Antibodies to myelin oligodendrocyte glycoprotein in HIV-1 associated neurocognitive disorder: a cross-sectional cohort study

HIV encephalopathy (HIVE) leads to dementia and motor disorder and is the major direct central nervous system (CNS) manifestation of advanced HIV-1 infection. Since the availability of combination antiretroviral therapy (cART) its incidence has decreased, but to a lesser extent than the incidence of extra-cerebral AIDS-manifestations [1]. With the increasing life expectancy of HIV-infected individuals the prevalence of HIV associated neurocognitive disorder (HAND) has risen to 20-50% [2]. While it is generally accepted that HAND is treatable, the extent and sustainability of the effects of cART on cerebral functioning are still unclear. There is accumulating evidence of chronically progressive and, at times, fluctuating cognitive impairment in patients with effective cART in terms of suppression of plasma viral load [3, 4], compatible with the notion of quiescent and active disease [5]. While HIV is the agens movens of HIVE, it does not damage neuronal cells directly. Rather, a plethora of cellular and molecular immunological mechanisms leads to neurological dysfunction [6]. Demyelination has early been recognized as a feature in the pathological and radiological appearance of HIVE [7, 8], and cases with early-stage HIV infection clinically mimicking multiple sclerosis (MS) have been described [9]. Myelin breakdown products and antibodies against them have been implicated in this context. In particular myelin basic protein has been suggested to be of prognostic significance [10, 11].

Another myelin protein that has been extensively studied in MS is myelin oligodendrocyte glycoprotein (MOG) [12]. MOG is a quantitatively minor type I transmembrane protein exclusively expressed in the CNS, and its extracellular domain has been identified as a main target for immune responses in experimental allergic encephalitis (EAE), an animal model for MS [13]. However, in humans antibodies against MOG are mainly found in patients with acute demyelinating encephalomyelitis (ADEM) or childhood MS [14‐16] whereas their value in adult MS is still under debate [17]. Anti-MOG antibodies are also detected in infectious diseases of the CNS [18], and their presence correlates with the titers of antibodies to Epstein Barr Virus (EBV) [19]. To our knowledge, this cross-sectional cohort study is the first to evaluate the potential role of MOG antibodies in cerebrospinal fluid (CSF) and serum of patients with HIV as markers for disease course and response to antiviral therapy.

A total of 65 subjects with HIV infection were included in this study. For the assignment to groups see Methods. Demographic, clinical and CSF data of the respective patient groups are given in table 1.

The pathophysiology of HIVE remains incompletely understood. While in the cART era the incidence of AIDS-defining conditions showed a dramatic decrease, AIDS dementia complex is still observed in a comparably high percentage of patients and despite successful viral clearance [22]. Ongoing autoimmune mechanisms have been suggested as an explanation for this discrepancy [6]. Besides activation of brain microglia, complement-mediated mechanisms have been implicated in the development of neurological dysfunction [23, 24]. Complement is readily activated by antibodies binding to cellular epitopes. With respect to autoimmune phenomena, CSF antibody responses to CNS autoantigens such as myelin basic protein (MBP) have been demonstrated [11, 25]. These responses seemed to be more prominent in the CSF of neurologically symptomatic HIV subjects. However an association with serum antibodies was not found [25, 26].

In our study, we found higher anti-MOG reactivity in CSF and serum of patients with HAND as compared to OND patients. For autoantibodies to CNS antigens it is generally accepted that some degree of brain tissue damage as well as an increased blood-brain-barrier permeability is needed for the antigen to be presented to the immune system. Such a lesion to cerebral tissue may be assumed in acute HIV-associated opportunistic CNS-infections as well as in non-HIV-associated meningitis/encephalitis (the groups HIVOI and OIND). Therefore, our finding that anti-MOG antibodies were highest in the serum and CSF of patients with HAND is somewhat surprising, as accepted signs of inflammation such as CSF pleocytosis and contrast enhancement on MRI are conspicuously absent in HIVE.

A significant finding is the higher values of CSF and (less so) serum MOG antibodies in the HAND group as compared to the group of asymptomatic HIV subjects. As HIV invades the CNS in every infected subject from early on, the presence of the virus on its own cannot account for the higher MOG antibody titers in HAND subjects. In addition, our finding of higher MOG antibody titers in more demented patients adds to the impression that anti-MOG antibody formation, in the absence of grossly destructive inflammation, may be a relevant biomarker for HAND.

Another finding of clinical importance is the high discriminatory power of CSF anti-MOG antibodies (AUC = 0.881) compared to other routine CSF parameters in ROC analyses. Importantly, a cut-off value of 0.14 for CSF anti-MOG antibodies was highly sensitive and specific in distinguishing patients with active HAND from asymptomatic HIV patients. Since opportunistic infections and HIV-associated meningitis are easily detected by an elevated CSF cell count, elevated CSF anti-MOG antibodies could aid the clinician in the sometimes difficult diagnosis of HAND.

Our result of an association of anti-MOG antibody titers with the viral load in CSF, but not plasma, suggests that the immunological activity in the CNS may be dissociated from immune processes in the systemic compartment. This corresponds with the well recognized correlation of HIV dementia with CSF, but not plasma viral load [27], and the delayed suppression of CSF viral replication upon initiation of cART in HAND patients [28].

With successful treatment of a microbial assault to the brain, as represented in our study by suppression of HIV production in the CSF and plasma, one would expect the stimulus for the production of autoantibodies to eventually disappear. Accordingly, in microbial meningitis and systemic inflammatory diseases anti-MOG antibodies have been shown to be transient [18]. In this respect, the persistence of MOG antibodies over months to years in six of our patients and, even more so, the emergence of MOG antibodies in one patient only after initiation of cART is of particular interest.

A related observation was the increase of anti-MOG antibodies in the early samples after initiation of cART despite decrease of the plasma and (to a lesser extent) CSF viral load. This was particularly pronounced in one patient in whom follow-up brain MRI showed an intermittent marked increase in white matter lesions in contrast to clinical improvement of his HAND.

The discordance between anti-MOG antibodies and viral suppression in many patients suggests that immunological mechanisms are relatively independent from virus concentrations. Eden et al. showed sustained immunologic activity as judged by persistently elevated CSF neopterin levels in HIV-positive individuals with long-term suppression of plasma and CSF viral load [29].

In the patient presented as case vignette the progressive white matter lesions in parallel to increasing anti-MOG antibodies might be regarded as an immune reconstitution phenomenon in response to CNS autoantigens similar to the clinically overt immune reconstitution syndromes that have been described in HIV patients with opportunistic infections upon start of treatment [30].

This study has some limitations. First, the retrospective collection of some of the data may have introduced bias. While HAND, HIVOI and the OND group showed comparable age distribution, patients with primary HIV infection and asymptomatic HIV patients were younger and OIND patients were older. Although an age-dependent effect of immune response to myelin components cannot be excluded, it has so far not been observed by others [18, 21]. In addition, the main difference in serum anti-MOG activity was observed between HAND and OND patients who were well matched. Hence, we are confident that this difference should not adversely influence the interpretation of the data. Second due to the limited number of included HIV positive patients, the study was underpowered to detect differences among subgroups of HAND patients. Further studies will have to address the question if anti-MOG antibodies are able to differentiate HAND patient with asymptomatic neurocognitive impairment or mild neurocognitive disorder from neurologically asymptomatic patients as it was the case in patients with HAD in the current study. Third, in MS the role of MOG antibodies is still under debate, and a number of recent studies suggest that antibodies to native MOG might only be pathogenic in patients with acute disseminated encephalomyelitis (ADEM) or childhood MS [14‐16], whereas they are an epiphenomenon in "classical MS" that might be related to ongoing tissue damage [18, 31] or EBV infection [19]. In this study we have used recombinant human MOG extracellular Ig domain (amino acids 1-125) produced in E. coli bacteria to determine anti-MOG antibodies by ELISA. This recombinant protein was refolded after purification [32] and has been widely used for immunological studies in patients and animal models (reviewed by [13] and [12]) and for the determination of the three-dimensional structure of MOG [33]. Recent data indicate that pathogenic antibodies recognize the physiological membrane topology and glycosylation pattern of MOG [34], whereas antibodies to the MOG preparation used in the current study reflect CNS inflammation and a secondary response to myelin debris rather than a pathogenic antibody response. However, the exact role of antibodies to surface exposed conformational epitopes and glycosylated MOG in HAND require further studies.

In conclusion, our study demonstrates high rates of seropositivity and high titers of CSF antibodies to MOG in patients suffering from HAND as compared to asymptomatic HIV-infected and unifected subjects. In a proportion of HAND patients these autoantibodies persist or even develop upon initiation of cART with successful clearance of viral RNA from CSF, suggesting that antibody-mediated mechanisms of brain damage might play a role in the neuropathogenesis of AIDS. In addition, determination of serum anti-MOG antibodies might be a new non-invasive marker for the discrimination of active versus quiescent neuroinflammation in the context of HIVE.