Abstract 5878
Background
GEP NEN are classified according to morphology and proliferation index. According to the WHO 2010 classification, grade 3 (G3) GEP NEN are defined as having a Ki67 >20%. However, G3 well-differentiated NETs (WD-NETs) and poorly-differentiated neuroendocrine carcinomas (PD-NECs) may overlap in their proliferation index leading to prognostic and therapeutic uncertainties. Recently, WHO 2017, defined a new subgroup of G3 pancreatic NETs (PNET); G3a for WD-NETs and G3b for PD-NECs.
Methods
We retrospectively identified patients with G3 GEP NEN and divided them into WD-NETs and PD-NECs according to histological reports. The relationship between baseline characteristics and OS was analysed using the Kaplan Meier logrank test. Univariate and multivariate analyses were performed using the Cox proportional hazards model. Optimal Ki67 cut-points were explored using R version 2.15.0.
Results
145 patients with G3 GEP NENs had median follow-up 17 mo (59% male, 50% PD-NEC, 37% PNEN). There was a trend in improved survival in the WD-NET cohort (median Ki67 30%) compared to PD-NECs (median Ki67 60%); 29 vs 22 months (p = 0.1) more marked in the non-PNET cohort (median OS WD-NET 44 mo vs PD-NEC 20 mo; p = 0.1). Of the entire cohort, an independent effect of Ki67 was demonstrated on risk of mortality (p = 0.02). A Ki67 cutoff of 50% was found to have the highest logrank statistic. Ki67 ≥50% was associated with poorer OS compared to lower Ki67 index (median OS 38 mo vs 20 mo; p = 0.005). Somatostatin receptor imaging (SRI) was positive in 81% of WD-NETs compared to 45% of PD-NECs.
Conclusions
Our findings suggest that morphology and proliferative index are important factors in the prognostic evaluation of G3 GEP NEN of pancreatic and non-pancreatic origin. Ki67 remains the most reliable prognostic factor and further work is required to refine optimal Ki67 cutoffs in the G3 GEP NEN cohort to guide prognosis and treatment.
Clinical trial identification
Legal entity responsible for the study
Aimee Hayes.
Funding
Has not received any funding.
Editorial Acknowledgement
Not applicable
Disclosure
C. Thirlwell: Advisory board, Speaker honoraria: Ipsen. C. Toumpanakis: Speaker honoraria: Ipsen, Novartis and AAA. M. Caplin: Advisory Board, Speaker honoraria: Novartis, Ipsen, AAA, Lexicon. All other authors have declared no conflicts of interest.