MUC1 in lung adenocarcinoma: cross-sectional genetic and serological study

Five hundred and fifty-two patients with previously untreated lung ADC which was pathologically diagnosed and subsequently treated at Hiroshima University Hospital (Hiroshima, Japan) between April 2003 and June 2012 were included in this study (Fig. 1). The inclusion criteria, shown in Fig. 1, were almost identical to those described in our previous studies [15, 16, 18–21]. Disease staging in all 552 cases included computed tomography scans of the chest and abdomen, bone scintigraphy or F-18 fluordeoxyglucose positron emission tomography, and magnetic resonance imaging of the head. The clinical or pathologic TNM stage was determined according to the 7th edition of the TNM classification of malignant tumors [22]. Two hundred and seventy-six HVs who attended health check-ups were also enrolled. The HVs had no history or current diagnosis of malignancy or apparent lung disease. Serum and peripheral venous whole blood samples were obtained from patients and HVs at enrollment with their written informed consents for using their samples and publishing the data. Serum samples were available for 354 of the 552 patients with ADC and for all 276 HVs. Among the 354 patients with ADC, EGFR gene mutation status was known for 304 patients that were included in the serum analyses. In addition, DNA samples were available for 172 patients and these were also included in the genomic analyses (Table 1). This study was approved by the Ethics Committee of Hiroshima University Hospital (IRB M33 and 326) and conducted in accordance with the ethical standards established by the Helsinki Declaration of 1975.

EGFR mutation status was evaluated in 304 patients, for whom serum samples were available, in one of the following sample types: cytology liquid samples derived from bronchoscopy or thoracentesis, paraffin-embedded transbronchial lung biopsy samples, or surgically resected tumor tissues. To evaluate EGFR mutations, the peptide nucleic acid-locked nucleic acid polymerase chain reaction clamp test that detects the G719C, G719S, G719A, L858R, L861Q, and T790 M mutations and seven different exon 19 deletions was used, as previously described [23].

DNA was extracted from peripheral venous whole blood samples using the phenol-chloroform extraction and ethanol precipitation methods, as previously described [24–26]. The rs4072037 genotype was determined by real-time polymerase chain reaction (PCR) using a commercially available SNP genotyping assay (TaqMan SNP Genotyping Assay C 27532642–10; Life Technologies Corp. Carlsbad, CA, USA) and the Applied Biosystems 7500 Fast RT-PCR System (Life Technologies Corp.)

Serum KL-6 levels were measured with a sandwich-type electrochemiluminescence immunoassay on a Picolumi 8220 Analyzer (Sanko Junyaku, Tokyo, Japan), as previously described [15, 16, 20, 27].

The numerical data were presented as the mean ± standard error of the mean. The Mann–Whitney U-test was used to analyze data between 2 groups; for 3 or more groups, the Kruskal-Wallis test followed by multiple comparisons using rank sums was performed [28]. When dividing patients into clinical subgroups according to performance status (PS) or TNM factors, we performed receiver operating characteristic (ROC) analysis and determined the appropriate cut-off point to predict the prognosis of the patients. Survival analysis was performed using the log-rank test and Cox proportional hazards models. Concordance (C)-statistics were used to evaluate and compare the Cox models. To test for deviations from Hardy-Weinberg equilibrium, genotype frequencies were determined by direct counting and the chi-square test or Fisher’s exact test were used as appropriate. Allelic frequencies were calculated based on the genotype frequencies, and the association between the minor G allele of rs4072037 and lung ADC was tested using the chi-square test or Fisher’s exact test. All statistical analyses were performed using SPSS for Windows, version 18.0 (SPSS Inc. Chicago, USA).

To investigate the association between rs4072037 and susceptibility to lung ADC and/or its prognosis, we extracted DNA from peripheral venous blood samples and evaluated the presence of the rs4072037 genotype in 172 ADC patients and 276 HVs. The clinical characteristics of the studied subjects are shown in Table 1. The HVs were predominantly male and relatively young, compared with patients with lung ADC. The characteristics of patients with ADC included in the genomic analyses and those included in the serum analyses were similar. As shown in Table 2, the genotype distributions of rs4072037 were in Hardy–Weinberg equilibrium for both patients and HVs (P = 0.891 and P = 0.979, respectively). Unexpectedly, the genotype distributions of rs4072037 did not differ significantly between patients with ADC and HVs (P = 0.933). This result was consistent even when we limited the patients to those with advanced stage disease (Table 2). In addition, genotypes of rs4072037 showed no significant association with the risk of lung ADC both in dominant and recessive models (Table 3). According to the Kaplan-Meier analysis, the mean survival times for patients with AA, AG, and GG genotypes were 2416.8 ± 257.2 days, 1988.8 ± 232.5 days and 486.0 ± 206.8 days, respectively. Patients with the GG genotype tended to have a poorer prognosis than the other patients, although this was not statistically significant according to the log-rank test (P = 0.173). We also have to pay attention to the fact that there was only 4 patients with the GG genotype.

Next, we investigated the correlation between the rs4072037 genotype and serum KL-6 levels. As shown in Fig. 2a, there was a significant correlation between the rs4072037 genotype and serum KL-6 levels in HVs, which was in line with the results of previous studies [24, 29]. In patients with lung ADC, a significant correlation between the rs4072037 genotype and serum KL-6 levels was observed in those with T1 or T2 (Fig. 2b, P < 0.001), N0 or N1 (Fig. 2c, P = 0.002) and M0 (Fig. 2d, P < 0.001), but not in those with T3 or T4 (Fig. 2b, P = 0.882), N2 or N3 (Fig. 2c, P = 0.616) and M1a or M1b (Fig. 2d, P = 0.501).

To confirm the correlation between serum KL-6 levels and the presence of lung ADC, we compared serum KL-6 levels between 304 patients with ADC and 276 HVs. As shown in Fig. 3a, serum KL-6 levels were significantly higher in patients with ADC than in HVs (880.8 ± 166.3 U/mL and 241.0 ± 5.2 U/mL, P < 0.001). In addition, high serum KL-6 levels were associated with poor PS (P < 0.001, Fig. 3b), and advanced T, N and M factors (P < 0.001, P < 0.001 and P < 0.001, respectively; Fig. 3b). On the other hand, serum KL-6 levels did not differ between patients with ADC with wild-type EGFR and mutant EGFR status (768.4 ± 170.4 U/mL and 1048.5 ± 327.7 U/mL, P = 0.920, Additional file 1).

We performed ROC analysis to evaluate the predictive ability of serum KL-6 in patients with ADC. The area under the curve for 1-year, 3-year, and 5-year survival was 0.698 (95% confidence interval [CI] = 0.618–0.778; P < 0.001, Fig. 4a), 0.686 (95% CI = 0.622–0.749; P < 0.001, Fig. 4b) and 0.656 (95% CI = 0.593–0.719; P < 0.001, Fig. 4c), respectively. The optimal cut-off value for serum KL-6 levels to discriminate between survivors and non-survivors was set at 600 U/mL, in accordance with the ROC analyses. According to the Kaplan-Meier analysis, the mean survival times for patients with higher and lower serum KL-6 levels were 777.8 ± 117.7 days and 2807.6 ± 224.2 days, respectively. Patients with higher serum KL-6 levels had a significantly poorer prognosis than those with lower serum KL-6 levels, as tested by the log-rank test (P < 0.001, Fig. 4d). This result was confirmed when we perform additional log-rank test stratified by median value of serum KL-6 (P < 0.001, Fig. 4e). To further investigate the predictive ability of serum KL-6, we performed Cox proportional hazards regression analysis with stepwise selection including both patients with mutant EGFR and those with wild type EGFR. The initial covariates included age, gender, smoking history, PS, EGFR mutation status, TNM classifications and serum KL-6 levels. As shown in Table 4A, there was a significant association between serum KL-6 levels and prognosis in patients with ADC, independent of the other covariates (PS, EGFR mutation status, and N and M factors). In addition, this result was not changed when we forcibly include age or gender into the covariates (data not shown). Furthermore, comparison of the C-statistics of each prediction model, with or without serum KL-6, demonstrated a significant increase in C-index with the addition of serum KL-6 to the other covariates (Table 4B).