In geriatric research of non-fatal events, participants often die during the study follow-up without having the non-fatal event of interest. Cause-specific (CS) hazard regression and Fine-Gray (FG) ...subdistribution hazard regression are the two most common estimation approaches addressing such competing risk. We explain how the conventional CS approach and the FG approach differ and why many FG estimates of associations are counter-intuitive. Additionally, we clarify the indirect link between models for hazard and models for cumulative incidence. The methodologies are contrasted on data from the Cardiovascular Health Study, a population-based study in adults aged 65 years and older.
INTRODUCTION:Reduced estimated glomerular filtration rate (eGFR) is associated with increased risk of cardiovascular disease among people living with HIV (PLWH). It is unclear whether eGFR equations ...incorporating cystatin C (CysC) measurements are more predictive of preclinical cardiovascular disease than those using only creatinine (Cr).
OBJECTIVES:The study aimed to determine which of the 3 Chronic Kidney Disease Epidemiology (CKD-EPI) eGFR equations is most associated with carotid intima media thickness (CIMT) and coronary artery calcium (CAC) score.
METHODS:This cross-sectional analysis of pooled data from 3 large cohorts compared the associations between the 3 CKD-EPI eGFR equations (Cr, CysC, and Cr-CysC) with CIMT and CAC score using multivariable regression analysis. eGFR and CIMT were analyzed as continuous variables. CAC scores were analyzed as a binary variable (detectable calcification versus nondetectable) and as a log10 Agatston score in those with detectable CAC.
RESULTS:Thousand four hundred eighty-seven participants were included, and of these 910 (562 HIV+ and 348 HIV−) had CIMT measurements and 366 (296 HIV+ and 70 HIV−) had CAC measurements available. In HIV− participants, GFR estimated by any CKD-EPI equation did not significantly correlate with CIMT or CAC scores. When PLWH were analyzed separately including HIV-specific factors, only GFR estimated using Cr-Cys C correlated with CIMT β= −0.90, 95% CI(−1.67 to −0.13) μm; P = 0.023. Similarly, eGFR correlated with Agatston scores only when using cystatin C-based eGFR β= −8.63, 95% CI(−16.49 to −0.77) HU; P = 0.034. Associations between other eGFR formulas and CAC did not reach statistical significance.
CONCLUSIONS:In PLWH, preclinical atherosclerosis may be more closely correlated with eGFR using formulae that incorporate CysC measurements than Cr alone.
Context:
Fibroblast growth factor 23 (FGF23) is a phosphaturic hormone that also inhibits calcitriol synthesis.
Objective:
Our objective was to evaluate the relationships of plasma FGF23 ...concentrations with bone mineral density (BMD) and hip fracture in community-dwelling older adults.
Design and Setting:
Linear regression and Cox proportional hazard models were used to examine the associations of plasma FGF23 concentrations with BMD and incident hip fracture, respectively. Analyses were also stratified by chronic kidney disease.
Participants:
Participants included 2008 women and 1329 men ≥65 years from the 1996 to 1997 Cardiovascular Health Study visit.
Main Outcome Measures:
Dual x-ray absorptiometry measured total hip (TH) and lumbar spine (LS) BMD in 1291 participants. Hip fracture incidence was assessed prospectively through June 30, 2008 by hospitalization records in all participants.
Results:
Women had higher plasma FGF23 concentrations than men (75 56–107 vs 66 interquartile range = 52–92 relative units/mL; P < .001). After adjustment, higher FGF23 concentrations were associated with greater total hip and lumbar spine BMD in men only (β per doubling of FGF23 = 0.02, with 95% confidence interval CI = 0.001–0.04 g/cm2, and 0.03 with 95% CI = 0.01–0.06 g/cm2). During 9.6 ± 5.1–11.0 years of follow-up, 328 hip fractures occurred. Higher FGF23 concentrations were not associated with hip fracture risk in women or men (adjusted hazard ratio = 0.95, with 95% CI = 0.78–1.15, and 1.09 with 95% CI = 0.82–1.46 per doubling of FGF23). Results did not differ by chronic kidney disease status (P > .4 for interactions).
Conclusions:
In this large prospective cohort of community-dwelling older adults, higher FGF23 concentrations were weakly associated with greater lumbar spine and total hip BMD but not with hip fracture risk.
Context:
Data on thyroid function in the oldest old are sparse, and existing studies show conflicting evidence on the relationship between thyroid function and mortality in this age group.
Objective:
...We describe longitudinal changes in thyroid function in a cohort of elderly individuals and determine the relationship between thyroid function and mortality.
Design, Setting, and Participants:
Eight hundred forty-three participants in the Cardiovascular Health Study All Stars Study who were not taking thyroid medications and had thyroid function testing in 2005–2006 (mean age 85 yr).
Main Outcome Measure:
Thyroid-stimulating hormone (TSH), free T4 (FT4), total T3, and thyroid peroxidase antibody status were measured in 1992–1993 and 2005–2006. Deaths were ascertained through February 2011.
Results:
There was a statistically significant 13% increase in TSH, 1.7% increase in FT4, and 13% decrease in total T3 over the 13-yr period. Two hundred eighty-seven deaths occurred over a median follow-up of 5.1 yr. There was no association between subclinical hypothyroidismhazard ratio (HR) 0.97, 95% confidence interval (CI) 0.66–1.43, TSH level (HR per milliunits per liter 0.94, 95% CI 0.88–1.01), or persistent thyroid peroxidase antibody positivity (HR 1.09, 95% CI 0.62–1.92), and death. However, FT4 was positively associated with death (HR per nanograms per deciliter 2.57, 95% CI 1.32–5.02).
Conclusions:
TSH increased over time in these older individuals. This elevation was not associated with increased or decreased mortality, although higher FT4 levels were associated with death. These findings raise concern for treatment of mild elevations of TSH in advanced age. Further studies are needed to determine the potential benefit of treating age-related changes in thyroid function.
Circulating immune cells have gained interest as biomarkers of hepatic steatosis. Data on the relationships between immune cell subsets and early-stage steatosis in population-based cohorts are ...limited.
This study included 1,944 asymptomatic participants of the Multi-Ethnic Study of Atherosclerosis (MESA) with immune cell phenotyping and computed tomography measures of liver fat. Participants with heavy alcohol use were excluded. A liver-to-spleen ratio Hounsfield units (HU) <1.0 and liver attenuation <40 HU were used to diagnose liver fat presence and >30% liver fat content, respectively. Logistic regression estimated cross-sectional associations of immune cell subsets with liver fat parameters adjusted for risk factors. We hypothesized that higher proportions of non-classical monocytes, Th1, Th17, and memory CD4
T cells, and lower proportions of classical monocytes and naive CD4
T cells, were associated with liver fat. Exploratory analyses evaluated additional immune cell phenotypes (n = 19).
None of the hypothesized cells were associated with presence of liver fat. Higher memory CD4
T cells were associated with >30% liver fat content, but this was not significant after correction for multiple hypothesis testing (odds ratio (OR): 1.31, 95% confidence interval (CI): 1.03, 1.66). In exploratory analyses unadjusted for multiple testing, higher proportions of CD8
CD57
T cells were associated with liver fat presence (OR: 1.21, 95% CI: 1.02, 1.44) and >30% liver fat content (OR: 1.34, 95% CI: 1.07, 1.69).
Higher circulating memory CD4
T cells may reflect liver fat severity. CD8
CD57
cells were associated with liver fat presence and severity, but replication of findings is required.
Current knowledge of the genetic architecture of key reproductive events across the female life course is largely based on association studies of European descent women. The relevance of known loci ...for age at menarche (AAM) and age at natural menopause (ANM) in diverse populations remains unclear. We investigated 32 AAM and 14 ANM previously-identified loci and sought to identify novel loci in a trans-ethnic array-wide study of 196,483 SNPs on the MetaboChip (Illumina, Inc.). A total of 45,364 women of diverse ancestries (African, Hispanic/Latina, Asian American and American Indian/Alaskan Native) in the Population Architecture using Genomics and Epidemiology (PAGE) Study were included in cross-sectional analyses of AAM and ANM. Within each study we conducted a linear regression of SNP associations with self-reported or medical record-derived AAM or ANM (in years), adjusting for birth year, population stratification, and center/region, as appropriate, and meta-analyzed results across studies using multiple meta-analytic techniques. For both AAM and ANM, we observed more directionally consistent associations with the previously reported risk alleles than expected by chance (p-valuesbinomial≤0.01). Eight densely genotyped reproductive loci generalized significantly to at least one non-European population. We identified one trans-ethnic array-wide SNP association with AAM and two significant associations with ANM, which have not been described previously. Additionally, we observed evidence of independent secondary signals at three of six AAM trans-ethnic loci. Our findings support the transferability of reproductive trait loci discovered in European women to women of other race/ethnicities and indicate the presence of additional trans-ethnic associations both at both novel and established loci. These findings suggest the benefit of including diverse populations in future studies of the genetic architecture of female growth and development.
We performed a hypothesis-generating phenome-wide association study (PheWAS) to identify and characterize cross-phenotype associations, where one SNP is associated with two or more phenotypes, ...between thousands of genetic variants assayed on the Metabochip and hundreds of phenotypes in 5,897 African Americans as part of the Population Architecture using Genomics and Epidemiology (PAGE) I study. The PAGE I study was a National Human Genome Research Institute-funded collaboration of four study sites accessing diverse epidemiologic studies genotyped on the Metabochip, a custom genotyping chip that has dense coverage of regions in the genome previously associated with cardio-metabolic traits and outcomes in mostly European-descent populations. Here we focus on identifying novel phenome-genome relationships, where SNPs are associated with more than one phenotype. To do this, we performed a PheWAS, testing each SNP on the Metabochip for an association with up to 273 phenotypes in the participating PAGE I study sites. We identified 133 putative pleiotropic variants, defined as SNPs associated at an empirically derived p-value threshold of p<0.01 in two or more PAGE study sites for two or more phenotype classes. We further annotated these PheWAS-identified variants using publicly available functional data and local genetic ancestry. Amongst our novel findings is SPARC rs4958487, associated with increased glucose levels and hypertension. SPARC has been implicated in the pathogenesis of diabetes and is also known to have a potential role in fibrosis, a common consequence of multiple conditions including hypertension. The SPARC example and others highlight the potential that PheWAS approaches have in improving our understanding of complex disease architecture by identifying novel relationships between genetic variants and an array of common human phenotypes.
Background:
The association between changes in IGF-I and IGF binding protein (IGFBP) levels and mortality in older adults is unknown.
Study Design:
Participants were 997 persons 77 to 100 yr old ...enrolled in the Cardiovascular Health Study All Stars Study. Plasma levels of IGF-I, IGFBP-1, and IGFBP-3 were assessed at two study examinations (1996–1997 and 2005–2006). Mortality was assessed between 2006 and 2010.
Results:
Cumulative mortality (CM) was similar among individuals who had at least 10% decreases over time in IGF-I levels (CM = 29.6%), individuals who had at least 10% increases over time in IGF-I levels (CM = 24.7%), and individuals who had IGF-I levels remaining within ±10% over time (CM = 23.5%). Adjusted for age, sex, race, diabetes, body mass index, creatinine, albumin, and C-reactive protein, decreasing IGF-I level had no significant association with overall cancer mortality or noncancer mortality. Levels of IGFBP-1 increased markedly over time by 38% (median). Individuals with the largest increases in IGFBP-1 level over time had significantly increased risk of mortality. The adjusted hazard ratio per sd of IGFBP-1 change was 1.40 for overall cancer mortality (95% confidence interval = 1.10, 1.77; P = 0.01) and 1.14 for noncancer mortality (95% confidence interval = 1.02, 1.27; P = 0.02). Changes in IGFBP-3 levels were not significantly associated with mortality.
Conclusion:
Among older adults, decreasing IGF-I level over time does not predict subsequent all-cause mortality, whereas increasing IGFBP-1 predicts increased risk of mortality.
The Metabochip is a custom genotyping array designed for replication and fine mapping of metabolic, cardiovascular, and anthropometric trait loci and includes low frequency variation content ...identified from the 1000 Genomes Project. It has 196,725 SNPs concentrated in 257 genomic regions. We evaluated the Metabochip in 5,863 African Americans; 89% of all SNPs passed rigorous quality control with a call rate of 99.9%. Two examples illustrate the value of fine mapping with the Metabochip in African-ancestry populations. At CELSR2/PSRC1/SORT1, we found the strongest associated SNP for LDL-C to be rs12740374 (p = 3.5 × 10(-11)), a SNP indistinguishable from multiple SNPs in European ancestry samples due to high correlation. Its distinct signal supports functional studies elsewhere suggesting a causal role in LDL-C. At CETP we found rs17231520, with risk allele frequency 0.07 in African Americans, to be associated with HDL-C (p = 7.2 × 10(-36)). This variant is very rare in Europeans and not tagged in common GWAS arrays, but was identified as associated with HDL-C in African Americans in a single-gene study. Our results, one narrowing the risk interval and the other revealing an associated variant not found in Europeans, demonstrate the advantages of high-density genotyping of common and rare variation for fine mapping of trait loci in African American samples.
In genomic research phenotype transformations are commonly used as a straightforward way to reach normality of the model outcome. Many researchers still believe it to be necessary for proper ...inference. Using regression simulations, we show that phenotype transformations are typically not needed and, when used in phenotype with heteroscedasticity, result in inflated Type I error rates. We further explain that important is to address a combination of rare variant genotypes and heteroscedasticity. Incorrectly estimated parameter variability or incorrect choice of the distribution of the underlying test statistic provide spurious detection of associations. We conclude that it is a combination of heteroscedasticity, minor allele frequency, sample size, and to a much lesser extent the error distribution, that matter for proper statistical inference.