A total of 337 patients with resected mucosal HNSCC were enrolled, of which 88.7% (299/338) were men, 59.9% (202/338) were under 60 years old and 80.4% (271/388) were locally advanced cases. According to the postoperative pathological reports, 142 patients (41.2%) were staged as pT3-4, 136 patients (38.3%) were staged as pN2-3, 68 patients (21.1%) had PNI and 45 patients (12.8%) had LVI. The distribution of primary sites was as follows: oral cavity (42.4%, 143/337), larynx (38.6%, 130/337), hypopharynx (11.3%, 38/337) and oropharynx (8.0%, 27/337).

Initially, significant differences were observed in the N classification (p = 0.007) and overall stage (p = 0.002) between the two groups. After matching, 132 patients in the POCRT group were matched with 132 patients in the PORT group with balanced baseline characteristics (all p ≥ 0.05) (Table 1). In total, there were 93 patients suffering from local, regional and/or distant treatment failure, including 39 local recurrent cases, 38 regional recurrent cases and 47 metastatic cases. A total of 39.7% of the patients with failure received salvage treatments: 20 patients with surgery, 12 with chemotherapy, 4 with radiation and 1 with immunotherapy.

With a median follow-up duration of 52 months (range, 0–137 months), the 3- year and 5-year survival rates for the matched cohort were as follows: OS, 72.4% and 64.7%; LRFS, 79.3% and 75.1%; DMFS, 83.5% and 81.7%; and DFS, 62.5% and 56.8%, respectively.

Univariate survival analysis (Table 2) showed that patients with pT3-4 had worse DMFS than those with pT1-2 (p = 0.003, 3-year rate 76.6% vs. 88.1%) (Fig. 1). Patients with pN2-3 had worse survival outcomes than those with pN0-1 in OS (p < 0.001, 3-year rate 55.6% vs. 82.5%), DFS (p < 0.001, 3-year rate 44.9% vs. 73.9%), LRFS (p < 0.001, 3-year rate 64.2% vs. 88.3%) and DMFS (p < 0.001, 3-year rate 71.4% vs. 90.3%) (Fig. 2). LVI and PNI were not significantly associated with survival outcomes, including OS, DFS, LRFS or DMFS (all p > 0.05). Patients with stage III/IV disease had worse survival outcomes in OS (p = 0.001, 3-year rate 67.4% vs. 97.7%), DFS (p = 0.002, 3-year rate 57.8% vs. 86.2%), LRFS (p = 0.042, 3-year rate 77.4% vs. 88.2%) and DMFS (p = 0.002, 3-year rate 80.3% vs. 97.7%) when compared to patients with stage I/II disease.

The survival curves of pT1-2 versus pT3-4 in the matched cohort (N = 264)

The survival curves of pN0-1 versus pN2-3 in the matched cohort (N = 264)

Four adverse pathological features, including T classification, N classification, PNI and LVI were then included in the multivariate analysis. As a result, pN2-3 was still an independent risk factor for OS (p < 0.001, hazard ratio (HR) 2.310, 95% confidence interval (CI) 1.469–3.632), DFS (p < 0.001, HR 2.346, 95% CI 1.556–3.536), LRFS (p < 0.001, HR 3.527, 95% CI 1.925–6.461) and DMFS (p = 0.002, HR 2.820, 95% CI 1.484–5.357), whereas pT3-4 was an independent risk factor for DMFS (p = 0.005, HR 2.460, 95% CI 1.306–4.632).

In the matched cohort, patients receiving POCRT had no significant survival difference from those receiving PORT, with a 3-year OS rate of 76.4% vs. 67.6% (p = 0.062), 3-year DFS rate of 64.2% vs. 60.8% (p = 0.288), 3-year LRFS rate of 83.7% vs. 74.0% (p = 0.076) and 3-year DMFS rate of 82.1% vs. 85.0% (p = 0.692) in the univariate survival analysis (Fig. 3).

The survival curves of PORT versus POCRT in the matched cohort (N = 264)

The cohort was divided into different subgroups by four adverse pathological features, including T classification, N classification, PNI and LVI to explore the potential survival benefits of POCRT. First, there was no significant difference in the distribution of baseline characteristics between PORT and POCRT in the subgroup of patients with pT3-4, pN2-3, PNI or LVI, respectively (all p > 0.05). As shown in Fig. 4, patients with pN2-3 achieved better survival outcomes from POCRT than PORT in OS (p = 0.050, 3-year rate 63.9% vs. 47.9%) and LRFS (p = 0.019, 3-year rate 74.6% vs. 54.9%), whereas there was no significant survival improvement in DFS (p = 0.057, 3-year rate 52.3% vs. 38.2%) or DMFS (p = 0.349, 3-year rate 67.9% vs. 75.3%). Furthermore, POCRT is an independent favorable factor (p = 0.023, HR 0.463, 95% CI 0.238–0.901) in patients with pN2-3 for LRFS using multivariate survival analysis. In the subgroup of patients with pT3-4 (shown in Fig. 5), POCRT significantly improved LRFS (p = 0.014, 3-year rate 88.5% vs. 69.1%). Meanwhile, in multivariate survival analysis of patients with pT3-4, POCRT was still an independent favorable factor (p = 0.009, HR 0.310, 95% CI 0.128–0.747) for LRFS. But there were no significant differences between POCRT and PORT in OS (p = 0.442, 3-year rate 66.8% vs. 64.3%), DFS (p = 0.653, 3-year rate 54.2% vs. 57.7%) or DMFS (p = 0.648, 3-year rate 74.9% vs. 78.5%). POCRT gained similar survival outcomes in OS (p = 0.291, 3-year rate 74.1% vs. 65.0%), DFS (p = 0.689, 3-year rate 61.0% vs. 60.7%), LRFS (p = 0.514, 3-year rate 81.4% vs. 80.4%) and DMFS (p = 0.752, 3-year rate 81.3% vs. 83.5%) to PORT in patients with PNI. In the subgroup of patients with LVI, there were no significant differences in OS (p = 0.746, 3-year rate 75.5% vs. 71.4%), DFS (p = 0.846, 3-year rate 51.9% vs. 57.1%), LRFS (p = 0.783, 3-year rate 75.5% vs. 83.9%) or DMFS (p = 0.440, 3-year rate 82.1% vs. 69.2%) between POCRT and PORT.

The survival curves of PORT versus POCRT in the matched cohort (N = 104) with pN2-3

The survival curves of PORT vs. POCRT in the matched cohort (N = 106) with pT3-4