P-189
VALIDATION OF A PROGNOSTIC MODEL INCLUDING THE NUMBER OF HARVESTED LYMPH-NODES IN THE SETTING OF NON-SMALL-CELL LUNG CANCER PATIENTS UNDERGOING CURATIVE RESECTION: A MULTICENTRE ANALYSIS

Objectives: The aim of this study is to validate a prognostic score previously reported including the number of resected lymph-nodes (#RNs) in a large cohort of non-small cell lung cancer (NSCLC) patients.

Methods: From January 2002 to December 2012, data on 4858 NSCLC patients undergoing curative-intent surgery in 6 Institutions were retrospectively reviewed. The previous reported prognostic model, including #RNs and other clinico-pathological factors, was verified in our series. The model includes a panel of high-risk (HRFs: pathologic stage, nodal status, age, #RNs>10) and intermediate risk factors (IRFs: gender, grading, histology). Patients were divided as follow: high risk (HR): >1 HRFs, regardless of any IRFs; intermediate risk (IR): 1 HRF regardless of any IRF, or no HRFs and >1 IRFs; low risk (LR): ≤1 IRFs. To test the discriminative ability of the model, the Harrell C-statistic with 95% CI was determined. The Kaplan-Meier product-limit method was used to estimate overall (OS), cancer-specific (CSS) and disease-free survival (DFS) curves, and the log rank test was adopted to evaluate the differences between groups.

Results: Pathological stages were I in 46.5%, II in 24.1%, III in 27.8% and IV in 1.6% of the cases. Overall, 5-year OS, CSS and DFS were 54.6%, 76.7% and 44.8%, respectively. The prognostic model was validated in a dataset of 3948 patients having complete data. Stratifying the sample into low-risk (LR, #107), intermediate-risk (IR, #1268) and high-risk (HR, #2496) groups, the optimal prognostic discrimination power of this score was confirmed (C-statistics: 71. 95% CI 69–73). Specifically, in LR, IR and HR, 5-year OS were 83.5%, 66.4% and 46.2% (P < 0.0001), 5-year CSS were 95.8%, 89% and 69% (P < 0.0001), and 5-year DFS were 74.7%, 59.1% and 35.5% (P < 0.0001), respectively.

Conclusions: Our study confirms the optimal prognostic discrimination power of the previous prognostic model including the number of harvested nodes in NSCLC patients. Furthermore, this score may be used to better stratify OS, CSS and DFS and, eventually, can help physicians to tailor post-surgical treatment.

Disclosure: No significant relationships.