Variations in HSPA 1 B at 6 p 21 . 3 are Associated with Lung Cancer Risk and Prognosis in Chinese Populations

Authors' Affiliations: Institute of Occupational Medicine and Ministry of Education Key Lab for Environment and Health, School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China; Department of Etiology and Carcinogenesis, Cancer Institute and Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China; Department of Epidemiology and Biostatistics, Cancer Center of Nanjing Medical University, Nanjing, China; Department of Oncology, Wuhan Iron and Steel (Group) Corporation Staff-Worker Hospital, Wuhan, China; and Cancer Center, Union Hospital, Huazhong University of Science and Technology, Wuhan, China.


Introduction
Lung cancer is a leading cause of cancer-related deaths in the world (1).Although tobacco smoking accounts for more than 80% of all cases (2), only around 15% of all heavy smokers develop lung cancer, indicating that genetic factors may also play a crucial role in susceptibility to the disease.To date, prognosis for patients with non-small cell lung carcinoma (NSCLC) remains poor, with an overall 5-year survival rate of only 10% to 15% (1).Despite intense efforts in the past decades, researchers have still not identified specific biomarkers for lung cancer risk assessment and prognosis prediction.
Published genome-wide association studies (GWAS) in Caucasians have mapped a lung cancer susceptibility locus to chromosome 6p21.3,a gene-rich region of the major histocompatibility complex (MHC; refs.[3][4][5].The reported significant variants (rs3117582, rs3131379, and rs4324798) at 6p21.3 do not exist in Asians according to the HapMap project, and whether single nucleotide polymorphisms (SNP) in this region are related to lung cancer risk in Chinese populations is unknown.
The 3 Hsp70 genes (HSPA1L, HSPA1A, and HSPA1B), encoding 3 highly homologous Hsp70 proteins (Hsp70-1, Hsp70-2, and Hsp70-hom, respectively), are located parallel to each other within the MHC class-III region at 6p21.3,only approximately 50-kb away from rs3131379 (3).HSPs, and Hsp70 in particular, are the major stress-inducible proteins that function as key components in regulating cellular homeostasis, probably through their molecular chaperone activities to inhibit accumulation of protein aggregates (6,7) or through their cytoprotective properties to protect against DNA damage (8), thus promoting cell survival under stressful stimuli (9,10).Accumulating evidence indicates their potential associations with tumorigenesis and clinical outcomes of patients with cancer (11,12).A case-control study nested in a large-scale Japanese cohort indicated that serum Hsp70 level might be an important biomarker for lung cancer risk (13).Because of its peptide chaperone capacity and its ability to bind to professional antigen-presenting cells (APC), the immunologic properties of Hsp70 indicate implications of its potential clinical applications as personalized vaccines for human cancers (14).
Because of the vital role of Hsp70 in cancer development and survival progress and the results from GWAS in Caucasian populations that identified risk alleles at 6p21.3 near HSPA11L-HSPA1A-HSPA1B (3), we hypothesized that genetic variations in HSPA1L-HSPA1A-HSPA1B may modulate Hsp70 expression levels and/or protein structures and, thus, influence lung cancer susceptibility and prognosis in Chinese populations.

Case-control study to identify risk variants
In this study, we carried out 3 independent case-control analyses.The discovery set included 1,152 patients with lung cancer (recruited from Union Hospital Cancer Center, Wuhan Steel Group/Corporation Staff-Worker Hospital, and Wuhan Zhongnan Hospital from January 2003 to February 2009) and 1,152 age-(AE5 years) and sex frequency-matched controls (randomly selected from 4,073 individuals participating in a community screening program in the same regions during the same period; ref. 15).Data from patients with lung cancer and controls from northern China were used to validate the identified risk variants (validation set) and it included an additional 1,781 patients with lung cancer (recruited at the Cancer Hospital, Chinese Academy of Medical Sciences, between July 2000 and November 2008) as well as 1,038 age-(AE5 years) and sex-matched controls (cancer-free individuals selected from a community nutritional survey of 6,450 individuals in the same region during the same period; ref. 16).In addition, 500 newly enrolled patients with lung cancer (from Union Hospital Cancer Center and Wuhan Steel Group/Corporation Staff-Worker Hospital) and 500 age-(AE5 years) and sex frequencymatched controls were chosen for further fine-mapping analysis.In addition, these 500 control subjects were randomly selected from the 4,073 individuals used for selecting controls in the discovery set, and they were independent of the 1,152 controls used in the discovery set.

Follow-up study to identify prognostic variants
For the survival analysis, we followed up 450 patients with primary lung cancer of Wuhan Steel Group/Corporation Staff-Worker Hospital as the discovery set.These patients were admitted to the Department of Oncology of Wuhan Iron and Steel Group/Corporation Staff-Worker Hospital and had sufficient follow-up data available (15).Patients enrolled in this hospital lived in the same region and had a similar socioeconomic status.Of the 450 patients, 4 (0.1%) patients did not have TNM stage classification of tumors.Among the remaining 446 patients with completed follow-up/clinical information, 28 (6.3%) were patients with small cell lung cancer (SCLC) and 418 (93.7%) were patients with NSCLC.There were 88 (21.1%) patients with early NSCLC (stage I or II) and 330 (78.9%) patients with advanced NSCLC (stage III or IV).Eight of the patients with advanced NSCLC were lost to follow-up.As we know, the survival outcomes of patients with NSCLC lung cancer can be affected by many factors, including genetic background, treatments, and socioeconomic status.To minimize the bias due to patient selection, inconsistency of therapies, and patient economic status among different hospitals, we limited our survival analysis to the 330 patients with advanced NSCLC (stage III or IV) from Wuhan Steel Group/ Corporation Staff-Worker Hospital.
In addition, 353 patients with advanced NSCLC from the Cancer Hospital of Jiangsu Province and the First Affiliated Hospital of Nanjing Medical University were used as the validation set (17).Among these 353 patients, only 331 patients with sufficient DNA data were evaluated.In this data set, 300 patients received chemotherapy.Among these patients, 142 (47.3%) received a gemcitabine/cisplatin regimen, 14 (4.7%) received a vinorelbine/cisplatin regimen, 36 (12.0%) received a docetaxel/cisplatin regimen, 13 (4.3%) received an etoposide/ cisplatin regimen, and 95 (31.7%) received chemotherapy that was a combination of the above regimens.The TNM stage classification was evaluated by medical oncologists according to the Staging Manual of the AJCC/UICC (2003; ref. 18).Patients were followed up by telephone calls every 3 months.Survival duration was calculated from the date when patients first received confirmed diagnoses to the date of the last followup or death.Dates of death were obtained from inpatient and outpatient records or from the patients' families through telephonic follow-up.In the discovery set, the mean and median follow-up periods were 17.5 and 12.2 months, respectively, whereas the follow-up period ranged from 1.0 to 84.0 months.In the validation set, the mean and median follow-up periods were 19.6 and 16.0 months, respectively, whereas the follow-up period ranged from 1.0 to 70.2 months.
All subjects in this study were unrelated ethnic Han Chinese, and general information about all the subjects was collected through personal interviews.Subjects who had smoked an average of less than 1 cigarette per day and for less than 1 year in their lifetimes were defined as nonsmokers; those who stopped smoking more than 1 year previously were considered former smokers; and those who were still smoking in the previous year were defined as current smokers.This study was approved by the Institutional Review Board of Tongji Medical College, the Chinese Academy of Medical Sciences Cancer Institute, and Nanjing Medical University.
The risk polymorphisms [rs3117582 (chr6: 31728499) and rs3131379 (chr6: 31829012)] reported by GWAS in Caucasians did not exist in Asians.Because rs3117582 was located in the block from chr6: 31722081 to 31733520, we further carried out a fine-mapping analysis from this block to the HSPA1L-HSPA1A-HSPA1B cluster.The same HapMap-CHB database and selection criteria described above were used to select 29 tagSNPs at 6p21.3 (from 31722081 to 31906010; Supplementary Fig. S2).

Genotyping
Six tagSNPs located in the HSPA1L-HSPA1A-HSPA1B cluster were genotyped by TaqMan methods, and the additional 24 tagSNPs at 6p21.3 were genotyped by the MassArray system (Sequenom).However, genotyping of rs6905572 failed because no appropriate primers could be designed for this SNP.In the 331 patients with advanced NSCLC in the validation set, 6 samples (1.8%) failed in genotyping of rs2763979 and 8 samples (2.4%) failed in genotyping of rs6457452 due to their lower DNA quantity.As a result, 325 and 323 patients with advanced NSCLC in the validation set were successfully genotyped for HSPA1B gene rs2763979 and rs6457452, respectively.For all tested samples, a 5% random sample was reciprocally tested, and the reproducibility was 100%.All primers and probes used in this study are listed in Supplementary Table S1.

Luciferase reporter assay
Referring to the transcription initiation site of the HSPA1B gene as "þ1," we constructed 6 reporter plasmids based on the pGL3-Basic vector (Promega), including 4 fragments from À1031 to þ216 according to their haplotypes (rs2763979C/ rs6457452C, rs2763979C/rs6457452T, rs2763979T/rs6457452C, and rs2763979T/rs6457452T, respectively) and 2 core regulation regions from À456 to þ216 containing either the rs6457452C or the rs6457452T allele (Fig. 2A).These were all inserted into KpnI/HindIII enzyme sites of the pGL3-Basic.Primer pairs designed for plasmid constructs and site-specific mutagenesis are listed in Supplementary Table S1.The direction and sequence authenticity of the 6 constructs were validated by restriction analysis and DNA sequencing.
The human bronchial epithelial 16HBE cells were kindly provided by Dr. Yiguo Jiang of the Guangzhou Medical College, China, whereas the human lung carcinoma A549 cells and hepatoma HepG2 Cells were purchased from the China Center of Type Culture Collection in Wuhan, China.The 2 Â 10 4 16HBE, HepG2, and A549 cells were seeded into 96-well plates and co-transfected with 100 ng reporter plasmids and 1 ng pRL-SV40 (Promega) using Lipofectamine-2000 (Invitrogen).The relative luciferase activity (RLA) was determined 24 hours after transfection as previously described (15,20).

Statistical analysis
A 2-phase screening was used to search for tagSNPs in the HSPA1L-HSPA1A-HSPA1B cluster associated with lung cancer risk and survival (Fig. 1).Distributions of numerical data were examined by the 1-sample Komogorov-Smirnov normality

Results
Associations between tagSNPs of the HSPA1L-HSPA1A-HSPA1B cluster and lung cancer risk Distributions of selected characteristics for subjects in the discovery and validation sets and fine-mapping analyses are listed in Table 1.As shown in Table 2, in the discovery set, compared with the HSPA1B rs6457452CC genotype, subjects with the rs6457452CT or combined rs6457452CT þ TT genotypes had a 1.39-fold increased risk of lung cancer (P ¼ 0.013 and 0.010, respectively).There was a dose-response effect of the increasing number of the rs6457452T alleles on increased lung cancer risk (P trend ¼ 0.026).However, no associations were found between rs2075800, rs2227956, rs1008438, rs1043618, and rs2763979 and lung cancer risk (Table 2).In the validation set, logistic regression analysis indicated that compared with the HSPA1B rs6457452CC genotype, subjects with rs6457452CT or rs6457452CT þ TT genotypes had a significantly increased risk of lung cancer (adjusted OR ¼ 1.34 and 1.32; P ¼ 0.009 and 0.012, respectively).Pooled analysis of the discovery and validation sets showed that the adjusted ORs were 1.41 for the rs6457452CT genotype (P ¼ 3.0 Â 10 À5 ), and 1.41 for the rs6457452CT þ TT genotype (P ¼ 2.8 Â 10 À5 ), compared with the rs6457452CC genotype, respectively (Table 2).We did not observe a significant interaction of age at diagnosis, sex, and smoking habits with HSPA1B rs6457452CT þ TT genotypes on the risk of lung cancer in the pooled analysis (P ¼ 0.231, 0.982, and 0.291, respectively).

Fine-mapping analysis on chromosome 6p21.3
A total of 28 tagSNPs at chr6: 31722081 to 31906010 were successfully genotyped.The rs3115672 was not detected in any subject.The allele distributions of rs3130617 and rs805923 deviated from HWE among controls (P < 0.05), and were excluded from subsequent analysis.Analysis of the remaining 25 tagSNPs revealed that only the rs6457452T allele on HSPA1B 5 0 -UTR was significantly associated with an increased lung cancer risk (P ¼ 0.013; Supplementary Table S2).

HSPA1L-HSPA1A-HSPA1B tagSNPs and survival of advanced NSCLC patients
The demographic and clinical characteristics of patients with lung cancer in the survival discovery and validation sets are presented in Table 3.In the discovery set, in which the mean age was 63.25 AE 10.07 years, 257 (77.9%) patients died of lung cancer, 89 (27.0%) received surgical operations, 255 (77.3%) received chemotherapies, and 152 (46.1%) received radiotherapies.In the validation set, in which the mean age was 59.65 AE 9.57 years, 242 (73.1%) patients died of lung cancer, 152 (45.9%) underwent surgical operations, 300 (90.6%) received chemotherapies, and 123 (37.2%) received radiotherapies.The 5-year survival rates for the 2 sets were 8% and 10%, respectively.The Kaplan-Meier analysis, log-rank test, and univariate Cox analysis revealed that younger patients, patients with earlier stage, or those who had surgical operations and chemotherapy had a significantly longer median survival time (MST) and a decreased risk of death (P < 0.05 in Table 3).There were no significant effects of sex, smoking status, and radiotherapy.
In the discovery set, the Kaplan-Meier method and log-rank test showed that patients with the rs1008438AC or rs1008438CC genotype had a significantly shorter MST than patients with the rs1008438AA genotype (log-rank P ¼ 0.047 and 0.009, respectively; Table 4).However, this SNP was not associated with death risk for patients with advanced NSCLC after adjustment for the confounders.Multivariate proportional hazards regression models and log-rank test revealed that, when compared with the HSPA1B rs2763979CC genotype, the rs2763979CT and rs2763979TT genotypes were associated with poor survival (adjusted HR ¼ 1.43 and 1.80; P ¼ 0.008 and 0.013, respectively) and a shorter MST for patients with advanced NSCLC (log-rank P ¼ 0.011 and 0.031, respectively); patients with the rs6457452CT þ TT genotypes had a 51% higher death risk than those with the rs6457452CC genotype (P ¼ 0.032).The HSPA1B rs2763979C>T and rs6457452C>T polymorphisms were further tested in the validation set.In this data set, when compared with the rs2763979CC genotype, patients with advanced NSCLC with the rs2763979TT genotype had a 1.61-fold higher death risk (P ¼ 0.036).No significant association was found on the rs6457452C>T polymorphism.Pooled analysis of the 2 cohorts showed that, when compared with the rs2763979CC genotype, the rs2763979CT or rs2763979TT genotype had a 1.21-or 1.66fold higher death risk (P ¼ 0.045 and 0.002, respectively), whereas the rs2763979TT genotype was associated with a significantly shorter MST (Table 4; Supplementary Fig. S3).We did not observe a significant interaction of age at diagnosis, stage, surgical operations, and chemotherapy with HSPA1B 2763979CT þ TT genotypes on the death risk among patients with advanced NSCLC in the pooled analysis (P ¼ 0.302, 0.229, 0.971, and 0.574, respectively).

Luciferase reporter activity for HSPA1B rs2763979C>T and rs6457452C>T
Among 4 constructed reporter plasmids with distinct haplotypes, the RLAs driven by promoter containing rs2763979C/ rs6457452T, rs2763979T/rs6457452C, and rs2763979T/ rs6457452T haplotype were significantly lower than that driven by rs2763979C/rs6457452C haplotype (all P < 0.05); the RLA driven by promoter containing rs2763979T/rs6457452T haplotype was significantly lower than that driven by rs2763979C/ rs6457452T or rs2763979T/rs6457452C haplotype (all P < 0.05; Fig. 2B); however, the RLA driven by promoter containing the rs2763979C/rs6457452T or rs2763979T/rs6457452C haplotype showed no significant difference (P > 0.05).In 2 reporter plasmids containing the rs6457452C or rs6457452T allele of HSPA1B rs6457452C>T polymorphism, the rs6457452C allele  Data were calculated by univariate Cox regression analysis.c Others include large cell, bronchioalveolar, mixed cell, undifferentiated and pathologic not otherwise specified carcinomas.
was associated with significantly higher RLA than rs6457452T allele in 3 cell lines (P < 0.05).

Discussion
To our knowledge, this study is the first to show that genetic variants of HSPA1B are associated with lung cancer susceptibility and survival of patients with advanced NSCLC in Chinese populations.In this study, we found that the HSPA1B rs6457452T allele was associated with an increased lung cancer risk in 2 independent lung cancer case-control studies comprising 5,123 Chinese individuals.Fine-mapping analysis confirmed that only this SNP within the surrounding 185-kb at the previously identified chromosome 6p21.3locus (chr6: 31722081 to 31906010) was associated with lung cancer risk.In addition, we observed that the HSPA1B rs2763979TT genotype was associated with poor survival for patients with advanced NSCLC in 2 independent cohorts.Further analysis of the biologic functions for rs2763979C>T and rs6457452C>T by luciferase reporter assay indicated that the rs2763979T and rs6457452T alleles led to a significantly lower expression of Hsp70 in both normal bronchial epithelial and in malignant cancer cells.
Hung and colleagues reported an association between rs4324798 at 6p21.3 and lung cancer risk (P ¼ 5.6 Â 10 À6 ; ref. 5), but this association was not confirmed in a subsequent meta-analysis (21).Wang and colleagues reported that 2 other SNPs at 6p21.3 were significantly associated with lung cancer risk (rs3117582 and rs3131379; ref. 3).However, these 3 SNPs do not exist in Asian populations.In contrast, the published Japanese and Korean GWAS of lung adenocarcinoma did not find 6p21.3 to be a susceptible locus, possibly because the sample size at the discovery stage was relatively small or because this locus does not contribute to carcinogenesis for this specific histology (22,23).Our fine-mapping analysis of 25 tagSNPs from the block containing rs3117582 to HSPA1L-HSPA1A-HSPA1B cluster found that only the HSPA1B rs6457452C>T variant was associated with lung cancer risk.However, the associations with other SNPs in the region investigated were not significant, which may be attributable to the limited sample size.Because the LD region of 6p21.3 is extensive and contains numerous transcripts, dissection and elucidation of the causal variant require additional investigations with larger sample sizes and different ethnic groups.
The HSPA1B þ1267A/G (Gln351Gln, rs1061581) polymorphism was reported to be associated with susceptibility to nasopharyngeal carcinoma (24), non-Hodgkin lymphoma (25), and breast carcinoma (25,26).Data from the Environmental Genome Project showed that the HSPA1B rs1061581 was in LD with rs6457452 (D' ¼ 1.0).Because the rs1061581 is a silent polymorphism, it is likely to be a surrogate marker for the rs6457452 in HSPA1B 5 0 -UTR.In our study, the HSPA1B rs6457452T allele reduced the Hsp70 synthesis levels of normal-status cultured cells.Wang and colleagues previously found that the HSPA1A rs1043618C allele was associated with increased lung cancer risk in 159 patients with lung cancer and 202 controls in a Chinese population, but no significant association was found for HSPA1B rs1061581 and HSPA1L rs2227956 polymorphisms (27).Rusin and colleagues reported that the 1541-1542delGT heterozygous genotype of the constitutive heat shock cognate protein 70 (HSC70) gene was associated with a significantly decreased risk for lung cancer in 96 paired European patients with lung cancer and controls (28).However, the sample sizes for the above 2 studies were small and the results require further validation.
Polycyclic aryl hydrocarbons (PAH) are the main potent procarcinogens in tobacco smoke that can induce transformation of normal cells, causing malignancies in experimental animals (29).Our previous studies showed that Hsp70 could protect human bronchial epithelial cells against DNA damage caused by benzo[a]pyrene (30), and could also protect DNA from genotoxic damage introduced by PAHs contained in coke-oven emissions (31).Because accumulation of damage at different molecular and cellular levels is the underlying mechanism for lung carcinogenesis (32), the increased DNA damage levels caused by reduced Hsp70 levels may enhance the process for genetic instability and promote cancer development.
Cellular studies and animal models have extensively evaluated the functions of Hsp70 in the carcinogenesis process and in cell proliferation and survival.For example, high Hsp70 expression was correlated with poor prognosis of breast cancer and endometrial cancer, but with better prognosis of esophageal, pancreatic, and renal cancers (12).The contradictory prognostic implications of Hsp70 might be attributable to its versatile properties.For example, elevated Hsp70 may confer resistance to chemotherapy and apoptotic removal of tumor cells; in addition, Hsp70 can participate in tumor immunogenicity, because Hsp70 expressed on the surface of tumor cells can elicit tumor cell-specific immunologic responses through chaperoning and presentation of cancer-specific antigens (33,34).Tamura and colleagues found that immunization of mice with lung carcinoma with cancer-derived Hsp70-peptide complex led to retarded progression of primary cancer, a reduced metastasis, and a prolongation of life span (35).In the present study, we hypothesize that the low Hsp70 expression levels driven by the rs2763979T alleles in lung cancer cells may contribute to a reduced the immunogenic response, thus conferring unfavorable survival rates for patients with NSCLC.
Patients with NSCLC are often diagnosed at an advanced stage and have a poor prognosis.Individual heterogeneity is a major reason for distinct responses of treatments and outcomes (36).Recently, using high-throughput expression and genetic microarray platforms, molecular epidemiologic and pharmacogenetic studies have been designed that aim to identify molecular and clinical predictors for outcomes of patients with lung cancer with specific histology, TNM stage, or therapeutic strategy (37)(38)(39).Hsp70 has become one of the most exciting anticancer molecular targets, and the Hsp70antigen complex has proven effective in the treatment of rodent lung cancer in preclinical studies (35) and is now undergoing clinical trials for treatment of human cancers (40).Therefore, newly established molecular biomarkers as well as identifiable clinical and pathologic features in patients will help pave the way for individualized prevention and therapy for lung cancer.
The present study has several strengths and limitations.First, a total of 5,123 Chinese subjects, included in 2 independent panels, and 661 patients with advanced NSCLC, included in 2 case-only survival cohorts, consistently confirmed the associations of HSPA1B rs6457452C>T and rs2763979C>T polymorphisms with lung cancer risk and survival respectively, which provided an improved statistical power and decreased the probability of false positives.Second, the fine-mapping analysis provided additional confirmation for the potential causal variant of rs6457452T allele.Finally, our functional studies offered a biologically plausible explanation for the epidemiologic findings.However, the fine-mapping analysis carried out in our study was not sufficiently extensive.A future, more well-designed, and comprehensive fine-mapping analysis on a longer region surrounding the real hit of 6p21.3, which contains a large number of SNPs, is needed to investigate the real causal variant.Moreover, because all subjects enrolled in this study were ethnic Han Chinese, the associations of these SNPs with lung cancer risk and survival in other ethnic groups also needs further validation by larger prospective studies.
In conclusion, this study provided evidence, for the first time, that the HSPA1B 5 0 -UTR rs6457452T allele may contribute to an increased lung cancer risk, whereas the HSPA1B promoter rs2763979TT genotype may confer poor survival outcomes for patients with advanced NSCLC in Chinese populations.Extensive functional evaluations and additional population-based prospective studies with different ethnic groups and well-designed clinical investigations are warranted to confirm and extend our findings.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.

Figure 1 .
Figure 1.Patients and study strategy.

Figure 2 .
Figure 2. Luciferase reporter assays for the HSPA1B rs2763979C>T and rs6457452C>T polymorphisms.A, schematic representation of HSPA1B gene and 6 reporter gene constructs.B, Luciferase expression of these constructs.The 6 reporter constructs and empty pGL3-Basic were transfected into 16HBE, A549, and HepG2 cells.All constructs were co-transfected with pRL-SV40 to standardize the transfection efficiency.Fold increase of luciferase activity was measured by defining the activity of the empty pGL3-Basic vector as 1.Mean AE SD of relative luciferase activity (RLA) values were from 6 independent transfection experiments, each done in triplicate. b

6 tagSNPs 1,152 patients 1,152 controls HSPA1L: rs2075800, rs2227956 HSPA1A: rs1008438, rs1043618 HSPA1B: rs2763979, rs6457452
regression models after adjusting for the confounders.The proportional hazards assumption was examined by testing interactions between the genotypes and time (all P-values > 0.05).The statistical software packages R and QVALUE were used to calculate false discovery rate.Differences of RLA were compared using one-way ANOVA, Student-Newman-Keul test, and Student t test.Two-sided P < 0.05 was considered significant and the SPSS v.12.0 software (SPSS Inc.) was used for all analyses.

Table 1 .
Distribution of selected characteristics among patients with lung cancer and controls in 3 panels of Chinese populations a Others include large cell, bronchioalveolar, mixed cell, undifferentiated and pathologic not otherwise specified carcinomas.

Table 2 .
Genotype frequencies of HSPA1L-HSPA1A-HSPA1B polymorphisms among patients and controls and their associations with risk of lung cancer in Chinese populations

Table 3 .
Demographic and clinical characteristics of patients with lung cancer in the survival discovery and validation sets Abbreviations: HR, hazard ratio; SCC, squamous cell carcinoma.a Data are for the combined discovery and validation sets.

Table 4 .
Associations between HSPA1L-HSPA1A-HSPA1B genotypes and survival of patients with advanced NSCLCThe Cox regression analysis was adjusted for age, sex, pack-years smoked, histology, TNM stage, surgical operation, chemotherapy, and radiotherapy status. a