Identification of genetically predicted DNA methylation markers associated with non-small cell lung cancer risk among 34,964 cases and 448,579 controls.

BACKGROUND: Although the associations between genetic variations and lung cancer risk have been explored, the epigenetic consequences of DNA methylation in lung cancer development are largely unknown. Here, the genetically predicted DNA methylation markers associated with non-small cell lung cancer (NSCLC) risk by a two-stage case-control design were investigated. METHODS: The genetic prediction models for methylation levels based on genetic and methylation data of 1595 subjects from the Framingham Heart Study were established. The prediction models were applied to a fixed-effect meta-analysis of screening data sets with 27,120 NSCLC cases and 27,355 controls to identify the methylation markers, which were then replicated in independent data sets with 7844 lung cancer cases and 421,224 controls. Also performed was a multi-omics functional annotation for the identified CpGs by integrating genomics, epigenomics, and transcriptomics and investigation of the potential regulation pathways. RESULTS: Of the 29,894 CpG sites passing the quality control, 39 CpGs associated with NSCLC risk (Bonferroni-corrected p ≤ 1.67 × 10-6 ) were originally identified. Of these, 16 CpGs remained significant in the validation stage (Bonferroni-corrected p ≤ 1.28 × 10-3 ), including four novel CpGs. Multi-omics functional annotation showed nine of 16 CpGs were potentially functional biomarkers for NSCLC risk. Thirty-five genes within a 1-Mb window of 12 CpGs that might be involved in regulatory pathways of NSCLC risk were identified. CONCLUSIONS: Sixteen promising DNA methylation markers associated with NSCLC were identified. Changes of the methylation level at these CpGs might influence the development of NSCLC by regulating the expression of genes nearby. PLAIN LANGUAGE SUMMARY: The epigenetic consequences of DNA methylation in lung cancer development are still largely unknown. This study used summary data of large-scale genome-wide association studies to investigate the associations between genetically predicted levels of methylation biomarkers and non-small cell lung cancer risk at the first time. This study looked at how well larotrectinib worked in adult patients with sarcomas caused by TRK fusion proteins. These findings will provide a unique insight into the epigenetic susceptibility mechanisms of lung cancer.

SNP characteristics and validation success in genome wide association studies.

Genome wide association studies (GWASs) have identified tens of thousands of single nucleotide polymorphisms (SNPs) associated with human diseases and characteristics. A significant fraction of GWAS findings can be false positives. The gold standard for true positives is an independent validation. The goal of this study was to identify SNP features associated with validation success. Summary statistics from the Catalog of Published GWASs were used in the analysis. Since our goal was an analysis of reproducibility, we focused on the diseases/phenotypes targeted by at least 10 GWASs. GWASs were arranged in discovery-validation pairs based on the time of publication, with the discovery GWAS published before validation. We used four definitions of the validation success that differ by stringency. Associations of SNP features with validation success were consistent across the definitions. The strongest predictor of SNP validation was the level of statistical significance in the discovery GWAS. The magnitude of the effect size was associated with validation success in a non-linear manner. SNPs with risk allele frequencies in the range 30-70% showed a higher validation success rate compared to rarer or more common SNPs. Missense, 5'UTR, stop gained, and SNPs located in transcription factor binding sites had a higher validation success rate compared to intergenic, intronic and synonymous SNPs. There was a positive association between validation success and the level of evolutionary conservation of the sites. In addition, validation success was higher when discovery and validation GWASs targeted the same ethnicity. All predictors of validation success remained significant in a multivariate logistic regression model indicating their independent contribution. To conclude, we identified SNP features predicting validation success of GWAS hits. These features can be used to select SNPs for validation and downstream functional studies.

Human genes differ by their UV sensitivity estimated through analysis of UV-induced silent mutations in melanoma.

We hypothesized that human genes differ by their sensitivity to ultraviolet (UV) exposure. We used somatic mutations detected by genome-wide screens in melanoma and reported in the Catalog Of Somatic Mutations In Cancer. As a measure of UV sensitivity, we used the number of silent mutations generated by C>T transitions in pyrimidine dimers of a given transcript divided by the number of potential sites for this type of mutations in the transcript. We found that human genes varied by UV sensitivity by two orders of magnitude. We noted that the melanoma-associated tumor suppressor gene CDKN2A was among the top five most UV-sensitive genes in the human genome. Melanoma driver genes have a higher UV-sensitivity compared with other genes in the human genome. The difference was more prominent for tumor suppressors compared with oncogene. The results of this study suggest that differential sensitivity of human transcripts to UV light may explain melanoma specificity of some driver genes. Practical significance of the study relates to the fact that differences in UV sensitivity among human genes need to be taken into consideration whereas predicting melanoma-associated genes by the number of somatic mutations detected in a given gene.

Tumor somatic mutations also existing as germline polymorphisms may help to identify functional SNPs from genome-wide association studies.

We hypothesized that a joint analysis of cancer risk-associated single-nucleotide polymorphism (SNP) and somatic mutations in tumor samples can predict functional and potentially causal SNPs from GWASs. We used mutations reported in the Catalog of Somatic Mutations in Cancer (COSMIC). Confirmed somatic mutations were subdivided into two groups: (1) mutations reported as SNPs, which we call mutational/SNPs and (2) somatic mutations that are not reported as SNPs, which we call mutational/noSNPs. It is generally accepted that the number of times a somatic mutation is reported in COSMIC correlates with its selective advantage to tumors, with more frequently reported mutations being more functional and providing a stronger selective advantage to the tumor cell. We found that mutations reported ≥10 times in COSMIC-frequent mutational/SNPs (fmSNPs) are likely to be functional. We identified 12 cancer risk-associated SNPs reported in the Catalog of published GWASs at least 10 times as confirmed somatic mutations and therefore deemed to be functional. Additionally, we have identified 42 SNPs that are tightly linked (R2 ≥ 0.8) to SNPs reported in the Catalog of published GWASs as cancer risk associated and that are also reported as fmSNPs. As a result, 54 candidate functional/potentially causal cancer risk associated SNPs were identified. We found that fmSNPs are more likely to be located in evolutionarily conserved regions compared with cancer risk associated SNPs that are not fmSNPs. We also found that fmSNPs also underwent positive selection, which can explain why they exist as population polymorphisms.

Downstream targets of GWAS-detected genes for breast, lung, and prostate and colon cancer converge to G1/S transition pathway.

Genome-wide association studies (GWASs) identified over 500 single nucleotide polymorphisms (SNPs) influencing cancer risk. It is logical to expect the cancer-associated genes to cluster in pathways directly involved in carcinogenesis, e.g. cell cycle. Nevertheless, analyses of the GWAS-detected cancer risk genes usually show no or weak enrichment by known cancer genes.We hypothesized that GWAS-detected cancer risk-associated genes function as upstream regulators of the genes directly involved in carcinogenesis. We have analyzed four common cancers: breast, colon, lung, and prostate. To identify downstream targets of GWAS-detected cancer risk genes we used MedScan, which is a text mining tool offered by PathwayStudio. We also used data on protein/protein interactions reported by BioGRID database. Among all identified targets we have selected common downstream targets. A gene was considered a common downstream target if it was a downstream target for at least three GWAS-detected genes for a given cancer type. Common downstream targets were identified separately for each cancer type. We found that common downstream targets for all four cancer types were enriched by cell cycle genes, more specifically, the genes involved in G1/S transition. Common downstream targets for bipolar disorder, Crohn's disease, and type 2 diabetes did not show G1/S transition enrichment.The results of this analysis suggest that many cancer risk genes function as upstream regulators of the genes directly involved in G1/S transition and exert their risk effects by reducing threshold for G1/S transition, elevating the background level of cell proliferation and cancer risk.

Gene characteristics predicting missense, nonsense and frameshift mutations in tumor samples.

BACKGROUND: Because driver mutations provide selective advantage to the mutant clone, they tend to occur at a higher frequency in tumor samples compared to selectively neutral (passenger) mutations. However, mutation frequency alone is insufficient to identify cancer genes because mutability is influenced by many gene characteristics, such as size, nucleotide composition, etc. The goal of this study was to identify gene characteristics associated with the frequency of somatic mutations in the gene in tumor samples. RESULTS: We used data on somatic mutations detected by genome wide screens from the Catalog of Somatic Mutations in Cancer (COSMIC). Gene size, nucleotide composition, expression level of the gene, relative replication time in the cell cycle, level of evolutionary conservation and other gene characteristics (totaling 11) were used as predictors of the number of somatic mutations. We applied stepwise multiple linear regression to predict the number of mutations per gene. Because missense, nonsense, and frameshift mutations are associated with different sets of gene characteristics, they were modeled separately. Gene characteristics explain 88% of the variation in the number of missense, 40% of nonsense, and 23% of frameshift mutations. Comparisons of the observed and expected numbers of mutations identified genes with a higher than expected number of mutations- positive outliers. Many of these are known driver genes. A number of novel candidate driver genes was also identified. CONCLUSIONS: By comparing the observed and predicted number of mutations in a gene, we have identified known cancer-associated genes as well as 111 novel cancer associated genes. We also showed that adding the number of silent mutations per gene reported by genome/exome wide screens across all cancer type (COSMIC data) as a predictor substantially exceeds predicting accuracy of the most popular cancer gene predicting tool - MutsigCV.

Allelic Spectra of Risk SNPs Are Different for Environment/Lifestyle Dependent versus Independent Diseases.

Genome-wide association studies (GWAS) have generated sufficient data to assess the role of selection in shaping allelic diversity of disease-associated SNPs. Negative selection against disease risk variants is expected to reduce their frequencies making them overrepresented in the group of minor (<50%) alleles. Indeed, we found that the overall proportion of risk alleles was higher among alleles with frequency <50% (minor alleles) compared to that in the group of major alleles. We hypothesized that negative selection may have different effects on environment (or lifestyle)-dependent versus environment (or lifestyle)-independent diseases. We used an environment/lifestyle index (ELI) to assess influence of environmental/lifestyle factors on disease etiology. ELI was defined as the number of publications mentioning "environment" or "lifestyle" AND disease per 1,000 disease-mentioning publications. We found that the frequency distributions of the risk alleles for the diseases with strong environmental/lifestyle components follow the distribution expected under a selectively neutral model, while frequency distributions of the risk alleles for the diseases with weak environmental/lifestyle influences is shifted to the lower values indicating effects of negative selection. We hypothesized that previously selectively neutral variants become risk alleles when environment changes. The hypothesis of ancestrally neutral, currently disadvantageous risk-associated alleles predicts that the distribution of risk alleles for the environment/lifestyle dependent diseases will follow a neutral model since natural selection has not had enough time to influence allele frequencies. The results of our analysis suggest that prediction of SNP functionality based on the level of evolutionary conservation may not be useful for SNPs associated with environment/lifestyle dependent diseases.

Association of Smoking with Metabolic Volatile Organic Compounds in Exhaled Breath.

Lung cancer (LC) screening will be more efficient if it is applied to a well-defined high-risk population. Characteristics including metabolic byproducts may be taken into account to access LC risk more precisely. Breath examination provides a non-invasive method to monitor metabolic byproducts. However, the association between volatile organic compounds (VOCs) in exhaled breath and LC risk or LC risk factors is not studied. Exhaled breath samples from 122 healthy persons, who were given routine annual exam from December 2015 to December 2016, were analyzed using thermal desorption coupled with gas chromatography mass spectrometry (TD-GC-MS). Smoking characteristics, air quality, and other risk factors for lung cancer were collected. Univariate and multivariate analyses were used to evaluate the relationship between VOCs and LC risk factors. 7, 7, 11, and 27 VOCs were correlated with smoking status, smoking intensity, years of smoking, and depth of inhalation, respectively. Exhaled VOCs are related to smoking and might have a potential to evaluate LC risk more precisely. Both an assessment of temporal stability and testing in a prospective study are needed to establish the performance of VOCs such as 2,5-dimethylfuranm and 4-methyloctane as lung cancer risk biomarkers.

How to get the most from microarray data: advice from reverse genomics.

BACKGROUND: Whole-genome profiling of gene expression is a powerful tool for identifying cancer-associated genes. Genes differentially expressed between normal and tumorous tissues are usually considered to be cancer associated. We recently demonstrated that the analysis of interindividual variation in gene expression can be useful for identifying cancer associated genes. The goal of this study was to identify the best microarray data-derived predictor of known cancer associated genes. RESULTS: We found that the traditional approach of identifying cancer genes--identifying differentially expressed genes--is not very efficient. The analysis of interindividual variation of gene expression in tumor samples identifies cancer-associated genes more effectively. The results were consistent across 4 major types of cancer: breast, colorectal, lung, and prostate. We used recently reported cancer-associated genes (2011-2012) for validation and found that novel cancer-associated genes can be best identified by elevated variance of the gene expression in tumor samples. CONCLUSIONS: The observation that the high interindividual variation of gene expression in tumor tissues is the best predictor of cancer-associated genes is likely a result of tumor heterogeneity on gene level. Computer simulation demonstrates that in the case of heterogeneity, an assessment of variance in tumors provides a better identification of cancer genes than does the comparison of the expression in normal and tumor tissues. Our results thus challenge the current paradigm that comparing the mean expression between normal and tumorous tissues is the best approach to identifying cancer-associated genes; we found that the high interindividual variation in expression is a better approach, and that using variation would improve our chances of identifying cancer-associated genes.

SNP characteristics predict replication success in association studies.

Successful independent replication is the most direct approach for distinguishing real genotype-disease associations from false discoveries in genome-wide association studies (GWAS). Selecting SNPs for replication has been primarily based on P values from the discovery stage, although additional characteristics of SNPs may be used to improve replication success. We used disease-associated SNPs from more than 2,000 published GWASs to identify predictors of SNP reproducibility. SNP reproducibility was defined as a proportion of successful replications among all replication attempts. The study reporting association for the first time was considered to be discovery and all consequent studies targeting the same phenotype replications. We found that -Log(P), where P is a P value from the discovery study, is the strongest predictor of the SNP reproducibility. Other significant predictors include type of the SNP (e.g., missense vs intronic SNPs) and minor allele frequency. Features of the genes linked to the disease-associated SNP also predict SNP reproducibility. Based on empirically defined rules, we developed a reproducibility score (RS) to predict SNP reproducibility independently of -Log(P). We used data from two lung cancer GWAS studies as well as recently reported disease-associated SNPs to validate RS. Minus Log(P) outperforms RS when the very top SNPs are selected, while RS works better with relaxed selection criteria. In conclusion, we propose an empirical model to predict SNP reproducibility, which can be used to select SNPs for validation and prioritization.

Strength of selection in lung tumors correlates with clinical features better than tumor mutation burden.

Single nucleotide substitutions are the most common type of somatic mutations in cancer genome. The goal of this study was to use publicly available somatic mutation data to quantify negative and positive selection in individual lung tumors and test how strength of directional and absolute selection is associated with clinical features. The analysis found a significant variation in strength of selection (both negative and positive) among tumors, with median selection tending to be negative even though tumors with strong positive selection also exist. Strength of selection estimated as the density of missense mutations relative to the density of silent mutations showed only a weak correlation with tumor mutation burden. In the "all histology together" analysis we found that absolute strength of selection was strongly correlated with all clinically relevant features analyzed. In histology-stratified analysis selection was strongest in small cell lung cancer. Selection in adenocarcinoma was somewhat higher compared to squamous cell carcinoma. The study suggests that somatic mutation- based quantifying of directional and absolute selection in individual tumors can be a useful biomarker of tumor aggressiveness.

Confirmation of TNIP1 but not RHOB and PSORS1C1 as systemic sclerosis risk factors in a large independent replication study.

INTRODUCTION: A recent genome-wide association study in European systemic sclerosis (SSc) patients identified three loci (PSORS1C1, TNIP1 and RHOB) as novel genetic risk factors for the disease. The aim of this study was to replicate the previously mentioned findings in a large multicentre independent SSc cohort of Caucasian ancestry. METHODS: 4389 SSc patients and 7611 healthy controls from different European countries and the USA were included in the study. Six single nucleotide polymorphisms (SNP): rs342070, rs13021401 (RHOB), rs2233287, rs4958881, rs3792783 (TNIP1) and rs3130573 (PSORS1C1) were analysed. Overall significance was calculated by pooled analysis of all the cohorts. Haplotype analyses and conditional logistic regression analyses were carried out to explore further the genetic structure of the tested loci. RESULTS: Pooled analyses of all the analysed SNPs in TNIP1 revealed significant association with the whole disease (rs2233287 p(MH)=1.94×10(-4), OR 1.19; rs4958881 p(MH)=3.26×10(-5), OR 1.19; rs3792783 p(MH)=2.16×10(-4), OR 1.19). These associations were maintained in all the subgroups considered. PSORS1C1 comparison showed association with the complete set of patients and all the subsets except for the anti-centromere-positive patients. However, the association was dependent on different HLA class II alleles. The variants in the RHOB gene were not associated with SSc or any of its subsets. CONCLUSIONS: These data confirmed the influence of TNIP1 on an increased susceptibility to SSc and reinforced this locus as a common autoimmunity risk factor.

Methylation of nonessential genes in cutaneous melanoma - Rule Out hypothesis.

Differential methylation plays an important role in melanoma development and is associated with survival, progression and response to treatment. However, the mechanisms by which methylation promotes melanoma development are poorly understood. The traditional explanation of selective advantage provided by differential methylation postulates that hypermethylation of regulatory 5'-cytosine-phosphate-guanine-3' dinucleotides (CpGs) downregulates the expression of tumor suppressor genes and therefore promotes tumorigenesis. We believe that other (not necessarily alternative) explanations of the selective advantages of methylation are also possible. Here, we hypothesize that melanoma cells use methylation to shut down transcription of nonessential genes - those not required for cell survival and proliferation. Suppression of nonessential genes allows tumor cells to be more efficient in terms of energy and resource usage, providing them with a selective advantage over the tumor cells that transcribe and subsequently translate genes they do not need. We named the hypothesis the Rule Out (RO) hypothesis. The RO hypothesis predicts higher methylation of CpGs located in regulatory regions (CpG islands) of nonessential genes. It also predicts the higher methylation of regulatory CpGs linked to nonessential genes in melanomas compared to nevi and lower expression of nonessential genes in malignant (derived from melanoma) versus normal (derived from nonaffected skin) melanocytes. The analyses conducted using in-house and publicly available data found that all predictions derived from the RO hypothesis hold, providing observational support for the hypothesis.

IRF5 polymorphism predicts prognosis in patients with systemic sclerosis.

OBJECTIVE: The first genome-wide association study (GWAS) of systemic sclerosis (SSc) demonstrated three non-major histocompatibility complex (MHC) susceptibility loci. The goal of this study was to investigate the impact of these gene variants on survival and severity of interstitial lung disease (ILD) in SSc. METHODS: The authors examined 1443 Caucasian SSc patients enrolled in the Genetics versus Environment In Scleroderma Outcome Study (GENISOS) and Scleroderma Family Registry (n = 914 - discovery cohort) and The Johns Hopkins Scleroderma Cohort (n = 529 - replication cohort). Forced vital capacity (FVC)% predicted was used as a surrogate for ILD severity. Five single nucleotide polymorphisms, IRF5 (rs10488631, rs12537284, rs4728142), STAT4 (rs3821236), CD247 (rs2056626) reached genome-wide significance in the SSc-GWAS and were examined in the current study. RESULTS: Overall, 15.5% of the patients had died over the follow-up period of 5.5 years. The IRF5 rs4728142 minor allele was predictive of longer survival in the discovery cohort (p = 0.021) and in the independent replication cohort (p = 0.047) and combined group (HR: 0.75, 95% CI 0.62 to 0.90, p = 0.002). The association of this SNP with survival was independent of age at disease onset, disease type and autoantibody profile (anticentromere and antitopoisomerase antibodies). The minor allele frequency of IRF5 rs4728142 was 49.4%. Moreover, IRF5 rs4728142 minor allele correlated with higher FVC% predicted at enrolment (p = 0.019). Finally, the IRF5 rs4728142 minor allele was associated with lower IRF5 transcript expression in patients and controls (p = 0.016 and p = 0.034, respectively), suggesting that the IRF5, rs4728142 SNP, may be functionally relevant. CONCLUSION: An SNP in the IRF5 promoter region (rs4728142), associated with lower IRF5 transcript levels, was predictive of longer survival and milder ILD in patients with SSc.

Identification of lung cancer drivers by comparison of the observed and the expected numbers of missense and nonsense mutations in individual human genes.

Largely, cancer development is driven by acquisition and positive selection of somatic mutations that increase proliferation and survival of tumor cells. As a result, genes related to cancer development tend to have an excess of somatic mutations in them. An excess of missense and/or nonsense mutations in a gene is an indicator of its cancer relevance. To identify genes with an excess of potentially functional missense or nonsense mutations one needs to compare the observed and expected numbers of mutations in the gene. We estimated the expected numbers of missense and nonsense mutations in individual human genes using (i) the number of potential sites for missense and nonsense mutations in individual transcripts and (ii) histology-specific nucleotide context-dependent mutation rates. To estimate mutation rates defined as the number of mutations per site per tumor we used silent mutations reported in the Catalog Of Somatic Mutations In Cancer (COSMIC). The estimates were nucleotide context dependent. We have identified 26 genes with an excess of missense and/or nonsense mutations for lung adenocarcinoma, 18 genes for small cell lung cancer, and 26 genes for squamous cell carcinoma of the lung. These genes include known genes and novel lung cancer gene candidates.

A smoking-based carcinogenesis model for lung cancer risk prediction.

Lung cancer is the leading cancer killer for both men and women worldwide. Over 80% of lung cancers are attributed to smoking. In this analysis, the authors propose to use a two-stage clonal expansion (TSCE) model to predict an individual's lung cancer risk based on gender and smoking history. The TSCE model is traditionally fitted to prospective cohort data. Here, the authors describe a new method that allows for the reconstruction of cohort data from the combination of risk factor data obtained from a case-control study, and tabled incidence/mortality rate data, and discuss alternative approaches. The method is applied to fit a TSCE model based on smoking. The fitted model is validated against independent data from the control arm of a lung cancer chemoprevention trial, CARET, where it accurately predicted the number of lung cancer deaths observed.

Modeling the mortality reduction due to computed tomography screening for lung cancer.

BACKGROUND: The efficacy of computed tomography (CT) screening for lung cancer remains controversial because results from the National Lung Screening Trial are not yet available. In this study, the authors used data from a single-arm CT screening trial to estimate the mortality reduction using a modeling-based approach to construct a control comparison arm. METHODS: To estimate the potential lung cancer mortality reduction because of CT screening, a previously developed and validated model was applied to the screening trial to predict the number of lung cancer deaths in the absence of screening. By using age, gender, and smoking characteristics matching those of the trial participants, the model was used to simulate 5000 trials in the absence of CT screening to produce the expected number of lung cancer deaths along with 95% confidence intervals (95% CIs), while adjusting for healthy volunteer bias. RESULTS: There were 64 observed lung cancer deaths in the screening cohort (n = 7995), whereas the model predicted 117.7 deaths (95% CI, 98 deaths-139 deaths), indicating a mortality reduction of 45.6% (P < .001). When a more conservative healthy volunteer adjustment was applied, 111.3 lung cancer deaths were predicted (95% CI, 91 deaths-132 deaths), for a lung cancer-specific mortality reduction of 42.5% (P < .001). CONCLUSIONS: The results of the current study indicate that CT screening along with early stage treatment can reduce lung cancer-specific mortality. This mortality reduction is greatly influenced by the protocol of nodule follow-up and treatment, and the length of follow-up.

Shifting paradigm of association studies: value of rare single-nucleotide polymorphisms.

Currently, single-nucleotide polymorphisms (SNPs) with minor allele frequency (MAF) of >5% are preferentially used in case-control association studies of common human diseases. Recent technological developments enable inexpensive and accurate genotyping of a large number of SNPs in thousands of cases and controls, which can provide adequate statistical power to analyze SNPs with MAF <5%. Our purpose was to determine whether evaluating rare SNPs in case-control association studies could help identify causal SNPs for common diseases. We suggest that slightly deleterious SNPs (sdSNPs) subjected to weak purifying selection are major players in genetic control of susceptibility to common diseases. We compared the distribution of MAFs of synonymous SNPs with that of nonsynonymous SNPs (1) predicted to be benign, (2) predicted to be possibly damaging, and (3) predicted to be probably damaging by PolyPhen. Our sources of data were the International HapMap Project, ENCODE, and the SeattleSNPs project. We found that the MAF distribution of possibly and probably damaging SNPs was shifted toward rare SNPs compared with the MAF distribution of benign and synonymous SNPs that are not likely to be functional. We also found an inverse relationship between MAF and the proportion of nsSNPs predicted to be protein disturbing. On the basis of this relationship, we estimated the joint probability that a SNP is functional and would be detected as significant in a case-control study. Our analysis suggests that including rare SNPs in genotyping platforms will advance identification of causal SNPs in case-control association studies, particularly as sample sizes increase.

Dietary patterns affect lung cancer risk in never smokers.

A number of studies suggest a role of dietary factors as risk predictors of lung cancer in never smokers. However, it is difficult to interpret the observed associations of lung cancer risk with any particular dietary item due to high correlation among different dietary items. In this study, we derived uncorrelated patterns of dietary items in the never smokers and evaluated the association of these patterns with lung cancer risk, using food frequency data from 299 never-smoker lung cancer patients and 317 controls enrolled in an ongoing case-control lung cancer study. We identified 2 major dietary patterns in never smokers: a "healthy eating" pattern including vegetables, fruits, and low-fat food items, and a "mixed dishes" pattern including most foods with positive loadings. Using multivariable regression analysis, we show that the healthy eating pattern is associated with a significant reduction of lung cancer risk among never smokers. The effect of the healthy eating pattern remained significant after adjustment for age, gender, education, caloric intake, secondhand smoke exposure, and family history of cancer. This finding, if confirmed in a prospective study, has a clear preventive significance, by promoting interventions encouraging healthier diets.

Adjusting a cancer mortality-prediction model for disease status-related eligibility criteria.

BACKGROUND: Volunteering participants in disease studies tend to be healthier than the general population partially due to specific enrollment criteria. Using modeling to accurately predict outcomes of cohort studies enrolling volunteers requires adjusting for the bias introduced in this way. Here we propose a new method to account for the effect of a specific form of healthy volunteer bias resulting from imposing disease status-related eligibility criteria, on disease-specific mortality, by explicitly modeling the length of the time interval between the moment when the subject becomes ineligible for the study, and the outcome. METHODS: Using survival time data from 1190 newly diagnosed lung cancer patients at MD Anderson Cancer Center, we model the time from clinical lung cancer diagnosis to death using an exponential distribution to approximate the length of this interval for a study where lung cancer death serves as the outcome. Incorporating this interval into our previously developed lung cancer risk model, we adjust for the effect of disease status-related eligibility criteria in predicting the number of lung cancer deaths in the control arm of CARET. The effect of the adjustment using the MD Anderson-derived approximation is compared to that based on SEER data. RESULTS: Using the adjustment developed in conjunction with our existing lung cancer model, we are able to accurately predict the number of lung cancer deaths observed in the control arm of CARET. CONCLUSIONS: The resulting adjustment was accurate in predicting the lower rates of disease observed in the early years while still maintaining reasonable prediction ability in the later years of the trial. This method could be used to adjust for, or predict the duration and relative effect of any possible biases related to disease-specific eligibility criteria in modeling studies of volunteer-based cohorts.