Clinical characteristics and outcomes after new-onset seizure among Zambian children with HIV during the antiretroviral therapy era

1 INTRODUCTION

Underlying neurological disease in people living with human immunodeficiency virus (PLWHIV) often manifests as a new-onset seizure, especially prior to widespread availability of antiretroviral therapies (ART). Between 1993 and 2003, among Indian populations with HIV, 20% of adults with neurological disorders related to advanced stages of immunosuppression presented with new-onset seizure.¹ Early in the HIV pandemic, seizures were commonly identified as part of the clinical phenomenon of severe immunosuppression though often without a clear underlying etiology.^{2, 3} Common seizure etiologies in this pre-ART era included opportunistic infections (OI),^{4, 5} other systemic illnesses secondary to HIV, adverse drug reactions, and metabolic derangements.⁶

The common co-occurrence of seizures in PLWHIV extend to children as well.^{7, 8} Similar to adults, children with HIV (CLWHIV) are also at risk of seizures secondary to OIs⁹ and stroke.¹⁰ Per the UNAIDS 2020 report, the incidence of CLWHIV (aged 0-14 years) in 2019 accounted for 150 000 (9%) of the estimated total 1.7 million cases worldwide. Of those children, 84% of the cases originated from sub-Saharan Africa.¹¹ In this region, expeditious diagnosis and treatment of new-onset seizures may be challenging due to limited resources for diagnostic assessments and lack of clear guidance on long-term management.^{10, 12}

Treating seizures in PLWHIV is particularly hazardous in resource-limited African settings, where enzyme-inducing antiseizure medications (EI-ASMs) are heavily relied on for epilepsy management. The co-usage of EI-ASM with ART is not recommended due to concerns of drug-drug interactions, which may lead to toxicity and/or failure of either agent.^{10, 13-15} As seizures in PLWHIV are strongly associated with advanced stages of HIV disease, early ART initiation is likely critical to avoiding this conundrum.⁷

Antiretroviral therapy rolled out nationwide in Zambia in 2005, and today, HIV testing and treatment programs utilizing “test and treat” policies and including programs for the prevention of mother to child transmission (PMTCT) are ubiquitous. To examine clinical characteristics and outcomes of new onset seizures in CLWHIV during this ART era, we identified CLWHIV with new-onset seizures in both urban and rural settings in Zambia. We examined clinical and demographic data including HIV treatment history, seizure etiology, semiology, and seizure severity. Lastly, due to identification of the high burden of early mortality, 30-day cause of death was also assessed.

2 METHODS

3 RESULTS

A total of 365 children were screened at UTH, with 290 deemed ineligible. See Figure 1 for enrollment details. Reasons for ineligibility were not mutually exclusive and included negative HIV status (184, 63%), event not new-onset seizure (83, 30%), a known history of epilepsy (38, 13%), age (29, 10%), a seizure onset time >6 weeks prior to screening (5, 2%) and death (4, 1%). Of the 108 eligible children, 8 (11%) parents declined to participate, 11 (15%) died before consent could be obtained, and 2 (3%) were lost to follow-up (absconded or were transferred to another unit of the hospital and untraceable). Fifty-four (72%) urban children were enrolled.

At the three non-urban sites, 369 children were screened, with 337 deemed ineligible. Reasons for ineligibility included negative HIV status (319, 95%), event not new-onset seizure (5, 1.5%), a known history of epilepsy (36, 11%), age (2, 0.6%) and death (1, 0.3%). Two (6%) of the 32 eligible children's parents declined to participate and 30 children were enrolled. After 6 months of screening, no eligible patients had been identified at Nakambala Hospital, with an external retrospective chart review confirming the lack of any children with HIV presenting with new-onset seizure, so this site was closed for enrollment. Across all sites, 11 consented infants with a positive HIV antibody test but subsequent negative DNA PCR were withdrawn. Among HIV-positive children screened for enrollment, 20% of the urban children and 72% of the rural children did not have new-onset seizure and instead were either known to have epilepsy or had epilepsy by history but without prior diagnosis and as such were excluded from enrollment. Ultimately, 49 urban and 24 non-urban children provided data for this analysis.

Demographic data for the 73 children are provided in Table 1. Median age was 6 years (IQR 2.2-10). Just over half (39, 53%) were male children. Several children were orphans being cared for by guardians other than parents and these individuals could only provide limited family and medical history. Prior to the index seizure, a quarter of children had a history of a CNS OI and 6 (8%) had a history of a prior CNS insult including traumatic brain injury, cerebral malaria, and/or stroke. Almost half of the index seizures were focal, and over a third were multiple or prolonged.

Over half of those enrolled (44, 60%) had WHO stage 4 HIV disease at the time of presentation. Less than half (36, 49%) of children enrolled were on ART. HIV treatment histories for these children are detailed in Table 2. Among those on ART, 19 (56%) had been on treatment for >1 year. Unfortunately, viral suppression (defined as having <200 copies per mL) was evident in only 7 children representing 19% of children on ART and 10% of the overall study population. For children on ART for >3 months without viral suppression, study staff reviewed ART clinic files and interviewed families to determine whether formal investigations for ART failure were warranted. Among children without viral suppression on ART for at least 3 months, adherence was universally very poor. Based upon review of ART clinic records and parental/guardian interviews, all of these children had prolonged and often recurrent periods in which ART were not collected and/or administered. Long-time intervals (median 182 days, mean 383 days) between HIV diagnosis and ART initiation were notable. ART regimen was varied, with 31% of children on efavirenz (EFV).

Table 3 summarizes cause of death within 30 days of the index seizure. A total of 22 (30%) children died, and no children were lost to follow-up. Causes of death identified included advanced HIV, meningitis, sepsis, disseminated TB, and status epilepticus.

Table 4 outlines seizure etiologies stratified by urbanicity. Diagnostic evaluations were particularly limited in the non-urban settings with neuroimaging and EEG only available to urban children. Across all sites, lumbar puncture (LP) uptake was poor despite the assurance of supplies for LP completion and the availability of extensive, relevant diagnostic investigations for study participants. Only a third of children underwent lumbar puncture for CSF diagnostics. Neuroimaging was performed on 39 of 49 urban children, with 18 of 39 getting an MRI, 17 of 39 receiving a CT, and 4 of 39 receiving both. An EEG was obtained in 12 of 49 urban children.

Infectious etiologies were identified in 53% of the population, representing a combination of both opportunistic and non-opportunistic infections. Bacterial meningitis, tuberculous meningitis, and HIV encephalitis ¹ were the most common infectious etiologies identified. Renal failure with frank uremia was present in 6% (3) of children in the urban setting where kidney function could be ascertained relatively easily. Renal failure occurred in the setting of both treated and untreated HIV and among those on treatment was often in the setting of ART regimen with known renal toxicity (primarily tenofovir). Only 1 child had a seizure that was deemed febrile after full data review. Given the dearth of available diagnostic data, more children in the rural setting had an unknown seizure etiology compared with the urban children.

4 DISCUSSION

This study evaluated new-onset seizure in Zambian children during a period in which HIV testing and treatment options were ubiquitous and readily available for those children who had carers willing and able to seek and retain engagement with the healthcare system. Sadly, seizure etiologies and outcomes in this ART era differed little from studies conducted in South Africa before HIV treatment were widely available.⁸ Despite over a decade of widespread ART availability including PMTCT programs, among CLWHIV and new-onset seizures, seizures continue to largely reflect the pathophysiology of long-standing, advanced, and often untreated HIV disease.

This study was not designed to determine whether successful PMTCT and ART outreach programs reduce the burden of HIV-associated seizures in children; however, the absence of any identifiable eligible cases at Nakambala Hospital after 6 months of active screening indirectly support this supposition. HIV care for the Nakambala catchment population includes HIV services embedded within the routine antenatal care provided to all covered women, ART services close to people's residence, time off work for seeking monthly HIV care for oneself or one's child, and active tracking of infants identified to be HIV exposed or infected to assure early treatment. Work in Botswana further supports the role of early ART initiation in preventing seizures and seizure-related morbidity and mortality in CLWHIV.⁷

Interestingly, febrile seizures were determined to be the seizure etiology in only one child despite febrile seizures being one of the most common etiologies for acute seizures in Zambian children.^20-22 Over half the cohort presented with multiple recurrent seizures or status epilepticus. This may reflect delays in accessing care and may also have contributed to the high inpatient mortality. It is also possible that children with a brief single seizure may not have been brought for care if they seemed otherwise well given that febrile seizures are extremely common in these setting and pre-existing barriers in travel and cost. Other work in the region has shown that children with simple febrile seizures are often not brought for care.²⁰ Renal failure was responsible for 3 deaths total and 1 in the rural centers, but this is likely an underestimation as kidney function studies are difficult to acquire at the non-urban sites. Possible reasons for renal failure include ART side effects, underlying HIV disease, and acute kidney injury as a result of rhabdomyolysis from prolonged seizures. Malaria did not play a large role as the etiology for epilepsy in this population. Malaria remains a major cause of death in some regions of Zambia but not in Lusaka or the Southern Province where this study was conducted. Moreover, implementation of programs like the US President's Malaria initiative has contributed to significant reduction in infection rate for children under 5 in the last decade. Thus, a history of severe malaria as expected should be fairly uncommon in this age group and region.²³

Only half the patients in this study were on ART, and the majority of those on ART showed poor viral suppression in the setting of very poor adherence with multiple and prolonged periods of defaulting from treatment. Zambia met the 90:90:90 goals²⁴ as early as 2011. In general, ART enrollment and adherence among children in Zambia is likely much better than was observed in the children in this study who presented with new-onset seizure.²⁵ The most recent reports indicate that 84% of Zambian children with HIV are on ART with 87% of these being virally suppressed.²⁴ This is in significant contrast to the situation of children in this study of new-onset seizure who appear to “slip through the cracks” of HIV care and treatment services—a situation previously described among adults in Zambia.²⁶ Caregivers interviewed reported multiple barriers to ART adherence including socioeconomic factors and household instability particularly in homes that have assumed care of orphaned children. Additional factors included stigma and limited understanding of medication benefits. These reports are congruent with studies of ART adherence^{27, 28} and underscore the imperative to provide extensive support services for these vulnerable populations. Poor outcomes in this population undoubtedly occur due to a number of factors including delayed treatment initiation after diagnosis, poor adherence, lack of appropriate diagnostic resources, and delayed care-seeking for acute illness.

Despite assuring all study sites had the material resources for LP completion and offering extensive diagnostic evaluations if CSF was acquired, LP uptake was quite poor with only a third of children undergoing an LP. The poor uptake of LP in the region has previously been described in PLWHIV with new-onset seizure in Zambia.²⁶ This issue is increasingly recognized and appears to be a multifactorial problem warranting further study.^29-31 In the context of this study, the constrained diagnostic data negatively impacted the identification of seizure etiology and cause of death and may have contributed to the high 30-day mortality. Nonetheless, diagnostic data in this cohort are likely more extensive than are usually available. Imaging requires some user fees unless waived by social work, EEG consumables are often out of stock and indirect costs of bringing children from outside the capital city into the tertiary care centers for diagnostic assessments are prohibitive for most families.

There are a number of other study limitations. These includes the single-country setting, small sample size, and heterogeneity of the study population.^{5, 32} Rural and urban populations differed in terms of their access to imaging and EEG, but also differed in terms of demographic characteristics. Children in the rural areas were accessing care at the ART clinic associated with their residential catchment area. In contrast, the urban cohort were children seeking care at Zambia's largest tertiary care children's hospital and previous work has shown that children accessing care at this site are among the most economically deprived in the city.^{33, 34} Seizure severity burden was high with children presenting many hours after the index seizure onset with fulminant status epilepticus and/or coma. Acute seizure management options are quite constrained in Zambia relative to high-income settings and intubation for ventilation is not de facto available even at the Teaching Hospitals for such cases. A comparative evaluation of quality of care delivery for these children, in terms of acute seizure management, to determine whether care quality contributed to mortality was not possible due to limited documentation even in the context of this prospective study. Certainly, active seizures did not always result in escalation of acute ASMs, but given the need for clinicians to balance risks and benefits with one of the primary risks being death from ASM-induced respiratory failure, this may reflect sound clinical judgment. None of the deaths were attributed to respiratory failure from ASMs, suggesting that the Zambian clinicians caring for these children choose less aggressive seizure management with less optimal seizure control over the risk of an iatrogenic death. The limitations to conduct diagnostic work and research speak to the clinical challenges that remain present. Nonetheless, the findings provide insights into the etiology and burden of seizures among CLWHIV in a large sub-Saharan African country and offers some direction for program development and future research. With fewer than 1 neurologist per million people in Zambia, there are not enough specialists in neurology to be directly involved in the care of children with HIV and seizures, but clear care pathways developed with local resource and expertise in mind could prove extremely valuable.³⁵

Despite widespread HIV testing and treatment availability in Zambia, new-onset seizure in CLWHIV still occurs largely in the setting of advanced, active HIV disease. Enhanced programs to assure early ART initiation, improve adherence and expeditiously identify and address ART failure are needed to reduce the burden of neurological injury and premature death in CLWHIV. These supportive efforts are especially needed among the particularly vulnerable including the poor, rural populations, and orphans.

CONFLICT OF INTEREST

None of the authors have any conflict of interest to disclose.

ETHICAL APPROVAL

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.