the association of LODDS ( a – g ), LNR ( h – n ), and pLN ( o – u ) with T category ( a , h , and o ) (Extent of tumor; T1+2; T3+4); presence of distant metastasis (M0 or M1) ( b , i , and p ); hormonal functionality (functional or afunctional) ( c , j , and q ); grading (G1, G2 , or G3) ( d , k , and r ); localization (foregut, midgut or hindgut) ( e , l , and s ); age (<median or≥median) ( f , m , and t ); and sex (female or male) ( g , n , and u ). **  p<0.01; ** **  p<0.0001.

Fig. 2

ROC analysis of the different lymph node classification systems: ROC curves were generated for LODDS, LNR, and pLN as continuous variables to predict ( a ) 1-year, ( b ) 3-year and ( c ) 5-year OS.

Fig. 3

Relationship between pLN, LNR and LODDS: Scatter plots presenting the distribution of LODDS versus pLN ( a ), LNR versus pLN ( b ), and LODDS versus LNR ( c ). **** p<0.0001.

Fig. 4

Nomogram for predicting the probability of patient survival: (a) : A nomogram composed of the independent variables age, sex, presence of distant metastases (M0 or M1), grading (G1, G2, or G3), extent of tumor (T1+2 or T3+4) and LODDS according to Persiani et al. predicting 1-, 3-, and 5-year OS. (b) : The final model was validated by bootstrap resampling (B=100 times) based on our data set and calibration curve assessment.