HostSeq: a Canadian whole genome sequencing and clinical data resource.

HostSeq was launched in April 2020 as a national initiative to integrate whole genome sequencing data from 10,000 Canadians infected with SARS-CoV-2 with clinical information related to their disease experience. The mandate of HostSeq is to support the Canadian and international research communities in their efforts to understand the risk factors for disease and associated health outcomes and support the development of interventions such as vaccines and therapeutics. HostSeq is a collaboration among 13 independent epidemiological studies of SARS-CoV-2 across five provinces in Canada. Aggregated data collected by HostSeq are made available to the public through two data portals: a phenotype portal showing summaries of major variables and their distributions, and a variant search portal enabling queries in a genomic region. Individual-level data is available to the global research community for health research through a Data Access Agreement and Data Access Compliance Office approval. Here we provide an overview of the collective project design along with summary level information for HostSeq. We highlight several statistical considerations for researchers using the HostSeq platform regarding data aggregation, sampling mechanism, covariate adjustment, and X chromosome analysis. In addition to serving as a rich data source, the diversity of study designs, sample sizes, and research objectives among the participating studies provides unique opportunities for the research community.

Using cancer registry data: agreement in cause-of-death data between the Ontario Cancer Registry and a longitudinal study of breast cancer patients.

Data from the Ontario Cancer Registry (OCR) were compared with data from a multi-centred prospective cohort of 1655 node-negative breast cancer patients with intensive clinical follow-up. Agreement in cause of death was evaluated using kappa statistics. The accuracy of OCR classification was evaluated against the Mount Sinai Hospital (MSH) study oncologist's interpretation of intensely followed, cohort-collected data as the reference standard. The two sources showed a high level of agreement (kappa statistic [kappa] = 0.88; 95% confidence interval [CI]: 0.86, 0.90) in vital status and cause of death. Among those cases where both sources reported a death, the OCR had a sensitivity of 95% (95% CI: 90.5, 98.8) and a specificity of 88% (95% CI: 79.6, 92.4). The OCR is a valuable tool for epidemiologic studies of breast cancer to acquire adequate and easily attainable cause-of-death information.

A tree-based model for allele-sharing-based linkage analysis in human complex diseases.

By adapting a well-known affected-relative-pair linkage model that can incorporate covariate or sub-phenotype information [Olson, 1999: Am J Hum Genet 65:1760-1769], we have developed a recursive-partitioning (RP) algorithm (tree-based model) for identifying phenotype and covariate groupings that interact with the evidence for linkage. This strategy is designed to identify subgroups of affected relative pairs demonstrating increased evidence for linkage, where subgroups are defined by pair-level or family-level covariates. After growing a full tree, we identified optimal tree size through a form of tree pruning and chose the best covariate at each split by using bootstrap algorithms. Simulation studies showed that power to detect linkage can increase in the presence of gene-environment interactions, depending on the magnitude of the interaction. As expected, however, power can decrease by examining more covariates, despite the pruning to optimize tree size. The RP model correctly identifies tree structure in a large proportion of simulations. We applied the RP model to a dataset of families with bipolar affective disorder (BPAD) where linkage regions on chromosome 18 have been previously identified. Using the all-pairs score in Genehunter, the NPL tests showed no regions with strong linkage evidence on chromosome 18. However, using the RP model, several suggestive regions were found on chromosome 18. Two covariates appeared to influence the degree of linkage: the type II BPAD subtype and a pattern of displaying mania before or after a depressive episode. The RP model has the potential to identify previously unknown gene-environmental interactions; here we have demonstrated the practical utility and potential this new methodology holds.

Genetic variation in the 5q31 cytokine gene cluster confers susceptibility to Crohn disease.

Linkage disequilibrium (LD) mapping provides a powerful method for fine-structure localization of rare disease genes, but has not yet been widely applied to common disease. We sought to design a systematic approach for LD mapping and apply it to the localization of a gene (IBD5) conferring susceptibility to Crohn disease. The key issues are: (i) to detect a significant LD signal (ii) to rigorously bound the critical region and (iii) to identify the causal genetic variant within this region. We previously mapped the IBD5 locus to a large region spanning 18 cM of chromosome 5q31 (P<10(-4)). Using dense genetic maps of microsatellite markers and single-nucleotide polymorphisms (SNPs) across the entire region, we found strong evidence of LD. We bound the region to a common haplotype spanning 250 kb that shows strong association with the disease (P< 2 x 10(-7)) and contains the cytokine gene cluster. This finding provides overwhelming evidence that a specific common haplotype of the cytokine region in 5q31 confers susceptibility to Crohn disease. However, genetic evidence alone is not sufficient to identify the causal mutation within this region, as strong LD across the region results in multiple SNPs having equivalent genetic evidence-each consistent with the expected properties of the IBD5 locus. These results have important implications for Crohn disease in particular and LD mapping in general.

Design considerations for association studies of candidate genes in families.

In genetic epidemiologic studies, investigators often use generalized linear models to evaluate the relationships between a disease trait and covariates, such as one or more candidate genes or an environmental exposure. Recently, attention has turned to study designs that mandate the inclusion of family members in addition to a proband. Standard models for analysis assume independent observations, which is unlikely to be true for family data, and the usual standard errors for the regression parameter estimates may be too large or too small, depending on the distribution of the covariates within and between families. The consequences of familial correlation on the study efficiency can be measured by a design effect that is equivalent to the relative information in a sample of unrelated individuals compared to a sample of families with the same number of individuals. We examine design effects for studies in association, and illustrate how the design effect is influenced by the intra-familial distribution of covariate values such as would be expected for a candidate gene. Typical design effects for a candidate gene range between 1.1 and 2.4, depending on the size of the family and the amount of unexplained familial correlation. These values correspond to a modest 10% increase in the required sample size up to more than doubling the requirements. Design effect values are useful in study design to compare the efficiency of studies that sample families versus independent individuals and to determine sample size requirements that account for familial correlation.

The generalizability of trial data; a comparison of beta-blocker trial participants with a prospective cohort of women taking beta-blockers in pregnancy.

OBJECTIVE: To evaluate the generalizability of randomized controlled trial (RCT) data to obstetric practice, using the example of beta-blocker therapy. STUDY DESIGN: Descriptive comparison of characteristics of participants, interventions, and outcomes assessed between a meta-analysis of trials of beta-blocker therapy in pregnancy, and a prospective cohort of beta-blocker exposed callers to Motherisk, a Teratology Information Service. RESULTS: 72 women (cohort) and 34 trials with 2474 participants (meta-analysis) were compared. The generalizability of trials was limited by an inadequate description of maternal demographics and indications for beta-blockers, and a focus on the effectiveness of beta-blockers as antihypertensives rather than on reproductive risks. CONCLUSIONS: Some of these limitations could be rectified. Others cannot, such as evaluation of teratogenicity (for ethical reasons) or all indications for a drug (for practical reasons). Reference to observational literature is, and will likely remain, necessary to address the 'harm' side of the therapeutic equation in pregnancy.

Modeling complex disease with demographic and environmental covariates and a candidate gene marker.

We randomly chose replicates 28 and 29 of the simulated data sets of Genetic Analysis Workshop 12 to model the dependence of affection status on covariates, quantitative traits, and genes using all living pedigree members. First we explored the relationship of affection status to demographic and environmental factors using logistic regression and the Cox proportional hazards models. In the second stage of our analyses the generalized transmission disequilibrium test (GTDT) was applied to nuclear families with at least two affected siblings to select single markers and high-risk alleles, which were tested in the population association analyses including all pedigree members. Multiple logistic regression models were fitted to investigate the joint contributions of genetic and nongenetic factors and a block-recursive modeling approach was adopted to study inherent hierarchical dependence structure in the data. We found that allele 2 on marker 35 of chromosome 6 is associated with higher risk compared with the other 3 alleles of this marker. In addition to this significant genetic effect, age at exam and four of the five quantitative traits (QT1, QT2, QT4, and QT5) had a significant association with the disease. Our results were obtained without knowledge of the true disease generating models.

Does simultaneous consideration of multiple regions improve disease gene localization?

Improvement in localization of disease susceptibility genes by simultaneous consideration of multiple interacting loci was assessed using the Genetic Analysis Workshop 12 simulated data. Evidence of linkage at primary loci was used to weight families for analyses at secondary loci. To identify regions linked to disease susceptibility genes, parametric and allele-sharing genome scans were performed in the extended pedigrees and nuclear families, respectively. The position of the peak allele-sharing lod was used as the estimate of a disease gene location. In weighted analyses, the positions where the greatest lod increases occurred were taken as alternative estimates of the gene locations. Variability of the location estimates of disease genes given by the unweighted and weighted analyses was compared. Similar analyses were carried out using true disease loci to determine weights. Weighted analyses did not in general improve the localization of disease genes in this data set, even with a large sample of 1,928 nuclear families, due to the features of the underlying additive liability threshold model.

Heterogeneity among smokers and non-smokers in attitudes and behaviour regarding smoking and smoking restrictions.

OBJECTIVE: To determine if smokers and non-smokers cluster into meaningful, discrete subgroups with distinguishable attitudes and behaviours regarding smoking and smoking restrictions. DESIGN: Qualitative research with 45 smokers guided development of questionnaire items applied in a population based telephone survey of 432 current smokers and 1332 non-smokers in Ontario, Canada. METHODS: Cluster analysis of questionnaire items used to categorise adult smokers and non-smokers; comparison of clusters on sociodemographic characteristics and composite knowledge and attitude scores. RESULTS: Smokers clustered in three groups. "Reluctant" smokers (16%) show more concern about other people discovering that they smoke, but parallel "easygoing" smokers (42%) in supporting restrictions on smoking and not smoking around others. "Adamant" smokers (42%) feel restrictions have gone too far, and are less likely to accommodate non-smokers. Significant gradients across categories in the expected direction were observed with respect to smoking status, stage of change, knowledge, and attitude scores, and predicted compliance with restrictions, validating the proposed typology. Non-smokers also clustered into three groups, of which the "adamant" non-smokers (45%) are the least favourably disposed to smoking. "Unempowered" non-smokers (34%) also oppose smoking, but tend not to act on it. "Laissez-faire" non-smokers (21%) are less opposed to smoking in both attitude and behaviour. A significant gradient across categories in the expected direction was observed with respect to composite scores regarding knowledge of the health effects of active and passive smoking and a composite score on support for restrictions on smoking in public places. CONCLUSION: Recognition and consideration of the types of smokers and non-smokers in the population and their distinguishing characteristics could inform the development of tobacco control policies and programmes and suggest strategies to assist implementation.

MDR1 gene expression and outcome in osteosarcoma: a prospective, multicenter study.

PURPOSE: Increased expression of the multidrug resistance gene (MDR1) has been implicated in osteosarcoma prognosis. This study represents the first prospective assessment of the prognostic value of MDR1 mRNA expression in patients with newly diagnosed extremity osteosarcoma. PATIENTS AND METHODS: A series of patients with high-grade, nonmetastatic extremity osteosarcoma were enrolled from six tertiary care institutions and observed prospectively for tumor recurrence (median follow-up duration, 30 months). All patients were treated with (neo)adjuvant chemotherapy and surgery. Tumors from 123 patients were analyzed for MDR1 mRNA expression. The association of the level of MDR1 expression with the risk of systemic recurrence was examined using survival analyses with traditional and histologic markers as prognostic factors. RESULTS: Using the highest MDR1 value for each patient, a dose-response relationship was not identified between the level of MDR1 expression and systemic relapse (relative risk, 1.15; P =.44). Analyses based on biopsy or resection values alone gave similar results (P =.11 and.41, respectively, log rank test). In multivariate analysis, large tumor size (> 9 cm) was the only significant independent predictor of systemic outcome (relative risk, 2.8; P =.002). CONCLUSION: We did not identify any correlation between MDR1 mRNA expression and disease progression in patients with osteosarcoma. It is likely that alterations in other genes are involved in resistance to chemotherapy in osteosarcoma and that they play a more critical role than MDR1 in this disease.

A study of smoking, p53 tumor suppressor gene alterations and non-small cell lung cancer.

PURPOSE: The purpose of this study was to investigate the relationship between smoking and p53 tumor suppressor gene alterations, and their association with clinicopathologic features and prognosis in non-small cell lung cancer (NSCLC). METHODS: For 111 of 119 stage I-III NSCLC patients that had been followed prospectively, tumor p53 protein accumulation was measured immunohistochemically (IHC). Staining was evaluated as a score (p53IHCS) combining intensity and percent distribution. RESULTS: Forty-eight of 111 (43%) tumors had p53IHCS > 1. p53 IHC was associated with increasing tumor size (T) (p = 0.035), nodal status (N) (p = 0.091), stage (p = 0.054), and histology: squamous cell carcinoma (70%) and adenocarcinoma (27%) (p = 0.0002). In logistic regression analysis, p53 IHC was associated with squamous cell histology versus other histotypes [adjusted odds ratio (OR)5.90, 95% confidence interval (CI) 2.34-14.90]. p53 IHC was not associated with smoking variables. In multivariate proportional hazards analysis, p53IHCS and pack-years smoked (PY), both as continuous variables, were negative prognostic factors. The adjusted hazard ratios (HR) for the survival outcome recurrence for p53IHCS and PY were 1.20 (95% CI 1.02-1.40) and 1.03 (95% CI 1.01-1.04), and for death due to recurrent disease (DRD) were 1.35 (95% CI 1.11-1.64) and 1.03 (95% CI 1.01-1.04), respectively. Comparing the 75th percentile to the baseline 0, the adjusted HR for p53IHCS (5 vs. 0) was 4.5 and for PY (55 vs. 0) was 5.1 for the outcome DRD. Both variables demonstrated a dose-response relationship with survival. CONCLUSIONS: PY and p53IHCS are significant, independent and important predictors of recurrence and DRD in stage I-III NSCLC.

Genomewide search in Canadian families with inflammatory bowel disease reveals two novel susceptibility loci.

The chronic inflammatory bowel diseases (IBDs)-Crohn disease (CD) and ulcerative colitis (UC)-are idiopathic, inflammatory disorders of the gastrointestinal tract. These conditions have a peak incidence in early adulthood and a combined prevalence of approximately 100-200/100,000. Although the etiology of IBD is multifactorial, a significant genetic contribution to disease susceptibility is implied by epidemiological data revealing a sibling risk of approximately 35-fold for CD and approximately 15-fold for UC. To elucidate the genetic basis for these disorders, we undertook a genomewide scan in 158 Canadian sib-pair families and identified three regions of suggestive linkage (3p, 5q31-33, and 6p) and one region of significant linkage to 19p13 (LOD score 4.6). Higher-density mapping in the 5q31-q33 region revealed a locus of genomewide significance (LOD score 3.9) that contributes to CD susceptibility in families with early-onset disease. Both of these genomic regions contain numerous genes that are important to the immune and inflammatory systems and that provide good targets for future candidate-gene studies.

Risks and benefits of beta-receptor blockers for pregnancy hypertension: overview of the randomized trials.

OBJECTIVE: Examine the benefits/risks of beta-blockers for pregnancy hypertension. STUDY DESIGN: Meta-analysis of relevant trials identified by comprehensive literature review (1966-97). RESULTS: Included were 30 trials for pregnancy hypertension, and four others for perinatal outcomes only. For mild chronic hypertension treated throughout pregnancy (n=2 trials), oral beta-blockers (compared with no therapy) were associated with an inconsistent increase in small for gestational age (SGA) infants (OR 2.46 [1.02, 5.92]). For mild-moderate 'late-onset' pregnancy hypertension (i.e. either chronic treated only late in pregnancy, or pregnancy-induced) (n=8 trials), oral beta-blockers (compared with no therapy) were associated with a decrease in severe hypertension (OR 0.27 [0.16, 0.451), borderline decrease in development of proteinuria (OR 0.69 [0.48, 1.02]), decrease in RDS (OR 0.33 [0.13, 0.85]), but a borderline increase in SGA infants (OR 1.47 [0.96, 2.26]). Beta-blockers were equivalent to other agents (n=15 trials). For severe 'late-onset' pregnancy hypertension (n=5 trials), i.v. labetalol produced less maternal hypotension (OR 0.13 [0.03, 0.71]) and fewer cesareans (OR 0.23 [0.13, 0.63]) than i.v. hydralazine/diazoxide. CONCLUSIONS: It is not clear that the benefits outweigh the risks when beta-blockers are used to treat mild to moderate chronic or pregnancy-induced hypertension, given the unknown overall effect on perinatal outcomes. For severe 'late-onset' pregnancy hypertension, i.v. labetalol is safer than i.v. hydralazine or diazoxide.

Fall in mean arterial pressure and fetal growth restriction in pregnancy hypertension: a meta-analysis.

BACKGROUND: We investigated the relation between fetoplacental growth and the use of oral antihypertensive medication to treat mild-to-moderate pregnancy hypertension. METHODS: The study design was a metaregression analysis of published data from randomised controlled trials. Data from a paper that was regarded as an extreme statistical outliner were excluded from primary analyses. The change in (group) mean arterial pressure (MAP) from enrolment to delivery was compared with indicators of fetoplacental growth. FINDINGS: Greater mean difference in MAP with antihypertensive therapy was associated with the birth of a higher proportion of small-for-gestational-age (SGA) infants (slope: 0.09 [SD 0.03], r2=0.48, p=0.006, 14 trials) and lower mean birthweight significant after exclusion of data from another paper regarded as an extreme statistical outliner (slope: -14.49 [6.98] r=0.16, p=0.049, 27). No relation with mean placental weight was seen (slope -2.01 [1.62], r2=0.15, p=0.25, 11 trials). INTERPRETATION: Treatment-induced falls in maternal blood pressure may adversely affect fetal growth. Given the small maternal benefits that are likely to be derived from therapy, new data are urgently needed to elucidate the relative maternal and fetal benefits and risks of oral antihypertensive drug treatment of mild-to-moderate pregnancy hypertension.

Tumor microsatellite instability and clinical outcome in young patients with colorectal cancer.

BACKGROUND: Colorectal cancer can arise through two distinct mutational pathways: microsatellite instability or chromosomal instability. We tested the hypothesis that colorectal cancers arising from the microsatellite-instability pathway have distinctive clinical attributes that affect clinical outcome. METHODS: We tested specimens of colorectal cancer from a population-based series of 607 patients (50 years of age or younger at diagnosis) for microsatellite instability. We compared the clinical features and survival of patients who had colorectal cancer characterized by high-frequency microsatellite instability with these characteristics in patients who had colorectal cancers with microsatellite stability. RESULT: We found high-frequency microsatellite instability in 17 percent of the colorectal cancers in 607 patients, and in a multivariate analysis, microsatellite instability was associated with a significant survival advantage independently of all standard prognostic factors, including tumor stage (hazard ratio, 0.42; 95 percent confidence interval, 0.27 to 0.67; P< 0.001). Furthermore, regardless of the depth of tumor invasion, colorectal cancers with high-frequency microsatellite instability had a decreased likelihood of metastasizing to regional lymph nodes (odds ratio, 0.33; 95 percent confidence interval, 0.21 to 0.53; P< 0.001) or distant organs (odds ratio, 0.49; 95 percent confidence interval, 0.27 to 0.89; P=0.02). CONCLUSION: High-frequency microsatellite instability in colorectal cancer is independently predictive of a relatively favorable outcome and, in addition, reduces the likelihood of metastases.

Disease-marker associations: power and heterogeneity in independent population samples.

We applied generalized transmission disequilibrium testing (TDT) models in combined replicates 1 through 5 from each of four simulated population samples. All analyses were conducted without knowledge of the generating models. To assess power and consistency of results within and between samples, analyses were repeated in all 25 replicates combined and in each replicate. With the exception of sample-specific findings for locus D, power was generally low to detect linkage in a genome scan or to confirm linkages detected by allele sharing in affected relatives, due to lack of linkage disequilibrium. We proposed likelihood ratio and Wald tests to detect heterogeneity among samples in disease-marker associations. Pooling data across heterogeneous populations may not improve power of the TDT method.

Meta-analyses of p53 tumor suppressor gene alterations and clinicopathological features in resected lung cancers.

p53 alterations are the most common genetic lesions observed in lung cancers. Because of the limited size of individual studies, the distributions of p53 alterations by clinicopathological features have not been well characterized. Here, we present meta-analyses describing the occurrence of p53 alterations by patient/tumor characteristics in resected lung cancer. The association between p53 alterations (gene and/or protein) and a variety of variables were evaluated by calculating pooled odds ratios (ORs) and confidence intervals (CIs). p53 alterations were detected in 46.8% of 4684 non-small cell lung cancers. p53 alterations occurred more frequently in the more strongly smoking-associated histotypes: squamous cell (51.2%) and large cell (53.7%) carcinomas versus adenocarcinomas [38.8%; OR (squamous versus adenocarcinoma) = 1.81, 95% CI = 1.55-2.11]. p53 alterations were found to be associated with T1-4, N0-3, stage I-III, differentiation, and sex: OR (T3 versus T1) = 1.62 (95% CI = 0.99-2.65), OR (N1-3 versus N0) = 1.65 (95% CI = 1.27-2.15), OR (stage III versus stage I) = 1.98 (95% CI = 1.35-2.89), OR (poorly and moderately versus well-differentiated) = 3.04 (95% CI = 1.56-5.94), and OR (male versus female) = 1.39 (95% CI = 1.10-1.75). No strong associations between p53 and ras or aneuploidy were observed. Lung cancer studies of p53 and smoking need to consider the effect of histotype, and prognostic studies of p53 should adjust for the effects of T and N or stage and histotype. The apparent association between p53 and sex may be confounded by histotype and must be evaluated by multivariate studies.

Analysis of affected sib pairs, with covariates--with and without constraints.

Covariate models have previously been developed as an extension to affected-sib-pair methods in which the covariate effects are jointly estimated with the degree of excess allele sharing. These models can estimate the differences in sib-pair allele sharing that are associated with measurable environment or genes. When there are no covariates, the pattern of identical-by-descent allele sharing in affected sib pairs is expected to fall within a small triangular region of the potential parameter space, under most genetic models. By restriction of the estimated allele sharing to this triangle, improved power is obtained in tests for genetic linkage. When the affected-sib-pair model is generalized to allow for covariates that affect allele sharing, however, new constraints and new methods for the application of constraints are required. Three generalized constraint methods are proposed and evaluated by use of simulated data. The results compare the power of the different methods, with and without covariates, for a single-gene model with age-dependent onset and for quantitative and qualitative gene-environment and gene-gene interaction models. Covariates can improve the power to detect linkage and can be particularly valuable when there are qualitative gene-environment interactions. In most situations, the best strategy is to assume that there is no dominance variance and to obtain constrained estimates for covariate models under this assumption.

Hypertension aggregates in families of kidney stone patients with high urinary excretion of uric acid.

OBJECTIVE: To determine whether kidney stone disease (KSD) and hypertension (HTN) share a common familial component that is determined by a specific urinary biochemical abnormality. DESIGN: Familial aggregation study. PATIENTS: Two hundred and twelve KSD patients, aged 18-50 years, collected a 24-h urine sample to measure the urinary excretion of uric acid, calcium, oxalate, magnesium and citrate, and were interviewed about the occurrence of HTN among first-degree relatives. OUTCOME: Positive family history (FHx) of HTN defined as two or more relatives with HTN, and HTN occurring in the fathers, mothers and siblings. RESULTS: Positive FHx of HTN was significantly associated with increasing urinary excretion of uric acid (P = 0.03) but not with the excretion of the other substances. When the patients were divided into those with and without hyperuricosuria, the adjusted odds ratio (OR) for positive FHx of HTN in a hyperuricosuric KSD patient was 3.8 (95% CI, 1.22-11.66). Separate analysis on the occurrence of HTN in the fathers, mothers and siblings of the probands indicated that hyperuricosuria is positively related to HTN occurring in the siblings of the patients (P < 0.001) but not in the fathers or in the mothers. The adjusted OR for HTN occurring in siblings of hyperuricosuric patients compared with siblings of non-hyperuricosuric patients was 3.8 (2.12-6.67). CONCLUSION: Siblings of KSD patients with hyperuricosuria had a significantly increased prevalence of HTN that could not be accounted for by age, family size, body-mass index and personal history of HTN of the probands. Additional studies need to be undertaken to determine whether this familial clustering has a genetic or environmental origin.