Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Oct 1.
Published in final edited form as: Brain Imaging Behav. 2018 Oct;12(5):1229–1238. doi: 10.1007/s11682-017-9769-9

White matter fiber bundle lengths are shorter in cART naive HIV: An analysis of quantitative diffusion tractography in South Africa

Jodi M Heaps-Woodruff a, John Joska b, Ryan Cabeen c, Laurie M Baker a, Lauren E Salminen d, Jacqueline Hoare b, David H Laidlaw c, Rachel Wamser-Nanney e, Chun-Zi Peng a, Susan Engelbrecht g, Soraya Seedat g, Dan J Stein h, Robert H Paul a
PMCID: PMC5936682  NIHMSID: NIHMS918624  PMID: 29110194

Abstract

This study examines white matter microstructure using quantitative tractography diffusion magnetic resonance imaging (qtdMRI) in HIV+ individuals from South Africa who were naïve or early in the initiation of antiretroviral therapy. Fiber bundle length (FBL) metrics, generated from qtdMRI, for whole brain and six white matter tracts of interest (TOI) were assessed for 135 HIV+ and 21 HIV− individuals. The association between FBL metrics, measures of disease burden, and neuropsychological performance were also investigated. Results indicate significantly reduced sum of whole brain fiber bundle lengths (FBL, p<0.001), but not average whole brain FBL in the HIV+ group compared to the HIV− controls. The HIV+ group exhibited significantly shorter sum of FBL in all six TOIs examined: the anterior thalamic radiation, cingulum bundle, inferior and superior longitudinal fasciculi, inferior frontal occipital fasciculus, and the uncinate fasciculus. Additionally, average FBLs were significantly shorter select TOIs including the inferior longitudinal fasciculus, cingulum bundle, and the anterior thalamic radiation. Shorter whole brain FBL sum metrics were associated with poorer neuropsychological performance, but were not associated with markers of disease burden. Taken together these findings suggest HIV affects white matter architecture primarily through reductions in white matter fiber numbers and, to a lesser degree, the shortening of fibers along a bundle path.

Introduction

HIV-positive (HIV+) individuals, even when on effective treatment for HIV, continue to experience chronic symptoms in multiple organ systems, including the brain (Nath, 2015). Many studies have shown that HIV impacts both gray and white matter in the brain, resulting in abnormalities in neuroimaging markers of brain integrity and neurocognitive impairment (Ances & Hammoud, 2014). Clinical studies reveal that the deleterious impact of HIV on brain structure and function is independent of viral clade (Bush et al, 2016), including clade C (HIV-C) which is dominant in South Africa ((de Almeida et al., 2013);(Paul et al., 2014);(Hoare et al., 2011; Joska, Fincham, Stein, Paul, & Seedat, 2010; Joska et al., 2012; Joska et al., 2011)).

Neuroimaging outcomes in HIV-C correspond with the extant literature in HIV clade B (HIV-B), indicating smaller brain volumes (Heaps et al., 2012; Ortega et al., 2013) in treatment naïve HIV+ individuals compared to uninfected controls. Smaller volumes in gray and white matter, and the thalamus correlated with worse performance on brief measures of cognitive functioning such as the international HIV dementia screen and a four item neuropsychological battery (e.g. NPZ-4) (Heaps et al., 2012; Ortega et al., 2013). Additionally, these smaller brain volumes correspond to measures of disease burden including lower CD4 cell counts and greater HIV viral load (Heaps et al., 2012; Ortega et al., 2013). In general, HIV studies of white matter using diffusion tensor imaging (DTI) show lower fractional anisotropy and increased diffusion in HIV+ individuals compared to controls (for review see McMasters and Ances, 2014 and Thompson and Jahanshad, 2015). These changes may reflect demyelination and axonal degradation (Alexander et al., 2011), are evident in early infection (Ragin, et al. 2015), in cART treated populations (Wright 2013), and may be intensified with age (Seider et al., 2016). White matter microstructure has also been examined in pediatric and adult HIV+ individuals in South Africa, with abnormal DTI scalar metrics evident in HIV+ individuals independent of the genetic polymorphism in the viral Tat protein common in clade C disease (Paul et al., 2017). Furthermore, studies have reported HIV-associated abnormal white matter microstructure in individuals with the APOE4 genotype (Hoare et al., 2013), poor prospective memory (Hoare et al., 2012), apathy (Hoare et al., 2010), and children receiving antiretroviral therapy (ART;(Hoare et al., 2015). Most recent studies in South Africa have found disruptions of structural brain networks in HIV+ individuals compared to healthy participants (Baker et al., 2017).

One limitation of structural imaging and traditional approaches to DTI metrics is the lack of detail about the organization and microstructure of white matter tracts due to data averaging at the voxel or regional level. Additionally, single tensor models only capture the directionality of tracts along the primary eigenvector, although it is possible to use multi-tensor models. Quantitative tractography diffusion magnetic resonance imaging (qtdMRI) is an alternative diffusion processing technique with high sensitivity to changes in the microstructure of white matter fibers. Fiber bundle lengths (FBLs) are computed using scalar DTI metrics, such as FA, in concert with tractography methods to delineate bundles of nerve fibers. A number of metrics are available, such as average FBL weighted by scalar DTI metrics, or a summation of FBL in the whole brain or in a given white matter tract. The metrics provide information about the underlying microstructure of white matter in better detail than previous methods. Additional information about the length, volume, and organization of fibers within a tract of interest (TOI) can be obtained using qdtMRI (Correia et al., 2008)). Previous investigations have demonstrated the clinical relevance of these metrics across several conditions (Baker et al., 2014) (Bolzenius et al., 2013) (Correia et al., 2008), but limited work has been conducted in HIV. Tate et al. utilized an early version of qtdMRI based on single fiber modeling to examine white matter integrity in 23 treatment-naive individuals with HIV-B compared to 20 HIV− controls. Results of the preliminary study revealed reduced tractography FA among HIV+ individuals at a global level, and a significant correlation between the imaging metric and worse cognitive performance. To date, no studies have examined white matter integrity in HIV-C using qtdMRI, and prior work has not utilized multi-fiber modeling which reduces limitations of single-tensor models such as partial volume effects or crossing fiber effects of prior tractography methods (Cabeen, Bastin, & Laidlaw, 2016).

The purpose of this study was to examine differences in qtdMRI FBL in both whole brain and six TOI between HIV− and HIV+ individuals with HIV-C. The TOIs were selected to allow comparison to prior DTI studies in HIV (Hoare et al., 2012; D. F. Tate et al., 2010) and other FBL studies (Baker et al., 2016; Behrman-Lay et al., 2015; Bolzenius et al., 2015). We selected average FBL and summed FBL metrics in both whole brain and selected tracts. The metrics were selected as they reflect complementary information about the underlying architecture of the white matter fibers that cannot be determined with standard DTI metrics such as FA and MD. Additionally, we chose long-range fiber bundles that connect the frontal lobe to the temportal, parietal, and occipital lobes. Average FBL reflects the average length of all the fibers within a given tract, or throughout the whole brain whereas summed FBL reflects the summation of all FBL within a tract or throughout the brain. In group comparisons, if average FBL is reduced in one group compared to the other this suggests an overall shortening of fibers. If summed FBL is reduced then it may reflect an overall loss of fibers, a change in the underlying architecture of the fiber, or reduction in fiber density. If they are both reduced, then this suggests both a loss of fibers as well as a shortening of fibers. We anticipated that HIV+ individuals would have shorter average FBL and summed FBL compared to HIV− individuals. We further examined the relationships between FBL metrics, markers of HIV disease burden (i.e., duration of infection, viral load, current CD4 cell count), and neuropsychological performance. We expected that FBL metrics would correlate with clinical laboratory variables and cognitive performance.

Methods

Participants

A total of 135 HIV+ individuals and 21 HIV− individuals completed neuroimaging at the University of Cape Town (UCT) in South Africa. All individuals consented to the protocol as approved by the ethics committee at UCT. In order to be included in the study, individuals were required to be between the ages of 18–45, speak Xhosa as their primary language, and have completed a minimum of five years of formal education. Additional inclusion criteria for the HIV+ group included: 1) Documented HIV serostatus by Elisa and confirmed by Western Blot; and 2) Naïve to (ART) or be within the first 30 days of ART initiation. Criteria for exclusion from the study for all participants were as follows: 1) Any current or past major DSM-IV Axis I psychiatric condition that could significantly affect cognitive status (e.g., schizophrenia or bipolar disorder); 2) Confounding neurological disorders including multiple sclerosis and other CNS conditions; 3) Head injury with loss of consciousness greater than 30 minutes; 4) Clinical evidence of opportunistic CNS infections (e.g., toxoplasmosis, progressive multifocal leukoencephalopathy, neoplasms); 5) Current substance use or alcohol use disorder as determined by a structured interview; and 6) Visual evidence of brain injury (e.g. stroke) on neuroimaging.

Recruitment of HIV+ participants occurred at community ART clinics in Cape Town, South Africa. Interested participants consented to participate and then completed a detailed medical and demographic survey. Study participation was voluntary and individuals were free to withdraw from the study at any point. Participants were compensated for assessments and transportation was provided for each study visit. HIV− participants were recruited from regional Voluntary Counseling and Testing Clinics in Cape Town, South Africa and were required to have laboratory-confirmed seronegative status. Recruitment from these clinics was intended to minimize differences in key demographic variables between the two groups (e.g., language, education, socioeconomic status). The study was approved by the Human Ethics Committee of the University of Cape Town, South Africa and the University of Missouri-St. Louis Institutional Review Board.

HIV Viral load, CD4+ cell counts, and duration of infection

Blood samples were collected within one week of the MRI acquisition. RNA was isolated from patient samples using the Abbott RealTime HIV-1 amplification reagent kit, according to the manufacturer’s instructions. Viral load was calculated using the Abbott m2000sp and the Abbott m2000rt analysers (Abbott laboratories, Abbott Park, Illinois, USA). CD4+ cell count was analyzed from blood samples and completed on the FACSCalibur flow cytometer in conjunction with the MultiSET V1.1.2 software (BD Biosciences, San Jose, CA, USA). Date of HIV diagnosis was available from clinic records for the majority (n=111) of HIV+ individuals

PCR amplification and sequencing of the tat exon 1 region

The tat exon 1 region was amplified (HXB2 position 5831 – 6045) by polymerase chain reaction (PCR) using the Promega GoTaq Flexi Kit (Promega, Madison, WI) according to manufacturer’s instructions. Details have been described previously (Paul et al., 2014). In brief, The PCR products were purified, sequenced and analyzed. Nucleotide sequences were used to translate info into amino acid sequences. HIV subtypes were determined to confirm HIV-C clade status for study inclusion.

Neuroimaging Acquisition

Imaging acquisition took place using a head-only Magnetom Allegra 3 T MRI scanning system (Siemens Medical Solutions, Erlangen, Germany) with a 4-channel phased-array head coil. A head stabilizer was used to minimize movement. Head placement was confirmed by a preliminary scout scan composed of three orthogonal planes. A T1-weighted magnetization-prepared rapid acquisition gradient echo (MP-RAGE) sequence was acquired using a T1-weighted three dimensional magnetization-prepared rapid acquisition gradient echo (MPRAGE) sequence (Time of repetition (TR)/inversion time (TI)/echo time (TE) = 2400/1000/2.38 milliseconds, flip angle = 8°, and voxel size = 1 × 1 × 1 mm3, as described in our initial study (Heaps et al., 2012). Additional MRI sequences were obtained, including T2-weighted scans that were used primarily for the detection of white matter disease or other injury, but were not used in the processing of DWI images.

Axial diffusion-weighted images were acquired using a customized in-house single-shot multi-slice echo-planar pulse sequence. The tensor was encoded using 30 non-collinear diffusion-encoded directions that were repeated to provide 60 diffusion weighted volumes. All directions were acquired with a gradient weight of b=1000 s/mm2. Six baseline acquisitions with a diffusion weighting of ~0 s/mm2 were also obtained and interleaved with the diffusion weighted volumes. The following parameters were used: TE- 103ms, TR 10s, matrix 128×128, FOV 218×218 mm 70 contiguous isotropic 1.7 mm3 voxels, full Fourier transform. Prior to image analysis, images were visually inspected and individuals who had evidence of white matter disease or visible lesions on T1 or T2 scans were excluded from further study participation.

Image analysis

Diffusion-weighted images (DWIs) were preprocessed using FSL 5.0 (Jenkinson, Beckmann, Behrens, Woolrich, & Smith, 2012). DWIs were corrected for motion and eddy-current induced artifacts through affine registration to the first baseline volume using FSL FLIRT (Jenkinson & Smith, 2001) with the mutual information criteria. The orientations of the gradient encoding directions were corrected by the rotation induced by these registrations (Leemans & Jones, 2009), and brain tissue was extracted using FSL brain extractions tool (BET) (Smith, 2002) with a fraction threshold of 0.45. Diffusion tensor images were estimated for each subject using FSL DTIFIT.

Following this, a study-specific white matter atlas was created using DTI-TK (Zhang et al., 2007). The template diffusion tensor image was computed by iteratively deforming and averaging the population imaging data using the tensor-based deformable registration algorithm in DTI-TK (Zhang, Yushkevich, Alexander, & Gee, 2006) with finite strain tensor reorientation and the deviatoric tensor similarity metric. This template was used to define inclusion and exclusion ROI masks for the anterior thalamic radiation (ATR), cingulum bundle (CING), the inferior frontal occipital fasciculus (IFOF), inferior longitudinal fasciculus (ILF), superior longitudinal fasciculus (SLF), and the uncinate fasciculus (UNC) (Mori & van Zijl, 2002). Whole brain tractography was performed in the template image, and subsets of curves were interactively selected to represent each TOI. For each bundle, two inclusion TOI masks and one exclusion TOI mask were drawn in template space using ITK-SNAP (Yushkevich et al., 2006). The exclusion mask was used to remove erroneous stray fibers that can occur due to limitations such as noise and voxel size (Zhang et al., 2010). These masks were placed at opposite ends of each template tractography bundle, and drawn in reference to standard white matter atlases ((Catani & Thiebaut de Schotten, 2012); (Oishi et al., 2009)).

Subject-specific fiber bundle metrics were computed as follows. First, the TOI inclusion and exclusion masks were deformed to subject native space using the DTI-TK registration. Whole brain tractography was then performed in subject native space and a subset of curves in the TOI was selected using the two inclusion and exclusion masks. Diffusion modeling for tractography used FSL XFIBRES to obtain ball-and-sticks diffusion models in each voxel (Behrens, Berg, Jbabdi, Rushworth, & Woolrich, 2007). Model fitting was performed with two stick compartments to improve tractography in areas with complex anatomy, such as crossing fibers. Tractography was performed using an extension of the standard streamline approach to use multiple fibers per voxels with the following parameters: four seeds per voxel, an angle threshold of 50 degrees, a minimum length of 10 mm, and a minimum volume fraction of 0.1. During tracking, a kernel regression estimation framework (Cabeen et al., 2016) was used for smooth interpolation of the multi-fiber ball-and-sticks models with a Gaussian kernel with a spatial bandwidth of 1.5 mm and voxel neighborhood of 7×7×7. Finally, the number of paths in each fiber bundle were calculated, then the average and sum FBL were computed from the distribution of streamline lengths in each bundle and retained for statistical analysis. Whole brain FBL-sum was calculated by summing the lengths of fiber bundles. Whole brain average FBL (FBL-avg) was calculated by averaging the lengths of fiber bundles. The FBL-avg of each TOI was calculated by computing the average length of all the fibers in the TOI. The FBL-sum of each TOI was calculated by summing the lengths of all the fibers in the TOI. Tracts of interest include: ATR, CING, IFOF, ILF, SLF, and UNC and were selected based on DTI findings in HIV (Hoare et al., 2011) as well as qtdMRI findings in other populations (Baker et al., 2014; Behrman-Lay et al., 2015; Bolzenius et al., 2013; Salminen et al., 2013). Intracranial volume was computed from the T1-weighted MRI using Freesurfer version 5.1.0 (Fischl, 2012).

Neuropsychological measures and depression screening

Neuropsychological methods are described in detail in Paul et al.(Paul et al., 2014). Briefly, a trained research technician fluent in Xhosa administered tests in the following domains: Learning/Memory, Executive Function, and Psychomotor speed (Table 1). HIV+ participants were screened for depression using either the Mini-International Neuropsychiatric Interview (M.I.N.I) (Sheehan et al., 1998) or total score on the Center for Epidemiologic Studies depression screen (CESD; (Radloff, 1991). Individuals who screened positively for clinical depression were not included in the study.

Table 1.

Neuropsychological measures

Learning/Memory
Verbal
HVLT learning
HVLT recall
Visual
BVMT learning
BVMT recall
Executive Function
Color word interference
Verbal fluency- animals
Working memory
Digit symbol
Color trails 2
Visuoconstruction/Planning
Rey-O copy
Block design
Processing speed
Trails A
Color trails 1
Open in a new tab

HVLT= Hopkins verbal learning test; BVMT= Brief visual memory test

Statistical Analysis

Analyses were conducted with SPSS 23 (IBM; New York, NY). Primary analyses utilized chi-square tests and independent t-tests to examine demographic differences between groups on age, sex, and education (see Table 2). Additionally, we employed correlations between all of the FBL metrics, age, education, sex, and intracranial volume to determine the need for covariates in subsequent analyses. All variables were checked prior to analyses for normality of distributions. Holm’s sequential Bonferroni was used to correct for multiple comparisons.

Table 2.

Demographic information for the HIV+ and HIV− groups

HIV+ n=135 HIV− n=21
Mean (SD) Median (IQR) Mean (SD) Median (IQR) p=
Age 31.81 (5.21) 31 (28–35.5) 24.24 (4.19) 24 (20–26) 0.0001
Education 10.26 (1.58) 11.14 (1.19) 0.015
Sex 81% 52% 0.0001
Current CD4 227 189 (115–315)
LogVL n=124 4.17 4.3 (3.4–4.87)
% detectable VL 100%
Duration of Infection n=111 12.62
Intracranial volume (ICV) cm3 1292.6 (261.17) 1385.0 (246.70) 0.13
Open in a new tab

Whole brain FBL-avg and FBL-sum between HIV− and HIV+ were separately examined using univariate analyses (SPSS General Linear Model Univariate procedure). In each model, HIV status was the primary independent variable with age, sex, and intracranial volume entered as covariates and whole brain FBL-avg the dependent variable in the first analysis, and whole brain FBL-sum as the dependent variable in the second analysis. Separate multivariate linear models were used (SPSS General Linear Model Multivariate procedure) to test the effect of HIV status on the six TOIs. HIV status was the primary independent variable with age, gender, and intracranial volume included as covariates (ATR, CING, IFOF, ILF, SLF, and UNC). The FBL-avg values for each TOI were dependent variables i the first analysis, with FBL-sum TOIs as dependent variables in the second analysis. Post-hoc univariate analyses, adjusted for covariates, were used to determine which TOIs differed significantly between the groups. As a supplementary analysis, the effect of HIV status (with age, sex, and ICV as covariates) on the number of FBL paths in each TOI was examined using a multivariate model with the six TOIs as dependent variables. Spearman’s correlations were completed to determine the degree of correspondence between the FBL metrics, duration of infection, current CD4 count, and viral load. CD4 cell count and viral load were natural log and log transformed, respectively, prior to analyses. Additional partial correlations controlling for age and intracranial volume were completed to inform the relationships between whole brain FBL metrics and the neuropsychological measures.

Results

Analysis of the demographic characteristics between the HIV+ and HIV− groups indicated significant group differences for age, sex, and education. The HIV+ group was older, on average (m=31.8 years old, range 22–46) and had less education (m=10.3 years) than the HIV− group (m=24.2 years old, range 20–35, and m=11.1 years of education). The HIV+ group was 81% female and the HIV− group was 52% female. All (100%) of the HIV+ group had a detectable viral load and 16.5% (n= 27) had initiated ART, with 6 of those individuals having received ART for less than 14 days. Education did not significantly correlate with any FBL measures (all rs < 0.2, ps > 0.05) and therefore we did not include education as a covariate in any of the analyses. Age, sex and intracranial volume were significantly correlated with several of the FBL measures and were therefore included as covariates in all of the FBL analyses.

HIV+ vs HIV− whole brain and tract-based FBL measures

Whole brain FBL-avg was similar between the two groups (F(4,153) =1.91, p =0.17, n2=0.012). In contrast, whole brain FBL-sum was significantly higher in the HIV− group (m=9811.4) compared to the HIV+ group (m=8028.3, F(4, 153)= 24.93, p<0.001, n2=0.14, see Table 3).

Table 3.

Shows the mean average length of the whole brain metrics for the HIV+ and HIV− group after adjustment for age, sex, and intracranial volume. Only the whole brain FBL sum metric was significantly different between the HIV+ and the HIV− group

Univariate analysis of whole brain FBL metrics
FBL average
Group Mean Std. Error 95% Confidence Interval
Lower Bound Upper Bound
HIV− 61.1 0.62 59.87 62.33
HIV+ 60.16 0.22 59.72 60.59
FBL sum
HIV− 9811456.1 327035.04 9165368.86 10457543.4
HIV+ 8028280.8 115356.43 7800383.75 8256177.85
Open in a new tab

Covariates were evaluated at the following values: sex = .23, participant_age = 30.85, intracranialvol = 1304894.91 mm3.

The multivariate analysis for tract-based FBL-avg indicated a significant multivariate effect of HIV status (Wilks’ Lambda = 0.876, F(6, 146) = 3.45, p =0.003, n2=0.12) with the HIV+ group exhibiting significantly shorter FBL-avg compared to the HIV− group. In post-hoc univariate comparisons, HIV+ individuals exhibited shorter FBL-avg in the ILF (p<0.01) and ATR (p < 0.01) after multiple comparison corrections (Holm-Bonferroni; (Holm, 1979). The multivariate model for tract-based FBL-sum revealed a significant multivariate effect of HIV on FBL-sum (Wilks’ Lambda = 0.875, F(6, 146) = 3.462, p = 0.003, n2=0.13, see Table 4). All six tracts for FBL-sum were significantly less in the HIV+ group compared to the HIV− group in each TOI (Table 4). A visualization of the tracts in a representative HIV+ subject compared to the template is provided in Figure 1.

Table 4.

Shows the mean values for each TOI for the FBL mean and FBL sum analyses

HIV+ HIV−
FBL average
TOI 95% CI 95% CI
Mean SE Lower Upper Mean SE Lower Upper p
ATR 158.0 0.7 156.6 159.4 164.1 2.0 160.2 168.0 0.005*
IFOF 273.2 1.3 270.7 275.7 280.8 3.6 273.7 287.8 0.056
ILF 184.3 1.1 182.2 186.4 198.1 3.1 192.0 204.1 <0.001*
SLF 112.9 0.9 111.0 114.7 115.6 2.6 110.4 120.8 0.341
CING 157.3 1.5 154.4 160.2 167.1 4.1 159.0 175.3 0.031*
UNC 127.8 1.5 124.9 130.7 139.1 4.1 130.9 147.2 0.013*
FBL sum
ATR 116.2 4.2 107.9 124.5 143.1 11.8 119.8 166.6 0.039*
IFOF 460.2 14.0 430.7 489.7 613.7 42.0 530.7 696.8 0.001*
ILF 284.3 9.3 265.8 302.8 349.6 26.4 297.5 401.6 0.021*
SLF 955.9 3.0 896.5 101.5 121.5 8.5 104.8 138.3 0.006*
CING 118.0 4.0 110.1 125.9 148.3 11.3 126.0 170.5 0.015*
UNC 38.4 2.0 34.4 42.4 55.1 5.7 43.8 66.3 0.008*
Open in a new tab

Covariates were evaluated at the following values: sex = .23, participant_age = 30.85, intracranialvol = 1306322.86.

*

indicates TOI significantly different between groups after correction for multiple comparisons (Holm-Sidak)

all values for FBL-sum have been adjusted by 10^-3 to reduce significant figures

CI=confidence interval; SE=standard error; TOI=tract of interest; ATR=anterior thalamic radiation

IFOF= inferior fronto-occipical fasciculus; ILF= inferior longitudinal fasciculus; SLF= superior longitudinal fasciculus; CING=cingulum bundle; UNC=uncinate fasciculus

Figure 1.

Figure 1

Open in a new tab

As a supplementary analysis, we examined the number of FB paths for the TOIs using the same multivariate model describe above with HIV status as the primary independent variable and age, gender, and intracranial volume as covariates. The HIV+ group had fewer FB paths in all of the TOI, with significant group differences noted in four of the six tracts (CING, IFOF, SLF, UNC) compared to the HIV− group (see supplementary table 1). In a follow-up analysis, the whole brain and TOI analyses were repeated with the 27 individuals who had initiated ART removed from the HIV+ group leaving 108 HIV+ participants in the analysis. The pattern of results remained the same, with the HIV+ group having shorter FBL-sum, and shorter TOI in 4 of the 6 tracts examined (see supplementary table 2 for details).

Correlations between FBL metrics, disease burden, inflammatory markers, and neuropsychological performance

Neither CD4 count, nor time since diagnosis, nor viral load correlated with any of the FBL metrics. Whole brain FBL metrics were modestly correlated with neuropsychological performance with the most robust covariance between FBL-sum and spatial learning and memory (Table 5). These correlations were repeated, excluding the 27 individuals who had initiated ART removed from the HIV+ group leaving 108 HIV+ individuals and 21 HIV− in the analysis. With the HIV+ individuals on ART removed from the analysis the relationship between the whole brain FBL metrics and neuropsychological measures changed such that visual learning was no longer significantly associated with whole brain FBL-sum. Results are detailed in Supplementary table 3.

Table 5.

Shows the values of the partial correlations between the whole brain qtdMRI metrics and neuropsychological measures controlling for age and intracranial volume. Measures that did not have a significant relationship with the qtdMRI metrics were not included in the table.

Whole Brain FBL sum
BVMT learn .32**
BVMT recall .31**
WCST per errors −.20*
Digit Symbol .15*
Verbal Fluency .21**
Block Design .20**
Open in a new tab
*

p<0.05;

**

p<0.01

BVMT= Brief Visual Memory Test, WCST=Wisconsin Card Sort Test

Discussion

This study extends the current literature examining the effect of HIV on white matter microstructure by examining the whole brain and TOI abnormalities in HIV-C. Differences were evident in whole brain FBL-sum, but not whole brain FBL-avg. Tract-specific FBL differences were also examined between the HIV+ and HIV− groups in six tracts of interest. Our results provide the first in vivo evidence of shorter FBL-sum in the ILF, SLF, ATR, IFOF, CING and UNC in HIV+ individuals. Additionally, there was an effect of HIV on the FBL-avg metric in the TOI, with the IFOF, ILF, CING and ATR tracts significantly shorter in the HIV+ group. Whole brain FBL metrics were modestly associated with neuropsychological performance, with higher FBL-sum corresponding with better performance. None of the FBL metrics were associated with markers of disease burden such as CD4 cell count, viral load, or duration of infection. These results provide seminal evidence of reduced sum of FBL in HIV whole brain and TOI, with shorter average FBL in select TOI.

FBL provides unique characterizations of white matter fibers that are meant to represent the biological lengths of fiber bundles. Shorter average FBL theoretically reflect a shorter average length of all the fibers within a bundle, or group of axons. Similarly, FBL-sum reflects the summed length of all the fibers within a bundle. Smaller FBL-sum may reflect a combination of both shortening of the fibers and/or fewer fibers within the bundle resulting from factors, such as inflammation and viral toxicity, that promote neuronal degradation (Scaravilli, Bazille, & Gray, 2007). The whole brain FBL metrics each provide unique information about the architecture of white matter fiber bundles in the brain. However, unlike biological bundle length, FBL is also likely to be sensitive to microstructural changes within bundles, to some extent, for example, when changes to the microstructure are sufficiently abnormal to change the course of tractography constructions. In the current analysis, mean differences in FBL were not observed in the whole brain measures; however, HIV+ individuals exhibited shorter FBL-avg in the ILF, and the ATR. FBL-sum was reduced in all six TOI, and our supplementary analysis of number of FBL confirmed that there were significantly fewer fiber bundles in the ATR, CING, IFOF, and UNC. Taken together, our findings suggest that HIV neuropathogenesis in HIV-C initially involves reductions in the number of white matter fibers to a greater extent than average tract length. It is also possible that the changes along a fiber bundle take place at earlier stages of disease, and ultimately result in the reduced number of fibers, resulting in smaller FBL-sum values. Future work can help to better understand these factors by examining the relation between FBL and microstructural properties such as fiber orientation dispersion and neurite density.

Reductions in FBL-avg and FBL-sum likely result from immune activation evident in untreated HIV and downstream effects on anisotropic diffusion of water along fiber bundles or myelin (Beaulieu, 2002). Studies have associated abnormalities in DTI metrics within specific tracts to cognitive deficits in the domains of processing speed, working memory, and executive functions(Biesbroek et al., 2013) as well as affective states, including apathy (Zappala, Thiebaut de Schotten, & Eslinger, 2012). Deficits in these cognitive domains are commonly reported in HIV (Heaton et al., 2010) (Paul et al., 2005; D. Tate et al., 2003). The results of this study show an association between shorter whole brain FBL sum and poorer performance on measures of learning, memory, visuoconstruction, verbal fluency and executive function. However, the strength of the relationships is limited, and the cross-sectional nature of this study does not inform long-term outcomes after sustained ART with viral suppression. Since the results of this study indicate that reduced FBL sum is associated with poorer neuropsychological performance there is an indication that this is due to fewer FBL, as opposed to shorter fiber bundle length due to the lack of correlation between FBL average and the neuropsychological measures. It is also possible that adequate connections have atrophied, or become “disconnected” resulting in poorer neuropsychological performance.

One longitudinal study of DTI in HIV reported significant changes in diffusion metrics indicating progressive white matter and cognitive decline in an ART stable HIV+ group after one year (Chang et al., 2008). Conversely, a recent longitudinal study of HIV+ individuals on stable highly-active ART did not find changes to DTI scalar metrics after approximately 26 months, but cognitive function was not assessed (Correa et al., 2016). Additional studies will be critical to better understand whether there is potential for recovery of FBL through therapeutic interventions, and how FBL changes affect cognition and daily living functions.

There are several limitations to our study that warrant discussion. This study involved a cross-sectional design and as such we cannot determine the progression of FBL changes over time. Longitudinal studies will be key to understanding how FBL can change across time, with ART or other conjunctive therapies to protect the brain. A number of individuals in this study had recently initiated ART therapy (within 30 days of their MRI), which may impact the degree of correspondence between markers of CD4 and viral load to FBL. A relatively small number of HIV− controls were available for the study; however, there was sufficient power to detect strong differences between groups using FBL as a measure of white matter integrity. Untreated HIV has been associated with the development of white matter disease, and our exclusion of individuals with overt white matter disease may not be fully representative of individuals with untreated HIV. As indicated in our supplementary analysis, it is possible that ART initiation in the small percentage of participants in the study influenced the relationship between cognitive performance and FBL-sum TOI metrics. However, duration of treatment was under 30 days and all of the participants had detectable viral loads. Therefore, it is unlikely that ART initiation in this subset significantly affected the outcomes.

Our results confirm abnormalities in FBL-sum in whole brain and TOI among HIV+ individuals ART in South Africa. These abnormalities indicate that long-range fiber bundles connecting the frontal lobe to each the temporal, parietal, and occipital lobes have shorter summed FBL, on average, than HIV− individuals. Additionally, the results suggest global measurements of FBL-sum may be more powerful to detect white matter abnormalities in HIV than FBL-avg. Shorter sum of FBL indicates either reduced FBL fiber volume, or fibers that have shortened along the anatomical tract. However, a significant whole brain difference in FBL-avg was not evident; therefore, it is likely that a reduction in the number of fibers are driving the FBL-sum metric lower in HIV, and not shortening of fibers along a path. Given that the study population was naïve to ART or had newly initiated ART future work will be critical to understanding how stable treatment with ART may affect FBL metrics and neuropsychological performance. This study provides a snapshot of white matter integrity, and future studies following individuals over time are necessary to understand the trajectory of brain health using qtdMRI and potential legacy effects of ART in HIV. Furthermore, there are a number of host and viral factors (e.g. resilience, inflammatory markers, and HIV co-receptor tropism) that may affect how FBL change in HIV that remain to be examined.

Supplementary Material

11682_2017_9769_MOESM1_ESM

Supplementary Table 1 shows the calculated number of fibers in each TOI indicating that in all tracts with the exception of the SLF the HIV+ group had fewer number of FBL in the TOIs compared to the HIV− group

Supplementary Table 2- Results of Multivariate analysis with 27 individuals on ART removed from HIV+ group

Supplementary Table 3- Results of partial correlation with HIV+ on ART removed from analysis

Acknowledgments

Funding for this study was provided by the National Institute of Mental Health to Robert H. Paul R01MH085604

Footnotes

Conflicts of Interest: Authors have no conflicts of interest to declare

Ethical approval: All procedures were approved by the institutional review boards for human subjects’ research at the University of Missouri-St. Louis and the University of Cape Town in accordance with the ethical standards national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent: Informed consent was obtained from all individual participants included in the study.

References

  1. Alexander, et al. Characterization of Cerebral White Matter Properties Using Quantitative Magnetic Resonance Imaging Stains. Brain Connectivity. 2011;1(6):432–446. doi: 10.1089/brain.2011.0071. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Baker LM, Cooley SA, Cabeen RP, Laidlaw DH, Joska JA, Hoare J, … Paul RH. Topological Organization of Whole-Brain White Matter in HIV Infection. Brain Connect. 2017;7(2):115–122. doi: 10.1089/brain.2016.0457. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Baker LM, Laidlaw DH, Cabeen R, Akbudak E, Conturo TE, Correia S, … Paul RH. Cognitive reserve moderates the relationship between neuropsychological performance and white matter fiber bundle length in healthy older adults. Brain Imaging Behav. 2016 doi: 10.1007/s11682-016-9540-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Baker LM, Laidlaw DH, Conturo TE, Hogan J, Zhao Y, Luo X, … Paul RH. White matter changes with age utilizing quantitative diffusion MRI. Neurology. 2014;83(3):247–252. doi: 10.1212/WNL.0000000000000597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Beaulieu C. The basis of anisotropic water diffusion in the nervous system - a technical review. NMR Biomed. 2002;15(7–8):435–455. doi: 10.1002/nbm.782. [DOI] [PubMed] [Google Scholar]
  6. Behrens TE, Berg HJ, Jbabdi S, Rushworth MF, Woolrich MW. Probabilistic diffusion tractography with multiple fibre orientations: What can we gain? Neuroimage. 2007;34(1):144–155. doi: 10.1016/j.neuroimage.2006.09.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Behrman-Lay AM, Usher C, Conturo TE, Correia S, Laidlaw DH, Lane EM, … Paul RH. Fiber bundle length and cognition: a length-based tractography MRI study. Brain Imaging Behav. 2015;9(4):765–775. doi: 10.1007/s11682-014-9334-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Biesbroek JM, Kuijf HJ, van der Graaf Y, Vincken KL, Postma A, Mali WP, … Group SS. Association between subcortical vascular lesion location and cognition: a voxel-based and tract-based lesion-symptom mapping study. The SMART-MR study. PLoS One. 2013;8(4):e60541. doi: 10.1371/journal.pone.0060541. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bolzenius JD, Laidlaw DH, Cabeen RP, Conturo TE, McMichael AR, Lane EM, … Paul RH. Impact of body mass index on neuronal fiber bundle lengths among healthy older adults. Brain Imaging Behav. 2013;7(3):300–306. doi: 10.1007/s11682-013-9230-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Bolzenius JD, Laidlaw DH, Cabeen RP, Conturo TE, McMichael AR, Lane EM, … Paul RH. Brain structure and cognitive correlates of body mass index in healthy older adults. Behav Brain Res. 2015;278:342–347. doi: 10.1016/j.bbr.2014.10.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cabeen RP, Bastin ME, Laidlaw DH. Kernel regression estimation of fiber orientation mixtures in diffusion MRI. Neuroimage. 2016;127:158–172. doi: 10.1016/j.neuroimage.2015.11.061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Catani M, Thiebaut de Schotten M. Atlas of human brain connections. Oxford; New York: Oxford University Press; 2012. [Google Scholar]
  13. Chang L, Wong V, Nakama H, Watters M, Ramones D, Miller EN, … Ernst T. Greater than age-related changes in brain diffusion of HIV patients after 1 year. J Neuroimmune Pharmacol. 2008;3(4):265–274. doi: 10.1007/s11481-008-9120-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Correa DG, Zimmermann N, Tukamoto G, Doring T, Ventura N, Leite SC, … Gasparetto EL. Longitudinal assessment of subcortical gray matter volume, cortical thickness, and white matter integrity in HIV-positive patients. J Magn Reson Imaging. 2016 doi: 10.1002/jmri.25263. [DOI] [PubMed] [Google Scholar]
  15. Correia S, Lee SY, Voorn T, Tate DF, Paul RH, Zhang S, … Laidlaw DH. Quantitative tractography metrics of white matter integrity in diffusion-tensor MRI. Neuroimage. 2008;42(2):568–581. doi: 10.1016/j.neuroimage.2008.05.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. de Almeida SM, Ribeiro CE, de Pereira AP, Badiee J, Cherner M, Smith D, … Ellis RJ. Neurocognitive impairment in HIV-1 clade C- versus B-infected individuals in Southern Brazil. J Neurovirol. 2013;19(6):550–556. doi: 10.1007/s13365-013-0215-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Fischl B. FreeSurfer. Neuroimage. 2012;62(2):774–781. doi: 10.1016/j.neuroimage.2012.01.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Heaps JM, Joska J, Hoare J, Ortega M, Agrawal A, Seedat S, … Paul R. Neuroimaging markers of human immunodeficiency virus infection in South Africa. J Neurovirol. 2012;18(3):151–156. doi: 10.1007/s13365-012-0090-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Heaton RK, Clifford DB, Franklin DR, Jr, Woods SP, Ake C, Vaida F, … Group C. HIV-associated neurocognitive disorders persist in the era of potent antiretroviral therapy: CHARTER Study. Neurology. 2010;75(23):2087–2096. doi: 10.1212/WNL.0b013e318200d727. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hoare J, Fouche JP, Phillips N, Joska JA, Donald KA, Thomas K, Stein DJ. Clinical associations of white matter damage in cART-treated HIV-positive children in South Africa. J Neurovirol. 2015;21(2):120–128. doi: 10.1007/s13365-014-0311-1. [DOI] [PubMed] [Google Scholar]
  21. Hoare J, Fouche JP, Spottiswoode B, Joska JA, Schoeman R, Stein DJ, Carey PD. White matter correlates of apathy in HIV-positive subjects: a diffusion tensor imaging study. J Neuropsychiatry Clin Neurosci. 2010;22(3):313–320. doi: 10.1176/appi.neuropsych.22.3.313. doi: 10.1176/appi.neuropsych.22.3.313. [DOI] [PubMed] [Google Scholar]
  22. Hoare J, Fouche JP, Spottiswoode B, Sorsdahl K, Combrinck M, Stein DJ, … Joska JA. White-Matter damage in Clade C HIV-positive subjects: a diffusion tensor imaging study. J Neuropsychiatry Clin Neurosci. 2011;23(3):308–315. doi: 10.1176/appi.neuropsych.23.3.308. doi: 10.1176/appi.neuropsych.23.3.308. [DOI] [PubMed] [Google Scholar]
  23. Hoare J, Westgarth-Taylor J, Fouche JP, Combrinck M, Spottiswoode B, Stein DJ, Joska JA. Relationship between apolipoprotein E4 genotype and white matter integrity in HIV-positive young adults in South Africa. Eur Arch Psychiatry Clin Neurosci. 2013;263(3):189–195. doi: 10.1007/s00406-012-0341-8. [DOI] [PubMed] [Google Scholar]
  24. Hoare J, Westgarth-Taylor J, Fouche JP, Spottiswoode B, Paul R, Thomas K, … Joska J. A diffusion tensor imaging and neuropsychological study of prospective memory impairment in South African HIV positive individuals. Metab Brain Dis. 2012;27(3):289–297. doi: 10.1007/s11011-012-9311-0. [DOI] [PubMed] [Google Scholar]
  25. Holm S. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics. 1979;6(2):65–70. [Google Scholar]
  26. Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith SM. Fsl. Neuroimage. 2012;62(2):782–790. doi: 10.1016/j.neuroimage.2011.09.015. [DOI] [PubMed] [Google Scholar]
  27. Jenkinson M, Smith S. A global optimisation method for robust affine registration of brain images. Med Image Anal. 2001;5(2):143–156. doi: 10.1016/s1361-8415(01)00036-6. [DOI] [PubMed] [Google Scholar]
  28. Joska JA, Fincham DS, Stein DJ, Paul RH, Seedat S. Clinical correlates of HIV-associated neurocognitive disorders in South Africa. AIDS Behav. 2010;14(2):371–378. doi: 10.1007/s10461-009-9538-x. [DOI] [PubMed] [Google Scholar]
  29. Joska JA, Westgarth-Taylor J, Hoare J, Thomas KG, Paul R, Myer L, Stein DJ. Neuropsychological outcomes in adults commencing highly active anti-retroviral treatment in South Africa: a prospective study. BMC Infect Dis. 2012;12:39. doi: 10.1186/1471-2334-12-39. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Joska JA, Westgarth-Taylor J, Myer L, Hoare J, Thomas KG, Combrinck M, … Flisher AJ. Characterization of HIV-Associated Neurocognitive Disorders among individuals starting antiretroviral therapy in South Africa. AIDS Behav. 2011;15(6):1197–1203. doi: 10.1007/s10461-010-9744-6. [DOI] [PubMed] [Google Scholar]
  31. Leemans A, Jones DK. The B-matrix must be rotated when correcting for subject motion in DTI data. Magn Reson Med. 2009;61(6):1336–1349. doi: 10.1002/mrm.21890. [DOI] [PubMed] [Google Scholar]
  32. Mori S, van Zijl PC. Fiber tracking: principles and strategies - a technical review. NMR Biomed. 2002;15(7–8):468–480. doi: 10.1002/nbm.781. [DOI] [PubMed] [Google Scholar]
  33. Oishi K, Faria A, Jiang H, Li X, Akhter K, Zhang J, … Mori S. Atlas-based whole brain white matter analysis using large deformation diffeomorphic metric mapping: application to normal elderly and Alzheimer’s disease participants. Neuroimage. 2009;46(2):486–499. doi: 10.1016/j.neuroimage.2009.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Ortega M, Heaps JM, Joska J, Vaida F, Seedat S, Stein DJ, … Ances BM. HIV clades B and C are associated with reduced brain volumetrics. J Neurovirol. 2013;19(5):479–487. doi: 10.1007/s13365-013-0202-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Paul RH, Brickman AM, Navia B, Hinkin C, Malloy PF, Jefferson AL, … Flanigan TP. Apathy is associated with volume of the nucleus accumbens in patients infected with HIV. J Neuropsychiatry Clin Neurosci. 2005;17(2):167–171. doi: 10.1176/jnp.17.2.167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Paul RH, Joska JA, Woods C, Seedat S, Engelbrecht S, Hoare J, … Stein DJ. Impact of the HIV Tat C30C31S dicysteine substitution on neuropsychological function in patients with clade C disease. J Neurovirol. 2014;20(6):627–635. doi: 10.1007/s13365-014-0293-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Paul RH, Phillips S, Hoare J, Laidlaw DH, Cabeen R, Olbricht GR, … Joska J. Neuroimaging abnormalities in clade C HIV are independent of Tat genetic diversity. J Neurovirol. 2017;23(2):319–328. doi: 10.1007/s13365-016-0503-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Radloff LS. The use of the Center for Epidemiologic Studies Depression Scale in adolescents and young adults. J Youth Adolesc. 1991;20(2):149–166. doi: 10.1007/BF01537606. [DOI] [PubMed] [Google Scholar]
  39. Salminen LE, Schofield PR, Lane EM, Heaps JM, Pierce KD, Cabeen R, … Paul RH. Neuronal fiber bundle lengths in healthy adult carriers of the ApoE4 allele: a quantitative tractography DTI study. Brain Imaging Behav. 2013;7(3):274–281. doi: 10.1007/s11682-013-9225-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Scaravilli F, Bazille C, Gray F. Neuropathologic contributions to understanding AIDS and the central nervous system. Brain Pathol. 2007;17(2):197–208. doi: 10.1111/j.1750-3639.2007.00047.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, … Dunbar GC. The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J Clin Psychiatry. 1998;59(Suppl 20):22–33. quiz 34–57. [PubMed] [Google Scholar]
  42. Smith SM. Fast robust automated brain extraction. Hum Brain Mapp. 2002;17(3):143–155. doi: 10.1002/hbm.10062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Tate D, Paul RH, Flanigan TP, Tashima K, Nash J, Adair C, … Cohen RA. The impact of apathy and depression on quality of life in patients infected with HIV. AIDS Patient Care STDS. 2003;17(3):115–120. doi: 10.1089/108729103763807936. [DOI] [PubMed] [Google Scholar]
  44. Tate DF, Conley J, Paul RH, Coop K, Zhang S, Zhou W, … Tashima K. Quantitative diffusion tensor imaging tractography metrics are associated with cognitive performance among HIV-infected patients. Brain Imaging Behav. 2010;4(1):68–79. doi: 10.1007/s11682-009-9086-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Yushkevich PA, Piven J, Hazlett HC, Smith RG, Ho S, Gee JC, Gerig G. User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability. Neuroimage. 2006;31(3):1116–1128. doi: 10.1016/j.neuroimage.2006.01.015. [DOI] [PubMed] [Google Scholar]
  46. Zappala G, Thiebaut de Schotten M, Eslinger PJ. Traumatic brain injury and the frontal lobes: what can we gain with diffusion tensor imaging? Cortex. 2012;48(2):156–165. doi: 10.1016/j.cortex.2011.06.020. [DOI] [PubMed] [Google Scholar]
  47. Zhang H, Avants BB, Yushkevich PA, Woo JH, Wang S, McCluskey LF, … Gee JC. High-dimensional spatial normalization of diffusion tensor images improves the detection of white matter differences: an example study using amyotrophic lateral sclerosis. IEEE Trans Med Imaging. 2007;26(11):1585–1597. doi: 10.1109/TMI.2007.906784. [DOI] [PubMed] [Google Scholar]
  48. Zhang H, Yushkevich PA, Alexander DC, Gee JC. Deformable registration of diffusion tensor MR images with explicit orientation optimization. Med Image Anal. 2006;10(5):764–785. doi: 10.1016/j.media.2006.06.004. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

11682_2017_9769_MOESM1_ESM

Supplementary Table 1 shows the calculated number of fibers in each TOI indicating that in all tracts with the exception of the SLF the HIV+ group had fewer number of FBL in the TOIs compared to the HIV− group

Supplementary Table 2- Results of Multivariate analysis with 27 individuals on ART removed from HIV+ group

Supplementary Table 3- Results of partial correlation with HIV+ on ART removed from analysis

NIHMS918624-supplement-11682_2017_9769_MOESM1_ESM.docx (26.5KB, docx)

RESOURCES