Demographical and clinicopathological characteristics
In table 1, we compared the demographical and clinicopathological characteristics of patients with different LODDS ranges. A total of 2059 patients in the SEER database were enrolled in this study and divided into three groups, patients with LODDS<-1.07, -1.07£LODDS<-0.27, and LODDS³-0.27. There was no difference between different groups in age, marital status, surgery type, and radiotherapy (all P>0.05). However, the variables of gender, race, laterality, primary site, histology type, differentiation, and chemotherapy were significantly different among the three groups (All P<0.05). The patients with LODDS³-0.27 had higher proportions of female, right laterality, the primary site of lower lobe, adenocarcinoma, low differentiation grade, low T1 stage, and chemotherapy. Since LODDS was calculated with the NDLN and the NPLN, the patients with LODDS³-0.27 had a higher N stage, more regional nodes examined and positive. We conducted the IPTW to eliminate the demographical and clinicopathological characteristics of patients with different LODDS ranges. As shown in Figure S1, the absolute standardized differences of variables were decreased under 0.2 and mostly under 0.1, indicating that the three groups were well matched after IPTW.
Univariate and multivariate Cox regression analysis
We conducted the univariate and multivariate Cox regression analysis to confirm the independent risk factors for the patients' survival, shown in Table 2 and Table 3. Before IPTW, the univariate analysis demonstrated that LODDS, age, gender, T stage, N stage, and radiotherapy were significantly associated with the OS of the patients (All P<0.05). However, the multivariate analysis showed that LODDS, age, gender, T stage, and radiotherapy were independent risk factors for patients' survival (All P<0.05), with N stage excluded.
After IPTW, the univariate analysis results were similar to the previous results, showing that LODDS, age, gender, T stage, N stage, and radiotherapy were variables with statistical significance, while the race, marital status, primary site, histologic type, differentiation, and surgery type were newly added variables (All P<0.05). Furthermore, the multivariate regression analysis indicated that LODDS, age, gender, race, marital status, primary site, differentiation, and T stage were independent risk factors for patients' survival (All P<0.05), with N stage excluded. Whether with or without IPTW, the LODDS was an independent risk factor for the prognosis of patients receiving neoadjuvant therapy followed by lung surgery.
We also conducted a subgroup analysis to validate the significance of LODDS further. By dividing the patients into different subgroups through the variable, we further compared the relative risk of different LODDS ranges. We found that the higher LODDS was associated with higher risks in most subgroups, as shown in Table 4. However, in subgroups of middle lobe, overlapping primary site, grade I differentiation, grade IV differentiation, and N3 stage, there was no statistical significance among different LODDS ranges, which could be caused by the relatively lower sample size.
Survival analysis
We compared the long-term survival of patients with different N classifications, shown in Figure 2A. Although the patients with different N stages presented different survival curves with a P value of 0.036, the curve was not separate and mostly overlapped. Nevertheless, when we divided the patients into three groups of LODDS ranges, we found that the curve was much more separate (Figure 2B). The patients with LODDS<-1.07 demonstrated the best survival status than the other two groups, while the middle group (-1.07£LODDS<-0.27) had better overall survival than the patients with LODDS³-0.27 (P<0.0001). Even after IPTW, the survival curve was still significant among three groups (P<0.0001), shown in Figure 3.
ROC analysis
In addition, we compared the accuracy and prognostic value of N classification, LODDS, and a multivariate model by ROC curve and AUC comparison. We used a multivariate model with five variables which had been shown to be independent prognostic indicatiors in multivariate analysis in Table 2: LODDS, age, gender, T stage, and radiotherapy. As shown in Figure 4, LODDS had significantly higher AUC than N classification in 1-year (P= 0.008), 3-year (P= 0.007), 5-year OS (P= 0.010), but not in 10-year OS (P=0.228). However, the multivariate model had significantly higher AUC than LODDS and N classification in 1-year, 3-year, 5-year, and 10-year OS (All P<0.001). We also compared the IDI and NRI of N classification, LODDS, and multivariate model, as shown in Table 5. When listing the LODDS as the reference, we found the IDI and NRI of N classification were negative. At the same time, those of the multivariate model was positive, suggesting that the LODDS had significantly higher predictive accuracy than N classification but was lower than the multivariate model (P<0.05).