Introduction

Nasopharyngeal carcinoma (NPC) is a head and neck epithelial malignant tumor that is located on the surface of the nasopharynx and has distinct ethnic and regional characteristics. The incidence of NPC is much greater in southeast Asia and southern China than in the United States1. In high-prevalence areas, such as southern China, 90%–95% of NPC cases are classified as nonkeratinized undifferentiated carcinoma2. Owing to its pathological features and the radiosensitivity of NPC, radiotherapy has been a major modality in the treatment of NPC3. Concurrent chemoradiotherapy (CCRT) either with or without induction (neoadjuvant) chemotherapy is more beneficial than radiotherapy alone for the treatment of advanced NPC patients4,5,6.

There is little doubt that chemoradiotherapy improves treatment outcomes and reduces recurrence and death rates. However, there is also considerable inter-individual variability in chemoradiotherapy reactions in tumor patients receiving the same or similar treatment protocols7,8. Because TNM stage is the main criterion for predicting prognosis in NPC patients, patients with the same TNM stage still exhibit variable efficacy and toxic effects, owing to heterogeneity among the tumors and the hosts9. Xerostomia, myelosuppression, dermatitis and oral mucositis are common side effects induced by chemoradiotherapy. Myelosuppression occurs with leucopenia, neutropenia, anemia and thrombocytopenia, and sometimes is even life threatening10. In addition to the clinical and demographic factors, genetic background plays an important role in inter-individual variations in response to chemoradiotherapy11,12,13.

Protein synthesis and folding, intracellular calcium levels, and lipid and sterol synthesis are all regulated by the endoplasmic reticulum (ER)14. ER stress may be activated by perturbation of the tumor microenvironment by factors such as hypoxia, pH changes, and oxidative stress induced by both chemotherapy and radiotherapy15,16. ER stress is also involved in cancer cell proliferation, cytoprotective autophagy, apoptosis, angiogenesis, and the inflammatory response mediated by the unfolded protein response (UPR), thereby resulting in cancer development and metastases17. Functional polymorphisms in ER stress pathway related genes, especially single nucleotide polymorphisms (SNPs) in the coding regions, may affect the expression or activity of proteins. The expression level of glucose-regulated protein of molecular mass 78 (GRP78, a major ER chaperone) has prognostic value in neoadjuvant chemoradiotherapy and laparoscopic surgery for locally advanced rectal cancer16. Peng et al have found that the G/G genotype at rs2269577, compared with the homozygous C/C genotype, is significantly associated with severe gastrointestinal toxicity in advanced non-small cell lung cancer (NSCLC) patients undergoing platinum-based chemotherapy18. The SNP rs2074549 in the valosin-containing protein (VCP) gene is significantly associated with severe neutropenia in Chinese advanced NSCLC patients on platinum-based regimens19. As a molecular folding chaperone responding to ER stress, heat shock protein 90 beta family member 1 (HSP90B1) has been implicated in chemoresistance and radioresistance and often is an anticancer therapeutic target20,21. These findings suggest that the ER stress pathway may play an important role in cancer development and metastases. However, there is a lack of direct clinical evidence of the relationship between ER stress gene SNPs and chemoradiotherapy efficacy or toxic reactions, thus leading to difficulties in determining which genes are risk factors.

In this study, we speculated that SNPs in ER stress pathway genes might be predictive factors for chemoradiotherapy efficacy and toxic reactions in NPC patients and may account for the inter-patient heterogeneity in clinical efficacy and toxic reactions. Thus, we explored the association between SNPs in ER stress response-related genes, including VCP (valosin-containing protein) rs2074549, HSP90B1 rs17034943, CANX (calnexin) rs7566, HSPA5 (heat shock protein family A (Hsp70) member 5) rs430397, CALCR (calcitonin receptor) rs2528521, and XBP1 (X-box binding protein 1) rs2269577, and clinical efficacy and toxic reactions.

Materials and methods

Study population

This study was approved by the Ethics Committee of Jiangxi Cancer Hospital (China) and was registered on the official website of the Chinese Clinical Trial Registry (www.chictr.org, registration number: ChiCTR-OPC-14005257). Between April 2014 and September 2015, a total of 150 patients with histopathologically confirmed NPC were enrolled at the Radiology Department of Jiangxi Cancer Hospital, China. Only patients whose performance status (PS) scores were ≤1.0 without severe heart, lung, liver or kidney dysfunction and whose lesions were measurable according to the WTO evaluation criteria, were eligible for this study. The other inclusion criteria included no anti-tumor treatment prior to chemoradiotherapy and the completion of radical treatment. The exclusion criteria included pregnancy or lactation, active infection before treatment, history of cancer, or complication with other tumors. All patients provided signed informed consent. Basal demographic and clinical information was collected.

Treatment regimen

All patients received intensity-modulated radiotherapy (IMRT) with 6-MV photons at a dose of 2 Gy/d, 5 times weekly. The doses were prescribed on the basis of the gross tumor volume (GTV), clinical target volume (CTV), and planning target volume (PTV). The accumulated doses for nasopharyngeal primary lesions and positive cervical lymph nodes were 66–77 Gy and 54–60 Gy, respectively, and were applied to the cervical prevention area in 30–33 fractions.

All patients received concurrent chemoradiotherapy (CRT). Platinum-based chemoradiotherapy was used as induction chemotherapy and adjuvant chemotherapy.

Evaluation of treatment efficacy and toxic reactions

On the basis of the Response Evaluation Criteria In Solid Tumors (RECIST) of the World Health Organization (WHO), the short-term effects on the primary tumor and lymph node after RT were evaluated with computed tomography (CT) and magnetic resonance imaging (MRI) immediately and 3 months after RT. The RECIST definition of efficacy included complete remission (CR), partial remission (PR), stable disease (SD) and progression disease (PD). In this study, for specific comparison with the CR group, the PR, SD and PD groups were consolidated into a single non-CR group.

Acute radiation-induced toxic reactions were evaluated according to the guidelines of the Radiation Therapy Oncology Group or European Organization for Research and Treatment of Cancer (RTOG/EORTC). In this study, acute toxic reactions of xerostomia, myelosuppression, dermatitis and oral mucositis were evaluated. On the basis of the RTOG/EORTC scores of the toxic reactions, the patients were further divided into subgroups. Patients with xerostomia of grade 2 or higher (grades 2–3) were defined as the chemoradiation sensitive subgroup, whereas those with mild toxicity (grade 1) were classified as the chemoradiation insensitive subgroup. For myelosuppression, patients were subdivided into the grade 0–2 (grades 0, 1, and 2) group and the grade 3+ (grades 3 and 4) group. For dermatitis, patients were subdivided into a grade 1–2 group and a grade 3+ (grades 3 and 4) group. For oral mucositis, patients were subdivided into a grade 0–2 (grades 0, 1, and 2) group and a grade 3+ (grades 3 and 4) group.

Genotyping assay

Peripheral blood samples (3 mL) were collected for genotyping before radiotherapy and were stored at −20 °C in EDTA anticoagulant tubes. DNA was extracted using a Wizard Genomic DNA Purification Kit (Promega, Madison, USA), per the instructions. Polymorphisms in 6 genes were analyzed with a Sequenom MassARRAY system (BioMiao Biological, Beijing, China).

Statistical analysis

All data were analyzed with SPSS 19.0 software. All statistical tests were two-sided. P values <0.05 were considered to be statistically significant. Descriptive statistics were used to report the demographics and treatment characteristics. Chi-square tests were performed to evaluate the effect of clinical factors on the short-term treatment efficacy (including primary tumor efficacy and cervical lymph node efficacy) and the risk of occurrence of xerostomia, myelosuppression, dermatitis and oral mucositis. The association between polymorphisms in ER stress pathway genes and short-term efficacy as well as the risk of acute toxic reactions was evaluated by estimating the odds ratios (ORs) and 95% confidence intervals (CIs) through univariate and multivariate logistic regression analyses. Multivariate analyses were adjusted for age, sex, drinking, smoking, family history, body mass index (BMI), TNM stage and clinical stage.

Results

Population characteristics and clinical outcome

A total of 150 NPC patients with a defined pathological diagnosis of non-keratinized nasopharyngeal carcinoma (NPC) were recruited. There were 100 males and 50 females with a mean age of 49.56 years (ranging from 14–71 years). On the basis of the clinical criteria of the 7th Union for International Cancer Control (UICC) staging standard, there were no NPC patients at stage I, 12 patients at stage II, 67 patients at stage III, and 71 patients at stage IV. Their demographics and treatment characteristics are summarized in Table 1. Epstein-Barr virus (pEBV) was detectable in the plasma of 109 patients (72.7%; pEBV-positive population). However, the pEBV-positive rate after CRT decreased to 2.2%.

Table 1 Baseline demographics and clinical characteristics.
Full size table

The associations between population characteristics and short-term effects in NPC patients immediately after and 3 months after CRT are listed separately in Table S1 (supplementary materials) and Table 2. We primarily evaluated primary tumor efficacy and cervical lymph node efficacy in NPC patients. For short-term effects immediately after CRT, the distribution of primary tumor efficacy was significantly different in the two tumor-size groups (T1–T2 group versus the T3–T4 group; P=0.036) in that the T3–T4 group showed poorer efficacy than the T1–T2 group (OR: 3.581; 95% CI: 1.019–12.582; P=0.047). The association between cervical lymph node efficacy and stages of lymph node metastasis was nearly significant (OR: 3.358; 95% CI: 0.919–12.272; P=0.067; N2–N3 vs N0–N1). There was no correlation between the short-term effect 3 months after CRT and clinical factors.

Table 2 Association of clinical factors and the short term treatment effect 3 months after CRT.
Full size table

When assessing the associations between clinical factors and the risk of occurrence of toxic reactions such as xerostomia, myelosuppression, dermatitis and oral mucositis, we found that a family history of NPC was the most relevant risk factor for xerostomia (P=0.008), and patients who had a family history of NPC were at lower risk of developing grade 2+ xerostomia (OR: 0.239; 95% CI: 0.078–0.735; P=0.013). The results are shown in Table 3. Additionally, compared with patients with a body mass index (BMI)≤24.0, patients with a BMI>24.0 had a lower risk of developing grade 3+ oral mucositis (OR: 0.431; 95% CI: 0.215–0.865; P=0.018), as shown in Table S2.

Table 3 Association of clinical factors and the risk of xerostomia and myelosuppression after CRT.
Full size table

Relationship between gene polymorphisms in the ER stress response pathway and short-term effects in NPC patients

As shown in Tables S3 and S4, there were no statistically significant associations between SNPs and primary tumor efficacy or cervical lymph node efficacy immediately after chemoradiotherapy in either the univariate analysis or the multivariate analysis.

The CT and CT+TT genotypes of rs7566 in the CANX gene were significantly associated with better primary tumor efficacy 3 months after CRT, as compared with the CC genotype (CT OR: 0.133; 95% CI: 0.020–0.886; P=0.037 and CT+TT OR: 0.122; 95% CI: 0.018–0.815; P=0.030). The CT and CT+CC genotypes of rs2528521 in the CALCR gene were significant factors predicting cervical lymph node efficacy 3 months after CRT in NPC patients (CT OR: 15.294; 95% CI: 1.625–143.976; P=0.017 and CT+CC OR: 13.478; 95% CI: 1.470–123.575; P=0.021). The results are listed in Tables 4 and 5.

Table 4 Association of the selected SNPs with primary tumor efficacy 3 months after CRT.
Full size table
Table 5 Association of the selected SNPs with the cervical lymph node efficacy 3 months after CRT.
Full size table

Relationship between gene polymorphisms in the ER stress response pathway and the risk of toxicity

The associations between genotypes of ER stress response pathway genes and xerostomia and myelosuppression are shown in Tables 6 and 7. The results showed that the CC and CT+CC genotypes of VCP rs2074549 were associated with a significantly increased risk of developing grade 3+ myelosuppression, by using the TT genotype as a reference (CC OR: 3.040; 95% CI: 1.336–6.914; P=0.008 and CT+CC OR: 2.768; 95% CI: 1.243–6.162; P=0.013). A significant association was detected between the CT and CT+TT genotypes of CANX rs7566, compared with the CC genotype, and there was an increased risk of occurrence of grade 3+ myelosuppression (CT OR: 2.459; 95% CI: 1.096–5.518; P=0.029 and CT+TT OR: 2.382; 95% CI: 1.112–5.104; P=0.026). There was no association between the selected SNPs and the risk of occurrence of xerostomia. As shown in Table S5, no relationship was found between SNPs in ER stress response pathway genes and the risk of dermatitis. We identified only a borderline significant association between the CT genotype of rs2074549 in the VCP gene and the risk of developing grade 3+ oral mucositis (OR: 3.939; 95% CI: 0.976–15.902; P=0.054), as shown in Table S6.

Table 6 Association of the selected SNPs with the risk of xerostomia after CRT.
Full size table
Table 7 Association of the selected SNPs with the risk of radiation-induced myelosuppression after CRT.
Full size table

Stratification analyses of patients at tumor stages T3 and T4

To evaluate the effects of SNPs in ER stress pathway genes on clinical efficacy and toxic reactions of patients at only tumor stages T3 and T4, we performed additional stratification analyses. As shown in Table 8, the CT and CT+TT genotypes of rs7566 in the CANX gene were significantly associated with better primary tumor efficacy 3 months after CRT in NPC patients at stages T3 and T4 only (CT OR: 0.112; 95% CI: 0.013–0.941; P=0.044 and CT+TT OR: 0.101; 95% CI: 0.012–0.849; P=0.035). Regarding the myelosuppression in patients at only the T3 and T4 stages, we obtained similar results: the CC and CT+CC genotypes of VCP rs2074549 and the CT and CT+TT genotypes of CANX rs7566 were significantly associated with increased risk of grade 3+ myelosuppression (Table 9). In addition, the association was more significant for the entire patient population.

Table 8 Association between rs7566 and primary tumor efficacy 3 months after CRT for patients at T3 and T4 stage.
Full size table
Table 9 Association between rs2074549, rs7566 and risk of myelosuppression for patients at T3 and T4 stage after CRT.
Full size table

Discussion

In this study, the effects of 6 SNPs in ER stress pathway genes on the chemoradiotherapy responses in NPC patients were evaluated. Our data demonstrated that the T allele of CANX rs7566 was a favorable factor related to primary tumor treatment efficacy 3 months after CRT, but it was also significantly associated with the development of grade 3+ myelosuppression in Chinese NPC patients. The C allele of CALCR rs2528521 was significantly correlated with cervical lymph node efficacy 3 months after CRT. The C allele of VCP rs2074549 was significantly associated with the occurrence of 3+ myelosuppression. To our knowledge, this is the first study to investigate the effects of polymorphisms in ER stress pathway genes on chemoradiotherapy reactions in NPC patients.

Through correlation analysis of population characteristics and clinical outcomes, NPCs classified as T3 and T4 showed poorer primary tumor efficacy after chemoradiotherapy than those classified as T1 and T2. TNM stage is a major prognostic predictor in NPC patients, and patients classified as T3 and T4 often have a poor prognosis22. Additionally, patients with a family history of NPC tended to have a lower risk of developing grade 2+ xerostomia as well as an improved survival rate23. Thus, it was surmised that patients with a family history of xerostomia might be more tolerant than those without this family history. Our results revealed that patients with a BMI >24.0 were at lower risk of developing grade 3+ oral mucositis. Pretreatment BMI was a significant independent prognostic factor for patients with NPC. Compared with their low-BMI counterparts, high-BMI patients have better 5-year failure-free survival rates (FFS)24. High-BMI patients may be more tolerant of oral mucositis than low-BMI patients. However, the underlying mechanism remains unknown, and further studies are wanted.

Valosin-containing protein (VCP) is a conserved structural protein in eukaryotes that belongs to the ATPase family. The expression of VCP has been found to be related to the pathological state and clinical outcome of human tumors such as childhood acute lymphoblastic leukemia25. The results from our study indicated that the C allele of VCP rs2074549 was significantly associated with the risk of occurrence of grade 3+ myelosuppression directly after CRT in NPC patients (P=0.008), particularly in the subgroup of patients at stage T3 and T4 (P=0.003). Rs2074549 is located in the intronic region of the VCP gene, and mutations at this SNP may influence the expression of VCP by affecting mRNA splicing, localization and stability26. Another conceivable mechanism is that the real functional variant mediating this process is not SNP rs2074549 but other polymorphisms in linkage disequilibrium (LD) with this SNP27. This SNP is significantly associated with severe neutropenia19. VCP is supposedly involved in the dissociation and proteasomal degradation of I-κB, an inhibitor of nuclear factor-kappaB (NF-κB)28. Interestingly, the NF-κB pathway is related to neutrophil apoptosis29,30, but its detailed mechanism requires further study. Variation in this SNP is probably related to the regulation of neutrophil apoptosis, thus leading to myelosuppression. However, there was no association between rs2074549 and treatment efficacy in NPC patients.

Calcitonin receptor (CALCR), one of seven transmembrane-spanning G protein–coupled receptors, is involved in the maintenance of calcium homeostasis and the regulation of bone resorption. The C1377T polymorphism in the CALCR gene has been found to be associated with bone mineral density (BMID) in the lumbar spine in a postmenopausal Han Chinese population31. Furthermore, the CT and CT+CC genotypes at rs2528521 were significantly correlated with cervical lymph node efficacy 3 months after CRT both in the whole population and in patients at stages T3 and T4. Rs2528521 is a C/T single-nucleotide variation located in the promoter region of the CALCR gene. SNPs located within promoters might influence overall protein activity or gene expression32. Few studies have examined the association between CALCR gene variations and disease, and the relationship between CALCR gene variations and toxic reaction in response to CRT has not been reported. An elevated intracellular calcium level may induce cell apoptosis33,34. We speculate that functional mutation of this SNP may decrease the intracellular concentration of Ca2+, thereby leading to cervical lymph node cell survival.

We also analyzed the relationship between CANX (calnexin) gene polymorphisms and clinical outcomes of chemoradiation in NPC patients. Our data indicated that the C/T single-nucleotide variation in CANX rs7566 is a favorable prognostic factor 3 months after CRT, and it was significantly associated with the development of chemoradiotherapy-induced myelosuppression in Chinese NPC patients. Stratification analyses of NPC patients at stages T3 and T4 yielded similar results regarding the risk of myelosuppression. As a calcium-binding molecular chaperone of ER, CANX interacts with newly synthesized N-linked glycoproteins in a timely manner and thus plays an important role in protein synthesis and folding35. Accumulating evidence indicates that CANX is involved in apoptosis induced by ER stress36,37. CANX silencing in murine neonatal cardiomyocytes induces Ca2+ cycling defects, ER stress, and apoptosis38. The over-expression of CANX provokes apoptosis39,40. In addition, CANX rs7566 is located in the 3'-terminal untranslated region (3'-UTR), which can be fully or partly bound by microRNAs (miRNAs), thereby triggering specific mRNA cleavage or translational repression41,42. Thus, we speculate that rs7566 located at the miRNA-binding sites may influence the binding ability of miRNAs and consequently the expression of target genes, thereby leading to alterations in susceptibility to CRT and increasing CRT efficacy and the risk of development of myelosuppression.

There were several limitations of this study. First, clinical evaluations of acute radiation-induced toxic reactions, were somewhat dependent on the patients' subjective feelings. For example, reporting and tolerance degree of xerostomia may vary greatly among patients. Second, patients with different CRT regimens may have different tumor responses and side effects and thus might have confounded our analyses. Third, the number of enrolled patients in this study was small, thus potentially limiting the statistical power of the correlation analysis. Nevertheless, we will verify our current findings in our upcoming study with a larger sample size.

In conclusion, we identified the VCP rs2074549, CANX rs7566 and CALCR rs2528521 SNPs as novel chemoradiation associated genes that can be regarded as predictors of clinical outcome of CRT in NPC patients. Knowledge of chemoradiation-related genes may aid in identification of targets for innovative therapies and promote individualized CRT. To our knowledge, our study is the first attempt to explore the association between ER stress pathway gene variations and clinical outcomes of chemoradiation. The underlying mechanism by which ER stress pathway genes modulate the response to chemoradiation in NPC patients requires further study.

Author contribution

Xiao-bin GUO, Wan-le MA, and Cheng-xian GUO were responsible for the research design, data analysis and manuscript composition; Li-juan LIU, Yu-ling HUANG, and Jing WANG performed case collection, data entry and data interpretation; and Shao-jun CHEN, Li-hua HUANG, Hui WANG, and Xiang-dong PENG were responsible for verifying and analyzing the clinical data; Ji-ye YIN, Jin-gao LI, and Guo-ping YANG provided logistic assistance.