Search results

A total of 4295 studies were found in our initial search. There were 2889 studies left after excluding 1406 repetitive studies. 2311 studies were excluded because they were considered not to meet the inclusion criteria after screening the titles and abstracts. 508 reviews and 51 other unrelated articles were eliminated by reading the full text. Eventually, this meta-analysis included 19 studies involving 851 patients with LA-NSCLC. We outline the study selection process for review using the PRISMA flowchart (Fig. 1).

Fig. 1

Flow chart of study selection

Study characteristics

Eight of the included studies were phase I / II clinical trials, and the rest were almost all case-series studies, except for one case-control study comparing the efficacy of proton therapy and photon therapy, in which we extracted the data as reported on proton therapy. There were three studies with 119 patients receiving PBT [31,32,33], 11 studies with 452 patients receiving CCRT [24, 34,35,36,37,38,39,40,41,42,43] (concurrent chemoradiotherapy, both PBT and concurrent chemotherapy), and four studies with 235 patients receiving CIRT [25, 26, 44, 45]. In another study, most patients received PBT, and others received CIRT [46]. The 2-year OS rate, PFS rate, and LC rate were reported by 15 studies, 13 studies, and 11 studies, respectively. Regarding the 5-year survival data, OS rates and PFS rates were reported by five studies. According to the MINORS criteria, the median quality score for case series studies was 11 (range 10–12), and the only control study was 18. The complete quality assessment is available in Supplementary Table 1. The characteristics of 19 included studies are summarized in Table 1.

Table 1 Study and population characteristics for included studies Efficacy

There were 15 studies reported 2-year OS, which included 563 patients. The pooled 2-year OS rate of LA-NSCLC treated by particle therapy was 61.3% (95%CI = 54.7-68.7%), with high inter-study heterogeneity (I2 = 68.2%, p = 0.00) (Fig. 2a). Based on the treatment type subgroup, the 2-year OS of CCRT group was the highest (67.3%, 95%CI = 58.0-78.0%, I2 = 69.7%, p = 0.002), the 2-year OS of PBT group was the lowest (49.1%, 95%CI = 39.3-61.4%, I2 = 11.1%, p = 0.325), the 2-year OS of CIRT group was 57.8% (95%CI = 51.8-64.4%) and the heterogeneity level was low (I2 = 0.0%, p = 0.644) (Fig. 2b). The pooled 2-year OS rate of stage II-III group (61%, 95%CI = 56-66%) was the same with that of the stage II-III group (61%, 95%CI = 51-73%) (Supplementary Fig. S1a).

Fig. 2

Meta-analysis of the 2-year overall survival rate (OS): (a) 2-year OS, overall; (b) 2-year OS, subgroup analysis stratified by treatment type. Abbreviations: PBT, proton beam therapy; CIRT, carbon-ion radiotherapy; CC, concurrent chemotherapy

The pooled 2-year PFS of the 13 included studies included 518 patients was 37.9% (95%CI = 33.8-42.6%), with low inter-study heterogeneity (I2 = 0.0%, p = 0.591) (Fig. 3a). The CCRT group showed the highest pooled 2-year PFS of 40.8% (95%CI = 34.0-48.8%), followed by the CIRT group for 38.7% (95%CI = 32.9-45.6%) and finally the PBT group for 24.6% (95%CI = 16.8-36.0%) (Fig. 3b). The pooled 2-year PFS was 40.2% (95%CI = 34.4-46.8%) for the stage II-III group and 35.3% (95%CI = 29.7-42.0%) for the stage III group, with no statistical difference between the two subgroups (P = 0.278) (Supplementary Fig. S1b).

Fig. 3

Meta-analysis of the 2-year progression-free survival (PFS): (a) 2-year PFS, overall; (b) 2-year PFS, subgroup analysis stratified by treatment type. Abbreviations: PBT, proton beam therapy; CIRT, carbon-ion radiotherapy; CC, concurrent chemotherapy

Data on 2-year LC rate to evaluate the effect of particle therapy were available in 11 studies, which included 505 patients, with a pooled value of 82.2% (95%CI = 78.7-85.9%, I2 = 15.2%, p = 0.299) (Supplementary Fig. S2a). 2-year LC in the PBT group, CIRT group, and CCRT group were 61.9%, 79.1%, and 85.0%, respectively (P = 0.013) (Supplementary Fig. S2b). 2-year LC of the stage II-III group was similar to that of the stage III group (81.6% VS 82.7%, P = 0.771) (Supplementary Fig. S2c).

Only five studies reported the OS and PFS data of patients at 5 years after particle therapy, two of which did not report the 95% confidence interval of survival data. Due to limited data, multiple imputation is not feasible, so we combined three studies that reported complete data in which all patients were stage III and received CCRT treatment. The pooled 5-year OS and PFS of patients treated with CCRT were 41.3% (95%CI = 27.1-63.1%) and 25.3% (95%CI = 16.3-39.4%), respectively (Fig. 4a-b). Three studies reported the data of the 5-year LC rate. For the above reasons, we integrated two studies with complete data in which all patients were stage III and received CCRT treatment. The pooled 5-year LC of patients treated with CCRT was 61.5% (95%CI = 50.7-74.6%) (Fig. 4c).

Fig. 4

Meta-analysis of the overall survival rate (OS) (a), progression-free survival (PFS) (b), and local control (LC) (c) at 5 year

Safety

The present study evaluated the occurrence rates of common adverse events, such as esophagitis, pneumonitis, and dermatitis, in LA-NSCLC patients who underwent particle therapy. All adverse events (both acute toxicity and late toxicity) were classified as grade 2 or grade 3/4.

The incidences of grade 2 and grade 3/4 dermatitis were 19.8% (95%CI = 14.8-25.3%) and 2.6% (95%CI = 0.5-5.7%), respectively (Supplementary Fig. S3a-b). The subgroup analysis of treatment type showed that the incidences of grade 2/3/4 dermatitis in the CCRT group were higher than in the PBT group and CIRT group, similar to that of esophagitis. The incidences of grade 2 dermatitis in PBT, CIRT, and CCRT groups were 15.5%, 12.2%, and 26.0%, respectively, and the incidences of grade 3/4 dermatitis were 4.2%, 0.0%, and 5.1% respectively (Supplementary Fig. S4a-b). Subgroup analysis stratified by stage showed that there was no difference in the incidences of dermatitis (grade 2/3/4) among groups (Supplementary Fig. S5a-b).

Regarding grade 2 pneumonitis, the incidence was 13.1%, and the PBT and CCRT groups had similar incidences (14.6% vs. 14.7%), both higher than that in the CIRT group (6.5%) (Supplementary Fig. S6a-b). The incidences of grade 2 pneumonitis did not vary significantly among subgroups stratified by stage (Supplementary Fig. S6c). The incidence of grade 3/4 pneumonitis was low, at 3.4% (95%CI = 1.4-6.0%), and the incidences were similar among the subgroups stratified by treatment type or stage (Fig. 5a and Supplementary Fig. S7a-b).

Fig. 5

Forest plots of the incidences of grade 3/4 pneumonitis (a) and oesophagitis (b)

The incidences of grade 2 and grade 3/4 oesophagitis were 23.2% (95%CI = 13.7-34.2%), 2.6% (95%CI = 0.4-6.0%), respectively (Supplementary Fig. S8a and Fig. 5b). The results of treatment type subgroup analysis showed that the incidences of grade 2/3/4 esophagitis was found to be significantly lower in patients treated with PBT or CIRT as compared to those who received CCRT. The incidences of grade 2 esophagitis in PBT, CIRT, and CCRT groups were 6.0%, 2.3%, and 33.1%, respectively (Supplementary Fig. S8b). The incidences of grade 3/4 esophagitis in PBT, CIRT, and CCRT groups were 0.4%, 0.0%, and 5.5%, respectively (Supplementary Fig. S9). There was no difference in the incidences of esophagitis (grade 2/3/4) between the two subgroups stratified by stage (Supplementary Fig. S10a-b).

Treatment‑related death

Two studies reported treatment-related deaths [32, 43]. Two patients who received PBT without concurrent chemotherapy and one who received PBT with concurrent chemotherapy encountered grade 5 adverse events. One patient who had preexisting severe inflammatory pneumonia before the diagnosis of NSCLC died of pneumonia during treatment. Another patient, who had undergone repeated biopsies of the irradiated bronchus, died of hemoptysis after treatment. In addition, One patient with tumor invasion of the pulmonary artery died of bronchopulmonary hemorrhage after treatment.

Sensitivity analysis

In the analysis of complete cases, the confidence intervals were found to overlap with those of the primary analysis, and the pooled estimate remained in close proximity to the overall estimate. The complete-case analysis shows the pooled 2-year OS, PFS, and LC rates were 58.9% (95% CI = 50.6-68.6%), 38.6% (95% CI = 33.4-44.7%), and 76.2% (95% CI = 68.6-84.6%), respectively (Supplementary Fig. S11-13). Similar outcomes were obtained when each cohort was excluded one at a time (Supplementary Fig. S14-16). The results of the sensitivity analysis indicated that the pooled estimate was reliable.