Prognostic signature of protocadherin 10 methylation in curatively resected pathological stage I non-small-cell lung cancer

Although curative resection is the current treatment of choice for localized non-small-cell lung cancer (NSCLC), patients show a wide spectrum of survival even after complete resection of pathological stage I NSCLC. Thus, identifying molecular biomarkers that help to accurately select patients at high risk of relapse is an important key to improving the treatment strategy. The purpose of this study was to evaluate the prognostic signature of protocadherin 10 (PCDH10) promoter methylation in curatively resected pathological stage I NSCLC. Using methylation-specific polymerase chain reaction assays, methylation of PCDH10 promoter was assessed in cancer tissues of 109 patients who underwent curative resection of pathological stage I NSCLC. Associations between PCDH10 methylation status and disease outcome was analyzed. PCDH10 promoter methylation was detected in 46/109 patients (42.2%). Patients with methylated PCDH10 showed significantly worse recurrence-free, overall, and disease-specific survival compared with those without methylation (P < 0.0001, P = 0.0004, P = 0.0002, respectively). Multivariate Cox proportional hazard regression analysis revealed that adjusted hazard ratios of methylated PCDH10 were 5.159 for recurrence-free, 1.817 for overall, and 5.478 for disease-specific survival (P = 0.0005, P = 0.1475, P = 0.0109, respectively). The pattern of recurrence was not significantly different between patients with and without PCDH10 methylation (P = 0.5074). PCDH10 methylation is a potential biomarker that predicts a poor prognosis after curative resection of pathological stage I NSCLC. Assessment of PCDH10 methylation status might assist in patient stratification for determining an appropriate adjuvant treatment and follow-up strategy.


Abstract
Although curative resection is the current treatment of choice for localized non-small-cell lung cancer (NSCLC), patients show a wide spectrum of survival even after complete resection of pathological stage I NSCLC. Thus, identifying molecular biomarkers that help to accurately select patients at high risk of relapse is an important key to improving the treatment strategy. The purpose of this study was to evaluate the prognostic signature of protocadherin 10 (PCDH10) promoter methylation in curatively resected pathological stage I NSCLC. Using methylation-specific polymerase chain reaction assays, methylation of PCDH10 promoter was assessed in cancer tissues of 109 patients who underwent curative resection of pathological stage I NSCLC. Associations between PCDH10 methylation status and disease outcome was analyzed. PCDH10 promoter methylation was detected in 46/109 patients (42.2%). Patients with methylated PCDH10 showed significantly worse recurrence-free, overall, and disease-specific survival compared with those without methylation (P < 0.0001, P = 0.0004, P = 0.0002, respectively). Multivariate Cox proportional hazard regression analysis revealed that adjusted hazard ratios of methylated PCDH10 were 5.159 for recurrence-free, 1.817 for overall, and 5.478 for disease-specific survival (P = 0.0005, P = 0.1475, P = 0.0109, respectively). The pattern of recurrence was not significantly different between patients with and without PCDH10 methylation (P = 0.5074). PCDH10 methylation is a potential biomarker that predicts a poor prognosis after curative resection of pathological stage I NSCLC. Assessment of PCDH10 methylation status might assist in patient stratification for determining an appropriate adjuvant treatment and follow-up strategy.

Introduction
Because non-small-cell lung cancer (NSCLC) is a molecularly heterogeneous disease, even patients with the same tumor, node, metastasis (TNM) classification in the American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) staging system show varied clinical outcomes and prognoses. Although surgical resection for early stage lung cancer is the best treatment for cure, even after curative resection for pathological stage I NSCLC, ~30% of patients eventually relapse after surgery [1]. For improved prognosis after resection, adjuvant chemotherapy has been widely applied; however, prospective randomized data showed that adjuvant chemotherapy for stage IB does not significantly improve survival; in fact, a detrimental effect was observed for stage IA [2][3][4]. For effective application of adjuvant therapy, identifying a novel biomarker able to predict patients with high chance of relapse is important in establishing a tailored treatment strategy for stage I NSCLC. Recent cumulative evidence indicates that aberrant promoter methylation and subsequent silencing of various specific genes play important roles in the development and/or progression of many human cancers [5][6][7][8][9], and promoter methylation status of specific genes has been reported as a promising biomarker for predicting disease outcome more accurately than TNM staging [10][11][12].
Protocadherins are a subfamily of the cadherin superfamily of genes that exhibit cell-to-cell adhesion activity with a mechanism thought to be distinct from that of classic cadherins [13,14]. Protocadherin 10 (PCDH10) belongs to the non-clustered protocadherin and is reported to be widely expressed but frequently silenced by promoter methylation in many carcinomas and hematological malignancies. Recent literature demonstrated that PCDH10 exerts a tumor-suppressor effect in gastric cancer by inducing apoptosis, controlling cell growth, and inhibiting cell invasion and metastasis [15]; however, the prognostic value of epigenetic silencing of PCDH10 in early-stage lung cancer remain to be investigated.
In this study, we employed a methylation-specific polymerase chain reaction (MS-PCR) assay to examine PCDH10 methylation status in curatively resected pathological stage I NSCLC. To evaluate the prognostic signature of PCDH10 promoter methylation in patients with curatively resected pathological stage I NSCLC, we analyzed whether methylation status is related to postoperative prognosis. We also assessed the association between PCDH10 methylation and clinicopathological features to investigate whether PCDH10 methylation is an independent prognostic marker.

Study population
This study was approved by the Institutional Review Boards of Kure Medical Center/Chugoku Cancer Center. In total, 109 patients who underwent complete resection for pathological stage I NSCLC at Kure Medical Center/Chugoku Cancer Center between June 2005 and November 2011 were enrolled. Detailed enrollment criteria were described previously [16]. Briefly, patients who received adjuvant therapy were included, and exclusion criteria were: neoadjuvant therapy, Noguchi Type A or B tumors (these types are not considered as candidates for adjuvant therapy because of their extremely excellent prognoses), small tumors precluding further sample extraction subsequent to routine histological examination, lost to follow-up or death from non cancerous causes <100 days postoperatively.
Video-assisted thoracic surgeries were performed in all patients. Systemic or selective lymph node dissection was conducted in patients who underwent lobectomy or segmentectomy. Wedge resection without lymph node dissection or concomitant with lymph node sampling was performed in selected patients, primarily because they had limited lung/cardiac function or multiple comorbidities.
The outpatient follow-up protocol was as described previously [16]. Briefly, chest and upper abdominal computed tomography were performed at 6-month intervals during the first 2 postoperative years and every year thereafter for at least 5 years. Other imaging examinations were performed as needed when patients presented with specific symptoms, or periodically at the discretion of the attending physician. For pulmonary nodules discovered during followup, diagnoses were determined by the institutional cancer board as previously described [16], primarily based on the modified Martini and Melamed classification scheme [17,18]. For metachronous adenocarcinomas with the same grade of differentiation as the primary tumor, secondary tumors lacking bronchoalveolar carcinoma components were classified as pulmonary metastases. Recurrence patterns were defined as previously described [16]; local recurrence was defined as the presence of a recurrent tumor in the same lobe or lymph node in the same ipsilateral hemithorax as the primary tumor. Pulmonary metastasis in a different lobe, even in the same ipsilateral hemithorax as the primary tumor, was defined as remote recurrence.

Statistical analysis
Comparisons were performed using the JMP for Windows statistical software (version 9.02; SAS Institute, Cary, NC). Associations between clinicopathological characteristics and PCDH10 methylation were determined, using Fisher exact test. Using the Kaplan-Meier method, the effect of PCDH10 methylation status on the time to death or recurrence was estimated, and differences between groups were analyzed by log-rank testing. Recurrence-free survival (RFS), overall survival (OS), and disease-specific survival (DSS) times were defined from the date of surgery until the date of event or the last follow-up. Cox proportional regression analysis was used to analyze the hazard ratios (HRs) of independent factors for survival. Adjusted HRs were calculated using factors that were significant by univariate analysis. Differences were considered significant for P < 0.05.

PCDH10 methylation status in the evaluation set
As the evaluation set, PCDH10 methylation status was analyzed in selected patients, who were matched with respect to gender, age, histology, stage, and differentiation. In this setting, patients with recurrence showed a significantly higher frequency of PCDH10 methylation compared to those without recurrence (data not shown).

Correlation between PCDH10 methylation and clinicopathological features
Representative results of MS-PCR are shown in Figure 1. The overall frequency of PCDH10 promoter methylation was 42.2% (46/109). The demographic and clinicopathological data of patients according to PCDH10 methylation are shown in Table 1.
Although we focused on methylation of the PCDH10 gene as a biomarker, we also performed immunohistochemical staining of the PCDH10 protein in several patients; representative results of this analysis are shown in Figures 1 and 2, which demonstrate that tumors with PCDH10 methylation showed a marked reduction in PCDH10 protein expression compared to those without methylation or noncancerous tissue.

Association of PCDH10 methylation status and disease outcome
The mean follow-up time of all patients was 1584 ± 681 days and that of nondeceased patients was 1751 ± 610 days. Five-year RFS, OS, and DSS rates for all patients were 71.1%, 74.8%, and 90.5%, respectively. The 5-year RFS rate was 43.0% in patients with PCDH10 methylation, which was significantly worse than 92.1% in patients without methylation (P < 0.0001, Fig. 3A). Table 2 shows the crude and adjusted HRs for RFS. Multivariate analysis revealed that PCDH10 methylation (P = 0.0005), age (P = 0.0143), and adjuvant therapy (P = 0.0071) were significantly independent poor prognostic factors for RFS. Table 3 shows that the recurrence pattern was not significantly different between patients with and without PCDH10 methylation (P = 0.5074). The 5-year OS rate was 60.1% in patients with PCDH10 methylation, which was significantly worse than 85.3% in patients without methylation (P = 0.0004, Fig. 3B). As shown in Table 4, multivariate analysis revealed that age (P = 0.0004) and histology (P = 0.0031) were significant independent risk factors; however, PCDH10 methylation was not an independent prognostic factor (P = 0.1475). The 5-year DSS rate was 79.7% in patients with PCDH10 methylation, which was significantly worse than 96.8% in patients without methylation (P = 0.0002, Fig. 3C). As shown in Table 5, multivariate analysis showed that PCDH10 methylation (P = 0.0109), age (P = 0.0452), and histology (P = 0.0386) were significantly independent risk factors for DSS. Table 3 shows that operative procedure did not significantly impact the recurrence pattern, and Tables 2, 4,  was not performed in patients who underwent wedge resection, a subgroup analysis was performed according to operative procedures. RFS, OS, and DSS were all significantly worse for patients with PCDH10 methylation, whether the operative procedure was lobectomy/segmentectomy (P = 0.0003, P = 0.0129, and P = 0.0023, respectively; Fig. 4A-C) or wedge resection (P = 0.0330, P = 0.0079, and P = 0.0410, respectively; Fig. 5A-C).

Discussion
Cumulative evidence indicates that epigenetic gene silencing of specific genes is associated with the onset and progression of human cancers [8], and the identification of specific gene methylation has been recognized as a powerful molecular biomarker for disease outcome of NSCLC [5,10]. Brock et al. showed that the methylation of two or more genes (p16, CDH13, RASSF1A, and APC) in tumor tissues and mediastinal lymph nodes of patients with curatively resected stage I NSCLC was associated with a lower 5-year RFS than methylation of fewer than two genes [10]. Sandoval et al. demonstrated that hypermethylation of 5 genes (HIST1H4F, PCDHGB6, NPBWR1, ALX1, and HOXA9) was significantly associated with shorter RFS in these patients [23]. The prognostic significance of promoter methylation in several of the above-mentioned genes was also assessed in our cohort (data not shown). Additionally, our group reported that methylation of BRCA1 and DLX4 affected disease outcome after curative resection of stage I NSCLC [12,16]. Future study by the combined analyses of multiple genes might contribute to identify the gene combination that more accurately predicts disease outcome.
Cadherin-mediated cell-cell interactions are critical events during the development and morphogenesis of cancer [24]. Cadherin 1, which is also known as E-cadherin, is widely recognized as a growth and invasion suppressor, and is used even for the diagnosis and prognosis of epithelial cancers. Additionally, the cancer-related functions of several other cadherins have been well studied [25,26]. Protocadherin is a member of the cadherin superfamily and contains several domains differing from those of the classical cadherins [25]. Although the specific roles of each protocadherin have not been fully investigated, protocadherin genes have been implicated as tumor suppressor genes with important functions, including signal transduction and growth control [27]. Several studies have concluded that PCDH10 suppresses tumor cell growth, migration, invasion, and colony formation, and is frequently inactivated epigenetically in colorectal, cervical, nasopharyngeal, esophageal, and other cancer types [26][27][28]. Recent study demonstrated an important role of p53 in regulating tumor cell migration through activating PCDH10 expression [29]. PCDH10 also appears to induce myeloma cell apoptosis by inhibiting the NF-kappaB pathway [30]. Here, although the specific biological role of PCDH10 in NSCLC still needs to be elucidated, we showed that PCDH10 promoter methylation predicted a poor prognosis in patients with curatively resected pathological stage I NSCLC by performing Kaplan-Meier survival analysis and log-rank testing. To further clarify the  prognostic value of PCDH10 methylation, multivariate analyses were conducted because a significant difference in PCDH10 methylation status was found in several clinicopathological features in this cohort. Furthermore, a recent study demonstrated that promoter methylation status of another protocadherin (PCDH20) was also associated with a shorter OS in lung cancer [31]; therefore, the role of other protocadherin methylation in NSCLC needs to be investigated in future studies.
The frequency of PCDH10 promoter methylation was 42.2% in this study, although Tang et al. reported that promoter methylation of PCDH10 was observed in 50% of NSCLC tissue and lung cancer cell lines [32]. The difference was probably because the present study was limited to patients with pathological stage I NSCLC. In this study, PCDH10 methylation was found to be an independent prognostic factor for RFS and DSS but not for OS by Cox proportional regression analysis, primarily because the proportion of death that was nondisease related was relatively high in this cohort. Additionally, it is unclear whether adjuvant chemotherapy for patients with PCDH10 methylation generates a better prognosis, because the number of patients in this study was limited and little is known about the association between PCDH10 and chemoresponse. We demonstrated that marked differences were not found in the patterns of recurrence between patients with and without PCDH10 methylation in spite of the difference in the risk of recurrence according to PCDH10 methylation status. It might be inferred from these data that PCDH10 possesses multifunctional tumorsuppressing effects in NSCLC. Although it is known that the prognosis of patients who underwent wedge resection is generally worse than that of patients who underwent anatomical resection (lobotomy or segmentectomy) with lymph node dissection, the subgroup analysis according to the operative procedure revealed that PCDH10 methylation status was a significant prognostic factor in the both groups of different operative procedures. Future studies using larger sample sizes, including other stages of NSCLC are necessary to more comprehensively elucidate the biological effects of PCDH10 in NSCLC, and to determine whether PCDH10 methylation status provides clinical information relevant to tailored adjuvant therapy and the postoperative surveillance strategy. Furthermore, since the clinical benefit of DNA demethylating agents is under investigation, clinical trials that assess the benefit of both demethylating agents and standard adjuvant chemotherapy are warranted.
There were several limitations to this study. First, this study was based on the data of patients in a single institution. The number of patients was relatively small, and the mean follow-up time was less than 5 years. Patients who did not undergo systemic lymph node dissection were included, although operative procedure was not an independent prognostic factor and did not significantly impact the recurrence pattern in this cohort. Additionally, the specific mechanism of PCDH10 methylation for NSCLC progression was not elucidated. Further investigations using larger sample sizes are needed to determine how PCDH10 methylation contributes to patient outcome.
In summary, this study demonstrated that promoter methylation of PCDH10 plays a significant role in NSCLC progression and might be a promising