Aims: Considering morphological heterogeneity of lung adenocarcinoma (LUAD) and no objective prognostic grading system existing currently, we aim to establish an ‘optimised architecture-based grading system’ (OAGS) to predict prognosis for resected LUAD.
Methods: A multicentral study involving three independent cohorts of LUAD was conducted. Predictive ability of the OAGS for recurrence-free probability (RFP) and overall survival (OS) was assessed in training cohort (n=228) by the area under the receiver operating characteristic curve (AUC), Harrell’s concordance index (C-index) and Kaplan-Meier survival analyses, which was validated in testing (n=135) and validation (n=226) cohorts.
Results: The OAGS consists of: grade 1 for lepidic, papillary or acinar predominant tumour with no or less than 5% of high-grade patterns (cribriform, solid and or micropapillary), grade 2 for lepidic, papillary or acinar predominant tumour with 5% or more of high-grade patterns, and grade 3 for cribriform, solid or micropapillary predominant tumour. In all stages, the OAGS outperformed the pattern-dominant grading system and IASLC grading system for predicting RFP (C-index, 0.649; AUC, 0.742) and OS (C-index, 0.685; AUC, 0.754). Multivariate analysis identified it as an independent predictor of both (RFP, p<0.001; OS, p<0.001). Furthermore, in pT1-2aN0M0 subgroup, the OAGS maintained its ability to predict recurrence (C-index, 0.699; AUC, 0.769) and stratified patients into different risk groups of RFP (p<0.001). These results were confirmed in testing and validation cohorts.
Conclusions: The OAGS is an independent prognostic factor and shows a robust ability to predict prognosis for resected LUAD.
Keywords: biomarkers; diagnosis; hospital; lung neoplasms; pathology department; tumour.