White matter of perinatally HIV infected older youths shows low frequency fluctuations that may reflect glial cycling

In perinatally HIV-infected (PHIV) children, neurodevelopment occurs in the presence of HIV-infection, and even with combination antiretroviral therapy (cART) the brain can be a reservoir for latent HIV. Consequently, patients often demonstrate long-term cognitive deficits and developmental delay, which may be reflected in altered functional brain activity. Our objective was to examine brain function in PHIV on cART by quantifying the amplitude of low frequency fluctuations (ALFF) and regional homogeneity (ReHo). Further, we studied ALFF and ReHo changes with neuropsychological performance and measures of immune health including CD4 count and viral loads in the HIV-infected youths. We found higher ALFF and ReHo in cerebral white matter in the medial orbital lobe for PHIV (N = 11, age mean ± sd = 22.5 ± 2.9 years) compared to controls (N = 16, age = 22.5 ± 3.0 years), with age and gender as co-variates. Bilateral cerebral white matter showed increased spontaneous regional activity in PHIV compared to healthy controls. No brain regions showed lower ALFF or ReHo in PHIV compared to controls. Higher log10 viral load was associated with higher ALFF and ReHo in PHIV in bilateral cerebral white matter and right cerebral white matter respectively after masking the outcomes intrinsic to the brain regions that showed significantly higher ALFF and ReHo in the PHIV compared to the control. Reductions in social cognition and abstract thinking in PHIV were correlated with higher ALFF at the left cerebral white matter in the left medial orbital gyrus and higher ReHo at the right cerebral white matter in the PHIV patients. Although neuroinflammation and associated neuro repair were not directly measured, the findings support their potential role in PHIV impacting neurodevelopment and cognition.

Study criteria. Study inclusion criteria were similar to our previous studies 26, 31 and consisted of the following: (1) 18-30 years of age; (2) perinatal acquisition of HIV or confirmation of HIV-uninfected status with Ora-Quick (OraSure Technologies, Bethlehem, PA 18015) buccal scraping (for HIV-subjects); (3) current treatment with combination antiretroviral medication for HIV-infected subjects; (4) post-menarchal status for all females since they were studied in the follicular phase of the menstrual cycle; (5) and for females negative urine pregnancy test on day of scanning. We excluded participants if they had: (1) a history of CNS opportunistic infection or other CNS condition (other than HIV); (2) severe metabolic disturbances, such as hepatic or renal failure; (3) metallic implants or braces or permanent retainers or other MRI exclusions; (4) claustrophobia; (5) Attention Deficit/Hyperactivity Disorder; (6) pregnancy (by interview and urine pregnancy test before scanning); (7) alcohol or other substance use/abuse including marijuana; (8) active psychiatric diagnosis; (9) use of chronic medication other than inhalers for asthma in control subjects; (10) severe school difficulties in control subjects; (11) female subjects pregnant or in luteal phase of menstrual cycle; (12) hepatitis C infection.
For HIV+ subjects, the following additional data were collected from chart review: age at first treatment for HIV, HIV viral load close to time of testing, highest known viral load, CD4 T cell counts close to time of testing, lowest known CD4, lowest known CD4%, current antiretroviral therapy, known presence of HIV encephalopathy, and history of maternal substance abuse during pregnancy. The clinical variables are summarized in Table 1 and demonstrate that these patients had been treated for many years with antiretroviral therapy. Data from a life time of HIV were not always complete and we have indicated in Table 1 such variables as the lowest known CD4 and CD4% and highest known viral load realizing that viral load was not a standard test when these patients were younger. We also note that we did not systematically assess adherence to medications and realize that adolescence is a time when patients can have poor compliance. Nonetheless, 54.5% of the study patients had an undetectable viral load. Of the 11 PHIV participants, four had a diagnosis of HIV encephalopathy while one patient was considered to have a probable diagnosis of HIV encephalopathy. Eight of these 11 patients had experienced school difficulties. In addition, two mothers had known substance abuse during pregnancy while information for one mother was not available, and for 8 there was no evidence of substance abuse during pregnancy. In addition, a high-resolution T 1 -weighted magnetization-prepared rapid gradient echo scan (MPRAGE) was acquired for anatomical information for better registration and overlay of brain activity. All the subjects were scanned at the same site.
Neurocognitive data. Patients performed a neurocognitive battery test at a separate visit from the MRI data collection. These tests were assessed in depth separately, but were included here to aid with interpretation of significant findings. All subjects were administered a comprehensive neuropsychological assessment battery by a clinical psychology trainee in the following fixed order: MATRICS Consensus Cognitive Battery (MCCB) 68   Data processing and analysis. All images were preprocessed by SPM12 software 77 and Matlab 2019 (Mathworks Inc., Natick, MA). The raw EPI images were realigned to the mean of the time series to correct for head motion using the standard SPM12 routine. We used the "DRIFTER" toolbox 78 for all rs-fMRI time-series to remove local oscillatory physiologic noise like cardiac and respiratory cycles. To account for whole brain influences we performed linear detrending. fMRI images were co-registered to the anatomical scans see Methods in 79 . The anatomical images were partitioned into gray matter, white matter and cerebrospinal fluid using SPM12's "DARTEL" procedure 80 . Each participant's deformation map, obtained from the anatomical image, was applied to the functional images for normalization into the Montreal Neurological Institute (MNI) space with an isotropic voxel size of 2 mm 3 .
We used the "DPABI: Data Processing & Analysis" software package 81 to calculate ALFF, f-ALFF and ReHo. In the software package, the time series was first converted to the frequency domain using a Fast Fourier Transform, and the averaged square root of the power spectrum for the predefined typical frequency interval 0.01-0.08 Hz was termed ALFF 81,82 . We applied a bandpass filter ranging from 0.01 to 0.08 Hz to all the ALFF and f-ALFF analyses. f-ALFF measures the power within the low frequency (0.01-0.08 Hz) divided by the total power in the entire detectable frequency range to represent the relative contribution of low frequency oscillations 55 . For ReHo we analyzed unsmoothed data as per DPABI recommendations 81 . We bandpass-filtered the data to 0.01-0.08 Hz and the ReHo cluster was for 27 voxels, along with smoothening the ReHo outcome (sm-ReHo) images by a 6 mm full-width-at-half-maximum Gaussian kernel similar to 81 . We inputted the z score signals (prefixed with z-ALFFmap, z-fALFFmap and szReHomap) outputted from DPABI 81 , for subsequent statistical analysis with SPM12 package 77 . Overlap in areas of difference of ALFF and ReHo indicates regions that are active at the specified frequency and are in sync with neighboring voxels, likely reflecting a large group of neurons firing together 55 .
Once we identified the brain regions showing significantly different ALFF or ReHo values compared to controls, we conducted additional correlation analysis between the pediatric HIV neurocognitive measures and average values for those regions.
Statistical analyses. The Statistical Package for the Social Sciences (SPSS, V 24.0, IBM, Chicago, IL) was used to examine demographic and clinical parameters. Independent samples t-tests were performed to examine age, and gender differences between PHIV-infected and healthy control groups. Pearson's correlation was performed to examine the association between cognitive measures and functional connections in the PHIVinfected youth group. The significance level was set at p = 0.05. Functions, Abstract Thinking), Maternal Substance Use, School Difficulties, and whether or not the PHIV had a diagnosis of HIV Encephalopathy. The significance level was set at p = 0.05. In order to avoid multicollinearity, we reported and removed from further analysis several psychological variables that were inter-correlated.
We used the SPM12 software package 77 for ANCOVA analyses of control (n = 16) and PHIV (n = 11) groups with age and sex as co-variates. Traditional neuroimaging findings are reported as t-statistic, where a t statistic is calculated at each voxel location. Groups of adjacent voxels identified as significant are termed clusters. Clusters of rs-fMRI differences are overlaid on anatomical backgrounds for visualization. Correction for multiple comparisons was performed with cluster thresholding, which consists of two stages. After thresholding with an uncorrected threshold of p < 0.001 and minimum cluster size of 3, clusters are each thresholded based on familywise error (FWE) correction at p < 0.05.
For the regions that showed significant differences in ALFF and ReHo between PHIV and healthy controls, we used intrinsic masking in SPM to correlate the ALFF and ReHo data in the 11 pediatric HIV patients with the clinical parameters of viral load, CD4 and neuropsychological variables. The significance level of contrasts was set to p = 0.001 with cluster size greater than or equal to 3.

Results
The patient and healthy control groups' demographic details are shown in Table 1; there were no significant differences in age or gender.
We found significantly higher ALFF and ReHo in the cerebral white matter in the medial orbital gyrus (or prefrontal cortex) for PHIV patients (n = 11) compared to controls (n = 16), with age and gender as co-variates. We overlaid clusters of ALFF and ReHo changes on average of the 27 subjects' anatomical scans (Figs. 1 and 2). Table 2 depicts the brain regions showing higher ALFF and ReHo in PHIV patients compared to controls. We found predominantly in the bilateral cerebral white matter an increased spontaneous regional neuronal activity in PHIV compared to healthy controls. There were no brain regions that showed significantly lower ALFF or ReHo in PHIV compared to control. We did not obtain a significant difference in fALFF between the patients and controls. Table 3 shows positive (p < 0.01, cluster size ≥ 3) associations of log 10 viral load with ALFF and ReHo for the 11 patients in regions of PHIV-Control differences (Table 1). Table 4 shows negative (p < 0.01, cluster size ≥ 3) associations of Social Cognition, Psychomotor Functioning and Abstract Thinking with ALFF at the left cerebral white matter in the left medial orbital gyrus and with the ReHo at the right cerebral white matter in the 11 PHIV patients in regions of PHIV-Control differences (Table 1). Table 4 also lists the only significantly positive (p < 0.01, cluster size ≥ 3) association of Social Cognition with ReHo, which appeared in the right central operculum/right cerebral white matter. www.nature.com/scientificreports/

Discussion
Youth perinatally infected with HIV showed altered resting state activity, reflecting differences in brain function relative to healthy counterparts. Specifically, we found higher activity of low frequency oscillations (ALFF) in PHIV youth compared to controls, especially in the cerebral matter of prefrontal cortex where it could indicate higher sympathetic activity. Per the original study by Biswal and colleagues, ALFF in a resting state reflects correlations between blood flow and oxygenation, which is interpreted as brain regions being functionally related 40 . Previous studies on acute traumatic brain injury reported higher ALFF and increased spontaneous activity in low frequency bands (0.01-0.08 Hz) 82 . We found a group of voxels in cerebral white matter in the medial orbital gyrus with higher ALFF together with a higher level of a marker of functional similarity, ReHo, in PHIV compared to controls. We did not observe a significant difference in the groups for in neural ALFF (f-ALFF); since this measure is considered gray matter specific 55 , the findings of altered ALFF likely reflect at least in part differences in non-neural physiology, including in the white matter; such fluctuations in the fMRI signal may reflect inflammation and glial activation in PHIV relative to the control group. Global effects such as motion or cerebral blood flow changes are unlikely to have influenced the findings. Low-frequency fluctuations in white matter are reduced relative to grey matter by 60% 40 and the significance of white matter spontaneous neuronal firing in resting fMRI data has not been reported previously for HIV adults or PHIV. We had removed physiological artifacts using DRIFTER toolbox 78 , and detrended the fMRI data with linear detrending tools (as in 40 ). Additionally, we found that some ReHo and ALFF differences occurred in   www.nature.com/scientificreports/ overlapping brain regions (cerebral white matter in the medial orbital gyrus). Thus, the findings of white matter differences in fMRI activity are unlikely to have been influenced by global effects. Higher ALFF in particular could be related to underlying glial cycling or mitosis of glial cells. Microgliosis and neuroinflammation are long-term consequences of traumatic brain injury and pathogenesis in general 83 and in PHIV we expect to find both microgliosis and neuroinflammation. HIV causes inflammation throughout the brain, which can persist despite control of the HIV virus in the peripheral blood 84 . The monocytes and T cells in the brain that are infected with HIV and have successfully crossed the blood-brain barrier can induce endothelial cells to release cytokines, consequently causing inflammation within the brain. In PHIV, this inflammation in the CNS may persist due to the difficulty of common medications being able to cross the blood-brain barrier into the CNS. The HIV-infected monocytes and T cells not only contaminate brain cells, but also release proinflammatory cytokines, viral proteins, and excitotoxins that can activate microglia, perivascular macrophages and astrocyte cells in the CNS and are potential reservoirs for the virus 85 . These are the main contributors to neuroinflammation in HIV infection and these cells release neurotoxic factors such as excitatory amino acids in addition to inflammatory mediators 86 . An HIV-infected CNS results in the increased activation of monocytes Table 3. Correlation of resting fMRI data with viral load. Brain regions masked with significantly higher regional neuronal activity in the pediatric HIV patients compared to healthy controls, showed positive correlation (peak uncorrected p < 0.01, cluster size ≥ 3) of log 10 viral load with regional neuronal activity in the 11 pediatric HIV patients. + ve indicate positive correlation and − ve indicate negative correlation.   www.nature.com/scientificreports/ and macrophage, resulting in astrocytosis and microglial activation 87 . Glial cycling could also result from such pathophysiology, which could explain why we find higher ALFF activity in PHIV. Our earlier finding of compromised white matter integrity found via Diffusion Tensor Imaging (DTI) in PHIV suggests structural differences may co-occur with the functional alterations, and inflammation is one possible cause of changes in both structure as seen with DTI 88 and ALFF as seen here.
Recent research in adult HIV patients receiving cART vs healthy controls has found rs-fMRI differences mostly in ALFF measures [89][90][91] . In these studies, various regions showed increased or decreased magnitude of ALFF which differs from the findings in our study [89][90][91][92] . While there may be systematic differences between our PHIV group and other HIV populations, our small number of subjects precludes making strong generalizations.
The fact that we found higher ALFF in the cerebral white matter of orbital and frontal gyri in PHIV patients at the brain regions correlated with cognitive and emotional response could be indicative of ongoing neuroinflammatory insults 93,94 . Similar to our present study, studies of leukoaraiosis (LA) have found a higher ALFF in cerebral white matter of superior orbital frontal gyrus in the periventricular and subcortical areas of the brain. Moreover, LA patients also show cognitive impairment 95 as found in our present study in the PHIV patients, suggesting there may be similar cognitive impairment and associated higher white matter ALFF activity and neuroinflammation in PHIV. Our findings are consistent with a previous study on postmortem brain tissue from patients with HIV-associated neurocognitive disorders, which showed signs of neuroinflammation 96 . It has been reported that some antiretroviral medications used to treat HIV can contribute to the likelihood of neurocognitive disorders 84 . Although new cART drugs are less toxic with fewer metabolic complications, chronic inflammation and other factors such as the irreparable damage of metabolic tissues suffered prior to the introduction of cART, side effects associated with other medications, and host genetic risk can still contribute to the neurocognitive impairment observed in PHIV-infected youth and in general HIV-infected patients.
Limitations of our study include the relatively small sample size, so further studies with larger cohorts are needed to confirm our findings. In addition, the cross-sectional design limited our ability to assess the impact of HIV on brain development over time. Future rs-fMRI study on PHIV youth should also include perinatally HIV-exposed uninfected youth apart from the HIV-unexposed healthy controls group for better distinguishing potential mechanisms.

Conclusions
The findings are consistent with the hypothesis that long-term higher neuroinflammation and associated neurorepair in perinatally HIV-infected patients may be reflected in the higher regional spontaneous activity that we observe in the white matter in PHIV patients compared to healthy controls. Moreover, the higher cerebral white matter spontaneous activity correlated with higher viral load and decreased cognition, suggesting a role for neuroinflammation in impaired cognition. Resting state fMRI, particularly ALFF data that has been utilized to interpret neuroinflammation in this study, shows promise as a future tool to follow the effects of HIV on brain function, which is an important measure since these PHIV youth survive many years into adulthood. Such noninvasive measures may detect subtle ongoing inflammation, which could potentially be targeted with antiinflammatory therapy or changes in antiretroviral treatment to preserve brain health in these surviving patients. In future, larger sample size studies should consider other neuroimaging techniques to confirm inflammation in the white matter in PHIV. www.nature.com/scientificreports/