Cerebral function in perinatally HIV-infected young adults and their HIV-uninfected sibling controls

Abstract

Background:

Perinatally acquired HIV-infected (PaHIV) young adults undergo neurodevelopment in the presence of HIV infection and antiretroviral therapy, which may lead to neurocognitive (NC) impairment. Knowledge of NC function in this group is sparse and control data lacking. We compared cerebral function in young adults with PaHIV infection to aged matched HIV negative family controls.

Methods:

16–25-year-old PaHIV young adults (Group 1, n = 33) and HIV-uninfected family controls (Group 2, n = 14) were recruited. Cerebral function was evaluated by: a computerized battery assessing NC function (CogState^TM), International HIV Dementia Scale (IHDS) and the prospective and retrospective memory questionnaire (PRMQ). Eight cases and four controls also underwent ¹H cerebral magnetic resonance spectroscopy (¹H-MRS) scanning measuring basal ganglia (BG) metabolites. Cases and controls were compared.

Results:

Group 1 mean (SD) CD4 count; 444 (319) cells/μl, plasma HIV viral load < 50 in 55%. There were no statistically significant differences between study groups in NC function or IHDS results (P>0.27 all observations). PRMQ scores were significantly higher (42 versus 35, P = 0.02) and MRS BG inflammatory-metabolites (choline- and myo-inositol- to creatine ratios) were significantly greater in Group 1 versus Group 2 (0.83 versus 0.63, P = 0.02 and 3.43 versus 3.03.P = 0.09 respectively). No significant association between PRMQ score and MRS metabolites was observed (P = 0.89).

Conclusion:

Statistically significant differences in cerebral function parameters were observed in PaHIV young adults compared to a well-matched control population. The cognitive deficit observed, in memory, rather than fine motor function, differs from the cerebral impairment often reported in HIV-infected adults.

Keywords:

• Neurocognitive function
• HIV
• Perinatal HIV infection
• Adolescents
• Magnetic resonance spectroscopy

Introduction

Since the advent of effective combination antiretroviral therapy (cART) increasing numbers of children born with HIV infection are surviving into adulthood¹ which has resulted in a year on year increase in the number of young adults living with perinatally acquired HIV (PaHIV) in the UK.^2,3 This group represents a new and rapidly growing cohort of patients with complex medical, psychological and social needs. Characterization of PaHIV-infected young adults can only now commence as previously few survived to adulthood and to date, little is known regarding the outcome, complications, needs and challenges arising when caring for such individuals.

In adult-acquired chronic HIV infection, in the era of cART, there has been a dramatic reduction in the incidence of severe HIV associated dementia.⁴ However reports suggest that a milder form of cerebral impairment, known as neurocognitive (NC) impairment^5,6 remain prevalent despite effective cART. This NC impairment has been associated with poorer treatment outcomes.⁷ In adult acquired HIV infection, neuro-development is complete prior to HIV infection whereas in PaHIV, neurodevelopment occurs in the presence of HIV-infection. Therefore, cerebral effects of chronic HIV-infection may differ in perinatally infected individuals where HIV infection has been present during neurodevelopment. Currently, data describing NC function in PaHIV-infected young adults are sparse and appropriate control data lacking.

Measurement of cerebral metabolites via ¹H magnetic resonance spectroscopy (¹H-MRS) is a non-invasive way of detecting both neuronal metabolites and markers of neuronal loss. Changes in cerebral metabolites such as a reduction in concentration of N-acetyl aspartate (NAA) have been associated with neuronal loss, and changes in Choline/Creatine (Chol/Cr) ratios and myo-inositol/Creatine (mI/Cr) ratios have been associated with neuronal inflammation. Within HIV positive adults, abnormalities in metabolite concentrations have been correlated with NC impairment.^8–12

Amongst PaHIV-infected children, there are fewer studies, but correlation between some neurometabolite levels and cognitive performance has been shown; higher NAA/Cho ratio has been associated with better arithmetic and comprehension abilities,¹³ choline concentration in the hippocampus of HIV+ children has been correlated with performance on a delayed spatial memory testing,¹⁴ and attention/processing speed has been associated with right hemisphere NAA/Cr.¹⁵

NC function of these young people may impact on adherence to medication¹⁶ and sexual behavior, and in turn impact on HIV associated morbidity and mortality. With increased survival times, elucidation of the extent and nature of PaHIV associated cognitive deficits in young adults surviving with HIV infection are urgently needed in order to plan for future service provision and to inform treatment decisions for the future in PaHIV positive children.

The aims of this study were to characterize cerebral function and to describe CNS metabolites in young adults with PaHIV infection and compare with HIV negative siblings or close family members of infected young people, as aged matched controls.

Methods

Prior to subject enrolment, the study protocol was approved by The St. Mary's Ethics Committee, London, UK (08/H0712/15) and all participants in the study provided written informed consent.

Patient selection and study procedures

Young persons (aged 16–25 years) with PaHIV, attending St Mary's Hospital, London, UK adolescent HIV clinic, or subjects' part of HYPNet (HIV in young person's network, http://www.hypnet.org.uk/), were eligible for inclusion and assigned as cases (Group 1, HIV-infected) and controls (Group 2, HIV-uninfected age matched family members). Subjects were recruited during attendance at clinic or HYPNet events, and recruitment was open to any eligible young person. Detailed clinical history was taken from all subjects and notes review was also carried out for subjects recruited from St Marys Hospital.

All subjects were proficient in English and Group 2 tested HIV negative and were aware of the family members HIV status. Exclusion criteria for all subjects included current neurological disease. Baseline epidemiological and demographic data were gathered for Groups 1 and 2, including gender, ethnic origin, education, and drug/alcohol use over the past 12 months and past medical history. Clinical data including HIV viral and immunological parameters and cART treatment history were collected for Group 1.

Study procedures included several modalities to assess cerebral function as follows; NC testing using a computerized battery, the International HIV Dementia Scale (IHDS) and the prospective and retrospective memory questionnaire (PRMQ). A subset of participants also underwent ¹H-MRS.

Cognitive testing

Clinical evaluation of NC function was performed using a computerized battery test which has been previously validated in HIV-infected subjects¹⁷ and been shown to correlate well with traditional, formal neuropsychiatric testing in adult HIV positive cohorts. Administration is standardized, timing of responses are recorded with high degrees of accuracy and tests are less time consuming than traditional neuropsychiatry testing, taking only 25 minutes. Computerized cognitive testing (CogState^TM, Melbourne, Australia) was undertaken and involved participants performing eight sequential tasks on a computer, in the form of card games. All participants in Groups 1 and 2 completed one full practice test before undertaking the study examination to obtain optimal performance at baseline.¹⁸ Scores were calculated for global speed, accuracy, and executive function from measurements across the seven domains.

International HIV Dementia Scale (IHDS)

This IHDS is a validated screening tool for HIV associated NC impairment.¹⁹ Subjects are asked to perform 3 tasks, and a total score between 0 (lowest score) and 12 (best score) is obtained. A score of 10 or below has been associated with a diagnosis of HIV associated dementia and warrants further assessment of NC function.¹⁹

Prospective and retrospective memory questionnaire (PRMQ)

The PRMQ²⁰ was administered to all subjects. This tool provides a self-reported measurement of prospective and retrospective memory and consists of 16 items, eight asking about prospective memory failures, and eight concerning retrospective failures; the higher the score the more self-reported memory problems are present.²⁰

Cerebral ¹H magnetic resonance spectroscopy

Cerebral proton ¹H-MRS was performed on an Achieva^TM 1.5 Tesla scanner (Phillips, MR Systems, Best, the Netherlands) at the Robert Steiner Magnetic Resonance Unit, Hammersmith Hospital, London, UK. Examination began with sagittal, coronal and axial tbl1-weighted MR images of the brain to enable accurate positioning of the voxels, and tbl2-weighted axial double spin echo images to exclude any visible cerebral pathology. ¹H-MRS was then performed in three anatomical voxel locations: right frontal white matter, mid-frontal gray matter and the right basal ganglia (RBG), details of which have previously been described.^21,22 In brief, MRS data were acquired by single voxel examination in the three areas using a double spin echo point resolved spectroscopy sequence with the following settings: echo time, 36 ms; repetition time, 3000 ms; 2048 data points; spectral width of 2500 Hz and 128 data acquisitions. Spectra were post-processed using the MRI machine manufacturer's software for automated water signal suppression and water shimming. Each examination lasted approximately 35 minutes. The magnetic resonance images were studied and reported by an experienced neuroradiologist. All spectra were analyzed and quantified by one observer (AW) using a Java-based version of the magnetic resonance user interface package (jMRUI Version Number: 3.0),²³ incorporating the AMARES algorithm²⁴ and expressing metabolites as ratios to cerebral creatine (Cr).

Statistical analysis

All statistical calculations were performed using SPSS (version 17.0; SPSS Inc., Chicago, IL, USA).

Global scores for NC speed, accuracy and executive function were calculated. Between-group differences in these NC scores were evaluated in a univariate model by linear regression (study Group 1 versus 2). Within group differences for Group 1 were also investigated to look for associations between clinical factors and NC scores in univariate models; potential factors included patient baseline characteristics such as age, current CD4+ lymphocyte count, plasma HIV viral load and length of time on cART. Similarly between and within group differences for IHDS and PRMQ were also evaluated in a univariate model by linear regression and associations with clinical and other study findings investigated.

Results

Patient characteristics

Thirty-three and 14 subjects were recruited in Groups 1 and 2, respectively with baseline characteristics shown in Table 1. HIV clinical parameters for Group 1 are shown in Table 2. Groups 1 and 2 were well matched in terms of age, ethnicity and education. Seventy-nine per cent of Group 1 were on cART and 55% had an HIV viral load < 50 copies/ml. Mean (IQR) CD4+ lymphocyte count was 444 (174–725) cells/μl (21%).

Table 1. Baseline characteristics

Parameter	Group 1 (PaHIV-infected group)	Group 2 (Control group)
Number of subjects	33	14
Age, years (mean, range)	19.6, 17–23	19.9, 16–24
Black/mixed ethnicity %	85	86
Male gender, n (%)	11 (33)	4 (29)
Recent recreational drug use (%)	2 (6)	1 (7)
Previous recreational drug use (%)	13 (39)	6 (43)
English is first language (%)	29 (88)	13 (93)
Number of years education (years)	14	15

Note: mean values unless otherwise stated.

Table 2. Group 1 HIV clinical parameter

Parameter	Group 1
Current plasma CD4 count (cells/μl) (IQR)	444 (174–725)
Current plasma CD4 cell % (IQR)	21 (9–35)
HIV VL < 50 copies/ml, n (%)	18 (55)
Age at HIV diagnosis, years (IQR)	5 (0–9)
Years since HIV diagnosis (IQR)	15 (13–20)
Currently taking ARV, n (%)	26 (79)
PI containing regimen, n (%)	18 (55)
Age at ARV commencement, years (IQR)	13 (8–16)
Years since first ARV treatment (IQR)	8.5 (4–13)

Note: mean values unless otherwise stated. IQR = interquartile range, VL = viral load, ARV = antiretroviral therapy, PI = protease inhibitor.

Neurocognitive, PRMQ and IHDS results

The results of the study assessments are shown in Table 3. There were no significant differences in NC scores or IHDS scores between Group 1 and Group 2 for all parameters measured. Statistically significant higher PRMQ scores were observed in Group 1 (mean score (SD) 42 (8.2)) compared to Group 2 (mean score (SD) 35 (9.0), P = 0.023), with a higher score representing poorer response.

Table 3. Results of neurocognitive testing

	Mean scores (standard deviation)
Domain	Total score (n = 47)	Group 1 (n = 34)	Group 2 (n = 14)	P value
Global speed score	10.64 (0.28)	10.66 (0.30)	10.57 (0.24)	0.27
Global accuracy score**	3.02 (0.49)	3.03 (0.47)	2.99 (0.42)	0.78
Executive function	17.83 (8.93)	18.18 (9.79)	17.00 (6.75)	0.68
IHDS	11.3 (0.9)	11.3 (0.9)	11.3 (0.8)	0.861
PRMQ	40 (8.9)	42 (8.2)	35 (9.0)	0.023

Note: IHDS = International HIV Dementia Scale, PRMQ = prospective and retrospective memory questionnaire, *P value for between group differences.

* Speed assessed in MSES and executive function in total errors. The lower the score achieved (i.e. faster speed or lower number of errors) the better the performance

** Accuracy assessed as number of correct responses where a greater score represents better performance.

Within Group 1, no statistically significant differences in testing parameters were noted between those subjects with and without plasma HIV RNA 50 copies/ml (P>0.1 all observations) – although numbers were small.

Cerebral metabolite ratios

Cerebral metabolite ratios for Groups 1 and 2 are shown in Table 4. No statistically significant differences were observed in cerebral metabolite ratios between Groups 1 and 2 in the frontal gray matter or frontal white matter (Table 4). In the RBG, choline/creatine (cho/Cr) and myo-inositol/creatine (mI/Cr) ratios were significantly higher in Group 1 compared to Group 2 (0.83 versus 0.63, P = 0.02 and 3.43 versus 3.03. P = 0.09 respectively). There were no significant differences in the ratio of N-acetyl aspartate/creatine (NAA/Cr) within the RBG between Group 1 and Group 2. Lastly, there were no associations between cerebral metabolite ratios and PRMQ scores (P = 0.86)

Table 4. Cerebral metabolite ratios

	Mean (standard deviation)
		Total score	Group 1 (n = 8)	Group 2 (n = 4)	P value
Frontal gray matter	NAA/Cr	1.66 (0.28)	1.75 (0.31)	1.49 (0.07)	0.14
	Cho/Cr	0.63 (0.10)	0.65 (0.12)	0.59 (0.06)	0.36
	mI/Cr	3.59 (0.97)	3.80 (1.03)	3.15 (0.76)	0.29
Frontal white matter	NAA/Cr	1.88 (0.86)	2.04 (1.03)	1.54 (0.15)	0.37
	Cho/Cr	0.92 (0.18)	0.97 (0.20)	0.81 (0.09)	0.17
	mI/Cr	4.15 (0.68)	4.17 (0.71)	4.12 (0.70)	0.92
Basal ganglia	NAA/Cr	2.01 (0.42)	2.13 (0.46)	1.77 (0.16)	0.17
	Cho/Cr	0.76 (0.15)	0.83 (0.14)	0.63 (0.03)	0.02
	mI/Cr	3.30 (0.39)	3.43 (0.35)	3.03 (0.34)	0.09

Note: Cr = creatine; Chol = choline; mI = myo-inositol; NAA = N-acetyl aspartate,

* P value for between group differences (P < 0.05 shown in bold).

Discussion

This study represents the first comparison of ¹H-MRS and NC assessment between PaHIV young adults and matched uninfected controls. Furthermore, in addition to NC testing results, biomarker results in the form of CMR are also available for some of the subjects recruited. In this cohort of PaHIV-infected young people, we observed statistically significant greater inflammatory cerebral metabolite factors in the right BG compared to age matched sibling controls, namely greater cho/Cr and mI/Cr ratios. Differences in cerebral metabolites seen on ¹H-MRS have been demonstrated in several chronic viral infections including HIV, and are thought to reflect neuropathology.^10,12,25 Myo-inositol is an osmo-sensitive glial marker and plays a crucial role in cell volume regulation²⁶ and increased concentrations are thought to indicate cell shrinkage within that region of the brain. The greater cho/Cr and mI/Cr ratios observed in this study suggests an ongoing inflammatory process and we postulate that this could reflect ongoing immune activation within the CNS. Our results contrast with findings from an early study conducted by Lu et al. before widespread availability of cART, where lower Cho/Cr in the basal ganglia of HIV-infected children without a diagnosis of encephalopathy was observed compared to healthy aged matched controls.²⁷

In contrast, the neuronal metabolite ratios we have assessed, namely NAA/Cr ratios, were not significantly different between Groups 1 and 2. These findings differ from a previous early study performed before cART usage was widespread, which showed subjects with encepholpathy had significantly lower mean NAA/Cr ratios compared with age-matched controls.²⁷ This may reflect effective HIV virological control observed in 5/8 subjects in this group and the resulting maintenance of neuronal integrity.²¹ These findings suggest that whilst inflammation is observed in this area of the brain, neuronal integrity may not be a significantly disrupted in these young people at this stage in their life. The small numbers of cases and controls enrolled in this sub study limit the interpretation of these findings. The potential for a selection bias in the MRS sub study may limit interpretation of results, as those young people best able to plan ahead may have been more likely to volunteer and attend for MRS screening. Future studies require larger groups to further characterize this cohort.

We also observed a statistically significant difference in PRMQ scores between cases and controls with cases (Group 1) self-reporting more memory difficulties than age-matched controls. Poorer PRMQ scores have been shown to correlate with executive function deficits in HIV positive adults²⁸ and prospective memory failures in particular may have a greater impact on individuals and may be an early indicator of disease.²⁹ It is important to note that the PRMQ score is a subjective measure of memory and as such, may be subject to influences such as low self-esteem and stigma, which may be prevalent amongst PaHIV adolescents.

There were no statistically significant differences in IHDS scores between groups. The IHDS has been shown to be 80% sensitive and 55% specific at detecting HIV associated dementia in horizontally infected HIV positive adults;¹⁹ however, it has not been assessed as a tool for detecting impairment in PaHIV subjects and in our cohort no differences were observed between study groups.

Interestingly, no differences in scores for NC testing when using a computerized battery were observed between study groups. This is in contrast to previous pilot data where PaHIV-infected young people have been compared to nearest age matched HIV negative population controls with high rates of impairment reported.³⁰ These differences could in part be explained by lack of appropriate controls used in previous studies. For instance many of the PaHIV positive young adults have lost one or both parents to HIV related disease and are immigrant families from sub-Saharan Africa, often with English as a second language and with complex social situations.^31,32 As a result, control data that takes into account this bias introduced by social and genetic factors are essential when the biological effect of HIV on NC function in these young people is being evaluated. In this study the inclusion of siblings and close family members as controls was a useful strategy to control for any potential bias exerted by social and genetic influences. An alternative explanation for the lack of difference observed between cases and controls is that the computerized NC battery used in this study did not capture the areas of deficit present in this cohort.

Because many of the subjects were born in Africa and subsequently came to the UK during childhood, historical data wsere not available for detailed review of all subjects and so previous biologic markers, treatment and diagnoses such as encephalopathy could not always be confirmed. Additionally the influence of HIV clade on NC functioning in this group of young people may also be important given the potential effect on clinical course³³ and would benefit from further study.

The group of PaHIV-infected young people enrolled in this study are a unique cohort who have all survived as young children in an era before cART was available, and as such may be atypical and subject to a “survivor phenomena”; and hence may not include the potential impact of cART on NC development. Despite the late introduction of cART treatment in this group, markers for neuronal injury were not found in this study, nor were correlations to measures of NC functioning. Continued monitoring of NC function in future PaHIV-infected individuals surviving into adulthood is therefore essential in order to elucidate any effect of cART during early neurodevelopment.

This is the first series to report in detail the NC and MRS findings in a cohort of PaHIV-infected young adults and compare them to age matched family controls. The numbers are small and findings must be interpreted with caution. Better understanding of the neuro-developmental sequelae of PaHIV infection will contribute vitally when considering issues such as competency and consent. Given that individuals with PaHIV have a significantly improved survival the long-term consequences of uncontrolled HIV virus at the time of neuronal development must be balanced against the risks of drug induced toxicities and development of drug resistance through teenage years. Future service provision relies upon further characterization of NC function in this cohort and studies are needed to evaluate the effects of cART within this cohort.

Disclaimer Statements

Contributors

Jane Ashby, Caroline Foster, Lucy Garvey, Sarah Fidler and Alan Winston: research, analysis, manuscript preparation. Tania Wan, Joanna Allsop, Yasotharan Paramesparan and Simon D Taylor-Robinson: research.

Funding

This study was partly supported by a research grant from the British HIV Association. We are grateful for support from the NIHR Biomedical Research Centre funding scheme at Imperial College Healthcare NHS Trust, London, UK for infrastructure funding support.

Conflicts of interest

None.

Ethics approval

Granted.