The Tumour Response to Induction Chemotherapy has Prognostic Value for Long-Term Survival Outcomes after Intensity-Modulated Radiation Therapy in Nasopharyngeal Carcinoma

The prognostic value of the tumour response to induction chemotherapy (IC) for long-term survival outcomes after intensity-modulated radiation therapy in nasopharyngeal carcinoma (NPC) remains unknown. We retrospectively reviewed 1811 consecutive patients with newly diagnosed NPC treated using IMRT, and 399 eligible patients with pre- and post-induction chemotherapy magnetic resonance images were recruited. The clinicopathological features of patients with different tumour responses were compared using the Chi-square test or Fisher’s exact test. Prognostic value was assessed using a multivariate Cox proportional hazards model. After IC, 101/399 (25.3%) patients had a complete tumour response overall (CR), 262 (65.7%) had a partial response (PR) and 36 (9.0%) had stable disease (SD). The 4-year disease-free survival (DFS), overall survival (OS), distant metastasis-free survival (DMFS) and locoregional relapse-free survival (LRRFS) rates for CR vs. PR vs. SD were 90.0% vs. 79.0% vs. 58.2% (CR vs. PR: P1 = 0.007; CR vs. SD: P2 < 0.001; PR vs. SD: P3 = 0.004), 95.7% vs. 88.7% vs. 70.2% (P1 = 0.017, P2 < 0.001, P3 = 0.005), 92.0% vs. 87.4% vs. 74.3% (P1 = 0.162, P2 = 0.005, P3 = 0.029) and 95.9% vs. 88.8% vs. 81.8% (P1 = 0.024, P2 = 0.006, P3 = 0.268), respectively. Multivariate analysis identified that the tumour response to IC was an independent prognostic factor for DFS, OS and LRRFS.

However, despite these negative outcomes, IC may have potential clinical value. Previous studies reported that the response to chemotherapy correlate with clinical outcome [13][14][15] . Most recently, Liu et al. revealed that the unsatisfactory tumour response after induction chemotherapy could predict poor prognosis for patients with advanced-stage NPC 16 . However, the sample was relative small. Moreover, the prognostic difference was only discussed between stable/progressive disease (SD/PD) and complete/partial response (CR/PR) groups, and was not investigated between CR and PR groups. Therefore, on the basis of this premise, we conducted a retrospective study to further analyse the prognostic value of different tumour responses to induction chemotherapy in NPC patients who received intensity-modulated radiation therapy (IMRT).

Materials and Methods
Patient Selection. Of the 1811 patients with newly diagnosed non-metastatic NPC treated between November 2009 and February 2012 at Sun Yat-sen University Cancer Center, and the 399 patients for whom both pre-and post-induction chemotherapy magnetic resonance (MR) images of the nasopharynx and cervical region were available were retrospectively analysed. This study was conducted in compliance with the institutional policy regarding the protection of patients' private information and approved by the Research Ethics Committee of Sun Yat-sen University Cancer Center. All the methods were carried out in accordance with the approved guidelines of Sun Yat-sen University Cancer Center. Written informed consent was obtained from all patients prior to therapy. Clinical Staging. Routine staging workup included a complete history, clinical examinations of the head and neck, direct fibre-optic nasopharyngoscopy, magnetic resonance imaging (MRI) of skull base and whole neck, chest radiography, whole-body bone scan, abdominal sonography, and positron emission tomography-CT if clinically-indicated. Immunoglobulin A antibodies against EBV viral caspid antigen (VCA-IgA) and Epstein Barr virus early antigen (EA-IgA) were quantified. All patients underwent dental evaluations before RT.
Patients were restaged according to the 7 th edition of the International Union against Cancer/American Joint Committee on Cancer (UICC/AJCC) staging system 17 . All MRI and clinical records were reviewed to minimize heterogeneity in restaging. Two radiologists evaluated all scans separately, and disagreements were resolved by consensus.
Imaging Protocol. All patients underwent MRI of the region from the suprasellar cistern to the inferior margin at the sternal end of the clavicle using a head-and-neck coil with a 3 Tesla system (Trio Tim; Siemens, Erlangen, Germany). T1-weighted fast spin-echo images in the axial, coronal and sagittal planes (repetition time [TR]/echo time [TE] = 650 ms/9 ms), T2-weighted fast spin-echo MR images in the axial plane (TR/TE = 2470 ms/90 ms) and a spin-echo echo-planar DWI sequence (matrix = 192 × 192; TR/TE, 5100 ms/96 ms; b-values = 0 and 1000 s/mm 2 ; three signal averages) were obtained before contrast injection. After intravenous administration of gadopentetate dimeglumine (0.1 mmol/kg body weight; Magnevist, Schering, Berlin, Germany), axial and sagittal T1-weighted spin-echo and coronal T1-weighted fat-suppressed spin-echo sequences were performed (same parameters as without contrast; 5 mm-thick sections with 1 mm interslice gap for axial plane, 6 mm-thick sections with 1 mm interslice gap for coronal and sagittal planes; matrix size, 512 × 512).
Diagnostic Criteria for Tumour Response. Tumour response was assessed using MRI images after completion of IC based on Response Evaluation Criteria in Solid Tumours (RECIST) 18 . Complete response (CR) was defined as disappearance of target lesion (short radius of target neck pathological lymph nodes < 10 mm, short radius of retropharyngeal lymph nodes < 5 mm). Partial response (PR) was defined as a reduction of at least 30% in the sum of longest diameter of target lesions (using baseline sum longest diameter as reference). Progressive disease (PD) was defined as an increase of at least 20% in the sum of longest diameter of target lesions (using smallest sum longest diameter since treatment started or appearance of one or more new lesions as reference). Stable disease (SD) was defined as neither sufficient shrinkage to qualify for PR nor sufficient increase to qualify for PD (using smallest sum longest diameter since treatment started as reference). Overall tumor response, local tumor response and regional tumor response was defined as per Table 1. Chemotherapy. According to institutional guidelines, we recommended RT alone for stage I, concurrent chemoradiotherapy (CCRT) for stage II, and CCRT + /− IC/adjuvant chemotherapy (ACT) for stage III to IVA-B. IC consisted of cisplatin (80 mg/m 2 ) with 5-fluorouracil (1000 mg/m 2 ) (PF), cisplatin (75 mg/m 2 ) with docetaxel (75 mg/m 2 ) (TP), or cisplatin (60 mg/m 2 ) with 5-fluorouracil (600 mg/m 2 ) and docetaxel (60 mg/m 2 ) (TPF) every three weeks for two or three or more cycles. Concurrent chemotherapy was cisplatin weekly (30-40 mg/m 2 ) or on weeks 1, 4 and 7 (80-100 mg/m 2 ) of radiotherapy.

Follow-Up and Statistical Analysis.
Follow-up was measured from first day of therapy to last examination or death. Patients were followed by MRI and plasma EBV DNA at least every 3 months during first 2 years, then every 6 months thereafter (or until death). The end points (time to first defining event) were disease-free survival (DFS), overall survival (OS), distant metastasis-free survival (DMFS), locoregional relapse-free survival (LRRFS), local relapse-free survival (LRFS) and regional relapse-free survival (RRFS).
The Chi-square test or Fisher's exact test were used to compare categorical variables and treatment failure patterns between the CR, PR and SD groups. Life-table estimation was performed using the Kaplan-Meier method and log-rank test. The multivariate Cox proportional hazards model was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs); age, gender, smoking, drinking, pathological type, T category, N category, concurrent chemotherapy, overall response were included as variables. All tests were two-sided; P < 0.05 was considered significant. Stata Statistical Package 12 (StataCorp LP, College Station, TX, USA) was used for all analyses.

Results
Patient Characteristics. For the 399 patients, the male (n = 308)-to-female (n = 91) ratio was 3.4:1. The median age for the whole cohort was 45 years (range, 14-76 years). The baseline characteristics of patients were list in Table 2. Sixty-eight (17.0%) patients did not received concurrent chemotherapy with 16 (23.5%) in CR group, 45 (66.2%) in PR group and 7 (10.3%) in SD group, and no significant difference was found between the three groups with respect to the omission of concurrent chemotherapy.
With regards to overall tumour response after induction chemotherapy, patients with SD had higher rates of local failure (11.1% vs. 2.0%, P = 0.022) and distant failure (25% vs. 7.9%, P = 0.008) after IMRT than patients with a CR ( Table 3). The distant failure rates of patients with overall PR and SD after induction chemotherapy were almost significantly different (13.0% vs. 25%, P = 0.054).
Patients with local SD after induction chemotherapy had higher rates of local failure after IMRT compared to patients with a local CR (10.8% vs. 2.2%, P = 0.008). The local failure rates of the local PR and CR groups were almost significantly different (6.7% vs. 2.2%, P = 0.057). Patients with a regional CR after induction chemotherapy had a lower rate of distant failure after IMRT than patients with a regional PR (7.2% vs. 14.5%, P = 0.033) and regional SD (7.2% vs. 25.5%, P < 0.001). Additionally, patients with a regional CR had a lower rate of regional failure after IMRT than patients with a regional PR (2.4% vs. 8.4%, P = 0.015).
Univariate Analysis. The  Multivariate Analysis. Multivariate analysis was performed to adjust for potential prognostic factors, including age, gender, smoking, drinking, pathological type, T category, N category and concurrent chemotherapy. Consistent with the results of univariate analysis, the overall tumour response after induction chemotherapy was an independent prognostic factor for DFS, OS and LRRFS (Table 4).

Discussion
The outcomes of this current study revealed that satisfactory tumour response to IC was associated with significantly improved DFS, OS and LRRFS for patients with NPC in the era of IMRT. The reported rates of overall CR, PR, and SD after induction chemotherapy range from 8% to 27%, 55% to 64%, and 11% to 17.6%, respectively 7,8,19,20 , and the overall tumour responses of the patients in this study are consistent with these results. Liu et al. 16 proved that patients with SD had poorer prognosis compared with patients with CR/PR. However, the prognostic difference was not clear between CR and PR groups. In this larger cohort study, we provided more detailed results about the prognostic value of different tumour response to induction chemotherapy.
Of note, the SD group had a higher percentage of smoking compared with that of CR or PR group, and this may influenced the survival outcomes since smoking was previously reported as a prognostic factor in advanced NPC 21 . Moreover, more patients in the SD group had T3-4 stage disease compared with that of patients in CR or      PR group. This unbalanced distribution of host and tumour factors between the three groups may influence the significant results of univariate analysis for DMFS since multivariate analysis did not identify tumour response as an independent prognostic factor for DMFS. On multivariate analysis of LRRFS, patients with CR had better LRFFS compared with patients with PR. However, this significant difference did not exist between CR/PR and SD groups. Reasonable explanation was that only 6 (16.7%) patients in the SD group experienced locoregional failure. Therefore, the small sample resulted in a low power to detect significant differences between the CR/PR and SD groups. In addition, 68 (17.0%) patients did not received concurrent chemotherapy after induction chemotherapy. The main reason for the omission of concurrent chemotherapy was that patients had poor status and could not tolerate concurrent chemotherapy. NPC is highly chemosensitive and radiosensitive. Although most patients with advanced NPC respond well to chemotherapy, recurrence of distant metastases is the major cause of treatment failure and has a poor prognosis 22 . Patients with SD may develop resistance to anticancer drugs which is a major factor that leads to this pattern of failure 23 . A number of molecular mechanisms have been linked to chemoresistance and poor prognosis. For example, positive tumour expression of multidrug resistance 1 (MDR1) is associated with poor overall survival in patients with recurrent or metastatic NPC receiving chemotherapy 23 . Multidrug resistance-associated protein-1 (MRP1) expression was identified as an independent prognostic factor for 5-year overall survival and disease-free survival 24 . Moreover, upregulation of the drug-resistance gene annexin I (ANX-I) is associated with a poorer prognosis in NPC, breast cancer and gastric cancer 25 , and radio-and chemo-sensitivity could be induced by suppressing expression of Jab1/CSN5 26 . These molecular mechanisms may have potential as therapeutic targets to improve the prognosis of patients with distant metastasis.
Despite the outcome that unsatisfactory (PR or SD) tumour response after IC indicated poor prognosis, we should pay attention to the factors which may affect the result. Firstly, using the RECIST standard 18 , it is difficult to define the local response in advanced-stage NPC with skull base invasion. The imaging changes in the skull base are not as obvious as changes in the nasopharynx tumour or neck lymph nodes, which may lead to inaccurate evaluation of local tumour response. More appropriate evaluation criteria for defining the tumour responses in NPC should be established. Secondly, the patients in this retrospective study received different induction chemotherapy regimens and varying numbers of cycles: 60.4% (61/101) of patients in CR group received equal or more than 3 cycles of induction chemotherapy, comparing with 45.0% (118/262) in PR group and 25.0% (9/36) in SD group. This should be reasonable because clinicians would quit IC and turn to RT immediately if the evaluation after two cycles IC was SD. Therefore, patients in SD group would receive less cycles of IC. The heterogeneity of IC regimens should be small and would not influence the results because most patients in the study received the standard treatment regimens according the NCCN guidelines.
The current study showed that patients with unsatisfactory tumour responses after IC have poorer prognosis compared with that of patients with satisfactory tumour responses. Therefore, we should pay more attention to patients with SD after induction chemotherapy in clinical practice. More intensive treatment regimens like higher radiation dose and adjuvant chemotherapy should be considered. Moreover, regular follow-up should also be provided to detect early recurrence and reduce the potential risk of distant failure.
This study is limited by its retrospective nature and fact the follow-up time may have been insufficient, though we selected DFS as the major endpoint to address this shortcoming. Moreover, many other prognostic factors including pre-treatment plasma Epstein-Barr virus DNA load [27][28][29] , primary tumour volume 30,31 and the pre-treatment serum lactate dehydrogenase (LDH) level 32 were not assessed in this study. Of note, only 2 (0.5%) patients had type I NPC, which is characteristic of endemic NPC areas and hence these results might not apply to other non-endemic regions. Further prospective studies of larger cohorts are warranted to confirm the prognostic value of the tumour response to induction chemotherapy in patients with NPC treated with IMRT.  Table 4. Multivariate analysis of prognostic factors associated with clinical outcomes based on the overall tumour responses after induction chemotherapy. Abbreviations: DFS = disease-free survival; OS = overall survival; DMFS = distant metastases-free survival; LRRFS = locoregional relapse-free survival; HR = hazard ratio; CI = confidence interval; CR = complete response; PR = partial response; SD = stable disease. Conclusions Tumour response to induction chemotherapy is an independent prognostic factor for patients with NPC receiving IMRT. Assessing tumour response after induction chemotherapy may assist prognostication and refine the treatment of high-risk patients with NPC. The outcomes in this current study need to be confirmed by prospective studies with large cohorts.