The role of atherosclerosis in HIV‐associated vasculopathy in young South African stroke patients

Abstract

Background

HIV‐associated vasculopathy is known to cause stroke in people living with HIV (PLWH). The role of atherosclerosis is unclear. We assessed the aetiology of vasculopathy in PLWH and tested the utility of markers of subclinical atherosclerosis to distinguish atherosclerotic (AV) from non‐atherosclerotic vasculopathy (NAV).

Methods

This cross‐sectional study recruited PLWH with stroke at a hospital in Johannesburg, South Africa, from 2014 to 2017. Patients with meningitis were excluded. Cerebrospinal fluid (CSF) was tested for multi‐viral polymerase chain reaction, including varicella zoster virus (VZV). Once an aetiological category was assigned, carotid intima‐media thickness (cIMT) and aortic pulse wave velocity (PWV) were compared in AV and NAV, and to predetermined thresholds for subclinical atherosclerosis (cIMT≥0.70 mm, PWV≥10.00 m/s).

Results

Some 28/85 PLWH (32.9%) vs. 9/109 (8.3%, p < 0.0001) people‐without‐HIV had vasculopathy on computed tomography angiography. Only four PLWH had AV. Compared with NAV (n = 11), those with AV were older (50.0 ± 4.1 vs. 39.2 ± 9.2 years, p = 0.04) and had more cardiovascular risk factors (median 2.0 [IQR 1.5–2.5] vs. 0.0 [IQR 0.0–1.0], p = 0.02). cIMT in AV was higher than in NAV (1.01 ± 0.07 mm [n = 4] vs. 0.63 ± 0.04 mm [n = 9], p < 0.001). All with AV had cIMT and PWV above the predetermined thresholds, while all except one with NAV were below. We found evidence of VZV in eight PLWH and HIV‐associated vasculitis in six.

Conclusions

Vasculopathy in PLWH in our region appears to be predominantly non‐atherosclerotic. cIMT and PWV were useful adjuncts in distinguishing AV from NAV. Despite excluding meningitis, VZV was implicated in a large proportion, emphasizing the likely underdiagnosis of this treatable infection. We thus recommend CSF VZV testing in all PLWH with stroke.

INTRODUCTION

HIV‐associated vasculopathy has been identified as a cause of stroke in up to a third of cases in people living with HIV (PLWH) [1, 2]. It is a broad term, encompassing abnormalities of blood vessels which may include an atherosclerotic vasculopathy (AV), non‐atherosclerotic vasculopathy (NAV), HIV‐associated vasculitis and small vessel disease (SVD) [3].

AV and NAV both affect large‐ to medium‐sized vessels, resulting in stenotic lesions. However in the case of NAV, there is no histopathological evidence of atheroma [4, 5, 6]. The clinical differentiation of AV from NAV is based predominantly on the patient's age (>45 years), the presence of traditional cardiovascular risk factors (TRFs) or the presumed presence of atherosclerosis elsewhere in the body [3]. There are, however, no definitive radiological or clinical features to distinguish the two entities, nor is there any biomarker which has been tested in this setting.

Carotid intima media thickness (cIMT) is a well‐recognized and validated surrogate marker for atherosclerosis in the cardiovascular system [7]. The presence of an increased cIMT is associated with an increased risk of cardiovascular disease, including stroke [7, 8]. Similarly, aortic pulse wave velocity (PWV) non‐invasively measures large artery stiffness and is a marker of subclinical arteriosclerosis and atherosclerosis [9]. Hence, cIMT and PWV measurements may assist in objectively differentiating AV and NAV in PLWH.

There is insufficient evidence to suggest that atherosclerosis is a major contributor to HIV‐related stroke in sub‐Saharan Africa (SSA). On the contrary, opportunistic infections have been shown to contribute to 23%–55% of strokes in PLWH, often on the basis of a vasculitis [1]. Varicella zoster virus (VZV) reactivation in particular has a known association with stroke in PLWH [10, 11, 12, 13]. From our large cohort of stroke patients from the epicentre of the world's HIV pandemic, we investigated the pathogenesis of HIV‐associated vasculopathy and determined the roles of atherosclerosis and VZV in this disease process. We also assessed the utility of cIMT and PWV to distinguish AV from NAV.

METHODS

This cross‐sectional study took place at the Charlotte Maxeke Hospital in Johannesburg South Africa, from August 2014 to November 2017.

We recruited consecutive patients (age ≥18 years) with ischaemic stroke (using the World Health Organization case definition) [14]. Patients diagnosed with meningitis on cerebrospinal fluid (CSF) examination were excluded (see Supporting Information Methods).

All patients were assessed by the same specialist neurologist (ES). Stroke aetiology classification was guided by a consensus paper on HIV‐related stroke classification [3]. TRFs (i.e., hypertension [HT], diabetes mellitus [DM], dyslipidaemia and obesity) were defined as detailed in the Supporting Information Methods, together with the full protocol of laboratory investigations. HIV testing (enzyme‐linked immunosorbent assay [ELISA]) was performed in all consenting patients. A suppressed viral load (VL) was defined as <50 copies/mL. A lumbar puncture was performed in all PLWH if not contraindicated. Cerebrospinal fluid (CSF) was examined for routine biochemistry, microscopy and a multi‐viral polymerase chain reaction (PCR) panel, including VZV (see Supporting Information Methods).

Computed tomography (CT) scanning of the brain was performed in all patients and was examined by a specialist radiologist and neurologist (ES). All patients had carotid ultrasound performed by a specialist radiologist, assessing for occlusive atherosclerotic plaque. CT angiography (aortic arch, neck and brain) was performed in all patients with no clearly identified cause for the stroke. The cardiac evaluation was performed by a specialist cardiologist (detailed in the Supporting Information Methods).

Additional carotid Doppler studies to assess cIMT were performed using high‐resolution B‐mode ultrasound, and large artery stiffness was indexed from aortic PWV using applanation tonometry. The cIMT and PWV examinations were performed by a single operator (blinded to the HIV status of the patients) who had undergone dedicated training in the relevant techniques, which are described elsewhere and in the Supporting Information Methods [15, 16]. The operator was not part of the patient's clinical management team and was blinded to the aetiological category which had been already assigned to the patient. No changes were made to the classification after cIMT and PWV measurements. A carotid IMT≥0.7 mm and a PWV≥10.00 m/s were considered suggestive of subclinical atherosclerosis as per previously published data [7, 9, 17, 18].

Definitions of vasculopathies

Only patients with vascular abnormalities on angiography were classified as vasculopathy. These included the presence of monofocal or multifocal occlusions, alternating constrictions and dilatations (‘beading’) or aneurysmal/dolichoectatic changes in the intracranial or extracranial vessels. In order to diagnose AV we required the visualization of atherosclerotic plaque causing a >50% occlusion of the internal carotid (ICA) or common carotid artery (CCA) ipsilateral to the stroke on the ultrasound studies performed by the radiologist. As our objective was to test the utility of cIMT and PWV to distinguish AV and NAV, we did not use these measurements in the diagnostic workup of the patient but only compared them once a diagnostic category was allocated. HIV‐associated vasculitis was diagnosed in the setting of angiographic features of vasculitis without an identified causative opportunistic infection [3]. SVD was defined as the presence of a traditional lacunar syndrome with an infarct <20 mm in size [3].

Statistical analysis

Data were analysed using Statistica Ver14.0.0.15, TIBCO Software Inc. Continuous variables were evaluated for normality using Shapiro–Wilk and Kolmogorov–Smirnov tests. Chi square, Fisher's exact test, Student's t‐test and Mann–Whitney statistical tests were performed for parametric and non‐parametric variables where appropriate. Continuous data are reported as mean ± SD unless otherwise specified. Missing data were not imputed. P‐values <0.05 were considered significant.

Informed consent was obtained as per the Declaration of Helsinki [19]. Ethics approval was granted by the University of the Witwatersrand Human Research Ethics Committee (Certificates M140429/M190688/M240682).

RESULTS

A total of 85 PLWH with ischaemic stroke were recruited and compared with 109 age‐ and sex‐matched people‐without‐HIV with ischaemic stroke recruited during the same time period (Figure S1). PLWH had less HT, DM and median number of TRFs than people‐without‐HIV, with a higher prevalence of vasculopathy (Table 1). Some 82/85 (96.5%) PLWH underwent angiography; the remaining three patients presented with unequivocal cardioembolic strokes.

TABLE 1.

Comparing traditional risk factors and stroke aetiology in people living with HIV and people‐without‐HIV.

Parameter	PLWH (n = 85)	People‐without‐HIV (n = 109)	P‐value
Age (years)	43.4 ± 12.5	45.0 ± 10.5	0.33
Male (%)	49.4	47.7	0.81
HT (%)	43.5	60.6	0.018
DM (%)	5.9	22.9	0.001
Dyslipidaemia (%)	24.7	30.3	0.39
Smoking (%)	25.9	18.4	0.21
Obesity (%)	25.8 (16/62)	34.6 (28/81)	0.26
Median number of TRFs (IQR)	1.0 (0–2)	2.0 (1–2)	0.02
Stroke aetiology
Cardioembolic (%)	20.0	30.3
SVD (%)	16.5	18.4
Vasculopathy (%)	32.9	8.3	<0.0001
Other (%)	4.7 a	6.4 b
Undetermined (%)	25.9	36.7

Abbreviations: DM, diabetes mellitus; HT, hypertension; IQR, interquartile range; SVD, small vessel disease; TRFs, traditional cardiovascular risk factors; VZV, varicella zoster virus.

^a Including recreational drug usage (n = 2), VZV with normal angiography (n = 2).

^b Including recreational drug usage (n = 3), hypercoagulable/hyperviscosity syndrome (n = 4).

Vasculopathy in PLWH

Twenty‐eight PLWH (32.9%) had vascular abnormalities on angiography. Two distinct types of vasculopathy were identified: a large vessel monofocal occlusion (n = 16) and a multifocal vasculitis (n = 12) (Figure 1).

FIGURE 1.

The aetiology of vasculopathy in people living with HIV (PLWH). *All four patients showed extracranial stenosis at the carotid bulb/proximal internal carotid artery (ICA). ^†The location was intracranial in five (all middle cerebral artery, MCA), and extracranial in six (common carotid artery (CCA) in three, ICA in two and vertebral artery in one). ^‡Two patients with non‐lacunar small vessel strokes also had evidence of recent VZV infection (not classified as ‘vasculopathy’ as angiography was normal). ^§The involved intracranial vessels included the bilateral ICA, MCA and/or anterior cerebral arteries (ACA). ^¶The intracranial blood vessels involved included the ICA, MCA and ACA, while the two patients with extracranial multifocal vasculitis had involvement of the CCA and ICA, or the ICA alone (which extended partially intracranially). ^#One patient with multifocal vasculitis had multiple plausible aetiologies for stroke (recreational drug usage and neurosyphilis). This patient (43 years old, CD4 470 cells/ul, VL12000 copies/mL, antiretroviral therapy naïve) had bilateral irregularly beaded MCAs, and a saccular intracranial ICA aneurysm. This was the only patient in the cohort who had clear aneurysmal changes. **Excluding the two patients described in ‡ above. AV, atherosclerotic vasculopathy; NAV, non‐atherosclerotic vasculopathy; VZV, varicella zoster virus.

Monofocal occlusion

Some 11/16 PLWH (68.7%) had no evidence of atherosclerosis (NAV). Only four PLWH had AV, while one PLWH with monofocal occlusion of their left MCA had a recent zoster rash and was thus categorized separately as monofocal occlusion secondary to VZV‐vasculitis (Figure 1). PLWH with AV were older than those with NAV, had more TRFs and a higher mean cIMT. PWV was only possible in one patient but was also markedly elevated (Table 2).

TABLE 2.

Comparison of atherosclerotic and non‐atherosclerotic vasculopathy in people living with HIV.

Parameter	Atherosclerotic vasculopathy (AV) (n = 4)	Non‐atherosclerotic vasculopathy (NAV) (n = 11)	P‐value
Age (years)	50.0 ± 4.1	39.2 ± 9.2	0.04
Male (%)	100.0	36.4	0.03
HT (%)	75.0	18.2	0.04
DM (%)	0.0	0.0	1.00
Dyslipidaemia (%)	25.0	9.1	0.42
Smoking (%)	100.0	9.1	0.001
Obesity (%)	0.0 (n = 2)	12.5 (n = 7)	0.60
Median number of TRFs (IQR)	2 (1.5–2.5)	0 (0–1)	0.02
CD4 (cells/μL)	474.5 ± 255.3	351.2 ± 312.5	0.49
Median VL (copies/mL) (IQR)	2200 (0–192 800) (n = 3)	2156 (0–25 476)	0.87
ART‐exposed (%)	50.0	45.5	0.88
Median ART duration (months) (IQR)	102 (60–144) (n = 2)	10.0 (4–60)	0.10
VL suppressed (%)	50.0	60.0	0.81
cIMT (mm)	1.01 ± 0.07 (n = 4)	0.63 ± 0.04 (n = 9)	<0.001
PWV (m/s)	12.60 (n = 1)	6.80 ± 2.14 (n = 8)	0.04

Abbreviations: ART, antiretroviral therapy; AV, atherosclerotic vasculopathy; cIMT, carotid intima‐media thickness; DM, diabetes mellitus; HT, hypertension; IQR, interquartile range; NAV, non‐atherosclerotic vasculopathy; PWV, pulse wave velocity; TRFs, traditional cardiovascular risk factors; VL, viral load.

We then tested the utility of cIMT and PWV in differentiating AV and NAV. All patients with NAV had a cIMT below the threshold of 0.70 mm, while all of those with AV were above this threshold (Figure 2a). One outlier with NAV had a PWV of 11.1 m/s (a 34‐year‐old with no TRFs and cIMT of 0.67 mm) (Figure 2b). However the overall mean was 6.45 ± 1.77 m/s, with no other NAV patient having a PWV over 8.2 m/s.

FIGURE 2.

Box plots of (A) carotid intima media thickness (cIMT) and (B) aortic pulse wave velocity (PWV) to distinguish atherosclerotic (AV) from non‐atherosclerotic vasculopathy (NAV). Comparison of cIMT and PWV to the predefined cut‐off values (horizontal lines, cIMT = 0.70 mm, PWV = 10.00 m/s) revealed that all patients with NAV had a cIMT below the threshold, while all of those with AV were above this threshold. One outlier with NAV had a PWV of 11.1 m/s. *p < 0.001, **p = 0.04, (Student's t‐test). PWV was only possible in one patient with AV (12.60 m/s).

Multifocal vasculitis

A multifocal large vessel vasculitis was noted in 12 patients (Figure 1). VZV was implicated in five of these (three were CSF VZV PCR‐positive, while two had recent cutaneous zoster infection, but negative CSF VZV PCR). There was also evidence of recent cutaneous zoster infection in one patient with monofocal MCA occlusion, and a further two patients with non‐lacunar small vessel strokes. As these two patients had normal angiography, we elected not to classify them as ‘vasculopathy’, even though a normal CT angiogram does not exclude a small vessel vasculitis. VZV was thus potentially implicated in 21.4% (6/28) of PLWH with confirmed vasculopathy on angiography, or 26.7% (8/30) of those with presumed vasculopathy (Figure 1). PLWH with VZV‐related stroke had a lower mean CD4 count compared with PLWH with non‐VZV related stroke and were younger, with a lower body mass index (BMI) and fewer TRFs (Table 3). While 5/8 VZV patients were antiretroviral therapy (ART)‐exposed, only one had VL‐suppression. The cIMT and PWV were lower in VZV‐related stroke patients (Table 3); however, multivariate regressions revealed these differences to be due to the difference in age, and not VZV status (Table S1).

TABLE 3.

Comparisons of people living with HIV with varicella zoster virus (VZV)‐related stroke, non‐VZV related stroke and HIV‐associated vasculitis.

Parameter	VZV‐related (n = 8)	Non‐VZV related (n = 71) a	P‐value (vs. VZV)	HIV‐associated vasculitis (n = 6)	P‐value (vs. VZV)
Age (years)	28.5 ± 4.7	45.7 ± 12.1	<0.001	35.0 ± 4.9	0.03
Male (%)	37.5	52.1	0.43	33.3	0.65
Median number of TRFs (IQR)	0.0 (0.0–1.0)	1.0 (0.0–2.0)	0.01	1.0 (1.0–1.0)	0.12
BMI (kg/m2)	19.7 ± 3.4 (n = 6)	27.8 ± 6.1 (n = 51)	0.003	28.5 ± 5.3 (n = 5)	0.01
CD4 (cells/μL)	92.9 ± 77.1	357.0 ± 201.5	<0.001	287.3 ± 217.6	0.04
Median VL (copies/mL) (IQR)	101 079 (1112–292 548)	1600 (0–44 331)	0.06	0 (0–19 200)	0.19
ART‐exposed (%)	62.5	45.1	0.35	50.0	0.53
Median ART duration (months) (IQR)	8 (4–10) (n = 5)	36 (5–60) (n = 26)	0.24	60 (13–108)	0.18
VL suppressed (%)	20.0	74.2	0.02	100.0	0.07
cIMT (mm)	0.56 ± 0.05 (n = 7)	0.71 ± 0.17 (n = 54)	0.02	0.63 ± 0.08 (n = 3)	0.12
PWV (m/s)	5.30 ± 1.10 (n = 5)	8.11 ± 2.77 (n = 42)	0.03	7.25 ± 0.30 (n = 4)	0.01

Abbreviations: ART, antiretroviral therapy; BMI, body mass index; cIMT, carotid intima‐media thickness; IQR, interquartile range; PWV, pulse wave velocity; TRFs, traditional cardiovascular risk factors; VL, viral load.

^a This excludes the six patients with HIV‐associated vasculitis analysed in the following columns.

In six PLWH with multifocal vasculitis the aetiology was unknown and was thus categorized as ‘HIV‐associated vasculitis’ after the exclusion of autoimmune, inflammatory and other infective vasculitides (detailed in the Supporting Information Methods). The remaining patient with multifocal vasculitis had multiple plausible aetiologies (detailed in Figure 1). There was no angiographic evidence of a small vessel vasculitis in any patient. When comparing VZV‐vasculitis to HIV‐associated vasculitis, patients with VZV had a lower age, lower CD4 count and lower mean BMI, but no difference in VL or ART usage (Table 3). The PWV was higher in HIV‐associated vasculitis, but below the predefined cut‐off value for atherosclerosis (Table 3).

Comparison of patients with intracranial and extracranial vasculopathy revealed no differences in demographics, TRFs, CD4, VL, ART exposure, cIMT or PWV (Table S2). We then compared NAV (n = 11) to multifocal vasculitis (n = 12) under the assumption that there is no proven association between atherosclerosis and multifocal vasculitis. There were no differences in the above variables between the two groups (Table S2).

Small vessel disease

SVD was identified in 14 PLWH. Compared with non‐SVD patients (n = 71), they had a higher mean age (53.6 ± 8.4 vs. 41.4 ± 12.2 years, p < 0.001), greater TRF prevalence and higher mean cIMT (0.78 ± 0.12 vs. 0.66 ± 0.17 mm, p = 0.03) (Table S3). Multivariate regressions revealed the difference in cIMT to be due to age, and not TRF burden (b = 0.00817, SE = 0.0014, p < 0.001, n = 64). PWV was similar in the two groups. SVD patients had a higher mean CD4, despite similar ART usage. Comparison of SVD in PLWH and people‐without‐HIV revealed no difference in age, sex, TRFs, cIMT or PWV (Table S3).

Effect of ART on cIMT and PWV

Some 40/85 PLWH (47.1%) were on ART at the time of their stroke. 37/40 were on a regimen containing emtricitabine, tenofovir and efavirenz. The three patients on protease‐inhibitor‐containing regimens had no evidence of an atherosclerosis‐mediated aetiology (two with multifocal vasculitis, and one with cardioembolism from a metallic heart valve, all with normal cIMT and PWV measurements). There was no difference in demographics, TRFs, mean CD4, cIMT or PWV in the ART‐exposed compared with ART‐naïve patients (Table S4).

Vasculopathy in people‐without‐HIV

Some 9/109 people‐without‐HIV (8.3%) had vasculopathy, compared with 32.9% of PLWH (p < 0.001, Table 1). Three had extracranial AV, while six had intracranial multifocal vasculitis of large/medium sized vessels. The aetiology was vasculitis secondary to systemic autoimmune disease (n = 3), Takayasu's arteritis (n = 2) and postpartum vasculitis (n = 1). There were no people‐without‐HIV with monofocal NAV.

DISCUSSION

We investigated the aetiology of vasculopathy in PLWH and explored the role of atherosclerosis in one of the largest PLWH stroke studies from SSA [1]. We described the presence of vasculopathy in almost one‐third of strokes in PLWH, compared with 8.3% of age and sex‐matched people‐without‐HIV. AV was uncommon in PLWH (4.7% of all strokes). We demonstrated that measurements of cIMT and PWV correlated well with the clinico‐radiological distinction of AV from NAV, and no evidence of subclinical atherosclerosis in NAV was shown. Despite the exclusion of meningitis from our cohort, we still implicated VZV in approximately one‐quarter of PLWH with vasculopathy. This points to the likely underdiagnosis of this condition in patients who appeared to have had infections reasonably excluded by means of normal basic CSF parameters.

Two distinct angiographic phenotypes of vasculopathy were noted: a monofocal occlusion of a single blood vessel or a multifocal vasculitis.

Monofocal vessel occlusion is mostly non‐atherosclerotic

AV had a very low prevalence in PLWH, and it appears to be driven by TRFs, and not by any HIV‐related factors including ART. This low AV prevalence may be partially ascribed to the low mean age of our cohort. However, historical data from SSA have also demonstrated a low prevalence of large vessel atherosclerotic stroke, both in PLWH and people‐without‐HIV [1, 20]. A study from Malawi ascribed 23/64 (38%) of strokes in PLWH to HIV‐associated vasculopathy. Of these, seven were judged to be atherosclerotic (10.9% of strokes in PLWH) [2]. The criteria used to distinguish AV from NAV were clearly defined; however, the distinction was highly reliant on patient age and the presence of only one TRF. For example a patient with age >45 years (or age ≤45 years with exposure to a single TRF) with a previous transient ischaemic attack was considered as ‘clinical history suggestive of atherosclerosis’, and diagnosed with ‘probable AV’. While this is plausible in a high TRF‐prevalence population, it may be presumptive in the setting of HIV‐related stroke in SSA, where the role of TRFs and atherosclerosis is yet to be established. A history of a previous cardiovascular event does not necessarily equate to the presence of atherosclerosis. cIMT or PWV was not reported in this or in other previous SSA studies.

Our findings of non‐atherosclerotic monofocal vessel occlusions (NAV) in 11 patients were of particular interest. Our clinico‐radiological diagnosis of NAV correlated with cIMT and PWV measurements, which were significantly lower than those in AV, and well below the predefined cut‐off values for subclinical atherosclerosis. There have been very few descriptions of similar findings in the literature, and detailed angiographic descriptions or reports on cIMT and PWV are lacking [1, 21]. While the extracranial arteries are relatively easily accessible to Doppler examination for atherosclerotic plaque, the assessment of intracranial atherosclerosis remains a challenge. Our data show that intracranial monofocal occlusion of a large vessel is not associated with surrogate markers of subclinical atherosclerosis.

The almost identical TRF profile, cIMT and PWV between patients with NAV and those with multifocal vasculitis (including VZV, which has no known association with atherosclerosis) again implies that the described monofocal occlusive vasculopathy (i.e., NAV) is unlikely to be atherosclerotic in nature. It is unknown whether they are part of a spectrum ranging from monofocal occlusion to multifocal vasculitis, or distinct disease entities, but both appear to be independent of atherosclerosis.

Multifocal vasculitis: VZV is an underdiagnosed cause

Angiographic features of multifocal vasculitis have been described previously, often secondary to opportunistic infections or meningitis. VZV in particular has a known association with stroke [10, 11, 12]. Our most noteworthy finding was the identification of VZV in a large number of vasculopathy‐related strokes in PLWH, despite us having excluded all patients with meningitis from our cohort. This critical observation suggests that VZV may be associated with stroke by causing a vasculitis in the absence of any CSF response and thus may be overlooked in patients with a normal basic CSF examination. The cutaneous features of VZV are not always present and may precede stroke by over 2 years [10]. We also did not have the availability of testing for VZV IgG on CSF, which is known to be more sensitive than PCR [10, 13]. The true prevalence of VZV‐associated strokes is thus likely to be higher than what we have detected, and the index of suspicion needs to remain high to detect this potentially treatable cause of stroke. The vital clinical implication is that based on these findings, we recommend testing for VZV in the CSF of all PLWH with stroke, even if the basic CSF examination is within normal parameters. Whether or not treating VZV in this context will prevent ongoing vessel damage and reduce future stroke risk is unproven, but sufficiently plausible to consider antiviral treatment.

The vasculopathy associated with VZV was exclusively intracranial, with five of six causing a multifocal vasculitis, with one intracranial monofocal occlusion. Previous descriptions of VZV‐related stroke indicate that any vessel size may be affected [10]. We failed to demonstrate any patients with small vessel vasculitis on CT angiography. Our patients with VZV‐related stroke had evidence of greater immunocompromise than non‐VZV‐related stroke, which is in keeping with most opportunistic infections.

HIV‐related vasculitis was diagnosed in 21.4% of vasculopathy patients. This is a diagnosis of exclusion, after ruling out other infectious and non‐infectious causes of vasculitis. Our differences in patient profiles in HIV‐associated vasculitis when compared with VZV‐vasculopathy suggests that these are likely distinct clinical entities. HIV‐associated vasculitis occurs in less severely immunocompromised patients and thus appears less likely to be related to opportunistic infections, compared with VZV, which was seen in more profoundly immunocompromised PLWH. VZV reactivation has previously also been shown to be more common in patients with more advanced immunocompromise [10, 13].

We observed with interest how monofocal occlusion was more prevalent extracranially, with multifocal vasculitis more prevalent intracranially. However we did not demonstrate any differences in demographics, TRFs or HIV‐related factors between intracranial and extracranial vasculopathy. This is in contrast to a previous description of extracranial vasculopathy occurring in patients with higher CD4 counts than those with intracranial vasculopathy [5]. The presence or absence of atherosclerosis was not commented on. Aneurysmal changes were also described in this case series; however, we found only one patient with an aneurysm (accompanying vasculitic changes in other intracerebral vessels).

Small vessel disease has a high TRF burden

SVD was an uncommon cause of stroke in PLWH (16.5%; Table 1). These patients' greater age and higher TRF burden (with elevated mean cIMT) suggests that the process of SVD stroke appears to be primarily TRF‐driven, and possibly incidental to HIV (with similar SVD patient profiles in PLWH and people‐without‐HIV). Whether SVD is truly an HIV‐associated vasculopathy is uncertain. The normal PWV measurements were expected, as PWV is a measure of large artery stiffness, not SVD [9, 18].

The study found a lower prevalence of TRFs in PLWH stroke patients compared with age‐ and sex‐matched people‐without‐HIV with stroke, in accordance with most recent SSA data [22, 23, 24, 25, 26]. Thus, while TRFs do play a role in certain stroke types (AV and SVD), the majority of stroke in PLWH in our setting appears to be independent of TRFs and atherosclerosis, even after the exclusion of meningitis. This is in stark contrast to studies from high‐income countries (HICs) [27, 28, 29, 30]. Our findings may represent a low‐middle‐income country still in transition from one with infection‐mediated to TRF‐mediated cardiovascular disease in PLWH [20, 21, 31]. We also noted no increased TRF prevalence nor any difference in cIMT or PWV in the ART‐exposed group compared with the ART‐naïve PLWH. This supports previous data suggesting that stroke in PLWH in SSA does not appear to be related to ART‐induced metabolic side effects, as opposed to in HICs [1, 24, 30]. Our low rates of protease inhibitor usage may be partly responsible for this.

Clinical implications

There are potential clinical implications of our findings of vasculopathy with a low atherosclerosis burden. Current risk‐prediction tools target atherosclerosis‐mediated stroke and have been shown to be inaccurate predictors of stroke risk in PLWH [32, 33]. To our knowledge, this is the first study reporting on markers of subclinical atherosclerosis in PLWH with stroke in SSA. While cIMT and PWV are admittedly surrogate measures, they are both well‐validated and highly sensitive markers of subclinical atherosclerosis, making it unlikely that we have missed significant subclinical disease [7, 8, 9]. If future studies confirm that stroke in PLWH in SSA is largely unrelated to atherosclerosis, a paradigm shift in our investigation and management may be needed. This includes a higher index of suspicion for subclinical infections despite their apparent exclusion, as we have demonstrated with VZV.

Limitations of our study included small sample sizes and the lack of a PLWH non‐stroke control group, which precluded direct comparisons of cIMT and PWV data. Our low AV prevalence did not permit for age‐matching within the PLWH cohort and also precluded well‐powered comparative cIMT and PWV analyses with other stroke aetiologies. We thus chose to analyse cIMT and PWV according to internationally accepted cut‐off values for atherosclerosis; however, we acknowledge that these have not been extensively tested in PLWH in our population. We chose to utilize cut‐off values on the lower end of normality, to increase the sensitivity of the analysis. While the cIMT and PWV operator was blinded to the patient's HIV status, it was not possible to blind the specialist neurologist and radiologist, which may have biased other assessments. The addition of other markers of atherosclerosis, such as arterial calcification on CT, would have been of interest, however this data was unavailable. We acknowledge that the angiographic features of monofocal occlusion may also be observed in patients with emboli from a proximal source. This was considered unlikely due to normal angiography of the proximal vasculature and normal cardiac investigations.

CONCLUSIONS

We have shown that vasculopathy contributed to almost a third of strokes in young South African PLWH. A largely non‐atherosclerotic pathogenesis is suggested, and cIMT and PWV have been shown to be useful tools in distinguishing AV from NAV.

Despite excluding meningitis from our cohort and the non‐availability of optimal diagnostic assays, we still implicated VZV in a large number of these vasculopathies, emphasizing the likely underdiagnosis of this potentially treatable infection. We thus recommend testing for VZV in the CSF of all PLWH with strokes and maintaining a very high index of suspicion for covert VZV infection (even in the absence of CSF pleocytosis or recent skin rash).

Our findings regarding VZV emphasize that HIV‐related stroke in SSA still appears to have a strong link with opportunistic infections. The appropriate treatment of HIV is therefore of major importance for primary stroke prevention, as opposed to the targeting of TRFs and atherosclerosis, of which we have found minimal evidence. Our findings are in stark contrast to data from HICs with their high atherosclerosis burden, and emphasizes the importance of region‐specific research to guide the local management and prevention of HIV‐related stroke.

AUTHOR CONTRIBUTIONS

ES: Conceptualisation, study design, investigation, data acquisition, data analysis, writing and editing. AW: Conceptualisation, study design, data analysis, supervision, writing and editing. GT: Investigation, data acquisition. JN: Writing and editing. GN: Conceptualisation, study design, supervision and editing. GM: Conceptualisation, study design, supervision and editing.

FUNDING INFORMATION

ES was supported by the Discovery Foundation Academic Fellowship Award (Ref. 035747). The funder played no role in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

Ethics approval was granted by the University of the Witwatersrand Human Research Ethics Committee (Certificates M140429/M190688/M240682).

Supporting information

Data S1: Supplementary Method: Traditional cardiovascular risk factor (TRF) definitions, Laboratory investigations, Cardiac evaluation and Techniques for carotid intima‐media thickness (cIMT) and pulse wave velocity (PWV) measurements.

FIGURE S1: Study flowchart.

TABLE S1: Factors associated with carotid intima‐media thickness (cIMT) and pulse wave velocity (PWV) on multivariate regression analyses.

TABLE S2: Comparing traditional cardiovascular risk factors (TRFs) and HIV‐related factors between different locations and types of vasculopathy.

TABLE S3: Comparisons of small vessel disease (SVD) strokes in people living with HIV (PLWH) to non‐SVD strokes in PLWH, and to SVD strokes in people‐without‐HIV.

TABLE S4: Comparing traditional cardiovascular risk factors (TRFs) and HIV‐related factors between antiretroviral therapy (ART)‐exposed and ART‐naïve people living with HIV (PLWH).

Sadiq E, Woodiwiss A, Tade G, Nel J, Norton G, Modi G. The role of atherosclerosis in HIV‐associated vasculopathy in young South African stroke patients. HIV Med. 2025;26(4):633‐642. doi: 10.1111/hiv.13764

DATA AVAILABILITY STATEMENT

All data will be made available to qualified researchers upon reasonable request.