Abstract 51P
Background
Lung NET are a heterogeneous group of tumors, with variegated clinical presentation and course. The histological subtype of typical carcinoids (TC) versus atypical carcinoids (AC), and TNM stage are well-established prognostic factors. In a previous work by our group, we have developed a nomogram that integrates relevant prognostic factors to predict lung NET outcomes. Our model, named Rachel score, demonstrated to a reliable and useful tool to predict survival rates in this population.
Methods
We selected a lung NET patients’ external cohort from Bologna University Hospital to evaluate the applicability and eventually validate the previously built nomogram and Rachel score. This external cohort was composed by 121 lung NET: median age was 59 years, 41 were male (33.88%), 43 were left-sided tumors (35.54%), 20 patients had positive nodal status (16.53%), 54 were atypical carcinoids (44.63%), 20 cases presented necrosis (16.53%), median Ki-67 was 7. Median OS was 89 months and median PFS was 72 months. Of note, Ki67 value was missing in 31 cases (25.62%).
Results
Overall survival (OS) multivariable models confirmed a prognostic role for male sex (p=0.071 in the model with Ki67, p=0.027 in the model without Ki67) and nodal status (p=0.054 in the model with Ki67). In progression free survival (PFS) survival model, AC was a negative prognostic factor (p= 0.001 in the model without KI67). Ki67 was associated both with OS and PFS (P= 0.086 and p=0.003). By applying Rachel score to the validation cohort, we confirmed three prognostic groups (specifically, high, medium and low risk), supporting the validity of this tool for prognostic stratification but only in the model without Ki67 (OS model p = 0.0028 and PFS model p < 0.0001). Unfortunately, in survival models including Ki67, the curves show a lower ability to discriminate between risk groups. This difference is probably related to the high percentage of cases lacking Ki67 value.
Conclusions
OS and PFS Kaplan-Meier curves in the validation cohort confirmed significant differences among the three groups identified by Rachel score by applying the model without Ki67 value, reinforcing a potential value of this tool for patients’ stratification.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.