ATM gene polymorphisms are associated with poor prognosis of non-small cell lung cancer receiving radiation therapy

General characteristics of the study population

We enrolled 720 NSCLC patients, including 335 males (46.5%) and 385 females (53.5%). We obtained their clinical and follow-up data to evaluate DFS and OS outcomes. Among these 720 patients, 497 (69.0%) were above 60 years, and 223 (31.0%) were ≤ 60 years of age. Furthermore, these patients included 370 (51.4%) smokers, 158 (21.9%) with squamous cell carcinoma, 358 (49.7%) with adenocarcinomas, and 204 (28.3%) with other cancer types. Moreover, 333 patients were diagnosed with stage IIIA and 387 with stage IIIB NSCLC. The cohort included 380 patients that received radiation therapy alone, whereas, the remaining 340 patients received chemoradiation therapy. While 357 patients received intensity-modulated radiation therapy (IMRT), 363 patients received three-dimensional conformal radiotherapy (3D-CRT) or another kind of radiotherapy. The median radiation dose received by this cohort was 65 Gy (419 patients received < 65 Gy and 419 received ≥ 65 Gy). The four ATM SNPs analyzed in this study were rs664143 (GG, GA and AA genotypes at nucleotide positions 148, 381 and 191, respectively), rs664677 (TT, TC, and CC genotypes at nucleotide positions 265, 282, and 173, respectively), rs189037 (AA, AG, and GG genotypes at nucleotide positions 172, 283, and 165, respectively), and rs373759 (TT, TC, and CC genotypes at nucleotide positions 227, 294 and 199, respectively). The baseline clinical characteristics of the study population are listed in Table 1.

Clinical characteristics and prognosis of NSCLC patients

The median follow-up time was 36.4 months (range: 2.9–94.6 months) and 293 patients died before the last follow-up. The median DFS was 24.5 months, and 383 patients showed disease progression. Kaplan–Meier survival curve analysis showed that gender (P = 0.003), smoking (P < 0.001), TNM stage (P < 0.001), treatment modality (P < 0.001), and radiation dose (P = 0.001) were significantly associated with DFS and OS. Patients with stage IIIB tumors showed significantly shorter DFS (HR = 1.67, 95%CI = 1.36–2.05, P < 0.001) and OS (HR = 1.39, 95%CI = 1.12–1.76, P = 0.006) than stage IIIA lung cancer patients. The survival time was significantly longer for patients treated with chemoradiation therapy than those treated with radiation therapy alone (40.4 vs. 22.0 months, P < 0.001). Moreover, patients treated with radiation therapy alone had worse survival outcomes than those treated with combination of chemo and radiation therapy (HR = 1.34, 95%CI = 1.16–1.69, P = 0.014). The OS rates for smoking NSCLC patients was significantly shorter than the non-smoking NSCLC patients (HR = 1.69, 95%CI = 1.69-1.33–2.15, P < 0.001). Patients that received a radiation dose < 65 Gy showed worse DFS rates (HR = 1.42, 95%CI = 1.15–1.75, P = 0.001) and OS (HR = 1.29, 95%CI = 1.25–1.51, P = 0.013) than NSCLC patients that received a radiation dose ≥ 65 Gy.

ATM SNPs and survival outcomes of NSCLC patients

Univariate analyses showed that two ATM SNPs, rs664143 and rs189037, as shown in Supplementary Figure 1, were significantly associated with DFS and OS, but, the other two ATM SNPs, rs664677 and rs373759, did not show any significant correlation with survival outcomes (Table 2).

The patients with ATM rs664143 AA and GA genotypes showed significantly shorter DFS (median survival time or MST (month): 22.7 vs. 24.3 vs. 33.4; P = 0.008; Figure 1A) and OS (MST: 33.6 vs. 37.8 vs. 52.9 months; P = 0.029; Figure 1C) than patients with the rs664143 GG genotype. After adjusting for clinical parameters, Cox regression analysis showed that patients with ATM rs664143 AA or GA genotypes showed worse survival outcomes than those with the rs664143 GG phenotype (AA vs. GG, HR: 1.38, 95%CI = 1.01-1.19, P = 0.034; GA vs. GG, 1.41, 95%CI = 1.05–1.89, P = 0.023). The rs664143 G allele showed dominant effect than the rs664143 A allele (GA/AA vs. GG: P = 0.002 for DFS, Figure 1B; P = 0.023 for OS, Figure 1D). According to multivariate Cox regression analysis, patients with rs664143 GA and rs664143 AA genotypes showed worse DFS (HR = 1.40, 95%CI = 1.05–1.86, P = 0.021) and OS (HR = 1.28, 95%CI = 1.12–1.78, P = 0.040) than those with the rs664143 GG phenotype. Patients with the ATM rs189037 GG and rs189037 AG genotypes showed significantly shorter DFS (MST: 24.5 vs. 23.8 vs. 31.0 months, P = 0.007, Figure 2A) and OS (MST: 33.5 vs. 34.9 vs. 43.5, P < 0.001, Figure 2C) than those with the rs189037 AA genotype. After adjusting for clinical parameters, Cox regression analysis showed that patients with rs189037 GG or AG genotypes had worse DFS (GG vs. AA: HR = 1.32, 95%CI = 1.17–1.91, P = 0.012; AG vs. AA: HR = 1.48, 95%CI = 1.18–2.03, P = 0.014; Table 2) and OS (GG vs. AA: HR = 1.23, 95%CI = 1.13–1.83, P = 0.029; AG vs. AA: HR = 1.76, 95%CI = 1.21–2.45, P = 0.001) than those with rs189037 AA. The rs189037 G allele exhibited dominant effect than the rs189037 A allele (DFS: P = 0.004, Figure 2B; OS: P = 0.001, Figure 2D). Multivariate Cox regression analysis showed worse survival outcomes for patients with the rs189037 AG and GG genotypes than those with the rs189037 AA genotype (AG/GG vs. AA: DFS, HR = 1.44, 95%CI = 1.06–1.95, P = 0.019; OS, HR = 1.16, 95%CI = 1.16–1.17–2.21, P = 0.004; Table 2).

We also compared the association between the number of risk alleles (rs664143 and 189037) and survival outcomes, DFS and OS. The results showed that patients with the risk alleles were associated with worse DFS and OS compared to those without a risk allele (Table 2). As shown in Table 3, three SNPs (rs664143, rs664677, and rs373759) showed a prior false-positive probability of 0.1.

Subgroup analysis of ATM SNPs and NSCLC prognosis

We performed subgroup analysis with clinical factors such as smoking, histology, tumor stage, treatment, and radiation dose that are significantly associated with DFS and OS. The subgroup analysis showed that rs664143 and rs189037 were significantly associated with DFS and OS in both IIIA and IIIB stages (Table 4 and Table 5). NSCLC patients with ATM rs664143 GA and AA genotypes showed worse DFS (stage IIIA: adjusted HR = 1.34, 95%CI = 1.18–2.05, P = 0.022; stage IIIB: adjusted HR = 1.52, 95%CI = 1.04–2.22, P = 0.030) and OS (stage IIIA: adjusted HR = 1.29, 95%CI = 1.18–2.09, P = 0.028; stage IIIB: adjusted HR = 1.51, 95%CI = 1.17–2.37, P = 0.021) than patients with the rs664143 GG genotype. Stage IIIA and IIIB NSCLC patients with rs189037 AG/GG genotypes showed significantly shorter median survival times than those with the rs189037 AA genotype (Stage IIIA: 39.0 vs. 58.6 months; stage IIIB: 27.1 vs. 43.7 months; Table 4). Cox regression analysis after adjustment showed shorter DFS and OS for the rs189037 AG/GG genotypes in than those with the rs189037 AA genotype in stage IIIA NSCLC (DFS: adjusted HR, 1.43; 95% CI = 1.31–2.22, P = 0.037; OS: adjusted HR, 1.62;(95%CI = 1.02–2.57, P = 0.042; Table 5) and stage IIIB NSCLC (DFS: adjusted HR,1.55; 95% CI = 1.05–2.38, P = 0.029; OS: adjusted HR, 1.69; 95% CI = 1.09–2.65, P = 0.020; Table 5). Moreover, patients with rs664143 and rs189037 SNPs that were smokers and received radiation therapy (RT) with a radiation dose > 65 Gy showed worse survival outcomes than patients that were non-smokers and received chemo-radiation therapy (CRT) with a radiation dose ≤65 Gy, respectively (Tables 4, 5). Furthermore, based on histology, the rs664143 and rs189037 gene polymorphisms were associated with DFS and OS among the patients with adenocarcinomas (ADC) as shown in Tables 4, 5.

Study population

This two-center follow-up study was conducted at Yantai Yuhuangding Hospital, the Affiliated Hospital of Qingdao University and Tangshan People’s Hospital from January 2009 to December 2017. The inclusion criteria were: (1) newly diagnosed Han Chinese NSCLC patients that were confirmed by biopsy with TNM stage IIIA–IIIB NCSLC tumors; (2) availability of follow-up clinical data, and (3) patients underwent radiotherapy or chemoradiotherapy. Patients that (1) underwent surgery or stereotactic ablative radiation therapy, and (2) with an history of lung cancer and recurrent disease, severe cardiovascular diseases, cerebral apoplexy, or depression were excluded from the study. Tumors were classified based on histology according to the World Health Organization system for NSCLC. The tumor staging was according to the revised American Joint Committee on Cancer/Union for International Cancer Control (AJCC/UICC 7^th Edition). This study was approved by the Ethics Committee of Yantai Yuhuangding Hospital, the Affiliated Hospital of Qingdao University, and conducted in accordance with the World Medical Association Declaration of Helsinki. We obtained written informed consent from all study subjects.

Radiation therapy

All patients underwent computed tomography (CT) scans of 5 mm slice thickness in a supine position with hands crossed before placing the forehead and thermoplastic body model fixation. The scans were performed from the mastoid process to the bottom of the lung and the images were transmitted over the network to a three-dimensional (3-D) treatment planning system. The target area delineation was according to ICRU50 and ICRU62 report guidelines. The gross tumor volume (GTV) was defined as the tumor volume seen below the lung window. The GTV node (GTVnd) was defined as a metastatic lymph node seen below the mediastinal window. The characteristics of the GTVnd were: short diameter > LCM; multiple fusions; necrosis or envelope invasion; and confirmed by positron emission tomography (PET) or mediastinoscopy. For concurrent chemoradiotherapy or radiotherapy alone, the double-lung V20 was limited to < 28% and < 30%, respectively, and adjusted according to the patient’s physical condition, age, complications, and basic lung-function parameters. The other tissue limits included a maximum dose of the spinal cord of < 45 Gy, a cardiac V40 of < 30%, and an esophageal V50 of < 40–50%. A physician confirmed the treatment plan developed by the radiologist. The plan was further verified by calibration on the CT simulation positioning machine. Finally, the treatment was performed using a Vanaii linear accelerator.

Clinical data and follow-up

We obtained clinical data from the medical records of the patients for parameters such as age at treatment, gender, smoking, histologic type, TNM stage, treatment method, radiation technique, and radiation dose. The survival data was collected by follow-up through the telephone or from the outpatient medical records. The patients underwent physical examinations, CT or PET/CT, laboratory tests, and lung function evaluations at 6 weeks after therapy. The follow-up interval was every 3 months for 2 years, and 6 months thereafter. The primary follow-up outcome was overall survival (OS). OS was defined as the time period from the start of treatment to the last follow-up or death. We also evaluated disease-free survival (DFS), which was defined as the time period from the start of treatment until the date of the first local recurrence or metastasis at the last follow-up. Patients without progression were censored at the last follow-up date.

Selection and genotyping of ATM SNPs

We first searched for the ATM gene SNPs in the dbSNP database (https://www.ncbi.nlm.nih.gov/snp/) using a minor allele frequency of > 5%. Next, we selected tagSNPs with r² > 0.8 among the Han Chinese population using the International HapMap project database and identified four SNPs (rs664143, rs664677, rs189037, and rs373759) with moderate linkage disequilibrium by analyzing the 1000GENOMES project database.

We obtained 5 mL of fasting venous blood in ethylenediaminetetraacetic acid (EDTA)-coated tubes and isolated the genomic DNA from peripheral blood leukocytes using the Biospin Whole Blood Genomic DNA Extraction Kit (Bioer Technology Co., Ltd., China) according to the manufacturer’s instructions. The samples were stored at −20°C.

The SNP sequences were PCR amplified using primer sequences that were designed using Primer Premier 5.0, and synthesized by Sangon Biotech (Shanghai, China) as shown in Supplementary Table 1. The PCR protocol was: initial denaturation cycle at 95°C for 7 min; 35 cycles of denaturation at 95°C for 1 min, annealing at 56°C for 1 min, and extension at 72°C for 1 min; final extension cycle at 72°C for 10 min. The PCR products were separated by 2% agarose gel electrophoresis, purified by ExoSAP-IT (USB Corp., Cleveland, OH, USA), sequenced in an Applied Biosystems 3730xl automated sequencer (Applied Biosystems, Foster City, CA, USA), and analyzed using the Vector NTI software.

Statistical analysis

The continuous variables were converted into categorical variables based on the mean age or median dose. The median OS and DFS were determined by Kaplan-Meir survival curve analyses using the log-rank test. The relationship between clinical parameters, ATM SNPs and survival parameters, OS and DFS, was determined using univariate and multivariate cox proportional hazards regression models. The multivariate Cox proportional regression model was adjusted for age, gender, smoking, histology, stage, treatment, radiation technique, and dose and the corresponding hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated. Subgroup analyses were performed for significant variables such as smoking, tumor stage, treatment, and radiation dose. The false-positive probability analysis was conducted by assuming that the HR for a risk allele was 1.5 and the HR for a protective allele was 1/1.5 times below the prior probability of 0.01 for each SNP. A significant association was defined as a false-positive value of < 0.20. Statistical significance was defined as a two-sided P-value < 0.05. Statistical analysis was performed using the SPSS 23.0 software.