In total, 140 patients who met the inclusion criteria were included in this study. Among them, 76 patients were in the RCIT cohort and 64 patients in the CIT cohort. Before PSM, the proportion of patients with ≥ 65 years old, cervical and upper tumor, only non-regional lymph node metastasis, and only single metastatic organ in RCIT group was higher than that in CIT group. After PSM, 61 well-paired patients were selected in the oligometastasis cohort. There were no significant differences in baseline characteristics between two groups after PSM (Table 1). The status of patients’ distant metastasis is presented in Table 2.

As of the last follow-up on July 15, 2023, the median follow-up time was 34.7 months (95%CI: 30.6–38.8 months). Ninety-nine (70.7%) patients had died at the point of data cutoff.

Before PSM, PFS and OS were superior in the RCIT group compared to the CIT group. The median PFS was 10.9 months in the RCIT group (95%CI: 8.7–13.2 months) versus 6.8 months in the CIT group (95%CI: 5.8.–7.8 months; P < 0.001; Fig. 1a). OS time was significantly longer in the RCIT group (median, 19.4 months; 95%CI: 14.2–24.6 months) than in the CIT group (median, 12.6 months; 95%CI: 11.4–13.7 months; P = 0.008; Fig. 1b). The 12-, 24- and 36-month OS rate of the RCIT group were 76.3%, 44.5 and 32.5%, respectively, while the CIT group were 56.3%, 26.1 and 20.4%, respectively.

Progression-free survival (a) and overall survival (b) for all patients in RCIT and CIT group before PSM. Progression-free survival (c) and overall survival (d) for all patients in RCIT and CIT group after PSM

Before PSM, we also found that there were significant differences in PFS (P < 0.001) and OS (P = 0.007) among the three groups of patients: the group receiving RT for all lesions, the group receiving RT for partial lesions, and the CIT group (Fig. 2a and 2b). The group that received RT for all lesions had superior PFS and OS (median PFS: 13.5 vs. 8.6 months, P = 0.001; median OS: 29.4 vs. 16.1 months, P = 0.046) compared to the group that received RT for partial lesions. However, there was no significant difference in PFS and OS (median PFS: 8.6 vs. 6.8 months, P = 0.472; median OS: 16.1 vs. 12.6 months, P = 0.519) between the group receiving RT for partial lesions and the CIT group.

Progression-free survival (a) and overall survival (b) for patients with different RT sites before PSM. Progression-free survival (c) and overall survival (d) for patients with different RT sites after PSM

Forest plots show factors associated with progression-free survival (a) and overall survival (b) in entire cohort

After PSM, it also revealed significant differences between RCIT group and CIT group in PFS and OS. The median PFS of RCIT group and CIT group was 10.9 (95%CI: 9.4–12.4) months and 7.3 (95%CI: 6.0–8.7) months, respectively (P = 0.004) (Fig. 1c). The median OS of RCIT group and CIT group was 22.4 (95%CI: 17.5–27.4) months and 13.4 (95%CI: 10.9–15.9) months, respectively (P = 0.031) (Fig. 1d). The 12-, 24- and 36-month OS rate of the RCIT group were 77.0%, 44.1%, and 30.5%, respectively, while the CIT group were 62.3%, 27.4%, and 21.4%, respectively.

Post-PSM analysis showed a significant difference in PFS (P = 0.003) among the three groups mentioned above, while OS was on the threshold of significance (P = 0.052) (Fig. 2c and 2d). The group that received RT for all lesions has a better trend in PFS (P = 0.061) compared to the group that received RT for partial lesions. Although OS revealed no significant differences (P = 0.201), the group that received RT for all lesions demonstrated prolonged OS (median OS: 29.4 vs. 17.3 months). There was no significant difference in PFS and OS (median PFS: 8.6 vs. 7.3 months, P = 0.353; median OS: 17.3 vs. 13.4 months, P = 0.429) between the group receiving RT for partial lesions and the CIT group.

Then we analyzed the failure patterns of two groups. In the RCIT and CIT groups, 32.9% (20/61) and 39.3% (24/61) of patients only experienced locoregional failure, 44.3% (27/61) and 27.9% (17/61) of patients only experienced distant metastasis, 3.3% (2/61) and 14.8% (9/61) of patients experienced both, 6.6% (4/61) and 16.4% (10/61) of patients were unable to confirm the failure patterns due to death, respectively. Patients in the RCIT group have a lower probability of experiencing locoregional failure compared to the CIT group (36.1% vs 54.1%; χ2 = 4.006, P = 0.045). In addition, in the RCIT group, 41.0% (25/61) of patients only experienced infield failure, 34.4% (21/61) of patients only experienced outfield failure, 4.9% (3/61) of patients experienced both, and 6.6% (4/61) of patients were unable to confirm the failure patterns due to death, respectively.

We conducted subgroup analysis between RCIT and CIT groups. In the subgroup analysis of PFS, patients with male gender (HR:0.600, P = 0.018), age ≥ 65 years old (HR:0.434, P = 0.002), ECOG performance status of 0–1 (HR:0.589, P = 0.008), BMI ≥ 18.5 (HR:0.603, P = 0.016), clinical T stage of 2–3 (HR:0.544, P = 0.006), clinical N stage of 1–2 (HR:0.554, P = 0.007), extra-regional lymph node metastasis only (HR:0.556, P = 0.017), 1–2 metastatic lesions (HR:0.441, P = 0.003), and single metastatic organ (HR:0.607, P = 0.019) experienced significantly prolonged PFS in the RCIT group (Fig. 3a). In the subgroup analysis of OS, patients with male gender (HR:0.593, P = 0.034) and age ≥ 65 years old (HR:0.556, P = 0.042) exhibited significantly prolonged OS in the RCIT group (Fig. 3b).

The grade 3 or higher (G3+) TRAEs of the entire cohort are shown in Table 3. The most common G3+ TRAEs were hematological toxicities, including anemia, thrombocytopenia, leukopenia, and lymphopenia. Before PSM, there was no significant difference between the two groups in the incidence of G3+ TRAEs (27[35.5%] in the RCIT group vs 18[28.1%] in the CIT group, P = 0.350). However, the incidence of G3+ pneumonitis (11.8% vs 1.6%, P = 0.043) was higher in the RCIT group than in the CIT group. After PSM, there is still no significant difference in the incidence of G3+TRAE between the two groups (23[37.7%] in the RCIT group vs 17[27.9%] in the CIT group, P = 0.247). The major G3+ TRAEs of the RCIT group were hematological toxicity (n = 18, 29.5%) and pneumonitis (n = 8, 13.1%). There were statistically significantly higher rates of pneumonitis in the RCIT group than in the CIT group (13.1% vs 1.6%, P = 0.038). Detailed individual information of 9 patients with G3+ pneumonitis is shown in Supplement Table 1. Notably, one patient (1.8%) in the RCIT group experienced a grade 5 serious TRAE. The patient developed refractory pneumonitis two months after completing concurrent RT and two cycles of pembrolizumab. Despite receiving corticosteroid treatment and mechanical ventilation support, the patient died of pneumonitis two months later.