| |
HIV disease and diabetes interact to affect brain white
matter hyperintensities and cognition
|
| |
| |
Download the PDF here
Wu, Minjiea; Fatukasi, Omalaraa; Yang, Shaolinb; Alger, Jefferyc; Barker, Peter B.d; Hetherington, Hobye; Kim, Taee; Levine, Andrewf; Martin, Eileeng; Munro, Cynthia A.h; Parrish, Toddi; Ragin, Anni; Sacktor, Nedh; Seaberg, Ericj; Becker, James T.a,k,l for the Neuropsychology Working Group of the Multicenter AIDS Cohort Study
Background: Since the onset of combination antiretroviral therapy use, the incidence of HIV-associated dementia and of HIV encephalitis has fallen dramatically. The present study investigates the extent of white matter hyperintensities (WMHs) among individuals with HIV disease, and factors that predict their presence and their impact on psychomotor speed.
Methods: A total of 322 men participating in the Multicenter AIDS Cohort Study (185 HIV-infected, age: 57.5 ± 6.0) underwent MRI scans of the brain. T1-weighted magnetization-prepared rapid gradient-echo (MP-RAGE) and T2-weighted Fluid Attenuated Inversion Recovery (FLAIR) images were obtained and processed using an automated method for identifying and measuring WMHs. WMH burden was expressed as the log10 transformed percentage of total white matter.
Results: There were no significant associations between WMHs and HIV disease. However, the extent of WMHs was predicted by age more than 60 (β = 0.17), non-white race (β = 0.14), glomerular filtration rate (β = −0.11), and the presence of diabetes (β = 0.12). There were no interactions between HIV status and age (β = −0.03) or between age and diabetes (β = 0.07). However, the interaction between HIV infection and diabetes was significant (β = 0.26). The extent of WMHs was significantly associated with performance on measures of psychomotor speed (β = 0.15).
Conclusion: In today's therapeutic environment, in HIV-infected and HIV seronegative individuals, those factors which affect the cerebrovasculature are the best predictors of WMHs. Diabetes has a specific impact among HIV-infected, but not uninfected, men, suggesting the need for more aggressive treatment even in the prediabetes state, especially as WMHs affect cognitive functions.
Introduction
In the early days of the HIV epidemic, brain damage and subsequent cognitive impairment were a common sequela. If HIV-associated dementia (HAD) was the initial AIDS-defining condition, death could occur within 6 months [1]. At autopsy, HIV encephalitis was common with multinucleated giant cells and 'white matter pallor' [2,3]. In spite of the obvious changes to cortical white matter at death, in-vivo evidence of these changes was relatively lacking [4].
Since the introduction of combination antiretroviral therapy (cART), the incidence and prevalence of HAD have fallen precipitously among those individuals with access to good medical care [5]; in its place, less severe forms of cognitive dysfunction remain [6]. However, there are no generally accepted biomarkers, especially from neuroimaging, that can aid in the diagnosis of these conditions [7]. One candidate for such a biomarker would be consistent changes to white matter. On the contrary, the data in this regard are inconsistent. In two recent meta-analyses [8,9] significant between-study variability was found. The effects of HIV disease on microstructure of the white matter vary considerably likely because of between-study differences in sampling, acquisition, and data analysis. Where there were HIV-related effects, the consistency of the differences across time suggested that the changes were not necessarily tracking with HIV disease. In terms of macrostructural changes, there are reliable differences as a function of serostatus in total gray matter volume and cerebrospinal fluid volume. The same, however, is not true for white matter volume. Again, findings are inconsistent among studies, and differences in data analysis may have contributed to many of disparate findings. Of some interest is the fact that the largest group differences in terms of gray matter volumes were observed early in the epidemic (e.g. [10]) and less so more recently (e.g. [11]).
Critical for the current analysis is the fact that many age-related factors such as diabetes are more common in HIV-infected individuals and can have substantial impact on cognition and brain structure (e.g. [12]). For example, insulin resistance is linked to impaired cognition [13], and among people without diabetes, fasting glucose levels are higher with increasing cognitive impairment [14]. The presence of the APOE*4 allele increases the risk of dementia in older adults [15], although in the Multicenter AIDS Cohort Study (MACS) we found no relationship between E*4 and incident cognitive impairment [16].
Others have found that among HIV-infected individuals, white matter hyperintensities (WMHs) are associated with age and SBP [17] accompanied by significant gray matter loss in the frontal cortex [18]. There is also an association between abnormal glucose metabolism and microstructural abnormalities in the basal ganglia and in the hippocampus [19]. Taken together, these findings indicate that there is small vessel disease in the brain and that it is likely more related to cardiovascular disease (CVD)-related factors than to HIV-associated neuropathology.
A subset of men participating in the MACS underwent brain MRI scans using the sequences developed for the Alzheimer's Disease Neuroimaging Initiative [20]. WMHs seen on T2-weighted (T2w) FLAIR MR brain images are closely linked to CVD and potentially to HIV infection. Here we investigated the predictors of WMHs expressed as a proportion of the total brain white matter. We also examined the relationship between the WMHs and a composite measure of psychomotor speed. We predicted that WMHs would be linked to psychomotor speed, and that both CVD and HIV Disease would be associated with greater WMH volume.
Discussion
There are four principal findings from this study. First, there is no association between the volume of WMHs and HIV serostatus in this group of gay and bisexual men either in unadjusted or adjusted analyses. Second, there is a significant association between WMH volume, diabetes, kidney function, and coronary artery calcium deposition, but only until age and race are accounted for. Third, there is a significant interaction between HIV serostatus and diabetes such that those individuals who are both diabetic and HIV-infected have significantly greater volumes of WMHs than people without diabetes and individuals without HIV disease. Finally, WMH volume, but not HIV infection, was a significant predictor of psychomotor speed.
With regard to the first point, there is significant between-study variability in the association between HIV infection and WMHs. In our first study with visual ratings of white matter lesions (WML) prior to the use of any pharmacological agents to combat HIV disease [4], we found no link between HIV serostatus and WMHs in spite of the fact that HIV encephalitis (HIVE) was (relatively) common. Since the beginning of the use of cART in the mid-1990s, HAD and HIVE have all but disappeared in areas of the world with access to adequate medical care. A recent meta-analysis of brain macrostructural changes [9]
concluded that the extent of white matter volume differences as a function of HIV disease has diminished rapidly in the cART era. However, O'Connor et al. also concluded that microstructural changes may persist in HIV infected individuals using cART, suggesting that DTI may be more sensitive in detecting the more subtle impact of HIV disease on brain structural connectivity (e.g., [24]). In a second meta-analysis examining microstructural changes, it was not possible to reach strong conclusions regarding the relationship between HIV disease and white matter as there was substantial between-study variability in methodologies, and it was also difficult to disentangle effects of treatment, age, and duration of infection [9]. Nevertheless, the authors concluded that because of the stability of the small alterations in white matter microstructure 3 and 6 months after the initiation of cART, there may be a residual impact of low-level viral replication and cellular inflammation on brain structure.
Here we used a validated, automated method for identifying WMHs and calculating the lesion volumes. We expressed lesion volume as a percentage of total white matter volume (i.e. healthy and unhealthy) and then transformed these values for parametric analysis. Although we found significant effects of age and race on WMH volume, HIV serostatus was not associated with lesion volume, either in unadjusted or adjusted models.
However, in spite of the lack of association with HIV infection, we did find a significant link between WMHs and diabetes, renal function, and coronary artery calcium deposition. This is important as these findings are consistent with the literature linking WMHs to cerebral small vessel disease, which is also highly correlated with changes in kidney function (e.g. [25]) in cognitively normal elderly individuals. The finding that these associations were significantly attenuated when age and race were entered into the model is consistent with the body of literature demonstrating increased rates of diabetes, heart disease, and kidney dysfunction among African-Americans as well as an increase in cerebrovascular disease as a function of age.
It is important to note that we did not study total white matter (i.e. normal appearing WM as a proportion of intracranial volume), but only that WM tissue that appeared hyperintense. These lesions are likely a consequence of systemic vascular disease can cause vascular hypertrophy and microvascular remodeling. The vascular disease promotes arteriosclerosis in large vessels and lypohyalinosis in penetrating arterioles, with subsequent dysfunction cerebral blood flow [26-28] which can lead to strokes and WMHs [29-31].
Numerous studies have found that diabetes mellitus is associated with poor brain health, as well as diseases of the kidney and the vasculature (both peripheral and central). Among individuals with a history of stroke or transient ischemic attacks, diabetes mellitus and impaired renal function are both associated with smaller brain volumes and poor white matter microstructural integrity [32]. Among patients scheduled for carotid endarterectomy, WMH volume was significantly associated with diabetes mellitus but not HTN [33]. In a large study analyzing brain microstructural differences, diabetes mellitus was associated with reduced WM fractional anisotropy (an index of WM integrity) as well as WML volume [34]. A study from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort found lower total brain volume in individuals with both mild cognitive impairment (MCI) and diabetes mellitus relative to individuals who had MCI but not diabetes. Those authors concluded that the presence of diabetes may accelerate cognitive deterioration among patients with MCI by affecting brain volume [35]. Among African-Americans with diabetesmellitus, greater gray matter volume (i.e., better structural integrity) was associated with lower levels of proteinuria as well as with a higher eGFR (i.e., better kidney function). Consistent with the present results, higher WML load was associated with lower eGFR as well as higher proteinuria, and proteinuria was associated with slower psychomotor speed [36].
Among a community sample of elderly Japanese individuals, brain volumes were found to be significantly smaller among those with diabetes mellitus. A longer duration of diabetes mellitus was also significantly associated with lower brain volumes; those individuals whose diabetes was diagnosed in midlife had poorer brain structural integrity relative to those who were diagnosed later in life [37]. The AGES-Reykjavik Study found that individuals with midlife cardiovascular risk factors, including diabetes, had smaller total brain volumes in late life than those without such risk factors [38]. Although there was a trend in this direction among the HIV-infected men in the current study, the small number of people with diabetes overall (n = 82) rendered that analysis underpowered. In another community-based study, there was a significant relationship between WML volume, HTN, diabetes, smoking, and education level [39].
Although alterations in brain structure are important, how these alterations may affect structural interconnections between brain regions is becoming increasingly critical to our understanding of the pathophysiology of dementia syndromes. In a small study comparing individuals with and without diabetes mellitus, those with disease had fewer white matter connections between the hippocampus and the prefrontal cortex. Those individuals with fewer structural connections between these brain regions also had poor performance on memory tests, suggesting a possible link between structural/functional connectivity and behavior [40].
Consistent control of diabetes is also important; intraindividual variability in HbA1c is significantly associated with higher WML volume among individuals carrying the APOE*4 allele (after controlling for a range of variables [41]). In a large study of individuals with diabetes mellitus followed for approximately 3.5 years, those with persistent proteinuria had a greater increase in new WMLs than those without proteinuria or those for whom protein was observed only intermittently. This difference was attenuated after adjustment for age and for SBP [42], suggesting that it was the persistent HTN causing increases in small vessel disease that were responsible for the increases in WMLs.
WMH volume may be affected by specific lifestyle interventions. After 10 years of aggressive behavioral interventions to promote weight loss and to increase physical activity, mean WMH volume was significantly lower among the treated individuals compared with the usual care group [43].
The observed interaction between the presence of diabetes and HIV infection is of particular significance. Although HAD and HIVE prevalence has fallen, milder forms of impairment remain common. The HIV-by-diabetes interaction might help explain apparent HIV-related cognitive impairment (esp. psychomotor slowing), whereas in fact poorer performance among infected individuals may be due to the greater extent of WMHs, and not to their HIV infection. From the perspective of treating physicians, the greater volume of WMHs among the HIV-infected individuals suggests that they should screen for and aggressively treat diabetes and prediabetes among their patients with HIV disease. As noted above, the longer that an individual suffers from systemic vascular disease, the greater the burden on the cerebral vasculature (resulting in WMHs). The mechanism for HIV-by-diabetes interaction is unclear and worthy of additional careful study, but as we did not find an association between HIV infection and diabetes, this is not the result of a spurious correlation.
Finally, these data raise important conceptual questions regarding the perspective from which investigators of HIV disease and HIV-associated neurological complications may want to consider their research questions. In the past, especially prior to the use of cART, we would routinely ask how various comorbid conditions affect the natural and treated history of HIV disease. It may be worth asking now about the impact HIV infection has on the expression, severity, natural history, and treatment/management of 'normal' age-related medical conditions? In the long run, as HIV infection becomes more of a chronic disease than it is even now, this may provide us with the most patient-relevant information.
|
|
| |
| |
|
|
|
