Neurocognitive functioning in MDR-TB patients with and without HIV in KwaZulu-Natal, South Africa

Summary

OBJECTIVES

Optimising medication adherence is one of the essential factors in reversing the tide of a TB-HIV syndemic in sub-Saharan Africa, especially South Africa. Impairment in key neurocognitive domains may impair patients’ ability to maintain adherence to treatment, but the level of cognition and its relationship to HIV status has not been examined in individuals with drug-resistant TB. We therefore investigated performance on several key neurocognitive domains in relationship to HIV status in a multidrug-resistant tuberculosis patients (MDR-TB) sample.

METHODS

We enrolled microbiologically confirmed MDR-TB inpatients at a TB-specialist referral hospital in KwaZulu-Natal province, South Africa. We collected cross-sectional data on sociodemographic, clinical and neurocognitive function (e.g. attention, memory, executive functioning, language fluency, visual-spatial, eye–hand coordination). For the primary analysis, we excluded participants with major depressive episode/substance use disorder (MDE/SUD). We fitted adjusted Poisson regression models to explore the association between HIV and neurocognitive function.

RESULTS

We enrolled 200 people with MDR-TB; 33 had MDE/SUD, and data of 167 were analysed (151 HIV+, 16 HIV−). The mean age of participants was 34.2 years; the majority were female (83%), and 53% had not completed secondary school. There was evidence of impaired neurocognitive functioning across all domains in both HIV+/− study participants. Based on the regression analyses, individuals with co-infection (MDR-TB/HIV+), as well as those who had longer duration of hospital stays experienced significantly lower cognitive performance in several domains. Poor cognitive performance was significantly related to older age and lower educational attainment. The presence of major depression or substance use disorders did not influence the significance of the findings.

CONCLUSIONS

Adults with MDR-TB have significant neurocognitive impairment, especially if HIV positive. An integrated approach is necessary in the management of MDR-TB as cognitive health influences the ability to adhere to chronic treatment, clinical outcomes and functionality.

Introduction

Although the incidence of tuberculosis (TB) has begun to decline in South Africa [1], the disease continues to be the leading cause of mortality [2] and years lived with disability [3] in the country. The TB epidemic is driven by the high prevalence of human immunodeficiency virus (HIV), and most people with TB are HIV positive. Multidrug-resistant tuberculosis (MDR-TB), caused by Mycobacterium tuberculosis strains resistant to rifampicin and isoniazid, creates particular public health challenges. Treatment is complex and lengthy with a high pill burden, especially when combined with antiretroviral therapy (ART). As a result, outcomes are poor, to the extent that only around half of all people with MDR-TB are successfully treated. This leads not only to the amplification of drug resistance but also to the ongoing transmission of drug-resistant TB in the community.

Improving outcomes for MDR-TB and reducing transmission requires high levels of retention and adherence to anti-TB drug therapy and to ART. While wide ranging factors can influence medication adherence [4], neurocognitive functions, such as attention, concentration, memory and executive functions, are important for medication adherence. Although neurocognitive impairment is well described in HIV infection, there is much less understanding of whether and how TB, and specifically MDR-TB, might affect neurocognitive function. Improving our understanding of this is a first step in developing better adherence and retention support strategies for people with MDR-TB. In this study, we characterise neurocognitive function across a range of domains in HIV-positive and HIV-negative adults with MDR-TB and explore the association with HIV infection.

Methods

Study design and participants

We enrolled inpatients with culture-confirmed MDR-TB (i.e. pulmonary or extra pulmonary) admitted to a specialised TB hospital in KwaZulu-Natal (KZN), South Africa, from September 2015 through October 2016. The inclusion criteria were consenting isiZulu- or English-speaking individuals aged between 21 and 59 years who were clinically stable to participate in assessments. The exclusion criteria were individuals with a previous inpatient admission to the study site, and/or prior receipt of MDR-TB treatment outside that study site; this was to limit possible confounding factors due to greater length and severity of illness and longer exposure to anti-TB treatment. Those who had not completed primary school level of education were not eligible as this is a minimum requirement for completion of the cognitive assessment tools. Patients with clinical diagnoses of TB meningitis or acute psychiatric disorders were not eligible. We obtained written informed consent from all participants. The University of KwaZulu-Natal Biomedical Research Ethics Committee (ref. BF251/14) approved the study protocol.

Measurement

We conducted face-to-face structured interviews and administered tests in English or isiZulu. A Masters-level clinical psychologist administered a sociodemographic questionnaire, a battery of five neuropsychological tests and a mental health diagnostic interview, under the supervision of a senior clinical psychologist, both of whom were blinded to the HIV status of the participant. We obtained information on participants’ demographic and socioeconomic status and their laboratory investigations. The five neuropsychological tests assessed six cognitive domains and, while there were no previously validated norms for the South African population available for these tests, their selection was based on the extensive clinical experience of one of the authors (TN) using these tests successfully in this population. The social cognition domain was not assessed as local use of existing tools raised questions about their validity in this population.

Digit span test

The Digit Span Test (DST) is a measure of attention and short-term memory [27], which requires participants to repeat a sequence of numbers presented verbally (i.e. digits forward) by the interviewer, and to then listen to a sequence of numbers and repeat them in reverse order (i.e. digits backward). In both parts (digits backward/forward), the length of each sequence of numbers presented increases as the subject responds correctly. The total score ranges from 0 to 30, with higher scores indicating better performance.

Trail making test

The Trail Making Test (TMT) is an assessment of attention, speed, visual-motor tracking and mental flexibility [5,6]. In part A (TMT A), participants are asked to connect consecutively numbered circles on a sheet. In Part B (TMT B), participants connect consecutively numbered and lettered circles alternately. The score is expressed as the time in seconds taken to complete the tasks, with lower scores indicating better performance.

Animal naming test

The Animal Naming Test (ANT), a measure of verbal fluency, requires subjects to name as many animals as they can in one minute, with higher numbers indicating better performance. The total number of animals named less than 14 is used for detection of mild dementia [7].

Rey complex figure

The Rey Complex Figure (RCF) is a test of visuo-constructive skill, visual memory, attention and planning [5]. The participant is first asked to copy a complex figure presented to him/her and is expected to reproduce the figure from memory immediately, with a higher score indicating better recall. The range of minimum and maximum score is 0-36.

Grooved pegboard test

The Grooved Pegboard Test (GPT) is a test of manual dexterity consisting of 25 holes with randomly positioned slots [6]. Pegs with a key along one side must be rotated to match the hole before they can be inserted, requiring complex visual-motor coordination. The score is the time (in seconds) required to place the peg, with higher scores indicating poorer performance.

Major depression episode and substance use disorder

Major depressive episode (MDE) and substance use disorder (SUD) encompassing substance abuse and dependence were diagnosed using the Mini International Neuropsychiatric Interview version 6.0 (MINI 6.0), a structured diagnostic schedule utilising DSM-IV criteria.

Data analysis

The analysis consisted of three components. First, descriptive statistics were used to summarise sociodemographic, clinical and neurocognitive function. Second, differences in the median neurocognitive function scores in MDR-TB participants with and without HIV were compared using non-parametric methods (i.e. Wilcoxon rank sum test) given the expected small number of HIV-negative cases in the MDR-TB sample. Third, regression analyses were performed to identify significant predictors of neurocognitive function scores. The regression was fitted to investigate the association between HIV status and neurocognitive function controlling for sociodemographic and clinical characteristics. As a crude indicator of severity of disease, we also controlled for length of hospital inpatient stay which was defined as the difference between the date of assessment and hospital admission. We did not have access to the information pertaining to the actual date of hospital discharge. Poisson models were fitted owing to positive skewedness of the dependent variables in the models. To assess the sensitivity of the findings, the three abovementioned analyses were undertaken both before and after major depression episode and substance use disorder (MDE/SUD) exclusion to better tease out the role of HIV on neurocognitive function, consistent with recent South African studies [8,9]. The data were analysed using STATA version 15 (StataCorp, College Station, TX), with their significance set at P < 0.05.

Results

Sociodemographic and clinical profiles

We enrolled 200 people with MDR-TB. After excluding 33 with MDE/SUD, our primary analyses included data from 167 participants, of whom 151 (90.4%) were HIV positive. The mean age was 34.2 years (SD = 7.2) with most of the participants (n = 89; 53.3%) not having completed secondary-level education. The complete sociodemographic characteristics are presented in Table 1.

Table 1.

Background of characteristics of MDR-TB study participants

	Total (N = 200)		Excluding SUD/MDE (n = 167)
	N	%	n	%
Gender
Female	161	80.5	139	83.2
Male	39	19.5	28	16.8
Age category
21–29	65	32.5	55	32.9
30–39	87	43.5	69	41.3
40–55	48	24.0	43	25.8
Race/Ethnicity
Black	196	98.0	164	98.2
Coloured/Indian	4	2.0	3	1.8
Education
<Grade 12	109	54.5	89	53.3
≥Grade 12	91	45.5	78	46.7
Marital status
Married/Stable partnership	113	56.5	95	56.9
Casual partnership	20	10.0	13	7.8
No relationship/partnership	67	33.5	59	35.3
Income per month
≤R1000	119	59.5	98	58.7
R1001–R5000	58	29.0	48	28.7
≥R5001	23	11.5	21	12.6
HIV status
No	21	10.5	16	9.6
Yes	179	89.5	151	90.4

Neuropsychological performance

The overall median neuropsychological performance scores are presented in Table 2. The bivariate analysis of comparison in neuropsychological performance scores (Table 3) indicated lower functioning on the TMT B (i.e. for visual attention and task switching; executive functioning) and RCF recall (i.e. visuo-spatial; executive functioning and organisation) in HIV-positive participants compared to HIV-negative participants.

Table 2.

Summary of neurocognitive functioning score

Tests	Purpose	All Groups			After Exclusion
		n	Median	IQR	n	Median	IQR
Digit Span Test - Forward (↑)	Attention and short- term memory	200	8.5	3	167	9	3
Digit Span Test - Backward (↑)	Attention and short- term memory	200	5	2	167	5	2
Trail Making Test - Part A (↓)	Visual attention and task switching; executive functioning	199	43	31	166	43	30
Trail Making Test - Part B (↓)	Visual attention and task switching; executive functioning	198	109	69	166	110	73
Animal Naming Test (↑)	Visual attention and task switching; executive functioning	200	13	3.5	167	13	4
Rey Complex Figure – Copy (↑)	Visuo-spatial; executive functioning (organisation)	200	29	12	167	29	10
Rey Complex Figure – Recall (↑)	Visuo-spatial; executive functioning (organisation)	200	11	11	167	12	12
Grooved Pegboard Test - Right (↓)	Eye–hand coordination and motor speed	195	77	24	164	77	22.5
Grooved Pegboard Test - Left (↓)	Eye–hand coordination and motor speed	195	84	28	164	84	27.5

Arrow indicates general directionality of better function.

Table 3.

Neurocognitive functioning score by HIV status (before and after exclusion of MDD/SUD)

Tests	MDR-TB and HIV		MDR-TB without HIV		Median difference statistics (rank sum) Z, p
	Median	IQR	Median	IQR
Before exclusion
Digit Span Test - Forward (↑)	8	3	9	3	1.38, 0.17
Digit Span Test - Backward (↑)	5	2	6	1	1.67, 0.09
Trail Making Test - Part A (↓)	44.5	31	35	17	2.96, <0.01
Trail Making Test - Part B (↓)	110	69	85	63	2.32, 0.02
Animal Naming Test (↑)	12	3	14	4	−0.86, 0.39
Rey Complex Figure – Copy (↑)	29	14	31	5	−1.62, 0.11
Rey Complex Figure – Recall (↑)	10	11	16	10	−2.97, <0.01
Grooved Pegboard Test - Right (↓)	77	24	70	24	1.95, 0.05
Grooved Pegboard Test - Left (↓)	84.5	29	81	19	1.23, 0.22
After exclusion
Digit Span Test - Forward (↑)	9	3	8.5	3	0.50, 0.62
Digit Span Test - Backward (↑)	5	2	6	2	2.17, 0.03
Trail Making Test Part A (↓)	44	31	32.5	21	2.60, <0.01
Trail Making Test Part B (↓)	116.5	71	78	53	2.63, <0.01
Animal Naming Test (↑)	13	4	14	5	−0.01, 0.99
Rey Complex Figure – Copy (↑)	29	13	31.5	5	−1.07, 0.28
Rey Complex Figure – Recall (↑)	11	12	17	9	−2.95, <0.01
Grooved Pegboard Test - Right (↓)	77	23	73	27	1.53, 0.12
Grooved Pegboard Test - Left (↓)	84	28	81.5	33	0.89, 0.38

Arrow indicates general directionality of better function.

Regression analysis

The adjusted regression analysis yielded three main results (Table 4). First, HIV co-infection in MDR-TB patients was associated with significantly lower cognitive performance in four domains, namely TMT A (adj β = 0.40, P < 0.01), TMT B (adj β = 0.35, P < 0.01), RCF Recall (adj β = −0.37, P < 0.01) and GPT Right (adj β = 0.03, P = 0.04). Second, longer hospital stay was significantly associated with lower cognitive outcomes in five domains, namely TMT A (adj β = 0.01 P < 0.01), TMT B (adj β = 0.35, P < 0.01), RCF Recall (adj β = −0.01, P < 0.01), GPT Right (adj β = 0.01, P < 0.01) and GPT Left (adj β = 0.01, P < 0.01). Lastly, poor cognitive performance was significantly related to older age, and lower educational attainment. The significance of the findings did not change when MDE/SUD samples were included. Males performed significantly better on the Trails A and B, RCF recall and the GPB-left.

Table 4.

Regression analyses (after exclusion of MDD/SUD)

		Digit Span Test - Forward (↑)		Digit Span Test – Backward (↑)		Trail Making Test Part A (↓)		Trail Making Test Part B (↓)		Animal Naming (↑)		Rey Complex Figure – Copy (↑)		Rey Complex Figure – Recall (↑)		Grooved pegboard Right (↓)		Grooved pegboard Left (↓)
		B	SE	B	SE	B	SE	B	SE	B	SE	B	SE	B	SE	B	SE	B	SE
Gender	[Female]
	Male	0.05	0.07	−0.04	0.1	0.12***	0.03	0.11***	0.02	0.01	0.06	0.07	0.04	0.17**	0.06	0.03	0.02	0.09***	0.02
Age category	[21–29]
	30–39	−0.02	0.07	0.08	0.09	0.12***	0.03	0.07***	0.02	0.02	0.05	−0.05	0.04	−0.07	0.06	0.06*	0.02	0.04*	0.02
	40–55	−0.12	0.08	−0.02	0.11	0.40***	0.03	0.37***	0.02	−0.01	0.07	−0.06	0.05	−0.21**	0.07	0.26***	0.03	0.27***	0.02
Education	[>Grade 12]
	≥Grade 12	0.09	0.06	0.20*	0.08	−0.22***	0.03	−0.28***	0.02	0.10*	0.05	0.12**	0.03	0.12*	0.05	−0.07***	0.02	−0.05**	0.02
Marital status	[No relationship/partnership]
	Married/Stable partnership	−0.02	0.06	−0.06	0.08	−0.01	0.02	−0.10***	0.02	0.01	0.05	0.02	0.03	−0.07	0.05	−0.08***	0.02	−0.09***	0.02
	Casual partnership	−0.16	0.11	−0.11	0.14	0.05	0.04	−0.17***	0.03	0.06	0.09	−0.06	0.06	0.04	0.09	−0.05	0.04	−0.07*	0.03
Income per	[R1001–R5000]
month	≤R1000†	−0.05	0.06	0.11	0.09	−0.05	0.02	−0.20***	0.02	0.03	0.05	0.05	0.04	−0.07	0.05	0.01	0.02	0.01	0.02
	≥R5001	0.12	0.10	0.04	0.13	0.15***	0.04	−0.27***	0.03	0.04	0.08	−0.05	0.06	−0.18*	0.09	−0.05	0.03	−0.04	0.03
Infection Status	[Mono-infection]
	Co-infection	0.01	0.09	−0.15	0.12	0.40***	0.04	0.35***	0.03	0.05	0.08	−0.08	0.05	−0.37***	0.07	0.06*	0.03	0.03	0.03
Length of inpatient stay	Days	−0.01	<0.01	−0.01	<0.01	0.01***	0.01	0.01***	0.01	−0.01	<0.01	−0.01	<0.01	−0.01***	<0.01	0.01***	<0.01	0.01***	<0.01

Reference category in bracket.

Arrow indicates general directionality of better function indicators.

^*P < 0.05

^**P < 0.01

^***P < 0.001.

Discussion

Treatment outcomes for adults with multidrug-resistant tuberculosis in low- and middle-income settings are poor, especially in the context of HIV co-infection. In this study from KwaZulu-Natal, South Africa, we document evidence of neurocognitive impairment in a group of adults with MDR-TB that may have implications for treatment adherence. Four factors were significantly associated with poorer cognitive function: HIV co-infection, longer duration of hospital stay, lower education level and older age. To our knowledge, this is the first report of such a comprehensive assessment of neurocognitive function in adults with MDR-TB in Africa. Employing a range of cognitive assessment tools, we also demonstrate that (1) visual attention and task switching based on the Trail Making Test parts A and B, (2) visuo-spatial and executive functioning for planning and organisation based on Rey Complex Figure recall and (3) complex visual-motor coordination based on Grooved Pegboard Test were significantly worse in HIV co-infected individuals even after controlling for age and levels of education.

Normative data exist for South African adults for all the tests used in our evaluation, although for some tests a small sample size was used to derive the normative scores [8,10-12]. The most marked impairment in our group seemed to be in hand-eye coordination and motor speed, as reflected in a median score on the grooved pegboard test that was substantially higher than healthy HIV-negative adults in four provinces of SA, and even higher than the mean score of asymptomatic HIV-positive adults [12]. In fact, the scores matched most closely with those for people with evidence of HIV-associated neurocognitive disorder (HAND) [12]. While the scores on the trail making test part A were similar to those for a population of healthy HIV-negative adults in KwaZulu-Natal, the scores for our study population for part B seemed to be worse [8]. This might suggest a subtle deficit in visual attention and executive functioning. Otherwise, the DST scores were similar to those for HIV-negative adults aged 18–50 years from KwaZulu-Natal [8], RCF scores were similar to those from a sample of unskilled workers from Eastern Cape, although HIV status was not ascertained in that study [11] and the ANT scores were similar to those from a group of HIV-positive and HIV-negative adults aged 19–34 years in the Western Cape [10].

More than half (53.3%) of our participants had ≤ 12 years of education; among asymptomatic HIV-positive subjects, individuals with ≤ 12 years of education have been shown to have a significantly higher prevalence of neuropsychological impairment, an association not evident in seronegative subjects [13]. Education is an important contributor to cognitive reserve. TB and HIV are chronic life-threatening illnesses that impact the brain directly through infection of brain tissue, and indirectly through a vicious cycle of poverty, malnutrition, compromised lifestyle, psychiatric co-morbidity, stress, maladaptive coping mechanisms and high risk behaviours [14]. These factors, together with low levels of formal education, not only increase vulnerability to infections, but contribute to low cognitive reserve. TB-HIV infections are closely linked to poverty and low education level.

Old age also adversely affects cognitive reserve; our sample comprised individuals < 60 years of age yet neurocognitive performance was poorer with increasing age. With increased longevity being conferred by the use of ART in HIV-positive individuals, further age-related cognitive decline can be anticipated. HIV-positive subjects with low cognitive reserve have been shown to have significantly greater impairment in attention and information processing speed, verbal learning and memory, as well as executive functioning and visuo-spatial functioning [15]. Advancing age in a Zambian HIV-positive sample was however not found to be associated with poorer cognitive performance,rather the younger HIV-positive group with less education performed worse on learning and recall than the younger HIV-positive group with more education. Further, the motor functioning scores on the GPT were normal across all education and age groups [16]. However, caution is required in interpreting their results as their ‘older’ group was defined at age 47 years compared to studies in general defining ‘old’ as above age 60. Apart from education effects, the contribution to cognitive reserve by education, age-effects and the nature and extent of cognitive stimulation from occupational activities all impact on performance scores on cognitive tests. The extent to which TB may have impacted on the results in our study are not known; poverty, low education, poor cognitive reserve, risk for TB-HIV co-/infection, increased vulnerability to neurocognitive impairment and reduced health compliance represent a vicious cycle of factors that impact outcomes in chronic infectious diseases. As successful treatment of TB-HIV enables longevity, optimum cognitive functioning is required to ensure health compliance and functionality.

There is a paucity of evidence as to whether and to what extent TB (other than diseases affecting the central nervous system) affects neurocognitive function. In one study from Zambia, NCI was more frequently observed among HIV-positive adults with pulmonary TB than in the HIV-positive adults without pulmonary TB [17]. This association was maintained after adjustment for age and education, and the finding was consistent across a range of cognitive domains. In contrast, in the study from a KwaZulu-Natal HIV clinic, active TB was not independently associated with NCI [18]. Social co-morbidities co-exist with physical and psychiatric co-morbidities and are difficult to modify as both TB and HIV are prevalent in under-resourced and economically marginalised communities [26]. Given the high local burden of both TB and HIV, further studies are necessary to define the impact of TB-HIV co-infection on cognition.

HIV co-infection was associated with worse performance on all tests with few exceptions; after adjustment for age and education, there was strong evidence of association between HIV and cognitive impairment on the recall component of the Rey Complex Figure test. Multiple neuropsychological domains, including psychomotor skills, executive functioning, attention, information processing speed and memory, are impacted upon by HIV [22,23,24]. While qualitative and quantitative differences in neurocognitive impairment increase with the stage of HIV disease, the changing profiles of the deficits suggest alternate intracerebral pathogenetic mechanisms, and implicate subcortical and fronto-striatal pathways [19]. Neuropsychological deficits remain common, although diminished, in the ART era, even in patients with undetectable plasma viral load, with some neuropsychological functions improving (attention, verbal fluency, visuo-construction deficits) while others deteriorate (learning efficiency, complex attention). The role of potential antiretroviral neurotoxicity, resistance to ART, disease chronicity and co-morbidities may be explanations for the continued presence of cognitive impairment, albeit attenuated in severity and profile of impaired domains post-ART [20]. The CNS penetration of ART may also impact on treatment response, with good penetration being associated with poorer neurocognitive performance and greater penetration impacting on CSF viral replication [21], thus confounding the interpretation of treatment effects.

Study limitations

Interpretation of our findings should be considered in light of a number of limitations. We recruited participants from the specialist TB referral hospital for KwaZulu-Natal Province. In general, this facility accepts referral of more complex cases and people with significant co-morbidities. Therefore, we cannot be certain that our participants are representative of all people with MDR-TB in the province. Due to incomplete clinical records, the exact details of exposure to specific anti-TB and antiretroviral drugs could not be captured for all patients. Most of these drugs are known to have cognitive and or psychiatric adverse effects which can influence cognitive performance. The possible contribution of these agents to the cognitive status of patients could therefore not be factored into our analyses. We also did not perform detailed clinical assessments over and above what was done for routine care, so we cannot exclude the possibility that some participants had subclinical central nervous system involvement with TB, HIV or other HIV-associated conditions. CD4 counts were not uniformly available, precluding an analysis of the relationship between immune status and cognitive impairment. We performed the cognitive assessment at a single time point only so any assessment of how TB treatment and antiretroviral therapy influenced cognitive function at an individual level was precluded. Composite scores of overall cognitive performance or domains were not computed as we felt it more useful to report on individual domain scores, especially in the absence of local norms for this clinical population. Finally, we did not include control groups of HIV-positive or HIV-negative adults without MDR-TB, therefore it is difficult to separate out the independent effects of HIV and TB on cognitive function. Notwithstanding these limitations, the main strength of this study is the detailed neurocognitive assessment of a group of people with well-defined clinical phenotypes relating to MDR-TB and HIV. Although the clinical complexity of MDR-TB and HIV co-infection, and the potential effects of their treatment, create difficulties for the reliable assessment of neurocognitive function, this complexity reflects the real-world situation for this challenging patient population. To our knowledge, this is the most comprehensive assessment of neurocognitive function in people with MDR-TB. We used a number of well-validated instruments for the assessment, many of which have previously been successfully used in South Africa and many of which have suitable reference data from healthy HIV-negative adults without TB.

Conclusion

Our study highlights the importance of including the assessment and study of neurocognition in TB/HIV patients, this being a necessary step in the scientific evidence-based and integrated approach being advocated to improve both the TB cure rate and the health outcomes for HIV [25]. If the WHO’s 90-90-90 and the Sustainable Development Goal of ending the tuberculosis epidemic by 2030 is to be realised, further longitudinal research is necessary to explore the course and causes of neurocognitive decline in the local TB-HIV population, as well as to quantify their implications for quality of life, medication adherence and both social and occupational functionality.