Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Jonathan Rosand

Neurocritical Care · Broad Institute

View

Dr. Jinsy Andrews

ALS · NYU Langone

View

Dr. Maisha Robinson

Hospice & Palliative Care · Mayo Clinic Florida

View

Dr. Michael Kentris

Epilepsy · St. Elizabeth Medical Center

View

Dr. Daniel Correa

Epilepsy · AECOM

View

Dr. Aaron Berkowitz

General Neurology · UCSF

View

Dr. Jose Hernan Posas

Sports Neurology · UVA

View

Dr. Michael Okun

Movement Disorders · UF

View

Dr. Joseph Sirven

Epilepsy · Mayo Clinic College of Medicine

View

Dr. Rachel Marie Salas

Sleep Medicine · Johns Hopkins

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Association of Lp(a) With Stroke and Cerebral Injury on MRI: Insights From the HCHS/SOL (Hispanic Community Health Study/Study of Latinos) and Investigation of Neurocognitive Aging MRI (SOL-INCA MRI).

    Apr 1, 2025DOI: 10.1161/strokeaha.124.048439Journal ArticleAHA

    Written by: Akhil Avunoori ChandraPriscilla Duran LucianoKatrina SwettRobert KaplanGregory A TalaveraMelissa LamarWassim TarrafFreddie MarquezParag H JoshiLinda GalloDaniela Sotres-AlvarezMorgan GianolaMartha L DaviglusDaniel L LabovitzHector GonzalezCharles DeCarliCarlos J Rodriguez

    Find these authors on Roon

    DL
    Dr. Daniel Labovitz
    Vascular Neurology · MMC

    Introduction:

    The association between ischemic/non-ischemic cerebrovascular disease with traditional lipid subtypes such as triglycerides, total cholesterol, low-density lipoprotein (LDL), and high-density lipoprotein (HDL) is well-documented.1–4 Lipoprotein(a) [Lp(a)], a lipoprotein molecule bound to apo-B100, has been extensively studied as an independent marker of cardiovascular disease, as well as a risk factor for atherosclerotic cardiovascular disease (ASCVD).5–7 The pro-thrombotic, pro-inflammatory, and pro-atherogenic effects of Lp(a) have made this molecule an important risk factor, potentially more harmful than LDL, but its relation with cerebrovascular disease and stroke is not as clear.8Several studies, including meta-analyses, cross-sectional, prospective, and genetic epidemiological studies, have shown that elevated Lp(a) levels are associated with an increased risk of ischemic stroke in non-Hispanic (NH) whites.9–13 However, other studies have found no association between Lp(a) and ischemic stroke.14,15 Although stroke is a clinical syndrome of cerebrovascular disease, other cerebrovascular disease processes can also adversely affect the structure and function of the brain and cerebral vasculature, including atherosclerosis and white matter hyperintensities (WMH) which can be detected through brain magnetic resonance imaging (bMRI).16 As of now, there is a paucity of literature regarding the association between Lp(a) and bMRI markers of cerebrovascular disease, particularly when studying underrepresented racial and ethnic groups, such as the Hispanic/Latino population, for whom stroke is the fourth most common17 disease–related underlying cause of death.To address these gaps, we leveraged a large, population-based, multi-center cohort study of Hispanic/Latino adults in the United States to specifically investigate the association between Lp(a) and multiple cerebrovascular disease related outcomes: self-reported stroke or transient ischemic attack (TIA), bMRI evidence of stroke, WMH, and silent brain infarct.

    Methods:

    The datasets generated and/or analyzed during the current study are publicly available but may be constrained by terms of specific data use and sharing agreements set forth by the National Institutes of Health (NIH) for HCHS/SOL and SOL-INCA-MRI. Data may be available upon reasonable request to the senior author and/or corresponding author.The HCHS/SOL is an ongoing, longitudinal population-based cohort study of diverse community-dwelling self-identified Hispanic/Latino adults enrolled from four US urban cities (Bronx, NY; Chicago, IL; Miami, FL and San Diego, CA). At baseline (Visit 1, 2008–2011), HCHS/SOL enrolled 16,415 individuals aged 18–74 during its baseline visit. The sampling design of HCHS/SOL has been described elsewhere.18,19 The Study of Latinos-Investigation of Neurocognitive Aging (SOL-INCA) is an HCHS/SOL ancillary study that evaluated the cognitive performance of 6377 HCHS/SOL individuals ≥50 years old, who had cognitive assessments performed at visits 1–2, and consented for an expanded cognitive test battery assessment through this ancillary study. The design and objectives of the SOL-INCA (2014–2017) study have been published elsewhere.20 SOL-INCA individuals (recruited from HCHS/SOL) served as the main source of participants for the Study of Latinos-Investigation of Neurocognitive Aging Magnetic Resonance Imaging (SOL-INCA-MRI), a sub-study of HCHS/SOL and SOL-INCA (Figure 1). The SOL-INCA-MRI (2018–2022) is an ancillary study that leverages the HCHS/SOL cohort and neurocognitive data from SOL-INCA to examine brain health and the role of vascular risk factor burden on cerebrovascular pathology and Alzheimer’s disease risk among the Hispanic/Latino population by incorporating bMRI techniques.20SOL-INCA-MRI participant selection was oversampled with individuals with cognitive impairment (defined by the National Institute on Aging Alzheimer’s Association Criteria21,22), and the cognitively healthy subjects were randomly sampled from each field center to match the individuals with cognitive impairment. In addition, younger individuals (35 to 50 years old) were sampled at random from the parent HCHS/SOL study to achieve a lifespan perspective on Hispanic/Latino brain health. Data collection for SOL-INCA-MRI is complete and includes a total of 2,668 individuals with bMRIs. Of the 2,668 participants with bMRIs, n=26 had missing Lp(a) values.For the current study, we used all available data of SOL-INCA MRI, as of August 2023, including an unweighted total of 2,642 individuals, aged 35–85 years old, who had Lp(a) levels measured in addition to completed bMRI scans. We excluded from the final analysis all entries with incomplete/missing data regarding each measured outcome: n=26 for cerebral injury, n=67 for silent brain infarction, and n=15 for WMH. This study conformed to the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) reporting guidelines.Lp(a) ascertainment in HCHS/SOL has been described.23 In brief, biorepository blood samples collected from 16,415 individuals at Visit 1 in 2008 to 2011 were used to measure Lp(a) levels. Lp(a) was measured in nanomoles per liter (nmol/L) by a validated latex-enhanced turbidimetric method (Tina-quant® Lipoprotein[a] Gen. 2 kit [Roche]). Results of 100 fresh-frozen samples were compared with those obtained by a double monoclonal antibody ELISA to test for accuracy, revealing a very low percent of observed bias, mostly due to the size polymorphism of apolipoprotein(a). Approximately 1,166 individuals had Lp(a) levels below the 2.8 nmol/L limit of quantification of the assay and were then assigned an arbitrary value of 1.4 nmol/L.bMRI acquisition and imaging analysis in SOL-INCA-MRI (2018–2022) has been described previously.24–27 In brief, SOL-INCA-MRI participants were scanned in three field centers using a GE 3T 750 MRI scanner and a Philips 3T Achieve TX MRI scanner in one field center. The MRI sequences obtained included T1 volume ≈1 mm3 resolution, three-dimensional (3D) fluid-attenuated inversion recovery (3D FLAIR), T2, T2*, and arterial spin labeling (ASL) sequences. All images were quality checked visually at every step of the segmentation process and skull stripping (removal of non-brain tissue) was performed26. bMRI images were analyzed by a neuroradiologist and a neurologist at a central core lab, University of California Davis’ Imaging of Dementia and Aging (IDeA) laboratory. WMH volumes (cm3) were obtained by a modified Bayesian probability structure–based algorithm performed on FLAIR and 3D T1 images28, and likelihood values from the Bayesian model were dichotomized to create a binary WMH mask. The presence of stroke on bMRI was determined from the size, location, and imaging characteristics of the lesion as previously described.29 The subtraction image, the proton density image, and the T2 weighted image were superimposed by the image analysis system at a three-times magnified view, aiding in interpretation of lesion characteristics.27 The signal void, best seen on T2-weighted images, was indicative of a vessel. Only lesions ≥3 mm were considered as cerebral infarcts.27The main outcomes of our study are 1) self-reported stroke or TIA (collected at the HCHS/SOL baseline visit); 2) cerebral injury, a composite outcome defined as the presence of at least one of the following: bMRI evidence of one or more brain infarcts; or self-reported stroke; or self-reported TIA; and secondary outcomes 3) log-transformed WMH, and 4) silent brain infarct (defined as no history of self-reported stroke or TIA but with positive bMRI findings of cerebral infarct). Self-reported history (yes/no) of prevalent stroke or TIA was collected at HCHS/SOL Visit 1 from response to the questions “Has a doctor ever said that you had a stroke?” and “Has a doctor ever said that you had a mini-stroke of TIA (transient ischemic attack)?”. Total cerebrum WMH volume (cm3) values were log transformed and adjusted for head size by dividing WMH by total cranial volume [TCV], to account for variance (all WMH values were >0).30Baseline characteristics included age (years), sex (female/male), body mass index (BMI, kg/m2), self-reported past medical history (hypertension, hypercholesterolemia, diabetes mellitus), current cigarette smoking, current alcohol use, systolic and diastolic blood pressure (mmHg), anti-hypertensive medication use, Hispanic/Latino (Central American, Cuban, Dominican, Mexican, Puerto-Rican, South American), education, annual household income, and acculturation measures (born outside of the United States, Spanish language preference and US residence >10 years).We included 16,333 individuals with measured Lp(a) levels (nmol/L) and data on self-reported history of prevalent stroke or TIA from HCHS/SOL Visit 1. Data from SOL-INCA MRI was utilized to identify a subset of 2,642 individuals with bMRI and Lp(a) levels. Baseline characteristics of SOL-INCA MRI individuals with versus without cerebral injury were compared using the Chi-square test for categorical variables or student’s t-test for continuous variables. The distribution of Lp(a) was presented in the form of histograms and median interquartile (IQR) range with bottom and top quartiles (Q1, Q2). Linear and multivariate logistic regression was used to obtain odds ratios (OR) and beta coefficient estimates (ß estimates) of the association of Lp(a) with primary outcomes of 1. Self-reported stroke or TIA; 2. Cerebral injury; and secondary outcomes of 3. Log-transformed WMH volume; and 4. Silent brain infarct. Lp(a) levels were log-transformed, to address the skewness of its distribution in both HCHS/SOL and SOL-INCA MRI, and modeled as a continuous variable [log-Lp(a)] and per quintiles (Q) as follows: Q1, 0-<6 nmol/L; Q2, 6-<13 nmol/L; Q3, 13-<32 nmol/L; Q4, 32-<77 nmol/L; and Q5, >77 nmol/L. For our primary analyses, outcomes of self-reported stroke or TIA and cerebral injury were assessed in the target population of HCHS/SOL and SOL-INCA MRI, respectively, while secondary analyses of log-transformed WMH and silent brain infarct were only assessed in SOL-INCA MRI. Covariates in multivariable analyses included age (in years at the time of a participant’s initial visit), sex (male vs female), LDL (mg/dL), smoking (defined as never/former vs current) and waist circumference (in cm). Multivariate analyses were categorized as follows: Model 1 (unadjusted), Model 2 (adjusted for age and sex), and fully adjusted model, Model 3 (adjusted for age, sex, LDL, smoking and waist circumference). The top Lp(a) quintile (Q5) was compared to the bottom quintile (Q1) as reference (ref.) in the primary analysis. We observed a “J-shaped” association for cerebral injury and Lp(a) and thus performed a secondary analysis comparing Q5 with the quintile with the lowest proportion of the observed outcome, Q2 (ref.). Additional sensitivity analyses were performed adjusting for 1. dyslipidemia (defined as HDL<40 mg/dL, LDL >160 mg/dL, or triglycerides ≥ 200 mg/dL); 2. hypertension (defined as blood pressure 140/90 mmHg or self-reported antihypertensive use); 3. HDL (mg/dL) and LDL (mg/dL); 4. systolic blood pressure [SBP] (mmHg) and diastolic blood pressure [DBP] (mmHg); 5. antithrombotic use; and 6. statin use. The relationship between log-Lp(a) and prevalent stroke or TIA, cerebral injury and silent brain infarct was graphically displayed using restricted cubic splines (RCS) fitted by three knots (10th, 50th, and 90th percentile). Similarly, the relationship between log-Lp(a) and log-WMH was displayed with a scatter plot with best fit linear regression line and 95% confidence lines. Complex survey analysis and sample weights are utilized for all analyses to account for the HCHS/SOL complex sample design.18 Weighting was used to account for the oversampling certain strata at screening to increase the likelihood that a selected address yielded a Hispanic/Latino household to facilitate examination of target outcomes and to partially adjust for any bias effects of non-response at the household and individual person levels at each field center. As a result, participants included in HCHS/SOL cohort were selected with unequal probabilities of selection, and these probabilities are taken into account with survey analysis and sample weights to appropriately represent the target population.Accordingly, we present unweighted counts (N’s), weighted means (with standard error, SE) and weighted percentages, %. A two-sided value of P < 0.05 was considered as statistically significant. Statistical analyses were completed using Statistical Analysis System SAS version 9.4 (RRID:SCR008567), SAS Institute, Gary, NC. Manuscript figures were created using GraphPad Prism version 8.0.0 for Windows (RRID:SCR002798), SAS 9.4 and Biorender.com. Graphical abstract was created in BioRender. Duran Luciano, P. (2025) https://BioRender.com/c66t520

    Results

    Among the overall HCHS/SOL target population, mean age ± SE was 41.1± 0.3 years, 52.0% were female, and their median IQR (Q1, Q3) Lp(a) level was 19.7 (7.3–60.6) nmol/L. Lp(a) distribution at 50th percentile was 19.9 nmol/L, 80th percentile was 76.1 nmol/L, 90th percentile was 135.9 nmol/L, 95th percentile was 184.7 nmol/L, and 99th percentile was 314.0 nmol/L [Figure S1].A total of 2.3% of the HCHS/SOL target population had a history of stroke or TIA. Median IQR (Q1, Q3) Lp(a) levels were significantly higher among individuals with self-reported stroke or TIA compared to individuals without stroke or TIA, 28.17 (9.02, 89.51) versus 19.6 (7.24, 60.29), p=0.007. The proportions of self-reported stroke or TIA were higher among the upper quintiles of Lp(a), p<0.05 [Figure 2A]. Compared to Q1, the highest proportion of individuals with self-reported stroke or TIA was noted in the Q5, p=0.08 [Figure 2A]. A log-transformed one unit increase in Lp(a) was significantly associated with higher odds of self-reported stroke or TIA in the unadjusted model and persisted after adjusting for age, sex, LDL, smoking, waist circumference [Table 2]. This significant association had an apparent J-shape, as observed in the RCS [Figure 3]. Q5 was significantly associated with higher risk of self-reported prevalent stroke or TIA compared to Q1 in unadjusted models and this association persisted when controlling for age, sex, LDL, smoking, waist circumference [Table 3]. Additional sensitivity analyses (comparing both log-Lp(a) unit increases and Lp(a) quintiles with self-reported stroke, cerebral injury, and silent brain infarct) in Models 1–4 did not significantly alter our primary result except after adjustments for antithrombotic and statin medications, which attenuated and rendered the association non-significant [Table S1–S2].SOL-INCA MRI target population had a mean age ± SE of 49.9 ± 0.4 years, 56.4% were female, and had a median IQR (Q1, Q3) Lp(a) level of 21.7 (8.1–62.9) nmol/L. Lp(a) distribution in the SOL-INCA MRI at 50th percentile was 21.1 nmol/L, 90th percentile was 137.1 nmol/L, 95th percentile was 183.5 nmol/L, and 99th percentile was 293.7 nmol/L [Figure S2].Of 2,642 individuals identified with cerebral injury in the SOL-INCA-MRI cohort, 7.7% had evidence of cerebral injury (self-reported stroke or TIA history or evidence of stroke on bMRI). Baseline demographics of participants with cerebral injury vs without from the SOL-INCA MRI subset can be found in Table 1. Individuals with cerebral injury were older, of Cuban background, had greater comorbidities (hypertension, dyslipidemia, diabetes mellitus) and reported higher use of statins and anti-hypertensives, all p<0.05. Median IQR (Q1, Q3) Lp(a) levels were significantly higher among individuals with cerebral injury than those without [Table 1]. The highest proportion of cerebral injury was noted in Q5 while the lowest proportion was noted in Q2 with no significant difference when comparing proportions of cerebral injury in Q5 vs Q1 [Figure 2B]. A possible J-shaped relationship between Lp(a) and cerebral injury was noted, showing a non-significant trend between increasing log-Lp(a) and cerebral injury (p=0.08) [Figure 2B, Figure 3]. A statistically non-significant trend was also noted between increases in log-Lp(a) and cerebral injury [Table 2]. Q5 compared to Q1 did not yield statistical significance for cerebral injury [Table 3] which remained consistent in sensitivity analyses [Table S1–S2]. However, our secondary analysis, comparing Q5 to Q2 (the quintile with the lowest observed proportion of cerebral injury), revealed a significant association between Lp(a) and cerebral injury, in fully-adjusted models and sensitivity analyses [Table 4, S3].A scatter plot with best fit linear regression line and 95% CI lines demonstrated a shape with a positive correlation in which every log-transformed one-unit increase in Lp(a) among SOL-INCA MRI participants was associated with a 0.1 increase in log-WMH (β 0.1, p=0.005) [Figure 4].Participants with silent brain infarct had a significantly higher median IQR (Q1, Q3) Lp(a) compared to participants without silent brain infarct (30.39 (9.54, 90.68) vs. 20.73 (8.13, 60.06) nmol/L, p=0.04). However, there was no significant difference across proportions of silent brain infarct per Lp(a) quintiles (p=0.36) [Figure 2C] and no associations were found between log-Lp(a) when assessed continuously or per quintiles during primary and sensitivity analysis (p=0.24) [Table 2–3, Figure 3, Tables S1–S2].

    Discussion

    In this population-based cohort of Hispanic/Latino individuals, increasing Lp(a) levels were associated with higher proportions of self-reported stroke or TIA. For every logarithmic unit increase in Lp(a), the odds of stroke or TIAs increased by 1%, an association that persisted after fully adjusting for age, sex, LDL, smoking, and waist circumference. Likewise, after fully adjusting, Q5 was independently and significantly associated with increased odds of stroke or TIA compared to Q1. There was also a statistically non-significant trend noted between increasing logarithmic units of Lp(a) and cerebral injury. In contrast, Q5 was independently associated with a two-fold increase in the odds of cerebral injury when compared with Q2, the quintile with the lowest observed proportion of cerebral injury. When further assessing bMRI data, we observed that increasing log-Lp(a) levels were positively associated with increasing log-WHM volumes. There was no significant association noted between Lp(a) and silent brain infarcts.The mechanistic reasons for why Lp(a) is associated with ASCVD have been previously enumerated. Lp(a)’s low-density lipoprotein particle component appears to lead to cholesterol deposition, increased thrombosis risk, and pro-inflammatory factors.12 Thus, the negative association of Lp(a) with brain health and positive association with stroke may reflect shared pathophysiological mechanisms. Although there is no validated, universally accepted transfer equation from nmol/L to mg/dL, since these are measuring two different things (molar concentration vs mass), we used ~2.5 nmol/L for every 1 mg/dL conversion equation for the sake of facilitating comparison with prior studies.31,32 In a population-based study of community dwelling Chinese adults Lp(a) levels >100 nmol/L (>40 mg/dl) were associated with the presence of a specific subtype of ischemic stroke (intracranial atherosclerotic stroke) diagnosed on bMRI.7 Similarly, elevated Lp(a) levels were associated with intracranial stenoses, as markers of intracranial stenotic atherogenesis, found on transcranial Doppler of patients who suffered their first TIA or stroke.33Results from meta-analyses and large genetic epidemiological studies have provided evidence of a causative relationship of Lp(a) with the risk of stroke9,12,13, although the risk seems to be weaker than that of Lp(a) with cardiovascular disease and seems to differ among racial and ethnic and sex groups.10,34,35 In a meta-analysis of eleven studies, Kumar et al found an association of increased Lp(a) levels with risk of stroke for Asian populations but not for non-Hispanic (NH) whites populations.10 In a prior case-control study utilizing the multiethnic Northern Manhattan Study (NOMAS), elevated Lp(a) levels were significantly and independently associated with increased stroke risk, through adjudicated events.34 However, in that study, Lp(a)> 75 nmol/L (30 mg/dL) was independently and significantly associated with an increased risk of stroke in the entire cohort with a significant linear dose-response relationship between Lp(a) and stroke, but this effect was lost in the stratified analysis of Hispanic/Latinos (n=317, 52% Hispanic).34 Contrary to NOMAS, we report among Hispanic/Latinos an independent and significant association of Lp(a) with self-reported stroke or TIA in HCHS/SOL. Our association however was not significant when assessing cerebral injury within SOL-INCA-MRI, albeit the relationship approached a non-significant statistical trend, in agreement with recent findings on the UK Biobank36.We can speculate a few reasons for the discrepancy between findings in NOMAS vs. HCHS/SOL: 1) NOMAS study could have been underpowered to assess this association among the Hispanic/Latino population, 2) limited generalizability of NOMAS to the heterogeneous Hispanic/Latino population of the US as compared to HCHS/SOL which includes diverse Hispanic/Latino background groups from four US cities, and 3) differences between NOMAS use of adjudicated events versus self-reported outcomes. Likewise, we can speculate some potential reasons for the discrepancy in findings between HCHS/SOL and SOL-INCA MRI: 1) lower power in the bMRI subset to detect such a relationship; 2) unmeasured confounders native to the SOL-INCA MRI cohort such as cognitive impairment and decline (participants with cognitive impairment were oversampled in SOL-INCA MRI), which has been linked to contrasting associations with Lp(a) among racial groups in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study37; or 3) it may be that the relationship between cerebral injury and Lp(a) may be driven by extreme Lp(a) values above 90–95th percentile risk which deserves further study12,38–40.We noted a possible J-shaped relationship when assessing Lp(a) levels across quintiles of cerebral injury and silent brain infarct. A similar association was found in a recent UK Biobank study assessing risk of incident ischemic stroke by Lp(a) among participants without baseline CVD and not taking a statin36. J-Shaped associations between LDL and cardiovascular events have been limitedly reported and there is a paucity of data regarding its association with Lp(a).41 Our sensitivity analysis comparing Q5 with Q2 noted a significant association between Lp(a) and cerebral injury in SOL-INCA MRI in unadjusted and fully adjusted models except when adjusting for statin use. Perhaps indicating that Lp(a) levels in Q2, a range which is closer to but still below the median Lp(a) levels in Hispanics/Latinos, may be a true baseline for this patient population. It may also be possible that the higher proportions of cerebral injury in Q1 rather than Q2 may be due to unexplained confounders that are not accounted for in this current study.Through a 13.5-year follow-up of 14,221 NH black and NH white participants from the Atherosclerosis Risk in Communities (ARIC) study, participants with Lp(a) ≥ 75 nmol/L (≥300 μg/mL) were observed to have 79% higher age-sex-race-adjusted risk of stroke vs. Lp(a) <25 nmol/L (<100 μg/mL).35 However, in sex-specific analysis within ARIC, the risk of stroke differed per race and sex: elevated Lp(a) levels were found to be associated with higher risk of stroke among NH black women (84% risk) and white women (2-fold risk) but not among NH black and white men.35 Notably, in the REGARDS study, racial differences are also noted in the risk of stroke conferred by elevated Lp(a) levels.42 The overall NH black population in REGARDS with Lp(a) ≥ 142.5 nmol/L (≥57 mg/dL) had 96% increased risk of stroke when compared to NH blacks with Lp(a) < 32 nmol/L (<13 mg/dL), however no associations were found among the overall NH white population after adjusting for age, sex and stroke risk factors.42 We describe Lp(a) levels > 77 nmol/L among Hispanic/Latinos, irrespective of sex, as associated with higher odds of self-reported stroke or TIA, signaling that cut-offs for elevated Lp(a) that confer risk of stroke within the Hispanic/Latino populations might be more similar to that of NH blacks versus NH whites. The reason for these racial discrepancies in conferred risk of Lp(a) might be multifactorial, including factors such as genetic ancestry (since Lp(a) levels are genetically influenced), access to lipid-lowering therapies, and others. Data from the REGARDS study may be used to further speculate that perhaps a Lp(a) >= 90th percentile may be more associated with stroke as compared to a quintile analysis. Nonetheless, our findings ultimately support the idea that elevated levels of Lp(a) confer a risk for stroke that is complex in nature and differs among races and ethnicities.Lp(a) has been previously studied with regard to disease outcomes such as stroke, however, little is known about the role that Lp(a) plays in assessing vascular-related declines in brain structural parameters and the association of Lp(a) with neuroimaging markers of vascular brain health (i.e. WMH – a previously described marker of declining brain function).3,16 A novel aspect of our work is that we utilize MRI parameters to analyze the relationship between Lp(a) and cerebrovascular disease, particularly stroke, silent brain infarction, and to assess neuroimaging markers of brain health (i.e. WMH).WMH has been described as a neuroimaging marker of cerebral microvascular disease that is also a predictor of overt disease such as stroke and dementia.43,44 The casual pathway between WMH and cerebrovascular disease with dyslipidemia is complex with paucity of studies examining the association of Lp(a) with WMH. Some studies have reported elevated LDL levels as an independent predictor of the progression of WMH.45 Nonetheless, it remains unclear whether lipid subtypes such as Lp(a) are in the causative pathway of WMH since research has been limited and some data are contradictory. For example, in a small case control-study of hospital patients with acute stroke or TIA, elevated Lp(a) was associated with WMH and extensive perivascular space.46 In contrast, Pan et al observed that participants with high levels of Lp(a) 265 nmol/L (106 mg/L) compared to 10 nmol/L (40 mg/L) were less likely to have WMH, but these findings were not significant.7 In our study of community-dwelling Hispanic/Latino individuals, we found that log WMH increased by 0.1 with each 1 unit increase in log-Lp(a), suggesting that Lp(a) might have an atherosclerotic effect in the pathophysiology of WMH.Data on the association between Lp(a) and silent brain infarct has been limited. A small case-control study conducted by Fukunaga et al. found that increasing Lp(a) levels were associated with silent brain infarct among hemodialysis dependent Japanese patients.47 In the present study, there was no significant association among Lp(a) levels and silent brain infarct. This may be impacted by factors such as socioeconomic status and lack of access to comprehensive medical care as silent brain infarcts are only observed using objective findings (such as brain imaging) in the absence of symptoms; thus, limiting the opportunity to diagnose it in populations with limited access to comprehensive medical care.From a clinical perspective, data to support treatment of Lp(a) has been limited since the currently available lipid-lowering treatments, such as statins and proprotein convertase subtilisin/kexin type 9 inhibitors, insufficiently reduce Lp(a) levels.48 However, with the advent of clinical trials targeting Lp(a) through gene transcription or Lp(a) assembly, such as lepodisiran49, olpasiran50 and pelacarsen51, clinicians may be able to address Lp(a) more appropriately in the future. Our present data supports that Lp(a) >77 nmol/L among Hispanic/Latino individuals is a special high-risk category susceptible to stroke that would benefit from inclusion in randomized-control trials with lipid-lowering therapies.The current study is not free of limitations. First, the HCHS/SOL overall sample is limited to four urban cities in the US, which limits the generalizability of the findings to the overall Hispanic/Latino population but these cities are where >70% of the US Hispanic/Latino population reside.52 Even though Lp(a) levels were measured before the MRI images were performed, Lp(a) levels are genetically predetermined and stable over time, making a one-time in the lifespan measurement adequate for risk assessment.53 Nonetheless, our study is the largest to date to study the association of Lp(a) with stroke in Hispanics/Latinos. The data presented combined with previously reported literature23,40 supports that Lp(a) should be considered in the context of one’s race and ethnic group and that, when compared to other ethnic groups8,54, universal cut-offs might not be appropriate methods of screening for conferred risk55. This is due to the wide variation in Lp(a) levels, with Hispanics/Latinos presenting with lower mean and median levels as compared with NH whites and NH blacks.23 As such, we suggest that future randomized clinical trials assess disease risk by race-ethnic specific Lp(a) reference values.

    Conclusions

    Lp(a) is independently and significantly associated with prevalent self-reported stroke or TIA and WMH, in a large diverse population of Hispanics/Latinos, suggesting that Lp(a) might be a modifiable risk factor for cerebrovascular disease. Lp(a) >77 nmol/L is associated with greater odds for stroke among the Hispanic/Latino population, representing a cutoff that could be targeted with the advent of new lipid-lowering drugs for Lp(a).
    DLDr. Daniel Labovitz

    Dr. Daniel Labovitz

    Neurology · Vascular Neurology
    Montefiore Medical Center
    daniell
    Bronx, NY, USA