*3.4. Multivariate Cox Proportional Hazards Analysis*

Multivariate Cox proportional hazards analysis was conducted to measure the effect of reduced NME1 expression on RFS in early stage NSCLC, after adjusting for potential confounding effects of variables. Considering the pathological stage, hazard ratios for RFS ranged from 1.64 to 3.93, after adjusting for patient age, sex, β-catenin expression, adjuvant chemotherapy, and histology (Supplementary Table S3). The hazard ratio for RFS in a total of 425 patients was 2.27 (95% CI = 1.70–3.03, *p* < 0.0001) times worse in patients with reduced NME1 expression than in those without (Supplementary Table S4). However, the hazard ratio was not associated with age (HR = 1.01, 95% CI = 0.99–1.03, *p* = 0.18), sex (HR = 1.02, 95% CI = 0.71–1.47, *p* = 0.93), adjuvant chemotherapy (HR = 1.03, 95% CI = 0.86–2.23, *p* = 0.89), and β-catenin expression (HR = 0.99, 95% CI = 0.75–1.34, *p* = 0.99).

To test our hypothesis that the effect of NME1 expression on RFS may be affected by Wnt signal and cisplatin treatment, we stratified patients according to adjuvant chemotherapy and β-catenin expression. The negative effect of reduced NME1 expression on RFS was exacerbated by cisplatin-based adjuvant chemotherapy (adjusted hazard ratio = 3.26, 95% CI = 1.16–9.17, *p* = 0.03; Supplementary Table S5). For patients who did not receive cisplatin-based adjuvant chemotherapy, the hazard ratio of reduced NME1 expression was increased from 1.89 (95% confidence interval (CI) = 1.21–2.95, *p* = 0.005) to 2.54 (95% CI = 1.66–3.89, *p* < 0.0001; Table 1) by β-catenin overexpression after adjusting for patient age, sex, histology, and pathologic stage. Among patients with β-catenin overexpression who received cisplatin-based adjuvant chemotherapy, patients with reduced NME1 expression were determined to have 15.52 (95% CI = 2.94–82.38, *p* = 0.001; Table 1) times poorer RFS than those without. These results suggest that the negative effect of reduced NME1 expression on RFS may be worsen by cisplatin-based adjuvant chemotherapy and β-catenin overexpression in early stage NSCLC.


**Table 1.** Cox proportional hazards analysis <sup>a</sup> of RFS according to NME1 in early stage NSCLC stratified by cisplatin-based adjuvant chemotherapy and β-catenin overexpression

<sup>a</sup> Adjusted for age, sex, histology, and pathologic stage. Abbreviations: HR, hazard ratio; CI, confidence interval.
