3.1 Patient characteristics
A total of 156 patients with recurrent NPC were treated at our center from January 2015 to December 2018 and 69 cases were excluded, including 19 cases with single lymph node regional recurrence, 17 cases with distant metastases, 1 case with other malignant tumors, 9 cases who did not complete treatment, 11 cases who did not receive IMRT treatment (including 9 cases treated with surgery, 1 case treated with brachytherapy and 1 case with palliative chemotherapy) and 12 cases who were lost to follow-up following treatment (Figure 1). The exclusion of these patients was performed to ensure the accuracy and authenticity of the data.
The remaining 87 patients with LR-NPC were included in the analysis, of whom 69 (76.7%) were men and 21 (23.3%) were women. A total of 32 (36.8%) patients received RT combined with N and were included in the N group, whereas 55 (63.2%) received CRT and were included in the CRT group. In the CRT group, 30 cases (54.5%) received induction chemotherapy, whereas 8 cases (14.5%) received concurrent chemotherapy and 17 cases (30.9%) both induction chemotherapy and concurrent chemotherapy. A total of 4 cases (all in CRT group) were treated with a re-irradiation dose of 50 Gy/25 fraction due to poor RT tolerance. The analysis of the clinical baseline characteristics of the patients in each group demonstrated no significant differences by gender, age, histology, recurrent T classification, recurrent N classification and recurrent clinical stage between the N and the CRT groups (P>0.05). The baseline characteristics of each group are detailed in Table 1.
3.2 Efficacy
All 87 patients completed radiation therapy, with a median follow-up time of 47.8 months (3.7-72.8 months) and a median survival time of 36.2 months for the entire group. The 4-year OS of the N and the CRT groups were 37.1% and 40.7% (P=0.735), respectively, whereas the 4-year PFS of the N and the CRT groups were 20.0% and 28.6%, respectively (P=0.713). The 4-year LRFFS of the N and the CRT groups were 50.3% and 45.3%, respectively (P=0.375), whereas the 4-year DMFS of both groups (N and CRT) was 80.0% (P=0.813). No significant differences were noted between the two groups in terms of OS, PFS, LRFFS and DMFS. The survival curves are shown in Figure 2.
3.3 Patterns of failure and causes of death
A total of 12 cases (13.7%) out of the total number of patients exhibited distant metastasis from the end of RT to the cut-off time point of the follow-up period. The median time period to metastasis was 12 months (range, 8.1-35.3 months). A total of 5 cases (15.6%) with metastases were noted in the N group, including 1 case of bone metastasis, 2 cases of lung metastasis, 1 case of liver metastasis and 1 case of distal lymph node metastasis. A total of 7 cases with metastases (12.7%) were noted in the CRT group, including 3 cases of bone metastasis, 2 cases of lung metastasis and 2 cases of distal lymph node metastasis. A second local or regional recurrence occurred in 29 cases (33.3%), with a median recurrence time of 18.7 months (range 5.2-42 months), of which 10 cases (31.3%) were noted in the N group and 19 cases (34.5%) in the CRT group. At the time of the analysis, 49 (56.3%) deaths occurred, of which 21 (65.6%) were noted in the N group and 28 (50.9%) in the CRT group. The most common cause of death was radiation therapy-associated complications, followed by tumor progression. A total of 26 cases (29.9%) did not survive due to RT-associated complications, of which 8 cases (25.0%) were from the N group and 18 cases (32.7%) were from the CRT group. A total of 15 cases (17.2%) did not survive due to tumor progression, of which 8 cases (25.0%) were from the N group and 7 cases (12.7%) were from the CRT group. The failure patterns and causes of death are shown in Table 2 and no significant differences were observed between the two groups with regard to failure patterns and causes of death (P>0.05).
3.4 Toxicity
All patients completed the established RT schedule. Mild to moderate acute toxic reactions, including mucositis and xerostomia, were common during RT and the majority of the acute toxicities were considered as tolerable. During treatment, mild (grade 0-1) hematological toxic reactions were reported in group N. The proportion and degree of hepatotoxicity, weight loss, dermatitis, nausea and oral mucositis were significantly lower in the N group than those noted in the CRT group (Table 3).
A total of 48 cases (55.2%) exhibited grade ≥3 late radiation injuries, of which 12 cases (37.5%) were reported in the N group and 36 cases (65.5%) in the CRT group. A total of 34 cases (39.1%) exhibited nasopharyngeal necrosis, of which 18 cases (20.7%) presented with necrosis of the internal carotid artery or other vessels, eventually leading to nasopharyngeal hemorrhage. A total of 19 cases (21.8%) presented with temporal lobe necrosis, whereas 16 cases (18.4%) exhibited cranial nerve injury, including 3 cases of facial nerve injury, 4 cases of optic nerve injury, 1 case of actinic nerve injury, 3 cases of auditory nerve injury, 7 cases of vagus nerve injury, 3 cases of glossopharyngeal nerve injury, 1 case of olfactory nerve injury and 2 cases of trigeminal nerve injury (> 1 type of cranial nerve injury may have occurred in each patient). A total of 7 cases (8.0%) presented with trismus (≤1 cm). Although no significant differences were noted in all complications between the two groups with the exception of trismus (P>0.05), the overall incidence of late radiation injuries was significantly lower in the N group than that noted in the CRT group (P=0.011) (Table 4).
3.5 Prognosis
Univariate analysis indicated that recurrent T classification, tumor response following RT and GTV-T volume were associated with OS. GTV-T volume was also associated with PFS, whereas plasma EBV DNA following RT was associated with DMFS. However, these factors were not significantly correlated with LRFFS (Table 5).
Multivariate analysis indicated that age ≥50 years and rT3-4 stage were independent prognostic factors affecting OS (HR=2.369, P=0.005; HR=4.875, P=0.003). OS was lower for patients aged ≥50 years compared with that for patients aged <50 years (37.0% vs. 58.5%, respectively; P=0.044). Moreover, OS was significantly lower for cases with rT3-4 stage compared with that noted for cases with rT1-2 stage (33.1% vs. 67.0%, respectively; P<0.01). GTV-T volume ≥30 cm3 was an independent prognostic factor for PFS (HR=2.031, P=0.009). The latter was significantly lower for cases with GTV-T volume ≥30 cm3 compared with that noted for cases with GTV-T volume <30 cm3 (17.0% vs. 35.2%, respectively; P<0.01). Positive plasma EBV DNA following RT was an independent prognostic factor for DMFS (HR=3.294, P=0.047). DMFS was significantly lower in EBV DNA-positive cases following RT compared with that noted in negative cases (72.6% vs. 84.2%, respectively; P=0.036). The use of chemotherapy or N was not a factor affecting prognosis (Table 6).
3.6 Toxicity-associated analysis
The chi-squared test indicated that the factors associated with severe late radiation injuries were recurrent T classification, tumor response following RT, GTV-T volume, treatment modality and low hemoglobin levels during RT. These five factors were included in the logistic regression analysis and the final model obtained was statistically significant (P<0.001). The model was used to correctly classify 70.1% of the study subjects, with a model sensitivity specificity, positive predictive value and negative predictive value of 71.8%, 68.8%, 75.0% and 65.1%, respectively. The model results indicated that treatment modality and recurrent T classification were poor prognostic factors associated with severe late radiation injuries, with a 4.49-fold risk of grade ≥3 late radiation injury in the CRT group compared with that of the N group (P=0.003). Moreover, a 9.06-fold risk of grade ≥3 late radiation injury was noted in the rT3-4 staging group compared with that of the rT1-2 staging group (P=0.003).