Axonal chronic injury in treatment-naïve HIV+ adults with asymptomatic neurocognitive impairment and its relationship with clinical variables and cognitive status

HIV can enter the central nervous system (CNS) soon after seroconversion and cause persistent CNS inflammation [1]. With ongoing injury to the brain, it may lead to cognitive, behavioural and motor abnormalities, which are called HIV-associated neurocognitive disorders (HAND) [2]. HAND can be clinically subdivided into three categories: asymptomatic neurocognitive impairment (ANI), mild neurocognitive disorder (MND) and HIV-associated dementia (HAD) [3]. ANI is the mildest and most common type of HAND (accounting for 70%), which is characterized by mild cognitive impairment on neuropsychological performance tests without obvious accompanying difficulties in daily functioning [4]. HAD is the most severe form. These definitions are mainly based upon an individual’s performance on multiple cognitive domains and a brief self-report of cognitive difficulties in daily life. These neuropsychological tests are time consuming (~ 3 h) and are usually performed in specific research institutions [3], so patients in ANI stages are rarely diagnosed in traditional outpatient visits (15–30 min).

Conventional structural magnetic resonance imaging (MRI) scans are unable to detect early HIV-associated brain white matter (WM) abnormalities [5]. As a noninvasive and rapidly evolving MRI technique, diffusion tensor imaging (DTI) can measure the diffusion of water molecules in WM, and more recently, it has become a popular method for studying HIV-induced WM microstructural integrity [6–12]. So far, tract-based spatial statistics (TBSS) is the most frequently recommended and employed method of analysis for DTI [13], which overcomes the limitations of conventional methods (region of interest, ROI; voxel-based analysis, VBA). There are four parameters for DTI, including fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD). As a marker of the diffusion directionality of water molecules, FA can reflect the deviation of water motion and provide information about the microstructural integrity of highly oriented microstructures [14]. MD is a marker of the molecular motion speed and can reflect the average diffusion in all three directions [15]. AD is assumed to reflect diffusivity parallel to the WM tract, and RD represents diffusion perpendicular to the tract [16]. Generally, FA and MD are influenced by AD and RD. Decreased FA and increased MD are measures of neuronal injury, increased AD is a measure of axonal chronic damage, and increased RD is a measure of myelin damage [7, 17].

Though many DTI studies on HIV have reported a loss of WM integrity [6–12], few studies have focused on the WM microstructure in neuroasymptomatic HIV+ individuals without treatment. Zhu T et al. found that WM injuries in neurologically asymptomatic HIV patients are mainly located in the posterior part of both hemispheres (MD, AD, RD increased significantly) [7]. Wang B et al. observed a decrease in FA in the corpus callosum and anterior corona radiata and an increase in MD, RD, and AD in most skeleton locations [8]. In both studies, the neuropsychological tests were assessed by AIDS Dementia Complex (ADC) staging according to the Memorial Sloan Kettering (MSK) staging system, and an ADC score of 0 or 0.5 was considered neurocognitive asymptomatic. Zhuang Y et al. investigated WM changes in ANI patients diagnosed according to the Frascati criteria [3] and found no significant WM microstructural differences between HIV-infected and healthy controls [18]. Cysique LA et al. reported that the location of WM injury in ANI cases was the anterior limb of the internal capsule [19]. The above findings still exhibit differences.

HIV-associated WM damage includes demyelination and axonal injury; however, the relationship between them and the neuropathology of HIV-related WM impairment is still unclear. The primary oligodendrocyte and myelin damage leading to secondary axonal damage (outside-in) or primary axonopathy triggering oligodendrocyte injury and demyelination (inside-out) are indistinguishable [20]. WM alterations may contribute to cognitive deficits in HIV-infected patients [7, 18, 21]. It is worth noting that the relationship between WM abnormalities and cognitive status has not been well characterized or systematically assessed.

Moreover, the effects of multiple clinical contributors on cerebral WM integrity merit further investigation. The various potential clinical influencing variables include factors that are directly related to HIV disease (i.e., CD4⁺ level, CD4⁺/CD8⁺ ratio, plasma viral load and duration of HIV infection) and factors that can affect the CNS, such as ageing. Recent studies have identified that HIV duration was significantly correlated with DTI parameters [7, 22]. FA values in the corpus callosum were negatively correlated with the duration of infection in antiretroviral-naïve primary HIV infection patients [23]. Cohen RA et al. reported that the CD4 nadir and the duration of HIV infection may be risk factors for cerebral injury [24]. However, other studies showed dissenting results that age can exacerbate HIV-associated WM abnormalities [11, 25]. Regrettably, there were discrepancies in previous findings.

In the present study, we aimed to investigate the WM microstructural changes in treatment-naïve HIV patients with ANI through DTI-TBSS technology. In particular, the current study was confined to treatment-naïve patients to rule out the possibility of antiretroviral therapy (ART) erosion on WM integrity. We also wanted to quantify the relationships between WM damage and age, CD4⁺ counts, CD4⁺/CD8⁺ ratio, plasma viral load, duration of HIV infection and cognitive status.

This study not only supports but also further extends previous DTI findings in neuroasymptomatic HIV-positive individuals. One purpose of this study is to explore the microstructure changes of WM in treatment-naïve ANI patients employing the DTI-TBSS method. Compared with healthy controls, ANI patients exhibited significantly increased MD and AD in the corpus callosum and anterior and superior corona radiate. The corpus callosum and corona radiata were distributed around the lateral ventricle. The anterior limb of internal capsule, external capsule, retrolenticular part of the internal capsule, posterior corona radiata, posterior thalamic radiation, sagittal stratum, and superior longitudinal fasciculus presented significantly increased AD, and we found that most of them were also close to the ventricles. It is speculated that HIV-associated WM involvement is selective rather than random. Periventricular WM, especially the corpus callosum and corona radiata, are more vulnerable to viral invasion and neuroinflammation in early HIV infection in adults, which are consistent with previous studies [6–8, 32, 33]. Ragin et al. even found a loss of WM integrity in the corpus callosum within 100 days of HIV infection [33]. It is not clear why these regions were vulnerable to viral invasion and neuroinflammation. One possible explanation is that the choroid plexus is the blood-cerebrospinal fluid (CSF) barrier, and HIV can accumulate in the CSF when it is destroyed. Studies have shown that HIV can infiltrate the CSF as early as 8 days after exposure [34], and CSF serves as a proxy for the brain parenchyma and the reservoir for monocytes linked to HIV neuropathogenesis [35]. Early neuroinvasion was identified by measurable markers of CSF inflammation [34], and the WM tracts around the ventricle might be affected gradually. The interpretation of these results requires caution, and future investigation will be needed to better characterize them.

WM tracts with increased AD are more extensive than those of increased MD. The overlap and differences of the two DTI parameters (MD and AD) in significantly altered cerebral regions reflect differences in the nature and degree of WM injury. Increased MD indicates an increase of the water molecules’ diffusion speed, which is caused by cell degeneration and a decrease of membrane density. Increased MD may reflect inflammation or increased glial activation, a measure of neuronal injury. Increased AD is a marker of axonal chronic damage [7, 36]. Increased RD is associated with the destruction of myelin integrity and is used as a marker for demyelination [16, 17, 37]. RD abnormalities were not found in the current ANI study. It was concluded that MD changes in ANI subjects were mainly attributable to increased AD, suggesting chronic axonal injury rather than the disruption integrity of myelin in early HIV infection. A similar study found elevated CSF neurofilament light chain concentration and its correlation with MRS-based metabolites in primary HIV infection [32], and the neurofilament light chain is a sensitive marker of axonal injury. These findings demonstrate that axonopathy may exist before myelin injury, and this may be a novel observation. However, whether can axonal injury trigger demyelination (inside-out) is not yet clear. In recent DTI studies on ultra-early HIV infection, the authors focused only on FA and MD, and AD and RD were not calculated [23, 33]. AD and RD are also important, as they provide information on the nature of the WM microstructure alterations observed in HIV patients. Researching on multiple metrics of DTI may help us to comprehend the pathophysiology of HIV-related WM injury. The relationship of axonal and myelin injury needs to be better characterized in future HAND pathology studies.

FA abnormalities were also not found in ANI patients, which implies that RD maybe the predominant factor that contributes to decreased FA. Similar findings also have been reported, which showed that FA changes were attributable to increased RD [14, 38]. In addition, several DTI studies noted that significant alterations in FA were found mainly in cognitively impaired HIV-infected patients [7, 39]. Thus, MD may be a more sensitive biomarker than FA in evaluating WM injury in early HIV infection.

Our previous DTI study in early HIV infection showed significant differences in MD, AD, and RD values between a therapy naïve HIV+ group and healthy control group [29]. One resemblance between the two studies is the distribution of white matter abnormalities. Another similarity is that white matter abnormalities are all reflected in the changes of MD and AD values, rather than FA values. The difference is that the previous study has a wider range of WM injury. Additionally, changes of RD values can be seen in a few WM tracts (genu of corpus callosum and superior corona radiate), which indicates myelin damage. A possible reason is that HIV-infected patients in the previous study were classified into ADC stage 0 according to the MSK classification. While MSK is a decent scale to globally express the state of cognitive functioning, it is not very sensitive to changes in less affected patients. The Frascati scale used in the current study may be more sensitive to identify and classify individuals with subclinical impairment. More detailed neuropsychological assessment for earlier HIV-infected patients is the novel element of the current study relative to the previous study. Corrêa et al. found that HIV patients with planning deficits had significantly decreased FA, increased MD and RD values, predominantly in frontal lobes, genu and splenium of the corpus callosum, and much less widespread abnormalities were seen in the AD values compared with normal controls. HIV+ patients with planning deficits also had significantly decreased FA values and increased MD and RD values in some white matter regions compared to those without planning deficits [40]. No significant abnormalities AD values were seen between the two groups. These results indicated that RD abnormal values predominated in the areas of decreased FA compared to AD values, suggesting that demyelination could play a role in the physiopathology of HIV-related WM injury, which is not completely consistent with our results. The possible reason for the difference between the two results was that participants in the previous article all received ART, and with longer known infection. Antiretroviral drugs may be injurious to brain cell elements. The influence of treatment on brain structure and function are less clear [41]. HIV+ patients on low CNS penetration ART had a significantly greater fMRI response amplitude compared to the HIV+ patients on high CNS penetration ART or normal controls [42]. To the best of our knowledge, no studies have detected the effects of ART regimen CNS penetration effectiveness on WM microstructure. Effects of treatment should be explored in future studies.

The MD values in the right anterior corona radiata and AD values in the left posterior corona radiata have a significantly negative correlation with CD4⁺/CD8⁺ ratios, and the regression coefficient is 0.437 and 0.488, respectively; in other words, WM microstructure changes (43.7% in the anterior corona radiata, 48.8% in the posterior corona radiate) can be influenced by CD4⁺/CD8⁺ ratios. The abnormality of AD values was related to axonal chronic injury. Clinically, the lower CD4⁺/CD8⁺ ratios were related to immunosenescence [43]. This might imply that immunosenescence among the ANI patients would accelerate the axonal chronic injury in the corona radiata, and the lower CD4⁺/CD8⁺ ratio might be an important predictor of WM injury in the corona radiata. Furthermore, we found that plasma viral load remained independently associated with AD values in the right anterior limb of the internal capsule, and the regression coefficient was 0.533. The higher the plasma viral load, the higher are the AD values in the right anterior limb of the internal capsule. The higher plasma viral load represents the activity and replication of HIV in the human internal environment. The positive correlation indicates that the WM microstructure in the anterior limb of the internal capsule was susceptible to viral replication in HIV infection. The current findings were not fully consistent with previous studies [7, 11, 22–25], and the results of previous studies also varied. This may be due to the difference and heterogeneity in participants with diverse treatments, sample size, cognitive status, and disease durations. Further studies are needed to resolve this incongruity and reliability.

Associations between WM microstructure alterations and cognitive impairment were observed in the current study. Several other studies have reported that WM changes were related to HIV-associated cognitive difficulties [44–46]. WM microstructure changes in the anterior and superior corona radiata and anterior limb of the internal capsule were significantly correlated with poorer verbal fluency. WM damage in the genu of the corpus callosum and anterior and superior corona radiata were significantly correlated with poorer memory (learning and delayed recall) and slower fine motor speed. The corona radiata is the radiated projection fibre connecting the internal capsule to the cerebral cortex. The corona radiata and internal capsule are important WM nodes that promote the transfer of sensorimotor information between the brain stem, thalamus and frontostriatal circuit [47]. The anterior corona radiate connects the anterior and medial nuclei of the thalamus to the prefrontal cortex. The superior corona radiate involves corticospinal tracts and the posterior frontal part of the anterior thalamic radiation [7]. The corpus callosum is the largest and most prominent WM tract, which is responsible for the communication of interhemispheric information. The genu of the corpus callosum contains the posterior frontal part of callosal fibres [7]. The corpus callosum and corona radiate are pivotal in extensive cognitive function, such as verbal fluency, attention, memory, psychomotor speed and executive functioning. The significant correlation between neurocognitive performance and MD and AD values from multiple WM microstructures suggests that WM abnormalities have functional consequences. HIV-related cognitive impairment may be associated with cortical and subcortical track loss caused by WM fibre bundle damage, and the WM microstructure may serve as an indicator to objectively predict cognitive deficits and progression. However, a multivariate model also showed that the WM microstructure alterations in the superior corona radiate and anterior limb of the internal capsule better predicted sustained attention/working memory. WM injury in the genu of the corpus callosum, anterior and superior corona radiata, anterior limb of the internal capsule and posterior thalamic radiation were significantly correlated with faster speed of information processing. A potential explanation for this may be some sort of compensatory mechanism, and it needs to be further verified in future multimodal studies (DTI combined functional MRI).

There were several limitations in the current study. First, the study was limited to a small sample size. A larger sample size would be more helpful to improve the power of the statistical analysis. Second, the participants were almost exclusively male, which may prevent the generalization of these results to HIV-infected women. Certainly, given that the gender gap is narrowing with rates of infection increasing in women, we are trying our best to extend our studies to include female patients in the future. Third, a cognitively intact HIV-positive group will be studied in further work. Fourth, a longitudinal follow-up study is imperative to observe the dynamic changes of WM after ART.

The observations of the current study strengthen the possibility that HIV-infected individuals at the ANI stage have underlying WM fibre abnormalities, which could be measured by increased MD, and the pathogenesis of this damage is likely to be predominantly the axonal chronic injury associated with increased AD. DTI has the potential to promote a better understanding of the pathogenesis of brain WM changes. Specific brain regions around the ventricle, especially the corpus callosum and corona radiata, are susceptible to be involved. Relationship exists between WM damage, HIV-related clinical factors, and cognitive status. HIV patients with a history of advanced immune suppression and higher viral load may be at high risk of WM injury. WM damage and disconnection to the cortex probably contribute to cognitive impairments.