Predictive Value Of A Prognostic Model Based On Pathologic Features In Lung Invasive Adenocarcinoma

Objective: The 2015 World Health Organization(WHO)classification introduced tumor spread through air spaces(STAS)officially as a new tumor invasion characteristic and presented it as a new exclusion criterion for minimally invasive adenocarcinoma.Tumor STAS is defined as tumor cells within the air spaces in the lung parenchyma beyond the edge of the main tumor and is composed of three morphological patterns: single cells,micropapillary structures,and solid nests or tumor islands.STAS was recently reported as a novel risk factor for the prognosis of patients with resected lung adenocarcinoma that indicates invasive tumor behavior.The purpose of this study was to build a prognostic model consisting of STAS and other pathologic features including visceral pleural invasion(VPI),vascular invasion(VI)and histological subtype(HS)in lung invasive adenocarcinoma.Methods: A total of 289 patients with resected lung invasive adenocarcinomas ?4 cm were analyzed retrospectively to evaluate the potential prognostic value of STAS,VPI,VI and HS for recurrence-free survival(RFS)and overall survival(OS).Furthermore,we also included a validation cohort that was composed of 91 patients.Hematoxylin and eosin(HE)-stained tumor slides of resected tumor specimens were microscopically reviewed and evaluated by two pathologists who were blinded to the clinical characteristics and survival outcomes of patients.The pathological features of the tumor were recorded,including tumor size,HS,VPI,VI,and STAS.Two pathologists diagnosed STAS when any of the morphological subtypes were present.In the event of disagreement,a consensus was reached after discussion.Results: STAS was observed in 143 patients(49.5%)in the study cohort and 50 patients(54.9%)in the validation cohort.There were significant correlations between the presence of STAS and several pathologic features,including larger tumor size(mean,2.471 cm versus 1.841 cm,p<0.001),higher pathologic T stage(p<0.001),higher pathologic N stage(p<0.001),higher pathologic stage(p<0.001),more invasive histological subtype(p<0.001),the presence of VPI(p<0.001),and the presence of VI(p<0.001).The univariate analysis showed that patients with tumor STAS had a worse RFS and OS than those without tumor STAS(hazard ratio(HR),9.401;95% confidence interval(CI),3.704-23.862;p<0.001)versus(HR,13.473;95% CI,4.143-43.817;p<0.001).Multivariate analysis showed that STAS,VPI,VI and HS were significant prognostic factors for poorer RFS and OS.Thus,a prognostic model including STAS,VPI,VI and HS was built using the results of the multivariate analysis.Nomograms were developed to predict the 5-year RFS and OS.The predictive accuracy for RFS and OS is shown by the calibration curves in the internal validation.The calibration plots for the predictive probability showed a great correlation between the nomogram-estimated prediction and the actual observed result.The discrimination and the prognostic accuracy of the prognostic model for the 5-year RFS and OS was calculated by the concordance index(C-index)in the internal validation,which was 0.8122 for predicting 5-year RFS and 0.8539 for predicting 5-year OS.Moreover,we also evaluated the predictive accuracy of the prognostic model in a validation cohort.The C-index of the prognostic model was 0.7928 for predicting 5-year RFS and 0.8249 for predicting 5-year OS.The C-index demonstrated that the model showed excellent discriminative ability in the external validation.Further,the nomograms of the prognostic model were externally validated by the calibration curves for 5-year RFS and OS,which showed that the nomograms were calibrated well.The C-index showed that the nomogram can discern a patient with the event from a patient without the event approximately 80% of the RFS time and approximately 84% of the OS time.In both the study cohort and validation cohort,the calibration curves of the nomograms showed a good correlation between the nomogram-estimated risk and the actual observed risk in the 5-year RFS and OS.Thus,the nomogram of the prognostic model was almost accurate for predicting the 5-year RFS and OS.Conclusion: STAS,VPI,VI and HS were significant prognostic factors for poorer RFS and OS.The prognostic model including STAS,VPI,VI and HS could effectively predict prognosis.