High Weight Loss during Radiation Treatment Changes the Prognosis in Under-/Normal Weight Nasopharyngeal Carcinoma Patients for the Worse: A Retrospective Analysis of 2433 Cases

Background Although weight loss is common in nasopharyngeal carcinoma (NPC) patients receiving radiotherapy, the prognostic influence of weight loss and its impact modified by body mass index (BMI) are still unclear. Methods 2433 NPC patients receiving radical radiotherapy at Sun Yat-sen University Cancer Center from November, 2000 to December, 2004 were enrolled. Weight change during radiation treatment was categorized into high weight loss (HWL) and low weight loss (LWL). The associations of HWL with overall survival (OS) and disease-specific survival (DSS) were analyzed by Cox regression. Results Among underweight patients, HWL was independently associated with poor OS (hazard ratio [HR], 2.06; 95% CI 1.36–3.11) and DSS (HR, 2.27; 95% CI 1.38–3.73), as compared with LWL, after adjusting for covariates. In normal weight patients, the impact of HWL on OS (HR, 1.47; 95% CI 1.19–1.80) and DSS (HR, 1.59; 95% CI 1.24–2.03) was moderate. Among overweight/obese patients, no significant association between HWL and OS (HR, 1.22; 95% CI 0.95–1.55), or DSS (HR, 1.23; 95% CI 0.93–1.64) was found. Conclusion Except for overweight/obese patients, high weight loss during radiation treatment was independently associated with poor survival in NPC. This impact was more prominent in the underweight patient group.


Introduction
Nasopharyngeal carcinoma (NPC) is an endemic head and neck epithelial malignancy. Approximately 60% patients with nasopharyngeal carcinoma present stage III or IV disease at initial diagnosis [1,2]. Although the treatment effect of NPC has improved primarily due to the progress in diagnostic imaging, radiation techniques and chemotherapy regimens, 20-30% of the patients will die due to cancer recurrence or/and distant metastasis [3][4][5]. Identification of those high-risk patients may provide new clues in developing clinical intervention to improve their survival.
Body mass index (BMI), a commonly utilized measure for indicating nutritional status in adults, has been shown closely associated with the prognosis of NPC patients [6,7]. Various studies have focused on the striking association between low BMI at diagnosis and poor prognosis due to higher local-regional recurrence and mortality. As emphasized by Shen et al., the hazard ratio for death was 0.66 for overweight (95% CI, 0.48 to 0.90) and 0.47 (95% CI, 0.23 to 0.97) for obese patients comparing to the baseline of normal weight or underweight patients [7].
In contrast to the considerable amount of research on BMI, the impact of body weight loss on NPC recurrence and death has not been addressed. More than 60% of NPC patients receiving curative-intent radiotherapy suffered from a weight loss greater than 5% during the treatment [8,9]. Advanced tumors, concurrent chemotherapy, and high BMI are important predisposing factors of weight loss, and it is reported that a critical weight loss (.5%) during the radiation was associated with poorer treatment tolerance and worse prognosis in head and neck cancer (HNC) patients [10,11].
Is weight loss also an unfavorable prognostic factor for NPC? Since BMI affects both weight loss and survival among NPC patients [9], analysis combining these two weight-related factors may be required. Therefore, in this study, we investigated the prognostic influence of weight loss and its corresponding potential effect-modification by BMI.

Patients
The medical records of 2820 newly diagnosed NPC patients without distant metastasis in Sun Yat-sen University Cancer Center (SYSUCC) from November, 2000 to December, 2004 were reviewed. Sun Yat-sen University Cancer Center Hospital Ethics Committee approved this study (No. YP201012). This was a retrospective analysis of routine data and therefore we requested and were granted a waiver of individual informed consent from the ethics committee. The data were collected by trained SYSUCC interviewers and analyzed anonymously. All patients received radical radiotherapy and completed the prescribed course of treatment. Exclusion criteria are any of the following: (i) missing weight measurement at baseline and/or at the end of radiotherapy (200); (ii) follow-up period less than 5 years (178); (iii) age less than 18 years old (9). A total of 2433 patients were enrolled. All patients received routinely nasopharyngeal CT or MRI examination before treatment and were staged according to the sixth edition of UICC staging system [12].

Measurement and Grouping
Pre-radiation treatment (pre-RT) weight was measured within 7 days before radiotherapy (RT), and post-radiation treatment (post-RT) weight was measured within 7 days after completion of RT. BMI was defined as pre-RT weight (kg) divided by the square of height (meter) and was categorized according to the WHO recommendations for Asian population [13]; because the proportion of obese patients was relatively small (6.5%), we merged overweight and obese patients and obtained three BMI groups: ,18.5 kg/m 2 , underweight, UW; 18.5-,23.0 kg/m 2 , normal weight, NW; $23.0 kg/m 2 , overweight/obese, OW ( Figure 1).
Weight loss (DW) was calculated as the relative percent of weight changes between weight measurement before and after the radiation treatment, A number of studies suggest that a greater than 5% body weight loss during radiation for HNC patients indicates malnutrition and is clinically meaningful [9,14,15]. DW was accordingly categorized into high weight loss (HWL; $5%) and low weight loss (LWL; ,5%).

Treatment
Radiotherapy included two-dimensional conformal radiotherapy (2D-CRT), three-dimensional conformal radiotherapy (3D-CRT), and intensity-modulated radiotherapy (IMRT). Opposing lateral facial-cervical fields were used in the 2D-CRT to cover the nasopharynx and upper cervical lymphatic drainage region, with 1 lower anterior cervical field to cover the lower cervical region. After 36 to 40 Gy, opposing lateral preauricular fields were used for the primary region, and anterior split neck fields were used for the cervical region. The primary tumor was irradiated to a dose of 60 to 78 Gy. The dose of prophylactic irradiation for patients with 2D-CRT was 50-54 Gy to the prophylactic areas. For 3D-CRT, the total prescribed dose was 66-72 Gy to the gross tumor volume of nasopharynx (GTVnx), 60 to 70 Gy to the region involved by the metastatic lymph nodes (GTVnd), 60 Gy to high risk microscopic (CTV1, the GTVnx and an additional 5-to 10-mm margin), and 50-54 Gy to the low risk microscopic (CTV2). For IMRT, the target definition and delineation were the same as for 3D-CRT. The prescription dose was 68 Gy to the GTVnx, 60 to 64 Gy to the GTVnd of neck, 60 Gy to the CTV1, and 54 Gy to CTV2.
Chemotherapy included induction chemotherapy, concomitant chemotherapy, and adjuvant chemotherapy. The induction or adjuvant chemotherapy regimen was fluorouracil (5-Fu) plus cisplatin. The concurrent chemotherapy regimens were either 5-Fu plus cisplatin or cisplatin alone; the 5-Fu plus cisplatin regimen was 70 to 100 mg/m 2 of cisplatin on day 1 plus 500 to 750 mg/m 2 5-Fu on day 2 till day 5 every 3-4 weeks, for 2 to 3 cycles, and the cisplatin regimen was 30 to 40 mg/m 2 of cisplatin every week, for 6 to 7 cycles.

Follow-up and End Points
After completion of RT, patients were followed up every 3 months for the first 3 years by clinic visits, telephone, letter and so on. The intervals gradually increased to 6-12 months after 3 years. The data of follow-up was last reviewed on February 2011.
The primary outcomes were overall survival (OS) and diseasespecific survival (DSS). Secondary outcomes were local-regional recurrence free survival (LRFS) and distant metastasis free survival (DMFS). OS was defined as time from diagnosis to death by any causes. DSS was defined as the time from the start of RT to death caused by disease progression or treatment-related complications. LRFS was defined as time from the start of RT to the first occurrence of locoregional failure or death from the primary cancer without a documented site of recurrence or metastasis.

Statistical Analysis
Wilcoxon rank sum and Chi-square test were used to compare ordinal and categorical variables between groups, respectively. Rates of DSS, OS, LRFS, and DMFS were estimated by means of the Kaplan-Meier method and were compared between the subgroups with the use of the log-rank test. Cox regression model including the two main effect parameters (weight loss and BMI; in categorical form) and their interaction effect parameters was utilized to test the interaction effect between weight loss status and BMI levels. The multiple-adjusted Cox regression model was used to evaluate the survival of 263 weight-based subgroups divided by weight loss status (HWL vs LWL) and BMI levels (UW vs NW vs OW), with the covariates including age, sex, UICC T stage, UICC N stage, and treatment (with or without chemotherapy) [16,17]; the 263 subgroups were as follows: NW+LWL (group 1, reference), NW+HWL (group 2), UW+LWL (group 3), UW+HWL (group 4), OW+LWL (group 5), and OW+HWL (group 6). Stratified multiple-adjusted Cox model by BMI levels was further conducted to evaluate the prognostic significance of weight loss status in each BMI group. A P value ,0.05 was considered significant. Statistical analysis was performed using SPSS 20.0 software.

Characteristics of the Population
Listed in Table 1 were baseline characteristics of patients. The median age was 46 years (range, 18-78 years). Among all the patients, 2166 (89.0%) had undifferentiated non-keratinizing carcinoma, 203 (8.3%) had differentiated non-keratinizing carcinoma and 64 (2.6%) had other types; 983 (40.4%) had early-stage disease (stage I/II), and 1450 (59.6%) had advanced disease (stage III/IV). All patients received definitive-intent radiotherapy during For the entire population, the mean BMI was 22.60 kg/m 2 . A total of 212 patients (8.7%) were underweight, 1157 (47.6%) were normal weight, and 1064 (43.7%) were overweight/obese. During radiation treatment, the mean weight loss was 4.33 kg, and 53.6% (1303) patients had high weight loss (HWL; DW$5%). The proportion of HWL was higher in patients with advanced T stage, N stage, high BMI level and who received chemotherapy. No significant differences in age and sex were found across the weight loss subgroups (Table 1).
In unadjusted analysis, underweight patients had significantly lower 5-year OS rate compared to normal weight (65% vs 73%, P = 0.004) and overweight/obese patients (65% vs 78%, P,0.001); HWL patients presented with significantly lower 5-year OS rate compared to LWL patients (P,0.001). Age, sex, UICC T stage, UICC N stage, and treatment also significantly influenced OS (P,0.001 for all). Results on 5-year DSS and LRFS were similar to that of OS, while no difference in DMFS was found across the weight loss groups ( Table 2).
The BMI6Weight loss interaction terms were tested for OS, and a borderline significance was identified for the differential impact of weight loss between underweight and overweight/obese patients (normal weight6HWL vs underweight6LWL, P = 0.239; overweight/obese6HWL vs underweight6LWL, P = 0.054; Table  S1). Therefore, we firstly divided the whole population into 263 weight-based subgroups according to BMI and weight loss. OS, DSS and LRFS were identified as significant in unadjusted analysis among the six subgroups, with joint P value ,0.001 for all ( Figure 2; Figure S1-S2); yet no significant difference was found for DMFS (P = 0.089  (Table 3).
To assess the prognostic impact of weight loss for patients with different BMI levels, further stratified analysis was conducted (  (Figure 3).

Discussion
BMI is one of the criteria commonly used in the assessment of nutritional status, whose prognostic impact has been extensively studied among NPC patients [6,7]. However, percentage weight loss, a sensitive and specific tool in assessment of recently developed malnutrition [18,19], has been shown correlated with poor prognosis in head and neck cancer patients [11,20], and should be taken into account in NPC prognosis.
Currently, no data specifically focused on the prognostic impact of weight loss among NPC patients was available. In our series, it's  intriguing to find a potential differential impact of weight loss between underweight and overweight/obese patients (P = 0.054 The reason underlying the differential prognostic impact of weight loss on NPC patients of different BMI levels may mainly involve malnutrition. During the radiation treatment, underweight patients were more susceptible to severe side effects, such as mucositis, dysphagia, etc., than normal weight patients, which may affect their food intake and lead to body wasting [21][22][23]. Further high weight loss during the treatment may serve as a prominent indicator of malnutrition [20], which is known to be correlated with compromised immunity [23,24], unplanned treatment breaks and poor treatment outcomes in head and neck cancer [25][26][27]; by contrast, low weight loss in underweight patients may represent a well-tolerant nutritional status and is therefore associated with better prognosis [7,28]. The unfavorable impact of weight loss was moderate in normal weight patients, and even weaker in overweight/obese patients; some studies reported that NPC patients with normal or higher baseline BMI may be more resistance to malnutrition [7,29,30], and this can partly explain our results. Apart from the impact of treatment, factors like advanced tumors, severe comorbidities and even insomnia can also contribute to body weight loss of NPC patients during radiation treatment [9]. Patients with advanced tumor are more likely to develop cancer cachexia syndrome, which is characterized by nutritionally irreversible weight loss and will greatly increase the mortality rate [31]. Comorbidities like diabetes, chronic obstructive pulmonary disease, and cardiovascular diseases will also worsen the prognosis of NPC patients and, on the other hand, exacerbate body wasting [32,33]. A recent study of 1001 newly diagnosed NPC patients showed that 15.5% elderly NPC patients ($70 years) presented with moderate to severe comorbidities (ACE-27 score$2) before radiation treatment; these patients had lower OS (HR, 2.63, 95% CI 1.45-4.76) than patients had none or mild comorbidities [33]. However, in our population, no patients presented with cachexia before treatment, and the proportion of elderly patients was rather small (16pts, 0.8%). Therefore, this explanation may not be the main consideration.
Upon 263 grouping based on assortment of BMI and weight loss, we found UW+HWL patient group had lowest OS and DSS as compared with the other five groups, while e.g. the difference in survival rate between UW+LWL and NW+LWL patients was small ( Figure 2). These data suggested that appropriate intervention towards weight loss, or malnutrition might be beneficial to underweight patients. For the past decade, there has been ample evidence that adequate nutrition support (NS) before and during radiotherapy can decrease the impact of side effect of treatment, minimize weight loss, and improve outcome for head and neck cancer patients [34,35]. Weight loss of more than 1-2% per week, or 5% in less than a month, should prompt further patient assessment, nutritional counseling, and more aggressive interventions [36]. However, Rabinovitch et al. analyzed the data of 1073 advanced HNC patients treated with definitive radiotherapy and found that patients with nutrition support during treatment had better 5-year locoregional control and 5-year overall survival rate as compared with patients receiving pre-treatment nutrition support, and that the pre-treatment nutritional support group even had poorer survival as compared with those with no nutritional support at all [37]. Therefore, how and when to adopt these nutritional interventions in NPC await further study.
Our study has several limitations. First, it is a retrospective study. Second, the proportion of obese patients in this study was relatively small (6.5%), which might impede our interpretation of the prognostic value of HWL in this group; therefore, we merged the overweight and obese group. Third, the modes of chemotherapy applied varied, which might have a confounding effect. For these reasons, we need to validate our finding in a multiinstitutional prospective study in the future. Conclusion Underweight NPC patients with high weight loss during radiation had worst survival compared with other subgroups assorted by BMI levels and weight loss. High weight loss is independently associated with increased mortality among underweight and normal weight, but not overweight/obese, NPC patients. Figure S1 DSS for NPC patients after radical radiotherapy in 263 grouping by BMI levels and weight loss status.

Supporting Information
(TIF) Figure S2 LRFS for NPC patients after radical radiotherapy in 263 grouping by BMI levels and weight loss status. (TIF)