Cardiometabolic risk in young adults with Down syndrome

Studies regarding cardiometabolic risk (CMR) for individuals with Down syndrome (DS) conflict. Our previous research in youth with DS, aged 10–20 years, found increased prevalence of dyslipidemia and prediabetes compared to matched peers without DS. Herein, we compare CMR in young adults with DS, aged 18–35 years, to a similar population‐based sample from the 2001–2018 National Health and Nutrition Examination Survey (NHANES). The group with DS had higher NonHDL‐C (mean DS 131.9 mg/dL; NHANES 126.1 p < 0.001), lower HDL‐C (DS 47.5 mg/dL; NHANES 52.2 p < 0.001), higher LDL‐C (DS 109.3 mg/dL; NHANES 105.4 p < 0.001), higher triglycerides (DS 102.9 mg/dL; NHANES 86.9 p < 0.001), but lower fasting glucose (DS 85.8 mg/dL; NHANES 95.2 p < 0.0001), lower HOMA‐IR (DS 2.17; NHANES 2.24 p = 0.0006), lower systolic (DS 109.7 mmHg; NHANES 114.6 p < 0.0001) and lower diastolic (DS 60.9 mmHg; NHANES 67.8 p < 0.0001) blood pressures. There was relationship of higher HDL‐C, triglycerides, glucose, systolic, and diastolic blood pressure with increasing BMI in the NHANES cohort which was dampened in the group with DS. These results indicate that more information is needed to guide clinicians in screening for CMR in individuals with DS.


| INTRODUCTION
Down syndrome (DS) affects 1 in 700 births Down Syndrome Fact Sheet" [National Down Syndrome Society, 2022) and is one of the most common causes of developmental disability in the United States. An estimated 200,000-250,000 individuals with DS live in the United States (2008)(2009)(2010); of these $125,000 are adults (≥18 years) (de Graaf et al., 2017;Presson et al., 2013). The increase in the size of the population with DS is in part attributable to increased life expectancy (Presson et al., 2013), as we now celebrate a life expectancy of nearly 60 years in DS ("2022 Down Syndrome Fact Sheet" [National Down Syndrome Society, 2022]). With improved survival, adult co-morbidities have emerged in DS including age-related dementia and early onset Alzheimer's disease (AD) (Nieuwenhuis-Mark, 2009), obstructive sleep apnea (OSA), and overweight/obesity (Capone et al., 2018); however, type 2 diabetes (T2D) and cardiovascular disease (CVD) have received limited attention. In adults without DS, insulin resistance, hypercholesterolemia, and inflammation have been associated with increased dementia risk (Pal et al., 2018). These findings highlight the strong potential relevance of cardiometabolic risk (CMR) in adults with DS. Additionally, in the general population, overweight/obesity, OSA, and reduced physical activity (all common in DS) (Foerste et al., 2016;Nordstrøm et al., 2013;Pitetti et al., 2013) are risk factors for T2D, adverse cardiac structure and function, and downstream cardiac outcomes such as heart failure and atrial fibrillation. Their relevance to outcomes in DS has not been systematically examined.
Individuals with DS are routinely excluded from clinical research studies of T2D and acquired CVD, and available data are conflicting.
Autopsy studies from the 1970's showed the absence of atheroma in the coronary arteries of five institutionalized adults with DS compared to five institutionalized adults without DS (Murdoch et al., 1977) and suggested that individuals with DS were protected from atherosclerotic disease. Similarly, a 2017 study comparing deceased individuals with DS to deceased individuals with sporadic AD demonstrated lower rates and severity of atherosclerosis in DS and no specific loss of arteriole wall elasticity (Head et al., 2017). Some recent epidemiologic studies have showed lower odds ratio of cardiovascular heart disease and coronary atherosclerosis in individuals with DS compared to controls (Chicoine et al., 2021;Fitzpatrick et al., 2020). A more recent multi-site study suggested that individuals with DS do not have the same relationship of biochemical markers of CMR to obesity as that seen in individuals without DS (Oreskovic et al., 2022). However, a 2003 epidemiologic study reported standardized mortality ratios (SMRs) for CV-related disease of 6, ischemic heart disease of 3.9 and diabetes-related of 11.4 for adults with DS (Hill et al., 2003). Two other large cohort studies reported increased mortality due to CVD in DS, after excluding deaths caused by congenital heart disease (CHD) (Day et al., 2005;Hermon et al., 2001;Hill et al., 2003). Thus, data regarding the CMR in adults with DS has been conflicting Our group recently compared 150 youth with DS to 100 typically developing youth (ages 10-20 years) matched for age, sex, race, ethnicity, and BMI%ile, and demonstrated that youth with DS had an increased prevalence of prediabetes and a more atherosclerotic lipid and lipoprotein profile . These data showed increased CMR for a given BMI z-score in youth with DS. However, whether this risk persists into adulthood is unclear. In a large population-based study, Aslam et al. found a similar incidence of T2DM across the lifespan between individuals with DS and those without, but this incidence was higher in youth and young adults with DS. Increased life expectancy in DS demands that we better understand the risk of obesity, characterize CMR, and identify the best risk screening measures to minimize health disparities (Virji-Babul et al., 2007) in this at-risk population The aims of this study were to: (a) examine the relationship between CMR and BMI in adults with DS and comparable controls and (b) estimate the prevalence of CMR (increased non-HDL cholesterol, increased fasting glucose, abnormal blood pressure) in young adults with DS and controls.

| MATERIALS AND METHODS
This cross-sectional study compared a historical cohort of participants with DS to a similar control group from a nationally representative sample of young adults based on 2001-2018 National Health and Nutrition Examination Survey (NHANES) data.

| Participants with Down syndrome
Using a retrospective chart review, the study identified adults aged 18-35 years with DS seen over the last 25 years in the Down Syndrome Clinic at Kennedy Krieger Institute in Baltimore, MD with a clinic visit and laboratory collection within 6 months of one another.
Excluded were those with a diagnosis of leukemia, T1D, and/or pregnancy. The study was approved by the Johns Hopkins Institutional Review Board (IRB) in clinic patients who had consented to participate in clinical data collection for research purposes.
Participants with DS had weight, height, and blood pressure measurements collected at a clinic visit. All metabolic labs, including fasting glucose, insulin and lipids were collected at outpatient laboratories following a requested minimum of 8 hours fasting. In those with data from multiple DS clinic visits/labs over time, only one set of data per participant was selected with priority given to (a) the most complete set of laboratory data and (b) shortest interval between laboratory data and clinic visit. When these factors were equal, the earliest visit was chosen, as participants would be less likely to have been exposed to external factors such as medications and comorbid diagnoses which might influence outcomes.
Clinical data were recorded in a Microsoft Access database, which was exported for use in statistical analysis software.

| NHANES participants
Participants from NHANES 2001-2018 were used as the DS comparison group as these years most closely matched the years of DS data collection. NHANES controls included: participants aged 18-35 years at visit with fasting (≥8 h) bloodwork and/or complete blood pressure measurements. NHANES participants who were pregnant, had BMI < 18 kg/m 2 , history of leukemia, history of presumed T1D or metformin use primarily to prevent diabetes, treat ovarian dysfunction, disease of stomach and duodenum, polycystic ovarian syndrome, or for irregular menstruation, were excluded. Presumption of T1D was based on diagnosis with current use of insulin, no metformin use, and age of onset for diabetes <10 years. Between 2001 through 2018, methodology for laboratory testing in NHANES changed. While most assessments remained comparable over study years, glucose and insulin did not; thus, to maintain comparability, we used the NHANES subset of years (2013-2018) for these assessments in our analyses.

| Defining abnormal metabolic parameters
Abnormal lipids levels were defined using the National Institutes of Health National Institute of Heart, Lung, and Blood Institute Adult Treatment Panel III (ATP III) guidelines (Program, 2001 In the report, NonHDL-C < 150 mg/dL is defined as acceptable and thus values ≥ 150 mg/dL were considered abnormal (T oth et al., 2012). As per the ATP III guidelines total fasting cholesterol level ≥ 200 mg/dL, high density lipoprotein cholesterol (HDL-C) < 40 mg/dL, and triglycerides ≥ 150 mg/dL were considered abnormal (Program, 2001). Low density lipoprotein cholesterol (LDL-C) was defined using two different cut off points. ATP III defines LDL-C < 130 mg/dL as optimal or near optimal and LDL-C ≥130 and <160 mg/dL as borderline high (Program, 2001). Thus, we studied the prevalence of LDL-C ≥130 and ≥160 mg/dL. Many research studies define HOMA-IR abnormality based on cut points above 1.7 to above 3.875 (Tang et al., 2015). In our study, the threshold used to define insulin resistance was HOMA-IR ≥ 3.
Abnormal blood pressure parameters were defined based on a 2017 report published by the American Heart Association and the American College of Cardiology (Whelton et al., 2018), with systolic blood pressure (BP) ≥ 130 mmHg or a diastolic BP ≥ 80 mmHg defined as elevated. Examination Surveys, 2018]) using the DOMAIN statement and the SAS or Stata SURVEY suite of procedures. Survey years were combined into one dataset, and weights were adjusted by dividing the original weight by nine to account for the number of 2-year cycles included in analysis. Weighted prevalence rates with corresponding 95% confidence intervals (CIs) were calculated, and relative standard errors were examined to confirm that none were above 30% per NHANES analytic guidelines for precision and reliability (Division of the National Health and Nutrition Examination Surveys, 2018). Bivariate analyses compared factors of interest between the group with DS and NHANES participants using Rao-Scott chi-squared tests and students' t tests for discrete and continuous variables, respectively. In the univariate analysis of lipids, we excluded participants with T2D, as well as those using metformin or statins. In the univariate analysis of fasting glucose and HOMA-IR, we excluded participants with T2D and/or metformin use.

| Statistical analysis
Multiple linear regression analyses were run with PROC SUR-VEYREG (SAS) or the survey form of general linear model analysis in Stata to assess differences between the group with DS and NHANES in measures of interest. All models included age, sex, and race/ethnicity as covariates. Models for non-HDL-, total, and LDLcholesterol also included BMI as a covariate, and models for HOMA-IR included atypical antipsychotic medication use as a covariate. Participants reporting use of antihyperlipidemic medications or as having T2D were not included in the analysis of the lipid domain, while participants reporting T2D were not included in the analysis of the glucose or HOMA-IR domains. Models evaluated evidence of interaction effects of study group with BMI and study group with atypical antipsychotic use as well as higher order (squared) effects of BMI. Only interaction and squared effects with p-values <0.05 were retained in models.
For prevalence analysis, participants reporting use of antihyperlipidemic medications were categorized as hyperlipidemia positive for each lipid categorization of interest (i.e., nonHDL-C ≥ 150 = yes, etc.) and participants reporting T2D were categorized as abnormal for glucose and HOMA-IR. Sensitivity analysis was used to examine additional cut points for LDL-C and HOMA-IR. p-Values <0.05 were considered statistically significant in all analyses.

| RESULTS
Participants with DS had higher BMI and a higher percentage of non-Hispanic White and other race/ethnicity compared to the NHANES group (Table 1). The NHANES cohort was slightly older. The groups were similar in sex distributions (Table 1). Participants with DS had substantially greater atypical antipsychotic (AAP) medication use.

| Lipids
The group with DS had worse dyslipidemia with higher nonHDL-C, lower HDL-C, higher LDL-C, and higher triglycerides (Table 2). Given the high percentage of individuals with DS on AAPs, we examined whether a relationship between lipid levels and AAP use was present. We found only weak, statistically nonsignificant relationships between each of the lipids and AAP use. Thus, AAP use was not included in regression models. In the linear regression analysis with NHANES as the reference group, individuals with DS had higher nonHDL-C, total cholesterol and LDL-C after adjusting for age, sex, race, and BMI (Table 3). There was evidence of an interaction effect between group and BMI characterized by a slower rate of linear change in HDL-C ( p < 0.001) and triglycerides (p = 0.003) across BMI levels in DS, as illustrated in Figure 1.

| Metabolic parameters
The group with DS had lower fasting glucose and HOMA-IR (Table 2).
Regression models generally agreed but introduced more nuance by finding interactions between group and BMI (p = 0.027; Figure 2) and AAP use ( p = 0.04; Figure 3) that affected fasting glucose concentration. These results showed the predicted mean glucose for NHANES participants was higher across all BMIs and the difference between NHANES and participants with DS increased with increasing BMI ( Figure 2). Among NHANES participants, the predicted mean fasting glucose did not differ by AAP use. However, among participants with DS, the predicted mean fasting glucose was lower for the subset on AAPs (Figure 3).

T A B L E 3 Linear regression predicting lipids in DS group with NHANES as the reference group.
Adjusted for age, sex, and race Adjusted for age, sex, race, and BMI

| Blood pressure
The group with DS had lower systolic and diastolic blood pressures ( Table 2). As above, regression modeling introduced nuance especially in regard to the impact of BMI. Not only was there an interaction by group, but there was also evidence of a curvilinear (BMI squared) component as illustrated in Figure 4. Predicted systolic blood pressure increased with increasing BMI for both groups, but the increase was less in the group with DS (p < 0.0001). Diastolic blood pressure was relatively unchanged across BMI levels for participants with DS but increased exponentially across BMIs for the NHANES group (p < 0.0001).
Regression analyses showed higher predicted SBP and DBP in the NHANES group compared to DS ( Figure 5). Similar to fasting glucose, the differences in SBP and DBP between the group with DS versus NHANES differed by AAP use (p = 0.017 for SBP; p = 0.005 for DBP); both were higher in NHANES. Those NHANES participants on AAPs presented higher mean SBP and DBP, while both were lower in DS using AAPs.   (Santoro et al., 2020). Our data show an increased difference between the two groups for BMI ≥ 30 and especially those on AAPs. These results, along with our previous study in youth, provide preliminary data demonstrating changes in F I G U R E 5 Comparison of predicted systolic and diastolic blood pressures for participants with DS (white) +/À AAP use compared to NHANES (black) +/À AAP use adjusted for age, sex and race ( p = 0.017 for SBP; p = 0.005 for DBP), demonstrating interaction by AAP use.

| Dyslipidemia
In 2020, the GLOBAL Workgroup published medical care guidelines for adults with DS (Tsou et al., 2020). The workgroup recommended that if individuals with DS have no coronary heart disease history, providers should screen for the appropriateness of statin therapy every 5 years starting at age 40 years, using a 10-year risk calculator as recommended by the United States. Preventive Services Task Force for adults in general (Tsou et al., 2020). This recommendation was based on two trials showing reduced risk of ischemic heart disease in adults with DS (Alexander et al., 2016;Sobey et al., 2015) and the workgroup's experience. "The clinical experience of GLOBAL Workgroup members suggests ASCVD is less common in adults with Down syndrome, but given the limitations of existing research, the GLOBAL Workgroup felt there was not sufficient justification to recommend adults with Down syndrome be treated differently than adults without Down syndrome (Tsou et al., 2020)." This statement highlights the lack of evidence available for creating screening guidelines.
In our study, young adults with DS had higher nonHDL-C, lower HDL-C, and higher LDL-C than a population-based sample of typical adults. Given the increased dyslipidemia in youth and adults with DS, studies are needed to investigate whether this translates into increased atherosclerotic disease. Since the publication of the GLOBAL workgroup guidelines, investigators have examined the prevalence of heart disease in large retrospective cohorts of adults with DS compared to control groups. Fitzpatrick et al. found that adults with DS had statistically significant lower odds of coronary heart disease, angina, atherosclerosis and myocardial infarction in comparison to the general population (Fitzpatrick et al., 2020). While Chicoine et al. also found decreased odds of coronary atherosclerosis and other heart disease in DS compared to a control group, they found no difference in rates of MI, which they postulate could be related to increased CHD and sleep apnea in individuals with sleep apnea (Chicoine et al., 2021). More studies will be needed to clarify these findings.
In addition, our previous research in youth with DS found a lack of relationship between pulse wave velocity (a non-invasive measure of cardiovascular disease risk associated with ischemic events in adults ) and its traditional determinants in non-DS

| Metabolic parameters
Our study demonstrated that patients with DS had lower fasting glucose and HOMA-IR. Our findings are supported by recent studies which found significantly lower odds ratio of T2D in people with DS compared to controls (Rivelli et al., 2022) but are in contrast to previous epidemiologic studies which showed higher prevalence of diabetes in adults with DS, although not all studies differentiated between T1D and T2D (Day et al., 2005;Hermon et al., 2001;Hill et al., 2003).
These previous epidemiologic studies also did not adjust for BMI or medication use. Based on these studies, the GLOBAL Workgroup recommended that earlier diabetes screening is warranted for adults with DS (Tsou et al., 2020). They recommend screening every 3 years beginning at age 30 rather than the American Diabetes Association electronic health records, Aslam et al. found that T2D incidence was overall similar between those with and without DS, however when broken down by age, youth, and young adults 5-34 years old with DS had a higher incidence of T2D than the control group (Aslam et al., 2022). Although not consistent with our current findings, the Aslam et al. results are in line with our group's previously published results showing an increased prevalence of prediabetes in 10-20 year-old individuals with DS compared to typically-developing controls .
Complicating the interpretation of our findings is the high use of AAPs in adults with DS ($30% in our study population). AAPs are a group of medications that have been associated with varying degrees of weight gain and hyperglycemia (Wei Xin Chong et al., 2016). There are few studies addressing the contribution of AAPs to metabolic syndrome in adults with DS and intellectual disabilities (ID), who are commonly prescribed AAPs for behavioral disorders such as self-harm and aggression. One small observational study comparing 138 antipsychotic-treated and 64 antipsychotic-naive participants with intellectual disabilities found that there was no difference in dyslipidemia between the two groups (Frighi et al., 2011). In our study, why individuals with DS on AAPs had lower fasting glucose and HOMA-IR than controls is not clear, but selection bias in which the provider was less likely to prescribe AAPs to patients with prediabetes, diabetes, or morbid obesity could be a factor. However, this is merely conjecture, and our findings should be considered preliminary to relate to future studies. The implications of AAPs therapy in individuals with DS on CMR need further study.

| Low blood pressure
Both historical reports and recent research have shown lower blood pressures in children and adults with DS (Morrison et al., 1996;Santoro et al., 2020). Our study supports the existing data. The underlying cause of low blood pressures in individuals with DS has yet to be identified, though it has been hypothesized that it is related to dysautonomia (Morrison et al., 1996). Given the increased rate of comorbid conditions in DS such as CHD and moyamoya syndrome that involve monitoring of blood pressures, it is important to take into account the relative lower blood pressures at baseline of those with DS, when assessing and caring for these comorbid conditions. Similar to the relationship of fasting glucose with BMI, the difference in blood pressure between NHANES and the group with DS becomes larger with increasing BMI. Additionally, the group with DS on AAPs had lower blood pressure readings. The role of AAPs in tempering blood pressure in individuals with DS again warrants further study.

| Limitations
One of the primary limitations of this study is its historical nature, with much of the data for individuals with DS accrued across a 25-year time span and in a clinical setting. Clinical settings are generally not as rigorous and controlled as a clinical research setting. Laboratory data collection was not standardized among participants with DS and it is likely that different assays were used in the outpatient clinical labs, especially given the prolonged study period. While vital sign measurements for the group with DS were obtained in a single clinic, these measurements we not done with the same precision and accuracy as done in a research setting and were not always done by the same person. Additionally, our cohort with DS is from a single clinical site specializing in the treatment of individuals with DS and thus may not be generalizable to all individuals with DS. AAP use was disproportionally higher in participants with DS compared to the NHANES group, complicating analyses and requiring further studies to confirm our results.
Of note, in the NHANES cohort, we were unable to exclude individuals with DS as this syndrome is not specified, although the numbers are likely a very small percentage of the NHANES group, given the prevalence of DS. Additionally, while we attempted to exclude individuals with T1D based on age of diagnosis and insulin use, some individuals with T1D may not have been excluded from the NHANES group. However, given the large size of the NHANES cohort and relatively low prevalence of T1D in the population, these limitations are not likely to have a significant impact on results.

| CONCLUSIONS
This analysis of CMR in young adults with DS compared to an NHANES control group serves as preliminary data for future studies.
Our results demonstrated that while individuals with DS appeared to have worse dyslipidemia, this difference was attenuated at higher BMIs. Additionally, individuals with DS had lower fasting glucose and blood pressure, and this difference was magnified at higher BMI. The impact of AAPs on insulin resistance and blood pressure in DS is unclear and warrants further study. Our results and inconsistencies in the existing literature point to the need for rigorous data to help inform the clinical care of adults with DS. Future studies will also need to investigate the relationships of traditional risk factors such as dyslipidemia to relevant clinical outcomes, such as micro-and macro-vascular disease and potentially dementia. Evidence-informed recommendations on CMR screening of adults with DS are critically needed to guide clinicians caring for individuals with DS, who are living longer lives.