Association of White Matter Hyperintensities With HIV Status and Vascular Risk Factors

Abstract

Objective

To test the hypothesis that brain white matter hyperintensities (WMH) are more common in people living with HIV (PLWH), even in the setting of well-controlled infection, and to identify clinical measures that correlate with these abnormalities.

Methods

Research brain MRI scans, acquired within longitudinal studies evaluating neurocognitive outcomes, were reviewed to determine WMH load using the Fazekas visual rating scale in PLWH with well-controlled infection (antiretroviral therapy for at least 1 year and plasma viral load <200 copies/mL) and in sociodemographically matched controls without HIV (CWOH). The primary outcome measure of this cross-sectional analysis was increased WMH load, determined by total Fazekas score ≥2. Multiple logistic regression analysis was performed to evaluate the effect of HIV serostatus on WMH load and to identify MRI, CSF, and clinical variables that associate with WMH in the PLWH group.

Results

The study included 203 PLWH and 58 CWOH who completed a brain MRI scan between April 2014 and March 2019. The multiple logistic regression analysis, with age and history of tobacco use as covariates, showed that the adjusted odds ratio of the PLWH group for increased WMH load is 3.7 (95% confidence interval 1.8–7.5; p = 0.0004). For the PLWH group, increased WMH load was associated with older age, male sex, tobacco use, hypertension, and hepatitis C virus coinfection, and also with the presence of measurable tumor necrosis factor α in CSF.

Conclusion

Our results suggest that HIV serostatus affects the extent of brain WMH. This effect is mainly associated with aging and modifiable comorbidities.

Brain white matter hyperintensities (WMH) are a common MRI finding,¹ visualized as areas of hyperintense signal in the periventricular or deep white matter on T2-weighted and fluid-attenuated inversion recovery (FLAIR) sequences. They are generally considered nonspecific in etiology but are thought to mostly result from cerebral small vessel disease (CSVD)² and have been associated with increasing age,³ cerebrovascular disorders,⁴ and an increased risk of dementia.⁵

The long-term consequences of HIV infection on the brain in people living with HIV (PLWH) well-controlled on antiretroviral drugs is unknown. These drugs prevent viral replication but do not prevent the formation of viral transcripts or proteins that might be detrimental to neuroglial or vascular systems.⁶

Brain MRI abnormalities have been consistently reported in PLWH,⁷ even in the setting of well-controlled infection, including brain atrophy⁸ and abnormal diffusion tensor imaging measurements,⁹ correlating with cognitive impairment. There is also evidence that WMH are more common in PLWH,^10,11 and that this might correlate with impaired cognition.^12,13 However, the various methods used to estimate WMH in heterogenous groups of PLWH, with some studies lacking a control group, leaves the true relevance and pathogenesis of WMH in HIV disease uncertain.

In this study, we evaluated the frequency, severity, and clinical correlations of WMH in our cohort of patients with well-controlled HIV infection compared to controls without HIV (CWOH). In evaluating WMH, we used the Fazekas visual rating scale,¹⁴ a validated tool for quantification of CSVD,¹⁵ which has been confirmed to be reliable in PLWH.^11,13

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

Participants were recruited through ongoing longitudinal studies at the NIH and the Department of Defense (DoD), which evaluate the natural course of neurocognitive outcomes in PLWH and controls. These studies were approved by the institutional review boards of the National Institute of Allergy and Infectious Diseases at the NIH (NCT01875588) and of the Uniformed Services University of the Health Sciences (NCT00893815) and its participating sites (Walter Reed National Military Medical Center and Naval Medical Center San Diego). Written informed consent was obtained from all participants in the study. At the NIH, PLWH and controls at least 18 years of age were recruited from the same clinics and communities for sociodemographic matching and the DoD participants were recruited from beneficiaries, including those in active duty service, postretirement, or medical separation with continued benefits. People were eligible for participation if they met the following inclusion criteria: no history of CNS infection, no concurrent unstable psychiatric illness, no history of traumatic brain injury, and no contraindication to MRI scanning. Eligibility criteria for the present data analysis included having completed neurocognitive tests and brain MRI and all PLWH were required to be on antiretroviral therapy for at least 1 year and to have plasma HIV RNA <200 copies/mL.

Demographic and Clinical Variables

Information on medication, cardiovascular risk factors, and history of substance use disorders as well as specific information for PLWH (including current and prior combination antiretroviral therapy [cART], time since HIV infection, time since cART initiation, and nadir CD4+ T-cell count) were obtained at interviews on site and through review of medical records. Data collected on all groups included body mass index, blood pressure measurements, and blood tests (lipid profile, fasting blood glucose, hemoglobin A1c levels, estimated glomerular filtration rate, C-reactive protein, and hepatitis C virus [HCV] antibody). A 10-year atherosclerotic cardiovascular disease (ASCVD) risk score was calculated for all participants 40–79 years of age (tools.acc.org/ASCVD-Risk-Estimator-Plus).

An optional lumbar puncture was performed for research for CSF evaluation including HIV viral load, white blood cell count, protein, and glucose. In a subset of participants from whom sufficient CSF was available, 2 additional tests were used: (1) quantification of transactivator of transcription (Tat) by ELISA, performed with the limit of detection for this assay being 200 pg/mL, as previously described^6,16; (2) a cytokine/chemokine magnetic bead panel (Millipore), per manufacturer's instructions, for measurement of 30 different cytokines/chemokines. Results from only 7 potentially relevant cytokines/chemokines^17–19 were used for the analysis (interleukin-6, interleukin-8, interleukin-10, interferon-α-2, interferon-γ, monocyte-chemoattractant protein-1, and tumor necrosis factor–α [TNF-α]), and the level of detection was 1–3 pg/mL for each cytokine/chemokine.

Neuropsychological testing was performed on all participants as previously described.²⁰ An overall average T score (a continuous measurement of average neuropsychological test results across all 7 domains) and the individual domain T scores were used for the analysis as measures of cognition. For evaluation of depression symptoms, participants completed the Beck Depression Inventory II (BDI-II) questionnaire.²¹

All procedures performed as part of these studies were done for research only and there were no clinical indications that prompted the procedures.

MRI Acquisition and WMH Grading

All participants completed a brain MRI scan on a 3T Philips Achieva scanner (Philips Medical Systems, Best, the Netherlands) with an 8-channel head coil, using a standardized protocol that included whole-brain 3D-FLAIR sequences (repetition time/effective echo time/inversion time 8,000/331/2,400 ms; voxel size 1.0 × 1.0 × 1.0). WMH were evaluated on 3D-FLAIR images and manually rated using the Fazekas visual rating scale,^13,14 with a score of 0–3 for periventricular hyperintensities (0 = no lesions, 1 = pencil thin lining, 2 = smooth halo, 3 = irregular with extension into deep white matter) and 0 to 3 for deep WMH (0 = no lesions, 1 = punctate foci, 2 = beginning confluence of foci, 3 = large confluent areas). In an attempt to further characterize the extent of deep WMH, a modified deep WMH score was also implemented (0 = no lesions, 1 = punctate foci [1–5 lesions], 2 = punctate foci [6–9 lesions], 3 = punctate foci [≥10 lesions], 4 = beginning confluence of foci, 5 = large confluent areas).

The first 10 MRIs were reviewed jointly by 2 neurologists and a neuroradiologist (Y.M., B.R.S., and D.A.H.) to standardize the application of the scale. All MRIs were then reviewed and scored by the same rater (Y.M.), blinded to clinical and demographic data. To establish intrarater and interrater reliability, 2 neurologists (Y.M. and B.R.S.) then rated independently a randomly selected subset of MRIs (n = 49). For participants who completed the 3-year visit (n = 30) at the time of data analysis, an evaluation and grading of the MRI was done for longitudinal comparison. Further imaging data included extraction of information from the MRI reports written by a neuroradiologist (D.S.R.) regarding findings of lacunar strokes and foci of leptomeningeal enhancement.²²

Outcome Measures

The primary outcome measure was increased WMH load, defined as having a total Fazekas score ≥2.²³ This binary outcome was chosen due to a non-normal distribution of the score and allowed us to dichotomize the participants into groups with and without higher total Fazekas scores. Secondary outcome measures were also used: total Fazekas score as a continuous variable (square root transformed), periventricular WMH score ≥2, deep WMH score ≥2, and modified deep WMH score ≥3. The latter 3 binary outcomes were chosen to represent scores from ostensibly more abnormal-appearing MRIs.

Statistical Analyses

Differences in demographic and clinical variables between PLWH and CWOH groups were examined using Fisher exact test for categorical variables, 2-sample t test for continuous variables with normal distribution, and Wilcoxon 2-sample test for continuous variables with non-normal distribution.

Fisher exact test was performed to evaluate the differences in distribution of raw Fazekas scores between PLWH and CWOH groups.

A multiple logistic regression model was used to evaluate the effect of group (PLWH vs CWOH) on the primary outcome measure. First, each of the 19 candidate covariates (demographic and clinical variables, see table 1) was analyzed in a simple logistic regression model using a significance level of 0.1 to determine whether it was associated with the primary outcome. This was done separately in each group because some of the variables had incomplete data and some were found to be significantly different between the groups. Then, any candidate covariate that was found to be associated with the primary outcome in either group was included in a stepwise logistic regression analysis using 0.1 as the entry and stay levels for the final multiple logistic regression model.

Table 1.

Demographic, Clinical, and Laboratory Characteristics of the Groups and Their Associations With Increased White Matter Hyperintensities (WMH) Load

To address potentially confounding variables that were significantly different between the 2 groups (α = 0.05), and that were also associated with the primary outcome measure based on the simple logistic regression (α = 0.1), we performed subgroup analysis to investigate whether the effect of HIV serostatus on the primary outcome measure was consistent in the 2 subgroups, for example, with and without history of cocaine use disorder.

To evaluate the effect of group (PLWH vs CWOH) on the secondary outcome measures, both simple and multiple linear (for total Fazekas score) or logistic (for the binary outcomes) regression models were applied, and due to strong correlations between the different outcome measures, the multiple regression models included the same set of covariates that were selected for the multiple regression model of the primary outcome measure.

To explore the association of WMH with the imaging, laboratory, and clinical variables in the PLWH group, the multiple logistic regression model was applied including the same set of covariates as in the above analysis and each of the imaging, laboratory, or clinical variables as a predictor variable.

Cohen-weighted kappa was used to assess interrater and intrarater reliability for the Fazekas scale. The Shapiro-Wilk test was used to examine the normality assumption. The statistical analyses were performed using SAS version 9.4. A significance level (α) of 0.05 was used for all statistical tests.

Data Availability

The data that support the findings of this study are available on request from the corresponding author.

Results

Clinical and Demographic Characteristics

The study included 203 PLWH and 58 CWOH with an available brain MRI scan between April 2014 and March 2019. Clinical and demographic comparisons between the groups are presented in table 1. Notable differences between the groups were the higher percentage of male participants in the PLWH group (75% vs 53%, p = 0.002) and higher rates of hypertension (33% vs 12%, p = 0.002), dyslipidemia (41% vs 22%, p = 0.01), and history of cocaine use disorder (29% vs 16%, p = 0.04).

HIV disease characteristics of the PLWH cohort including cART regimen are presented in table 2. As per the inclusion criteria, all PLWH had a viral load <200 copies/mL and were on cART for at least 1 year.

Table 2.

Disease Characteristics of the Cohort of People Living With HIV (PLWH) and Their Associations With Increased White Matter Hyperintensities (WMH) Load

WMH Differences Between Groups

The proportions of raw scores (total Fazekas score, periventricular WMH score, and deep WMH score) for each group are presented in figure 1, and it was evident that the distributions of the total score and periventricular WMH score were different between the groups (p = 0.003 and p = 0.004, respectively), while differences were smaller but still significant in the distributions of deep WMH and modified deep WMH scores (p = 0.01 and p = 0.03, respectively).

Figure 1. Comparison of Proportions of Raw Scores in Total, Periventricular, and Deep White Matter Hyperintensities (WMH) Scales Among the Groups.

Distribution of scores differed between the groups (p value represents comparison of distribution using Fisher exact test), reflected most significantly in the periventricular (PV) WMH score. Among the people living with HIV (PLWH) group, there is a relative rarity of a score of zero for both periventricular and deep WMH. CWOH = controls without HIV.

Covariate selection using the simple logistic regression found that age, sex, history of tobacco use, history of cocaine use disorder, hypertension, and HCV seropositivity were associated (p ≤ 0.1) with increased WMH load (binary total Fazekas score); the individual effects of different clinical and laboratory variables on this primary outcome measure are presented in table 1. In the simple logistic regression, the prevalence of increased WMH load was higher in the PLWH group with an odds ratio (OR) of 3.0 (95% confidence interval [CI] 1.6–5.6, p = 0.0006). In the stepwise logistic regression, only age and history of tobacco were selected. The multiple logistic regression analysis with age and history of tobacco use as covariates (table 3) showed that the adjusted OR (aOR) of the PLWH group for increased WMH load was 3.7 (95% CI 1.8–7.5, p = 0.0004). Images from representative MRI scans are shown in figure 2.

Table 3.

HIV Serostatus Effect on White Matter Hyperintensities (WMH) Based on Regression Models

Figure 2. Representative Comparison of People Living With HIV (PLWH) to Controls Without HIV (CWOH).

Representative T2–fluid-attenuated inversion recovery images from MRI scans of (A) a 53-year-old man with HIV infection diagnosed 23 years prior to the scan, plasma viral load undetectable, and CD4+ count of 696 cells/mm³ at the time of scan, no known vascular risk factors, with a total Fazekas score of 3 (2 for periventricular and 1 for deep) and a mild degree of atrophy; and (B) a 51-year-old male control without HIV, with no known vascular risk factors, with a total Fazekas score of 1 (1 for periventricular and 0 for deep). CWOH = control without HIV; PLWH = people living with HIV.

Because frequencies of male participants as well as history of cocaine use disorder and diagnosis of hypertension were significantly different between PLWH and CWOH groups, and these also correlated with increased WMH load in at least one of the groups, subgroup analyses were performed and are presented in table 4. Within the male subgroup, PLWH had a higher risk for increased WMH load compared to CWOH (aOR 5.7 [95% CI 2.2–14.8], p = 0.0004). However, in the female subgroup, which had to be stratified by history of tobacco use because of different frequencies (positive history of tobacco use in 78% of female PLWH vs 41% of female CWOH, p = 0.002), there was no significant effect of HIV serostatus on the primary outcome. Within the subgroup of those negative for history of cocaine use disorder, PLWH had a higher risk for increased WMH load compared to CWOH (aOR 4.8 [95% CI 2.1–10.9], p = 0.0002). Similarly, in the subgroup without hypertension, PLWH had a higher risk for increased WMH load compared to CWOH (aOR 3.3 [95% CI 1.5–7.2], p = 0.0003). However, these differences between PLWH and CWOH were not evident in subgroups that were positive for cocaine use disorder or hypertension.

Table 4.

Subgroup Analysis of Risk for Increased White Matter Hyperintensities Load

Multiple regression analysis for secondary outcome measures (table 3) showed that after adjusting for age and history of tobacco use, the PLWH group had a higher mean total Fazekas score (2.1 [95% CI 2.0–2.2] vs 1.7 [95% CI 1.5–1.9], p = 0.002) but the frequency of periventricular WMH score ≥2, deep WMH score ≥2 or modified deep WMH score ≥3 did not differ between the groups.

Determinants and Correlates of WMH in the PLWH Group

There were no significant correlations between the WMH outcome measures and any of the individual cognitive domain T scores or the composite average T score in either group. In addition, no correlation was found with depression symptoms represented by BDI-II scores.

In terms of disease characteristics of the PLWH cohort, after adjusting for age and history of tobacco use, the only factor that was associated with decreased WMH load was being on a cART regimen that included non-nucleoside reverse transcriptase inhibitors (NNRTIs) with an aOR of 0.3 (95% CI 0.1–0.6, p = 0.002) for total Fazekas ≥2 (table 2). There was some increase in the risk for total Fazekas ≥2 among participants being treated with integrase inhibitors (INIs) (aOR 2.6 [95% CI 1–6.5], p = 0.039). Other factors such as current and nadir CD4+ count, duration of HIV infection, or duration of untreated infection did not correlate significantly with any of the outcome measures.

CSF analysis was available in 75/203 (37%) participants in the PLWH group and CSF cytokines/chemokines were measured in 73 of these samples. No significant correlation was found between WMH and CSF measures like white blood cell count, CSF protein, and CSF viral load. However, within the array of CSF cytokines, one difference was found: PLWH participants with measurable TNF-α in the CSF (n = 34) had a higher mean total Fazekas score compared to the rest of the group (2.5 [95% CI 2.2–2.8] vs 2.0 [95% CI 1.8–2.2], p = 0.007). Thirty-three participants (45%) from the PLWH group had measurable Tat in the CSF, but this did not correlate with any of the outcome measures.

Regarding other imaging findings, lacunar strokes were identified in only 7/203 (3%) PLWH cases and 2/58 (3%) CWOH, and all of these cases had increased WMH load. Foci of leptomeningeal enhancement were identified much more commonly in PLWH (33/203%, 16% vs 2/58%, 3%, p = 0.01), and this finding was associated with a higher mean total Fazekas score (2.5 [95% CI 2.2–2.8] vs 2.1 [95% CI 1.9–2.2], p = 0.02).

None of the clinical, CSF, or imaging variables correlated with outcome measures representing a more severe WMH load (periventricular WMH score ≥2, deep WMH score ≥2, modified deep WMH score ≥3).

Longitudinal Evaluation of WMH

Thirty available 3-year-visit scans were scored and compared to baseline scoring (23 PLWH, 7 CWOH). No significant difference was found in periventricular or deep WMH score in any CWOH cases. For the PLWH group, in 4 cases (17%) the score changed (2 cases: deep WMH score increased by 1; 1 case: periventricular score increased by 1; 1 case: periventricular score decreased by 1). A representative case with increase in periventricular component of the score is presented in figure 3.

Figure 3. Representative Longitudinal Change in White Matter Hyperintensities (WMH).

Representative T2–fluid-attenuated inversion recovery images from the baseline scan (A) and the 3-year visit scan (B) of a 53-year-old man with HIV infection diagnosed 25 years prior to the first scan, plasma viral load less than the limit of detection, and CD4+ count of 585 cells/mm,³ with history of hypertension and tobacco use, showing an evident progression of the periventricular hyperintense signal (arrow, score increased from 1 to 2). There is also an evident progression of atrophy and of the deep WMH, although this did not result in a change of this component of the score.

Reliability Analysis

Reliability analysis using Cohen-weighted kappa for the different components of the scale indicated that both intrarater and interrater agreement were very high: periventricular WMH (0.87, 0.73), deep WMH (0.99, 0.81), modified deep WMH (0.97, 0.93), and total score (0.92, 0.83) for intrarater and interrater kappa, respectively.

Discussion

Our data show that the extent of WMH is higher in PLWH with well-controlled infection compared to sociodemographically similar controls without HIV. After controlling for age and history of tobacco use, PLWH had an almost 4-fold higher odds of increased WMH load compared to CWOH, which seems to be mainly attributable to the periventricular component of the score being so differently distributed between groups. Thus, despite differences in sex and some cardiovascular risk factors between the groups, this analysis supports the hypothesis that there is a contribution of HIV serostatus to the extent of WMH.

The subgroup analysis suggests that the effect of HIV serostatus is more notable among subgroups without confounding factors such as hypertension, history of cocaine use disorder, or male sex. This specific effect in male patients is consistent with the fact that previous studies showing higher burden of WMH in PLWH relied solely¹¹ or mainly²⁴ on men. Interestingly, studies in the general population have shown that somewhat counterintuitively, male patients typically have lower WMH burden compared to female patients.^25,26 Our subgroup analysis shows that in female patients and in groups with hypertension or a history of cocaine use disorder, there were no significant differences by HIV serostatus. While this could be attributed partially to smaller sample sizes in the subgroups, it possibly suggests that the effect of HIV was overwhelmed by the effects of these other predisposing factors on WMH. Because hypertension and cocaine use are modifiable risk factors for WMH, this analysis highlights the importance of identifying and treating these factors for patients with well-controlled HIV disease.

Within the PLWH group, older age, male sex, history of tobacco use, hypertension, and HCV coinfection were all associated with a higher rate of WMH. Tobacco use and HCV coinfection are additional modifiable risk factors that had not been previously identified to be clearly associated with WMH. A previous study using diffusion tensor imaging showed that tobacco use in PLWH was associated with abnormalities in the white matter tracts,²⁷ and similarly, HCV coinfection has been associated with increased white matter abnormalities.²⁸ However, these studies did not define the pattern of abnormalities, and another study using diffusion tensor imaging did not find an additive effect of HCV in PLWH.²⁹

Our focus on WMH in PLWH with well-controlled infection and the use of a simple and reliable rating scale provides insights into the different factors associated with white matter changes in the brain, while supporting the reliability of this tool in clinical practice to estimate the severity of findings in an individual case. One of our major findings is that although we were able to show differences between PLWH and controls in terms of WMH, we did not show that these WMH are associated with any of the clinical outcomes investigated. This point is especially relevant for neurologists evaluating patients with cognitive complaints and an MRI brain that shows WMH. Our results suggest that what we see in the white matter on MRI might not have a clear association with cognitive outcomes, at least cross-sectionally, and so neurologists need to continue diligently assessing other possibilities for cognitive symptoms rather than attributing them to these common findings on MRI.

Our preliminary longitudinal analysis also identifies the potential value of following changes in WMH of PLWH. While some differences might be attributed to technical differences between scans or rating scale reliability, others may represent true progression. Further longitudinal analysis could help identify any patient with progressive white matter changes and evaluate for possible cognitive changes that were not evident in this cross-sectional analysis, despite literature supporting this correlation in the general population⁵ and specifically in PLWH.^12,13

Most previous studies on WMH in HIV were missing data on viral load or antiretroviral treatment^12,30 or lacked a control group.^13,24 Two major studies did use a population with well-controlled infection and similarly estimated a higher extent of WMH compared to a control group, but they were limited to some degree by a focus on male patients, use of different scanners,¹¹ and inclusion of other measurements like microbleeds or silent infarcts in the analysis that do not necessarily share the same pathophysiology.¹⁰ Our study had a breadth of data available, with cognitive, neuroimaging, CSF, and psychosocial history available for each participant, thus we were able to look more broadly for possible correlates of WMH.

The pathophysiology of WMH in the brains of PLWH remains uncertain. In the general population, WMH are correlated with age and cardiovascular risk factors.^3,31 The higher extent of WMH in PLWH could be related to persistence of viral reservoirs in the CNS, having a direct effect on the brain parenchyma, or an indirect effect by means of accelerated aging and CSVD.³² However, this could also be related to residual damage from the acute infection,³³ and this might be supported by the lack of association with many of the specific disease measures, although we also did not find a correlation with the time of untreated infection. Another possibility is some degree of contribution from the antiretroviral compounds used to treat HIV.²⁴

The most commonly proposed pathophysiology of WMH in PLWH is that of microvascular alterations associated with concomitant cardiovascular risk factors, which has been suggested to be a result of accelerated age-related changes.³⁴ This is supported by the demonstration of a general prematurity of age-related comorbidities among PLWH,³⁵ exemplified by a joint effect of aging and HIV on microstructure of white matter bundles,³⁶ an increased risk of stroke in PLWH even with a well-controlled infection,³⁷ and a specific interaction of HIV with factors like age and diabetes mellitus in regards to WMH.^30,38 Our results provide support for this mechanism, as age and other cardiovascular risk factors were highly correlated with WMH in the PLWH group, requiring a process of controlling and stratifying for other risk factors to illustrate differences between PLWH and controls. In addition, despite multiple methods of analysis, few correlations were found with measures related to HIV, and these were not consistent across higher levels of WMH load. Thus, the clinical significance of WMH specifically in the setting of HIV infection remains unclear.

Accumulation of white matter injury is likely multifactorial,³⁹ including persistence of CNS HIV reservoir,⁴⁰ direct neurotoxicity of viral factors,⁴¹ and secondary processes of neuroinflammation⁴² and neurodegeneration.⁴³ In our exploratory analysis, we found a correlation between WMH and foci of leptomeningeal enhancement, as well as measurable TNF-α in the CSF. While these findings should be interpreted cautiously, they could support existence of viral-induced inflammation or alternatively be a marker of accelerated aging, as TNF-α levels in the CSF have been shown to increase linearly with age,⁴⁴ and the rest of our analysis did not show any correlation with measures related to the duration of infection or level of viral infection or inflammation based on CSF viral load, viral proteins such as Tat, or the other cytokines measured. We also did not find evidence that WMH correlate with the degree of previous immunodeficiency, although this has been suggested in some studies.^10,45

As for the contribution of cART to WMH, there is accumulating evidence for white matter injury,⁴⁶ and specifically cerebrovascular toxicity⁴⁷ caused by cART. Some studies suggest higher WMH load in treated patients²⁴ and a specific contribution of protease inhibitors (PIs) to cerebrovascular changes.⁴⁸ Our findings suggest that patients on an NNRTI-containing regimen had lower WMH load, which might be attributed to better CNS viral control or to avoidance of other classes of cART, such as PIs or INIs. This is supported by the finding that INIs were associated with higher WMH load, although we did not find a correlation with more severe degrees of WMH and this effect of INIs was independent of other classes of antiretroviral drugs. However, these findings should be addressed with caution, as our study design and data collection on cART regimen do not allow for a full evaluation of the history of exposure, and also we did not compare treated patients to untreated patients, and there is no doubt about the systemic and neurologic long-term accumulative benefits of cART. Hence, further studies are needed to fully understand the effects of these drugs on the brain and white matter changes.

There is evidence that the distinction between deep and periventricular WMH reflects different histopathologic and etiologic features,^4,49 and that periventricular WMH, especially caps and smooth halos, have more features of ependymitis and subependymal gliosis compared to the ischemic nature of deep WMH and irregular periventricular WMH.⁵⁰ Our data suggest that the main difference between PLWH and CWOH was due to the periventricular component of the Fazekas score, which could imply that this is indeed a distinct virus-related process, although the score is inherently biased towards periventricular changes.

Strengths of our study are the research-based use of a consistent MRI protocol, the large number of participants (relative to prior studies) with well-controlled infection, and a sociodemographically similar control group. Limitations of our study include the lack of matching for sex and some comorbid conditions, which had to be addressed by adjustments and stratifications, as well as the use of a single manual rating scale and the extraction of some imaging data (i.e., lacunar strokes and foci of leptomeningeal enhancement) from prewritten reports. Another limitation is the exploratory nature of the analysis used in search of clinical and laboratory correlations, which could not be adjusted for multiple comparisons and requires caution in the interpretation of this part of the results.

Our results suggest that HIV serostatus affects the extent of brain WMH even in the setting of well-controlled infection. This contribution is possibly mediated by accelerated aging and modifiable cardiovascular comorbidities, as few significant correlations were found with specific HIV disease measures. Longitudinal changes and the contribution of inflammation and antiretroviral treatment regimen to white matter injury in the setting of HIV infection should be further studied in long-term cohorts to determine the clinical significance of these findings.

Glossary

aOR

adjusted odds ratio

ASCVD

atherosclerotic cardiovascular disease

BDI-II

Beck Depression Inventory II

cART

combination antiretroviral therapy

CI

confidence interval

CSVD

cerebral small vessel disease

CWOH

controls without HIV

DoD

Department of Defense

FLAIR

fluid-attenuated inversion recovery

HCV

hepatitis C virus

INI

integrase inhibitor

NNRTI

non-nucleoside reverse transcriptase inhibitor

OR

odds ratio

PI

protease inhibitor

PLWH

people living with HIV

Tat

transactivator of transcription

TNF

tumor necrosis factor

WMH

white matter hyperintensities

Appendix 1. Authors

Appendix 2. Co-investigators

Study Funding

This work was supported by the Intramural Research Programs of the National Institute of Neurologic Diseases and Stroke, the NIH Clinical Center, the National Institute of Mental Health, the National Institute for Allergy and Infectious Diseases, and the National Institute on Alcohol Abuse and Alcoholism, NIH. This study was also supported by the Infectious Disease Clinical Research Program (IDCRP), a Department of Defense (DoD) program executed by the Uniformed Services University of the Health Sciences (USUHS) through a cooperative agreement with The Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc. This project was also supported in part with federal funds from the National Institute of Allergy and Infectious Diseases, NIH, under Inter-Agency Agreement Y1-AI-5072 and from the Defense Health Program, US DoD, under award HU0001190002.

Disclosure

The authors report no disclosures relevant to the manuscript. The views expressed are those of the authors and do not reflect the official views of the Uniformed Services University of the Health Sciences, Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc., the National Institutes of Health, the Department of Health and Human Services, the Department of Defense, or the Departments of the Army, Navy or Air Force. Mention of trade names, commercial products, or organizations does not imply endorsement by the US government. Go to Neurology.org/N for full disclosures.