Association between HIV and cytomegalovirus and neurocognitive outcomes among children with HIV

Abstract

Objective(s):

Children with HIV may experience adverse neurocognitive outcomes despite antiretroviral therapy (ART). Cytomegalovirus (CMV) is common in children with HIV. Among children on ART, we examined the influences of early HIV viral load (VL) and CMV DNA on neurocognition.

Design:

We determined the association between pre-ART VL, cumulative VL, and CMV viremia and neurocognition using data from a cohort study.

Methods:

Children who initiated ART before 12 months of age were enrolled from 2007-2010 in Nairobi, Kenya. Blood was collected at enrollment and every 6 months thereafter. Four neurocognitive assessments with 12 domains were conducted when children were a median age of 7 years. Primary outcomes included cognitive ability, executive function, attention, and motor. Generalized linear models were used to determine associations between HIV VL (pre-ART and cumulative; N=38) and peak CMV DNA (by 24 months of age; N=20) and neurocognitive outcomes.

Results:

In adjusted models, higher peak CMV viremia by 24 months of age was associated with lower cognitive ability and motor z-scores. Higher pre-ART HIV VL was associated with lower executive function z-scores. Among secondary outcomes, higher pre-ART VL was associated with lower mean nonverbal and metacognition z-scores.

Conclusion:

Higher pre-ART VL and CMV DNA in infancy were associated with lower executive function, nonverbal and metacognition scores and cognitive ability scores in childhood, respectively. These findings suggest long-term benefits of early HIV viral suppression and CMV control on neurocognition.

INTRODUCTION

Children with HIV often experience neurodevelopmental delays.[1,2] While early antiretroviral therapy (ART) initiation improves neurodevelopmental outcomes,[3-5] it does not fully mitigate adverse outcomes.[2,6,7] Viral suppression during infancy and early childhood has been associated with improved neurocognition among older children.[8-10] However, among children 5-7 years of age, central nervous system (CNS) damage persists despite early ART and viral suppression.[11-13]

The CNS is a protected site in terms of medication penetration and immune cell migration. Therefore, HIV viral replication in the CNS may contribute to long-term neurocognitive effects despite effective ART and systemic viral suppression. Persistence of HIV in the CNS could disrupt brain development and neuronal migration and function.[14]

Children with HIV have a higher risk of congenital cytomegalovirus (CMV) infection and earlier postnatal acquisition of CMV.[15,16] Congenital CMV can cause profound damage to the CNS with long-term effects on neurodevelopmental outcomes.[17] CMV is also an important opportunistic infection in people living with HIV,[18] and even in the absence of CMV-specific disease, has been associated with accelerated HIV disease progression.[19,20] There is also accumulating evidence that postnatal CMV acquisition could influence neurocognitive outcomes.[21,22] Understanding early biomarkers that predict neurocognitive outcomes could inform interventions and facilitate identification of children at high risk for neurocognitive delays. Using data from an ongoing cohort study, we aimed to determine the association between early viral load (VL) and CMV viremia on neurocognitive outcomes in school-aged children with HIV.

METHODS

This secondary longitudinal analysis was nested in the Optimizing Pediatric HIV treatment study (OPH, NCT00428116), which enrolled children who initiated ART by 12 months of age from September 2007 to August 2010 at Kenyatta National Hospital (KNH) in Nairobi, Kenya. The trial was conducted prior to the WHO recommendation of lifelong ART for all people with HIV. A subset of children who met pre-specified criteria were randomized to either treatment interruption (median: 4.3 [interquartile range: 3.5, 8.9] months) or continued treatment after 24 months of ART.[23] Blood samples were collected every 3 months for the first 24 months of follow up and every 6 months thereafter. Neurocognitive assessments were conducted at a median of 7 years of age. The study was approved by the University of Washington and Fred Hutchinson Cancer Center Institutional Review Boards and Kenyatta National Hospital/University of Nairobi Ethics and Research Committee. Caregivers provided written informed consent for their children’s participation.

HIV VL was quantified using the Gen-Probe assay on plasma samples, which had a lower limit of detection of 150 copies/ml. HIV VL area under the curve (AUC) from birth to the neurocognitive assessment was calculated using the trapezoidal rule with cubic splines. CMV DNA levels were measured from stored plasma specimens using real-time quantitative PCR as previously described,[24] with a limit of detection of one copy/reaction. Values were transformed to express measurements in international units (IU)/ml by dividing by 1.4. The PCR limit of detection was >1 copy per reaction or 35.7 IU/ml.[25] Peak CMV viremia by 24 months of age was measured by quantitative PCR on plasma with a lower limit of detection of 50 copies/ml for CMV DNA. Undetectable HIV VL and CMV DNA were designated half the value of the limit of detection.

Neurocognitive assessments included the Kaufman Assessment Battery for Children 2nd Edition (KABC),[26,27] the Behavior Rating Inventory of Executive Functioning (BRIEF),[28] the Bruinick's-Oseretsky Test of Motor Proficiency 2nd Edition Brief Form (BOT),[29] and the Visual Test of Variables of Attention (TOVA).[30] Four primary outcomes included cognitive ability from KABC, executive function from BRIEF, motor from BOT, and attention from TOVA. Secondary outcomes included short-term memory, visual-spatial, learning, non-verbal test performance, and delayed memory from the KABC; behavior regulation and metacognition from the BRIEF; and processing speed from the TOVA. All neurocognitive outcomes were normalized and presented as z-scores. Children were included in this analysis if they had one or more exposures of interest. Inverse probability weighting was used to account for differential missingness by age of enrollment. Generalized linear models were used to determine the association between pre-ART HIV VL, HIV VL AUC, and CMV DNA levels, and neurocognitive outcomes (Supplementary Table 1). Adjustment variables included infant sex at birth, caregiver education, and age at neurocognitive test. Benjamini-Hochberg approach was used to account for multiple comparisons for primary and secondary outcomes. Stata version 18.0 (Stata Corporation, College Station, Texas USA) was used for all analyses.

RESULTS

Of 38 children who completed neurocognitive assessments, 38 had data on pre-ART VL and 22 had CMV DNA levels measured prior to 24 months of age. Median age at ART initiation was 4.6 (interquartile range [IQR]: 4.1-5.3) months, half of the children were female (19 [50%]) and 42% were on a first-line protease inhibitor (PI)-based regimen. Twenty (91%) participants with CMV DNA levels had detectable CMV by 24 months of age. The majority of caregivers [37 (97%)] were the child’s biological mother. Of 22 participants in this analysis who met randomization criteria, 11 (50%) were randomized to the treatment interruption arm. Median caregiver age was 26 (IQR: 23-34) years and median caregiver years of education was 10 (IQR: 8-12) years (Table 1).

Table 1.

Characteristics of children with neurocognitive assessments

	Participants with viral load data		Participants with CMV data
	N	Median (IQR) N=38	N	Median (IQR) N=22
Age at ART initiation (months)	38	4.6 (4.1-5.3)	22	4.4 (3.6-4.6)
Child age at neurocognitive assessment (years)	38	7.1 (6.6-7.6)	22	6.8 (6.3-7.2)
Female (REF: Male)	38	19 (50%)	22	10 (45%)
PI-based regimen (all children)	38	16 (42%)	22	11 (50%)
PI-based regimen (no switches)	22	15 (68%)	15	10 (67%)
Ever breastfed	36	33 (92%)	22	21 (95%)
Ever hospitalized	38	21 (55%)	22	11 (50%)
Caregiver age (in years)	37	26. (23-34)	21	26 (24-29)
Highest level of education (binary)	38		22
None/Primary		23 (61%)		14 (64%)
Secondary/College		15 (39%)		8 (36%)
Caregiver number of years of education	32	10.0 (8.0-12.0)	20	10.0 (8.0-11.0)
Primary caregiver: biological mother (REF: Other)	38	37 (97%)	22	22 (100%)
Crowding (>3 people per room)	25	16 (64%)	22	9 (64%)
Monthly house rent (KSH)	37	1500 (1200-3750)	20	1500 (1100-3000)
Number of children randomized to interrupted ART	22	11 (50%)	11	5 (45%)
Baseline VL log10copies/ml	38	6.6 (6.0-7.0)	22	6.6 (6.0-7.2)
Age at baseline VL (months)	38	4.2 (3.6-5.0)	22	3.9 (3.4-4.3)
VL AUC (log10copies/ml-months)	38	215 (187-278)	22	243.3 (173.2-281.7)
Number of VL measures prior to neurocognitive assessment	38	29 (15-44)	22	29 (15-43)
Age at first viral suppression (months)	38	10 (8-14)
Viral suppression (<1,000 copies/ml) at neurocognitive assessment	38	26 (68%)	22	14 (64%)
CD4 percent	34	17.6 (14.0-24.0)	22	17.1 (14.0-22.0)
Detectable pre-ART CMV (REF: no pre-ART CMV)	19	16 (84%)	19	16 (84%)
Detectable CMV by 24 months of age	22	20 (91%)	22	20 (91%)
Peak CMV by 24 months of age (log10copies/ml)	20	4.3 (3.8-5.0)	20	4.3 (3.8-5.0)

Primary outcomes

In adjusted models, one log₁₀ higher peak CMV viremia copies/ml was associated with 0.30 lower mean cognitive ability z-score (−0.30; 95% confidence interval [95%CI]: −0.56, −0.03; p=0.027). Peak CMV viremia was associated with lower mean motor z-scores (−0.24; 95%CI: −0.46, −0.01; p=0.04). Children with one log₁₀ higher pre-ART HIV VL (−0.55; 95%CI: −0.99, −0.11; p=0.014) had lower mean executive function z-scores (Figure 1).

Figure 1. Predictors of neurodevelopmental outcomes among school-age children with HIV.

Adjustment variables included infant sex at birth, caregiver education, and age at neurocognitive test. Pre-ART viral load HIV viral load area under the curve (AUC), and peak cytomegalovirus (CMV) were continuous exposures. Pre-ART viral load and peak cytomegalovirus DNA were log₁₀ transformed.

HIV viral load

CMV

Secondary outcomes

One log₁₀ higher pre-ART VL copies/ml was associated with 0.32 lower mean nonverbal and 0.77 lower mean metacognition z-scores (−0.32; 95%CI: −0.63, −0.02; p=0.037 and −0.77; 95%CI: −1.37, −0.18; p=0.011, respectively). Pre-ART HIV VL, HIV VL AUC, and peak CMV viremia were not associated with short-term memory, visual-spatial, learning, delayed memory, or processing speed z-scores (Supplemental Figure 1).

DISCUSSION

Higher pre-ART VL and early CMV levels were generally associated with poorer neurocognitive outcomes. Children with higher levels of pre-ART VL had lower mean executive function z-scores. Cognitive ability and motor scores were lower among children with higher CMV viremia.

Children with higher pre-ART HIV VL had significantly lower mean executive function, non-verbal, and metacognition z-scores, which aligns with existing evidence that has found that peak or early viral suppression is associated with better cognitive[8-10] and executive function scores.[19,31] In contrast to pre-ART VL, HIV VL AUC up to age at neurocognitive assessment was not associated with any neurocognitive outcomes, which aligned with our a priori hypothesis that pre-ART VL would be a better predictor of neurocognitive outcomes than AUC. The CNS is a sanctuary site for HIV-1, with evidence of persistent HIV replication despite ART.[32-34] The CNS is protected by a highly selective semipermeable blood-brain barrier, which makes it difficult for ART to enter the CNS.[34,35] HIV VL in plasma and in cerebrospinal fluid (CSF) are not well-correlated in children and adults after long duration of ART.[32,33,36-38] Some studies have found detectable HIV VL in cerebrospinal fluid among children and adults with sustained undetectable VL.[32,39] Plasma HIV VL during acute infection, however, has been associated with viral seeding of the CNS,[40] and there is evidence that VL in plasma and CSF are correlated immediately after ART initiation.[41-43] Thus, pre-ART VL better predicts long-term outcomes of children. Our findings suggest that obtaining VL prior to initiating ART could be useful to identify children who need more intensive follow-up for neurocognitive assessments. Our findings also underscore the importance of early infant HIV diagnosis programs for improving long-term neurocognitive outcomes among children with HIV. In 2021, only 52% of children with HIV were on ART;[44] innovative approaches to improve early infant diagnosis and linkage to care and treatment are necessary for improving long-term health and cognitive outcomes for children with HIV.

Few studies have evaluated CMV and neurocognitive outcomes in children, most of which have compared children with and without CMV infection, rather than assessing impact of levels of early CMV viremia. In the absence of maternal ART, up to 30% of children with in utero HIV infection may also acquire CMV in utero.[15,16] Given the age at which children in our cohort were enrolled, we were unable to determine which children may have had congenital CMV infection, which could account for the associations with neurocognition observed in our study. CMV causes a dramatic increase in cellular activation among diverse cell subsets.[45,46] Postnatal CMV acquisition could also affect neurocognitive outcomes through persistent immune activation and trafficking of activated innate cells, such as macrophages to the CNS. Consistent with this hypothesis, CMV infection has been found to be associated with increased levels of soluble CD163, an indicator of monocyte activation.[47] CMV seropositivity or infection has been associated with adverse neurodevelopmental[22] and neurocognitive outcomes,[21] , respectively. The latter study found that at 8 years of age CMV infection was associated with lower intelligence quotient but not any other neurocognitive outcomes.[21] We found that children with higher peak CMV viremia by 24 months of age had lower mean cognitive ability and mean motor z-scores at a median of 7 years of age. Our findings, in addition to the larger body of research that has found associations between CMV and neurocognition, suggest that measures to prevent CMV infection or reduce CMV viremia – including early ART initiation – could improve neurocognitive outcomes among children with HIV.

There are effective interventions to support children to overcome neurodevelopmental deficits.[48] A randomized controlled trial in Uganda found that computerized cognitive rehabilitation training can be effective for improving neurocognitive outcomes among children with HIV.[49] Additionally, caregiver training, non-computerized cognitive training, and interventions focused on improving child physical activity and nutrition have been effective in improving certain neurodevelopmental domains.[48] Our study found that pre-ART viral load and CMV viremia had impact on specific clinically relevant domains. Executive function and cognition have been associated with academic performance and social connection.[50,51] Motor and cognitive skills are often correlated, as motor skills can help children learn from their environment.[52] Therefore, early interventions to improve these neurocognitive domains could have a long-term impact. Given accumulating evidence that high pre-ART viral load and CMV viremia may influence neurocognition, it may be useful to focus on providing effective evidence-based interventions to children with higher pre-ART VL and CMV viremia during infancy.

This secondary analysis leveraged data from a longitudinal cohort of plasma pre-ART VL, repeated VL, and early CMV DNA measures with neurocognitive assessments conducted at a median of 7 years of age. We had small sample sizes of children with CMV DNA data which may have limited our ability to detect differences. We were not able to assess HIV VL or DNA in the CNS, which would be more biologically relevant to neurocognition. We also were unable to diagnose congenital CMV in this cohort.

Higher pre-ART VL and CMV viremia were associated with lower neurocognitive scores. Efforts to improve early ART initiation could improve neurocognitive outcomes for children with HIV. Findings from this study could inform future research and strategies to identify and support children with HIV who are most in need of effective interventions that could improve neurocognition.

Supplementary Material

Supplementary Figure 1

Supplementary Figure 1. HIV viral load

CMV