The experiments of this study were conducted at Beijing You An Hospital, the Capital Medical Hospital. The study received approval from the ethical committee of the Beijing Municipal Science and Technology Commission, the University of Science and Technology of China, and Wenzhou Medical University. The study complied with the code of ethics of the World Medical Association for animal experiments. The study involved ten male rhesus Monkeys of Chinese origin. We initially evaluated our animals to determine their health status. We conducted an indirect immunoassay (IFA) to identify and exclude the rhesus monkeys with simian immunodeficiency virus (SIV), Simian type D retrovirus (SRV), herpes B virus, or Simian T cell lymphotropic virus (STLV-1). All rhesus monkeys were in good health at the time of recruitment. We housed our rhesus Monkeys at the Institute of Laboratory Animal Sciences, Peking Union Medical College, while maintaining the optimal conditions—a temperature range of 16 to 26 degrees Celsius (°C), a humidity level between 40% and 70%, and a light/dark cycle of 12 hours each (light/dark =12hr/12hr). We gave the rhesus monkeys food twice a day without any restrictions on their diet, and we exposed them to unrestricted water access (water ad libitum).

Figure 1 shows a full experimental setup and design, including regimen combinations, laboratory tests involved, housing conditions, and procedures during blood/CSF/MRI data collections. The details of the specific days or weeks on which each examination was performed are summarized in Table 1 . Briefly, to accomplish our objective, which involves the identification of early inflammatory-related changes in the extracellular environment following SIVmac239 infection and the potential restorative effects of the TDF+EMTBL+DTG regimen, we designed and conducted a series of follow-up experiments for 48 weeks (336 days). We acquired MRI images at 7 stages, i.e., at the baseline, and on day 10, day 28, day 84, day 168, day 252, and day 336 following SIVmac239 infection ( Figure 1B ). MRI imaging closely followed the experiments for blood sample collection and assays, which were conducted at 7 stages as well, i.e., at the baseline, and on day 7, day 35, day 84, day 168, day 252, and day 336 after infection. Step by step description of the experiments is as follows. Briefly, at the baseline, i.e., before injecting rhesus monkeys with SIVmac239 dose, we set protocols for data collection. We collected blood samples from the forelimbs (specifically, from the brachial and inguinal veins) and the hindlimbs (specifically, from the saphenous vein). From the blood samples, we quantified the primary biomarkers of immunity levels. We first quantified the number of CD4 T-cells present in 1 μl of the blood collected from the same site for each monkey. We next quantified the number of CD8 T-cells present in 1 μl of the blood. We later calculated the ratio of the number of CD4 T-cells to the number of CD8 T-cells in 1 μl of the blood— this ratio reflects the stability of the immune system and is commonly referred to as the CD4/CD8 ratio. We then gave all monkeys intramuscular injections of anesthetic drugs, Ketamine hydrochloride (5 -10 ml) —and scanned their brains with T1-and diffusion-weighted protocols.

Neuroimaging of the brains of rhesus monkeys was performed on a Siemens 3T MRI Scanner (Siemens, Tim TRIO, and Germany Erlangen, Germany) at Beijing You An Hospital, the capital Medical hospital. The number of MRI scanning experiments conformed to the details shown in Figure 1B . We set protocols for the two imaging sequences—3D-T1-weighted anatomical and diffusion-weighted images. We used the following settings for 3D-T1-weighted imaging: TR/TE = 1,800 ms/4.12 ms, field of view (FOV) = 18.9 mm × 18.9 mm x 7.2 mm, Data matrix (matrix size) = 192 × 192 x 72, flip Angle = 90, slice thickness = 1 mm, and voxel size =0.984 mm × 0.984 mm × 1 mm; whereas we set the following parameters for diffusion-weighted imaging: Echo planar, 120 diffusion-encoded (60, b = 1,000 s/mm2; 60, b = 2,000 s/mm2), 2 references (b = 0 s/mm2), TR/TE = 5200 ms/100 ms, field of view (FOV) = 15.2 mm × 15.2 mm x 8.0 mm, Data matrix (matrix size) = 76 x 76 x 40, flip Angle = 900, slice thickness =2 mm, and voxel size =2 mm × 2 mm × 2 mm; time points = 122 volumes; oblique = 15.8720.

Macaque MRI data processing was in accordance with earlier studies (6, 17). We first preprocessed the data in the procedures that involve five steps. We began by extracting the monkey brains from the raw T1-WI and DWI data using three different approaches: (1) ‘3dSkullStrip’ tool with ‘monkey’ flag implemented in AFNI (AFNI, v23.1.08, https://afni.nimh.nih.gov/) (18); (2) ‘bet4animal’ tool implemented in FMRIB Software Library (Oxford, UK, FSL, v6.0.6.6, http://www.fmrib.ox.ac.uk/fsl/); and (3) ‘U-net model for brain extraction’ tool implemented in DeepBet-U-Net (DeepBetv1.0, https://github.com/HumanBrainED/NHP-BrainExtraction) (19). DeepBet extracts the brains of non-human primates using deep learning models. These models previously received initial training on human brains, and they were later updated with the brains of non-human primates by Wang et al. We further cross-trained one of the models with the brains of our rhesus monkeys. Here we used the ‘Site-All-T-epoch-36-update-with-Site-6-plus-7-epoch-09’ model, which was initially updated by Wang et al. with brains of 19 macaques from 13 sites specified by PRIDE-DE. Our two experts, Dr. BA. N and Dr. Y.L, conducted a thorough visual inspection for quality assessment of brain extraction performed by each of the three approaches. They then voted for the approach with the best performance on our data. Experts found DeepBet’s results sufficient for subsequent processing steps due to better smoothing and fine-area delineation (See Supplementary Figure S1A for the results of AFNI-3dSkullStrip and FSL-bet4animal, and Supplementary Figure S1B for the results of DeepBet-U-Net). We next corrected for eddy-current induced distortion and head-motion artifact using FSL’s ‘eddy-current’ tool (20). Before computing for water molecules (FW) diffusing freely in the extracellular space, we first fitted the DWI data using the standard FLS’s ‘DTI-FIT’ to obtain DTIFIT-derived raw FA maps. We then non-linearly registered the DTIFIT-derived raw FA map to a standard space defined by UWDTIRhesusWM template, using FSL’s registration tools (FLIRT/FNIRT). Then, the multi shell DWIs data encoded in 60 directions for b = 1000 s/mm2 and 60 directions for b = 2000 s/mm2 were again put into Python code to calculate FW, and FA and MD corrected for FW (FW-FA, FW-MD). Briefly, our code uses a bi-tensor model to split a single voxel information into extracellular FW and tissue compartments. We particularly used an optimized version of the model proposed by Hoy et al. (11), originally described by Pasternak et al. (21) in its expanded form: S ( g , b ) = S 0 ( 1 − f ) e − b g T D g + S 0 f e − b D i s o Where g and b are diffusion gradient direction and weighted, S(g,b) is the diffusion-weighted signal measured, S 0 is the signal in a measurement with no diffusion weighting, D is the diffusion tensor, f is the volume fraction of the free water component, and D i s o is the isotropic value of the free water diffusion(normally set to 3.0   t i m e s   10 − 3 m m 2 s − 1 The generated new metrics, FW-VF, FA-FW, and MD-FW, were further transformed into standard space using the transformation matrix obtained during the transformation of the DTI-derived raw FA map to a standard space defined by UWDTIRhesusWM template. Our two experienced neurologists (X.W. and B.N., with 10 and 6 years of experience, respectively) visually inspected our data for registration quality validation. It is important to note that the UWDTIRhesusWM template used here is rhesus macaque white matter based, developed using high-quality brain scans of 271 rhesus monkeys (https://www.nitrc.org/projects/rmdtitemplate/) (22). The template accounts for signal and spatial variability across species. It also addresses the previously identified challenges of low signal contrast, low signal-to-noise ratio, low sharpness, and low visibility of fine white matter structures and spatial features of other templates (22).

Briefly, all 10 rhesus monkeys were male, of Chinese origin, and aged 3.8 ± 0.3 years. At the time of recruitment, the rhesus monkeys were all healthy, with an average weight of 4.72 ± 0.58 kg. The ranges of their CD4 T-cell count, CD8 T-cell count, and CD4/CD8 ratio were 566—2342, 235—881, 0.97—3.0, respectively.

Before injections of TCID5O for SIVmac239 infection, the average scores of CD4 T-cell count, CD8 T-cell count, and CD4/CD8 ratio were 1120.79 ± 514.25, 605.81 ± 201.47, and 1.96 ± 0.78, respectively. After 7 days of TCID5O injection, the SIVmac239 transformed into SIV infection. The number of SIV copies per ml increased from 0 to 6.89 ± 0.44 on day 7 (P < 0.0001, in plasma, Table 2 ) and continued to rise to 6.92 ± 0.40 on day 14 (P < 0.0001, in CSF, Table 2 ) and to 7.17 ± 0.51 on day 34 (P < 0.0001, in plasma). For CD4 T-cells ( Table 3 ), there was no significant change in their number by day 7 (1187.68 ± 468.95 cells/µL, paired t-test, P = 0.6000). A statistically significant change in CD4 T-cells was first detected on day 14; the CD4 T-cells dropped to 756.29 ± 404.44 cells/µL (P = 0.0022). As the disease progressed to day 34, the CD4 T-cells continued to remain below < 1000 (837.66 ± 351.59 cells/µL, day 34, P = 0.0495). The CD8 T-cells, on the other hand, began to manifest change by day 7; they increased from 605.81 ± 201.47 cells/µL at baseline to 871.01 ± 467.53 cells/µL, although this increase was not statistically significant (P = 0.0907). By day 14, the change in CD8 T-cells was more evident, increasing to 1012.38 ± 237.49 cells/µL, P = 0.0010). This increase in CD8 T-cells persisted through day 34 (1114.42 ± 234.84 cells/µL, P = 0.0009) as the viral infection continued. For the CD4/CD8 ratio, we started to observe a trend of decline by day 7; the ratio decreased to 1.65 ± 0.93 from 1.96 ± 0.78 at baseline. However this change was not statistically significant (P = 0.1782). An intensified change in the CD4/CD8 ratio for our rhesus monkeys manifested on day 14 (dropped to 0.78 ± 0.40, P < 0.0001)—and this reduction in CD4/CD8 ratio progressed as the disease advanced (0.77 ± 0.36, P < 0.0001, day 34). These results suggest that a significant reduction in immunity levels—CD4 T-cells and CD4/CD8 ratio—begins as early as the first 14 days (two weeks) of infection and persists as the disease advances. The reduction in immunity levels and the increase in CD8 T-cell proliferation and differentiation coincided with a rise in the number of SIV copies.

On day 40 of SIVmac239 infection, 5 of 10 rhesus monkeys were initiated by injection and through diet with suppressive cART drugs [FTC/EMTBL+TDF+ DTG]. Five days before cART initiation (day 40 of SIV infection), these specific five rhesus monkeys had an average CD4 T-cell count of 742.53 ± 270.44, a CD8 T-cell count of 1062.66 ± 209.93, and a CD4/CD8 ratio of 0.69 ± 0.15. Their average number of SIV copies per ml was 7.38 ± 0.45 in plasma and 5.85 ± 0.44 in CSF. After treatment, the viral suppressive effects of our regimens began to manifest within the first 44 days of progressive cART (see Table 2 , bottom panel). By day 44(50) of the suppressive drug treatment, the number of SIV copies significantly declined; the number of plasma SIV copies reduced by 75.09%, dropping from 7.38 ± 0.45 to 3.34 ± 1.33, P = 0.0016). The SIV copies in CSF dropped as well from 5.85 ± 0.44 to 2.95 ± 1.39 (P = 0.0043). Our drug suppressive effects continued progressively over time. In short, 60% (3 of 5), 80% (4 of 5), and 100% (5 of 5) of rhesus macaques achieved a total virosuppression by day 44, day 128, and day 212, respectively. For immunity levels, we detected the immune-boosting effects of our drugs by day 16 of suppressive drug treatment (see Table 3 ). The CD4/CD8 ratio of the treated rhesus monkeys increased from 0.69 ± 0.15 to 1.4 ± 0.48 (P = 0.0100) by day 16 of suppressive drug treatment and continued to improve, reaching 1.9 ± 0.56 by day 212 (P = 0.0034). Meanwhile, the number of CD8 T cells was significantly reduced to 527.24 ± 125.96 (P = 0.0088) by day 16 and to 486.6 ± 149.49 (P = 0.0208) by day 212, from 1062.66 ± 209.93 before cART initiation. (See Tables 2 , 3 for statistics on the progression of immunity levels and SIV viral copies as the disease progresses, as well as with cART intervention; also see Figure 2 ; Supplementary Table S1 for overall comparisons before and after treatment).

On the other hand, the rhesus monkeys (5 out of 10) that did not receive cART intervention on day 40 continued to have elevated SIV copies on later days, similar to the levels seen on day 35 or earlier. On day 35, the plasma levels of SIV copies were 6.97 ± 0.53 cells/µL (P = 0.4401 vs. Baseline) and persisted on the follow-up days: day 90 (6.72 ± 0.65 cells/µL, P = 0.4401 vs. day 34), day 173 (6.99 ± 0.76 cells/µL, P = 0.9282 vs. day 34), day 256 (6.96 ± 0.95 cells/µL, P = 0.9957 vs. day 35), and day 340 (7.14 ± 0.61 cells/µL, P = 0.5349 vs. day 34). These untreated rhesus monkeys continued to experience severely worsened immunity levels as the disease advanced. Their CD4/CD8 ratio remained low(< 1) during the early days of infection, i.e., 0.76± 0.49 on day 14 (P = 0.0027 vs. baseline) and 0.84± 0.51 on day 34 (P = 0.0053 vs. baseline) —and continued to be low in later days: 1.07 ± 0.61 on day 56 (P = 0.0038 vs. baseline; P = 0.0170 vs. day 34), 0.91 ± 0.47 on day 90 (P = 0.0039 vs. baseline; P = 0.3845 vs. day 34), 0.83 ± 0.50 on day 173 (P = 0.0061 vs. baseline; P = 0.9454 vs. day 34), 0.97 ± 0.89 on day 256(P = 0.0177 vs. baseline; P = 0.5406 vs. day 34), and 1.1 ± 0.97 on day 340 (P = 0.0819 vs. baseline; P = 0.3318 vs. day 34). The number of CD4 T-cells declined progressively (below 1000) from the baseline (1120.01 ± 687.63) as the disease advanced. By day 34, the CD4 T-cells dropped to 932.78± 0.427(P = 0.0974 vs. baseline). After day 40 (when those in the other group were receiving cART), the CD4 T-cells of the untreated rhesus monkeys continued to decline to 753.65± 278.05 (P = 0.0688 vs. baseline; P = 0.2454 vs. day 34), 626.71 ± 273.24 (P = 0.0365 vs. baseline; P = 0.0833 vs. day 34), 745.97 ± 283.02 (P = 0.0943 vs. baseline; P = 0.3200 vs. day 34), 393.45 ± 202.53 (P = 0.0451 vs. baseline; P = 0.0568 vs. day 34), and 489 ± 384.1 cells/µL (P = 0.0147 vs. baseline; P = 0.0343 vs. day 34) by day 56, day 90, day 173, and day 340, respectively.

Before the injection of TCID5O, the average weight rhesus monkeys was 4.72 ± 0.58 kg. After 14 days of infection, they lost 6% of their weight (from 4.72 ± 0.58 to 4.45 ± 0.55 kg, P = 0.0145, Paired t-test, see Supplementary Table S2 ). By day 28 of progressive diet, they regained the 6% of the lost weight (to 4.68 ± 0.58, P = 0.6882). Over time, the weight continued to increase (to 4.81 ± 0.781 by day 96). We did not find strong evidence of group effects on weights after the cART intervention (P > 0.5); both rhesus monkeys, with and without cART, gained weight over time. However, those who did not receive cART intervention demonstrated a more pronounced weight gain trend than those who received it, especially from days 14 to 28 (cART-, 4.64 ± 0.72 to 4.9 ± 0.75, P = 0.0004; cART+, 4.26 ± 0.26 to 4.46 ± 0.26, P = 0.0217) and days 84 to 168 (cART-, 4.92 ± 0.65 to 5.28 ± 0.75, P = 0.0288; cART+, 4.18 ± 0.29 to 4.33 ± 0.49, P = 0.2344).

As early as within the first 28 days of SIV239 infection, both the extracellular water volumes and the diffusion of fiber tissues of the rhesus monkeys demonstrated changes (see Table 4 ). Briefly, free water volumes increased significantly for extracellular spaces surrounding the fiber tissues of the internal capsule (anterior, posterior and retrolenticular limbs), corona radiata (right and left anterior), thalamic radiata (right posterior), sagittal striatum (right), external capsule (right and left), longitudinal fasciculus (right and left superior), uncinate fasciculus(right), prefrontal WM(right and left of both dorsal and ventral parts), frontal gyrus WM(left inferior), temporal gyrus WM(left superior), and cingulate WM(left anterior). This increase was more pronounced (survived FDR-corrections for multiple comparisons) in the internal capsule (left anterior limb), corona radiata (left anterior), thalamic radiation (right posterior), external capsule (left), longitudinal fasciculus (left superior), uncinate fasciculus (right), and prefrontal WM (left ventral). Exceptionally, for the cerebellar peduncles (right and left, inferior and superior) and the tapetum (right), the extracellular FW volume decreased. An intensified decrease (surviving FDR corrections for multiple comparisons) was observed in the right inferior cerebellar peduncle and the left superior cerebellar peduncle.

Meanwhile, early alterations to the diffusion properties, i.e., FA and MD corrected for FW, were also observed within this 28-day period of SIV239 infection. Briefly, the values of FW-FA decreased progressively in several fiber tissues, particularly in the corticospinal tract (left), corona radiata (right superior), cingulum (left superior), central tegmental (left), pyramidal tracts, and temporal gyrus WM (left superior). The decrease was more severe (survived FDR-corrections for multiple comparisons) in the left central tegmental. Meanwhile, two fibers, the cerebellar peduncle (middle) and uncinate fasciculus (right and left), displayed abnormal pattern of increased FW-FA values in the early days of SIV infection, with the uncinate fasciculus fiber tissues demonstrating an intensified increase (survived FDR-corrections for multiple comparisons). Most fibers of rhesus monkeys demonstrated changes in FW-MD values— at these early stages of SIV infection. Notably, there was an increase in FW-MD values in two major fibers— the cerebellar peduncle (right and left inferior, and left superior) and the internal capsule (retrolenticular limb), whereas decreased FW-MD values were detected in 15 other fiber tissues. Such fiber tissues with decreased FW-MD values include the corpus callosum (the genu, body and splenium), fornix, internal capsule (right and left anterior limb, and left posterior limb), corona radiata (right and left anterior), sagittal striatum (left), external capsule (right and left), Stria terminalus (right and left), uncinate fasciculus (right), commissure (anterior), prefrontal WM(both right and left, dorsal and ventral), pyramidal tracts frontal gyrus WM(right and left inferior), temporal gyrus WM (right middle), midbrain WM(right), and cingulum WM(right and left anterior). 70% of these WM fiber structures had an intensified FW-MD decrease (survived FDR correction for multiple comparisons) (See Table 4 ).

From the analysis of interactive effects, we found MRI evidence of changes in the extracellular environment in rhesus monkeys following SIVmac239 infection and cART intervention. Briefly, there were interactive effects of cART treatment and time on the FW-VF of the extracellular spaces crossed by two fiber tracts—the left Sagittal Striatum (SS-L) and the right Central Tegmental Fiber Tract (CTFT-R) ( Figure 3 , also see Table 4 ). The FW-VF of these fiber tracts increased significantly over time for the rhesus monkeys that received cART intervention. Such an increase was significantly slowed in those who received a TFG+TDF+DTG regimen on day 40. It is important to highlight that, prior to cART intervention on day 40, all rhesus monkeys (from all groups/sets) had a similar trend of increasing FW-VF, with minimal variations (nearly negligible). The differences in extracellular FW-VF between the two groups of rhesus monkeys began to unfold after day 40 of cART intervention. The greatest differences in FW-VF (cART- vs. cART+, see Figure 3 ; Supplementary Table S8 for statistics) were identified on day 336 (time point 7) of MRI scanning (296 days after cART intervention).The analyses of independent effects revealed independent effects of time (arguably reflecting the impact of SIV239 infection) on the extracellular FW-VF in SIV-infected rhesus monkeys, independent of cART status (see Figure 4 , also see Supplementary Table S3 ). Fourteen (14) fiber tracts were identified with independent effects of time. Two patterns of FW-VF attenuation were identified within these fibers. The first pattern of FW-VF disruption was characterized by a continuous and progressive increase in extracellular FW-VF over time. This pattern was observed in eight(8) fiber tissues, including the superior cerebellar peduncle, medial longitudinal fasciculus, central tegmental, adjacent amygdala WM, fornix, Strial terminalus right, Strial terminalus left, and sagittal striatum. The second pattern of FW-VF disruption unfolded as a progressive decline in FW-VF and was more evident in four fiber tracts, i.e., the pontine crossing tract, the splenium of the corpus callosum, the superior corona radiata, and the superior fronto-occipital fasciculus. While we generally observed these two patterns of changes over time, we also found time-specific, region-specific, sharp, and excessive changes in FW-VF in the early days of SIV239 infection. In particular, there was a sharp increase in FW-VF in the extracellular spaces surrounding the superior corona radiata, fornix, and superior fronto-occipital fasciculus on day 28 following the infection of SIVmac239. Conversely, a sharp decrease in FW-VF was observed in the extracellular spaces surrounding the superior cerebellar peduncle and adjacent amygdala WM.

The interactive effects of cART treatment and time were also found in the FW-corrected diffusion properties of the two fiber tracts (see Figure 3 , Table 4 ). Those who received cART intervention showed a progressive increase in FW-corrected FA values over time, while those who did not receive it exhibited a decreasing pattern, particularly in the left sagittal striatum and right central tegmental fiber tracts. A post-hoc analysis revealed that the most apparent differences in FW-FA were exhibited on day 336 of MRI scanning, i.e., 296 days after cART intervention (cART-, vs. cART+ at T7, Supplementary Table S8 ). We also identified independent effects of time on FW-FA for several fiber tissues (see Figure 5 , also see Supplementary Table S4 ). Ten (10) fiber tracts demonstrated a significant reduction in FW-FA values over time, independent of cART status. These fiber tracts include the dorsal prefrontal WM, adjacent thalamus WM, genu of corpus callosum, perihippocampal cingulum, central tegmental, anterior cingulum WM, external capsule, adjacent amygdala WM, and midbrain WM.

The association between clinical measurements (i.e., SIV copies, CD4 T-cells, and CD4/CD8 ratio) and brain markers is shown in Figures 6 – 8 . A progressive increase in viral load (SIV copies per mL) and a continuous decline in the immune system (indicated by lower CD4 T-cells and CD4/CD8 ratio) occurred in parallel with changes in the extracellular spaces and diffusion properties of brain fiber tissues. In fact, a low CD4/CD8 ratio was associated with increased extracellular FW-VF across fiber tissues of the ACR-L, PTR-R, DPF-L, VPF-L, and IFG-L ( Figure 6A ). Similarly, decreased CD4 T-cell counts were strongly associated with increased FW-VF across the ACR-L, SS-L, CgH-L, VPF-L, and MTG-WM-R ( Figure 6C ). Conversely, an increased number of SIV copies, reflecting disease progression, was highly associated with increased FW-VF in the fiber tissues, particularly the SLF-L, DPF-R, PT, and MTG-WM-L ( Figure 6D )(Also see Supplementary Table S5 for statistics). For diffusion properties, we found significant associations between compromised immunity, increased SIV copies, and the diffusion properties of the fiber tissues ( Figure 7 ). In particular, a low CD4/CD8 ratio was associated with reduced FW-FA values; this was more apparent in the fiber tissues of ACR-L, SFO-R, and ACg-WM-L ( Figure 7A ). Decreased CD4 T-cells were also associated with reduced FW-FA values across multiple fiber tissues, especially the BCC, FX, SS-R, VPF-R, and VPF-L ( Figure 7C ). Conversely, increased SIV viral copies per ml were associated with reduced FW-FA values in the CgH-R and increased FW-FA values in the SLF-R ( Figure 7B ) (Also see Supplementary Table S6 for statistics of these results). On the other hand, the analysis of associations with FW-MD revealed that a low CD4/CD8 ratio was independently associated with reduced FW-MD values across the ACR-R, SS-L, CgH-L, IFG-WM-R, IFG-WM-L, and ACg-WM-R ( Figure 8A ). Additionally, decreased CD4 T-cells were associated with reduced FW-MD values for fiber tissues of BCC, FX, CgH-L, SLF-L, IFG-WM-R, and ACg-WM-R ( Figure 8C ), whereas increased viral copies as the disease advances was associated with reduced FW-MD values in the DPF-R fiber tissues ( Figure 8B ) (Also see Supplementary Table S7 for statistics of these results). Further analysis using correlation techniques to support the findings of this section is summarized in Supplementary Table S9 . And the most interesting results were in Supplementary Table S10 , where the rate of changes in CD4 T-cells over time was significantly correlated with the rate of changes in FW-VF, FV-FA, and FW-MD.

There are several limitations worth-mentioning in this study. First, while we generally agree that HIV or SIV can be considered neuroinflammatory diseases due to HIV or SIV-triggered activation of microglia, NLRP3 inflammasome, and caspase-1, as well as the release of IL-1β (39–41), the absence of a study-specific histopathological assessment may limit our conclusions. Second, although we applied several exclusion criteria to our rhesus monkeys to screen out those with conditions such as herpes B virus, STLV-I, and SRV, which may affect the brain, further studies are needed to account for other potential factors, such as early inflammation status, which is likely to affect the DTI-derived metrics. The current study design did not include a control cohort for comparison, which could have strengthened our findings. Thus, a control cohort is warranted to broaden the scope of our findings.