Louisa Gnatiuc

Background and Aims: Despite the high burden of cardiovascular disease (CVD) in Latin America and the Caribbean (LAC), risk prediction models adapted for use in LAC populations remain limited. The aim of this study was to recalibrate the Systematic COronary Risk Evaluation 2 (SCORE2) risk models for estimating 10-year CVD risk in individuals without prior CVD or diabetes mellitus in LAC, resulting in the SCORE2 Latin America and the Caribbean (SCORE2-LAC) models. Methods: Sex-specific, competing risk-adjusted SCORE2 models were recalibrated to contemporary CVD incidence across four risk regions, defined by country-level age- and sex-standardized CVD mortality rates, using region-specific risk factor distributions. Cardiovascular disease incidence was estimated from CVD mortality data and ratios of the total first-to-fatal CVD events, derived from 40 237 152 individuals (1 261 329 CVD events). External validation included 132 512 individuals without prior CVD or diabetes from 11 cohorts across seven countries (3938 CVD events). Discrimination was assessed using Harrell’s C-index. Results: In the external validation datasets, the pooled C-index of SCORE2-LAC was .740 [95% confidence interval (CI) .716–.763], with cohort-specific C-indices ranging from .674 (95% CI .588–.761) to .789 (95% CI .747–.831). Estimated CVD risk varied several-fold across the four LAC risk regions. For a 50-year-old non-smoker with a systolic blood pressure of 140 mmHg, total cholesterol of 213 mg/dL (5.5 mmol/L), and HDL cholesterol of 50 mg/dL (1.3 mmol/L), the predicted risk ranged from 3% in low-risk countries to 10% in very-high-risk countries for women, and from 4% to 10% for men. Conclusions: The SCORE2-LAC models provide region-specific contemporary estimates of 10-year CVD risk in apparently healthy individuals, enabling identification of individuals at high CVD risk across LAC.

Type 2 diabetes (T2D) is a highly heritable, polygenic disease with over 600 loci identified through genome-wide association studies (GWAS). However, despite possessing unique genetic variation shaped by demographic history and admixture, Latin American populations remain markedly underrepresented in global genomic research. To address this gap, we conducted genome- and exome-wide analyses of 19,431 T2D cases and 105,611 controls from the Mexico City Prospective Study (MCPS). We identified 86 independent GWAS associations, including 21 novel signals, 15 of which replicated in external cohorts. Risk alleles at novel loci were enriched in individuals with Indigenous American ancestry. Exome analyses revealed rare and ultra-rare missense variants with substantial risk effects at HNF1A and GCK , as well as a protein-damaging variant in SLC30A8 that reduced T2D risk by 45% in carriers. Integrative analyses indicate that T2D genetic architecture in Mexico is predominantly driven by common regulatory variation acting in the endocrine pancreas. Polygenic risk scores strongly stratified T2D risk and transferred to Indigenous Mexican populations. These findings demonstrate the power of large-scale genetic discovery in diverse populations to refine disease architecture and identify loci with potential therapeutic relevance.

Aims: SCORE2 estimates the 10-year risk of developing non-fatal and fatal cardiovascular disease (CVD), and it was designed for mostly Europeans. However, its transferability to non-European populations has not been extensively studied. The aim of this study was to evaluate the performance of the SCORE2 in ethnically and socioeconomically diverse populations, including those outside of Europe. Methods and results: We leveraged cohort studies from Mexico (N = 95,660, mean age 52.4 years, 32.6% men, mean follow-up 9.7 years), and Portugal (N = 1,216, 52.9 years, 36.9% men, 8.8 years follow-up). SCORE2 has differentially classified frequency of CVD events across levels of risk severity, ranging from 0.7–1.0% CVD events among those with low-moderate risk-profiles, vs. 6.6–9.0% among those with high risk-profiles, respectively. The area under the receiver operating curve (AROC) was 0.762 in Mexico and 0.718 in Portugal. Sensitivities (95% CI) were 85.7 (84.2–87.2) and 95.0 (86.1–99.0), and specificities were 42.8 (42.5–43.1) and 24.3 (21.9–26.9) for Mexico and Portugal, respectively. In all cases, the positive predictive values were below 7%, whereas the negative predictive values were all above 98%. Harrell’s C were 0.717 and 0.692, and Brier’s scores were 0.022 and 0.052 for Mexico and Portugal, respectively. In Mexico, SCORE2 showed good calibration, with a slight overestimation in the higher-risk categories. In Portugal, overestimation in higher-risk categories was more substantial. Conclusion: SCORE2 estimates modest to acceptable performance for predicting the 10-year CVD in Mexicans but performed less favourably in the Portuguese cohort, indicating a need for recalibration.

Aims: Most polygenic risk scores (PRSs) for coronary heart disease (CHD) are derived from European ancestry populations. This paper evaluates the performance of CHD PRSs in a Mexican population. Methods and results: 133 207 participants aged 35–79 years from the Mexico City Prospective Study were included. Eight PRSs (comprising 44–6 472 620 polymorphisms) were selected for prediction of CHD (defined as self-reported prior heart attack or angina, or CHD death before age 80 years). Logistic regression was adjusted for age, sex, and the first seven genetic principal components, before and after adjustment for other cardiovascular risk factors. The area under the receiving operating characteristic curve (‘C-statistic’) was also estimated. Of the included participants, 67% were women, the mean (±SD) age was 51 ± 12 years, and Indigenous American ancestry averaged 67%. CHD was documented for 5163 participants (3.9%), including 1901 prevalent and 3479 fatal cases. All eight PRSs were positively and log-linearly associated with CHD, with odds ratios (ORs) per 1 SD PRS increase ranging from 1.05 (95% CI, 1.03–1.08) to 1.29 (95% CI, 1.25–1.33). Associations were consistent across strata of age and ancestry and were independent of other vascular risk factors. For six PRSs, however, associations were substantially stronger in men than women. Multi-ancestry PRSs outperformed Eurocentric-ancestry PRSs. Despite remaining predictive of risk independently of established non-genetic risk factors, inclusion of a PRS into a risk model did not increase the C-statistic noticeably. Conclusion: In this Mexican population, existing PRSs predicted CHD independently of established vascular risk factors, particularly for men. PRSs better capturing genetic variation in Latin American people may further enhance risk prediction in such populations.

BACKGROUND: Observational studies relating blood pressure in middle age to mortality may underestimate lifelong effects. Mendelian randomization can reduce the impact of confounding and reverse causality and may better estimate lifelong effects of blood pressure on mortality. METHODS: Mendelian randomization analyses used 125 895 Mexico City Prospective Study participants aged 35 to 74 years at recruitment with valid genetic and other data. Cox regression, adjusted for confounders and regression dilution bias, related blood pressure to mortality in 133 027 participants aged 35 to 74 years without prior chronic disease (other than diabetes) at recruitment. RESULTS: In the genetic analyses (40 560 [32%] men; mean age 50 years, mean body mass index 29 kg/m ² ) there were 13 153 deaths before age 75 years (3478 cardiovascular, 2053 kidney, and 7622 other). Each 10 mm Hg higher genetically predicted lifelong systolic blood pressure was associated with 73% higher cardiovascular mortality at ages 35 to 74 years (rate ratio, 1.73 [95% CI, 1.44–2.06]), 42% higher kidney death (1.42 [95% CI, 1.15–1.75]), but no clear increase in death from other causes. These lifelong rate ratios were higher than those estimated by observational analyses relating blood pressure in middle age to risk. Mendelian randomization analyses of lifelong diastolic blood pressure confirmed strong associations with cardiovascular but not kidney death. Mortality rate ratios were similar for men and women and in those with versus without diabetes, and broadly similar at different ages and at different proportions of Indigenous American ancestry. Sensitivity analyses gave consistent results. CONCLUSIONS: In this Mexican population, genetically informed lifelong differences in blood pressure were strongly related to death from cardiovascular and kidney disease.

Background: Persistent hyperglycaemia in diabetes can cause weight loss, distorting the association of adiposity with mortality. We estimated the lifelong associations of genetically predicted body mass index (BMI) with 52 causes of death among 125 003 Mexican adults, in whom persistent hyperglycaemia in diabetes was common. Methods: A trans-ancestry genetic instrument for BMI (from 724 BMI-associated single-nucleotide polymorphisms) estimated the causal relevance of BMI to mortality before age 75 years, stratified by sex and adjusted for age and underlying ancestry structure, using a one-sample Mendelian randomization (MR) approach. Two-sample MR and other sensitivity analyses were also performed. Results: The genetic instrument explained 3% of the BMI variation and predicted BMI similarly in men and women. Each 5-kg/m2 higher genetically predicted BMI was associated with nearly a doubling in the risk of all-cause mortality at ages 35–74 years [13 066 deaths; hazard ratio (HR) 1.80, 95% confidence interval (CI) 1.63–2.00]. Hazard ratios were greater for vascular-metabolic (n = 7111; HR 2.15, 95% CI 1.87–2.48) than for non-vascular-metabolic causes (n = 5955; HR 1.47, 95% CI 1.27–1.71) and particularly strong for renal (n = 2034; HR 3.59, 95% CI 2.76–4.67), acute diabetic crises (n = 557; HR 2.70, 95% CI 1.64–4.44), and infective deaths (n = 811; HR 2.61, 95% CI 1.73–3.92). For all-cause mortality, HRs were somewhat greater at younger ages compared with older ages, and slightly larger in those with a higher proportion of Indigenous American ancestry. The strength of the association with mortality was reduced by more than half after simple adjustment for genetic predisposition to diabetes. Sensitivity analyses supported the main conclusions. Conclusion: In this Mexican population, genetically predicted lifelong BMI was strongly related to mortality and mediated substantially through diabetes.

Importance: There is limited population-based evidence on the prevalence of cognitive impairment in Mexico, a country with a rapidly aging population and where key risk factors, such as diabetes and obesity, are common. Objective: To describe the distribution of cognitive impairment in a sample of adults from Mexico City. Design, Setting, and Participants: This cross-sectional population-based study included participants from the Mexico City Prospective Study which 50,000 men and 100,000 women aged ≥35 years from two districts in 1998-2004. In 2015-2019 about 10,000 survivors were resurveyed with identical information from the original survey and additional assessments including a cognitive assessment. The main analyses included those aged 50-89 years with complete cognitive assessment and covariate data at resurvey. Main outcomes and measures: Cognition was assessed using the Mini Mental State Examination (MMSE) and those with cognitive impairment (MMSE ≤24) were identified. The distribution of MMSE scores and cognitive impairment by age, sex, and major disease risk factors (diabetes, hypertension, and adiposity) was analyzed among those with complete MMSE data and some degree of self-reported formal education. Results: Of the 9,288 participants aged 50-89 years at the 2015-2019 resurvey with complete data, 8,197 reported having at least some years of formal education. Among these (mean age 66 years; 31% men), their mean MMSE score was 26.2 (SD 3.6) points, 1,941 (24%) had cognitive impairment, mean body-mass index (BMI) was 28.6 (SD 5.5) kg/m ² , 3,008 (37%) had previously-diagnosed hypertension and 2,467 (30%) had previously-diagnosed diabetes. The sex- and district-standardised prevalence of cognitive impairment increased strongly with age, from 10% in those 50-59 years to 55% in those aged 80-89. At any given age, the prevalence of cognitive impairment was higher in women than in men. After accounting for the effects of age, sex, and district there was little difference in the prevalence of cognitive impairment between participants with or without diabetes, hypertension, overweight or obesity (BMI ≥25 km/m ² ), or high levels of fat mass. Conclusion and relevance: In this population of adults aged 50-89 years from two districts of Mexico City, the prevalence of cognitive impairment was high, particularly among women. The extent to which cognitive impairment relates to health outcomes in this population needs to be investigated. KEY POINTS: Question: What is the distribution of cognitive impairment among adults in Mexico City? Findings: In this cross-sectional study of 8,197 adults aged 50-89 years, the sex- and district-standardised prevalence of cognitive impairment ranged from 10% in those 50-59 years to 55% in those aged 80-89 and at any given age was higher in women than in men. There was little difference in the prevalence of cognitive impairment between participants with or without hypertension, diabetes or excess of adiposity. Meaning: Cognitive impairment is common in adults in Mexico City. Its relevance to major morbidity and mortality deserves future research.

Introduction Although higher risks of infectious diseases among individuals with diabetes have long been recognized, the magnitude of these risks is poorly described, particularly in lower income settings. This study sought to assess the risk of death from infection associated with diabetes in Mexico. Research design and methods Between 1998 and 2004, a total of 159 755 adults ≥35 years were recruited from Mexico City and followed up until January 2021 for cause-specific mortality. Cox regression yielded adjusted rate ratios (RR) for death due to infection associated with previously diagnosed and undiagnosed (HbA1c ≥6.5%) diabetes and, among participants with previously diagnosed diabetes, with duration of diabetes and with HbA1c. Results Among 130 997 participants aged 35–74 and without other prior chronic diseases at recruitment, 12.3% had previously diagnosed diabetes, with a mean (SD) HbA1c of 9.1% (2.5%), and 4.9% had undiagnosed diabetes. During 2.1 million person-years of follow-up, 2030 deaths due to infectious causes were recorded at ages 35–74. Previously diagnosed diabetes was associated with an RR for death from infection of 4.48 (95% CI 4.05–4.95), compared with participants without diabetes, with notably strong associations with death from urinary tract (9.68 (7.07–13.3)) and skin, bone and connective tissue (9.19 (5.92–14.3)) infections and septicemia (8.37 (5.97–11.7)). In those with previously diagnosed diabetes, longer diabetes duration (1.03 (1.02–1.05) per 1 year) and higher HbA1c (1.12 (1.08–1.15) per 1.0%) were independently associated with higher risk of death due to infection. Even among participants with undiagnosed diabetes, the risk of death due to infection was nearly treble the risk of those without diabetes (2.69 (2.31–3.13)). Conclusions In this study of Mexican adults, diabetes was common, frequently poorly controlled, and associated with much higher risks of death due to infection than observed previously, accounting for approximately one-third of all premature mortality due to infection.

Objective: Liver fat associates with obesity‐related metabolic disturbances and may precede incident diseases. Metabolomic profiles of liver fat in the UK Biobank were investigated.Methods: Regression models assessed the associations between 180 metabolites and proton density liver fat fraction (PDFF) measured 5 years later through magnetic resonance imaging, as the difference (in SD units) of each log metabolite measure with 1‐SD higher PDFF among those without chronic disease and not taking statins, and by diabetes and cardiovascular diseases.Results: After accounting for confounders, multiple metabolites were associated positively with liver fat (p < 0.0001 for 152 traits), particularly extremely large and very large lipoprotein particle concentrations, very low‐density lipoprotein triglycerides, small high‐density lipoprotein particles, glycoprotein acetyls, monounsaturated and saturated fatty acids, and amino acids. Extremely large and large high‐density lipoprotein concentrations had strong inverse associations with liver fat. Associations were broadly comparable among those with versus without vascular metabolic conditions, although negative, rather than positive, associations were observed between intermediate‐density and large low‐density lipoprotein particles among those with BMI ≥25 kg/m², diabetes, or cardiovascular diseases. Metabolite principal components showed a 15% significant improvement in risk prediction for PDFF relative to BMI, which was twice as great (but nonsignificant) compared with conventional high‐density lipoprotein cholesterol and triglycerides.Conclusions: Hazardous metabolomic profiles are associated with ectopic hepatic fat and are relevant to risk of vascular‐metabolic disease.