Study on LASSO regression prediction of lymph nodes around hepatic artery metastasis based on blood routine index in patients with distant gastric cancer / 中国医师进修杂志

ABSTRACT

Objective:To investigate the value of LASSO regression prediction of lymph nodes around hepatic artery metastasis based on blood routine index in patients with distant gastric cancer.Methods:The clinical data of 110 patients with distant gastric cancer from June 2018 to February 2022 in Jieshou People′s Hospital were retrospective analyzed. Among them, 43 patients had lymph nodes around hepatic artery metastases (metastasis group), and 67 patients have not lymph nodes around hepatic artery metastases (non-metastasis group). The basic clinical data were recorded; the routine blood test was detected, the indexes including white blood cell count, neutrophil percentage, lymphocyte count, platelet count, lymphocyte percentage, acidophil count, basophils count, hemoglobin, red blood cell distribution width (RDW), platelet distribution width (PDW) and neutrophil to lymphocyte ratio (NLR). The R language 4.1.0 software "grpreg" package was used to establish a Group LASSO Logistic regression analysis model to finally select the factors predicting lymph node around hepatic arterial metastasis in patients with distal gastric cancer. Nomogram were made using R language 3.5.3 software package and rms program package, calculated the consistency index (C-index), and the accuracy of the model was evaluated by the area under the curve (AUC) of the receiver operating characteristic (ROC) curve.Results:The white blood cell count, neutrophil percentage, NLR and lymphocyte percentage in metastasis group were significantly higher than those in non-metastasis group (12.16 ± 1.62) × 10 9/L vs. (9.38 ± 2.64) × 10 9/L, (73.36 ± 6.94)% vs. (52.21 ± 6.23)%, 3.23 ± 0.63 vs. 2.35 ± 0.13 and (48.62 ± 3.64)% vs. (31.02 ± 2.94)%, the acidophil count and basophils count were significantly lower than those in non-metastasis group (0.31 ± 0.03) × 10 9/L vs. (0.36 ± 0.04) × 10 9/L and (0.08 ± 0.01) × 10 9/L vs. (0.09 ± 0.02) × 10 9/L, the degree of differentiation and TNM stage were also worse than those in non-metastasis group, and there were statistical differences ( P<0.01). The nomogram was constructed to predict lymph nodes around hepatic artery metastasis based on the degree of differentiation, TNM stage, white blood cell count, neutrophil percentage, NLR, lymphocyte percentage, acidophil count and basophils count in patients with distant gastric cancer, the scores of each indexes were 20.64, 26.42, 32.83, 25.78, 38.46, 35.65, 29.06 and 18.26 scores, the total score (227.10 scores) was the incidence of the nomogram model (29.82%). The validation result showed C-index of 0.819 and 0.806 (95% CI 0.785 to 0.864 and 0.779 to 0.816), and the correction curves for both sets were fitted well to the ideal curve with AUC of 0.801 and 0.810 (95% CI 0.784 to 0.826 and 0.795 to 0.852), and the decision curve showed high net benefit value with threshold probability from 1% to 9%. Conclusions:LASSO regression model combined with white blood cell count, neutrophil percentage, NLR, lymphocyte percentage, acidophil count and basophils count is ideal to predict lymph nodes around hepatic artery metastasis in patients with distant gastric cancer.