Efficacy and Toxicity of Three Induction Chemotherapy Regimens in Locoregionally Advanced Nasopharyngeal Carcinoma: Outcomes of 10-Year Follow-Up

Background/Objective We aimed to compare the 10-year survival outcomes of induction docetaxel plus cisplatin and 5-fluorouracil (TPF), docetaxel plus cisplatin (TP), and cisplatin plus 5-fluorouracil (PF) regimens additional to concurrent chemoradiotherapy (CRT) in locoregionally advanced nasopharyngeal carcinoma (NPC). Methods Eligible patients with newly diagnosed stage III-IVA NPC were included. Propensity score matching (PSM) was used to balance prognostic covariates. Survival outcomes and toxicities between different groups were compared. Results A total of 855 patients between 2009 and 2012 were included, with 395 (46.2%), 258 (30.2%), and 202 (23.6%) receiving TPF plus CRT, TP plus CRT, and PF plus CRT regimens, respectively. After a median follow-up of 111.8 months, multivariate analysis both in the whole cohort and PSM selected 202 pairs showed that TPF plus CRT and TP plus CRT achieved significantly better 10-year overall survival (OS) than PF plus CRT. Sensitivity analysis after excluding patients with T3-4N0 disease demonstrated that TPF plus CRT still achieved significantly better OS than PF plus CRT (HR, 0.580; 95% CI, 0.395-0.852; P = 0.005), while the difference between TP plus CRT and PF plus CRT was marginally significant (HR, 0.712; 95% CI, 0.503-1.008; P = 0.056). With regard to toxicity profile, PF regimen achieved the lowest grade 3–5 toxicities (27.3%). Conclusion TPF plus CRT and TP plus CRT were better than PF plus CRT in improving the 10-year OS of patients with stage III-IVA NPC.


BACKGROUND
As an aggressive and relatively rare head and neck cancer, nasopharyngeal carcinoma (NPC) has an extremely unbalanced geographical distribution worldwide; it is endemic in Southern China and Southeast Asia but very rare in western countries (1,2). Different from other head and neck cancers, nondisseminated NPC is cured by radiotherapy. A 10-year overall survival (OS) of 87.1-100% could be achieved in patients with stage I-II disease; however, the corresponding survival outcome of patients with stage III-IVA disease was only 75.5-55.6%, and distant metastasis has come as the main failure patter (3). Unfortunately, more than 80% of patients presented with advanced diseases at initial diagnosis (2,4). Therefore, how to reduce distant metastasis and improve the therapeutic outcomes of patients with advanced disease have been widely studied.
Induction chemotherapy (IC), usually a combination of two or three cytotoxic drugs, is given before radiotherapy to eliminate clinically undetectable micrometastatic lesions, thereby reducing the rate of distant failure and improving survival. Indeed, several phase III clinical trials conducted in recent years have showed that IC additional to concurrent chemoradiotherapy (CRT) could improve both distant metastasis-free survival (DMFS) and OS (5)(6)(7)(8)(9). This evidence strengthened the role of IC in locoregionally advanced NPC, and IC plus CRT was therefore approved as the preferable treatment strategy worldwide for advanced NPC. Despite these advances, the most effective IC regimen, however, is still unknown since various regimens all achieved positive results (5-8, 10, 11). A previously retrospective study uncovered that a triple combination of docetaxel plus cisplatin and 5-fluorouracil (TPF) was better than docetaxel plus cisplatin (TP) and cisplatin plus 5-fluorouracil (PF) regimens in patients with stage III-IVA NPC (excluding T3N0) (12). However, the insufficient follow-up duration (median, 46.1 months) and inclusion of patients who did not receive concurrent chemotherapy made these results inconclusive.
Based on this premise, we conducted this study to compare the 10-year survival outcomes of patients with locoregionally advanced NPC receiving induction TPF, TP, or PF plus CRT in the era of intensity-modulated radiotherapy (IMRT).
We used PSM to balance independent prognostic factors identified above (tumor stage, alcohol intake, age, and gender) and further performed survival analysis in the selected 202 pairs (Supplementary Table S2 Table S3).

Sensitivity Analysis
We performed sensitivity analysis by excluding stage T3-4N0 disease, which was regarded as low risk of distant metastasis by previous trials (5,7). In total, 32 patients were excluded, and 189 pairs were selected by PSM from the remaining patients (Supplementary Table S4

Toxicity Comparison
Treatment adverse events of different IC regimens are shown in Table 3. As expected, the PF regimen achieved the lowest percentages of grade 3-5 toxicities (27.3%), and the TP regimen had the highest rate of grade 3-5 toxicities, which were mainly grade 3-5 neutropenia (97.1%) and febrile neutropenia (11.8%). This should be due to the application of

DISCUSSION
Our current study reported the 10-year survival outcomes of patients with stage III-IVA NPC receiving different IC regimens plus CRT in the era of IMRT. We found that TPF plus CRT and TP plus CRT achieved significantly better OS than PF plus CRT both in the whole cohort and the selected pairs by PSM. Toxicity analysis showed that the PF regimen had the lowest percentages of grade 3-5 adverse events. To date, our study is the first one to report the 10-year therapeutic outcomes of locoregionally advanced NPC treated by IC plus CRT in the era of IMRT.
Our study only recruited patients receiving induction TPF, TP, and PF regimens because these three regimens have been used most frequently and for the longest time in our center. Their efficacy in locoregionally advanced NPC has also been verified by clinical trials (5,6,(8)(9)(10)(11). Although gemcitabine plus cisplatin (GP) is also effective and may have fewer adverse events (7), the insufficient follow-up duration of patients receiving this regimen precludes them from being enrolled into this study. As previous study showed that two cycles of IC could achieve comparable outcomes as three or more cycles (13), we therefore only recruited patients receiving at least two cycles to reduce the impact of the IC cycle. Consistent with previous findings (13), results of multivariate analysis in our study also did not identify the IC cycle (2 vs. 3-4) as an independent prognostic factor. Another interesting finding was that patients receiving the TP regimen achieved lower CCD dose than those receiving the TPF regimen. The mainly responsible reason may be that a higher dose of docetaxel (75 mg/m 2 ) may reduce patients' tolerance to concurrent cisplatin/nedaplatin.
Notably, 39 new events occurred 5 years after radiotherapy, accounting for 13.1% of all events. Therefore, intensive follow-up is still needed after 5 years. Among the 39 new events, distant metastasis only accounted for 30.7% and noncancer-related death accounted for 33.3%. These results indicated that distant metastasis was no longer the main cause of treatment failure after 5 years for patients receiving IC plus CRT, and we should pay attention to noncancer-related death, which may be due to treatment-related sequelae.
When analyzing all the 855 patients together, both the TPF plus CRT and TP plus CRT groups achieved significantly better OS and DFS than the PF plus CRT group. However, some comparisons in PSM or sensitivity analysis only showed marginally significant difference. The main reason contributing to this should be the decreased sample sizes, which reduced statistical power in PSM and sensitivity analysis. Generally, induction TPF and TP regimens should be more effective than the PF regimen, which was consistent with the findings of a meta-analysis that taxanes-based IC could decrease the risk of distant metastasis by above 10% for patients with stage IVA disease (14). Moreover, the effect of docetaxel additional to induction PF regimen has also been verified in head and neck cancers (15)(16)(17)(18). Different from previous results that TPF was significantly better than TP with regard to DFS and OS endpoints (12), survival endpoints except LFFS did not significantly differ between the TPF plus CRT and TP plus CRT groups. There may be three reasons attributing to this discrepancy. First, the followup duration was much longer in our study. Second, patients with stage T3N0 were excluded in that study. Third, patients who did not receive concurrent chemotherapy were also included in that study. Limitations of our study should also be addressed. First, many potential bias may exist in our retrospective study. We therefore set strict enrollment criteria and balanced various prognostic factors by PSM to reduce bias. Moreover, we performed sensitivity analysis to further validate our findings. Second, toxic data during IC are unavailable for most of the patients due to the retrospective nature of our study. We therefore extracted these data from previously published clinical trials. To minimize the impact of ethnic differences on adverse events, we only recruited the three clinical trials (5, 6, 10) conducted in the endemic area (mainly South China). Despite this, this result should be interpreted discreetly. Third, the relatively small sample size, especially in PSM and sensitivity analysis, precluded the production of significant differences for some endpoints, although the survival curves showed obvious differences.

Patient Inclusion Criteria
We retrospectively reviewed the data of patients with newly diagnosed NPC who were treated at our center between 2009 and 2012. Patients would be enrolled for this study after meeting the following criteria: (1) newly diagnosed stage III-IVA disease; (2) receiving IC plus CRT; (3) IC regimens were TPF, TP, and PF and IC cycles ≥ 2; (4) concurrent chemotherapy regimen should be single-agent platinum; (5) treated by IMRT and received a total dose of at least 66 Gy; (6) no other malignancy. Notably, we included patients receiving TPF, TP, and PF regimens because they were most frequently used at that time. Moreover, the efficacy of these regimens has also been validated in

Pretreatment Staging Workup
Potential patients with indicated symptoms in our hospital would receive routine staging workup including physical examination, enhanced magnetic resonance imaging (MRI) of head and neck, chest computed tomography (CT) or X-ray, abdominal CT or sonography, and whole body bone scan. 18 F-FDG positron emission tomography (PET)-CT would also be recommended to patients who presented with massive lymph node or bilateral cervical lymph node metastasis. Patients were restaged by one radiologist (LT) and one radiation oncologist (YH) separately, both with more than 10-year experience in the diagnosis and treatment of NPC at our center, based on the imaging data and the eighth edition of the International Union against Cancer/American Joint Committee on Cancer (UICC/ AJCC) manual (19).

Radiotherapy
All the patients received pre-radiotherapy evaluation to exclude any contraindication. IMRT was delivered using the simultaneous integrated boost (SIB) technique. The prescribed doses were 66-70 Gy at 2.0-2.27 Gy per fraction to the planning target volume (PTV) of nasopharynx lesion and metastatic neck lymph nodes, 56-60 Gy at 30-35 fractions to the PTV of clinically high-risk regions, and 50-56 Gy at 30-35 fractions to the PTV of clinically low-risk regions. Radiotherapy fractions were delivered once per day from Monday to Friday each week.

Toxicity of Induction Chemotherapy
Due to the retrospective nature of our study, IC-related adverse events were not recorded for most of the patients. Therefore, we extracted toxicity data from three previous clinical trials conducted in endemic areas (mainly in South China) (5,6,10) to perform indirect comparisons between TPF, TP, and PF regimens.

Follow-Up Strategy and Endpoints
Patients finishing the treatment would be followed according to the institutional follow-up strategies, which included enhanced MRI of head and neck, chest CT or X-ray, abdominal CT or sonography, and whole bone scan (optional) every 3 months during the first 2 years after radiotherapy, every 6 months during the third to fifth years, and annually thereafter. For patients who lived far away from our hospital, we recommended them to receive these imaging workups at local medical centers and they would be followed by telephone. Disease recurrence including local, regional, and distant metastasis (except bone) was diagnosed by pathology. Bone metastasis was mainly confirmed by imaging methods like MRI, CT, or PET-CT.
Endpoints evaluated at our study included OS (time from diagnosis to death), disease-free survival (DFS, time from diagnosis to disease progression including noncancer-related death), distant metastasis-free survival (DMFS, time from diagnosis to first distant failure), and locoregional failure-free survival (LFFS, time from diagnosis to first local or regional recurrence or both).

Statistical Methods
T-test or one-way ANOVA was applied to determine the difference between continuous variables, and chi-square test or Fisher's exact test was used for categorical variables. Propensity score matching (PSM) (20) was employed to balance covariates (T category, N category, overall stage, gender, age, alcohol intake) between the three groups. Estimated 10-year survival outcomes of OS, DFS, DMFS, and LFFS were obtained from Kaplan-Meier methods, and the differences were compared by log-rank test. Independent prognostic factors and their corresponding hazard ratios (HRs) and 95% confidence intervals (CIs) were identified by the multivariate Cox proportional hazard model. All statistical analyses were conducted using the Stata Statistical Package 12 (StataCorp LP, College Station, TX, USA), and a two-sided P < 0.05 indicated statistical significance.

CONCLUSION
Based on the 10-year follow-up, our current study reported and compared the efficacy of three IC regimens and uncovered that TPF plus CRT and TP plus CRT were better than PF plus CRT in improving the OS of patients with locoregionally advanced NPC. Further comparisons of TPF or TP with the GP regimen by future studies are needed to identify the optimal treatment strategy for NPC patients with locoregionally advanced disease.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Sun Yat-sen University Cancer Center. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.