Abstract
Background
Chemoimmunotherapy is the first-line treatment of de novo metastatic nasopharyngeal carcinoma (dmNPC), with additional locoregional radiotherapy (LRRT) significantly prolonging patient survival. De novo metastatic nasopharyngeal carcinoma, however, demonstrates considerable heterogeneity, resulting in significant variability in patient outcomes. We developed and validated a prognostic tool for patients undergoing first-line chemoimmunotherapy plus LRRT and to evaluate the benefit of local therapy (LT) for distant metastases across different risk levels.
Patients and methods
We studied 364 dmNPC patients receiving initial platinum-based chemotherapy and anti-programmed cell death protein 1 immunotherapy followed by LRRT. Patients were randomly divided into training and validation cohorts (7 : 3 ratio). The primary endpoint was progression-free survival (PFS). A prognostic model for PFS was developed using recursive partitioning analysis (RPA).
Results
An RPA model categorized patients into five prognostic groups based on number of metastatic lesions, liver metastasis status, and post-treatment Epstein–Barr virus DNA levels. Survival analysis identified three distinct risk groups. High-risk patients had significantly poorer PFS compared with medium- and low-risk groups (2-year PFS rate: training cohort: 13.7% versus 69.4% versus 94.4%, P < 0.001; validation cohort: 7.8% versus 65.1% versus 87.3%, P < 0.001). We investigated the impact of LT for distant metastases across these risk groups and found that only patients in the medium-risk group derived benefit from LT (2-year PFS rate: 77.5% versus 64.0%; hazard ratio = 0.535, 95% confidence interval 0.297-0.966, P = 0.035). Conversely, no survival benefit from LT for distant metastases was observed in the low-risk (P = 0.218) and high-risk subgroups (P = 0.793).
Conclusions
Our RPA-based prognostic model integrates number of metastatic lesions, liver metastasis status, and post-treatment Epstein–Barr virus DNA levels to predict PFS in dmNPC patients undergoing chemoimmunotherapy plus LRRT. This model offers personalized treatment guidance, suggesting that patients in the medium-risk group may benefit from LT for distant metastases, while those in high- and low-risk groups may not.
Key words: de novo metastatic nasopharyngeal carcinoma, immunotherapy, radiotherapy, EBV DNA
Highlights
-
•
First study to develop a prognostic tool for individualized predictions in dmNPC treated with chemoimmunotherapy and LRRT.
-
•
Recursive partitioning analysis identified three risk groups with distinct PFS outcomes.
-
•
High-risk patients have poorer survival compared with those in the low- and medium-risk groups.
-
•
Only medium-risk patients derived benefit from additional local therapy for distant metastases.
Introduction
Nasopharyngeal carcinoma (NPC) is one of the most common head and neck cancers, primarily affecting populations in southern China, North Africa, and Southeast Asia, with notably lower incidence rates observed in other regions.1 As reported in Global Cancer Statistics 2020, 133 354 new cases of NPC were diagnosed globally,1 of which 4%-10% were classified as de novo metastatic NPCs (dmNPC).2, 3, 4 Among its clinical presentations, dmNPC poses a particularly challenging scenario due to its aggressive nature and historically poor prognosis. Recent studies indicate that combining immunotherapy with chemotherapy significantly improves survival in patients with recurrent or metastatic NPC compared with chemotherapy alone.5, 6, 7 Consequently, this combination has emerged as the preferred first-line treatment of dmNPC.8,9
Increasing evidence supports the combination of locoregional radiotherapy (LRRT) with palliative chemotherapy, significantly improving prognosis for patients with dmNPC compared with chemotherapy alone.10, 11, 12, 13, 14 In the era of immunotherapy, many dmNPC patients have received LRRT following first-line chemoimmunotherapy, with several retrospective studies demonstrating its benefits.15, 16, 17 The heterogeneous nature of dmNPC, however, remains a significant challenge for consistent treatment outcomes.18, 19, 20 Thus, there is critical need for refined prognostic tools to accurately stratify patients based on risk factors and guide personalized treatment decisions for LRRT.
Additionally, there is currently limited research and no consensus in clinical practice regarding the use of additional local therapy (LT) for distant metastatic lesions in patients with dmNPC. Typically, LT is administered to patients with oligometastases that are well controlled locally, with the aim of achieving a potential cure. For patients with widespread metastases, limited evidence supports survival benefits from LT for distant metastases, and palliative management is recommended for symptomatic patients.9 Therefore, there is a critical need for more precise prognostic stratification tools to identify the most suitable candidates for LT for distant metastases.
To address these gaps, our study aimed to develop and validate a prognostic model specifically tailored for dmNPC patients undergoing first-line chemoimmunotherapy followed by LRRT. We utilized recursive partitioning analysis (RPA) to identify robust prognostic factors and establish distinct risk groups based on their impact on progression-free survival (PFS). The development of this prognostic model seeks to provide clinicians with a practical tool for personalized treatment planning, potentially enhancing outcomes in this challenging patient population.
Methods
Patient selection
Between July 2018 and June 2023, a retrospective analysis was conducted on 364 patients diagnosed with dmNPC. Eligible patients included those with previously untreated, pathologically confirmed dmNPC. They received at least two cycles of platinum-based chemotherapy and anti-programmed cell death protein 1 (PD-1) immunotherapy followed by LRRT. Comprehensive clinical data from both the baseline and after treatment were required for efficacy assessments. The research protocol received approval from the Research Ethics Committee of Sun Yat-Sen University Cancer Center (approval number: B2024-431-01).
Baseline evaluation and treatment
Patients were categorized based on the eight edition of the Tumor, Node, Metastasis staging manual by the American Joint Committee on Cancer/Union for International Cancer Control. Each patient included in this study had their serum Epstein–Barr virus (EBV) levels measured within 2 weeks before treatment. Metastatic sites and lesion counts were assessed using imaging modalities before treatment, with 349 patients (95.9%) undergoing positron emission tomography-computed tomography (PET-CT) scans and the remaining 15 patients (4.1%) receiving chest and abdominal CT scans plus whole-body bone scan. All patients received platinum-based combined anti-PD-1 inhibitors as first-line therapy. Chemotherapy regimens included gemcitabine/platinum (GP), docetaxel/platinum (TP), platinum/5-fluorouracil (PF), docetaxel/platinum/5-fluorouracil (TPF), and docetaxel/platinum/capecitabine (TPC). The TPC regimen was grouped with TPF due to conversion of capecitabine to fluorouracil for action in the body. Anti-PD-1 inhibitors included toripalimab, camrelizumab, sintilimab, tislelizumab, pembrolizumab, and nivolumab. Due to limited numbers, patients receiving nivolumab (six patients) and pembrolizumab (nine patients) were combined and categorized as ‘others’ for analysis. Chemoimmunotherapy cycles were administered every 3 weeks. All patients underwent intensity-modulated radiotherapy-based LRRT. Detailed treatment information is summarized in Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2024.103960.
Follow-up and outcome
Patients underwent regular follow-up evaluations every two to four cycles before the completion of LRRT and every 3-6 months after that. These evaluations included plasma EBV DNA detection and radiologic assessments such as CT, magnetic resonance imaging, or PET/CT, which were selectively carried out to assess tumor response. Tumor response was measured according to the RECIST (version 1.1) guidelines.21 The primary endpoint of the study was PFS, calculated from treatment initiation to the first occurrence of disease progression or death from any cause.
Statistical analysis
The reverse Kaplan–Meier method was utilized to examine the median follow-up time. The cohort of 364 patients was randomly divided into a training group (n = 254) and a validation group (n = 110) in a 7 : 3 ratio. Statistical comparisons between groups were carried out utilizing chi-square or Fisher’s exact test. The Systemic Immune-Inflammation Index (SII) and Prognostic Nutritional Index (PNI) were dichotomized based on the optimal cut-off point determined by maximally selected rank statistics. The cut-off of more than five metastatic lesions was chosen based on previous studies.15,22 Cox proportional hazard models were employed to assess hazard ratios (HRs) and 95% confidence intervals (CIs) to examine the relationship between variables and PFS in the training cohort. Variables P < 0.1 in the univariate analyses were selected for further evaluation. Subsequently, a stepwise multivariable Cox regression model was constructed to identify independent predictors of PFS, adjusting for potential confounders and interactions among variables. Variables with a significance level of P < 0.05 in the multivariate analyses were retained for the prognostic model development.
Following Cox regression analysis, an RPA model was established utilizing the significant predictors identified. This tree-based approach segmented patients into risk groups based on their characteristics. Survival curves were estimated utilizing the Kaplan–Meier method and compared with the log-rank test. Similar survival curve groups were merged to form the final RPA model. A time-dependent receiver operating characteristic (ROC) curve analysis was conducted to assess the predictive accuracy of the developed RPA model on both the training and validation cohorts. To ensure comprehensive analysis across the entire cohort of 364 individuals, we utilized the complete dataset to assess the efficacy of LT in various risk subgroups. All statistical analyses were conducted utilizing R software (version 4.3.3). Two-tailed P values <0.05 were deemed statistically significant.
Results
Patient characteristics and outcomes
A total of 364 dmNPC patients who were treated with LRRT after first-line chemoimmunotherapy met the inclusion criteria. A comparison of the patient characteristics in the training and validation sets is shown in Table 1. Among all included patients, the median age was 45 years (interquartile range: 36-55 years), with males comprising 83.2% of the cohort. Most patients had T3-4 (90.7%) and/or N2-3 (85.7%) disease. Bone metastasis was the most common metastatic site (72.8%), followed by liver (28.8%) and lung (22.3%). At baseline, 148 patients (40.7%) had high plasma EBV DNA levels (>104 copies/ml), and 73.1% achieved EBV DNA clearance after first-line chemoimmunotherapy. After a median follow-up of 26.1 months (95% CI 24.5-28 months), the median PFS and overall survival (OS) were not achieved, with an objective response rate (ORR) of 91.8%. The 1-, 2-, and 3-year PFS rates were 83.7%, 67.6%, and 58.4%, respectively, and the corresponding OS rates were 97.8%, 87.9%, and 80.1%.
Table 1.
Comparison of patient characteristics within training and validation cohorts
| Variable | Training cohort |
Validation cohort |
P value |
|---|---|---|---|
| (n = 254) | (n = 110) | ||
| Age (years) | 0.095 | ||
| <45 | 115 (45.3) | 61 (55.5) | |
| ≥45 | 139 (54.7) | 49 (44.5) | |
| Sex | 1.000 | ||
| Male | 211 (83.1) | 92 (83.6) | |
| Female | 43 (16.9) | 18 (16.4) | |
| Smoking | 0.245 | ||
| No | 159 (62.6) | 61 (55.5) | |
| Yes | 95 (37.4) | 49 (44.5) | |
| Drinking | 0.253 | ||
| No | 215 (84.6) | 87 (79.1) | |
| Yes | 39 (15.4) | 23 (20.9) | |
| Comorbidities | 1.000 | ||
| No | 202 (79.5) | 87 (79.1) | |
| Yes | 52 (20.5) | 23 (20.9) | |
| Karnofsky score | 1.000 | ||
| ≤80 | 20 (7.9) | 8 (7.3) | |
| >80 | 234 (92.1) | 102 (92.7) | |
| BMI (kg/m2) | 0.598 | ||
| <18.5 | 19 (7.5) | 7 (6.4) | |
| 18.5-22.9 | 99 (39.0) | 50 (45.5) | |
| 23.0-27.4 | 114 (44.9) | 42 (38.2) | |
| ≥27.5 | 22 (8.7) | 11 (10.0) | |
| Histology | 1.000 | ||
| II | 1 (0.4) | 0 (0.0) | |
| III | 253 (99.6) | 110 (100.0) | |
| Tumor category | 0.660 | ||
| T1-2 | 23 (9.1) | 11 (10.0) | |
| T3 | 121 (47.6) | 57 (51.8) | |
| T4 | 110 (43.3) | 42 (38.2) | |
| Node category | 0.057 | ||
| N0-1 | 29 (11.4) | 23 (20.9) | |
| N2 | 76 (29.9) | 31 (28.2) | |
| N3 | 149 (58.7) | 56 (50.9) | |
| Liver metastases | 0.954 | ||
| No | 180 (70.9) | 79 (71.8) | |
| Yes | 74 (29.1) | 31 (28.2) | |
| Lung metastases | 0.579 | ||
| No | 200 (78.7) | 83 (75.5) | |
| Yes | 54 (21.3) | 27 (24.5) | |
| Bone metastases | 0.508 | ||
| No | 66 (26.0) | 33 (30.0) | |
| Yes | 188 (74.0) | 77 (70.0) | |
| No. of metastatic sites | 1.000 | ||
| ≤2 | 230 (90.6) | 100 (90.9) | |
| >2 | 24 (9.4) | 10 (9.1) | |
| No. of metastatic lesions | 1.000 | ||
| ≤5 | 130 (51.2) | 57 (51.8) | |
| >5 | 124 (48.8) | 53 (48.2) | |
| Lactate dehydrogenase (U/l) | 0.774 | ||
| ≤250 | 184 (72.4) | 82 (74.5) | |
| >250 | 70 (27.6) | 28 (25.5) | |
| C-reactive protein (mg/l) | 0.760 | ||
| <3 | 114 (44.9) | 52 (47.3) | |
| ≥3 | 140 (55.1) | 58 (52.7) | |
| SII | 0.108 | ||
| ≤597.86 | 63 (24.8) | 37 (33.6) | |
| >597.86 | 191 (75.2) | 73 (66.4) | |
| PNI | 1.000 | ||
| ≤49.00 | 29 (11.4) | 12 (10.9) | |
| >49.00 | 225 (88.6) | 98 (89.1) | |
| Pretreatment EBV DNA (copies/ml) | 0.148 | ||
| <10 000 | 144 (56.7) | 72 (65.5) | |
| ≥10 000 | 110 (43.3) | 38 (34.5) | |
| Post-treatment EBV DNA | 1.000 | ||
| Undetectable | 186 (73.2) | 80 (72.7) | |
| Detectable | 68 (26.8) | 30 (27.3) | |
| First-line PCT regimens | 0.672 | ||
| GP | 138 (54.3) | 56 (50.9) | |
| TP | 63 (24.8) | 34 (30.9) | |
| TPF | 47 (18.5) | 18 (16.4) | |
| PF | 6 (2.4) | 2 (1.8) | |
| Cycle of first-line PCT | 0.548 | ||
| <6 | 69 (27.2) | 34 (30.9) | |
| ≥6 | 185 (72.8) | 76 (69.1) | |
| Immunotherapy drugs | 0.004 | ||
| Camrelizumab | 87 (34.3) | 23 (20.9) | |
| Tislelizumab | 61 (24.0) | 36 (32.7) | |
| Toripalimab | 70 (27.6) | 24 (21.8) | |
| Sintilimab | 30 (11.8) | 18 (16.4) | |
| Othersa | 6 (2.4) | 9 (8.2) | |
| Cycle of immunotherapy | 1.000 | ||
| <6 | 13 (5.1) | 6 (5.5) | |
| ≥6 | 241 (94.9) | 104 (94.5) | |
| Radiotherapy dose (Gy) | 1.000 | ||
| <66 | 10 (3.9) | 4 (3.6) | |
| ≥66 | 244 (96.1) | 106 (96.4) | |
| Concurrent treatment | 0.562 | ||
| No | 17 (6.7) | 12 (10.9) | |
| Immunotherapy ± targeted | 103 (40.6) | 49 (44.5) | |
| Platinum ± targeted | 22 (8.7) | 9 (8.2) | |
| Platinum + immunotherapy ± targeted | 80 (31.5) | 26 (23.6) | |
| Capecitabine + immunotherapy ± targeted | 14 (5.5) | 5 (4.5) | |
| Othersb | 18 (7.1) | 9 (8.2) | |
| Local therapy for metastases | 0.981 | ||
| No | 159 (62.6) | 68 (61.8) | |
| Yes | 95 (37.4) | 42 (38.2) |
BMI, body mass index; EBV, Epstein–Barr virus; GP, gemcitabine and platinum; PCT, palliative chemotherapy; PF, platinum and 5-fluorouracil; PNI, prognostic nutritional index; SII, systemic immune-inflammation index; TP, docetaxel and platinum; TPF, docetaxel, platinum and 5-fluorouracil.
Including pembrolizumab and nivolumab.
Including capecitabine alone, targeted drugs alone, or a combination of both.
Efficacy of different chemoimmunotherapy and concurrent regimens
Univariable analysis of the training cohort showed no statistically significant differences in PFS among patients receiving different chemotherapy regimens, immunotherapy drugs, or concurrent regimens (referring to systemic treatments that were administered concurrently with radiotherapy). Consistently, no difference in median PFS was observed among patients who received different chemotherapy regimens (P = 0.719; Supplementary Figure S1A, available at https://doi.org/10.1016/j.esmoop.2024.103960) and anti-PD-1 inhibitors (P = 0.168; Supplementary Figure S1B, available at https://doi.org/10.1016/j.esmoop.2024.103960). In our study, 137 patients (37.6%) received concurrent platinum-based chemotherapy, 198 patients (54.3%) received other concurrent regimens, including immunotherapy, targeted therapy, capecitabine, or combinations thereof, and 29 patients (8.0%) did not receive concurrent therapy. No significant difference in PFS was observed among the different concurrent therapies (P = 0.527; Supplementary Figure S1C, available at https://doi.org/10.1016/j.esmoop.2024.103960).
Development and validation of the prognostic model
The univariate analysis results for PFS of the training cohort are summarized in Supplementary Table S2, available at https://doi.org/10.1016/j.esmoop.2024.103960. The variables with P < 0.1 were selected for multivariate Cox regression analysis. For the prognostic model, four predictors were selected based on Akaike information criterion (AIC) and stepwise variable selection. The final multivariate Cox regression model outcomes are shown in Table 2. Independent factors for PFS in dmNPC patients included liver metastases status (yes versus no, HR: 1.853, 95% CI 1.141-3.008, P = 0.013), number of metastatic lesions (>5 versus ≤5, HR: 1.795, 95% CI 1.089-2.965, P = 0.022), lactate dehydrogenase (>250 versus ≤250 U/L, HR: 1.874, 95% CI 1.185-2.962, P = 0.007), and post-treatment EBV DNA (detectable versus undetectable, HR: 5.090, 95% CI 3.177-8.154, P < 0.001).
Table 2.
Significant prognostic factors for PFS in the training cohort
| Variable | HR (95% CI) | P value |
|---|---|---|
| Liver metastases | ||
| No | Reference | |
| Yes | 1.853 (1.141-3.008) | 0.013 |
| No. of metastatic lesions | ||
| ≤5 | Reference | |
| >5 | 1.795 (1.089-2.965) | 0.022 |
| Lactate dehydrogenase (U/L) | ||
| ≤250 | Reference | |
| >250 | 1.874 (1.185-2.962) | 0.007 |
| Post treatment EBV DNA | ||
| Undetectable | Reference | |
| Detectable | 5.090 (3.177-8.154) | <0.001 |
CI, confidence interval; EBV, Epstein–Barr virus; HR, Hazard ratio; PFS, progression-free survival.
An integrated prognostic model utilizing RPA was developed based on the above independent factors. The model stratifies patients into five distinct groups: group A (undetectable post-treatment EBV DNA, absence of liver metastases, ≤5 metastatic lesions), group B (undetectable post-treatment EBV DNA, absence of liver metastases, >5 metastatic lesions), group C (undetectable post-treatment EBV DNA, presence of liver metastases), group D (detectable post-treatment EBV DNA, ≤5 metastatic lesions), and group E (detectable post-treatment EBV DNA, >5 metastatic lesions) (Figure 1A). The Kaplan–Meier curves for each group are illustrated in Figure 1B. Pairwise comparisons using the log-rank test indicated no significant difference in PFS between groups B and C or between groups C and D (Figure 1B). Therefore, we combined groups B, C, and D as the medium-risk group. Finally, 78 (30.7%) patients were categorized as low-risk (group A), 142 (55.9%) as medium-risk, and 34 (13.4%) as high-risk (group E). The corresponding 1-year PFS rates were 100%, 83.9%, and 47.1%, 2-year PFS rates were 94.4%, 69.4%, and 13.7%, while the 3-year PFS rates were 82.7%, 58.3%, and 0, respectively. With low-risk patients being the reference, the HRs for medium- and high-risk patients were 4.207 (95% CI 1.897-9.330; P < 0.001) and 20.120 (95% CI 8.687-46.600; P < 0.001), respectively. Survival curves stratified based on the three strata showed significant survival differences (global log-rank P < 0.001, Figure 2A).
Figure 1.
RPA model development and prognostic stratification of dmNPC patients. (A) RPA model for predicting PFS in the training cohort, (B) Kaplan–Meier curves of PFS for patients grouped by five prognostic groups. CI, confidence interval; EBV, Epstein–Barr virus; HR, hazard ratio; mPFS, median progression-free survival; NA, not available; NR, not reached; PFS, progression-free survival; RPA, recursive partitioning analysis.
Figure 2.
Risk stratification and model evaluation. Kaplan–Meier curves of PFS for patients stratified by different risk groups in (A) the training and (B) the validation cohort. The time-dependent ROC curves of the RPA model for predicting the 1-, 2-, and 3-year PFS rate in (C) the training and (D) the validation cohort. AUC, area under the curve; HR, hazard ratio; PFS, progression-free survival; ROC, receiver operating characteristic; RPA, recursive partitioning analysis.
In the validation cohort, patients were grouped into three risk categories through the model, with 35 patients classified in the low-risk group, 60 patients in the medium-risk group, and 15 patients in the high-risk group. Corresponding 1-year PFS rates were 97.1%, 84.8%, and 46.7%, 2-year PFS rates were 87.3%, 65.1%, and 7.8%, and 3-year PFS rates were 87.3%, 59.2%, and 0, respectively. With stratum one as reference, the HRs for strata two and three were 4.457 (95% CI 1.333-14.900; P = 0.015) and 20.003 (95% CI 5.635-71.010; P < 0.001), respectively. Survival curves stratified by the three risk groups also demonstrated significant survival differences (global log-rank P < 0.001, Figure 2B). These findings underscore the robust prognostic value of the developed model in predicting survival outcomes among dmNPC patients treated with LRRT following first-line chemoimmunotherapy.
Additionally, the performances of the RPA model were assessed by time-dependent receiver operating characteristic (ROC) curves, with time-area under the curve (AUC) values of 0.790, 0.776, and 0.721 for 1-, 2-, and 3-year PFS, respectively (Figure 2C). In the validation cohort, the time-AUC values were 0.770, 0.722, and 0.829 for 1-, 2-, and 3-year PFS, respectively (Figure 2D).
Identification of best-fit LT for metastases candidates
In the overall cohort, 137 patients (37.6%) received additional LT for metastases, with 118 patients (32.4%) receiving radiotherapy, 13 (3.6%) receiving ablation, 3 undergoing surgery, and 3 receiving a combination of ablation or surgery with radiotherapy. The specific locations are outlined in Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2024.103960. Among these patients, 118 received RT for their metastases in overall dataset. The clinical profiles of patients across various risk and treatment categories are detailed in Supplementary Table S3, available at https://doi.org/10.1016/j.esmoop.2024.103960. Interestingly, the impact of LT on metastases varied across the three risk groups. In the low-risk group, the PFS showed no significance between patients treated with and without LT (HR = 2.151, 95% CI 0.603-7.676, P = 0.227; Figure 3A). Similarly, in the high-risk group, PFS did not significantly differ between these treatment groups (HR = 0.618, 95% CI 0.284-1.347, P = 0.221; Figure 3A). While in the medium-risk group, patients who received LT for distant metastases showed improved PFS compared with those who did not (HR = 0.540, 95% CI 0.317-0.920, P = 0.021; Figure 3B).
Figure 3.
Comparison of PFS between patients in LT for metastases and non-LT for metastases groups in the overall cohort. (A) Low-risk patients, (B) medium-risk patients, and (C) high-risk patients. CI, confidence interval; HR, hazard ratio; LT, local therapy; PFS, progression-free survival.
Discussion
This study represents a pioneering effort in developing and validating an RPA-based prognostic model that combines baseline metastatic characteristics with post-treatment EBV DNA levels to provide individualized prediction and risk stratification for patients with dmNPC undergoing LRRT following chemoimmunotherapy. Notably, our additional analysis revealed that LT for distant metastases conferred significant therapeutic benefit specifically within the medium-risk subgroup, contrasting with the lack of such benefit observed in low- and high-risk patients.
NPC, formerly known as lymphoepithelioma due to the massive infiltration of non-malignant lymphocytes,23 is characterized by high PD-L1 expression, found in tumor cells or tumor-associated immune cells in 83%-92% of patients.24 These unique characteristics make NPC patients strong candidates for anti-PD-1 immunotherapy. Based on three successful phase III trials, the combination of anti-PD-1 immunotherapy with GP has become the recommended first-line treatment of recurrent or metastatic NPC in major guidelines.5, 6, 7 In our study, 194 patients (53.3%) were treated with a GP regimen, and we did not observe significant variance in PFS between patients receiving different chemotherapy regimens. Additionally, we did not observe significant differences in PFS among various PD-1 inhibitors.
Previous retrospective studies have indicated that adding LRRT to first-line palliative chemotherapy improves survival in patients with dmNPC,10, 11, 12, 13 which has been validated by a randomized phase III trial.14 In addition to directly targeting tumor cells, radiotherapy has the potential to stimulate immune responses, including reprogramming the tumor microenvironment through immunogenic cell death and the release of new antigens to activate the immune system. Radiotherapy also up-regulates PD-L1 expression in tumor and inflammatory cells, offering synergistic potential when combined with immunotherapy.25, 26, 27, 28, 29, 30 In the era of immunotherapy, several retrospective studies have highlighted the benefits of LRRT following first-line chemoimmunotherapy in dmNPC patients.15, 16, 17 A recent phase II clinical trial focused on patients responding to initial chemotherapy, followed by combined LRRT and anti-PD-1 therapy, reported a median PFS of 19.4 months.17 In our study, however, the median PFS has not yet been reached, exceeding 26.1 months of follow-up. This difference may be attributed to the early introduction of anti-PD-1 immunotherapy during the induction phase before radiotherapy, facilitating T-cell priming within the tumor microenvironment. These findings underscore the potential clinical advantages of initiating anti-PD-1 monoclonal antibodies earlier in treatment protocols.
In recent years, there has been ongoing debate regarding the LT of metastatic lesions in metastatic NPC. Previous studies have indicated that such treatments can prolong patient survival, including a large-scale matched study involving 2041 patients with metastatic NPC,20,31 which showed improved OS with LT of metastatic lesions. Subgroup analyses from this retrospective analysis suggested that this approach improved survival regardless of whether patients had dmNPC or received immunotherapy.31 A multicenter population study involving three cohorts with a total of 977 dmNPC patients demonstrated, however, that local treatment of metastatic lesions did not improve survival outcomes in patients receiving systemic chemotherapy and radical radiotherapy for primary tumors, even among those with oligometastases and no liver involvement.18 These inconsistent results further validate our findings, specifically that only medium-risk patients with dmNPC benefit from LT targeting metastatic lesions. We hypothesize that low-risk patients exhibit a minimal tumor burden (absence of liver metastases, ≤5 metastatic lesions) and are sensitive to chemoimmunotherapy (undetectable EBV DNA after chemoimmunotherapy), and that LRRT has a distal effect that is enhanced by the addition of immunotherapy. Consequently, low-risk patients achieve excellent outcomes with systemic chemoimmunotherapy followed by LRRT, obviating the necessity for additional LT for metastatic lesions. In contrast, patients categorized into the high-risk subgroup demonstrated unfavorable PFS after LRRT. These patients present with extensive tumor burden (>5 metastatic lesions) and resistance to chemoimmunotherapy (detectable post-treatment EBV DNA, which has also been previously reported to be associated with radiotherapy resistance32). As a result, LRRT is unlikely to provide substantial benefit due to rapid tumor progression, making additional LT of metastases ineffective. These findings suggest that aggressive approaches targeting the primary tumor and regional nodes or distant metastases to manage disease progression are ineffective in high-risk patients. Thus, these patients should be referred for sequential consolidation chemotherapy or chemoimmunotherapy strategies.
With the advent of the immunotherapy era, significant advancements have transformed the treatment landscape for patients with dmNPC, offering the potential for extended survival. The concept of managing primary tumor and regional nodes as well as distant metastases aggressively in this context is evolving rapidly. Without prospective data clarifying the ideal candidates for LRRT, using prognostic models can help refine patient selection, moving beyond reliance solely on clinical experience and avoiding unnecessary overtreatment. Thus, the selection of appropriate candidates for LRRT and local treatment of metastatic lesions based on our prognostic tool represents a critical advancement towards precision therapy for dmNPC.
This study, however, has several limitations. Firstly, being a retrospective single-institution study, the results warrant validation in prospective studies conducted across multiple centers. Secondly, the proposed prognostic model was not externally validated due to the single-center nature of patient recruitment. Thirdly, the LT for metastatic lesions predominantly involved radiotherapy in our study, with a smaller proportion of patients undergoing surgery or ablation. As a result, the impact of these other modalities requires further investigation. Fourthly, there may be a selection bias in how patients were chosen for LT of metastases. The criteria for selecting patients for these treatments were not randomized and could influence the results. Lastly, the study population was confined to an EBV-endemic region, potentially influencing tumor characteristics differently than in non-endemic areas. Therefore, further research is essential to determine the generalizability of our findings to NPC in non-endemic regions.
Conclusion
We developed and validated an integrated RPA-based prognostic model for PFS in patients with dmNPC, incorporating the independent predictors including the number of metastatic lesions, liver metastases, and post-treatment EBV DNA level. This model effectively stratifies patients into three distinct prognostic subgroups with significantly different PFS outcomes. Our findings also highlight that medium-risk patients may benefit from additional LT for distant metastases, whereas both low- and high-risk patients may not.
Acknowledgments
Funding
This work was supported by the National Key Research and Development Program of China [grant numbers 2022YFC2505800, 2022YFC2705005], National Natural Science Foundation of China [grant numbers 32200651, 82203776, 82203125, 82222050, 82272739, 82272882, 82173287, 82073003, 82003267, 82002852, 82373258, 82372980, 82361168664, 82303967], Guangdong Basic and Applied Basic Research Foundation [grant numbers 2021B1515230002, 2023B1515120092], Science and Technology Program of Guangzhou [grant numbers 202201011561, 2023A04J2127, 2024B03J1248], Sun Yat-sen University Clinical Research 5010 Program [grant numbers 201315, 2015021, 2017010, 2019023], Innovative Research Team of High-level Local Universities in Shanghai [grant number SSMU-ZLCX20180500], Postdoctoral Innovative Talent Support Program [grant number BX20220361], Planned Science and Technology Project of Guangdong Province [grant number 2019B020230002], Key Youth Teacher Cultivating Program of Sun Yat-sen University [grant number 20ykzd24], and Fundamental Research Funds for the Central Universities.
Disclosure
The authors have declared no conflicts of interest.
Contributor Information
Y. Liang, Email: liangyuj@sysucc.org.cn.
Q. Chen, Email: chenqy@sysucc.org.cn.
S. Guo, Email: guoshsh@sysucc.org.cn.
Supplementary data
References
- 1.Sung H., Ferlay J., Siegel R.L., et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–249. doi: 10.3322/caac.21660. [DOI] [PubMed] [Google Scholar]
- 2.Lee A.W.M., Ng W.T., Chan L.K., et al. The strength/weakness of the AJCC/UICC staging system (7th edition) for nasopharyngeal cancer and suggestions for future improvement. Oral Oncol. 2012;48(10):1007–1013. doi: 10.1016/j.oraloncology.2012.03.022. [DOI] [PubMed] [Google Scholar]
- 3.Pan J.J., Ng W.T., Zong J.F., et al. Proposal for the 8th edition of the AJCC/UICC staging system for nasopharyngeal cancer in the era of intensity-modulated radiotherapy. Cancer. 2016;122(4):546–558. doi: 10.1002/cncr.29795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Chen Y.P., Chan A.T.C., Le Q.T., Blanchard P., Sun Y., Ma J. Nasopharyngeal carcinoma. Lancet. 2019;394(10192):64–80. doi: 10.1016/S0140-6736(19)30956-0. [DOI] [PubMed] [Google Scholar]
- 5.Mai H.Q., Chen Q.Y., Chen D., et al. Toripalimab or placebo plus chemotherapy as first-line treatment in advanced nasopharyngeal carcinoma: a multicenter randomized phase 3 trial. Nat Med. 2021;27(9):1536–1543. doi: 10.1038/s41591-021-01444-0. [DOI] [PubMed] [Google Scholar]
- 6.Yang Y., Pan J., Wang H., et al. Tislelizumab plus chemotherapy as first-line treatment for recurrent or metastatic nasopharyngeal cancer: a multicenter phase 3 trial (RATIONALE-309) Cancer Cell. 2023;41(6):1061–1072.e1064. doi: 10.1016/j.ccell.2023.04.014. [DOI] [PubMed] [Google Scholar]
- 7.Chan A.T.C., Lee V.H.F., Hong R.L., et al. Pembrolizumab monotherapy versus chemotherapy in platinum-pretreated, recurrent or metastatic nasopharyngeal cancer (KEYNOTE-122): an open-label, randomized, phase III trial. Ann Oncol. 2023;34(3):251–261. doi: 10.1016/j.annonc.2022.12.007. [DOI] [PubMed] [Google Scholar]
- 8.Bossi P., Chan A.T., Even C., Machiels J.P. ESMO Guidelines Committee. ESMO-EURACAN Clinical Practice Guideline update for nasopharyngeal carcinoma: adjuvant therapy and first-line treatment of recurrent/metastatic disease. Ann Oncol. 2023;34(3):247–250. doi: 10.1016/j.annonc.2022.11.011. [DOI] [PubMed] [Google Scholar]
- 9.Network N.C.C. Head and Neck Cancers, Version 4. 2024. https://www.nccn.org/login?ReturnURL=https://www.nccn.org/professionals/physician_gls/pdf/head-and-neck.pdf Available at.
- 10.Rusthoven C.G., Lanning R.M., Jones B.L., et al. Metastatic nasopharyngeal carcinoma: patterns of care and survival for patients receiving chemotherapy with and without local radiotherapy. Radiother Oncol. 2017;124(1):139–146. doi: 10.1016/j.radonc.2017.03.019. [DOI] [PubMed] [Google Scholar]
- 11.Zeng L., Tian Y.M., Huang Y., et al. Retrospective analysis of 234 nasopharyngeal carcinoma patients with distant metastasis at initial diagnosis: therapeutic approaches and prognostic factors. PLoS One. 2014;9(9):e108070. doi: 10.1371/journal.pone.0108070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hu J., Kong L., Gao J., Hu W., Guan X., Lu J.J. Use of radiation therapy in metastatic nasopharyngeal cancer improves survival: a SEER analysis. Sci Rep. 2017;7(1):721. doi: 10.1038/s41598-017-00655-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hu S.X., He X.H., Dong M., et al. Systemic chemotherapy followed by locoregional definitive intensity-modulated radiation therapy yields prolonged survival in nasopharyngeal carcinoma patients with distant metastasis at initial diagnosis. Med Oncol. 2015;32(9):224. doi: 10.1007/s12032-015-0663-2. [DOI] [PubMed] [Google Scholar]
- 14.You R., Liu Y.P., Huang P.Y., et al. Efficacy and safety of locoregional radiotherapy with chemotherapy vs chemotherapy alone in de novo metastatic nasopharyngeal carcinoma: a multicenter phase 3 randomized clinical trial. JAMA Oncol. 2020;6(9):1345–1352. doi: 10.1001/jamaoncol.2020.1808. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hu Y.J., Lu T.Z., Zhang H., et al. Locoregional radiotherapy improves survival outcomes in de novo metastatic nasopharyngeal carcinoma treated with chemoimmunotherapy. ESMO Open. 2023;8(5):101629. doi: 10.1016/j.esmoop.2023.101629. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chen Y., Chen C., Peng H., et al. Risk-adapted locoregional radiotherapy strategies based on a prognostic nomogram for de novo metastatic nasopharyngeal carcinoma patients treated with chemoimmunotherapy. Sci Rep. 2024;14(1):3950. doi: 10.1038/s41598-024-54230-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chen S.Y., Duan X.T., Li H.F., et al. Efficacy of sequential chemoradiotherapy combined with toripalimab in de novo metastatic nasopharyngeal carcinoma: A phase II trial. Cell Rep Med. 2023;4(11):101279. doi: 10.1016/j.xcrm.2023.101279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Zou X., You R., Liu H., et al. Establishment and validation of M1 stage subdivisions for de novo metastatic nasopharyngeal carcinoma to better predict prognosis and guide treatment. Eur J Cancer. 2017;77:117–126. doi: 10.1016/j.ejca.2017.02.029. [DOI] [PubMed] [Google Scholar]
- 19.Shen L., Dong J., Li S., et al. M1 stage subdivision and treatment outcome of patients with bone-only metastasis of nasopharyngeal carcinoma. Oncologist. 2015;20(3):291–298. doi: 10.1634/theoncologist.2014-0206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Liang Y.J., Sun X.S., Yang Z.C., et al. Effect of local treatment for metastasis and its sequence with chemotherapy on prognosis of post-treatment metastatic nasopharyngeal carcinoma patients. Oral Oncol. 2019;92:40–45. doi: 10.1016/j.oraloncology.2019.03.015. [DOI] [PubMed] [Google Scholar]
- 21.Eisenhauer E.A., Therasse P., Bogaerts J., et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1) Eur J Cancer. 2009;45(2):228–247. doi: 10.1016/j.ejca.2008.10.026. [DOI] [PubMed] [Google Scholar]
- 22.Tian Y.H., Zou W.H., Xiao W.W., et al. Oligometastases in AJCC stage IVc nasopharyngeal carcinoma: A subset with better overall survival. Head Neck. 2016;38(8):1152–1157. doi: 10.1002/hed.24345. [DOI] [PubMed] [Google Scholar]
- 23.Young L.S., Dawson C.W. Epstein-Barr virus and nasopharyngeal carcinoma. Chin J Cancer. 2014;33(12):581–590. doi: 10.5732/cjc.014.10197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Mai H.Q., Chen Q.Y., Chen D., et al. Toripalimab plus chemotherapy for recurrent or metastatic nasopharyngeal carcinoma: the JUPITER-02 randomized clinical trial. JAMA. 2023;330(20):1961–1970. doi: 10.1001/jama.2023.20181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Herrera F.G., Bourhis J., Coukos G. Radiotherapy combination opportunities leveraging immunity for the next oncology practice. CA Cancer J Clin. 2017;67(1):65–85. doi: 10.3322/caac.21358. [DOI] [PubMed] [Google Scholar]
- 26.Sharabi A.B., Lim M., DeWeese T.L., Drake C.G. Radiation and checkpoint blockade immunotherapy: radiosensitisation and potential mechanisms of synergy. Lancet Oncol. 2015;16(13):e498–e509. doi: 10.1016/S1470-2045(15)00007-8. [DOI] [PubMed] [Google Scholar]
- 27.Weichselbaum R.R., Liang H., Deng L., Fu Y.X. Radiotherapy and immunotherapy: a beneficial liaison? Nat Rev Clin Oncol. 2017;14(6):365–379. doi: 10.1038/nrclinonc.2016.211. [DOI] [PubMed] [Google Scholar]
- 28.Twyman-Saint Victor C., Rech A.J., Maity A., et al. Radiation and dual checkpoint blockade activate non-redundant immune mechanisms in cancer. Nature. 2015;520(7547):373–377. doi: 10.1038/nature14292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Gupta A., Probst H.C., Vuong V., et al. Radiotherapy promotes tumor-specific effector CD8+ T cells via dendritic cell activation. J Immunol. 2012;189(2):558–566. doi: 10.4049/jimmunol.1200563. [DOI] [PubMed] [Google Scholar]
- 30.Deng L., Liang H., Burnette B., et al. Irradiation and anti-PD-L1 treatment synergistically promote antitumor immunity in mice. J Clin Investig. 2014;124(2):687–695. doi: 10.1172/JCI67313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Zhang M.X., Liu T., You R., et al. Efficacy of local therapy to metastatic foci in nasopharyngeal carcinoma: large-cohort strictly-matched retrospective study. Ther Adv Med Oncol. 2022;14 doi: 10.1177/17588359221112486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zhou X., Lin Y., Chen Y., et al. Epstein-Barr virus (EBV) encoded microRNA BART8-3p drives radioresistance-associated metastasis in nasopharyngeal carcinoma. J Cell Physiol. 2021;236(9):6457–6471. doi: 10.1002/jcp.30320. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.