Polygenic risk scores, radiation treatment exposures and subsequent cancer risk in childhood cancer survivors

Survivors of childhood cancer are at increased risk for subsequent cancers attributable to the late effects of radiotherapy and other treatment exposures; thus, further understanding of the impact of genetic predisposition on risk is needed. Combining genotype data for 11,220 5-year survivors from the Childhood Cancer Survivor Study and the St Jude Lifetime Cohort, we found that cancer-specific polygenic risk scores (PRSs) derived from general population, genome-wide association study, cancer loci identified survivors of European ancestry at increased risk of subsequent basal cell carcinoma (odds ratio per s.d. of the PRS: OR = 1.37, 95% confidence interval (CI) = 1.29-1.46), female breast cancer (OR = 1.42, 95% CI = 1.27-1.58), thyroid cancer (OR = 1.48, 95% CI = 1.31-1.67), squamous cell carcinoma (OR = 1.20, 95% CI = 1.00-1.44) and melanoma (OR = 1.60, 95% CI = 1.31-1.96); however, the association for colorectal cancer was not significant (OR = 1.19, 95% CI = 0.94-1.52). An investigation of joint associations between PRSs and radiotherapy found more than additive increased risks of basal cell carcinoma, and breast and thyroid cancers. For survivors with radiotherapy exposure, the cumulative incidence of subsequent cancer by age 50 years was increased for those with high versus low PRS. These findings suggest a degree of shared genetic etiology for these malignancy types in the general population and survivors, which remains evident in the context of strong radiotherapy-related risk.

Inflated expectations: Rare-variant association analysis using public controls

The use of publicly available sequencing datasets as controls (hereafter, "public controls") in studies of rare variant disease associations has great promise but can increase the risk of false-positive discovery. The specific factors that could contribute to inflated distribution of test statistics have not been systematically examined. Here, we leveraged both public controls, gnomAD v2.1 and several datasets sequenced in our laboratory to systematically investigate factors that could contribute to the false-positive discovery, as measured by λΔ95, a measure to quantify the degree of inflation in statistical significance. Analyses of datasets in this investigation found that 1) the significantly inflated distribution of test statistics decreased substantially when the same variant caller and filtering pipelines were employed, 2) differences in library prep kits and sequencers did not affect the false-positive discovery rate and, 3) joint vs. separate variant-calling of cases and controls did not contribute to the inflation of test statistics. Currently available methods do not adequately adjust for the high false-positive discovery. These results, especially if replicated, emphasize the risks of using public controls for rare-variant association tests in which individual-level data and the computational pipeline are not readily accessible, which prevents the use of the same variant-calling and filtering pipelines on both cases and controls. A plausible solution exists with the emergence of cloud-based computing, which can make it possible to bring containerized analytical pipelines to the data (rather than the data to the pipeline) and could avert or minimize these issues. It is suggested that future reports account for this issue and provide this as a limitation in reporting new findings based on studies that cannot practically analyze all data on a single pipeline.

In-utero exposure to zidovudine-containing antiretroviral therapy and clonal hematopoiesis in HIV-exposed uninfected newborns

OBJECTIVE

Zidovudine (ZDV) has been extensively used in pregnant women to prevent vertical transmission of HIV but few studies have evaluated potential mutagenic effects of ZDV during fetal development.

DESIGN

Our study investigated clonal hematopoiesis in HIV-exposed uninfected (HEU) newborns, 94 of whom were ZDV-exposed and 91 antiretroviral therapy (ART)-unexposed and matched for potential confounding factors.

METHODS

Utilizing high depth sequencing and genotyping arrays, we comprehensively examined blood samples collected during the first week after birth for potential clonal hematopoiesis associated with fetal ZDV exposure, including clonal single nucleotide variants (SNVs), small insertions and deletions (indels), and large structural copy number or copy neutral alterations.

RESULTS

We observed no statistically significant difference in the number of SNVs and indels per person in ZDV-exposed children (adjusted ratio [95% confidence interval, CI] for expected number of mutations = 0.79 [0.50--1.22], P = 0.3), and no difference in the number of large structural alterations. Mutations in common clonal hematopoiesis driver genes were not found in the study population. Mutational signature analyses on SNVs detected no novel signatures unique to the ZDV-exposed children and the mutational profiles were similar between the two groups.

CONCLUSION

Our results suggest that clonal hematopoiesis at levels detectable in our study is not strongly influenced by in-utero ZDV exposure; however, additional follow-up studies are needed to further evaluate the safety and potential long-term impacts of in-utero ZDV exposure in HEU children as well as better investigate genomic aberrations occurring late in pregnancy.

Radiation-related genomic profile of papillary thyroid carcinoma after the Chernobyl accident

The 1986 Chernobyl nuclear power plant accident increased papillary thyroid carcinoma (PTC) incidence in surrounding regions, particularly for radioactive iodine (¹³¹I)-exposed children. We analyzed genomic, transcriptomic, and epigenomic characteristics of 440 PTCs from Ukraine (from 359 individuals with estimated childhood ¹³¹I exposure and 81 unexposed children born after 1986). PTCs displayed radiation dose-dependent enrichment of fusion drivers, nearly all in the mitogen-activated protein kinase pathway, and increases in small deletions and simple/balanced structural variants that were clonal and bore hallmarks of nonhomologous end-joining repair. Radiation-related genomic alterations were more pronounced for individuals who were younger at exposure. Transcriptomic and epigenomic features were strongly associated with driver events but not radiation dose. Our results point to DNA double-strand breaks as early carcinogenic events that subsequently enable PTC growth after environmental radiation exposure.

Frequency of Pathogenic Germline Variants in Cancer-Susceptibility Genes in the Childhood Cancer Survivor Study

Background

Pediatric cancers are the leading cause of death by disease in children despite improved survival rates overall. The contribution of germline genetic susceptibility to pediatric cancer survivors has not been extensively characterized. We assessed the frequency of pathogenic or likely pathogenic (P/LP) variants in 5451 long-term pediatric cancer survivors from the Childhood Cancer Survivor Study.

Methods

Exome sequencing was conducted on germline DNA from 5451 pediatric cancer survivors (cases who survived ≥5 years from diagnosis; n = 5105 European) and 597 European cancer-free adults (controls). Analyses focused on comparing the frequency of rare P/LP variants in 237 cancer-susceptibility genes and a subset of 60 autosomal dominant high-to-moderate penetrance genes, for both case-case and case-control comparisons.

Results

Of European cases, 4.1% harbored a P/LP variant in high-to-moderate penetrance autosomal dominant genes compared with 1.3% in controls (2-sided P = 3 × 10-4). The highest frequency of P/LP variants was in genes typically associated with adult onset rather than pediatric cancers, including BRCA1/2, FH, PALB2, PMS2, and CDKN2A. A statistically significant excess of P/LP variants, after correction for multiple tests, was detected in patients with central nervous system cancers (NF1, SUFU, TSC1, PTCH2), Wilms tumor (WT1, REST), non-Hodgkin lymphoma (PMS2), and soft tissue sarcomas (SDHB, DICER1, TP53, ERCC4, FGFR3) compared with other pediatric cancers.

Conclusion

In long-term pediatric cancer survivors, we identified P/LP variants in cancer-susceptibility genes not previously associated with pediatric cancer as well as confirmed known associations. Further characterization of variants in these genes in pediatric cancer will be important to provide optimal genetic counseling for patients and their families.

Subsequent Neoplasm Risk Associated With Rare Variants in DNA Damage Response and Clinical Radiation Sensitivity Syndrome Genes in the Childhood Cancer Survivor Study

PURPOSE

Radiotherapy for childhood cancer is associated with elevated subsequent neoplasm (SN) risk, but the contribution of rare variants in DNA damage response and radiation sensitivity genes to SN risk is unknown.

PATIENTS AND METHODS

We conducted whole-exome sequencing in a cohort of childhood cancer survivors originally diagnosed during 1970 to 1986 (mean follow-up, 32.7 years), with reconstruction of doses to body regions from radiotherapy records. We identified patients who developed SN types previously reported to be related to radiotherapy (RT-SNs; eg, basal cell carcinoma [BCC], breast cancer, meningioma, thyroid cancer, sarcoma) and matched controls (sex, childhood cancer type/diagnosis, age, SN location, radiation dose, survival). Conditional logistic regression assessed SN risk associated with potentially protein-damaging rare variants (SnpEff, ClinVar) in 476 DNA damage response or radiation sensitivity genes with exact permutation-based P values using a Bonferroni-corrected significance threshold of P < 8.06 × 10^-5.

RESULTS

Among 5,105 childhood cancer survivors of European descent, 1,108 (21.7%) developed at least 1 RT-SN. Out-of-field RT-SN risk, excluding BCC, was associated with homologous recombination repair (HRR) gene variants (patient cases, 23.2%; controls, 10.8%; odds ratio [OR], 2.6; 95% CI, 1.7 to 3.9; P = 4.79 × 10^-5), most notably but nonsignificantly for FANCM (patient cases, 4.0%; matched controls, 0.6%; P = 9.64 × 10^-5). HRR variants were not associated with likely in/near-field RT-SNs, excluding BCC (patient cases, 12.7%; matched controls, 12.9%; P = .92). Irrespective of radiation dose, risk for RT-SNs was also associated with EXO1 variants (patient cases, 1.8%; controls, 0.4%; P = 3.31 × 10^-5), another gene implicated in DNA double-strand break repair.

CONCLUSION

In this large-scale discovery study, we identified novel associations between RT-SN risk after childhood cancer and potentially protein-damaging rare variants in genes involved in DNA double-strand break repair, particularly HRR. With replication, these results could affect screening recommendations for childhood cancer survivors and risk-benefit assessments of treatment approaches.

Alternative Isoform Analysis of Ttc8 Expression in the Rat Pineal Gland Using a Multi-Platform Sequencing Approach Reveals Neural Regulation

Alternative isoform regulation (AIR) vastly increases transcriptome diversity and plays an important role in numerous biological processes and pathologies. However, the detection and analysis of isoform-level differential regulation is difficult, particularly in the face of complex and incompletely-annotated transcriptomes. Here we have used Illumina short-read/high-throughput RNA-Seq to identify 55 genes that exhibit neurally-regulated AIR in the pineal gland, and then used two other complementary experimental platforms to further study and characterize the Ttc8 gene, which is involved in Bardet-Biedl syndrome and non-syndromic retinitis pigmentosa. Use of the JunctionSeq analysis tool led to the detection of several novel exons and splice junctions in this gene, including two novel alternative transcription start sites which were found to display disproportionately strong neurally-regulated differential expression in several independent experiments. These high-throughput sequencing results were validated and augmented via targeted qPCR and long-read Pacific Biosciences SMRT sequencing. We confirmed the existence of numerous novel splice junctions and the selective upregulation of the two novel start sites. In addition, we identified more than 20 novel isoforms of the Ttc8 gene that are co-expressed in this tissue. By using information from multiple independent platforms we not only greatly reduce the risk of errors, biases, and artifacts influencing our results, we also are able to characterize the regulation and splicing of the Ttc8 gene more deeply and more precisely than would be possible via any single platform. The hybrid method outlined here represents a powerful strategy in the study of the transcriptome.

Detection and visualization of differential splicing in RNA-Seq data with JunctionSeq

Although RNA-Seq data provide unprecedented isoform-level expression information, detection of alternative isoform regulation (AIR) remains difficult, particularly when working with an incomplete transcript annotation. We introduce JunctionSeq, a new method that builds on the statistical techniques used by the well-established DEXSeq package to detect differential usage of both exonic regions and splice junctions. In particular, JunctionSeq is capable of detecting differential usage of novel splice junctions without the need for an additional isoform assembly step, greatly improving performance when the available transcript annotation is flawed or incomplete. JunctionSeq also provides a powerful and streamlined visualization toolset that allows bioinformaticians to quickly and intuitively interpret their results. We tested our method on publicly available data from several experiments performed on the rat pineal gland and Toxoplasma gondii, successfully detecting known and previously validated AIR genes in 19 out of 19 gene-level hypothesis tests. Due to its ability to query novel splice sites, JunctionSeq is still able to detect these differences even when all alternative isoforms for these genes were not included in the transcript annotation. JunctionSeq thus provides a powerful method for detecting alternative isoform regulation even with low-quality annotations. An implementation of JunctionSeq is available as an R/Bioconductor package.

QoRTs: a comprehensive toolset for quality control and data processing of RNA-Seq experiments

BACKGROUND

High-throughput next-generation RNA sequencing has matured into a viable and powerful method for detecting variations in transcript expression and regulation. Proactive quality control is of critical importance as unanticipated biases, artifacts, or errors can potentially drive false associations and lead to flawed results.

RESULTS

We have developed the Quality of RNA-Seq Toolset, or QoRTs, a comprehensive, multifunction toolset that assists in quality control and data processing of high-throughput RNA sequencing data.

CONCLUSIONS

QoRTs generates an unmatched variety of quality control metrics, and can provide cross-comparisons of replicates contrasted by batch, biological sample, or experimental condition, revealing any outliers and/or systematic issues that could drive false associations or otherwise compromise downstream analyses. In addition, QoRTs simultaneously replaces the functionality of numerous other data-processing tools, and can quickly and efficiently generate quality control metrics, coverage counts (for genes, exons, and known/novel splice-junctions), and browser tracks. These functions can all be carried out as part of a single unified data-processing/quality control run, greatly reducing both the complexity and the total runtime of the analysis pipeline. The software, source code, and documentation are available online at http://hartleys.github.io/QoRTs.

A GCH1 haplotype confers sex-specific susceptibility to pain crises and altered endothelial function in adults with sickle cell anemia

GTP cyclohydrolase (GCH1) is rate limiting for tetrahydrobiopterin (BH4) synthesis, where BH4 is a cofactor for nitric oxide (NO) synthases and aromatic hydroxylases. GCH1 polymorphisms are implicated in the pathophysiology of pain, but have not been investigated in African populations. We examined GCH1 and pain in sickle cell anemia where GCH1 rs8007267 was a risk factor for pain crises in discovery (n = 228; odds ratio [OR] 2.26; P = 0.009) and replication (n = 513; OR 2.23; P = 0.004) cohorts. In vitro, cells from sickle cell anemia subjects homozygous for the risk allele produced higher BH4. In vivo physiological studies of traits likely to be modulated by GCH1 showed rs8007267 is associated with altered endothelial dependent blood flow in females with SCA (8.42% of variation; P = 0.002). The GCH1 pain association is attributable to an African haplotype with where its sickle cell anemia pain association is limited to females (OR 2.69; 95% CI 1.21-5.94; P = 0.01) and has the opposite directional association described in Europeans independent of global admixture. The presence of a GCH1 haplotype with high BH4 in populations of African ancestry could explain the association of rs8007267 with sickle cell anemia pain crises. The vascular effects of GCH1 and BH4 may also have broader implications for cardiovascular disease in populations of African ancestry.

PleioGRiP: genetic risk prediction with pleiotropy

MOTIVATION

Although several studies have used Bayesian classifiers for risk prediction using genome-wide single nucleotide polymorphism (SNP) datasets, no software can efficiently perform these analyses on massive genetic datasets and can accommodate multiple traits.

RESULTS

We describe the program PleioGRiP that performs a genome-wide Bayesian model search to identify SNPs associated with a discrete phenotype and uses SNPs ranked by Bayes factor to produce nested Bayesian classifiers. These classifiers can be used for genetic risk prediction, either selecting the classifier with optimal number of features or using an ensemble of classifiers. In addition, PleioGRiP implements an extension to the Bayesian search and classification and can search for pleiotropic relationships in which SNPs are simultaneously associated with two or more distinct phenotypes. These relationships can be used to generate connected Bayesian classifiers to predict the phenotype of interest either using genetic data alone or in combination with the secondary phenotype(s).

AVAILABILITY

PleioGRiP is implemented in Java, and it is available from http://hdl.handle.net/2144/4367.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Bayesian methods for multivariate modeling of pleiotropic SNP associations and genetic risk prediction

Genome-wide association studies (GWAS) have identified numerous associations between genetic loci and individual phenotypes; however, relatively few GWAS have attempted to detect pleiotropic associations, in which loci are simultaneously associated with multiple distinct phenotypes. We show that pleiotropic associations can be directly modeled via the construction of simple Bayesian networks, and that these models can be applied to produce single or ensembles of Bayesian classifiers that leverage pleiotropy to improve genetic risk prediction. The proposed method includes two phases: (1) Bayesian model comparison, to identify Single-Nucleotide Polymorphisms (SNPs) associated with one or more traits; and (2) cross-validation feature selection, in which a final set of SNPs is selected to optimize prediction. To demonstrate the capabilities and limitations of the method, a total of 1600 case-control GWAS datasets with two dichotomous phenotypes were simulated under 16 scenarios, varying the association strengths of causal SNPs, the size of the discovery sets, the balance between cases and controls, and the number of pleiotropic causal SNPs. Across the 16 scenarios, prediction accuracy varied from 90 to 50%. In the 14 scenarios that included pleiotropically associated SNPs, the pleiotropic model search and prediction methods consistently outperformed the naive model search and prediction. In the two scenarios in which there were no true pleiotropic SNPs, the differences between the pleiotropic and naive model searches were minimal. To further evaluate the method on real data, a discovery set of 1071 sickle cell disease (SCD) patients was used to search for pleiotropic associations between cerebral vascular accidents and fetal hemoglobin level. Classification was performed on a smaller validation set of 352 SCD patients, and showed that the inclusion of pleiotropic SNPs may slightly improve prediction, although the difference was not statistically significant. The proposed method is robust, computationally efficient, and provides a powerful new approach for detecting and modeling pleiotropic disease loci.

A genome-wide association study of total bilirubin and cholelithiasis risk in sickle cell anemia

Serum bilirubin levels have been associated with polymorphisms in the UGT1A1 promoter in normal populations and in patients with hemolytic anemias, including sickle cell anemia. When hemolysis occurs circulating heme increases, leading to elevated bilirubin levels and an increased incidence of cholelithiasis. We performed the first genome-wide association study (GWAS) of bilirubin levels and cholelithiasis risk in a discovery cohort of 1,117 sickle cell anemia patients. We found 15 single nucleotide polymorphisms (SNPs) associated with total bilirubin levels at the genome-wide significance level (p value <5 × 10(-8)). SNPs in UGT1A1, UGT1A3, UGT1A6, UGT1A8 and UGT1A10, different isoforms within the UGT1A locus, were identified (most significant rs887829, p = 9.08 × 10(-25)). All of these associations were validated in 4 independent sets of sickle cell anemia patients. We tested the association of the 15 SNPs with cholelithiasis in the discovery cohort and found a significant association (most significant p value 1.15 × 10(-4)). These results confirm that the UGT1A region is the major regulator of bilirubin metabolism in African Americans with sickle cell anemia, similar to what is observed in other ethnicities.

Ancestry of African Americans with sickle cell disease

The inheritance of genetic disease depends on ancestry that must be considered when interpreting genetic association studies and can provide insights when comparing traits in a population. We compared the genetic profiles of African Americans with sickle cell disease to those of Black Africans and Caucasian populations of European descent and found that they are less genetically admixed than other African Americans and have an ancestry similar to Yorubans, Mandenkas and Bantu.

Hippocampal morphometry in population-based incident Alzheimer's disease and vascular dementia: the HAAS

BACKGROUND

Hippocampal changes may be a useful biomarker for Alzheimer's disease if they are specific to dementia sub-type. We compare hippocampal volume and shape in population-based incident cases of Alzheimer's disease and vascular dementia (VaD).

METHODS

Participants are Japanese-American men from the Honolulu Asia Aging Study. The following analysis is based on a sub-group of men with mild incident Alzheimer's disease (n=24: age=82.5 ± 4.6) or incident VaD (n=14: age=80.5 ± 4.5). To estimate hippocampal volume, one reader, blinded to dementia diagnosis, manually outlined the left and right formation of the hippocampus using published criteria. We used 3-D mapping methods developed at the Laboratory of Neuro Imaging (LONI) to compare regional variation in hippocampal width between dementia groups.

RESULTS

Hippocampal volume was about 5% smaller in the Alzheimer's disease group compared to the VaD group, but the difference was not significant. Hippocampal shape differed between the two case groups for the left (p<0.04) but not right (p<0.21) hippocampus. The specific region of the hippocampus that most consistently differed between the Alzheimer's disease and VaD cases was in the lateral portion of the left hippocampus. Our interpretation of this region is that it intersects the CA1 sub-region to a great extent but also includes the dentate gyrus (and hilar region) and subiculum.

CONCLUSION

As indicated by shape analysis, there are some differences in atrophy localisation between the Alzheimer's disease and VaD cases, despite the finding that volume of the hippocampi did not differ. These findings suggest hippocampal atrophy in Alzheimer's disease may be more focal than in VaD.

Severe sickle cell anemia is associated with increased plasma levels of TNF-R1 and VCAM-1

Sickle cell anemia (SCA, HBB glu6val) is characterized by multiple complications and a high degree of phenotypic variability: some subjects have only sporadic pain crises and few acute hospitalizations, while others experience multiple serious complications, high levels of morbidity, and accelerated mortality [1]. The tumor necrosis factor-α (TNF-α) signaling pathway plays important roles in inflammation and the immune response; variation in this pathway might be expected to modify the overall severity of SCA through the pathway's effects on the vascular endothelium [2,3]. We examined plasma biomarkers of TNF-α activity and endothelial cell activation for associations with SCA severity in 24 adults (12 mild, 12 severe). Two biomarkers, tumor necrosis factor-α receptor-1 (TNF-R1) and vascular cell adhesion molecule-1 (VCAM-1) were significantly higher in subjects with severe SCA. Along with these biomarker differences, we also examined data from a genome-wide association study (GWAS) using SCA severity as a disease phenotype, and found evidence of genetic association between disease severity and a single nucleotide polymorphism (SNP) in VCAM1, which codes for VCAM-1, and several SNPs in ARFGEF2, a gene involved in TNF-R1 release [4].

Clustering by genetic ancestry using genome-wide SNP data

BACKGROUND

Population stratification can cause spurious associations in a genome-wide association study (GWAS), and occurs when differences in allele frequencies of single nucleotide polymorphisms (SNPs) are due to ancestral differences between cases and controls rather than the trait of interest. Principal components analysis (PCA) is the established approach to detect population substructure using genome-wide data and to adjust the genetic association for stratification by including the top principal components in the analysis. An alternative solution is genetic matching of cases and controls that requires, however, well defined population strata for appropriate selection of cases and controls.

RESULTS

We developed a novel algorithm to cluster individuals into groups with similar ancestral backgrounds based on the principal components computed by PCA. We demonstrate the effectiveness of our algorithm in real and simulated data, and show that matching cases and controls using the clusters assigned by the algorithm substantially reduces population stratification bias. Through simulation we show that the power of our method is higher than adjustment for PCs in certain situations.

CONCLUSIONS

In addition to reducing population stratification bias and improving power, matching creates a clean dataset free of population stratification which can then be used to build prediction models without including variables to adjust for ancestry. The cluster assignments also allow for the estimation of genetic heterogeneity by examining cluster specific effects.

Clustering by genetic ancestry using genome-wide SNP data

Background

Population stratification can cause spurious associations in a genome-wide association study (GWAS), and occurs when differences in allele frequencies of single nucleotide polymorphisms (SNPs) are due to ancestral differences between cases and controls rather than the trait of interest. Principal components analysis (PCA) is the established approach to detect population substructure using genome-wide data and to adjust the genetic association for stratification by including the top principal components in the analysis. An alternative solution is genetic matching of cases and controls that requires, however, well defined population strata for appropriate selection of cases and controls.

Results

We developed a novel algorithm to cluster individuals into groups with similar ancestral backgrounds based on the principal components computed by PCA. We demonstrate the effectiveness of our algorithm in real and simulated data, and show that matching cases and controls using the clusters assigned by the algorithm substantially reduces population stratification bias. Through simulation we show that the power of our method is higher than adjustment for PCs in certain situations.

Conclusions

In addition to reducing population stratification bias and improving power, matching creates a clean dataset free of population stratification which can then be used to build prediction models without including variables to adjust for ancestry. The cluster assignments also allow for the estimation of genetic heterogeneity by examining cluster specific effects.

Genetic signatures of exceptional longevity in humans

Healthy aging is thought to reflect the combined influence of environmental factors (lifestyle choices) and genetic factors. To explore the genetic contribution, we undertook a genome-wide association study of exceptional longevity (EL) in 1055 centenarians and 1267 controls. Using these data, we built a genetic model that includes 150 single-nucleotide polymorphisms (SNPs) and found that it could predict EL with 77% accuracy in an independent set of centenarians and controls. Further in silico analysis revealed that 90% of centenarians can be grouped into 19 clusters characterized by different combinations of SNP genotypes-or genetic signatures-of varying predictive value. The different signatures, which attest to the genetic complexity of EL, correlated with differences in the prevalence and age of onset of age-associated diseases (e.g., dementia, hypertension, and cardiovascular disease) and may help dissect this complex phenotype into subphenotypes of healthy aging.

Fetal hemoglobin in sickle cell anemia: genome-wide association studies suggest a regulatory region in the 5' olfactory receptor gene cluster

In a genome-wide association study of 848 blacks with sickle cell anemia, we identified single nucleotide polymorphisms (SNPs) associated with fetal hemoglobin concentration. The most significant SNPs in a discovery sample were tested in a replication set of 305 blacks with sickle cell anemia and in subjects with hemoglobin E or beta thalassemia trait from Thailand and Hong Kong. A novel region on chromosome 11 containing olfactory receptor genes OR51B5 and OR51B6 was identified by 6 SNPs (lowest P = 4.7E-08) and validated in the replication set. An additional olfactory receptor gene, OR51B2, was identified by a novel SNP set enrichment analysis. Genome-wide association studies also validated a previously identified SNP (rs766432) in BCL11A, a gene known to affect fetal hemoglobin levels (P = 2.6E-21) and in Thailand and Hong Kong subjects. Elements within the olfactory receptor gene cluster might play a regulatory role in gamma-globin gene expression.

Genetic modifiers of the severity of sickle cell anemia identified through a genome-wide association study

We conducted a genome-wide association study (GWAS) to discover single nucleotide polymorphisms (SNPs) associated with the severity of sickle cell anemia in 1,265 patients with either "severe" or "mild" disease based on a network model of disease severity. We analyzed data using single SNP analysis and a novel SNP set enrichment analysis (SSEA) developed to discover clusters of associated SNPs. Single SNP analysis discovered 40 SNPs that were strongly associated with sickle cell severity (odds for association >1,000); of the 32 that we could analyze in an independent set of 163 patients, five replicated, eight showed consistent effects although failed to reach statistical significance, whereas 19 did not show any convincing association. Among the replicated associations are SNPs in KCNK6 a K(+) channel gene. SSEA identified 27 genes with a strong enrichment of significant SNPs (P < 10(-6)); 20 were replicated with varying degrees of confidence. Among the novel findings identified by SSEA is the telomere length regulator gene TNKS. These studies are the first to use GWAS to understand the genetic diversity that accounts the phenotypic heterogeneity sickle cell anemia as estimated by an integrated model of severity. Additional validation, resequencing, and functional studies to understand the biology and reveal mechanisms by which candidate genes might have their effects are the future goals of this work.

Imputation of missing genotypes: an empirical evaluation of IMPUTE

Background

Imputation of missing genotypes is becoming a very popular solution for synchronizing genotype data collected with different microarray platforms but the effect of ethnic background, subject ascertainment, and amount of missing data on the accuracy of imputation are not well understood.

Results

We evaluated the accuracy of the program IMPUTE to generate the genotype data of partially or fully untyped single nucleotide polymorphisms (SNPs). The program uses a model-based approach to imputation that reconstructs the genotype distribution given a set of referent haplotypes and the observed data, and uses this distribution to compute the marginal probability of each missing genotype for each individual subject that is used to impute the missing data. We assembled genome-wide data from five different studies and three different ethnic groups comprising Caucasians, African Americans and Asians. We randomly removed genotype data and then compared the observed genotypes with those generated by IMPUTE. Our analysis shows 97% median accuracy in Caucasian subjects when less than 10% of the SNPs are untyped and missing genotypes are accepted regardless of their posterior probability. The median accuracy increases to 99% when we require 0.95 minimum posterior probability for an imputed genotype to be acceptable. The accuracy decreases to 86% or 94% when subjects are African Americans or Asians. We propose a strategy to improve the accuracy by leveraging the level of admixture in African Americans.

Conclusion

Our analysis suggests that IMPUTE is very accurate in samples of Caucasians origin, it is slightly less accurate in samples of Asians background, but substantially less accurate in samples of admixed background such as African Americans. Sample size and ascertainment do not seem to affect the accuracy of imputation.

A hierarchical and modular approach to the discovery of robust associations in genome-wide association studies from pooled DNA samples

BACKGROUND

One of the challenges of the analysis of pooling-based genome wide association studies is to identify authentic associations among potentially thousands of false positive associations.

RESULTS

We present a hierarchical and modular approach to the analysis of genome wide genotype data that incorporates quality control, linkage disequilibrium, physical distance and gene ontology to identify authentic associations among those found by statistical association tests. The method is developed for the allelic association analysis of pooled DNA samples, but it can be easily generalized to the analysis of individually genotyped samples. We evaluate the approach using data sets from diverse genome wide association studies including fetal hemoglobin levels in sickle cell anemia and a sample of centenarians and show that the approach is highly reproducible and allows for discovery at different levels of synthesis.

CONCLUSION

Results from the integration of Bayesian tests and other machine learning techniques with linkage disequilibrium data suggest that we do not need to use too stringent thresholds to reduce the number of false positive associations. This method yields increased power even with relatively small samples. In fact, our evaluation shows that the method can reach almost 70% sensitivity with samples of only 100 subjects.

Hippocampal shape analysis in Alzheimer's disease: a population-based study

BACKGROUND

Hippocampal atrophy--particularly of the CA1 region--may be useful as a biomarker for Alzheimer's disease (AD) or the risk for AD. The extent to which the AD hippocampus can be distinguished in vivo from changes due to normal aging or other processes that affect the hippocampus is of clinical importance and is an area of active research. In this study, we use structural imaging techniques to model hippocampal size and regional shape differences between elderly men with incident AD and a non-demented comparison group of elderly men.

METHODS

Participants are Japanese-American men from the Honolulu Asia Aging Study (HAAS). The HAAS cohort has been followed since 1965. The following analysis is based on a sub-group of men who underwent MRI examination in 1994-1996. Participants were diagnosed with incident AD (n=24: age=82.5+/-4.6) or were not demented (n=102: age=83.0+/-5.9). One reader, blinded to dementia diagnosis, manually outlined the left and right hippocampal formation using published criteria. We used 3D structural shape analysis methods developed at the Laboratory of Neuro Imaging (LONI) to compare regional variation in hippocampal diameter between the AD cases and the non-demented comparison group.

RESULTS

Mean total hippocampal volume was 11.5% smaller in the AD cases than the non-demented controls (4903+/-857 mm(3) vs. 5540+/-805 mm(3)), with a similar size difference for the median left (12.0%) and median right (11.6%) hippocampus. Shape analysis showed a regional pattern of shape difference between the AD and non-demented hippocampus, more evident for the hippocampal body than the head, and the appearance of more consistent differences in the left hippocampus than the right. While assignment to a specific sub-region is not possible with this method, the surface changes primarily intersect the area of the hippocampus body containing the CA1 region (and adjacent CA2 and distal CA3), subiculum, and the dentate gyrus-hilar region.

Analysis and validation of automated skull stripping tools: a validation study based on 296 MR images from the Honolulu Asia aging study

As population-based epidemiologic studies may acquire images from thousands of subjects, automated image post-processing is needed. However, error in these methods may be biased and related to subject characteristics relevant to the research question. Here, we compare two automated methods of brain extraction against manually segmented images and evaluate whether method accuracy is associated with subject demographic and health characteristics. MRI data (n = 296) are from the Honolulu Asia Aging Study, a population-based study of elderly Japanese-American men. The intracranial space was manually outlined on the axial proton density sequence by a single operator. The brain was extracted automatically using BET (Brain Extraction Tool) and BSE (Brain Surface Extractor) on axial proton density images. Total intracranial volume was calculated for the manually segmented images (ticvM), the BET segmented images (ticvBET) and the BSE segmented images (ticvBSE). Mean ticvBSE was closer to that of ticvM, but ticvBET was more highly correlated with ticvM than ticvBSE. BSE had significant over (positive error) and underestimated (negative error) ticv, but net error was relatively low. BET had large positive and very low negative error. Method accuracy, measured in percent positive and negative error, varied slightly with age, head circumference, presence of the apolipoprotein eepsilon4 polymorphism, subcortical and cortical infracts and enlarged ventricles. This epidemiologic approach to the assessment of potential bias in image post-processing tasks shows both skull-stripping programs performed well in this large image dataset when compared to manually segmented images. Although method accuracy was statistically associated with some subject characteristics, the extent of the misclassification (in terms of percent of brain volume) was small.