Winston - 2013. Neurocognitive Function in HIV+ART

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Neurocognitive Function in HIV Infected Patients on

Antiretroviral Therapy
Alan Winston1,2*, Alejandro Arenas-Pinto3,4, Wolfgang Stöhr3, Martin Fisher5, Chloe M. Orkin6,
Kazeem Aderogba7, Andrew De Burgh-Thomas8, Nigel O’Farrell9, Charles J.N. Lacey10, Clifford Leen11,12,
David Dunn3, Nicholas I. Paton3,13 for the PIVOT Trial Team"
1 Section of Infectious Diseases, Division of Medicine, Imperial College London, St Mary’s Hospital Campus, Norfolk Place, London, United Kingdom, 2 Department of HIV
and GU Medicine, Imperial College Healthcare NHS Trust, London, United Kingdom, 3 Medical Research Council Clinical Trials Unit, London, United Kingdom, 4 Centre for
Sexual Health and HIV Research, University College London, London, United Kingdom, 5 Brighton and Sussex University Hospitals NHS Trust, Brighton, United Kingdom,
6 Barts Health NHS Trust, Ambrose King Centre, Royal London Hospital, London, United Kingdom, 7 East Sussex Healthcare Trust, Avenue House, Eastbourne, United
Kingdom, 8 Gloucestershire Care Services, Hope House, Gloucestershire Royal Hospital, Gloucester, United Kingdom, 9 Pasteur Suite, Ealing Hospital, London, United
Kingdom, 10 Centre for Immunology and Infection, Hull York Medical School, University of York, York, United Kingdom, 11 Edinburgh Infectious Diseases, Edinburgh
University Regional Infectious Diseases Unit, Western General Hospital, Edinburgh, United Kingdom, 12 Edinburgh University, Edinburgh, United Kingdom, 13 National
University of Singapore, Singapore

Abstract
Objective: To describe factors associated with neurocognitive (NC) function in HIV-positive patients on stable combination
antiretroviral therapy.

Design: We undertook a cross-sectional analysis assessing NC data obtained at baseline in patients entering the Protease-
Inhibitor-Monotherapy-Versus-Ongoing-Triple therapy (PIVOT) trial.

Main outcome measure: NC testing comprised of 5 domains. Raw results were z-transformed using standard and
demographically adjusted normative datasets (ND). Global z-scores (NPZ-5) were derived from averaging the 5 domains and
percentage of subjects with test scores .1 standard deviation (SD) below population means in at least two domains
(abnormal Frascati score) calculated. Patient characteristics associated with NC results were assessed using multivariable
linear regression.

Results: Of the 587 patients in PIVOT, 557 had full NC results and were included. 77% were male, 68% Caucasian and 28% of
Black ethnicity. Mean (SD) baseline and nadir CD4+ lymphocyte counts were 553(217) and 177(117) cells/mL, respectively,
and HIV RNA was ,50 copies/mL in all. Median (IQR) NPZ-5 score was 20.5 (21.2/20) overall, and 20.3 (20.7/0.1) and 21.4
(22/20.8) in subjects of Caucasian and Black ethnicity, respectively. Abnormal Frascati scores using the standard-ND were
observed in 51%, 38%, and 81%, respectively, of subjects overall, Caucasian and Black ethnicity (p,0.001), but in 62% and
69% of Caucasian and Black subjects using demographically adjusted-ND (p = 0.20). In the multivariate analysis, only Black
ethnicity was associated with poorer NPZ-5 scores (P,0.001).

Conclusions: In this large group of HIV-infected subjects with viral load suppression, ethnicity but not HIV-disease factors is
closely associated with NC results. The prevalence of abnormal results is highly dependent on control datasets utilised.

Trial registry: ClinicalTrials.gov, NCT01230580

Citation: Winston A, Arenas-Pinto A, Stöhr W, Fisher M, Orkin CM, et al. (2013) Neurocognitive Function in HIV Infected Patients on Antiretroviral Therapy. PLoS
ONE 8(4): e61949. doi:10.1371/journal.pone.0061949
Editor: Shilpa J Buch, University of Nebraska Medical Center, United States of America
Received October 9, 2012; Accepted March 13, 2013; Published April 30, 2013
Copyright: ß 2013 Winston et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: PIVOT is funded by the National Institutes for Health Research Health (NIHR) Technology Assessment programme (project number 06/403/90). The
views and opinions expressed herein are those of the authors and do not necessarily reflect those of the Health Technology Assessment programme, NIHR, the
National Health Service or the Department of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of
the manuscript.
Competing Interests: The authors have read the journal’s policy and have the following conflicts: AW has received honoraria or research grants, or been a
consultant or investigator, in clinical trials sponsored by Abbott, Boehringer Ingelheim, Bristol-Myers Squibb, Gilead Sciences, GlaxoSmithKline, Janssen Cilag,
Roche, and Pfizer. All other authors have no conflict of interest. This does not alter the authors’ adherence to all the PLOS ONE policies on sharing data and
materials.
* E-mail: [email protected]
"
The members of the PIVOT Trial Team are listed in the Acknowledgements section

PLOS ONE | www.plosone.org 1 April 2013 | Volume 8 | Issue 4 | e61949


The PIVOT Neurocognitive Study

Introduction using Hopkins Verbal Learning Test-Revised (HVLT-R) for learning


and recall [20], fine motor skills assessed using Grooved Pegboard
Whilst effective combination antiretroviral therapy (cART) has [21], and attention and executive function assessed using Colour
dramatic effects on the incidence of serious neurological compli- Trails Tests (CTT) 1 and 2 [22]. These tests were specifically
cations secondary to chronic HIV infection, including severe HIV- chosen to assess the cognitive domains reported to be predomi-
associated dementia [1,2], high rates of more subtle cognitive nantly affected in chronic HIV infection [23] and were feasible to
decline in HIV infected subjects are being increasingly described undertake within a multicentre clinical study.
with neurocognitive (NC) impairment rates approaching 50% in Subjects also self-completed an anxiety and depression score
some cohorts [3,4,5]. Several factors have been implicated with questionnaire (EQ-5D, [24]) selecting one of the three following
the evolution of NC impairment in the cART era including older options: 1) I am not anxious or depressed, 2) I am moderately
age [3], low nadir CD4+ lymphocyte count [6], chronic hepatitis- anxious or depressed, 3) I am extremely anxious or depressed.
C-virus (HCV) co-infection [7,8] and possibly the use of
antiretroviral regimens with poor central-nervous-system (CNS)
exposure [9].
Normative standards
Raw scores for each cognitive test were transformed to z-scores
Given the changing picture of HIV associated CNS disease, a
using the manufacturers’ normative data [20,21,22] matched for
revised nomenclature system often known as the Frascati criteria,
each participant by age (for all tests) and also by years of education
has been proposed classifying subjects with abnormal NC testing
(only for the CTT). These reference data were of US origin and in
results into three categories based on patients symptoms measured
via the activities of daily living scale [10]. Subjects with abnormal general derived from a large number of subjects (over 1000) of
NC testing results, who are otherwise asymptomatic are classified predominantly Caucasian ethnicity. Additionally we utilise nor-
as HIV-associated asymptomatic NC impairment (ANI), who are mative data considering ethnicity which were available for the
mildly symptomatic are classified as HIV-associated mild NC CTT [22] and the HVLT domains [25]: Here, normative data
disorder (MND) and in whom are severely symptomatic classified were provided separately for subjects of Caucasians and Black
as HIV-associated dementia (HAD). (African-American) ethnicity although subject numbers for the
Although high rates of NC impairment in HIV infected subjects latter group (approximately 150 control subjects) were smaller. In
are reported, the clinical characteristics of these reported cohorts addition, for HVLT and CTT these alternative normative data
often differ and may include confounding variables which could did not distinguish between age groups and years of education,
influence NC function. For instance antiretroviral treatment status respectively, and for CTT the age categories for Black ethnicity
[11], levels of plasma HIV viraemia and HIV disease stage [12] did only cover the range of 20 to 50 years.
and subjects ethnicity [13] vary widely between reports.
In general NC function improves when initiating cART Statistical analyses
[14,15,16] and the majority of HIV infected people attending Only subjects in whom full NC testing results were available
for care are now on stable ART with full viral load suppression from all 5 domains were included in the analyses. Baseline patient
[17,18]. The aim of this study was to determine the prevalence of characteristics, and ART and HIV disease histories for all eligible
NC impairment and the factors associated with NC function in a subjects were tabulated separately for subjects of Caucasian and
large contemporary cohort of patients on stable ART. Black ethnicity. We calculated the CNS Penetration Effectiveness
(CPE) score based on the 2010 scoring system for study baseline
Methods antiretroviral regimens [26] and calculated each subjects 10 year
cardiovascular risk [27].
Subject selection For each cognitive domain, z-scores were calculated by
This cross-sectional analysis included patients whom had agreed subtracting the mean and dividing by the standard deviation
to take part in the Protease Inhibitor monotherapy Versus (SD) of test scores in reference populations using both the standard
Ongoing Triple-therapy in the long term management of HIV and demographically adjusted normative datasets. The z-score for
infection (PIVOT) [19] trial. The protocol for this study is the fine motor skills domain was obtained by taking the average of
available as supporting information; see Protocol S1. Eligibility the z-scores for the dominant and non-dominant hands on the
criteria included documented HIV infection; receiving a stable Grooved Pegboard tests. Global z-scores (NPZ-5) were derived
cART regimen comprising two nucleoside-reverse-transcriptase- from averaging the 5 domains. For all individual test z scores and
inhibitors (NRTIs) with either a non-nucleoside-reverse-transcriptase- the NPZ-5, scores above zero denote above-average neurocogni-
inhibitor (NNRTI) or a protease-inhibitor (PI) for at least 24 weeks tive function and scores below zero denote below-average
prior to study entry; a plasma HIV RNA ,50 copies/mL at neurocognitive function compared to the reference population. Z
screening and for at least 24 weeks prior to screening; and a CD4+ scores of each individual test were dichotomised based on a
lymphocyte count .100 cells mL. This paper reports the results of the threshold of one SD below the mean of the normative dataset (i.e.
NC tests done at baseline in patients entering the PIVOT study, prior ,21).
to randomisation. In addition we used two measures for defining abnormal NC
The PIVOT study is registered as ClinicalTrials.gov function. Firstly, the percentage of subjects having a global NPZ-
NCT01230580. National Research Ethics Service (NRES) ap- 5,21. Secondly, we calculated the percentage of subjects with
proval for the trial, including the NC assessments, was obtained test scores ,21 in at least two individual domains, corresponding
from the East of England Cambridge South Ethics Committee. to a previously published definition, frequently known as the
Frascati score [10]. We categorised subjects into having an
Neurocognitive measures ‘abnormal Frascati score’ if at least two cognitive domains scored
NC testing was administered by a designated study nurse or ,21 or otherwise categorised as having a ‘normal Frascati score’.
investigator at each study site who had completed standardised As no formal activities of daily living scale was assessed, we were
training to administer the test procedures. Five domains were not able to subcategorise subjects in the ANI, MCD and HAD
assessed as follows; Verbal learning and memory were assessed categories.

PLOS ONE | www.plosone.org 2 April 2013 | Volume 8 | Issue 4 | e61949


The PIVOT Neurocognitive Study

Prevalence of abnormal NC function using these measures were were very similar using univariable linear regression compared to
calculated using standard and ethnicity adjusted normative data, the statistics given in Table 2. These differences between ethnic
for the overall study population, and separately for subjects of groups remained nearly unchanged in a multivariable model, that
Caucasian and Black ethnicity. is after adjustment for demographic and clinical variables and
Proportions were compared using chi-square test. Simple other potential influence factors of NC performance. Both Black
comparisons of test scores were made using Mann-Whitney or (21.1 score points; 95% CI 21.3 to 20.9) and other ethnicity
Kruskal-Wallis rank tests. We used univariable and multivariable (20.6 score points; 95% CI 21.0 to 20.3) were associated with
linear regression models to identify baseline factors associated with poorer NPZ-5 scores compared with Caucasians (overall P value
test scores (NPZ-5, z-scores), investigating gender, age, ethnicity, for ethnic group ,0.001).
years of education, nadir and current CD4+ T-cell count, time Because of an apparent influence of the reference data used (see
since HIV RNA suppression, years on ART, current use of below), we decided to assess factors associated with NPZ-5 scores
NNRTIs, current or past use of efavirenz, CPE score of current separately for subjects of Caucasian and Black ethnicity; numbers
regimen, current and past smoking, cardiovascular (CV) risk, for other ethnicity were considered too small for meaningful
presence of hepatitis C virus (HCV) antibodies, baseline haemo- multivariable analyses.
globin, and anxiety/depression level. In subjects of Caucasian ethnicity, current nevirapine use (0.34
score points less in those on nevirapine; p = 0.021) and HCV (0.33
Results score points less in those with detectable antibodies; p = 0.047)
were associated with lower NPZ-5 scores (Table 3). Of note, the
Patient characteristics
association with nevirapine was adjusted for previous cessation of
Full NC results were available for 557 of the 587 subjects
efavirenz due to CNS toxicity, and results were very similar when
enrolled into PIVOT with baseline characteristics shown in Table 1.
instead adjusting for never, current, previous use of efavirenz (i.e.
Of interest, and of potential clinical relevance, although current
ignoring recorded cause of cessation). 20/43 of patients currently
CD4+ lymphocyte count (mean, SD) was relatively high at 553
on nevirapine have been on efavirenz before, of which 12 had
(217) cells/mL, nadir CD4+ lymphocyte count was 177 (117) cells/
stopped efavirenz due to CNS toxicity, however, in patients on
mL. Subjects had been receiving cART for a mean of 5 (SD 3)
nevirapine adjusted NPZ-5 scores were 20.63 and 20.54,
years with 53% of the cohort receiving an NNRTI based regimen
respectively, in those never and previously on efavirenz. There-
and 39% receiving an efavirenz containing regimen at baseline.
Thirty-one per cent of subjects reported moderate anxiety or fore, association with nevirapine was unlikely being confounded by
depression with only 2% of subjects reporting extreme anxiety. a previous switch from efavirenz.
The majority of subjects were of Caucasian (68%) or Black In subjects of Black ethnicity, smoking was the only factor
(28%) ethnicity with only 20 subjects (4%) of other ethnicities. In associated with NPZ-5 scores: current and ex-smokers had 0.68
general, baseline characteristics were similar between these groups and 0.37, respectively, score points higher than subjects who never
apart from the much higher proportion of men in the Caucasian smoked (overall p = 0.022) (Table 3).
group. Other HIV disease factors such as nadir CD4+ lymphocyte
count and antiretroviral factors such as CPE score were not
Neurocognitive test results utilising standard normative associated with NPZ-5 score in either ethnic group.
data
Using the standard normative dataset, median (IQR) NPZ-5 Neurocognitive test results utilising demographically
scores was 20.5 (21.2, 20) overall, and 20.3 (20.7, 0.1) and adjusted normative data
21.4 (22, 20.8), respectively, in subjects of Caucasian and Black Using the demographically adjusted normative data, the
ethnicity (Table 2). In all ethnic groups, NPZ-5 scores were differences between ethnicities were less marked with poorer
significantly below 0 (P,0.001 in all). Overall, 32% of subjects had results for subjects of Caucasian ethnicity and better results for
a global NPZ-5 score ,21, with a significantly different subjects of Black ethnicity compared with using the standard
prevalence in Caucasian and Black subjects (17% vs. 67%; normative data (Table 2). Although a statistically significant
P,0.001). When assessing the number of subjects with an difference in the number of subjects with NPZ-5 scores,21
abnormal Frascati score a similar trend was observed, although remained between ethnicities (P = 0.001), the difference was no
proportions of abnormal findings were generally higher, with 51%, longer statistically significant in numbers of subjects meeting the
38% and 81%, respectively, of subjects overall, of Caucasian and Frascati criteria (P = 0.20). Of note, whereas there was a significant
of Black ethnicity (P,0.001 for difference between Caucasian and difference in NPZ-5 scores between Caucasian and Black subjects
Black ethnicity). in univariable regression analysis (p,0.001), the difference
In 80% of subjects overall, the two definitions of impairment disappeared-in contrast to the analysis using the standard
were in agreement (Figure 1), and NPZ-5 score and the Frascati normative dataset-when adjusting for demographic, clinical and
criteria were either both ‘normal’ (48%) or both ‘abnormal’ (31%). other factors (p = 0.31).
In 1% of subjects, NPZ-5 score was,21 but Frascati criteria were
classified as normal, and vice versa in 20%. Hence, discordance Discussion
between findings based on NPZ-5 scores and the Frascati criteria
was predominantly due to an abnormal Frascati score definition in We have assessed NC function in a large group of HIV infected
the presence of a NPZ-5 score which was above our cut-off of 1 subjects on effective cART using a standard cognitive testing
SD below normative means. Agreement was 77% in Caucasians, battery. Although this is not the same as a formal neurophyscho-
and 85% in patients of Black ethnicity. logical assessment, the brief neurocognitive test battery we used
comprised validated components of standard formal neuropsy-
Factors associated with global cognitive scores chological tests [28] and we have made several interesting and
NPZ-5 scores were not quite normally distributed (slightly novel observations relevant to both current clinical practice and
skewed to the left). However, differences between ethnic groups future research in this field.

PLOS ONE | www.plosone.org 3 April 2013 | Volume 8 | Issue 4 | e61949


The PIVOT Neurocognitive Study

Table 1. Baseline characteristics, overall, and separately for subjects of Caucasian and Black ethnicity.

Variable Overall1 (n = 557) Caucasian ethnicity (n = 381) Black ethnicity (n = 156)

Gender, male 427 (77) 353 (93) 61 (39)


Age, years 44 (9) 45 (10) 44 (8)
Nadir CD4+, cells/mL 177 (117) 189 (117) 147 (109)
Baseline CD4+, cells/mL 553 (217) 569 (219) 512 (205)
Years undetectable HIV RNA 4 (3) 4 (3) 4 (3)
Years education 15 (4) 15 (4) 15 (4)
Years on cART 5 (3) 5 (3) 5 (3)
on NNRTI 296 (53) 187 (49) 102 (65)
Efavirenz
never on 217 (39) 147 (39) 59 (38)
not currently on but previously 125 (22) 93 (24) 27 (17)
currently on 215 (39) 141 (37) 70 (45)
efavirenz ever stopped due to CNS problems{ 67 (12) 55 (14) 11 (7)
baseline CPE score 7 (1) 7 (1) 8 (1)
Smoker
never 261 (47) 143 (38) 113 (72)
current 157 (28) 135 (35) 17 (11)
ex-smoker 139 (25) 103 (27) 26 (17)
CV risk
,10% 341 (61) 199 (52) 127 (82)
10–20% 187 (34) 158 (41) 26 (17)
.20–30% 28 (5) 24 (6) 2 (1)
Hepatitis C antibody positive 21 (4) 18 (5) 2 (1)
Baseline haemoglobin, g/dL 14 (1) 15 (1) 13 (1)
Anxiety/Depression
not anxious/depressed 369 (67) 249 (66) 106 (70)
moderately anxious/depressed 173 (31) 121 (32) 46 (30)
extremely anxious/depressed 9 (2) 9 (2) 0

1 {
Table 1 legend: only patients with complete neurocognitive data (including patients of ethnicity other than Caucasian or Black, n = 20); 5/67 (all Caucasian) were back
on efavirenz at baseline; cART = combination antiretroviral therapy, CPE = Clinical Penetration Effectiveness, CV = cardiovascular, NNRTI = non-nucleoside reverse-
transcriptase inhibitor; Data are number (%) or mean (standard deviation).
doi:10.1371/journal.pone.0061949.t001

In our study, the median global NPZ-5 score for Caucasian for this in the model, it may be that improvements in cognitive
subjects, using the standard normative control data, was 20.3 and function may continue for many years until global cognitive scores
therefore slightly lower than 0 which to be expected in a normal in a group of patients may start to match control population
population. However, these data are reassuring insofar as the values.
difference is quite small and unlikely to be functionally significant Despite global NPZ-5 scores in Caucasian subjects being similar
[29], suggesting that in effectively treated HIV infected subjects to the standard normative control data, high rates of abnormalities
global cognitive function is similar to normative control data. were observed when we assessed results using the Frascati criteria
Furthermore, the small difference may be driven by factors other with 37% of subjects having an abnormal Frascati score in this
than HIV infection that can influence neurocognitive function group. These results differ from a recent study conducted in the
such as differences in rates of smoking that may differ from the UK where Garvey et al described only 19% of Caucasian individuals
normative dataset. We did not have an HIV-negative comparison to have an abnormal Frascati score [31]. In this study, global NPZ
group in our study which might have provided further clarity on scores were also similar to population control data. The differences
this. Our study differs from other reports where poorer global may be due to the patient selection and nature of the testing
NPZ-5 scores in HIV infected cohort are described [5,6]. We performed but it is also possible that these differences could be due
postulate these differences may in part be due to differences in the to the normative datasets used in our study compared to the study
populations studied. In our study, all subjects were virologically conducted by Garvey et al. In our study we undertook traditional
suppressed at the time of study and had been so for a mean of 4 NC testing where the normative control dataset originate from
years. Previous studies have reported improvements in cognitive historic US cohorts, some dating from the 1960s and 1970s. Garvey
function in HIV infected subjects commencing antiretroviral et al undertook computerised NC testing (CogStateTM, Mel-
therapy with the dynamic of these improvements evident for at bourne, Australia) where the control population are predominant-
least a one year period [15,30]. Although we did not find evidence ly Caucasian male Australian subjects recruited within the last

PLOS ONE | www.plosone.org 4 April 2013 | Volume 8 | Issue 4 | e61949


The PIVOT Neurocognitive Study

Table 2. Neurocognitive testing results.

P-value for difference


Caucasian Black ethnicity Other ethnicity between Caucasian vs.
1
Parameter Overall (n = 557) ethnicity (n = 381) (n = 156) (n = 20) Black ethnicity

Standard NPZ-5 score 20.5 (21.2, 20) 20.3 (20.7, 0.1) 21.4 (22, 2.8) 20.8 (21.3, 20.4) ,0.001
normative data [95%CI 20.8, 20.6] [95%CI 20.4, 20.3] [95%CI 21.8, 21.4] [95%CI 21.3, 20.6]
Overall NPZ-5 score .1 SD 179 (32) 66 (17) 104 (67) 9 (45) ,0.001
below population mean
At least 2 individual 285 (51) 144 (38) 127 (81) 14 (70) ,0.001
tests with z-score .1 SD
below population mean
Adjusted NPZ-5 score 20.8 21.1 (21.6, 20.6) ,0.001
normative data" (21.3, 20.3)
Overall NPZ-5 score.1 SD 145 (38) 69 (55) 0.001
below population mean
At least 2 individual tests 236 (62) 86 (68) 0.20
with z-score.1 SD below
population mean

Table 2 legend: 1patients with complete neurocognitive data including patients of ethnicity other than Caucasian or Black (n = 20); IQR = interquartile range, SD =
standard deviation; "Excludes all patients of ethnicity other than Caucasian and Black, as well as 30 subjects of Black ethnicity who fell outside the age range covered by
the adjusted normative data; NPZ-5 results using standard normative data in these 30 patients were similar to those of the other 126 patients. Data are number (%) or
median (interquartile range).
doi:10.1371/journal.pone.0061949.t002

decade. With cultural changes over time, it is possible that the undertaken using instructions in English and it is possible language
CogStateTM control dataset is more representative of our difficulties could account in part for these differences. However we
Caucasian cohort recruited over the past few years in the UK do not believe this to be an important factor because the
and therefore when a NC score is calculated which encompasses differences were no more marked on the HVLT tests, which are
results from several cognitive domains, such as the Frascati score, highly language-dependent, than they were on the other tests
such differences in results become apparent. which do not depend on linguistic fluency. Furthermore, the
We observed poorer results within individual NC domain scores majority of the Black patients attending clinics in the UK are
(data not shown), global NPZ-5 scores and the categorical score either born in the UK, or are immigrants from African countries
using the Frascati criteria in our study for subjects of Black where English is widely spoken, and patients who were not able to
ethnicity compared to subjects of Caucasian ethnicity. There are understand the study instructions were not included. Patients who
several plausible explanations for these findings. All tests were appeared to understand the instructions but then performed the

Figure 1. Association between NPZ-5 score and Frascati score.


doi:10.1371/journal.pone.0061949.g001

PLOS ONE | www.plosone.org 5 April 2013 | Volume 8 | Issue 4 | e61949


The PIVOT Neurocognitive Study

Table 3. Linear regression analysis assessing factors associated with NPZ-5 scores.

Univariable Multivariable

Study parameter Caucasian ethnicity Black ethnicity Caucasian ethnicity Black ethnicity

Coef. [95% CI] P Coef. [95% CI] P Coef. [95% CI] P Coef. P

Gender, female vs. male 0.06 [20.19, 0.32] 0.63 20.10 [20.40, 0.21] 0.54 0.03 [20.25,0.32] 0.83 20.05 0.83
[20.49,0.40]
Anxious/depressed at baseline 20.12 [20.26, 0.02] 0.09 20.00 [20.33, 0.32] 0.99 20.11 0.14 0.04 [20.32,0.40] 0.83
(moderately or extremely vs. Not) [20.25,0.04]
Efavirenz, on at baseline 0.04 [20.09, 0.18] 0.54 20.12 [20.41, 0.18] 0.44 20.05 0.53 20.25 0.19
vs. not on [20.22,0.11] [20.62,0.12]
Nevirapine, on at baseline 20.28 [20.49, 20.07] 0.009 20.07 [20.44, 0.31] 0.72 20.34 20.63, 0.021 20.08 0.79
vs. not on 20.05 [20.62,0.47]
Ever stopped efavirenz due 0.07 [20.12, 0.26] 0.49 20.23 [20.80, 0.34] 0.42 0.09 [20.11,0.29] 0.39 20.16 0.62
to CNS problems, (yes v. No) [20.79,0.47]
Smoker, vs. never
current smoker 0.02 [20.14, 0.18] 0.80 0.63 [0.17, 1.10] 0.008 0.05 [20.13,0.22] 0.60 0.68 [0.11,1.25] 0.020
ex-smoker 0.13 [20.04, 0.30] 0.13 0.30 [20.09, 0.70] 0.13 0.12 [20.05,0.29] 0.17 0.37 [20.08,0.82] 0.11
Cardiovascular risk, . = 10% 20.06 [20.20, 0.07] 0.36 20.01 [20.40, 0.38] 0.96 20.05 0.53 20.32 0.18
vs. ,10% [20.20,0.10] [20.80,0.15]
Hepatitis C antibody, 20.30 [20.61, 0.01] 0.061 20.44 [21.74, 0.86] 0.51 20.33 [20.65, 0.047 20.26 0.70
positive vs. negative 20.00] [21.63,1.11]
Prior AIDS illness, yes vs. no 0.05 [20.13, 0.23] 0.58 20.10 [20.45, 0.25] 0.58 0.02 [20.17,0.22] 0.82 20.12 0.57
[20.54,0.30]
Age, per 10 years 20.05 [20.12, 0.02] 0.14 20.06 [20.25, 0.13] 0.55 20.05 0.21 20.04 0.74
[20.13,0.03] [20.26,0.18]
CD4+ nadir, per increasing 20.03 [20.09, 0.02] 0.26 0.00 [20.13, 0.14] 0.97 20.01 0.81 20.09 0.34
100 cells/mL [20.08,0.06] [20.26,0.09]
CD4+ at baseline, per increasing 20.03 [20.06, 0.00] 0.06 20.00 [20.07, 0.07] 0.95 20.02 0.20 0.01 [20.09,0.11] 0.80
100 cells/mL [20.06,0.01]
Years with undetectable HIV RNA 20.01 [20.03, 0.01] 0.45 20.04 [20.10, 0.02] 0.19 0.00 [20.03,0.04] 0.90 20.04 0.34
[20.13,0.05]
Years of known HIV infected 20.01 [20.02, 0.01] 0.41 20.00 [20.04, 0.04] 0.995 0.00 [20.01,0.02] 0.60 0.03 [20.03,0.09] 0.34
Years of education 20.08 [20.17, 0.01] 0.10 0.03 [20.16, 0.22] 0.75 20.02 0.089 0.00 [20.04,0.05] 0.91
[20.04,0.00]
Number of different ART 20.02 [20.06, 0.02] 0.31 20.12 [20.21, 20.02] 0.021 20.02 0.53 20.09 0.17
drugs received [20.07,0.03] [20.22,0.04]
Haemoglobin, per g/dl 0.01 [20.05, 0.07] 0.80 0.05 [20.06, 0.15] 0.39 0.02 [20.05,0.09] 0.60 20.00 0.97
[20.15,0.14]
CPE score of regimen 20.02 [20.09, 0.05] 0.59 20.00 [20.14, 0.14] 0.99 0.03 [20.06,0.13] 0.50 20.05 0.62
[20.24,0.15]

Table 3 legend: CI = confidence interval; CPE = Clinical Penetration Effectiveness; NPZ-5 = mean neuropsychiatric z-score for 5 domains; Coef = regression coefficient. P
values (P),0.05 shown in bold. "Overall p-value for smoking in subjects of Black ethnicity 0.016 (univariable) and 0.022 (multivariable).
doi:10.1371/journal.pone.0061949.t003

tests incorrectly would also have been excluded as the tests would studies using different normative datasets [31,32], quite different
have been deemed invalid. Differences in HIV disease factors or results are observed. Lastly, results of NC testing from control
other characteristics at baseline could also explain such differences. datasets which are recruited specifically to match HIV infected
However, baseline characteristics were similar for subjects of cohorts are described to differ substantially from traditional
Caucasian and Black ethnicity in many areas, and the influence of population control datasets [13]. Indeed the effects of ethnicity
ethnicity persisted after adjusting for baseline factors. on cognitive function and lack of appropriate control data have
We hypothesise the differences in NC test results between the previously been described in a female HIV-infected cohort [13].
ethnicities observed in our study are due to differences in the A finding from our study which remains challenging is to
characteristics of control datasets available for use for several explain the poorer NC testing results observed in Caucasian
reasons. Firstly, when utilising the demographically adjusted subjects when using the demographically adjusted dataset com-
normative dataset, the differences between ethnicities become less pared to the standard normative dataset. Again this may be related
marked and indeed no statistically significant difference is to differences in the Caucasian controls used within these control
observed in the Frascati criteria definition between Caucasian datasets however this area requires further investigation.
and Black ethnicities when we use this adjusted normative data. Nadir CD4+ lymphocyte count has been associated with greater
Secondly, as described above, for Caucasian subjects, in other risk of NC impairment in many reports [6,31,33], a finding we did

PLOS ONE | www.plosone.org 6 April 2013 | Volume 8 | Issue 4 | e61949


The PIVOT Neurocognitive Study

not observe within our cohort. The longer duration of effective Acknowledgments
antiretroviral therapy and a lack of historic documented virolog-
ical failure differs from other studies and is a possible explanation. We thank all the patients and staff from all the centres participating in the
PIVOT trial.
We observed an association between current nevirapine use in
Barking Hospital: S Obeyesekera; Barts & The London: C Orkin;
Caucasian subjects and poorer NC function which has not been Birmingham Heartlands Hospital: S Taylor; Elton John Centre, Brighton:
previously described. This is an unexpected finding as nevirapine is M Fisher; Central Middlesex Hospital: G Brook; Chelsea & Westminster
considered to have pharmacokinetic exposure at effective concen- Hospital: B Gazzard; Ealing Hospital: N O’Farrell; Avenue House/Ore
trations in the cerebrospinal fluid [34] and no in vitro data to our Clin Eastbourne: K Aderogba; Edith Cavell Hospital: S Jebakumar;
knowledge have reported any neuronal toxicities associated with Gloucester Royal Hospital: A DeBurgh-Thomas; Homerton University
nevirapine. We believe this may be a channelling bias whereby Hospital: I Reeves; Harrogate District Hospital: C Lacey; James Cook Uni
subjects with pre-existing NC impairment or psychiatric conditions Hosp Middlesbrough: D Chadwick; King’s College Hospital: F Post;
are preferentially commenced on nevirapine. In our multivariate Leicester Royal Infirmary: A Palfreeman; Luton & Dunstable Hospital: T
Balachandran; Manchester Royal Infirmary: V Lee; Royal Victoria Infirm
model, ever stopping efavirenz due to CNS toxicities was not
Newcastle: E Ong; Newham University Hospital: R O’Connell; Norfolk
associated with global NPZ-5 scores, which may suggest the and Norwich University Hospital: N David; North Manchester General: E
association we have observed with nevirapine is not associated with Wilkins; North Middlesex University: J Ainsworth; Northwick Park
subjects switching to nevirapine from other cART regimens with Hospital: M Kapembwa; Derrifor Hospital Plymouth: Z Fox; Royal
overt CNS toxicities. However this association does not address the Berkshire Hospital: F Chen; Royal Bournemouth Hospital: E Herieka;
channelling bias associated with commencing a nevirapine regimen Royal Free Hospital, London: M Johnson; Royal Hallamshire Hospital: D
de novo in subjects previously naive to antiretroviral therapy. Dockrell; Royal Liverpool University Hospital: N Beeching; Royal Victoria
Several findings from our study are highly relevant to this field of Hospital Belfast: S Pheng Quah; Southend University Hospital: J Day;
HIV medicine. Of important to clinicians caring for HIV infected Southmead Hospital Bristol: M Gompels; St James Hospital Leeds: J
Minton; St George’s Hospital London: P Hay; St Mary’s Hospital,
individuals and for HIV community groups, our description of a large
London: A Winston; St Mary’s Hospital, Portsmouth: V Harindra; St
cohort where overall NC function in subjects on cART for many years Thomas’ Hospital, London: J Fox; The Mortimer Market Centre, London:
is similar to a matched HIV seronegative control population, albeit I Williams; University Hospital of Coventry: S Allan; Western General
only within the Caucasian cohort in whom we believe the matched Edinburgh: C Leen; Queen Elizabeth Hosp Woolwich: S Kegg; Worcester
population datasets are the most relevant, is reassuring. Although we Royal Hospital: M Roberts; York Teaching Hospital: C Lacey
are not able to tease out pathogenic mechanisms associated with MRC Clinical Trials Unit: N Paton, A Arenas-Pinto, S Fleck, D
cerebral dysfunction, we are able to provide reassuring data that in Dunn, W Stöhr, K Scott, E Owen-Powell, I Kummeling, K Sanders, L
general NC function within a well cared for cohort, does not Vivas; Trial Steering Committee: A Freedman (Chair), D Mercey, S
substantially differ from a control population. Another seminal finding Fidler, E Torok, B Cromarty, N Paton, B Gazzard, C Orkin, A Babiker;
non-voting members: D Dunn, W Stöhr; Independent Data Monitor-
from our work for future clinical research, is the importance of
ing Committee: T Peto (Chair), D Lalloo, A Phillips, R James
recruiting well matched control populations to the HIV infected
populations being studied, in order to aid the interpretation of study
findings and in order to assess if findings are related to HIV-disease Author Contributions
itself or the cohorts being studied. Conceived and designed the experiments: AW NP DD AP WS. Performed
the experiments: AP WS. Analyzed the data: WS DD. Contributed
Supporting Information reagents/materials/analysis tools: WS DD AP. Wrote the paper: AW AP
WS DD NP MF CO KA ADT CJNL CL NO.
Protocol S1.
(PDF)

References
1. Dore GJ, Correll PK, Li Y, Kaldor JM, Cooper DA, et al. (1999) Changes to 10. Antinori A, Arendt G, Becker JT, Brew BJ, Byrd DA, et al. (2007) Updated
AIDS dementia complex in the era of highly active antiretroviral therapy. Aids research nosology for HIV-associated neurocognitive disorders. Neurology 69:
13: 1249–1253. 1789–1799.
2. Garvey L, Winston A, Walsh J, Post F, Porter K, et al. (2011) HIV-associated 11. Heaton RK, Franklin DR, Ellis RJ, McCutchan JA, Letendre SL, et al. (2011)
central nervous system diseases in the recent combination antiretroviral therapy HIV-associated neurocognitive disorders before and during the era of
era. Eur J Neurol 18: 527–534. combination antiretroviral therapy: differences in rates, nature, and predictors.
3. Becker JT, Lopez OL, Dew MA, Aizenstein HJ (2004) Prevalence of cognitive J Neurovirol 17: 3–16.
disorders differs as a function of age in HIV virus infection. Aids 18 Suppl 1: 12. Cysique LA, Brew BJ (2011) Prevalence of non-confounded HIV-associated
S11–18. neurocognitive impairment in the context of plasma HIV RNA suppression.
4. Joska JA, Westgarth-Taylor J, Myer L, Hoare J, Thomas KG, et al. (2010) J Neurovirol 17: 176–183.
Characterization of HIV-Associated Neurocognitive Disorders Among Individ- 13. Manly JJ, Smith C, Crystal HA, Richardson J, Golub ET, et al. (2011)
uals Starting Antiretroviral Therapy in South Africa. AIDS Behav. Relationship of ethnicity, age, education, and reading level to speed and
5. Robertson KR, Smurzynski M, Parsons TD, Wu K, Bosch RJ, et al. (2007) The executive function among HIV+ and HIV- women: the Women’s Interagency
prevalence and incidence of neurocognitive impairment in the HAART era. HIV Study (WIHS) Neurocognitive Substudy. J Clin Exp Neuropsychol 33:
Aids 21: 1915–1921. 853–863.
6. Ellis RJ, Badiee J, Vaida F, Letendre S, Heaton RK, et al. (2011) CD4 nadir is a 14. Al-Khindi T, Zakzanis KK, van Gorp WG (2011) Does Antiretroviral Therapy
predictor of HIV neurocognitive impairment in the era of combination Improve HIV-Associated Cognitive Impairment? A Quantitative Review of the
antiretroviral therapy. Aids 25: 1747–1751. Literature. J Int Neuropsychol Soc 17: 956–969.
7. Thiyagarajan A, Garvey LJ, Pflugrad H, Maruff P, Scullard G, et al. (2010) 15. Cysique LA, Vaida F, Letendre S, Gibson S, Cherner M, et al. (2009) Dynamics
Cerebral function tests reveal differences in HIV-infected subjects with and of cognitive change in impaired HIV-positive patients initiating antiretroviral
without chronic HCV co-infection. Clin Microbiol Infect 16: 1579–1584. therapy. Neurology.
8. Ryan EL, Morgello S, Isaacs K, Naseer M, Gerits P (2004) Neuropsychiatric 16. Winston A, Duncombe C, Li PC, Gill JM, Kerr SJ, et al. (2010) Does choice of
impact of hepatitis C on advanced HIV. Neurology 62: 957–962. combination antiretroviral therapy (cART) alter changes in cerebral function
9. Letendre S, Marquie-Beck J, Capparelli E, Best B, Clifford D, et al. (2008) testing after 48 weeks in treatment-naive, HIV-1-infected individuals commenc-
Validation of the CNS Penetration-Effectiveness rank for quantifying antiret- ing cART? A randomized, controlled study. Clin Infect Dis 50: 920–929.
roviral penetration into the central nervous system. Arch Neurol 65: 65–70. 17. Bansi L, Sabin C, Delpech V, Hill T, Fisher M, et al. (2010) Trends over
calendar time in antiretroviral treatment success and failure in HIV clinic
populations. HIV Med 11: 432–438.

PLOS ONE | www.plosone.org 7 April 2013 | Volume 8 | Issue 4 | e61949


The PIVOT Neurocognitive Study

18. Munderi P, Walker AS, Kityo C, Babiker AG, Ssali F, et al. (2010) Nevirapine/ Societies’ guidelines on prevention of cardiovascular disease in clinical practice.
zidovudine/lamivudine has superior immunological and virological responses Heart 91 Suppl 5: v1–52.
not reflected in clinical outcomes in a 48-week randomized comparison with 28. Carey CL, Woods SP, Rippeth JD, Gonzalez R, Moore DJ, et al. (2004) Initial
abacavir/zidovudine/lamivudine in HIV-infected Ugandan adults with low validation of a screening battery for the detection of HIV-associated cognitive
CD4 cell counts. HIV Med 11: 334–344. impairment. Clin Neuropsychol 18: 234–248.
19. Clinical Trials Website. Available at: http://clinicaltrials.gov/ct2/show/ 29. Price RW, Yiannoutsos CT, Clifford DB, Zaborski L, Tselis A, et al. (1999)
NCT01230580. Accessed 3rd April 2012. Neurological outcomes in late HIV infection: adverse impact of neurological
20. Brandt J, Benedict RHB (2001) Hopkins Verbal Learning test-Revised.: Odessa. impairment on survival and protective effect of antiviral therapy. AIDS Clinical
21. Trites R (1997) Neuropsychological test manua. Ontario. Trial Group and Neurological AIDS Research Consortium study team. Aids 13:
22. D’Elia LF (1996) Color Trails Test; Odessa, editor. 1677–1685.
23. Valcour V, Paul R, Chiao S, Wendelken LA, Miller B (2011) Screening for 30. Winston A, Puls R, Kerr S, Duncombe C, Li P, et al. (2011) Dynamics of
cognitive impairment in human immunodeficiency virus. Clin Infect Dis 53: cognitive change in HIV-infected individuals commencing three different initial
836–842. antiretroviral regimens: a randomized, controlled study. HIV Med.
24. Stavem K, Froland SS, Hellum KB (2005) Comparison of preference-based
31. Garvey L, Surendrakumar V, Winston A (2011) Low rates of neurocognitive
utilities of the 15D, EQ-5D and SF-6D in patients with HIV/AIDS. Qual Life
impairment are observed in neuro-asymptomatic HIV-infected subjects on
Res 14: 971–980.
effective antiretroviral therapy. HIV Clin Trials 12: 333–338.
25. Norman MA, Moore DJ, Taylor M, Franklin D Jr, Cysique L, et al. (2011)
32. Levine AJ, Sinsheimer JS, Bilder R, Shapshak P, Singer EJ (2011) Functional
Demographically corrected norms for African Americans and Caucasians on the
Hopkins Verbal Learning Test-Revised, Brief Visuospatial Memory Test- polymorphisms in dopamine-related genes: Effect on neurocognitive functioning
Revised, Stroop Color and Word Test, and Wisconsin Card Sorting Test 64- in HIV+ adults. J Clin Exp Neuropsychol 34: 78–91.
Card Version. J Clin Exp Neuropsychol 33: 793–804. 33. Ciccarelli N, Fabbiani M, Di Giambenedetto S, Fanti I, Baldonero E, et al.
26. Letendre S, Ellis R, Clifford D, Collier D, Gelman B, et al. (2010) Correlates of (2011) Efavirenz associated with cognitive disorders in otherwise asymptomatic
CSF viral loads in 1221 volunteers in the CHARTER Cohort; February; San HIV-infected patients. Neurology 76: 1403–1409.
Francisco, USA. 34. Gibbs JE, Gaffen Z, Thomas SA (2006) Nevirapine uptake into the central
27. British Cardiac Society, British Hypertension Society, Diabetes UK, HEART nervous system of the Guinea pig: an in situ brain perfusion study. J Pharmacol
UK, Primary Care Cardiovascular Society, et al. (2005) JBS 2: Joint British Exp Ther 317: 746–751.

PLOS ONE | www.plosone.org 8 April 2013 | Volume 8 | Issue 4 | e61949

You might also like