Establishment and Validation of Nomogram Based on Combination of Pretreatment C-Reactive Protein/Albumin Ratio–EBV DNA Grade in Nasopharyngeal Carcinoma Patients Who Received Concurrent Chemoradiotherapy

Background A higher ratio of pretreatment C-reactive protein/albumin ratio (CAR) is associated with poor prognosis in nasopharyngeal carcinoma (NPC), and Epstein–Barr virus (EBV) DNA level is known to not only participate in the occurrence of nasopharyngeal carcinoma but also affect the development and prognosis of the disease. Herein, we proposed that a combination of both these markers could improve the predictive prognostic ability. Methods In all, 842 NPC patients who received concurrent chemoradiotherapy (CCRT) were entered in this study. We collected all patients’ blood samples and EBV DNA copy numbers within one week before any treatment. Receiver operating characteristic (ROC) curve was used to determine the optimal cut-off. We employed the Kaplan–Meier method for survival analyses and the univariate and multivariate analyses (Cox proportional hazards regression model) for statistical analysis. A nomogram was constructed based on multivariate analyses results of the validation set. The model was internally validated using 1000 bootstrap samples to avoid overfitting. Another validation of 10-fold cross-validation was also applied. Calibration curves and concordance index (C-index) were calculated to determine predictive and discriminatory capacity. Results In the whole cohort, we observed that higher CAR, EBV DNA level, and CAR-EBV DNA (C-E) grade were associated with shorter overall survival (OS) and distant metastasis-free survival (DMFS) (all P<0.05). In univariate and multivariate analyses, C-E grade was an independent prognostic factor (all P<0.05). In the training set, we gained the similar results with the whole set. According to multivariate analyses of the training set, we constructed a nomogram. The results of bootstrap samples and 10-fold cross-validation showed favorable predictive efficacy. And calibration curves of the model provided credibility to its predictive capability. Conclusion C-E grade was confirmed as an independent prognostic predictor in patients with NPC who received CCRT. Higher level of pretreatment C-E grade could signify a higher risk of metastasis and shorter OS. The prognostic nomogram based on C-E grade was dependable in nasopharyngeal carcinoma patients.


INTRODUCTION
Nasopharyngeal carcinoma (NPC) is a relatively rare type of head and neck cancer, and is characterized by its unique ethnic and geographic distributions. Southern China has one of the highest incidence of NPC (1), at 20-30 per 100,000 people (2). In addition, the etiology of NPC is distinct from other carcinomas and is related with the Epstein-Barr virus (EBV) (3). During the last several decades, patient prognosis has improved dramatically owing to advances in disease management (4). The widespread application of intensity-modulated radiotherapy and optimization of chemotherapy strategies have contributed to improved survival with reduced toxicities (5,6). Concurrent chemoradiotherapy (CCRT) had proven to be superior in many clinical trials and is recommended for patients with locally advanced disease (7)(8)(9)(10)(11). Although the trends show that the incidence has declined gradually and the mortality reduced substantially, there are still a considerable number of patients with NPC. It is especially crucial to make an accurate assessment to form a suitable treatment plan. Thus far, the gold standard for evaluating the prognosis of NPC is the Union Internationale Contre le Cancer/American Joint Cancer Committee (UICC/ AJCC) TNM staging system (12). However, clinicians have noticed that the prognosis of different patients with the same TNM staging is quite distinct. One explanation for this is that the TNM staging system does not account for patients' personal conditions. Therefore, finding an individual, integrated, and robust indicator is very essential.
Inflammation is known to be associated with cancer (13)(14)(15), as it plays an important role in carcinogenesis and tumor progression (16)(17)(18)(19). Researchers have shown that inflammation-based indicators are relevant to prognosis of various cancers such as breast cancer (20), gastric cancer (21), rectal and colorectal cancer (22), and lung cancer (23). With respect to NPC, some studies have verified that the neutrophilto-lymphocyte ratio (NLR), (24), lymphocyte-to-monocyte ratio (LMR) (25), EBV DNA level (26), and C-reactive protein/ albumin ratio (CAR) (27) are prognostic biomarkers. Creactive protein is an accurate protein generated by liver under systemic inflammation and serum albumin is always seen as signal of nutrition status, therefore, CAR, as a high individual marker, could precisely reflect inflammatory nutritional state of objects. Besides, EBV plays an important role in the etiology of NPC. However, to our best knowledge, no study yet has focused on both CAR and EBV DNA levels. Therefore, CAR-EBV DNA (C-E) grade, which combines both the indices, can reflect patient prognosis from two aspects: etiology and personal condition.
At present, no published studies have reported a predictive marker that integrated CAR and EBV DNA level for overall survival (OS) in NPC patients. In this study, we aimed to establish and explore the prognostic ability of CAR-EBV DNA (C-E) grade in patients who received CCRT. We also demonstrated the relationship between CAR-EBV DNA (C-E) grade and the clinical characteristics of NPC. A nomogram based on CAR-EBV DNA (C-E) grade was built for an authentic prognostic prediction in NPC patients who received CCRT in order to assist clinical work.

Patients and Sample Selection
In all, 842 NPC patients were recruited at the Sun Yat-sen University Cancer Center (SYSUCC; Guangzhou, China) from December 2009 to December 2014. The inclusion criteria were as follows: 1) histologic diagnosis of NPC; 2) patients who received CRRT; and 3) clinical stage II-IVa based on the 8th AJCC staging system; The exclusion criteria were: 1) synchronal malignancies; 2) lack of pretreatment peripheral blood examination and EBV DNA copy number; and 3) insufficient follow-up data. All patients involved in this study provided written informed consent.
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.
All patients underwent pretreatment evaluation that included complete patient history, physical examination, hematology and biochemistry profiles, magnetic resonance imaging (MRI) of the neck and nasopharynx, chest radiography, abdominal sonography, and a whole-body bone scan using single photonemission computed tomography. All patients' data was from SYSUCC, and blood samples and EBV DNA copy number were obtained one week before initiation of any treatment.

Treatment
All selected patients were treated according to the SYSUCC treatment protocol for NPC. Generally, all patients received intensity-modulated radiotherapy (IMRT) at a total dose of 68-70 Gy to the primary lesions, while metastatic lymph node areas received 2 Gy/day, 5 times per week. The regional lymphatic drainage area was irradiated with a total dose of 50-54 Gy. All patients also underwent simultaneous chemotherapy during IMRT with cisplatin, carboplatin, or nedaplatin, using either a weekly or triweekly regimen.
Construction and Grades of CAR-EBV DNA CAR was calculated by dividing the C-reactive protein and albumin levels. The cut-off of CAR and EBV DNA level were obtained according to the Youden index based on receiver operating characteristic (ROC) curves. Based on the cut-off values, patients with both lower CAR and EBV DNA levels were assigned grade 0, those with either increased CAR or increased EBV DNA level were assigned grade 1; and those with both higher CAR and EBV DNA level were assigned grade 2.

Outcome and Follow-Up
Overall survival time was defined as the duration from the date of diagnosis to death or to the last follow-up. Recurrence-free survival (RFS) was defined as the time from the date of diagnosis to the date of first recurrence, death from any cause, or last follow-up. Distant metastasis-free survival (DMFS) time was defined as the period from the date of diagnosis to metastasis, death from any cause, or last follow-up. All patients were followedup by outpatient examination or telephonic interviews.

Statistical Analysis
All analyses were conducted using the SPSS software version 25.0 (IBM Corp., Armonk, NY); GraphPad Prism version 6.0 (GraphPad, La Jolla, CA); and R software (Version 5.1-0, Vanderbilt University, Nashville, TN). SPSS software was used to calculate the ROC curve to determine the cut-off values for CAR and EBV DNA. The relationships between CAR-EBV DNA (C-E) grade and other key clinicopathological characteristics were analyzed by chi-square test or Fisher's exact test. Univariate and multivariate analyses were performed using the Cox proportional hazards model. Two-tailed P values <0.05 were considered to indicate statistical significance. Survival curves were plotted by the Kaplan-Meier method, and significance was determined by the log-rank test. A nomogram was generated with endpoints of 3-year and 5-year OS and DMFS, using the R software "rms" package. The concordance index (Cindex) for OS and DMFS and the calibration plots were obtained. 1000 bootstrap samples and 10-fold cross-validation was also applied to avoid overfitting.

Patients' Characteristics and Relationships Between C-E Grade With Clinicopathological Features
In all, 842 NPC patients were enrolled in this study. The relationship between clinicopathological characteristics and CAR-EBV DNA of the whole cohort is presented in Table 1.
The relationship between features with CAR and EBV DNA is given in the Supplementary Materials. Briefly, the median age of The whole cohort was randomly divided into a training set and validation set (ratio: 9:1, respectively) ( Table 2). With respect to C-E grade, 362 (43.0%), 337 (40.0%), and 143 (17.0%) patients were assigned to grade 0, grade 1, grade 2, respectively.

Survival Analysis of CAR, EBV DNA, and C-E Grade
According to classifications of the three indicators, the whole cohort was divided into different groups. Figure 2 shows the significant survival differences among the different groups. According to optimal cut-off and gradation, we noticed that higher CAR, EBV DNA level, and C-E grade were associated with shorter OS and DMFS (all P<0.05). The OS, DMFS, and RFS for patients in the high and low CAR groups were 46 Figures 2G-I).
Furthermore, we performed univariate and multivariate analyses for OS and DMFS. We observed that C-E grades were a significant independent prognostic factor (Tables 3 and 4, all P<0.05). There were no obvious differences in the survivals for RFS, we did not procedure univariate and multivariate analyses for RFS.
In the training set, we conducted survival analysis and univariate and multivariate analyses (see Supplementary  Materials). The results were in line with those of the whole set. For OS and DMFS, high-level groups of CAR, EBV DNA level, and C-E grade had shorter survival time than low-level groups (all P<0.05). And no significant intergroup difference was found with respect to RFS (all P>0.05). The results of univariate and multivariate analyses were consistent with the whole set.

Construction and Validation of the Nomogram
A nomogram model was built based on the results of multivariate analysis. Independent prognostic indicators were integrated into the prediction of OS and DMFS times. With regard to OS time, T stage, N stage, and C-E grade were included (Figure 3). And for DMFS time, these factors included N stage, and C-E grade (Figure 4).
The prediction ability of the final model was assessed using the C-index. For OS, the value was 0.693 (95% CI: 0.661-0.725) as well as the bias-corrected C-index, which was estimated using bootstrap with 1000 iterations and noted to be 0.695 (95% CI: 0.657, 0.733). For DMFS, the values were 0.693 (95%CI: 0.664-0.722) and 0.690 (95% CI: 0.678, 0.702), respectively. In addition, the result of internal 10-fold cross-validation for OS (C-index:

DISCUSSION
In this retrospective study, 842 NPC patients who received CCRT were included. First, we found that CAR and EBV DNA levels were independent predictors of survival in a cohort of NPC patients. By combining pretreatment CAR with EBV DNA level, we developed a novel evaluation grade and proposed a new prognostic factor-C-E grade-in NPC patients. We demonstrated that patients in different groups according to optimal cut-off had clearly distinct survival times. The univariate and multivariate analyses showed that C-E grade was an independent prognostic indicator in NPC patients.
Next, we randomly divided patients into a training set and validation set. The results of the survival plot and univariate and multivariate analyses of the training set were consistent with the whole cohort. Based on the results of the multivariate analysis, a nomogram of C-E grade was constructed. Afterward, C-index and calibration in both sets were calculated, and the values confirmed the validity of this model. In survival analysis, we noticed that the level of CAR and EBV DNA had no relationship with RFS. Actually, we are not the first research that found this trend. Several studies reported that the level of CAR was associated with OS and DMFS in NPC patients (27)(28)(29)(30). For the level of EBV DNA, there were also some similar researches. In one research, the researchers made a conclusion that EBV DNA was more useful in detecting distant metastasis than recurrence (31). Another research investigated the relationship between EBV DNA level and survival of NPC patients, the researchers also reported that EBV DNA level was highly prognostic of long-term survival and distant metastasis in NPC patients (26). Our results were consistent with these researches that patients with higher level of CAR and EBV DNA had poorer OS and DMFS. C-E grade was obtained from CAR and EBV DNA level. Therefore, the survival curves of C-E grade were congruous with two of them so that the survival analysis of it for OS and DMFS was significant, but not for RFS. However, the underlying reason is unclear. A lot of related work need to be done to explore the differences.
Inflammatory markers have been closely associated with various tumors. For instance, Kuroda et al. showed that Controlling Nutritional Status (CONUT) is useful to not only estimate the nutritional status but also predict the long-term OS in gastric cancer patients after curative resection (32).  and particularly for those who are being considered for active treatment in operable non-small-cell lung cancer (34). Several research studies have attested that inflammation takes part in development, recurrence, and metastasis of cancer. C-Reactive Protein (CRP) is one of the acute-phase proteins and is synthesized by the liver (35). As an  inflammatory marker, it has high sensitivity but low specificity. However, elevated CRP has been verified to relate with poor prognosis in a broad variety of cancers such as colorectal cancer, prostate cancer, and breast cancer (36). Serum albumin (ALB) is the main serum protein that can reflect the nutritional status of humans (37). Recently, many studies had found that low ALB can influence the prognosis of malignancies in different cancers such as hepatocellular carcinoma (38), colorectal cancer (39), renal cancer (40), and NPC (41). Based on the CAR value, it was clear that a high value is a risk factor for tumors. Additionally, its association with poor prognosis had been confirmed in diverse tumors (42)(43)(44)(45)(46)(47) including NPC (48,49). Epstein-Barr virus (EBV) infection is an important etiological factor in NPC and its copy number is positively correlated with tumor stage (50). On the basis of studies, EBV DNA is widely used to screen and monitor disease change (51)(52)(53). Several studies have reported that EBV DNA level can judge the prognosis of NPC (26,54,55). In this study, we combined CAR and EBV DNA level as a novel biomarker to evaluate NPC prognosis. In theory, this marker could better reflect disease change and predict survival outcome. Our results showed that our hypothesis was correct, as proved by multiple statistical analyses.
We believe that C-E grade are extraordinary and comprehensive prognostic indices of NPC with respect to individual patient condition and etiology. In clinical application, C-E grade is easily available before initiation of treatment. As a risk factor, it could assist the treating clinician with an all-round and individualized understanding of the disease.
Our study has some limitations. First, selection bias was inevitable given the retrospective nature of the analysis. Second, we did not explore the relationship between C-E grade and other inflammatory indicators such as neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), which are reportedly associated with poor prognosis in NPC. Third, we concentrated on pretreatment C-E grade and did not follow-up and update the figure. It would be better to conduct a prospective study. Last, we did not validate our nomogram externally, rather only internally. Therefore, more research is required for further validation and improvement of results. In conclusion, we proposed a new C-E grade that can be easily applied in the clinic and help clinicians to better understand the disease and formulate a treatment plan. We have shown that a high C-E grade is usually suggestive of poor prognosis. C-E grade is an independent, integrated, and personalized prognostic indicator in NPC patients. Further,

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. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
Z-YY, WX, and J-JH contributed to conception and design of the study. X-WB and C-GS organized the database. XH performed the statistical analysis. Z-YH and WW wrote the first draft of the manuscript. Z-ZH, WW, and XH wrote sections of the manuscript. All authors contributed to the article and approved the submitted version.