Early Initiation of Antiretroviral Therapy is Protective Against Seizures in Children with HIV in Zambia: A Prospective Case-Control Study

Abstract

Background:

Seizures are relatively common among children with HIV in low-and middle-income countries, and are associated with significant morbidity and mortality. Early treatment with antiretroviral therapy may reduce this risk by decreasing rates of central nervous system infections and HIV encephalopathy.

Methods:

We conducted a prospective, unmatched case-control study. We enrolled children with new-onset seizure from University Teaching Hospital in Lusaka, Zambia as well as two regional hospitals in rural Zambia. Controls were children with HIV and no history of seizures. Recruitment took place from 2016–2019. Early treatment was defined as initiation of ART prior to 12 months of age, at a CD4 percentage greater than 15% in children ages 12 months to 60 months, or a CD4 count greater than 350 cell/mm³ for children 60 months or older. Logistic regression models were used to evaluate the association between potential risk factors and seizures.

Results:

We identified 73 children with new-onset seizure and compared them to 254 control children with HIV but no seizures. Early treatment with antiretroviral therapy was associated with a significant reduction in the odds of seizures (OR 0.04, 95% CI 0.02—0.09; p<0.001). Having an undetectable viral load at the time of enrollment was strongly protective against seizures (OR 0.03, p<0.001), while history of WHO Stage 4 disease (OR 2.2, p=0.05) or CD4 count <200 (OR 3.6, p<0.001) increased risk of seizures.

Conclusions:

Early initiation of antiretroviral therapy and successful viral suppression would likely reduce much of the excess seizure burden in children with HIV.

Introduction

Seizures are relatively common in people living with HIV (PLWH), occurring in up to 20% of untreated adults and 8% of children ^1–6. HIV may lead to seizures through a variety of mechanisms, including immune suppression leading to opportunistic infections, metabolic disturbances, toxicity of antiretroviral drugs, and chronic inflammation in the central nervous system ^7–16. Treatment of seizures is particularly challenging in populations with HIV in Sub-Saharan Africa, given difficulties accessing diagnostic tests and anti-seizure drugs as well as potential drug interactions between antiretroviral therapy (ART) and anti-seizure medications¹⁷. The widespread introduction of ART has transformed the HIV pandemic in low and middle income countries (LMICs), with severe immunosuppression becoming much less common ^18–22. In areas with access to ART, HIV has been transformed into a chronic, potentially manageable illness. However, new onset seizures in children with HIV continues to be a relatively common concern ^{5, 17}.

In a previous study conducted in Botswana from 2003–2009, we noted that early treatment with ART was protective against epilepsy ²³. We and others have noted that the primary risk factors for seizures in children with HIV included severe immunosuppression leading to opportunistic infections as well as HIV encephalopathy ^{5, 17, 23, 24}. However, risk factors for seizures in the current era of widely available ART are unclear.

Zambia is a lower-middle income country in Sub-Saharan Africa with relatively high rates of HIV infection, with approximately 12% of the adult population and 1.1% of children currently living with HIV ^{25, 26}. ART became widely available in Zambia in 2005, and a test-and-treat strategy was implemented to attempt to achieve viral suppression in all PLWH. However, recent estimates indicate that only about a third of children with HIV in Zambia are virally suppressed. ²⁵ In the current study, we sought to prospectively ascertain risk factors for new onset seizures among children with HIV in Zambia. Our primary hypothesis was that early initiation of combination antiretroviral therapy and sustained viral suppression would be associated with a reduced risk of seizures. Based on results from the HIV-associated Neurocognitive Disorders in Zambia (HANDZ) study suggesting that malnutrition is major risk factor for poor neurologic outcomes¹⁶, we also hypothesized that indicators of malnutrition would be associated with new-onset seizures.

Methods

Overview

We conducted a prospective unmatched case-control study among children ages 0–18 years with a history of perinatally-acquired HIV infection. Cases were children with new onset seizure presenting to the University Teaching Hospital in Lusaka or one of two regional hospitals in the Southern Province of Zambia. Controls were children with perinatally-acquired HIV recruited from medication refill visits at the Pediatric Center of Excellence with no history of seizures by the end of the study period.

Population and Setting

This research was completed as part of the Cohort of HIV-Associated Seizure and Epilepsy (CHASE) Study. Detailed methods of the CHASE study have previously been published ¹⁷. Briefly, from April of 2016 to June of 2019, children living with HIV (CLWH) presenting with new-onset seizures were prospectively identified at Zambia's University Teaching Hospitals (UTH) in the Lusaka, and at two non-urban sites (referred to collectively as Rural Health Consortium) located in Zambia’s Southern Province. Controls were recruited from the HIV-associated Neurocognitive Disorders in Zambia (HANDZ) study at the Pediatric Center of Excellence, a large referral center for children with HIV located on the UTH campus. Methods and further details of the population for the HANDZ study have been described in previous publications. ^27–30

Cases were identified based upon presentation with an event determined to be seizure by the study team, which included an epileptologist and pediatric neurologist. All potential cases at UTH were offered EEG and head imaging, although some died or were lost to follow up before these could be completed. EEG and imaging were not available to patients seen in the rural hospitals. Inclusion criteria for cases were as follows: (1) 1 month-18 years of age, (2) HIV infection, as documented by PCR or diagnostic immunoassay (3) having experienced the index seizure <6 weeks prior to screening, and (4) written consent from parent/guardian. Inclusion/exclusion criteria for controls were identical to that for cases, except that controls were excluded if they had ever had a seizure prior to enrollment, or if they developed new-onset seizures during the follow up period of the study.

Data Collected

Data collected included subject demographics, HIV and clinical history, anthropometry, and laboratory and imaging results. Information on risk factors was collected from parent interviews, patient interviews (where possible) and chart review of inpatient and outpatient records. Socioeconomic status variables (maternal education, possessions owned, electricity, and access to piped water) were combined into a socioeconomic status index (SESI). ^31–33 Nutritional status was assessed by measuring weight for height and weight for age to define stunting(an indicator of chronic malnutrition) and underweight (an indicator of chronic or acute malnutrition) according to standard WHO definitions ³⁴. In order to facilitate comparisons with our earlier study²³, early treatment was defined according to 2010 World Health Organization (WHO) consensus criteria as initiation of ART prior to 12 months of age, at a CD4 percentage greater than 15% in children ages 12 months to 60 months, or a CD4 count greater than 350 cell/mm³ for children 60 months or older. Clinical and immunologic staging was defined according to WHO Guidelines. ³⁵

Statistical Analyses:

Statistical analyses were performed using Stata 16 (StataCorp LP, College Station, TX). Comparisons between variables were performed using t-tests for normally-distributed continuous variables, Kruskal-Wallis rank tests for ordinal or non-normally distributed continuous variables, and Chi-squared tests for dichotomous or categorical variables. Bivariable and multivariable logistic regression was used to evaluate the association between potential risk factors and seizures. Multivariable regression models were constructed using Directed Acyclic Graphs (DAGs) for variable selection (see Figure 1) ^36–38. Sensitivity analyses looked at whether risk factors remained significant when limited only to cases recruited from the urban site. Mediation analysis was performed using the “med4way” package in Stata 16 ^{39, 40} with potential mediators selected based on the DAG. Where variables were highly collinear (e.g. stunting and low weight), the variable with greater explanatory power in the logistic model (based on R²⁾ was used in the mediation analysis. Missing data were treated with listwise deletion in regression analyses. Significance level was set at p=0.05. Given the number of cases identified, we had 90% power to identify risk factors with odds ratios of 2.5 or greater.

Figure 1:

“Directed Acylic Graph (DAG) demonstrating the postulated causal pathways linking the exposure (early treatment with antiretroviral therapy) with the outcome (seizures).”

Ethics Statement:

Ethics approvals were granted by the University of Zambia's Biomedical Research Ethics Committee, the Zambian National Health Research Association, the University of Rochester School of Medicine, and all other appropriate associated US institutions. Written and verbal informed consent was sought from parents or legal guardian, and where appropriate, assent was sought from participating children over the age of 12. All care for enrolled children was delivered by the healthcare providers routinely caring for such patients within the enrolling facilities.

Results

Characteristics of Cases and Controls

We identified 72 participants with new onset seizures and 254 controls with no history of seizures. 3 children initially included in the control group had to be excluded as they developed seizures during the follow up period. A summary of the characteristics of cases and controls is presented in Table 1. Children with seizures were younger than controls (median age 6 vs. 11, p<0.001), and were more likely to have a family history of epilepsy (15% vs. 1%, p=<0.001). Other conventional risk factors for seizures such as birth complications and head trauma were minimal in this cohort and were similar between cases and controls. All cases and controls were of Black race and were from Zambia. Cases had significantly lower socioeconomic status than controls (median SESI 3 vs. 5, p=0.01), although these differences were nonsignificant when limited to solely the urban population. Cases were almost twice as likely to be underweight and have growth stunting.

Table 1:

Characteristics of Cases and Controls

Demographics
	Cases	Controls	P-value
	73	254
Gender male, n (%)	39 (53)	138 (54)	0.89
Age (years)  Mean (SD)	6.6 (4.6)	10.5(3.4)	<0.001
Median(IQR)	6 (2.2 –10)	11(8–13)
Family history of epilepsy, n (%)	10 (15)	2 (1)	<0.001
History of complications during birth, n (%)	2 (3)	20 (8)	0.10
Mother’s Age, Mean (SD)	33.3 (6.9)	38.5 (6.5)	<0.001
Median Socioeconomic Status Index (IQR)	3 (2—6)	5 (3–7)	0.01
Mean years of maternal education (SD)	8.0 (3.0)	7.7 (3.3)	0.56
Mother attended post-secondary school (e.g. college, nursing school), n (%)	0	27 (10%)	0.004
Running water in home, n (%)	33 (59)	122 (48)	0.16
Improved toilet facilities in home, n (%)	4 (7)	79 (38)	<0.001
Electric lighting in home, n (%)	25 (44)	197 (78)	<0.001
Median Possession Index (IQR)	2 (1–3)	3 (1–4)	<0.001
Antenatal care during pregnancy, n(%)	N = 53	N=221	0.23
	49 (92)	191 (86)
PMTCT care during pregnancy, n (%)	N = 54	N=230
	13 (24%)	46 (19%)	0.11
Weight for age, n (%)	N = 56	N=251	<0.001
Underweight	27 (48%)	66 (26)
Normal	29(52)	185 (73)
Length/height for age, n (%)	N = 37	N=204	<0.001
Stunting	23 (62)	66 (32)
Normal	14 (37)	138 (67)

Cases and controls had significant differences on almost all HIV variables measured (see Table 2). Cases were much less likely to be taking ART at enrollment (49% vs 100%, p<0.001). Cases had lower CD4 counts and percentages, more advanced WHO Stage, and were substantially more likely to have detectable viral loads. Among those cases who were taking ART, median age of ART initiation was similar between cases and controls (72 months vs. 70 months, p=.46). However, cases were more likely to have been started on ART as a consequence of developing advanced disease, while controls were more likely to have been started on ART as part of a test and treat protocol.

Table 2:

HIV-specific characteristics of cases and controls

	Cases	Controls	P-values
CD4 Count	n=64	N=236	<0.001
Mean (SD)	603 (693)	840 (456)
Median (IQR)	410 (130–835)	774 (542–1020)
CD4 count less than 200, n (%)	20 (31)	17 (9)	<0.001
CD4 %	N=54	N=232	<0.001
Mean (SD)	19.8 (12.6)	30.6 (9.8)
Median (IQR)	21 (9–28)	32 (25–37)
Worst WHO Stage, n (%)			0.003
Stage 1	11 (15)	36 (15)
Stage 2	4 (5)	35 (15)
Stage 3	10 (14)	62 (26)
Stage 4	48 (66)	98 (41)
Last viral load undetectable, n (%)	7 (13)	199 (83)	<0.001
Plasma viral load in those with detectable viral load, in copies/ml	N = 47	N=33	0.05
Mean (SD)	324,737 (715,194)	70,685 (208,300)
Median	80,513 (2656–252,015)	7439 (361–27,198)
History of prior opportunistic infection, n (%)	18 (25%)	17 (7%)	<0.001
Taking ART, n(%)	36 (49%)	254 (100%)	<0.001
Duration of ART use, n (%)	N = 34
< 2 months
2 -- 6 months	6 (18)	0
6 months – 1 year	4 (12)	0
> 1 year	5 (15)	0
	19 (56)	100%
Mean age at time of ART initiation, in years (SD)	4.7 (3.4)	4.8 (3.6)	0.88

Characteristics and Etiology of Seizures.

Complete details on seizure etiology and outcomes in the CHASE cohort have previously been published¹⁷. Briefly, seizure semiology was focal in 30% and generalized in 70% of participants. A single child was identified who had likely simple febrile seizures. None of the participants fit into a typical genetic/idiopathic epilepsy syndrome (e.g. childhood absence epilepsy). Infectious causes were identified in over half (54%) of the cases, with bacterial meningitis, tuberculous meningitis, and HIV encephalitis, being the most common. No etiology could be identified in 12 (16%) children, most of whom died before the completion of clinical investigations. 56% of participants had abnormal MRI scans, with the most common abnormalities including generalized atrophy (in 49%) and subcortical white matter abnormalities (in 38%). Focal lesions were identified on MRI in 10% of patients, including prior stroke and subdural hematoma. 39% of children who had an EEG completed had focal or multifocal epileptiform discharges identified. Twenty-seven children (37%) presented with status epilepticus, and twenty-two (30%) children died within 30 days of the index seizure.

Risk Factors for Seizures

Early treatment with ART was strongly protective against seizures (OR, 0.04, 95% CI 0.02—0.09; p<0.001) in the unadjusted analysis; this effect was essentially unchanged in the multivariable model controlling for age and SESI (OR 0.05, 95% CI 0.02—0.09; p<0.001). This effect persisted when the analysis was limited only to the urban cases. This was primarily driven by cases being treated at low CD4 counts or being untreated as of presentation, as treatment initiation at age <12 months was uncommon in both groups and not significantly different between cases and controls (in 8% of cases and 15% of controls, p=0.1). Having an undetectable viral load at the time of enrollment was strongly protective against seizures (OR 0.03, 95% CI 0.01–0.07, p<0.001), while history of WHO Stage 4 disease (OR 2.2, 95% CI 1.3–3.9, p=0.05) or CD4 count <200 (OR 3.6, 95% CI 1.8–7.0, p<0.001) increased risk of seizures. Both stunting (OR 3.7, 95% CI 1.8–7.7, p<0.001) and being underweight (OR 7.2, 95% CI 3.8–13.7, p<0.001) were also significantly associated with risk of seizures. There was no significant association between any specific antiretroviral drug and seizures. In the mediation analysis, 39% of the effect of early treatment was mediated by viral suppression at the time of enrollment; while approximately 39% of the effect was mediated through being underweight.

Discussion

This study demonstrates decreased rates of new-onset seizures in children with HIV receiving early treatment with ART. The most common etiologies of seizures in our population included CNS infection and HIV encephalopathy, and both of these etiologies may be reduced by early treatment with ART ^{16, 19, 41}. Our study also notes that nutritional status including being underweight and stunted are strongly associated with seizures. This complements recent findings from the HANDZ study that chronic malnutrition in children with HIV may be a key factor contributing to poor neurologic outcomes. Of note, a significant number of cases (15%) had a family history of seizures, suggesting that either primary genetic epilepsies or a familial tendency towards seizures may have contributed in these cases.

The optimal timing of antiretroviral therapy in children is now generally considered to be “as soon as possible,” and almost all national health systems have now adopted a test-and-treat strategy ^{21, 24, 42–51}. In Zambia, prevention of mother to child (PMTCT) programs are well-established and quite effective, and rates of perinatal transmission have fallen dramatically over the last decade. However, a number of children with HIV fall through the cracks of the treatment cascade and end up presenting with very advanced disease, with seizures sometimes representing the initial presentation to the health care system in LMICs. ¹⁷ Our study is broadly in line with prior studies on seizures and HIV in children, including our own prior work in Botswana²³, and recent study completed in South Africa⁵². Prior studies of early treatment in diverse settings have demonstrated a range of benefits including reduced mortality, reduced rates of CNS infections, and reduced rates of HIV encephalopathy. Our study further suggests that programs which attempt to identify children with HIV early, initiate and maintain treatment, and encourage adherence could substantially reduce the burden of neurologic morbidity and mortality.

Our study has several limitations. First, controls were recruited from a single urban center, and thus may be substantially different from cases recruited from rural areas. However, in the sensitivity analysis, risk factors remained significant even when restricted only to the urban population. Several important mediators, such as the role of systemic inflammation, were unable to be measured in the current study. However, given that the cases were largely not virally suppressed and many were presenting in the setting of opportunistic infections, inflammatory markers are not likely to have been particularly informative. Finally, while we noted that a substantial portion of the effect of early treatment was mediated through being underweight, it is very likely that in this population weight is at least partially acting as a marker of untreated HIV ^53,54. Nevertheless, this study raises the possibility that nutritional support programs included into HIV treatment programs might further reduce the risk of neurologic complications.

Conclusion

In summary, early treatment with antiretroviral drugs is likely to be protective against seizures in children with HIV. Further research on is necessary to develop care pathways to ensure that children at risk for HIV infection are identified early, started on ART, and maintained on treatment throughout their lifespan. Finally, the role of malnutrition in contributing to neurologic morbidity needs to be further investigated, but a growing body of evidence suggests that chronic malnutrition may be a key factor contributing to neurologic comorbidities in children with HIV.