Patient characteristics
Between January 1, 2003 and December 31, 2012, there were 1057 new patients diagnosed with clinical stages III and IV HPSCC in four individual hospitals. Patients who received first treatment more than 90 days after diagnosis (n=221), underwent radiotherapy alone (n=21), or received chemotherapy alone (n=163) were excluded from the study (Fig. 1). Thus, a total of 652 patients were enrolled in this study and followed up until the end of 2015 or were censored due to death. During the study period, 213 patients (32.7%) underwent PS, and 439 patients (67.3%) underwent definitive CCRT as the initial treatment modality. In the PS group, 151 patients (70.8%) received postoperative adjuvant therapy, while salvage surgery was performed in 168 patients (38.3%) of the CCRT group. The characteristics of all subjects are demonstrated in Table 1.
Survival analysis in the advanced-stage HPSCC
The median follow-up duration of all subjects was 30.6 months, and the 5-year OS and DFS were 37.5% and 31.3%, respectively, for all patients. Clinical stage was a significant predictor of prognosis, as shown in Fig. 2. Kaplan-Meier survival analysis indicated that the 5-year OS rates of patients with stages III, IVA, and IVB were 54.3%, 39.1%, 19.4%, respectively, and the 5-year DFS rates of stages III, IVA, and IVB were 48.4%, 33.4%, and 13.3%, respectively (both log-rank p < 0.001; Fig 2A and 2B).
The median survival time was 50.6 months (95% confidence interval 37.9 - 66.3 months) in patients treated with PS and 24.3 months (95% confidence interval 18.4 - 28.7 months) in those who underwent definitive CCRT. The Kaplan-Meier estimate of the 5-year OS rates for patients undergoing PS and definitive CCRT were 45.0% and 33.1% (p < 0.001, Fig. 2C), while the 5-year DFS rates for patients undergoing PS and definitive CCRT were 36.2% and 28.9% (p = 0.003; Fig. 2D), respectively. These results indicated that the long-term prognosis of the PS group was superior to that of the definitive CCRT group.
We further analyzed the survival outcomes between PS and CCRT by clinical stage. In patients with stage III HPSCC, the 5-year OS and DFS rates in the PS and CCRT groups were similar but without statistical significance (Figs. 3A and 3B). As shown in Fig. 3C and 3D, for patients with clinical stage IVA HPSCC, the PS group (vs. CCRT group) showed significantly better 5-year OS rates (46.7% vs. 35.0%, respectively, p = 0.002) and 5-year DFS rates (38.1% vs. 31.0%, respectively, p = 0.041). For patients with stage IVB HPSCC, the 5-year OS rates were 22.7% and 19.3% in the PS and CCRT groups, respectively (p = 0.235, Fig. 3E), and the 5-year DFS rates were 15.1% and 9.1% in the PS and CCRT groups, respectively (p = 0.696, Fig. 3F). Although the PS seems to have better 5-year OS and DFS rates in stage IVB patients, these differences were not statistically significant.
Subgroup survival analysis of stage IVA HPSCC by treatment modality
To investigate which subgroup of patients with stage IVA HPSCC would benefit the most from PS treatment, we conducted subgroup analysis based on patient age (≥ 65 or <65 years old), T and N classification, and treatment methods. As shown in Fig. 4A and 4B, PS provided significantly better 5-year OS rates than CCRT treatment in patients both less than (44.4% vs. 34.9%, respectively, p = 0.018, Fig. 4A) and more than 65 years old (64.2% vs. 35.5%, respectively, p = 0.022, Fig. 4B). Notably, the difference between treatment methods is even more prominent in those more than 65 years old. For stage IVA patients stratified by early (T1-T3) and advanced (T4) T classification, the PS group showed better 5-year OS rates than those of the CCRT group, with marginal significance (51.8% vs. 39.4%, respectively, p = 0.052, fig. 4C) in those with T1-T3 disease. The PS group also had better 5-year OS rates than those of the CCRT group in the patients with advanced T4 classification (44.1% vs. 33.1%, respectively, p = 0.021, Fig. 4D). Similarly, when we stratified stage IV patients by early (N0-1) and advanced (N2) nodal spread, the PS group showed significantly better 5-year OS rates for patients with N2 classification (47.2% vs. 33.2% for the CCRT group, p = 0.002, Fig. 4F), but there was no significant difference with regard to OS in the patients with N0-N1 classification (44.5% for the PS group vs. 41.2% for the CCRT group, p = 0.551, fig. 4E).
In the subgroup analysis of stage IVA patients with regard to 5-year DFS, the PS group showed significantly better DFS rates for patients with N2 classification (38.8% vs. 28.1% for the CCRT group, p = 0.018, Fig. 5F), but there was no significant difference between the two groups (PS vs. CCRT) for patients less than 65 years old (37.0% vs. 29.9%, p = 0.077, Fig. 5A) or more than 65 years old (46.3% vs. 36.4%, p = 0.253, Fig. 5B) or with T1-T3 (42.2% vs. 31.0%, p = 0.182, Fig. 5C), T4 (36.0% vs. 31.2%, p = 0.113, Fig. 5D), or N0-N1 classification (40.5% vs. 34.7%, p = 0.99, Fig. 5E).
Multivariate analysis of survival outcomes
Multivariate analysis revealed that clinical stage as well as initial treatment modality (CCRT vs. PS) were independent predictors of 5-year OS and that advanced stages were also predictors for worse 5-year DFS (table 2). For patients with advanced HPSCC, initial treatment with PS appeared to be associated with a better prognosis for 5-year OS (hazard ratio [HR]: 0.73; 95% CI: 0.57 - 0.94, p = 0.014) and 5-year DFS (HR: 0.83; 95% CI: 0.66 - 1.04, p = 0.105, table 2).
In subgroup analysis of patients with stage IVA HPSCC, the multivariate Cox regression analysis showed that PS was associated with better 5-year OS (HR: 0.69; 95% CI, 0.51-0.93, p = 0.014, table 3) and DFS (HR: 0.78; 95% CI, 0.59-1.03, p = 0.079, table 3), although the difference was not statistically significant for DFS. The analysis also demonstrated that PS was profoundly associated with a 67% lower risk of death in stage IVA patients older than 65 years of age (HR: 0.33; 95% CI: 0.12 - 0.88, p = 0.027, Table 3). PS treatment seemed to show better outcomes in terms of OS and DFS compared with CCRT for stage IVA patients with early (T1-T3) and advanced (T4) T classification. However, the difference was not statistically significant except for a marginal significance shown for OS in patients with T4 classification (HR: 0.70; 95% CI: 0.48 – 1.00, p = 0.053 table 3). For patients with stage IVA HPSCC and N0-N1 classification, the difference in prognosis in terms of OS and DFS between the two groups were not statistically significant (HR: 0.92; 95% CI: 0.46 – 1.86, p = 0.816 for OS; HR: 1.11; 95% CI: 0.51 – 2.15, p = 0.762 for DFS, table 3). On the other hand, for stage IVA HPSCC patients with N2 classification, PS was associated with significantly lower risks of death in terms of OS (HR: 0.64; 95% CI: 0.46 – 0.90, p = 0.01, table 3) and DFS (HR: 0.71; 95% CI: 0.52 – 0.97, p = 0.033) compared to CCRT.