Value of number of negative lymph nodes in predicting the prognosis of patients with esophageal cancer after neoadjuvant therapy and the construction of nomogram prodiction model / 中华消化外科杂志

ABSTRACT

Objective:To investigate the value of number of negative lymph nodes (NLNs) in predicting the prognosis of patients with esophageal cancer after neoadjuvant therapy and the construction of nomogram prodiction model.Methods:The retrospective cohort study was conducted. The clinicopathological data of 1 924 patients with esophageal cancer after neoadjuvant therapy uploaded to the Surveillance, Epidemiology, and End Results Database of the National Cancer Institute from 2004 to 2015 were collected. There were 1 624 males and 300 females, aged 63 (range, 23?85)years. All 1 924 patients were randomly divided into the training dataset of 1 348 cases and the validation dataset of 576 cases with a ratio of 73 based on random number method in the R software (3.6.2 version). The training dataset was used to constructed the nomogram predic-tion model, and the validation dataset was used to validate the performance of the nomogrram prediction model. The optimal cutoff values of number of NLNs and number of examined lymph nodes (ELNs) were 8, 14 and 10, 14, respectively, determined by the X-tile software (3.6.1 version), and then data of NLNs and ELNs were converted into classification variables. Observation indicators (1) clinicopathological characteristics of patients in the training dataset and the validation dataset; (2) survival of patients in the training dataset and the validation dataset; (3) prognostic factors analysis of patients in the training dataset; (4) survival of patients in subgroup of the training dataset; (5) prognostic factors analysis in subgroup of the training dataset; (6) construction of nomogram prediction model and calibration curve. Measurement data with normal distribution were represented as Mean± SD, and comparison between groups was conducted using the t test. Measurement data with skewed distribution were represented as M(range), and comparison between groups was conducted using the Mann-Whitney U test. Count data were described as absolute numbers, and comparison between groups was conducted using the chi-square test. The Kaplan-Meier method was used to draw survival curve and Log-Rank test was used for survival analysis. The COX proportional hazard model was used for univariate and multivariate analyses. Based on the results of multivariate analysis, the nomogram prediction model was constructed. The prediction efficacy of nomogram prediction model was evaluated using the area under curve (AUC) of the receiver operating characteristic curve and the Harrell′s c index. Errors of the nomogram prediction model in predicting survival of patients for the training dataset and the validation dataset were evaluated using the calibration curve. Results:(1) Clinicopathological characteristics of patients in the training dataset and the validation dataset. There was no significant difference in clinicopatholo-gical characteristics between the 1 348 patients of the training dataset and the 576 patients of the validation dataset ( P>0.05). (2) Survival of patients in the training dataset and the validation dataset. All 1 924 patients were followed up for 50(range, 3?140)months, with 3-year and 5-year cumulative survival rate as 59.4% and 49.5%, respectively. The 3-year cumulative survival rate of patients with number of NLNs as <8, 8?14 and >14 in the training dataset was 46.7%, 62.0% and 66.0%, respectively, and the 5-year cumulative survival rate was 38.1%, 52.1% and 59.7%, respectively. There was a significant difference in the survival of these patients in the training dataset ( χ2=33.70, P<0.05). The 3-year cumulative survival rate of patients with number of NLNs as <8, 8?14 and >14 in the validation dataset was 51.1%, 54.9% and 71.2%, respectively, and the 5-year cumulative survival rate was 39.3%, 42.5% and 55.7%, respectively. There was a significant difference in the survival of these patients in the validation dataset ( χ2=14.49, P<0.05). The 3-year cumulative survival rate of patients with number of ELNs as <10, 10?14 and >14 in the training dataset was 53.9%, 60.0% and 62.7%, respectively, and the 5-year cumulative survival rate was 44.7%, 49.1% and 56.9%, respectively. There was a significant difference in the survival of these patients in the training dataset ( χ2=9.88, P<0.05). The 3-year cumulative survival rate of patients with number of ELNs as <10, 10?14 and >14 in the validation dataset was 56.2%, 47.9% and 69.3%, respectively, and the 5-year cumula-tive survival rate was 44.9%, 38.4% and 51.9%, respectively. There was a significant difference in the survival of these patients in the validation dataset ( χ2=9.30, P<0.05). (3) Prognostic factors analysis of patients in the training dataset. Results of multivariate analysis showed that gender, neoadjuvant pathological (yp) T staging, ypN staging (stage N1, stage N2, stage N3) and number of NLNs (8?14, >14) were independent influencing factors for the prognosis of patients with esophageal cancer after neoadjuvant therapy ( hazard ratio=0.65, 1.44, 1.96, 2.41, 4.12, 0.69, 0.56, 95% confidence interval as 0.49?0.87, 1.17?1.78, 1.59?2.42, 1.84?3.14, 2.89?5.88, 0.56?0.86, 0.45?0.70, P<0.05). (4) Survival of patients in subgroup of the training dataset. Of the patients with NLNs in the training dataset, the 3-year cumulative survival rate of patients with number of NLNs as <8, 8?14 and >14 was 61.1%, 71.6% and 76.8%, respectively, and the 5-year cumulative survival rate was 50.7%, 59.9% and 70.1%, respectively. There was a significant difference in the survival of these patients in the training dataset ( χ2=12.66, P<0.05). Of the patients with positive lymph nodes in the training dataset, the 3-year cumulative survival rate of patients with number of NLNs as <8, 8?14 and >14 was 26.1%, 42.9% and 44.7%, respectively, and the 5-year cumulative survival rate was 20.0%, 36.5% and 39.3%, respectively. There was a significant difference in the survival of these patients in the training dataset ( χ2=20.39, P<0.05). (5) Prognostic factors analysis in subgroup of the training dataset. Results of multivariate analysis in patients with NLNs in the training dataset showed that gender, ypT staging and number of NLNs (>14) were independent influencing factors for the prognosis of patients with esophageal cancer after neoadju-vant therapy ( hazard ratio=0.67, 1.44, 0.56, 95% confidence interval as 0.47?0.96, 1.09?1.90, 0.41?0.77, P<0.05). Results of multi-variate analysis in patients with positive lymph nodes in the training dataset showed that race as others, histological grade as G2, ypN staging as stage N3 and number of NLNs (8?14, >14) were independent influencing factors for the prognosis of patients with esophageal cancer after neoadjuvant therapy ( hazard ratio=2.73, 0.70, 2.08, 0.63, 0.59, 95% confidence interval as 1.43?5.21, 0.54?0.91, 1.44?3.02, 0.46?0.87, 0.44?0.78, P<0.05). (6) Construction of nomogram prediction model and calibration curve. Based on the multivariate analysis of prognosis in patients of the training dataset ,the nomogram prediction model for the prognosis of patients with esophageal cancer after neoadju-vant treatment was constructed based on the indicators of gender, ypT staging, ypN staging and number of NLNs. The AUC of nomogram prediction model in predicting the 3-, 5-year cumulative survival rate of patients in the training dataset and the validation dataset was 0.70, 0. 70 and 0.71, 0.71, respectively. The Harrell′s c index of nomogram prediction model of patients in the training dataset and the validation dataset was 0.66 and 0.63, respectively. Results of calibration curve showed that the predicted value of the nomogram prediction model of patients in the training dataset and the validation dataset was in good agreement with the actual observed value. Conclusion:The number of NLNs is an independent influencing factor for the prognosis of esophageal cancer patients after neoadjuvant therapy, and the nomogram prediction model based on number of NLNs can predict the prognosis of esophageal cancer patients after neoadjuvant therapy.