Central Nervous System Vasculopathy in HIV-Infected Children Enrolled in the Pediatric AIDS Clinical Trials Group 219/219C Study

Abstract

Background

Central nervous system (CNS) vasculopathy has been reported in human immunodeficiency virus–infected (HIV+) adults and children. In children, it often presents with HIV encephalopathy, stroke, or intracerebral aneurysms. The etiology, incidence, and risk factors of HIV-associated CNS vasculopathy in children are unknown.

Methods

We identified HIV+ children with a diagnosis of vasculopathy or other cerebrovascular events among children enrolled between 1993 and 2004 in 2 prospective, multicenter cohort studies. Demographic and laboratory data, history of antiretroviral use, and signs, symptoms, and diagnostic studies pertaining to the CNS event were reviewed.

Results

Among the 3338 HIV+ children, 51 had diagnoses that suggested CNS vasculopathy. Of these, 12 (24%) were included in this analysis, after excluding those with alternative diagnoses and those from closed sites. Among these 12, 4 (33%) were female, 4 (33%) were white, and 10 (83%) had perinatal HIV. Their average age at the event was 10.8 years with a median CD4 count of 22 cells/mm³ and median HIV-1 viral load of 94 304 copies/mL. Fifty-eight percent of subjects had a history of opportunistic infections before the CNS event. Fifty percent had cerebral aneurysms on imaging. The overall incidence among HIV+ subjects was 3.4 cases per 10 000 person-years (95% confidence interval, 1.8–6.0).

Conclusions

CNS vasculopathy in HIV+ children is uncommon but more common than in the general pediatric population. Cerebral aneurysms are the most common manifestation. Although the pathogenesis remains unclear, older children and those with low CD4 counts and high HIV viral loads are at the highest risk.

Despite the early recognition of central nervous system (CNS) disease (encephalopathy and developmental delay) in the pretreatment era of the pediatric human immunodeficiency virus (HIV) epidemic [1–3], reports on cerebrovascular events in the pediatric acquired immune deficiency syndrome (AIDS) population remain few in number [4–8]. Affected patients present with acute ischemic or hemorrhagic strokes or intracerebral aneurysms, presumably due to vasculopathy. The incidence of CNS aneurysms was estimated to be <1% among pediatric AIDS patients in the pre-combination antiretroviral therapy (cART) era [9]. The etiology of vasculopathy in both adults and children with HIV remains unknown [7]. However, direct infection by HIV-1 virus, immune complex deposition, and impaired regulation of the immune response have been implicated as likely causes [10]. The majority of HIV-infected (HIV+) children do not develop CNS opportunistic infections or neoplasms [2, 11]. Rather, they tend to develop severe early and progressive neurological deterioration, presenting mainly as HIV-1 encephalopathy or rarely as a stroke.

We sought to identify cases of CNS vasculopathy, and specifically arteriopathy, among a large, multicenter prospective cohort of children with HIV infection and to identify presenting signs and symptoms. The primary objective of the study is to estimate the incidence of CNS vasculopathy in children with HIV infection, and to investigate potential risk factors for CNS vasculopathy-related events.

METHODS

Pediatric AIDS Clinical Trials Group (PACTG) 219/219C was a prospective, multicenter cohort study designed to assess the long-term consequences of HIV-1 infection and its treatments in HIV+ infants, children, and adolescents, as well as in utero/neonatal exposure to antiretroviral (ARV) drugs in HIV-1–exposed but uninfected (HEU) infants born to infected women. The original PACTG 219 study opened to enrollment on April 2, 1993 and was limited to children who were co-enrolled in another PACTG treatment trial or of children whose mothers participated in such a trial. The revised version, PACTG 219C, implemented on September 15, 2000, extended enrollment to any HIV+ or HEU child aged ≤21 years, regardless of previous participation or co-enrollment into PACTG clinical trials. PACTG 219C closed enrollment on April 25, 2006 and closed to follow-up on May 31, 2007. Further details on enrollment criteria for PACTG 219/219C have been previously published [12, 13]. The study was approved by the Institutional Review Board (IRB) at each participating institution. Written informed consent was obtained from each child's parent or legal guardian, with assent obtained from older children according to local IRB requirements. The study population for these analyses primarily focused on 3338 HIV+ infants, children, and youth who were enrolled into PACTG 219, PACTG 219C, or both between 1993 and April 2004; a comparison group of the 2354 HEU-enrolled children was included. All analyses included follow-up information through April 2004.

Suspect cases were selected if they carried a diagnosis of vasculitis, vasculopathy, stroke (either ischemic or hemorrhagic), transient ischemic attack, CNS infarction, CNS bleed, arteriovenous malformation, or aneurysm, suggesting CNS vasculopathy. All suspect cases were reviewed in detail by one of the authors and excluded if they were not related to a CNS event. Information obtained from the database included study version, date of study enrollment, gender, race and ethnicity, age at time of event, route of acquisition of HIV (vertical, horizontal, or transfusion-related), CD4+ count, and HIV viral load prior to event, Centers for Disease Control and Prevention (CDC) HIV disease classification, signs and symptoms at the time of the event, history of use of ARV drugs, whether the subject was on cART (defined as 3 or more drugs from at least 2 classes) at the time of the event, and all diagnoses reported since study enrollment. Requests for additional information were then sent to the site where the subjects received care, using a standardized case report form. Requested information included the results of evaluations performed at the time of the event including lumbar punctures, electroencephalograms (EEGs), and computerized tomography (CT) and magnetic resonance imaging (MRI) scans. Information was requested on other relevant conditions including chronic vasculopathy, seizure disorders, hypertension, renal disease, autoimmune disease, coincident illnesses, and hypercoagulable states. Exclusion criteria included thrombocytopenia at the time of CNS event, documented CNS infection other than HIV, primary CNS lymphoma, systemic lupus erythematosus, or a defined autoimmune disorder.

STATISTICAL METHODS

Summary statistics for characteristics of the children and adolescents developing CNS vasculopathy are reported. Incidence rates and corresponding confidence intervals (CIs) for vasculopathy were calculated overall and by calendar period based on a Poisson distribution. Statistical analyses were conducted using SAS version 8 (SAS Institute Inc, Cary, NC) and STATA version 9 (StataCorp, College Station, TX). We assume that a historical diagnosis of CNS vasculopathy would have been captured in PACTG 219C, although this may not have been the case in PACTG 219 because this prior study version collected only diagnoses that occurred while on study. Thus, in calculating the incidence rates for the development of CNS vasculopathy, the person-years of exposure time are assumed to start from birth for PACTG 219C participants and at enrollment for those enrolled only in PACTG 219.

RESULTS

We identified 51 subjects with diagnoses suggestive of CNS vasculopathy among the 3338 HIV+ subjects. After review, we excluded 27 as not being consistent with CNS vasculopathy. These subjects had either no clinical data supporting a CNS event, vasculopathy of a different organ system, or a CNS abnormality not related to vasculopathy such as cerebral atrophy, demyelination, or cerebellar dysfunction syndrome. Queries were sent to the clinical sites to obtain additional information on the remaining 24 cases. Results of 4 of these queries were inadequate to confirm a diagnosis of CNS vasculopathy. Eight other possible cases were excluded because the participating site did not respond or the site no longer participated in the study. The remaining 12 subjects were included in the analyses. Four of these subjects (33%) were diagnosed with a CNS event before enrollment, but they had sufficient prior data to include them in the analysis.

Demographic and health characteristics of the 12 subjects are presented in Table 1. Among the 12 subjects, 4 (33%) were female, 4 (33%) were white, 7 (58%) were black, and 1 (8%) was Hispanic. The majority (83%) had been exposed to HIV-1 perinatally. Of the remaining 2 subjects, the mode of exposure was transfusion for 1 and unknown for the other. The average age at the time of CNS event was 10.8 years with a range of 0–19 years. Only 4 (33%) subjects were less than 10 years of age. The CDC clinical classification was known for 7 subjects, and 5 (71%) of these were CDC class C. CD4 counts were performed at study entry for the 8 subjects enrolled before their CNS event. Among these 8, their median CD4 count was 58.5 cells/mm³ (range, 9–813) and their median CD4 percentage was 7.5% (range, 0–20) at entry. A CD4 count before the event was available for 8 additional subjects (range, 1.5–38 months before event) and within 3 months after the event for another. These 9 subjects had a median CD4 count of 22 cells/mm³ (range, 0–275). Their median CD4 percentage was 3% (range, 0–18), and all but 1 had CD4% <15%. HIV-1 viral loads were available before the CNS event in 6 subjects and within 3 months after the event in another. The median HIV-1 viral load for these 7 subjects was 94 000 copies/mL. The HIV-1 viral load was >10⁵ copies/mL for 3 subjects, >10⁴ but <10⁵ for 2, and <50 copies/mL for 2. It should be noted that the CNS events occurring among subjects enrolled in PACTG 219 typically occurred before the availability or routine collection of HIV RNA viral load measures. Likewise, 4 CNS events occurred before enrollment in either PACTG 219 or 219C, and CD4 counts were not available.

Table 1.

Demographic, Health, and Antiretroviral Characteristics For HIV+ Vasculopathy Cases and the Overall 219/219C HIV+ Study Population

Characteristic	Vasculopathy Cases, at Event (n = 12)	Overall 219/219C Study Population,a at Study Entry (n = 3338)
Age in years (median, range)	11.2 (0–19.4)	7.0 (0–22)
Male gender	8 (67%)	1621 (49%)
Race/ethnicity
Black Non-Hispanic	7 (58%)	1879 (56%)
White Non-Hispanic	4 (33%)	491 (15%)
Hispanic	1 (8%)	909 (27%)
Other	0 (0%)	59 (2%)
Perinatal HIV exposure	10 (83%)	3032 (91%)
CD4 count in cells/mm3 (median, range)b	22 (0–275)	708 (0–9463)
CD4% (median, range)b	3 (0–18)	27 (0–73)
CDC Class Cc	5 (71%)	628 (25%)
ARV historyd
Any ARV	10 (91%)	3159 (95%)
cART within 12 months	8 (73%)	1538 (46%)
PI-based regimen	9 (82%)	1244 (37%)

Abbreviations: ARV, antiretroviral; cART, combination antiretroviral therapy, defined as 3 or more drugs from at least 2 classes; CDC, Centers for Disease Control and Prevention; HIV, human immunodeficiency virus; HIV+, HIV-infected; PI, protease inhibitor.

^aInfected children only; HIV-exposed, but uninfected children are not included.

^bHIV disease severity measures and ARV history measured closest to time of event for vasculitis cases and at study entry for overall population. CD4 count was only available for 8 of 12 subjects with CNS vasculitis.

^cCDC Class available for 7 of 12 subjects at time of vasculitis event, percentage calculated of those with available data.

^dARV history reported among those with information available, including 11 of those with vasculitis.

We compared the 12 subjects with CNS vasculopathy to the overall HIV+ PACTG 219/219C population at study entry (Table 1). The overall HIV+ population was slightly younger (7 vs 11 years) with a lower proportion of males (49% vs 67%). It also had a lower proportion of white non-Hispanics (15% vs 33%) and a greater proportion of Hispanics (27% vs 8%). The HIV+ PACTG 219/219C population was more likely to be immunocompetent with a median CD4 count of 708 cells/mm³ (vs 22 cells/mm³) and a greater CD4% (27% vs 3%). Only 25% of PACTG 219/219C HIV+ subjects were CDC Class C at study entry compared with 71% of subjects with CNS vasculopathy.

Seven subjects were diagnosed with 1 or more opportunistic infections before the CNS event. Of note, 6 subjects had recurrent herpes zoster before their CNS event. The median time between the last episode of zoster and the CNS event was 29 months (range, 1–112 months). Other opportunistic infections included 5 children with oral or esophageal candidiasis, 2 with recurrent herpes simplex virus (HSV), 2 with Pneumocystis jiroveci pneumonia (PCP), 2 with Mycobacterium avium intracellulare (MAI), and 1 with cryptosporidiosis. Other comorbid conditions were diagnosed in 7 of the subjects before their CNS vasculopathy event and included hypertension (4), renal dysfunction (3), idiopathic thrombocytopenic purpura (2), hypercoagulability (1), autoimmune disorder (1), cardiomyopathy (1), pancreatitis (1), and lymphocytic interstitial pneumonitis (1). Four subjects were reported to have a seizure disorder, but it is unclear whether it presented before the CNS event.

Antiretroviral history was available for 11 of the 12 subjects with confirmed CNS vasculopathy, and 10 had received ARVs before being diagnosed with CNS vasculopathy (Table 1). Eight of the subjects (67%) had been on cART in the 12 months before their CNS event. cART was initiated a median of 18.5 months before their CNS event (range, 3–48 months), and 7 (64%) were on cART at the time of their CNS event; another 2 (18%) were on non-cART regimens. All subjects receiving cART were on protease inhibitor-based regimens and were taking an average of 4.3 ARV drugs. No specific ART medication associations were apparent.

Eight of the 12 subjects had their CNS event after enrollment in the study. Each of these had either signs or symptoms of the event noted in the query or a specific diagnosis was reported. Four of the subjects presented with a complaint of intermittent blurred vision, apnea, coma, or left-sided weakness. Five of the subjects had a specific diagnosis reported: 2 (25%) subjects were diagnosed with stroke, 2 (8%) with transient ischemic attacks, and 2 (8%) with both transient ischemic attacks and stroke. The most common diagnoses made include hemorrhagic or ischemic stroke (67%), seizure (33%), hemiparesis (33%), aneurysm (33%), HIV vasculopathy (33%), recurrent transient ischemic attacks (25%), and Circle of Willis vasculopathy (25%). Almost all subjects had more than 2 diagnosis related to their CNS event.

Among the 12 confirmed cases, results of their diagnostic evaluation were available for 11. The child with the missing evaluation was diagnosed with vasculopathy prior to entry into the study. Two of the 5 with a cerebral spinal fluid evaluation had abnormal results. Both had elevated protein (64 mg/dL and 94 mg/dL), and 1 had an elevated white blood cell count (17 cells/mL, 77% lymphocytes and 21% monocytes). Only 1 subject had an EEG performed and it was interpreted as abnormal. All 11 subjects had brain imaging performed. Ten had a CT scan performed and all demonstrated abnormalities. All 11 had abnormalities on MRI (Table 2). Six subjects were diagnosed with cerebral aneurysms, 2 with ectasia, 2 with narrowing of a medium-sized artery, 3 with significant areas of infarctions or ischemia, 1 with hemorrhage, and 1 with HIV vasculopathy and hyperintensities of the putamen and caudate nucleus. Six (50%) subjects had more than 1 finding on the MRI.

Table 2.

Findings, Outcomes, and Sequelae For the 11 HIV+ Subjects With Central Nervous System Imaging^a

Patient Number	CNS Lesion Found by MRI	Neurologic Deficit	CD4 Count (%)b	Viral Load (copies/mL)	Therapy at CNS Event	Outcome
1	Circle of Willis ectasia, calcification of basal ganglia	Yes	22 (0)	183 962	DDI	Alive
2	Multiple aneurysms, multiple infarcts	Yes, blurred vision	5 (1)	41 000	DDI, D4T, NVP, SQV, NFV	Died
3	Internal carotid artery aneurysm	Yes; seizures, right hemiparesis	15 (3)	94 304	DDI, D4T, ABC, IDV, NFV	Died
4	Right MCA infarct	Yes	ND	ND	Not on therapy	Alive
5	Left MCA stroke	Yes	12 (3)	ND	DDI, D4T, EFV, RTV, IDV	Died
6	Left occipital/parietal hemorrhage	Yes	178 (3)	358 996	DDI	Died
7	Hyperintensities in putamen and caudate nucleus; HIV vasculopathy	Yes, coma	0 (0)	1 404 148	Not on therapy	Died
8	Aneurysm of left ACA and narrowing of right MCA	Yes	275 (18)	<50	ABC, ZDV, 3TC, TDF, LPV/r	Alive
9	Narrowing of right middle cerebral artery	Yes	144 (12)	<50	Unknown	Alive
10	Fusiform CNS aneurysm	Hemiplegia	ND	ND	DDI, EFV, IDV, RTV	Alive
11	Multiple CNS aneurysms	Yes	ND	ND	D4T, 3TC, IDV	Alive
12	Diffuse CNS aneurysms with ectasia	Yes	34 (6)	179 914	D4T, NVP	Died

Abbreviations: ABC, abacavir; ACA, anterior cerebral artery; CNS, central nervous system; D4T, stavudine; DDI, didanosine; EFV, efavirenz; HIV, human immunodeficiency virus; HIV + , HIV-infected; DV, indinavir; LPV/r, lopinavir/ritonavir; MCA, middle cerebral artery; MRI, magnetic resonance imaging; ND, test not done; NFV, nelfinavir; NVP, nevirapine; RTV, ritonavir; SQV, saquinavir; TDF, tenofovir; ZDV, zidovudine; 3TC, lamivudine.

^aAll subjects had an MRI except Patient 4, who was enrolled 4 years after the CNS event; the mode of imaging is unknown for this patient.

^bCD4 counts shown were performed prior to or within 3 months of the CNS event.

Of the 12 subjects, 6 (50%) died during the study period (Table 2). Their mean age at the time of death was 13 years, and death occurred a mean of 24 months after the CNS event (range, 2 weeks to 49 months). Causes of death included sepsis, aspiration pneumonia, cardiomyopathy, chronic hepatitis, acute renal failure, and PCP. Only 1 death resulted directly from the CNS event. The 6 surviving children had been diagnosed with CNS events an average of 6.3 years (range, 2–11 years) before this analysis. Four (33%) subjects had no reported long-term deficits. However, 2 of these subjects died 18 and 30 months, respectively, after the CNS event. Neurologic deficits continuing for >30 days after the CNS event were considered to be permanent sequelae. These included hemiparesis or hemiplegia (33%), seizure disorder (33%), and hearing loss (8%).

Follow-up time was censored at diagnosis of CNS vasculopathy for cases, and at the last visit date on study before April 1, 2004 for non-cases. The incidence rate is 3.4 cases per 10 000 person-years (95% CI, 1.8–6.0) among all HIV+ subjects, and 3.3 cases per 10 000 person-years (95% CI, 1.6–6.1) among perinatally infected subjects. Four cases of CNS vascular events were found among the 2354 HEU subjects. Of these, 3 were premature with intracranial hemorrhages in the neonatal period and were excluded. The fourth subject had a CNS event at 600 days of life. This single subject was used to calculate an incidence rate of 1.4 cases per 10 000 person-years (95% CI, 0.04–7.9) among the HEU subjects. The incidence rate ratio for HIV+ to HEU subjects is 2.44 (95% CI, 0.48–44.3).

DISCUSSION

We report an incidence rate of cerebrovascular events of 3.4 cases per 10 000 person-years (95% CI, 1.8–6.0) among all HIV+ subjects and 3.3 cases per 10 000 person-years (95% CI, 1.6–6.1) among perinatally infected subjects. Despite being a rare event, this incidence rate is higher than the reported incidence rate of stroke in the general pediatric population. Reports of stroke in North American children range from 0.23 to 0.27 cases per 10 000 children-years [14–16].

More than 80% of HIV+ children in the pre-cART era had evidence of CNS involvement including acquired microcephaly, diffuse cerebral atrophy, calcifications of the basal ganglia, and HIV-associated encephalitis [3, 7, 17, 18]. Park et al [18] described 7 HIV+ children with clinical and/or neuroradiological evidence of stroke. They estimated the clinical incidence of cerebrovascular events to be 1.3% per year. Another study of 380 HIV+ children found that 1% had radiologic evidence of a stroke [6].

Whereas aneurysms of the great vessels can occur in the adult HIV population, CNS vasculopathy is more common in the pediatric population [4, 5]. Intracerebral aneurysms are rare in adults, but multiple case reports can be found in children [1, 6–9, 19–27]. In pediatric HIV patients, aneurysms are characteristically diffuse, fusiform dilatations of the larger arteries around the Circle of Willis [7, 21]. Children tend to present with aneurysms late in childhood (8–13 years old), which suggests that damage to the vessels occurs over time. In our study, 67% of children with a new diagnosis consistent with CNS vasculopathy were 10 years or older.

Some authors have suggested that the inflammation related to CNS vasculopathy, and aneurysm formation in particular, may result from coinfections in immunocompromised individuals [28, 29]. Possible etiologic agents include varicella zoster virus (VZV), cytomegalovirus (CMV), syphilis, and toxoplasma [10, 30–36]. However, with the exception of VZV, opportunistic CNS infections are uncommon in children [7, 9, 18, 21]. Half of the cases in our series had a diagnosis of VZV or recurrent VZV. The mortality rate among cases with a diagnosis of VZV during the study period was 66% compared with only 33% of cases without a diagnosis of VZV; given the limited sample size, this difference in mortality was not significant. In addition, no difference was seen in CD4 count or viral load. Previous studies have suggested that VZV may be associated with unilateral lesions [7], but we did not find this association in our study.

The possibility that CNS vasculopathy results from an inflammatory response to HIV in the CNS has also been suggested [37, 38]. There is evidence that HIV invades the CNS early in perinatal infection [30, 39]. HIV DNA has been identified in the cerebral white matter and basal ganglia as well as in blood vessel walls [6, 11]. Invasion of endothelial surfaces likely leads to inflammatory changes.

A limitation of our study was the unavailability of biopsy-obtained histologic data to further investigate potential etiologies of the lesions identified. We excluded subjects with confirmed opportunistic infection at the time of the CNS vasculopathy presentation (stroke or bleed). Six of 12 subjects in our cohort had a history of recurrent zoster in the past, and 6 of 11 subjects had a history of opportunistic infections in the past (such as VZV, CMV, HSV, MAI, Candida, PCP, and Cryptococcus). However, further analysis to validate an association between opportunistic infections and CNS vasculopathy was not possible due to the small number of cases. We assumed that a historical diagnosis of CNS vasculopathy would have been captured in PACTG 219C, whereas this may not have been the case in PACTG 219 because this prior study version collected only diagnoses occurring while on study. Thus, in calculating the incidence rates for the development of CNS vasculopathy, the person-years of exposure time are assumed to start from birth for PACTG 219C participants and at enrollment for those only in PACTG 219. The CNS event occurred in 4 subjects before enrollment in 219/219C, and for these subjects the signs, symptoms, and diagnostic tests were largely unavailable. It is also possible that other children in the cohort developed cerebrovascular pathology that remained clinically silent. These children would be missed in our study and result in an underestimation of the incidence of vasculopathy.

In conclusion, we found an increased incidence of cerebrovascular events in HIV+ children, with the incidence among the HIV+ subjects 2.4 times that of HEU subjects. Because of the small number of cases in the HEU group, this difference was not statistically significant. When compared with the experience of cerebrovascular events reported in the general North American pediatric population, however, cerebrovascular events in HIV+ children are 14 times more common. Although the pathogenesis of cerebrovascular diseases in HIV+ pediatric subjects is still unclear, our study suggests an association between CNS vasculopathy and an extremely low CD4 count and high HIV viral load. This predicts that CNS vasculopathy will become increasingly uncommon with highly effective ART.