Andrade, Mariza de

The United Kingdom‐Primary Biliary Cholangitis (UK‐PBC) risk scores are a set of prognostic models that estimate the risk of end‐stage liver disease in patients with PBC at 5‐, 10‐ and 15‐year intervals. They have not been externally validated outside the United Kingdom. In this retrospective, external validation study, data were abstracted from outpatient charts and discrimination and calibration of the UK‐PBC risk scores were assessed. A total of 464 patients with PBC treated with ursodeoxycholic acid were included. The median diagnosis age was 52.4 years, and 88% were female patients. The cumulative incidence of events was 6%, 9%, and 15% at 5, 10, and 15 years, respectively. Concordance (c‐statistic) was 0.88, 0.85, and 0.84 using the 5‐, 10‐ and 15‐year risk scores, respectively, which was slightly lower than values observed in the United Kingdom validation cohort. Using the 5‐year risk score, more events were observed than predicted (25 versus 16.8; P = 0.046); using the 10‐year risk score, there was no difference between the observed and predicted number of events (35 versus 44.9; P = 0.14); conversely, using the 15‐year risk score, fewer events were observed than predicted (46 versus 67.5; P = 0.009). Limiting evaluation by the 15‐year UK‐PBC risk score to those with >10 years of follow‐up demonstrated no difference between observed and predicted events. Using the 5‐year risk score, patients within the highest quartile had statistically significant worse event‐free survival compared to the rest of the cohort: 82% versus 98% at 5 years, 73% versus 97% at 10 years, and 58% versus 93% at 15 years. Conclusion: In patients assessed at a North American tertiary medical center, the UK‐PBC risk score had excellent discrimination and was reasonably calibrated both in the short and long term. (Hepatology Communications 2018;2:676‐682)

Background: Certain diseases can occur with and without a trigger. We use Venous Thromboembolism (VTE) as our example to identify genetic interaction with pregnancy in women with VTE during pre- or postpartum. Pregnancy is one of the major risk factors for VTE as it accounts for 10% of maternal deaths. Methods: We performed a whole genome association analysis using the Cox Proportional Hazard (CoxPH) model adjusted for covariates to identify genetic variants associated with the time-to-event of VTE related to pre- or postpartum during the childbearing age of 18–45 years using a case-only design in a cohort of women with VTE. Women with a VTE event after 45 years of age were censored and contributed only follow-up time. Results: We identified two intragenic single nucleotide polymorphisms (SNPs) at genome-wide significance in the PURB gene located on chromosome 7, and two additional intragenic SNPs, one in the LINGO2 gene on chromosome 9 and one in RDXP2 on chromosome X. Conclusions: We showed that the time-to-event model is a useful approach for identifying potential hazard-modification of the genetic variants when the event of interest (VTE) occurs due to a risk factor (pre- or post-partum).

Gene-environment (GE) interaction has important implications in the etiology of complex diseases that are caused by a combination of genetic factors and environment variables. Several authors have developed GE analysis in the context of independent subjects or longitudinal data using a gene-set. In this paper, we propose to analyze GE interaction for discrete and continuous phenotypes in family studies by incorporating the relatedness among the relatives for each family into a generalized linear mixed model (GLMM) and by using a gene-based variance component test. In addition, we deal with collinearity problems arising from linkage disequilibrium among single nucleotide polymorphisms (SNPs) by considering their coefficients as random effects under the null model estimation. We show that the best linear unbiased predictor (BLUP) of such random effects in the GLMM is equivalent to the ridge regression estimator. This equivalence provides a simple method to estimate the ridge penalty parameter in comparison to other computationally-demanding estimation approaches based on cross-validation schemes. We evaluated the proposed test using simulation studies and applied it to real data from the Baependi Heart Study consisting of 76 families. Using our approach, we identified an interaction between BMI and the Peroxisome Proliferator Activated Receptor Gamma (PPARG) gene associated with diabetes.

Abstract Lung cancer is the leading cancer-related cause of death in the United States. We had previously performed multipoint linkage analysis on families with a strong family history of lung cancer and found significant linkage to the 6q25 region. In order to find the source of that linkage signal, we performed targeted sequencing 6q23-6q27 on 75 individuals from the 9 most highly linked families. We performed two types of parametric linkage analysis, using an autosomal dominant mode of inheritance with 10% penetrance for carriers and a 1% phenocopy rate using a disease allele frequency of 1%. The first type was a two-point analysis using an Elston-Stewart algorithm. This approach did not lead to any genome-wide significant results but demonstrated significant heterogeneity throughout the nine families.In an effort to recover the power from the original multipoint analysis, we performed a regional linkage analysis using SEQLinkage and MERLIN. SEQLinkage built regional haplotypes that corresponded to a gene or an intergenic region - the regions were based on a customized map of our design. The regional markers were multiallelic which allowed for greater information content and were similar to the microsatellites that were used in the original multipoint analysis. The regional markers were then analyzed in a two-point linkage analysis via MERLIN. This allowed us to identify two genome-wide significant signals at PACRG-AS1 at 6q26 (HLOD = 3.4) and SAMD5 at 6q24 (3.3). The PACRG-AS1 is novel, though it is associated with the known lung cancer gene PARK2. SAMD5 may be involved in lung cancer cell proliferation. The heterogeneity of the two signals was particularly interesting. The PACRG-AS1 was driven primarily by two families. The SAMD5 marker was not being strongly driven by any of the families and appears to be a small cumulative effect across the nine families. In addition, several families have large, but non-significant LOD scores at other loci across the region. This further reinforces the locus heterogeneity within the region, and it is likely that our two significant signals here are not the only variants affecting the phenotype. We further attempted to localize the signals by running SEQLinkage using a custom map where genes are broken into exons and introns. Though this resulted in no significant markers, the highest signal was located in the intronic region of SAMD5 (HLOD = 3) and several other suggestive signals were localized to intronic regions of good candidate genes like SASH1 and ARID1B. Examination of predicted effects of the candidate regulatory variants using eQTL databases is ongoing. This is a region full of promising candidates, and it is likely that the two significant signals found here are just part of many that could be affecting lung cancer risk. We plan to do further analysis within the individual families to elucidate any more genes affecting this signal. Citation Format: Anthony M. Musolf, Claire L. Simpson, Bilal A. Moiz, Candace Middlebrooks, Mariza de Andrade, Diptasri Mandal, Colette Gaba, Ping Yang, Yafang Li, Ming You, Elena Y. Kupert, Marshall W. Anderson, Ann G. Schwartz, Susan M. Pinney, Christopher I. Amos, Joan E. Bailey-Wilson. A study in locus heterogeneity: Targeted sequencing analysis of 6q reveals multiple significant loci as the source of a previous linkage peak in familial lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4290. doi:10.1158/1538-7445.AM2017-4290

Abstract Lung cancer (LC) is the leading cancer killer of Americans; an estimated 158,000 people will die in the U.S. from LC in 2016. While it is well-known that LC risk is affected by the environment, particularly tobacco smoking, there is a substantial genetic risk to LC also. We examined genotype data from Illumina HumanCore-12v1-0 array (297,000 SNPs) on 175 individuals in 25 extended families with a strong history of LC recruited by the Genetic Epidemiology of Lung Cancer Consortium. The purpose of this study is to determine which families are segregating a high-penetrance genetic risk haplotype so that the most informative families can be selected for DNA sequencing studies. Quality control was performed to remove SNPs and individuals with greater than 1% missingness as well as monomorphic SNPs. SNPs with Mendelian errors in more than one family were dropped. Identity-by-descent values were calculated to confirm correct familial relationships; 1 individual was dropped. After quality control we were left with approximately 245,000 SNPs for analysis. We performed three types of parametric linkage analyses using an autosomal dominant model with 40% penetrance in carriers and 1% penetrance in non-carriers. A disease allele frequency of 0.01 was used. Standard single variant two-point analysis between the disease and each marker was performed using TwoPointLods. Multipoint linkage analysis was performed using SimWalk2. We also performed regional-based linkage analyses using SEQLinkage and MERLIN. SEQLinkage builds a multiallelic regional marker (similar to a microsatellite) that corresponds to a gene or a portion of a gene. Two-point linkage analyses were then performed on the regional markers using MERLIN. We identified five loci that were genome-wide significant (HLOD score ≥ 3.3) from the regional-based linkage analyses on 18p11.23 (HLOD = 4.1), 2p22.2 (3.9), 14q13.1 (3.7), 16p13.1 (3.4), and 20q13.11 (3.4). It is particularly exciting that the scores centered on prospective cancer genes. Our highest score was centered on PTPRM, a protein tyrosine phosphatase on chromosome 18 that has been implicated as a LC oncogene. The signal on 20q13 also centered on another protein tyrosine phosphatase (PTPRT) that has been shown to be mutated in LC cells. The source of the signal on 16p13 was RNA binding protein RBFOX1, which has been shown to be deleted in LC cell lines and the source of the signal on 2p22 was LRP1B, a LDL receptor-related protein that is often inactivated in LC cells. The source of the final signal on 14q13 was NPAS3, a transcription factor that is a tumor suppressor in brain tumors. It should be noted that all previous evidence linking these genes to cancer was based on somatic mutations; this is the first time any of these genes has been shown to be significantly linked to germline disease risk in a family-based study. We plan to perform targeted sequencing on the linked regions to elucidate the exact causal variant. Citation Format: Joan E. Bailey-Wilson, Anthony M. Musolf, Claire L. Simpson, Mariza de Andrade, Diptasri Mandal, Colette G. Gaba, Ping Yang, Ming You, Elena Y. Kupert, Marshall W. Anderson, Ann G. Schwartz, Susan M. Pinney, Christopher I. Amos. Familial lung cancer is significantly linked to cancer-associated genes on five chromosomes [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4268. doi:10.1158/1538-7445.AM2017-4268

A number of familial diseases have an age‐of‐onset component, which can be considered as a censored quantitative trait. However, few software resources are available for the use of time‐to‐event endpoints in linkage analysis. The purpose of this analysis was to examine the use of martingale residuals from Cox survival models as quantitative traits for familial diseases with variable age at onset. We used these residuals as quantitative traits in variance components linkage scans for the 50 replicates of the general population simulated data for chromosomes 6 and 7. The region on chromosome 6 containing markers D06G034 and D06G035 demonstrated evidence for linkage, consistent with the underlying genetic model. This analysis demonstrates the applicability of using martingale residuals as a quantitative trait in linkage analyses of diseases that depend on age of onset. © 2001 Wiley‐Liss, Inc.

We report single‐point and multipoint results from three statistical genetics software packages using a variance components approach for quantitative traits. When possible, we have shown the effects of covariates. Results from GENEHUNTER, SOLAR, and ACT software packages are compared for the quantitative trait immunoglobulin E (IgE) using chromosome 5 asthma familial data from Oxfordshire (England), Perth (Australia), and Freiburg (Germany), and using the genome‐wide German data. Although few differences among variance components results for GENEHUNTER, SOLAR, and ACT were detected on a limited region of chromosome 5, these packages showed larger disagreement in the German data for chromosomes 1, 9, and 14. © 2001 Wiley‐Liss, Inc.

We discuss methods for detecting genetic linkage for quantitative data. The usual LOD score method uses a pseudolikelihood formulation and has optimal power provided all parameters are correctly specified, but can lead to erroneous estimates of the location for the locus influencing a trait under misspecification of parameters describing the variance of the trait. Alternative methods, in which attention focuses upon modelling covariation among relatives as a function of genetic marker, similarity lead to unbiased estimates of the location and major gene heritability of the trait influencing locus. The Haseman-Elston approach uses a regression method to perform linkage analysis and its properties have been widely studied. This method is generally less powerful than variance components procedures, but the maximum likelihood-based variance components procedures require normality of the trait to ensure robustness of the genetic linkage tests (i.e. a correct false positive rate). When samples are non-randomly selected an ascertainment correction is generally required in order to obtain unbiased parameter estimates when applying variance components methods. For quantitative traits, ascertainment corrections usually condition either on the proband exceeding a threshold, or on the trait value of the proband. We summarize simulations that show that both approaches lead to similar efficiencies for estimating genetic effects. Finally, we discuss methods for analysing diseases that include time-to-onset information. A variety of methods are available for the linkage analysis of quantitative traits. Here, we have reviewed the most commonly used methods.

Objective To determine if hepatocyte growth factor (HGF), a promising biomarker of coronary heart disease (CHD) given its release into circulation in response to endothelial damage, is associated with subclinical and clinical CHD in a racial/ethnic diverse population. Methods HGF was measured in 6738 participants of the Multi-Ethnic Study of Atherosclerosis (MESA). Highest mean HGF values (pg/mL) were observed in Hispanic, followed by African, non-Hispanic white, then Chinese Americans. Results In all races/ethnicities, HGF levels were associated with older age, higher systolic blood pressure (SBP) and body mass index, lower high-density lipoprotein, diabetes and current smoking. In fully adjusted models, each SD higher HGF was associated with an average increase in coronary artery calcium (CAC) of 55 Agatston units for non-Hispanic whites (p<0.001) and 51 Agatston units for African-Americans (p=0.007) but was not in the other race/ethnic groups (interaction p=0.02). There were 529 incident CHD events, and CHD risk was 41% higher in African (p<0.001), 17% in non-Hispanic white (p=0.026) and Chinese (p=0.36), and 6% in Hispanic Americans (p=0.56) per SD increase in HGF. Conclusion In a large and diverse population-based cohort, we report that HGF is associated with subclinical and incident CHD. We demonstrate evidence of racial/ethnic heterogeneity within these associations, as the results are most compelling in African-Americans and non-Hispanic white Americans. We provide evidence that HGF is a biomarker of atherosclerotic disease that is independent of traditional risk factors.