**Polymorphisms in** *ERCC5* **rs17655 and** *ERCC1* **rs735482 Genes Associated with the Survival of Male Patients with Postoperative Oral Squamous Cell Carcinoma Treated with Adjuvant Concurrent Chemoradiotherapy**

**Thomas Senghore 1,2, Huei-Tzu Chien 3,4, Wen-Chang Wang 5, You-Xin Chen 1, Chi-Kuang Young 6, Shiang-Fu Huang 3,7,\* and Chih-Ching Yeh 1,8,\***


Received: 26 November 2018; Accepted: 25 December 2018; Published: 1 January 2019

**Abstract:** The nucleotide excision repair (NER) pathway plays a major role in the repair of DNA damaged by exogenous agents, such as chemotherapeutic and radiotherapeutic agents. Thus, we investigated the association between key potentially functional single nucleotide polymorphisms (SNPs) in the NER pathway and clinical outcomes in oral squamous cell carcinoma (OSCC) patients treated with concurrent chemoradiotherapy (CCRT). Thirteen SNPs in five key NER genes were genotyped in 319 male OSCC patients using iPLEX MassARRAY. Cox proportional hazards models and Kaplan–Meier survival curves were used to estimate the risk of death or recurrence. Carriers of the *XPC* rs2228000 TT genotype showed a borderline significant increased risk of poor overall survival under the recessive model (hazard ratio (HR) = 1.81, 95% confidence interval (CI) = 0.99–3.29). The CC genotypes of *ERCC5* rs17655 (HR = 1.54, 95% CI = 1.03–2.29) and *ERCC1* rs735482 (HR = 1.65, 95% CI = 1.06–2.58) were associated with an increased risk of worse disease-free survival under the recessive model. In addition, participants carrying both the CC genotypes of *ERCC5* rs17655 and *ERCC1* rs735482 exhibited an enhanced susceptibility for recurrence (HR = 2.60, 95% CI = 1.11–6.09). However, no statistically significant interaction was observed between them. Our findings reveal that the *ERCC5* rs17655 CC and *ERCC1* rs735482 CC genotypes were associated with an increased risk of recurrence in male patients with OSCC treated with CCRT. Therefore, CCRT may not be beneficial, and alternative treatments are required for such patients.

**Keywords:** nucleotide excision repair; genetic polymorphism; oral squamous cell carcinoma; concurrent chemoradiotherapy; prognosis

### **1. Introduction**

Oral squamous cell carcinoma (OSCC) is the leading cause of cancer morbidity and mortality, especially among men in Taiwan [1]. Despite new advances in the diagnosis and therapeutic approaches, the 5-year survival remains low [1,2]. While relapse of OSCC remains a major clinical challenge, the incidence of relapse among patients varies, even for those with a similar stage of disease at diagnosis or those who undergo the same treatment [3]. This implies that other factors, such as genetic variations, may play an important role in disease prognosis.

Most patients with OSCC are diagnosed at an advanced stage of the disease [4]. For these patients, the treatment options are limited to mainly systemic therapy, often as concurrent chemoradiotherapy (CCRT) with platinum-based DNA damaging agents as either primary treatment or adjuvant postoperative therapy [5–7]. However, the overall survival (OS) for these patients remains poor because most of them experience recurrence or distance metastases [8–10]. Genetic variations in DNA repair genes affect susceptibility to the efficacy and survival outcome of a certain treatment [11,12]. Increased DNA repair capacity may affect the sensitivity of the tumor cells to chemotherapy and radiotherapy (RT) by allowing cancer cells to repair DNA that has been damaged by these agents. Single nucleotide polymorphism (SNP) in genes involved in the nucleotide excision repair (NER) pathway may modulate DNA repair capacity by influencing gene expression or activity, thereby affecting the anticancer effects of therapeutic agents and treatment response [13,14].

The excision repair cross-complementation genes, including groups 1 (*ERCC1*), 2 (*ERCC2*), and 5 (*ERCC5*) and xeroderma pigmentosium complementation group A (*XPA*) and C (*XPC*) encode proteins that are involved in the NER pathway; and together with other proteins, operate to recognize and repair damaged DNA [15]. The XPC together with XPF initially recognize the DNA lesion that is unwound and remodeled by helicase proteins ERCC3 and ERCC2 that binds to XPA and replication protein A (RPA). The ERCC1 and ERCC5 proteins are involved in the incision of the identified DNA lesion. The difference in treatment response and clinical outcome have been attributed to SNPs in genes that code of the above proteins [13,14]. Therefore, identifying genetic markers in the NER pathway may help develop personalized management strategies, thereby maximizing treatment success and improving survival.

Thus, we conducted a retrospective cohort study to test whether SNPs in genes involved in the NER pathway are associated with prognosis in male patients with OSCC treated with adjuvant CCRT. A total of 13 SNPs in *ERCC5*, *ERCC2*, *ERCC1*, *XPC*, and *XPA* genes, which have been found to affect the risk and/or survival of cancers, were selected in the present study [13,14,16–20]. Their associations with clinical outcomes were evaluated using alternative genetic models, including additive, dominant, and recessive models.

### **2. Material and Methods**

### *2.1. Study Population*

In total, 360 male participants newly diagnosed with histopathological confirmed advanced OSCC who received surgery plus adjuvant CCRT were recruited from the Head and Neck Surgery Department's Cancer Registry at Chang Gung Memorial Hospital, LinKou, Taiwan, from 1999 to 2016. A total of 41 participants were excluded, including 13 of aboriginal ethnicity, 23 with early-stage oral cancer (TNM stages I and II), and 5 with missing information on clinicopathologic variables (TNM stage, vascular invasion, and extracapsular spread). A final sample of 319 was included for analysis. Information on demographic characteristics (age, education, occupation, and ethnicity), lifestyle habit (cigarette smoking, alcohol drinking, and betel nut chewing), and family cancer history were collected through an interviewer-administered questionnaire. Lifestyle habits were categorized as either never (if the person never engaged in the habit continuously for more than a year) or ever (if the person ever engaged in the habit for more than a year). From the weight and height measurements, body mass index (BMI) was calculated as weight/height2 (kg/m2). Clinical information was also

collected before treatment through a detailed medical history, physical examination, completed blood count, routine blood chemistry, computed tomography (CT) or magnetic resonance imaging (MRI) of the head and neck, abdominal ultrasound, and whole body bone scan or positron emission tomography scan. This study was approved by the Chang Gung Memorial Hospital (IRB No. 201800213B0) and Taipei Medical University ethics review committees (IRB No. N201802083). All participants provided written informed consent after a detailed explanation of study objectives.

### *2.2. Sample Preparation and DNA Extraction*

For each participant, a pair of tumor and normal adjacent nontumor tissue samples were surgically removed, dissected into small pieces, and immediately stored in liquid nitrogen at −80 ◦C. The surgically removed samples were then sent for pathological examination and staging as per the seventh edition of the American Joint Committee on Cancer—TNM staging system [21]. Histology diagnosis was defined as squamous cell carcinoma, verrucous carcinoma, cylindric cell carcinoma, adenoid cystic carcinoma, mucoepidermoid carcinoma, and adenocarcinoma. For this study, only those with a diagnosis of squamous cell carcinoma were included. Venous blood samples were also collected and stored in heparinized tubes. Germline DNA was extracted from buffer-coated cells using the standard phenol-chloroform method and prepared for genotyping.

### *2.3. SNP Selection and Genotyping*

SNPs in the NER pathway were selected from studies that indicated that SNPs were associated with the risk or prognosis of malignancies in ethnic Chinese [16,18–20]. A total of 13 potentially functional SNPs in *ERCC5* (rs2094258, rs1047768, rs17655, and rs873601), *ERCC2* (rs13181 and rs1799793), *ERCC1* (rs735482, rs3212986, and rs11615), *XPC* (rs2228001 and rs2228000), and *XPA* (rs1800975 and rs10817938) genes were genotyped using the Sequenom iPLEX MassARRAY system (Sequenom, Inc., San Diego, CA, USA). A 10% random sample was reanalyzed, and it showed 100% concordance for all the polymorphisms.

### *2.4. Patient Treatment and Follow-Up*

All patients underwent radical tumor excision with clinical stage-based neck dissection after preoperative tumor survey. The primary tumors were resected above 1-cm safety margins (both peripheral and deep margins). Neck dissections were performed according to examination status. If the lesion invaded deeply and crossed the midline, as observed in tongue cancer, bilateral neck dissection was performed. Pathologic parameters included tumor stage, nodal status, extranodal extension (ENE), tumor cell differentiation, perineural invasion, skin invasion, bone invasion, and tumor depth. Postoperative RT was administered to patients with pT4 stage tumor, pathologically close margins (≤4 mm), or pathologically positive lymph nodes. The radiation dose lay between 6000 and 6600 cGy. CCRT with cisplatin-based agents was administered to patients with ENE or pathological multiple lymph node metastases 4 to 8 weeks after the surgical procedure. During the course of RT, 5-fluorouracil was administered orally.

Following commencement of treatment, the participants were monitored during their treatment and after treatment through regular clinical and radiological examinations. Follow-up involved monthly checkups for the first 6 months. This was followed by checkups every 2 months in the second 6 months, then checkups every 3 months within the second year, and checkups every 6 months thereafter. The follow-up included an analysis of medical history, physical examination (including complete oral examination), laboratory examination, X-rays, and CT or MRI. To confirm recurrence, histology of biopsy or imaging studies were conducted. Data for all deaths resulting from OSCC were based on death certificates.

### *2.5. Statistical Analysis*

Statistical analysis was performed using SAS (version 9.4 for Windows; SAS institute, Cary, NC, USA). Demographic and clinical characteristics were summarized as mean and standard deviation for continuous variables and frequency and proportions for categorical variables. The distribution of genotypes by clinical characteristics was assessed using Chi-square test. Major clinical outcomes were disease-free survival (DFS) and OS. DFS was measured from the first day of treatment to the time of recurrence, metastasis, or death due to any cause. OS was calculated as the time elapsed (in months) from the date of commencing RT to the date of death. Patients without an event at the date of the last contact were considered as being censured or subject to administrative censoring by the end of the follow-up period. Survival analysis was conducted using the Kaplan–Meier method, and survival curve differences among the genotypes were compared using the log-rank test. Univariate and multivariate Cox proportional hazards models were used to evaluate the effects of demographic, clinical characteristics, and SNPs on survival. Hazard ratios (HRs) and their 95% confidence intervals (CIs) were used to estimate the relative risk of death or recurrence. We evaluated the individual variants in different genetic models, including additive, dominant, and recessive models, on OSCC survival. Sociodemographic and clinical factors significant in the univariate analysis were adjusted in multivariate Cox regression models. Furthermore, multiplicative interactions were evaluated using the likelihood ratio test. Due to the location of multiple SNPs within the same chromosome or gene, linkage disequilibrium (LD) analysis was performed using Haploview (version 4.2). For those SNPs within the same block that were found to be in high LD with each other, further haplotypes analysis was performed using PHASE software (version 2.1) [22]. Statistical significance was set at *p* < 0.05 and was two-sided.

### **3. Results**

### *3.1. Demographic and Clinical Characteristics of Study Participants*

The demographic and clinical characteristics of the study patients are summarized in Table 1. The mean age was 49.72 ± 9.8. Most participants were under 50 years (51.41%) old, of Taiwanese descent (paternal 72.10% and maternal 74.92%), had normal BMI (49.22%), had smoked cigarettes (according to the "ever" criterion; 85.25%), drank alcohol (69.28%), and chewed betel nut (86.21%). A considerable number of patients exhibited poor clinical characteristics. In total, 55 (17.24%) had poorly differentiated tumors, 197 (61.76%) had primary tumor size in the T3 to T4 range, 217 (67.92%) had N2 to N3 nodal involvement, 177 (55.49%) had perineural invasion, 19 (5.96%) had vascular invasion, 40 (12.54%) had lymphatic invasion, 205 (64.24%) had ENE, and 277 (86.83%) had pathologic TNM stage IV. The genotype frequency distribution analysis showed a statistically significant difference in genotypes of *ERCC1* rs11615 in terms of tumor differentiation (*p* = 0.039), *XPC* rs2228000 in terms of vascular invasion (*p* = 0.045), *ERCC1* rs3212986 and *XPA* rs10817938 in terms of lymphatic invasion (*p* = 0.046 and 0.033, respectively), *XPC* rs2228001 in terms of pathologic TNM stage (*p* = 0.039), and *ERCC5* rs17655 in terms of DFS (*p* = 0.049) (Table S1).

### *3.2. Survival Analysis*

The median (range) follow-up duration was 15 months (1–199 months) and 12 months (1–199 months) for OS and DFS, respectively. In the univariate analysis, N2–N3 nodal involvement (HR = 2.41, 95% CI = 1.42–4.10), lymphatic invasion (HR = 2.18, 95% CI = 1.30–3.67), and ENE (HR = 3.91, 95% CI = 2.13–7.19) were significantly associated with OS, whereas primary tumor size in the range of T3 to T4 (HR = 1.72, 95% CI = 1.17–2.53), N2–N3 nodal involvement (HR = 1.63, 95% CI = 1.09–2.44), and ENE (HR = 1.79, 95% CI = 1.20–2.69) were significantly associated with DFS. However, no significant association was observed between demographic and lifestyle factors and survival (Table 2).


**Table 1.** Demographic and clinical characteristics of patients with oral squamous cell carcinoma (OSCC) treated with concurrent chemoradiotherapy (CCRT).

OSCC, oral squamous cell carcinoma; SD, standard deviation; BMI, body mass index.


**Table 2.** Univariate association of demographic and clinical factors with survival in patients with OSCC treated with CCRT.

BMI, body mass index; OSCC, oral squamous cell carcinoma; HR, hazard ratio; CI, confidence interval. \* *p* < 0.05.

In the univariate Cox proportional hazards models, the *ERCC1* rs735482 CC genotype was marginally significantly associated with poor DFS (HR = 1.53, 95% CI = 0.99–2.38; *p* = 0.058). The *XPA* rs10817938 CC genotype was significantly associated with an increased risk of worse OS (HR = 2.97, 95% CI = 1.20–7.35; *p* = 0.019), and DFS (HR = 2.61, 95% CI = 1.06–6.41; *p* = 0.037), respectively (Table S2).

The results for the multivariate Cox proportional hazards models with covariates adjusted for all selected SPNs are shown in Table 3. Only the *XPC* rs2228000 TT genotype (HR = 1.81, 95% CI = 0.99–3.29, *p* = 0.053) showed an increased risk of poor OS at borderline significance compared with the CC+CT genotypes. The *ERCC5* rs17655 CC (HR = 1.50, 95% CI = 1.01–2.24; *p* = 0.045) and *ERCC1* rs735482 CC (HR = 1.61, 95% CI = 1.04–2.51; *p* = 0.034) genotypes were significantly associated with an increased risk of DFS compared with their counterparts with the GG+GC and AA+AC genotypes, respectively, in the recessive models. The test of LD show that SNPs in *ERCC5* block 1 (rs2094258 and rs1047768; D' = 0.97, *R*<sup>2</sup> = 0.19) and block 2 (rs17655 and rs873601; D' = 0.98, *R*<sup>2</sup> = 0.89), *ERCC1* block (rs3212986 and rs11615; D' = 1.00, *R*<sup>2</sup> = 0.18), *XPC* block (rs2228001 and rs2228000; D' = 1.00, *R*<sup>2</sup> = 0.28), and *XPA* block (rs1800975 and rs10817938; D' = 1.00, *R*<sup>2</sup> = 0.24) were in LD with each other (Figure S1). Of the haplotype constructed from these blocks, only *XPA* GT haplotype (HR = 0.68, 95% CI = 0.47–0.99; *p* = 0.042) showed statistically significant association with OS (Table S3).

**Table 3.** Multivariate association between nucleotide excision repair (NER) candidate single nucleotide polymorphisms (SNPs) and OSCC survival in patients treated with CCRT.



OSCC, oral squamous cell carcinoma; SNPs, single nucleotide polymorphisms; HR, hazard ratio; CI, confidence interval. <sup>a</sup> Adjusted for age, BMI, N stage, lymphatic invasion, and extranodal extension. <sup>b</sup> Adjusted for age, T stage, N stage, and extranodal extension. \* *p* <0.05.

We further conducted a combination analysis for the *ERCC5* rs17655 and *ERCC1* rs735482 polymorphisms and DSF in patients with OSCC. The Kaplan–Meier curves showed borderline significant differences in DFS among the four genotypes (log-rank test *p* = 0.078) (Figure 1). The multivariate Cox proportional models indicated that patients with the combination of *ERCC5* rs17655 CC and *ERCC1* rs735482 CC genotypes exhibited a higher risk of disease recurrence than those with the combination of *ERCC5* rs17655 GG+GC and *ERCC1* rs735482 AA+AC genotypes (HR = 2.60, 95% CI = 1.11–6.09; *p* = 0.027) (Table 4). However, this gene-gene interaction was not statistically significant.

**Figure 1.** Kaplan–Meier survival curve for the combined *ERCC5* rs17655 and *ERCC1* rs735482 polymorphisms and disease-free survival in patients with oral squamous cell carcinoma treated with concurrent chemoradiotherapy. The figure illustrates a borderline significant difference in recurrence among the four groups (log-rank test *p* = 0.078).

**Table 4.** Interaction between the *ERCC5* rs17655 and *ERCC1* rs735482 polymorphisms on the disease-free survival of patients with OSCC treated with CCRT.


OSCC, oral squamous cell carcinoma; HR, hazard ratio; CI, confidence interval. <sup>a</sup> Adjusted for age, T stage, N stage, and extranodal extension. \* *p* < 0.05.

Table 5 shows the subgroup analysis for the association between significant SNPs and DFS stratified by demographic and clinopathological factors. Results show a significant interaction between *ERCC5* rs17655 polymorphism and perineural invasion on the risk for DFS (interaction *p* = 0.008). The *ERCC5* rs17655 CC genotype individuals with perineural invasion (HR = 2.46, 95% CI = 1.46–4.15; *p* < 0.001) had an increased risk for DFS compared to their counterparts with no perineural invasion. Although a significant interaction between *ERCC1* rs735482 polymorphism and vascular invasion was also observed (interaction *p* < 0.001), the harmful effect of CC genotype on recurrence was not present in any subgroup of vascular invasion.


**Table 5.** Association between the *ERCC5* rs17655 and *ERCC1* rs735482 polymorphisms and the disease-free survival in OSCC patients treated with CCRT stratified by demographic and clinopathological factors.

OSCC, oral squamous cell carcinoma; HR, hazard ratio; CI, confidence interval; Int, interaction. <sup>a</sup> Adjusted for age, T stage, N stage, and extranodal extension. \* *p* < 0.05.

### **4. Discussion**

In this study, we investigated the association between potentially functional SNPs in the NER pathway genes and clinical outcomes in male patients with OSCC treated with CCRT. Our findings suggest that the *XPC* rs2228000 TT genotype was marginally significantly associated with increased risk of death, whereas the *ERCC5* rs17655 CC and *ERCC1* rs735482 CC genotypes were significantly associated with the increased risk of relapse.

The NER pathway plays a major role in DNA repair through the removal of bulky DNA lesions formed by ultraviolet (UV) radiation, environmental mutagens, and other chemotherapeutic agents [23,24]. Studies have revealed that variations in DNA-repair capacity are related to cancer risk and prognosis [25,26]. In addition, SNPs in the NER genes modulate susceptibility to efficacy and survival outcome of the treatment in certain types of cancers [11,12]. Therefore, the same phenomena

may be exhibited in patients with OSCC, particularly those undergoing CCRT. Such information may be useful in identifying patients who may benefit from alternative therapies to achieve superior survival and improve quality of life.

The *XPC* gene is a key component of the XPC complex, which plays an important role in the early part of the global genome NER. The corresponding protein plays an important function in damage sensing and DNA binding [27]. SNPs in this gene have been found to affect clinical outcomes in various cancer types [18,28,29]. Li and colleagues observed that the *XPC* rs2228000 TT genotypes were associated with shorter OS than the CC+CT genotype individuals in a study of Japanese gastric cancer patients [30]. Another Chinese study demonstrated that patients with the CC genotype of *XPC* rs2228000 have a borderline significant decreased risk of developing gastric cancer compared with those with the CT+TT genotype [31]. This evidence suggests that the T-allele may have a high susceptibility for poor prognosis. Similarly, in our study, the *XPC* rs2228000 TT genotype shows an increased risk of death compared with the CC+CT genotype. Given the importance of the *XPC* gene in the NER pathway, it is possible that variants of *XPC* alter the DNA repair capacity and thereby affect sensitivity to therapeutic agents. However, the association observed in our study was borderline significant and must be interpreted with caution. Furthermore, large studies may be required to confirm these findings.

We also observed that those with the *ERCC5* rs17655 CC and *ERCC1* rs735482 CC genotypes have an increased risk of relapse compared with individuals with the GG+GC and AA+AC genotypes, respectively. Considered a central component in NER, *ERCC5* encodes a specific DNA endonuclease responsible for excision and repair of UV-induced DNA damage [23]. Evidence has linked *ERCC5* polymorphism to chemotherapeutic response and prognosis of tumors [19,20]. The ERCC5 mRNA expression levels were correlated with cytotoxicity of chemotherapy regiments [32]. Additionally, the rs17655 leads to an amino-acid substitution from histidine to aspartic acid, which may lead to differential interacting abilities, thus, influencing the DNA repair efficacy. Song et al. also observed that *ERCC5* rs17655 polymorphism has a moderately increased risk of recurrence in squamous cell carcinoma of the oropharynx [33]. ERCC1 is also a crucial member of the NER pathway that forms a complex with ERCC4, and together with ERCC5, is responsible for DNA incision [34]. Other studies have reported that *ERCC1* affects the clinical outcome and may serve as a potential biomarker for response to cisplatin-based therapy [35,36]. On the basis of these study results, we speculate that the CC genotypes of *ERCC5* rs17655 and *ERCC1* rs735482 may increase the DNA-repair capacity of cancer cells, leading to increased susceptibility to recurrence. Therefore, if these findings are confirmed by other studies, these SNPs may serve as therapeutic biomarkers for clinical outcome in patients with OSCC who undergo CCRT.

Our study has several limitations. First, the hospital-based nature of the patients may have led to selection bias. Secondly, not all SNPs in the entire NER pathway were used. Some rare functional SNPs may have an influence on survival. Finally, the human papilloma virus (HPV) status and inflammatory cytokines expression of patients was not included in the analysis and may limit the interpretation of our findings; hence, HPV and cytokines may affect survival [37,38]. However, a major strength of our study is that all the patients had a similar tumor stage and received the same treatment. This meant that the effect of different treatments was excluded, which might lead to different levels of DNA damage and repair.

### **5. Conclusions**

We investigated the association between key potentially functional SNPs in the NER pathway and susceptibility for death or relapse in male patients with advanced OSCC who were treated with adjuvant CCRT. Our findings showed that the CC genotypes of *ERCC5* rs17655 and *ERCC1* rs735482 were associated with an increased risk of recurrence. CCRT may not be beneficial for these patients; therefore, alternative treatments are required. To our knowledge, this is the largest study to investigate the association between NER polymorphisms and survival in patients with OSCC treated with CCRT

in ethnic Chinese. Our findings may require further confirmation in studies with a larger sample size or other ethnic populations.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/8/1/33/s1, Figure S1: Linkage disequilibrium (LD) analysis among single nucleotide polymorphisms (SNPs) within the same chromosome or gene, Table S1: Frequency distributions of NER polymorphisms by clinicopathological factors, Table S2: Univariate association between NER candidate SNPs and OSCC survival in CCRT treated patients, Table S3: Haplotype analysis of association between NER candidate SNPs and OSCC survival in patients treated with CCRT.

**Author Contributions:** Conceptualization: S.-F.H. and C.-C.Y.; investigation: H.-T.C. and C.-K.Y.; methodology: T.S., Y.-X.C., W.-C.W., S.-F.H., and C.-C.Y.; resources: H.-T.C., Y.-X.C., and C.-K.Y.; formal analysis: T.S. and Y.-X.C.; validation: T.S., H.-T.C., and Y.-X.C.; writing: (original draft preparation) T.S.; writing (review and editing): T.S., W.-C.W., S.-F.H., and C.-C.Y.; supervision: S.-F.H. and C.-C.Y.; and funding acquisition: S.-F.H. and C.-C.Y.

**Funding:** The research was supported by the Ministry of Science and Technology, Taiwan (MOST 107-2314-B-038-071).

**Acknowledgments:** The authors would like to thank Wallace Academic Editing for the English language review.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Distinct Expression and Clinical Significance of Zinc Finger AN-1-Type Containing 4 in Oral Squamous Cell Carcinomas**

### **Julián Suarez-Canto 1, Faustino Julián Suárez-Sánchez 2, Francisco Domínguez-Iglesias 2, Gonzalo Hernández-Vallejo 3, Juana M. García-Pedrero 4,5,6,\* and Juan C. de Vicente 5,7,8,\***


Received: 27 October 2018; Accepted: 7 December 2018; Published: 10 December 2018

**Abstract:** Zinc finger AN1-type containing 4 (ZFAND4) has emerged as a promising prognostic marker and predictor of metastasis for patients with oral squamous cell carcinoma (OSCC). However, further validation is fundamental before clinical implementation. Hence, this study evaluated the expression pattern of ZFAND4 protein expression by immunohistochemistry using an independent cohort of 125 patients with OSCC, and correlations with the clinicopathologic parameters and disease outcome. Remarkably, ZFAND4 expression, while negligible in normal epithelium, exhibited two distinct expression patterns in tumors that did not overlap. A gross granular staining was characteristic of the undifferentiated cells at the invasive front of tumors, whereas the most differentiated cells located at the center of the tumor nests showed diffuse non-granular staining. ZFAND4 staining was higher in undifferentiated than in differentiated areas of tumors. High ZFAND4 expression in differentiated cells was significantly associated to well-differentiated (*p* = 0.04) and non-recurrent tumors (*p* = 0.04), whereas ZFAND4 expression in undifferentiated cells correlated with tumor location (*p* = 0.005). No correlations between the ZFAND4 expression and patient survival were found. These data question the clinical relevance of ZFAND4 expression as a prognostic biomarker in OSCC, and also reveal distinct ZFAND4 expression patterns depending on the differentiation areas of tumors that should be evaluated separately.

**Keywords:** ZFAND4; ANUBL1; oral squamous cell carcinoma; immunohistochemistry; prognosis

### **1. Introduction**

Head and neck squamous cell carcinoma that includes, among others, oral squamous cell carcinoma (OSCC) is the sixth most common cancer in the world, with an annual prevalence of nearly 600,000 new cases worldwide [1,2]. It is generally accepted that OSCC initiates and progresses through a series of multiple genetic alterations caused by chronic exposure to carcinogens, such as alcohol, smoking, and human papilloma virus [3]. Multiple genetic and molecular studies have improved our understanding of the molecular basis of this disease. Indeed, several cellular signaling pathways have been found dysregulated in these tumors through genetic and epigenetic alterations. However, despite major advances in diagnosis and treatment, the survival rate of patients with OSCC has modestly improved over the past 40 years, and it remains at approximately 50% [4].

Sasahira et al. [5] investigated the transcriptional profiles of primary and recurrent OSCC, and found that one of the most upregulated genes identified in recurrent OSCC was zinc finger AN1-type containing 4 (ZFAND4), also known as AN1 and ubiquitin-like homolog (ANUBL1). Although the functional role of ZFAND4 in cancer is still unknown, Kurihara-Shimomura et al. [6] evaluated its prognostic utility in OSCC. They concluded that ZFAND4 could be a useful marker for predicting metastasis and poor prognosis in patients with OSCC. In addition, Tang et al. [7] demonstrated that ZFAND4 expression is upregulated in gastric cancer and positively associated with the grading of this disease.

In the light of these data, ZFAND4 emerges as a promising prognostic biomarker; however, further validation in independent study cohorts is fundamental for implementation to the clinic. Therefore, the objective of this study was to investigate the expression pattern and clinical relevance of ZFAND4 protein expression using an independent cohort of 125 patients with OSCC, and to establish correlations with the clinicopathologic parameters and disease outcome.

### **2. Experimental Section**

### *2.1. Patients and Tissue Specimens*

A retrospective study was designed. Surgical tissue specimens from 125 patients with OSCC who underwent surgical treatment with curative purposes at the Hospital Universitario Central de Asturias between 1996 and 2007 were retrospectively collected, in accordance to approved institutional review board guidelines. All experimental procedures were conducted in accordance to the Declaration of Helsinki and approved by the Institutional Ethics Committee of the Hospital Universitario Central de Asturias and by the Regional CEIC from Principado de Asturias. Informed consent was obtained from all patients. Clinicopathologic data were collected from medical records. Tissue specimens were obtained from the Biobanco del Principado de Asturias, and representative tissue sections from archival, formalin-fixed paraffin-embedded blocks.

### *2.2. Tissue Microarray Construction*

The original archived hematoxylin- and eosin-stained slides were reviewed by an experienced pathologist (FDI) to confirm histological diagnosis. Three representative tissue cores (1 mm diameter) were selected from each tumor block, and transferred to a recipient 'Master' block in a grid-like manner using a manual tissue microarray instrument. In addition, each tissue microarray also contained three cores of normal epithelium as an internal control. A section from each microarray was stained with hematoxylin and eosin, and examined by light microscopy to check the adequacy of tissue sampling.

### *2.3. Immunohistochemistry (IHC)*

TMA sections (4 μm) were cut and dried and dried on Flex IHC microscope slides (Dako). The sections were deparaffinized with standard xylene, hydrated through graded alcohols into water, and pretreated by hydrogen peroxide to quench the endogenous peroxidase activity. Antigen retrieval was performed using Envision Flex Target Retrieval solution (Dako), at room temperature on an automatic staining workstation (DakoAutostainer Plus, Dako, Glostrup, Denmark). Staining was carried out at room temperature on an automatic staining workstation (Dako Autostainer Plus) with anti-ZFAND4 antibody (Atlas Antibodies, Stockholm, Sweden) diluted to 0.5 Nμg/mL using the

Dako EnVision detection system (Dako, Glostrup, Denmark). Sections were counterstained with hematoxylin, dehydrated with ethanol, and permanently coverslipped. For negative control purposes, DakoCytomation mouse serum diluted at the same concentration as the primary antibody was used.

Staining was scored blinded to clinical data by two independent observers. ZFAND4 expression was evaluated according to the percentage of stained tumor cells and the staining intensity using the Allred score, as previously described [8]. The proportion of ZFAND4-positive cells was evaluated in both undifferentiated and differentiated areas of the tumors. In all these groups, proportional scores were categorized as: 0, no cells were stained; 1, 1/100 cells were stained; 2, 1/10 cells were stained; 4, 2/3 cells were stained; 5, all cells were stained. Staining intensity was scored as: 0, negative; 1, weak; 2, intermediate; and 3, strong. The total score was calculated by the sum of the proportional and intensity scores, ranging from 0 to 8. Similar to Kurihara-Shimomura et al. [6], the optimal cut-off score for ZFAND4 expression was selected using the receiver operating characteristic (ROC) curve according to the survival status. ZFAND4 staining was independently evaluated in undifferentiated areas of tumors mainly located at the invasive front, and in differentiated areas at the center of the tumor islands.

### *2.4. Statistical Analysis*

χ<sup>2</sup> and the Fisher's exact test were used for comparison between categorical variables. Disease-specific survival (DSS) was determined for the date of treatment completion to death for the tumor. For time-to-event analysis, survival curves were plotted using the Kaplan-Meier method. Differences between survival times were analyzed by the log-rank test. Hazard ratios (HR) with their 95% confidence intervals (CI) for clinicopathologic variables were calculated using the univariate Cox proportional hazards model analysis. All tests were two-sided and *p* values less than 0.05 were considered statistically significant. All statistical analyses were performed using SPSS version 21 (IBM Co., Armonk, NY, USA).

### **3. Results**

### *3.1. Patient Characteristics*

The cohort of 125 OSCC patients was composed of 82 men and 43 women, ranging from 28 to 91 years, with a median age of 57 years. Forty-one patients (33%) were never-smokers and 56 (45%) never-drinkers. The main clinicopathologic characteristics are summarized in Table 1. Forty-nine cases (39%) showed neck lymph node metastasis, more than 50% were well-differentiated tumors and advanced clinical stages (III or IV), and the most common site was the tongue (41%) followed by the floor of the mouth (30%). Adjuvant radiotherapy was administered to 75 patients (60%), and adjuvant chemotherapy to 14 patients (11.2%). Fifty-four cases (43%) showed loco-regional recurrence, and 19 (15%) suffered from a second primary carcinoma. Over a median follow-up of 61 months (range, 1 to 230 months) 53 deaths occurred.

**Table 1.** Clinical and pathological characteristics of 123 patients with oral squamous cell carcinoma and where zinc finger AN1-type containing 4 (ZFAND4) was valuable.



**Table 1.** *Cont.*

### *3.2. Immunohistochemical Analysis of ZFAND4 Expression in OSCC Tissue Specimens*

ZFAND4 staining was not valuable in two (1.6%) of 125 OSCC specimens. While ZFAND4 expression was negligible in normal epithelium, two distinct expression patterns were noted in the tumors that did not overlap in any of the samples (Figure 1). A gross granular staining was characteristic of the undifferentiated cells at the invasive front of tumors, whereas the most differentiated cells located at the center of the tumor nests showed diffuse non-granular staining. The mean percentages of positive ZFAND4 staining were 44.98 (standard deviation –SD-35.38) in undifferentiated cells and 17.18 (SD, 20.61) in differentiated cells. ZFAND4 staining intensity was also evaluated in both undifferentiated and differentiated areas. In undifferentiated cells, there were 13 (11%) negative cases, 37 (30%) weak, 43 (35%) intermediate, and 30 (24%) cases with strong staining. In differentiated cells, 47 (38%) cases were scored negative, 11 (9%) weak, 47 (38%) intermediate, and 18 (15%) had strong staining. Since each tumor was represented by three different tissue cores in the OSCC TMAs, the percentages of stained cells frequently varied in the three tumor

areas assessed. Taking this into consideration, the Allred score was determined in two different ways: Either considering the maximum value of ZFAND4 positivity or the mean value of the three tumor cores. Regarding the intensity of immunostaining, the maximum value was always used for all calculations. Finally, the total score was calculated by the sum of the percentages of staining and intensity scores. The resulting indexes ranged between 0 and 8. The receiver operating characteristics (ROC) curve was used to determine the best cut-off score to predict patients' survival, and this value was 4. Accordingly, those cases with an Allred score above 4 were considered as high ZFAND4 expression.

(**B**) **Figure 1.** *Cont*.

**Figure 1.** Immunoexpression of ZFAND4 in oral squamous cell carcinoma. (**A**) Staining in undifferentiated areas (arrow) and differentiated areas (\*). (**B**) Staining in undifferentiated cells, mainly located in the invasive front of tumor tissue. (**C**) Staining in differentiated cells located at the center of the tumor islands.

### *3.3. Associations of ZFAND4 with Clinicopathologic Characteristics*

We next assessed the correlations of high ZFAND4 expression with the clinical data. Table 2 shows the associations of high ZFAND4 expression determined by using the maximum value of the percentage of stained cells to calculate the Allred score. In differentiated areas, high ZFAND4 expression was significantly associated with well-differentiated (*p* = 0.04) and non-recurrent tumors (*p* = 0.04), whereas ZFAND4 expression in undifferentiated cells was significantly correlated with tumor location in the tongue (*p* = 0.005).

On the other hand, when the Allred score was calculated using the mean value of percentage of stained cells (Table 3), ZFAND4 expression in differentiated cells was found to be significantly associated with N status (*p* = 0.02), being more frequently detected in pN0 and pN1 cases compared to pN2. However, no significant relationship was found between ZFAND4 expression in undifferentiated cells and any clinicopathologic variable.


**Table 2.** Relationships between clinical and pathological variables and high ZFAND4 expression determined by using the maximum value of stained cells to calculate the Allred score.


**Table 2.** *Cont.*

**Table 3.** Relationships between clinical and pathological variables and high ZFAND4 expression determined by using the mean value of stained cells to calculate the Allred score.



**Table 3.** *Cont.*

### *3.4. ZFAND4 Expression and Patients' Survival*

Over a median follow-up period of 61 months, 27 patients (42.1%) harboring high ZFAND4 expression in undifferentiated cells calculated by using the mean Allred score died due to the index cancer, and 15 patients (41.6%) with high ZFAND4 expression in the differentiated cells. When the maximum Allred score was used, 31 patients (40.7%) with high ZFAND4 expression in the undifferentiated cells, and 20 patients (36.3%) with high ZFAND4 expression in the differentiated cells died due to the index cancer. Kaplan-Meier analysis showed that there were no statistically significant differences in disease-specific survival (DSS) between patients with high versus low ZFAND4 expression in either differentiated or undifferentiated cells (Table 4).


### **4. Discussion**

This study aimed to investigate the clinical relevance and prognostic significance of ZFAND4 in OSCC. The prevalence of OSCC is estimated at 264,000 cases and 128,000 deaths annually worldwide [3]. Since the completion of the Human Genome Project in 2003, the subsequent progress in understanding the biology of cancer has led to the development of personalized therapies based on the patient's unique molecular and genetic profile to target defective signaling pathways of tumor cells. It is generally accepted that OSCC arises from multiple genetic alterations, although the molecular basis of carcinogenesis is not fully understood. DNA sequencing technologies coupled with advances in algorithms have enormously contributed to the molecular and functional characterization of mutations, genes, and pathways altered in multiple cancers, including OSCC [9–12]. Furthermore, the majority of tumors showed alterations in multiple targetable genes that are candidates for combination therapy [11]. It is of paramount importance to identify molecular alterations involved in the development of recurrent and metastatic disease, which remains the main cause of morbidity and mortality in OSCC patients.

In a recent paper, Sasahira et al. [5] conducted a cDNA microarray analysis in order to compare the gene expression profile of primary and recurrent OSCC. Ten genes were found to be upregulated in recurrent OSCC compared with the primary tumors. Among these genes, ZFAND4 showed a 100-fold recurrent/primary, thus suggesting a possible role for this gene in tumorigenesis. Tang et al. [7] reported that ZFAND4 expression was consistently highly expressed in gastric cancer compared to normal tissue, and positively associated with increased stage. Functionally, ZFAND4 was found to downregulate the expression of the anti-proliferative miRNAs, miR-148b, miR-375, and miR-182, in SGC-7901 cells, thereby promoting cell proliferation by activation of cyclin-dependent kinase and downregulation of p21 and p53 [7], which supports the notion that ZFAND4 may act as an oncogene in gastric cancer.

In this study, we assumed the same methodology used by Kurihara-Shimomura et al. [6] in order to validate their results in an independent cohort of OSCC patients, and more importantly, the utility of ZFAND4 as a predictor of metastasis and poor prognostic marker. Interestingly, ZFAND4 staining consistently showed distinct expression patterns and distribution in our cohort of 125 OSCC, i.e., granular staining in undifferentiated areas and diffuse staining in differentiated areas of the tumors. These two expression patterns were analyzed separately to evaluate possible correlations with the clinical and follow-up data. Moreover, Allred scores were calculated using both maximum value and mean value of ZFAND4-positive cells for the three tissue cores selected from each tumor.

We found that ZFAND4 staining was higher in undifferentiated than in differentiated areas of tumors. However, ZFAND4 expression in differentiated cells showed the most relevant and significant associations with well-differentiated (*p* = 0.04) and non-recurrent tumors (*p* = 0.04). Nevertheless, ZFAND4 expression did not show a major impact on patient survival. Only a trend was observed between low ZFAND4 expression in differentiated cells and tumor-associated deaths (Table 4). Therefore, the prognostic relevance of ZFAND4 described by Kurihara-Shimomura et al. [6] has not been replicated in our series. Furthermore, Kurihara-Shimomura et al. [6] reported that ZFAND4 is essential for distant metastasis in OSCC, and hypothesized that ZFAND4 could facilitate metastasis to the lymph nodes and distant organs by promoting angiogenesis and/or lymphangiogenesis. In marked contrast, our results only proved a marginal (if any) relationship between ZFAND4 expression and the presence of lymph node metastasis. The limitations of our study are the retrospective design and the use of tissue microarrays to evaluate ZFAND4 immunostaining. Several factors could contribute to the discrepant results between these two studies. On one hand, etiological, clinical, and epidemiological differences in the patient cohorts as well as molecular/biological differences among tumors depending on the geographic areas. Moreover, ZFAND4 expression exhibited a highly heterogeneous pattern within the tumors depending on the differentiation status, which could certainly have a major contribution on the varying results. Particularly since we evaluated separately the clinical significance and correlations of the two distinct ZFAND4 expression patterns in undifferentiated and

differentiated areas of tumors, while Kurihara-Shimomura et al. [6] did not distinguish expression patterns within the different tumor areas. Thus, positive ZFAND4 immunostaining was scored in the whole tumor as the diffuse staining in tumor islands, cords, or sheets, irrespective of the cell differentiation status and the grade of keratinization. Since ZFAND4 expression showed two distinct non-overlapping expression patterns depending on the differentiation areas of tumors, each with a clearly distinct clinical significance, evaluation of ZFAND4 expression in the whole tumor could therefore be misleading.

### **5. Conclusions**

The herein presented data question the clinical relevance of ZFAND4 expression as a prognostic biomarker in OSCC, and also revealed distinct ZFAND4 expression patterns depending on the differentiation areas of tumors that should be evaluated separately. Further studies are necessary to fully elucidate the pathobiological role of ZFAND4 in OSCC and the potential clinical implications.

**Author Contributions:** Conceptualization, J.C.d.V.; Funding acquisition, J.C.d.V. and J.M.G.-P.; Investigation, J.S.C., F.J.S.S. and F.D.I.; Methodology, J.S.C., F.J.S.S.; Resources, F.D.I., G.H.V. and J.C.d.V.; Writing—original draft, J.C.d.V.; Writing—review & editing, J.M.G.-P.

**Funding:** This study was supported by grants from the Plan Nacional de I + D + I ISCIII PI16/00280 and CIBERONC (CB16/12/00390), the Instituto de Investigación Sanitaria del Principado de Asturias (ISPA), Fundación Bancaria Caja de Ahorros de Asturias-IUOPA and the FEDER Funding Program from the European Union.

**Acknowledgments:** We thank the samples and technical assistance kindly provided by the Principado de Asturias BioBank (PT13/0010/0046), financed jointly by Servicio de Salud del Principado de Asturias, Instituto de Salud Carlos III and Fundación Bancaria Cajastur and integrated in the Spanish National Biobanks Network. This study was supported by grants from the Plan Nacional de I + D + I ISCIII PI16/00280 and CIBERONC (CB16/12/00390), the Instituto de Investigación Sanitaria del Principado de Asturias (ISPA), Fundación Bancaria Caja de Ahorros de Asturias-IUOPA and the FEDER Funding Program from the European Union.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **The Prognostic Significance of Neutrophil-to-Lymphocyte Ratio in Head and Neck Cancer Patients Treated with Radiotherapy**

**Yeona Cho 1, Jun Won Kim 1, Hong In Yoon 2, Chang Geol Lee 2, Ki Chang Keum 2,\* and Ik Jae Lee 1,\***


Received: 12 November 2018; Accepted: 29 November 2018; Published: 3 December 2018

**Abstract:** Background: To investigate the prognostic value of pre-treatment neutrophil/lymphocyte ratio (NLR) in patients treated with definitive radiotherapy (RT) for head and neck cancer. Methods: We retrospectively analyzed 621 patients who received definitive RT for nasopharyngeal, oropharyngeal, hypopharyngeal, and laryngeal cancer. An NLR cut-off value of 2.7 was identified using a receiver operating characteristic curve analysis, with overall survival (OS) as an endpoint. Results: The 5-year progression-free survival (PFS) and OS for all patients were 62.3% and 72.1%, respectively. The patients with a high NLR (68%) had a significantly lower 5-year PFS and OS than their counterparts with a low NLR (32%) (PFS: 39.2% vs. 75.8%, *p* < 0.001; OS: 50.9% vs. 83.8%, *p* < 0.001). In a subgroup analysis according to primary site, a high NLR also correlated with a lower PFS and OS, except in oropharyngeal cancer, where a high NLR only exhibited a trend towards lower survival. In a multivariate analysis, a high NLR remained an independent prognostic factor for PFS and OS. Conclusion: Head and neck cancer tends to be more aggressive in patients with a high NLR, leading to a poorer outcome after RT. The optimal therapeutic approaches for these patients should be reevaluated, given the unfavorable prognosis.

**Keywords:** head and neck cancer; radiotherapy; neutrophil/lymphocyte ratio; survival

### **1. Introduction**

Currently, definitive radiotherapy (RT) is one of the main modalities used to treat locally advanced head and neck cancer. However, patients exhibit varying degrees of RT response and may develop recurrences even after a complete response. Although various clinical and molecular predictors of treatment outcomes after definitive RT for head and neck cancer have been investigated, no clear consensus regarding reliable predictive biomarkers has been reached.

Treatment outcomes are known to be affected by both tumor characteristics and host-related factors, including age, sex, and performance status. Recent reports have also described close associations of systemic inflammation with tumorigenesis and treatment outcomes [1,2], and several laboratory markers associated with systemic inflammatory processes, including albumin, hemoglobin, absolute white blood cell (WBC) count or WBC components, and platelet count, have been investigated as prognostic and predictive markers in various types of cancer [3,4]. Inflammation plays a key role in cancer physiology by promoting carcinogenesis, dedifferentiation, and primary tumor growth, and by stimulating tumor cell proliferation by inhibiting apoptosis and increasing mitotic rates [5].

Tumor–host interactions can induce systemic inflammatory responses that affect the numbers of circulating WBCs and the neutrophil/lymphocyte ratio (NLR) in certain types of cancers [6]. Normal NLR values are in the range of 0.78–3.53 in the general population [7]; high NLR values are associated with poor outcomes not only in cancer patients but also in patients with cardiovascular disease [8–10]. To date, some studies of head and neck cancer have suggested an association of high NLR with a poor prognosis. However, data regarding the prognostic significance of a high NLR are limited, especially among patients undergoing definitive RT [11–13]. The antitumor immune response is thought to be part of the ionizing radiation-induced tumor cell death process. Therefore, tumor shrinkage caused by the host immune response may be a direct effect of radiation [14,15]. Accordingly, we postulated that the host immune status, as reflected by the NLR, may predict recurrence after RT in head and neck cancer patients. This study aimed to evaluate the relationships of pretreatment NLR and other hematologic markers with tumor recurrence and survival in patients undergoing definitive RT for head and neck cancer.

### **2. Materials and Methods**

### *2.1. Patient Selection and Treatment Protocols*

This study was approved by the institutional review board of the Gangnam Severance Hospital (Protocol number: 3-2017-0387). We retrospectively reviewed the medical records of patients who underwent definitive RT with or without concurrent chemotherapy for cancers of the head and neck (including nasopharyngeal, oropharyngeal, hypopharyngeal, and laryngeal cancers) at our institution between 2006 and 2015. Patients who underwent surgery before or after RT, received RT of <30 Gy, had a distant metastasis at the initial diagnosis or previous history of other primary cancer, and whose pre-RT common blood test results were unavailable were excluded. The remaining 621 patients included in the analysis were staged according to the 7th edition of the TNM classification of the American Joint Committee on Cancer (AJCC). Human Papilloma Virus (HPV) infection status was evaluated in oropharyngeal cancer patients. To assess HPV status of each tumor, we used formalin-fixed, paraffin-embedded biopsy tissue to examine p16 expression, which is recognized as a surrogate marker for HPV infection in the oropharynx. Details of this process are described in a previous report of our institution [16].

Patients were treated with definitive RT alone, concurrent chemoradiotherapy (CCRT), or induction chemotherapy followed by CCRT. The choice of treatment was determined by the primary tumor and stage, risk factors, and/or the physicians' discretion. External beam RT comprised either 3D-conformal RT or intensity-modulated RT (IMRT) and was administered 5 days per week in daily fractions of 1.8–2.5 Gy to yield total doses to the primary tumor of 66–75 Gy.

Concurrent chemotherapy regimens included weekly cisplatin (DDP; 40 mg/m2); weekly 5-fluorouracil (5-FU) and cisplatin (FP; 750 mg/m2 and 20 mg/m2, respectively); and 5-FU, taxotere, and cisplatin (TPF; 750 mg/m2, 70 mg/m2, and 75 mg/m2, respectively) every 3 weeks. The anti-epidermal growth factor receptor mAb, cetuximab (Erbitux), was also used. The induction chemotherapy regimen consisted of FP every 3 weeks for 3 cycles or TPF every 3 weeks for 2 cycles.

### *2.2. Hematologic Markers*

The patients' blood counts were evaluated prior to performing diagnostic procedures or administering treatments. The WBC count, hemoglobin (Hb) level, absolute neutrophil count (ANC), absolute lymphocyte count (ALC), platelet count, and albumin level were recorded. A diagnosis of anemia was based on a hemoglobin level of <13 g/dL in men and <12 g/dL in women, and hypoalbuminemia was defined as a serum albumin level <3.5 g/dL. Onodera's prognostic nutritional index (PNI) was calculated as 10 × Albumin + 0.005 × ALC.

The NLR was calculated as the neutrophil count divided by the lymphocyte count, and the platelet/lymphocyte ratio (PLR) was calculated as the platelet count divided by the lymphocyte count. The optimum NLR cut-off values were identified via a receiver operating characteristic (ROC) curve analysis, using overall survival (OS) as an end point (Supplement Figure S1), and patients were categorized into high NLR (NLR ≥ 2.7) and low (NLR < 2.7) NLR groups.

### *2.3. Outcome Assessment*

All patients were followed up for 4–6 weeks after RT, and subsequently at 3-month intervals for the first and second years, 6-month intervals for the third year, and annually for the fourth and fifth years. Progression was defined as regrowth of the primary tumor or the involvement of cervical lymph node(s) (LN) or detection of any new lesion(s) in follow-up imaging studies. Progression-free survival (PFS) was defined as the interval between the date of initial treatment to the detection of first recurrence, death from any cause, or the last follow-up. OS was defined as the interval between the date of initial treatment and death from any cause or the last follow-up.

### *2.4. Statistical Analysis*

Categorical data were analyzed using Fisher's exact test or the χ<sup>2</sup> test. Continuous data were compared between groups using the Mann–Whitney U test. The Kaplan–Meier method and log-rank test were used to estimate and compare the PFS and OS rates. Hazard ratios (HRs) were obtained using the cumulative survivor function and are reported with corresponding 95% confidence intervals (CIs). Univariate and multivariate analyses of factors related to OS and PFS were conducted using the Cox proportional hazards model, and multivariate analysis included all variables with *p* values < 0.05 in the univariate analysis. A *p* value < 0.05 was considered statistically significant. All analyses were performed using IBM SPSS, version 20.0 (SPSS, Chicago, IL, USA).

### **3. Results**

### *3.1. Patient and Treatment Characteristics*

Table 1 presents the demographic and treatment characteristics of the 621 included patients, of whom 425 (68.4%) and 196 (31.6%) were stratified into the low and high NLR groups, respectively. Laryngeal cancer was the most frequent type of primary cancer in the low NLR group, whereas nasopharyngeal cancer cases comprised the majority in the high NLR group. Patients with a high NLR tended to have a more advanced clinical T classification and higher frequency of LN metastasis and a significantly higher frequency of systemic chemotherapy (75.2% vs. 48.5%, *p* < 0.001) than those in the low NLR group.

The groups did not differ significantly in terms of the use of IMRT. Of the 16 patients who did not complete the planned course of RT, four (2.0%) and 12 (2.8%) belonged to the high and low NLR groups, respectively (*p* = 0.787). However, more patients in the high NLR group received radiation doses of ≥70 Gy (equivalent dose in 2 Gy fractions, α/β = 10), and the overall duration of RT tended to be longer in this group.


**Table 1.** Patient characteristics.


**Table 1.** *Cont.*

Abbreviations: <sup>a</sup> Equivalent dose in 2 Gy fractions, <sup>b</sup> unevaluable, <sup>c</sup> concurrent chemoradiotherapy, <sup>d</sup> 3-dimensional conformal radiotherapy, <sup>e</sup> intensity modulated radiotherapy.

### *3.2. Hematologic Markers*

Table 2 presents the values of the measured hematologic markers in all patients. The median baseline WBC count, ANC, and ALC were 6800/μL, 3900/μL, and 1810/μL, respectively, and the WBC and ANC values were significantly higher in the high NLR group. A higher proportion of patients with leukocytosis (WBC count ≥9000/μL) was also observed in the high NLR group (31.7% vs. 7.8% for low NLR, *p* < 0.001). The high NLR group also had higher platelet counts, which expectedly yielded a higher PLR, and was more likely to present with anemia and hypoalbuminemia at the time of diagnosis. Patients with a high NLR also had a significantly lower Onodera's PNI (49.7 vs. 55 for low NLR, *p* < 0.001).

**Table 2.** Hematologic markers.



**Table 2.** *Cont.*

Abbreviations: <sup>a</sup> Absolute neutrophil count, <sup>b</sup> absolute lymphocyte count, <sup>c</sup> platelet/ lymphocyte ratio, <sup>d</sup> prognostic nutritional index, <sup>e</sup> unevaluable.

### *3.3. Survival Analysis*

The patients were followed up for a median of 39 (range, 2–130) months. During the follow-up period, 148 patients died and 156 experienced a recurrence. The 5-year PFS and OS rates for all patients were 63.8% and 72.9%, respectively, and both rates were significantly lower in the high NLR group than in the low NLR group (PFS: 39.2% vs. 75.8%, *p* < 0.001; OS: 50.9% vs. 83.8%, *p* < 0.001) (Figure 1). In survival analyses stratified by early- (stage I–II) or advanced-stage disease (stage III–IV), a high NLR remained significantly associated with a poor PFS and OS (Figure 2).

An additional subgroup analysis was performed after stratifying cases by the primary site (nasopharynx, oropharynx, hypopharynx, and larynx). As shown in Figures 3 and 4, patients with nasopharyngeal, hypopharyngeal, and laryngeal cancer and the high NLR group had poorer PFS and OS rates. Among patients with oropharyngeal cancer, however, a high NLR status exhibited only borderline significance in terms of 5-year PFS (42.0% vs. 54.0%, *p* = 0.059) (Figure 3B) and only exhibited a trend with reduced OS (51.5% vs. 65.3%, *p* = 0.215) (Figure 4B).

We also conducted additional analysis according to the treatment scheme. Both PFS and OS were significantly worse in high NLR patients receiving any type of treatment: Patients receiving RT alone, 5-year PFS 84.8% vs. 56.1%, *p* < 0.001 and OS 90.6% vs. 68.6%, *p* < 0.001; patients receiving CCRT, 5-year PFS 66.2% vs. 31.5%, *p* < 0.001 and OS 74.3% vs. 41.7%, *p* < 0.001; patients receiving induction chemotherapy + CCRT, 5-year PFS 68.1% vs. 40.7%, *p* < 0.001 and OS 78.9% vs. 55.2%, *p* < 0.001.

**Figure 1.** (**A**) Progression-free survival and (**B**) overall survival of patients according to neutrophil/lymphocyte ratio (NLR) status.

**Figure 2.** Progression-free survival and overall survival of patients with head and neck cancer at (**A**,**B**) stage I–II, and (**C**,**D**) stage III–IV.

**Figure 3.** Subgroup analysis of progression-free survival according to primary tumor site: Nasopharynx (**A**), oropharynx (**B**), hypopharynx (**C**) and larynx (**D**).

**Figure 4.** Subgroup analysis of overall survival according to primary tumor site: nasopharynx (**A**), oropharynx (**B**), hypopharynx (**C**) and larynx (**D**).

#### *3.4. Analysis of Prognostic Factors*

The results of univariate and multivariate analyses performed to identify prognostic factors for PFS and OS are shown in Table 3. The multivariate analysis revealed significant associations of a high NLR with poor PFS, an older age, and an advanced T classification. An elevated PLR was also found to be associated with a poor PFS. An older age, advanced T classification, and high NLR were also found to associate significantly with OS, and an elevated PLR exhibited a borderline significant association with a poor OS. Primary hypopharyngeal cancer was associated with both a poor PFS and a poor OS. A low albumin level (<3.3 g/dL) exhibited a negative trend with PFS, but not with OS. LN metastasis, overall stage, leukocytosis (WBC count ≥9000/μL), anemia, and Onodera's PNI were not identified as prognostic factors for PFS or OS.


survivalandoverallsurvival.

#### *J. Clin. Med.* **2018** , *7*, 512


**Table 3.** *Cont.* ≥50 0.48 0.36–0.65 <0.001 0.80 0.53–1.23 0.316 0.44 0.31–0.63 <0.001 0.74 0.45–1.24 0.259Abbreviations: a Absolute neutrophil count, b absolute lymphocyte count, c platelet/ lymphocyte ratio, d prognostic nutritional index, e concurrent chemoradiotherapy, f 3-dimensionalconformal radiotherapy, g intensity modulated radiotherapy.

 **Analysis**

 **95%CI**

 0.69–3.42

 0.288

*p*

### *3.5. p16 Status and Hematologic Markers*

We also evaluated the relationship of the p16 status with the levels of various hematologic markers. This status was available for 38 of 94 patients with oropharyngeal cancer (40.4%). In total, 26 patients had p16-positive oropharyngeal cancer and 12 patients had p16-negative oropharyngeal cancer. Patients with a positive p16 status tended to have a lower NLR than those with a negative status (median NLR: 2.1 vs. 2.8, *p* = 0.103). In addition, the WBC count and ANC were lower in p16-positive patients than in their p16-negative counterparts, although these differences were not statistically significant (median WBC: 7100 vs. 8300/μL, *p* = 0.073; median ANC: 4000 vs. 5200/μL, *p* = 0.119) (Supplement Figure S2).

Among the 26 patients with p16-positive oropharyngeal cancer, PFS and OS were not different between the low and high NLR groups (5-year PFS 56.1% vs. 57.1%, *p* = 0.781 and 5-year OS 86.5% vs. 80.0%, *p* = 0.646). In patients with p16-negative oropharyngeal cancer (*n* = 12), PFS was significantly lower in the high NLR group than in the low NLR group (5-year PFS 0% vs. 100%, *p* = 0.009), while OS showed no difference, which may be due to the limited number of cases.

### **4. Discussion**

During the past decade, various markers of systemic inflammation have been evaluated with the aims of refining patient stratification to treatment and predictions of survival. Of these markers, the NLR, which is derived from the ANC and ALC of a full blood count, is routinely available. Accordingly, in this study, we evaluated the significance of the pre-treatment NLR in patients who received RT for head and neck cancer and observed significant associations of a high NLR with disease recurrence and OS in the patient sample.

As noted above, we stratified the patients into two groups according to NLR status and found that those with a high NLR had a more advanced clinical stage and therefore more frequently received concurrent chemotherapy with a higher total radiation dose and longer total duration of RT. Despite this more aggressive treatment, however, patients with a high NLR were more likely to experience unfavorable outcomes, and these results remained consistent regardless of disease stage. We further found that adverse hematologic features, such as an elevated WBC, ANC, and platelet count as well as anemia, hypoalbuminemia, and a low Onodera's PNI, were more frequently observed in the high NLR group. These findings suggest that certain types of tumors elicit an enhanced systemic inflammatory response, which may reflect the aggressive nature of the tumor.

We also performed a subgroup analysis according to the primary site of head and neck carcinoma and determined that a high NLR was significantly associated with a poorer PFS and OS among patients with nasopharyngeal, hypopharyngeal, and laryngeal cancers. However, among patients with oropharyngeal cancer, a high NLR was only borderline significant as a prognosticator of PFS and exhibited a trend with OS. Other studies of oropharyngeal cancer have identified NLR as a significant prognostic factor for disease control, and most have used a relatively higher NLR cut-off value (≥5) [17,18] than those used for other primary sites in the head and neck [19,20]. Additionally, some studies of oropharyngeal cancer have identified the prognostic significance of circulating neutrophil and lymphocyte counts only for recurrence-free survival (RFS), but not for OS [21].

Our above findings may be attributable to the limited number of oropharyngeal cancer patients in our study (*n* = 94) or to the effects of HPV infection. Previous studies suggested that HPV infection might affect the distribution of WBC components and alter inflammatory responses in patients with oropharyngeal cancer [11]. Huang et al. reported that HPV-positive patients had lower levels of circulating neutrophil and monocyte counts when compared to their HPV-negative counterparts, despite similar levels of lymphocyte counts [21]. In the HPV-positive cohort, a high neutrophil or monocyte count was found to correlate with reductions in OS and RFS, whereas in the HPV-negative cohort, the neutrophil and lymphocyte counts were not predictive of either survival parameter. Another study also reported a significantly lower NLR among HPV-positive patients than among their HPV-negative counterparts. Given these findings, we suggest that oropharyngeal cancer may exhibit

behaviors or inflammatory responses that are distinct from those of head and neck cancers at other primary sites. Although our study did not reveal significant differences in inflammatory markers among oropharyngeal cancer patients according to p16 status, HPV infection might have altered the WBC distribution and affected disease control and OS.

Some scientists have recommended that more accurate diagnostic tools are needed for risk stratification according to HPV infection. Previous reports have suggested that HPV specific tests such as DNA in situ hybridization and polymerase chain reaction should be used to confirm HPV status, although p16 protein over-expression is very sensitive to the presence of transcriptionally-active HPV, correlates strongly with patient outcomes, is widely available, and is easy to interpret [22]. Gupta et al. also suggested that circulating HPV16 DNA in the plasma may be a clinically useful biomarker [23]. These biomarkers can be measured using a simple blood test, and the level of ctDNA is predictive of an early treatment response. If used in combination with inflammatory markers, this technique will be helpful in predicting a patient s prognosis and determining an appropriate treatment plan.

As noted, NLR is an easily and routinely determined biomarker of systemic inflammation, and in our study population a high NLR correlated strongly and independently with a poorer PFS and OS. Previous studies have also identified a high NLR as an independent prognostic factor for many other types of cancers, including colorectal cancer, renal cell cancer, pancreatic cancer, and head and neck squamous cell carcinoma [2,24–28]. Although the tumorigenic mechanism underlying this relationship with the NLR has not been clearly elucidated, it appears likely that increased levels of several inflammatory cytokines contribute to a microenvironment that promotes carcinogenesis and tumor progression [29]. Several growth factors, including epidermal growth factor, vascular endothelial growth factor, and transforming growth factor-α, also contribute to the creation of microenvironments supportive of angiogenesis and tumor proliferation [30]. Our results are therefore consistent with the concept that a high NLR contributes to poor disease control by suppressing the cytolytic activities of activated effector T cells and the peritumoral infiltration of immuno-suppressive cells such as macrophages [31,32].

We further demonstrated that a higher PLR was associated with PFS in our multivariate analysis. Again, the significance of interactions between platelets and the tumor microenvironment remains somewhat unclear. The platelet count provides an additional index of systemic inflammation elicited by the tumor and degranulation. This inflammation, together with the consequent release of platelet-derived proangiogenic mediators within the tumor microvasculature, may also serve as an important determinant of tumor growth [33–35].

This study had a few limitations. First, the study had a retrospective design and included various primary sites of head and neck cancer. Additionally, patients in the high NLR group tended to have more advanced-stage disease. However, the NLR remained a significant prognostic factor for survival in the multivariate analysis, as well as in an additional analysis according to disease stage. Second, the selection of treatment modalities and regimens was heterogeneous and was determined in accordance with the primary site and physicians' discretion. Third, we did not explore the association of LN metastasis with patient outcomes because of the use of clinical stage and likelihood of upstaging for LNs. Therefore, our findings should be interpreted cautiously. Nevertheless, this is one of the largest studies to evaluate the prognostic significance of systemic inflammation in patients who underwent RT for head and neck cancer, and we have revealed potential differences in the patients' characteristics and outcomes according to their NLR status. In addition, we observed different associations of the NLR status among patients with oropharyngeal cancer versus those with other head and neck cancers, which underscores the need for further studies of the relationship between HPV infection and the NLR.

In conclusion, the results of our large population study validate the suggested association of a high NLR with poorer outcomes after adjusting for potential confounding factors. Although further studies of the biological mechanisms underlying the relationship between inflammation and aggressiveness are needed, our results suggest that a classification system based on pretreatment hematologic markers could identify patients with a high risk of recurrence and poor survival.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/7/12/512/s1, Figure S1: The receiver operating characteristic (ROC) curve analysis for neutrophil/lymphocyte ratio (A) using overall survival OS as an end point. Figure S2. The white blood cell count (A) and neutrophil/lymphocyte ratio (B) in p16 negative and p16 positive patients.

**Author Contributions:** Conceptualization, Y.C., H.I.Y., I.J.L; Data Curation, Y.C., C.G.L., K.C.K.; Formal Analysis, Y.C., J.W.K., H.I.Y.; Resources, C.G.L., K.C.K., I.J.L.; Supervision, K.C.K., I.J.L.; Writing—Original draft, Y.C., J.W.K.; Writing—Review and editing, H.I.Y., C.G.L., K.C.K., I.J.L.

**Funding:** This work was supported by the National Research Foundation of Korea Grant funded by the Korean Government (number NRF-2018R1D1A1B07048234).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


### *J. Clin. Med.* **2018**, *7*, 512


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article*
