IAS–USA Topics in Antiviral Medicine

Invited Review CROI 2021: Neurologic Complications of HIV-1 or Covid-19

Beau M. Ances, MD, PhD; Albert M. Anderson, MD; Scott L. Letendre, MD

The 2021 Conference on Retroviruses and Opportunistic (CROI) living location and conditions. A sub- featured a timely review of the neurologic complications of COVID-19 as well stantial proportion of people with HIV as new research findings on mechanisms by which SARS-CoV-2 may affect the brain. CROI included new and important findings about the neurologic In the CHARTER study, complications of HIV-1, (also known as JC ), and cognitive decline over cryptococcus. New long-term analyses of cognition in people with HIV-1 identified that cognitive decline over time is associated with multimorbidity, 12 years was associated particularly diabetes, chronic lung disease, and vascular disease risk conditions. with chronic lung These conditions are associated with aging, and the question of whether people with HIV are at risk for premature aging was addressed by several disease, diabetes reports. New findings from large analyses of resting state networks also pro- mellitus, major vided valuable information on the structural and functional networks that depressive disorder, are affected by HIV-1 infection and cognitive impairment. Several reports addressed changes after initiating or switching antiretroviral therapy (ART). and hypertension Findings that will improve understanding of the biologic mechanisms of brain injury in people with HIV were also presented and included evidence contracted their infection more than that host (eg, myeloid activation, inflammation, and endothelial activation) 20 years ago and are now aging into and viral (eg, transcriptional activity and compartmentalization) factors ad- their seventh decade and beyond be- versely affect brain health. Other research focused on adjunctive therapies cause of the benefits of ART. In the to treat HIV-1 and its complications in the . This sum- United States, this group has been the mary will review these and other findings in greater detail and identify key particular focus of research cohorts gaps and opportunities for researchers and clinicians. such as the MWCCS (MACS/WIHS Com- bined Cohort Study) and the CHARTER Keywords: HIV, CROI 2021, COVID-19, SARS-CoV-2, neurologic complications, (CNS Antiretroviral Therapy Effects cognition, brain, CSF, neuroimaging, aging, neurotoxicity Research) study. CHARTER investiga- tors reported on cognition and depres- Introduction HIV age into their seventh decade and sion in people with HIV after more than beyond. New data also provide encour- a decade of follow up (Abstract 101). The effects of HIV-1 and SARS-CoV-2 in aging news for reducing the neurotox- The 397 participants had been follow- the central nervous system (CNS) was icity of antiretroviral therapy (ART), for ed up for a mean of 12.4 years and had an important theme of several presenta- treating cognitive impairment, and for a mean age of 56 years. Nearly all took tions at the 2021 virtual Conference on advancing the HIV cure agenda. This ART (mean duration, 15.3 years) and Retroviruses and Opportunistic Infec- review will focus on major thematic 91.9% had a plasma HIV RNA level tions (CROI). Presentations focused on areas that may inform new research below 200 copies/mL. Nearly a quarter HIV pathogenesis and CNS reservoirs' and stimulate further discussion of clin- (23.4%) had evidence of cognitive de- persistent neurologic dysfunctions (as ical management of HIV infection. cline, compared with 5% of a norma- assessed by neuropsychiatric testing, tive population of people without HIV. imaging, or [CSF] Observational Findings on the Decline was associated with the pres- evaluations) in virologically controlled Effects of HIV-1 on the Brain ence of several comorbid conditions, in- people with HIV. New data were also cluding chronic lung disease (P=.021), presented on premature aging and the New Data on Cognition and Mood diabetes (P=.004), major depressive dis- effects of aging-related comorbidities People with HIV are highly diverse and order (P=.016), and hypertension (P= on brain function, which have become differ in many characteristics including .021), as well as longer duration of ART increasingly important as people with age, sex, race, ethnicity, drug use, and (P=.048), nonuse of antihypertensive

Dr Ances is the Daniel J. Brennan, MD, Professor of Neurology at Washington University at St. Louis in Missouri. Dr Anderson is Associate Profes- sor at Emory University in Atlanta, Georgia. Dr Letendre is Professor of Medicine at University of California San Diego. Send correspondence to Scott Letendre at University of California, San Diego, Antiviral Research Center, 220 Dickinson St, Suite A, Mailcode 8208, San Diego, CA 92103 or email [email protected]. 334 CROI 2021: Neurologic Complications Volume 29 Issue 2 April/May 2021

drugs (P=.001), and lifetime cannabis Permanente system and compared the at subclinical levels after 24 weeks. In use disorder (P=.043) (model P<.0001). incidence of dementia between peo- those who maintained viral suppression Speed of information processing, work- ple with HIV on ART (n=11,302) and on ART for 6 years, depressive symp- ing memory, motor functioning, and people without HIV (n=154,620) who toms remained stable or improved. verbal fluency were the cognitive abili- were followed up in primary care clin- As people with HIV age, they appear ties that contributed the most to global ics between 2000 and 2016. Incident to develop aging-related diseases such decline. all-cause dementia diagnoses were as diabetes and hypertension more fre- The HNRP (HIV Neurobehavioral Pro- identified using the International Clas- quently than people without HIV. This gram) at the University of California San sification of Diseases (ICD) codes that has raised concerns about premature Diego reported on cognitive decline in a were confirmed via chart review for aging of people with HIV, which may larger group of participants (n=1195), more than 300 randomly selected pa- be linked to the inflammation that can but over a shorter period of observation tients. A total of 264 people with HIV persist even during suppressive ART. (mean, 7.1 years), finding that a com- and 2006 people without HIV devel- An important question is whether peo- monly used index of comorbid condi- oped dementia during follow up. In- ple with HIV will also develop Alzhei- tions, the Charlson Comorbidity Index, cidence of dementia was 4.4 (people mer’s disease more frequently or at a was associated with more rapid global with HIV) and 2.1 (people without HIV) younger age than people without HIV. cognitive decline, particularly in work- per 1000 person-years, for an incident Several research groups have identi- ing memory (P=.007) and executive rate ratio of 2.0 (95% confidence in- fied that cognitive impairment in peo- functioning (P=.001) (Abstract 328). terval [CI], 1.8-20). The incidence rate ple with HIV has different phenotypes, The medical conditions that were most ratio was highest in younger people and this has been categorized by the strongly associated with cognitive de- with HIV but declined in older people National Institute of Mental Health in cline were diabetes, mild liver disease, with HIV, a pattern that has been iden- a Research Domain Criteria (RDoC) and congestive heart failure. tified elsewhere. framework.1 Using criteria from the A continuing debate in the field con- Chan and colleagues reported results Alzheimer’s disease field, Moore and cerns the extent to which HIV and its from people who were first assessed in colleagues previously identified that a effects on the immune system drive Thailand shortly after contracting HIV subgroup of people who were catego- brain injury compared with comorbid in the SEARCH (South East Asia Re- rized as having HIV-associated neu- conditions like diabetes, as this influ- search Collaboration on HIV) program. rocognitive disorder (HAND) may actu- ences treatment approaches to man- This group focused on longer term ally have amnestic mild cognitive im- aging conditions like cognitive impair- changes (6 years) with regard to de- pairment (aMCI), a distinction that ment and depression. The CHARTER pression instead of cognition (Abstract has implications for therapeutic man- analysis did not find that either HIV 340). Prior studies have found that agement. In another analysis based at RNA level or CD4+ cell count was as- depressive symptoms, like cognitive the University of California San Diego, sociated with cognitive decline but the performance, often improve after ini- Moore and colleagues classified 264 HNRP analysis found that lower CD4+ tiation of ART, but most have reported people with HIV as having either aMCI cell count was statistically significantly on shorter term change in people with (n=25), HAND (n=65), neither (n= associated with cognitive decline in chronic HIV infection. The focus of the 81), or both (n=95) in order to deter- with in-person analyses. Of note, nei- Thai cohort on people with acute HIV mine if this influenced their progression ther analysis included measures rel- infection is a strength that better allows over time (Abstract 331). The 4 groups evant to activity of the HIV reservoir indexing changes from the time of in- did not differ in duration of HIV dis- among people with HIV who are tak- fection and ART initiation. Participants ease, ART use, or plasma HIV RNA ing suppressive ART, such as low-level completed the 9-item Patient Health level, but did in age, CD4+ cell count, HIV RNA production (eg, using a sin- Questionnaire (PHQ-9) for depression and baseline cognitive performance, gle copy assay) or cell-associated HIV symptoms and the Distress Thermom- with the impaired groups performing DNA. Since HIV-1 infection increases eter (DT) for anxiety/stress at baseline worse than the group with neither the risk for many of these conditions (before ART initiation) and then again aMCI nor HAND. Over a mean of 4.2 (eg, diabetes, heart disease) compared 12, 24, and 96 weeks after ART initia- years of follow up, none of the groups with the general population, it may tion and every 48 weeks thereafter. At had significant decline based on fully indirectly be responsible for cognitive baseline, the prevalence of moderate adjusted models, which was contrary impairment even in the absence of di- (21%) and moderate to severe (27%) de- to hypotheses. The absence of cogni- rect effects on the brain. pression symptoms were high. Those tive decline would seem to be reas- Another approach was used by Lam individuals who had a higher baseline suring, but since 3 of the 4 groups had and colleagues to determine if people PHQ-9 score also had a higher plasma cognitive impairment at baseline, this with HIV were at greater risk for inci- HIV RNA level and a greater frequency also means that the impairment did dent dementia (Abstract 330). They of acute retroviral syndrome. As ex- not improve, which could mean that analyzed data from more than 165,000 pected, both the PHQ-9 and DT scores the injury is irreversible or that new electronic health records from the Kaiser improved after starting ART and were therapies are needed for people with

335 IAS–USA Topics in Antiviral Medicine

HIV who remain cognitively impaired clusion that people with HIV do not (n=297) on ART and people without even if they are already taking suppres- only have persistent systemic inflam- HIV (n=1509) to identify resting-state sive ART. This group also analyzed mation during suppressive ART but networks that relate to cognitive im- Alzheimer’s disease-related biomark- persistent neuroinflammation as well. pairment. The salience (SAL) and pari- ers (amyloid-β 1-42 [Aβ42] and phos- In Abstract 318, the potential differ- etal memory networks most strongly phorylated tau p-tau] 181) in CSF from ence in neuroimaging measures as a distinguished people with HIV (either function of sex was examined. Kelly cognitively impaired or unimpaired) NAA, a marker of neu- and colleagues compared brain volume from people without HIV. Comparing proportion in virologically-controlled cognitively impaired people with HIV ronal health, was lower people with HIV (n=286) and people with those without HIV, additional in- at baseline but normal- without HIV (n=105) of both sexes to volvement of the frontal parietal net- assess how biologic sex may modify work (FPN) was observed. Limiting the ized after 24 months of the effects of HIV on the brain. Total dataset to just people with HIV, the ART. In contrast, choline, brain, white matter, and gray matter SAL, FPN, basal ganglia, and ventral at- an inflammation marker, proportions were plotted as a func- tention network most strongly distin- tion of age. In general, brain volume guished cognitively impaired from un- continued to be elevated proportions were higher in women impaired participants. A deep learning 24 months after ART than in men irrespective of HIV status. model was also used to generate vox- initiation Greater loss of white matter volume elwise maps of resting-state networks over time was seen in men with HIV that identified changes in resting-state than in women with HIV. Consistent network topology among the 3 groups. 31 people with HIV who were older with this observation, men with HIV Anatomically, the greatest differences than 50 years and enrolled in the NNTC also had higher concentrations of the in resting-state network topology iden- (National NeuroAIDS Tissue Consor- axonal biomarker, neurofilament light, tified by deep learning occurred in the tium), finding that a higher p-tau/A 42 β than women with HIV (P=.02). These dorsal and rostral lateral prefrontal cor- ratio, which is typical of people with findings reinforce the importance of tex, anterior cingulate, parietal regions, Alzheimer’s disease, was associated designing mechanistic and therapeutic and caudate. These results support with worse Learning and Delayed Re- research that focuses on potential sex- that different resting-state networks call. Together, these (and other) find- based differences. are affected by HIV and cognitive im- ings support that people with HIV may With the growth of imaging datasets pairment. develop Alzheimer’s disease-type path- in recent years, high dimension ana- Petersen and colleagues also used ology, but they do not support the con- lytical methods, such as machine learn- machine learning to study brain ag- clusion that people with HIV are at ing, are increasingly used in neuroHIV ing among people with HIV. They greater risk than the general popula- examined brain structure (gray and tion. white matter volumes) and function The greatest differences (cerebral cortical blood flow) to char- New Data on Imaging acterize deviations from typical aging in resting-state network across the lifespan, finding evidence In addition to the depression findings topology identified by of accelerated brain aging in people (Abstract 340), the SEARCH program deep learning occurred with HIV, that is, machine learning- also reported on the effects on white based overestimation of the actual age matter brain metabolism of 2 years of in the dorsal and rostral of older people with HIV, compared ART that was initiated following acute lateral prefrontal cortex, with people without HIV or younger HIV infection (Abstract 163). A total people with HIV. They also showed that of 37 people with HIV were assessed anterior cingulate, cerebral blood flow reductions were by magnetic resonance spectroscopy parietal regions, and specific to older people with HIV who (MRS) within the left frontal white mat- caudate had detectable HIV RNA, and people ter (FWM) prior to ART (baseline) and with HIV with suppressed viral load 24 months after ART initiation. Com- were similar to people without HIV. pared with people without HIV, peo- research. For example, Luckett and col- Finally, imaging-based estimates of ple with HIV had lower total N-acetyl- leagues identified potential relation- brain aging correlated with cognition, aspartate (NAA) level, a marker of neu- ships between HIV infection, structural especially motor functioning. These ronal health, at baseline but values and functional organization of the brain, findings suggest that imaging-based normalized after 24 months of ART. cognition, and aging (Abstract 146). brain aging may be a useful nonin- In contrast, choline, an inflammation These authors used a machine learn- vasive biomarker of neuropathology marker, continued to be elevated 24 ing-based approach within a cohort of and cognitive impairment in people months after ART, supporting the con- virologically controlled people with HIV with HIV.

336 CROI 2021: Neurologic Complications Volume 29 Issue 2 April/May 2021

Viral Mechanisms in the trajectory following ART initiation re- integration sites. Based on snRNA-seq, Pathogenesis of HIV Disease mains to be proven, but prior research the investigators were able to identify in the Brain from this group suggests that it may.2 multiple clusters of cells (eg, excitatory Using brain tissue from the NNTC, the neurons, ) but HIV expres- The observational findings in the pre- same group also examined the phylo- sion was only present in those who ceding section are very valuable, but genetic relationship between envelope had HIV encephalitis, primarily in mi- they do not address the biologic mech- sequences from brain, CSF, and blood croglia. Differential expression analy- anisms of brain injury, which are es- from 5 people with HIV who had sub- sis revealed that microglia with active sential for designing interventions type B HIV-1 infection, had HIV-asso- viral transcription had greater expres- to treat the CNS complications of HIV ciated dementia, and were not taking sion of core markers of activation (se- disease. These mechanisms can be ART at the time of death. Three of the creted phosphoprotein 1[SPP1], lipo- broadly categorized into viral mecha- participants had brain- and CSF-derived protein lipase apolipoprotein E [LPL nisms and host mechanisms. Before sequences that were compartmental- APOE], FMS-related receptor tyrosine ized from blood-derived sequences. Al- kinase 1[FLT1]) and decreased expres- Higher compartmen- though the divergence of brain- and sion of markers of proliferation. These talized CSF viral load blood-derived sequences has been findings are somewhat diminished by was associated with demonstrated many times, the investi- many of the findings being present only gators extended their work to show in those who had HIV encephalitis, a worse global cognitive that HIV DNA concentrations were condition that is typically characterized performance, more highest in tissue from the basal ganglia, by high levels of HIV expression and the frontal lobe (white matter), and the that has a greatly reduced incidence in depressive symptoms, occipital lobe. Using the Affinofile assay the modern ART era. Even when no HIV and worse self-reported that allows the density of CD4+ expres- transcripts were detected, however, the functioning sion to vary, they also found that Env investigators still found evidence of en- proteins derived from the compart- hanced microglial activation, suggest- the widespread use of potent ART, pub- mentalized sequences from CSF and ing that low-level HIV may still be pres- lished data supported that compart- brain had greater affinity for CD4 and ent even if they did not detect it. mentalization of HIV in the CNS oc- thus enter cells with low CD4 density Two groups reported on HIV RNA or curred in a substantial proportion of more efficiently, which is typical of the DNA quantification in peripheral blood people with HIV and conferred greater myeloid cells that HIV-1 productively in- mononuclear cells (PBMCs), CD4+ T risk of brain injury. With expanding use fects in the brain. cells, or monocytes. The neuroHIV of ART, clues that HIV itself may con- Compartmentalization of HIV-1 in the group at St. Vincent’s Hospital in Syd- tinue to affect brain health trajectory brain has crucial implications for the ney, Australia, extracted cellular DNA are emerging. These clues include data HIV cure agenda. To better understand and RNA from CSF and blood collected on cell-associated HIV DNA, CSF viral the nature of HIV-1 persistence in the from 20 people with HIV who were tak- escape, and HIV proteins such as Tat. brain, the Manhattan Brain Bank site ing suppressive ART and used a more Joseph and colleagues reported on a of the NNTC isolated neuronal and glial sensitive method than has typically new concept, the compartmentalized nuclei from postmortem frontal cortex been used to quantify HIV-1 transcripts CSF viral load (CCVL, specifically the collected from 27 decedents (6 people and HIV-1 DNA, specifically the Dou- product of the percent of CSF-derived with HIV encephalitis, 15 people with ble-R assay that is based on πCode sequences in variable regions of the HIV without encephalitis, and 6 people MicroDiscs platform (Abstract 162). An HIV envelope [V1-V3] that are CSF-spe- without HIV) and created integration 18-color flow cytometry showed that cific and the quantity of HIV RNA in site sequencing libraries 10X chromi- cells in the CSF were 91% memory T CSF) (Abstract 341). Among 50 people um single nucleus RNA-sequencing cells, with roughly equal memory CD4+ with HIV assessed in the Rakai Cohort (snRNA-seq). They identified 1279 in- and CD8+ T cells. Other CSF cells were in Uganda, individuals with higher tegration sites, predominantly from the 3.1% CD14+CD16+ monocytes, 2.0% CCVL had worse global cognitive per- glial cell fraction from those with HIV natural killer (NK) cells, and 0.4% B formance (P=.049, particularly speed encephalitis. Glial integration sites were cells. HIV-1 RNA transcripts were quan- of information processing), more de- found preferentially in introns, gene tified in CSF in 90% of participants pressive symptoms (P=.003), and worse dense regions, and active regions of the (compared with about 10% in prior re- self-reported functioning (P=.044) com- . Glial integration sites showed ports) and HIV-1 DNA was quantified in pared with individuals with lower CCVL. a stronger preference for integration CSF in 80% of participants (compared This composite measure had larger ef- into short interspersed nuclear element with about 50% of historical controls). fect sizes in relation to these outcomes repeats than T-cell integration sites and Concentrations of both were higher in than did each of the individual compo- contained a significantly lower propor- CD4+ T cells from CSF than in PBMCs, nents used to calculate it. Whether the tion of clonal (5% vs 18%; P<.0001) although the significance of this finding pre-ART CCVL influences brain health and recurrent (13% vs 30%;P<.0001) is reduced since the comparison of the

337 IAS–USA Topics in Antiviral Medicine

CD4+ T-cell subset in CSF with the total or, when HIV-1 RNA in blood is de- impairment from HAND (Abstract 320). mononuclear cell fraction from blood tectable, having HIV-1 RNA at least 0.5 They measured soluble biomarkers of will amplify the differences between log10 copies/mL higher in CSF than in vascular injury (ICAM-1, VCAM-1, C- the compartments. Participants were blood. Its presence can occur with se- reactive protein), inflammation (inter- vere neurologic symptoms or with no feron [IFN]-α, interleukin [IL]-1β, IL-6, Greater HIV transcrip- symptoms at all and some, but not all, IL-8, IL-15, C-X-C motif chemokine li- studies have found associations with gand 10 [CXCL10], chemokine ligand tional activity in CSF ART (eg, use of HIV protease inhibi- 2 [CCL2], vascular endothelial growth cells correlated with tors) and HIV-1 (eg, presence of drug factor [VEGF]), and brain injury (total worse neuronal resistance mutations) characteristics. tau, glial fibrillary acidic protein [GFAP], Investigators from the Swiss HIV Co- YKL-40) in CSF and blood from 143 integrity in frontal hort and the NAMACO (Neurocognitive people with HIV on ART and 64 people white matter and Assessment in the Metabolic and Aging without HIV from the NNTC and the Cohort) study sought to validate these CHARTER Study. Overall, people with posterior cingulate findings by reporting on CSF collected HIV had higher levels of intercellular cortex from 287 people with HIV. CSF viral adhesion molecule (ICAM)-1, C-reactive escape was present in 29 (10.1%), of protein, IL-8, IL-15, CXCL10, and VEGF also assessed with 1H-magenetic reso- whom 18 (62%) had suppressed plasma in plasma and higher levels of C-reac- nance spectroscopy and greater trans- HIV-1 RNA and 11 (38%) had detectable tive protein, CXCL10, VEGF, and GFAP criptional activity in CSF cells corre- plasma HIV-1 RNA. Characteristics of in CSF than people without HIV. Among lated with lower NAA (ie, worse neu- patients were comparable whether or people with HIV, those with HAND had ronal integrity) in FWM (P=.04) and not they had CSF viral escape, includ- higher plasma ICAM-1, vascular cell posterior cingulate cortex (P=.055). ing demographics, cardiovascular and adhesion protein (VCAM)-1, C-reactive Using the same overall design as metabolic comorbidities, time since protein, and YKL-40 as well as brain prior SEARCH program analyses (ie, HIV diagnosis (12 vs 16 years, respec- injury biomarkers (CSF total Tau, GFAP, identifying people with acute HIV in- tively; P=.40), CD4+ T-cell count (553 YKL-40) than those without HAND. Fur- fection and then following them up vs 611 cells/µL; P=.10), CNS Penetra- thermore, cerebrovascular disease was over time after ART initiation), inves- tion-Effectiveness score (7 vs 8; P=.20), more prevalent among people with HIV tigators isolated monocytes from cryo- neurocognitive diagnosis, or presence who had HAND than among those with- preserved PBMCs from 30 people with of magnetic resonance imaging (MRI) out HAND and was associated with HIV and quantified total monocyte abnormalities. These findings confirm higher VCAM-1 and YKL-40 levels in HIV-1 RNA by real-time polymerase that CSF viral escape occurs in a minor- plasma and higher total Tau and YKL-40 chain reaction prior to ART and 96 ity of people with HIV, but its pathologic levels in the CSF. Overall, these results weeks after ART initiation. Monocyte significance continues to be unclear, support that vascular disease may be HIV-1 RNA was detected in 17 (57%) although it may have implications for more closely related to brain injury in participants at baseline, but only in 3 achieving functional cure in the people people with HIV on ART than inflam- of 30 (10%) at 96 weeks. The investi- who have it. mation. The importance of vascular gators compared these findings with a disease was further supported by Cooley panel of soluble myeloid activation and Host Pathogenesis of HIV and colleagues, who examined the ef- proinflammatory biomarkers in blood Disease in the Brain fect of risk on and performance on a screening bat- white matter integrity, as measured by tery of 3 neuropsychologic tests and The persistent inflammation that oc- diffusion tensor imaging in people found that participants who had de- curs in people with HIV increases their with HIV (Abstract 319). The Framing- tectable monocyte HIV RNA at baseline risk for numerous aging-related comor- ham cardiovascular disease risk score had higher neopterin concentrations bidities, including vascular disease and was calculated for 166 virologically and worse performance on 2 of the diabetes. Since inflammation and its well-controlled people with HIV and neuropsychologic tests (Color Trails 1, downstream consequences have been 48 people without HIV. Cognitive per- P=.014, and Trailmaking A, P=.05). linked to worse cognition and depres- formance and fractional anisotropy of Taken together, these findings support sion, understanding which of these most major white matter tracts in the brain that HIV itself may continue to affect strongly influences brain health trajec- were compared between low, moder- brain health trajectory, even during tory is important for directing therapy at ate, and high cardiovascular disease suppressive ART, but sufficiently sen- the most impactful target. Guha and col- risk for the 2 groups. Results indicated sitive methods are needed. leagues studied the potential role of ce- that a moderate or high cardiovascular Many groups have found evidence rebrovascular disease and its contribu- disease risk was associated with worse of CSF viral escape, which is generally tion to cognitive impairment in peo- cognitive performance (psychomotor defined as detectable HIV-1 RNA in ple with HIV by focusing on biomark- speed) and lower fractional anisot- CSF when it is undetectable in blood ers that distinguish vascular cognitive ropy within several major white matter

338 CROI 2021: Neurologic Complications Volume 29 Issue 2 April/May 2021

tracts including the frontal aslant, fron- of inflammation in blood (IL-6, soluble proportions from 2 independent co- tal occipital, and inferior longitudinal tumor necrosis factor receptor (TNFR)- horts of virologically suppressed wom- fasciculus for both people with HIV and I, soluble TNFR-II, C-reactive protein). en with HIV-1 infection in Baltimore people without HIV. Since the Framing- Higher body fat composition (total per- (n=19) and New York City (n=18) (Ab- ham cardiovascular disease risk score cent fat and visceral adipose tissue) stract 322). A higher proportion of contains several modifiable compo- was also associated with worse cogni- intermediate (CD14+CD16+) mono- nents (eg, smoking, blood pressure, tion. These results contrast with those cytes, which express medium levels of and cholesterol), the results suggest of Abstract 320 and support target- CCR2, high levels of CX3CR1, are (C-C that treating these conditions may im- ing inflammation but, of note, these chemokine receptor type 5) CCR5 posi- prove white matter injury in people investigators did not include vascu- tive, and have proinflammatory effects, with HIV. lar biomarkers in their analysis for was associated with lower global cogni- Molsberry and colleagues also inves- comparison. tive performance at the time of cog- tigated cardiovascular disease risk fac- Vecchio and colleagues evaluated nitive testing (P=.006) and approxi- tors and explored the relationship be- sex differences in 16 soluble biomark- mately 1 year before cognitive testing tween statin use and cognitive perfor- ers of inflammation and immune ac- (P=.02). A higher proportion of classi- mance over time in participants from tivation in CSF in relation to cognitive cal monocytes was also associated the MACS (Multicenter AIDS Cohort performance in a cohort of virally sup- with better cognition (P<.05). In con- Study) (n=1407) (Abstract 338). In pressed people with HIV (n=83) in trast, no associations were found be- addition to their lipid lowering effect, Uganda (Abstract 321). Overall, men tween monocyte subsets and mental statins have anti-inflammatory prop- performed worse on various cogni- health indicators, such as symptoms of erties that can improve endothelial tive measures including timed gait, depression or anxiety, although lower function and enhance dynamic cere- motor skills, executive performance, C D 4+ T-cell proportion was associated bral blood flow. These effects could and semantic fluency than women, with higher perceived stress (P=.03). potentially improve cognitive perfor- and men had stronger associations be- Collazo-Rodriguez and colleagues mance. Using multivariable-adjusted tween biomarkers and cognition than aimed to determine if exosomes from linear regression to compare cognitive did women. Thus, inflammation may blood of women with HIV-1 infection in- test performance prior to and after stat- more strongly influence cognition in duced a shift in uninfected monocytes in initiation, statin use was not associ- Ugandan men than women with HIV. toward this pathologic intermediate ated with improved performance on Further supporting a role for inflamma- phenotype. They found that exosomes any neuropsychologic test on the first tion in men, Anderson and colleagues from women with HIV-1 were taken test completion after statin initia- presented data from the MACS in the up by both uninfected classical and in- tion. Further analysis identified that United States on GlycA, a composite termediate monocytes and that they people with HIV who initiated statins blood biomarker of glycosylated pro- caused a shift in phenotype toward in- tended to have, on average, a faster teins that reflects acute phase reactants termediate monocytes and the extent rate of cognitive decline. HIV serostatus (including alpha1-acid glycoprotein, of the shift did not depend on whether did not modify this association. These haptoglobin, and others) in 843 men, women had cognitive impairment or results caution against use of statins to 63% of whom had HIV-1 infection (Ab- not, although exosomes from women protect or improve cognition. stract 324). In multivariable analy- who had cognitive impairment caused El-Kamari and colleagues studied ses, higher GlycA was associated with a more rapid shift toward the interme- the association between cognitive per- impairment when incorporating 5 other diate monocyte phenotype. Because of formance and biomarkers of inflam- biomarkers individually (C-reactive pro- this well demonstrated importance of mation, insulin resistance, and body tein, IL-6, CCL2, sCD14, and soluble myeloid cells in the pathogenesis of fat composition (by dual-energy X-ray CD163). The association between GlycA HIV-1 in the brain, Hammonds and absorptiometry [DEXA]) in ART-treated and impairment was driven by peo- colleagues evaluated the performance people with HIV (n=65) and people ple with HIV, particularly those with a of 4 myeloid cell models, specifically without HIV (n=33) (Abstract 329). higher C-reactive protein level (odds 2 microglial model cell lines (C20, Cognitive function was evaluated using ratio, 1.84; 95% confidence interval, HMC3) and 2 sources of primary cell- Cognivue (6 cognitive domains and 2 1.17-2.89). This study provides more derived microglia (monocyte-derived performance parameters). People with evidence that systemic inflammation microglia [MMG] and induced plu- HIV had worse overall cognitive perfor- plays a role in cognition among people ripotent stem cell-derived microglia mance and had worse performance in with HIV. [iPSC-MG]) (Abstract 351). Significant multiple domains (visuospatial, mem- Myeloid cells remain critically im- differences were observed upon gene ory, executive function, naming/lan- portant in the pathogenesis of HIV-1 in expression profiling, with MMG and guage, delayed recall, and abstraction). the brain. Several reports extended iPSC-MG clustering closely with pri- Among people with HIV, worse cogni- work in this area, including Veenhuis mary human microglial cells, and C20 tive performance in multiple domains and colleagues, who used a test-valida- and HMC3 exhibited marked differ- was associated with higher biomarkers tion approach to assess myeloid cell ences. Consistent with these differences,

339 IAS–USA Topics in Antiviral Medicine

iPSC-MG and MMG were readily in- and either the quantity of mtDNA or discontinued (3 due to nasopharyn- fected with R5-tropic HIV-1, and C20 of the mitochondrial common deletion geal irritation). Compared with place- and HMC3 required pseudotyping for from a buccal specimen, finding that bo, intranasal insulin was associated infection. HIV replication dynamics people with HIV (n=124) had higher with improvements in global cogni- and HIV-1 particle capture, however, levels of both measures than people tion (P=.029) at 24 weeks, which was differed noticeably between MMG and without HIV-1 (n=25) (P<.0001) (Ab- iPSC-MG. Based on these and other stract 326). When they compared these findings, the investigators concluded mitochondrial measures with a panel Compared with placebo, that the iPSC microglia model provided of soluble biomarkers in CSF and blood, intranasal insulin was a more authentic HIV-1 model system they found that higher mtDNA level was associated with improve- than the alternatives. These investiga- associated with higher soluble TNFR- tors are developing a 3-dimensional II (P=.042) and higher amyloid-β 1-42 ment in global cognition cerebral organoid model using these (P=.0005) levels even after adjusting at 24 weeks, which was and other cells. for HIV serostatus and demographic As noted earlier, people with HIV characteristics. The association with driven by improvements may experience earlier brain aging than amyloid-β 1-42 was present in the sub- in verbal memory, visual people without HIV-1 and understand- group of people with HIV, even after memory, and attention ing the biologic mechanisms by which adjusting for duration of HIV and ART, and nadir and current CD4+ T cell count Higher epigenetic (model, P<.0001). A higher level of the driven by improvements in verbal mitochondrial common deletion was memory, visual memory, and atten- age acceleration was associated with higher soluble TNFR- tion. On a personal note, this study was associated with worse II level in blood ( =.004) but not for led by Dr Ned Sacktor, whose untimely performance on atten- CSF biomarkers. death in late 2020 cannot overshadow Volpe and colleagues also worked the many important contributions he tion and working mem- with mtDNA but used MutPred path- made to the field. ory in people with HIV ogenicity scores to evaluate the influ- Considering the importance of my- ence of mtDNA variants on cognitive eloid cells in the pathogenesis of HIV-1 but not in people with- performance (Abstract 325). Among disease in the brain, many have con- out HIV 744 people with HIV, the presence of sidered that CCR5 antagonists, such as any deleterious mtDNA variant was maraviroc, might have particular bene- this occurs is a key gap in the field. In associated with motor impairment fit in the brain. Shikuma and colleagues addition to imaging-based methods, (P=.03), even in multivariable analy- reported findings from a randomized, brain aging can also be inferred by es- ses. In ancestry-stratified multivariable placebo-controlled trial of adjunctive timating cellular aging using methods analyses, people of European ancestry therapy with maraviroc in people with such as DNA methylation or mitochon- (n=317) also trended toward having an HIV who had sustained plasma HIV drial DNA (mtDNA). Shiau and col- association with speed of information RNA suppression and at least mild cog leagues measured DNA methylation in processing and people of African an- nitive impairment (Abstract 333). Par- whole blood using Illumina EPIC Arrays cestry (n=357) trended toward having ticipants were randomized 2:1 and in 69 people with HIV and 38 people an association with working memory followed up for 48 weeks, resulting without HIV who lived in New York (P values=.06-.08). Overall, these re in 39 participants who completed all City. The National Institutes of Health sults support that biomarkers that have evaluations. At baseline, those ran- (NIH) Toolbox Cognition Battery was been linked to premature biologic ag- domly assigned to the maraviroc arm used to assess cognitive performance ing in people with HIV are also associ- had worse global (P=.002) and motor across 5 domains. Overall, chronologic ated pathologic events in the CNS. (P=.001) performance. Participants age correlated with DNA methylation- in the maraviroc arm had significantly greater improvements in the combined estimated biologic age, but people with Interventional Findings on the learning/memory domain (P=.009) HIV had higher mean epigenetic age Effects of HIV-1 on the Brain acceleration (EAA) and extrinsic epi- than participants in the placebo arm. genetic age acceleration (EEAA) than Several clinical trials related to the CNS No other domains significantly changed people without HIV. Higher EAA was were presented. Yacoub and colleagues between the treatment arms. Of note, associated with worse performance on reported findings from a randomized, the study included people with HIV attention and working memory in peo- placebo-controlled clinical trial of in- with very mild impairment who might ple with HIV but not in people with tranasal insulin in 21 people with HIV not meet criteria for HAND (as mini- out HIV. who were taking suppressive ART and mal as neuropsychological testing per- Solanky and colleagues assessed who had at least mild cognitive impair- formance (NPZ),-0.5), and the learn- associations between HIV-1 infection ment. Six participants prematurely ing/memory domain change was not

340 CROI 2021: Neurologic Complications Volume 29 Issue 2 April/May 2021

statistically significant after adjusting in psychiatric symptoms and 5 do- Although 2-drug regimens have largely for type I error. mains of cognition were assessed at 48 been safe for the brain in shorter term With ongoing development of new weeks, at which point symptoms of de- analyses, these findings raise concerns ART drugs and treatment strategies, pression and anxiety were less severe about their long-term effects on inflam- switch studies continue to provide im- and sleep problems were less com- mation and coagulation. portant, clinically relevant data for peo- mon. Global cognitive performance Relevant to the HIV-1 cure agenda, ple with HIV and their medical practi- also improved, particularly executive McMahan and colleagues presented an tioners. Perez-Valero and colleagues function, attention/working memory, interim analysis of a trial using the pro- performed an open-label switch study and learning/memory. Since this trial grammed cell death protein-1 (PD-1) in which participants either continued also switched several drugs, improve- a suppressive regimen of dolutegravir/ ments cannot be attributed to a single A single dose of lamivudine/abacavir or switched to drug. Importantly, the absence of both pembrolizumab reduced darunavir/cobicistat/emtricitabine/ten- a comparison arm and the incorpora- ofovir alafenamide (Abstract 339). Neu- tion of practice effects in the analysis cell-associated HIV DNA ropsychiatric symptoms were evaluated of the neuropsychologic test results are in CSF by 46% at week with the Hospital Anxiety Scale, Hospi- limitations of the study. tal Depression Scale, Pittsburgh Sleep Another report that has implications three with persistent Quality Index, and neuropsychiatric for switching ART regimens in the clin- decrease of 7.7% at adverse events by the NIH Division of ic is the report on doravirine concen- week 24 AIDS criteria. At 4 weeks, participants trations in CSF (Abstract 358). A total who switched had a reduction in symp- of 14 plasma and 15 CSF samples were blocker pembrolizumab (Abstract 345). toms in 3 of the 4 categories (anxiety, collected with most participants re- Six people with HIV with virologic sup- neuropsychiatric adverse events, and placing an integrase strand transfer pression for at least 12 months and sleep) compared with those who did inhibitor with doravirine. At week 4, 1 CD4+ T-cell count above 350/mL were not switch (n=69 evaluable partici- participant met a common definition given a single dose of pembrolizumab. pants). This was particularly true for of CSF viral escape with detectable HIV No grade 3 or 4 adverse events oc- participants who had moderate to se- RNA CSF (32 copies/mL) but undetect- curred but grade 1 or 2 adverse events vere disturbances in sleep quality and able in blood plasma. Total doravirine involving blood, metabolic, and gas- sleep latency. Because the switch en- concentrations in CSF were approxi- trointestinal systems were observed. tailed a change of several drugs, it is mately 13% of those in blood. In CSF, PD-1+ CD4+ and PD1+ CD8+ T cells unclear which medication(s) may have doravirine was mostly unbound to in CSF decreased at 3 weeks and ap- driven the results. drug-binding proteins (76.1%). The to- peared to rebound at 24 weeks. Cell- Calcagno and colleagues also re- tal CSF; unbound plasma ratio of do- associated HIV DNA in CSF decreased ported a switch study in which peo- ravirine was 0.99, supporting the con- by 46% after 3 weeks and remained ple with HIV were randomly assigned clusion that doravirine crosses the persistently decreased by 7.7% after to continue a suppressive ART regi- blood-brain barrier primarily via pas- 24 weeks. The findings suggest that men or to change to a regimen that sive diffusion. single-dose is generally safe in peo- was designed to minimize potential An additional switch study may have ple with HIV and may decrease HIV neurotoxicity (darunavir/cobicistat/em- implications for brain health. Serrano- in the CNS, but this was an interim tricitabine/maraviroc) (Abstract 335). Villar and colleagues reported the ef- analysis of a small number of partici- The study was interrupted for slow fects of switching from a 3-drug to a pants within an ongoing study. accrual after 38 people with HIV had 2-drug regimen on inflammation bio- There was also a CNS trial using been randomly assigned and had been markers over time in 148 people with the nonhuman primate model. Garcia- followed up for 24 weeks. In the switch HIV evaluated in the Spanish AIDS Mesa and colleagues presented a study arm, parietal delta waves on electro- Research Network (Abstract 527). In in SIV-infected rhesus macaques of di- encephalography, which can indicate this nonrandomized trial that included methyl fumarate, which regulates ex- brain injury, decreased (P=.022) as did many 3- and 2-drug ART regimens, in- pression of antioxidant, anti-inflamma phosphorylated Tau in CSF (P=.002) vestigators found that participants who tory, and cytoprotective genes via its and liver fibrosis as measured by Fi- remained on a 3-drug regimen expe- effects on nuclear erythroid 2-related broscan (P=.038). rienced a slow decline of numerous factor 2-antioxidant response element In another switch strategy, Vergori biomarkers over time (IL-6, C-reactive (Nrf2-ARE) pathway (Abstract 347). Five and colleagues reported the results of a protein, soluble CD14, soluble CD163 of 9 animals were treated with dimethyl single-arm study in which 109 people and D-dimer). In contrast, switching to fumarate, followed by examination of with HIV on efavirenz/emtricitabine/ a 2-drug regimen was associated with 11 brain regions. Several antioxidant tenofovir disoproxil fumarate switched increases in IL-6, C-reactive protein, enzymes (GPX-1, NQO1, HO-1, and to bictegravir/emtricitabine/tenofovir and D-dimer (all P values ≤ .01) over PRDX1) were higher in the brains of alafenamide (Abstract 334). Change 3 years, after adjusting for covariates. the -treated animals

341 IAS–USA Topics in Antiviral Medicine

in various brain regions. Dimethyl fu- for other countries to monitor neuro- reactivity. This was followed by incuba- marate was also associated with lower COVID cases. tion of brain sections with whole CSF oxidative stress end products (3-nitroty- In an oral presentation, Farhadian from SARS-CoV-2-infected patients, rosine and 8-hydroxydeoxyguanosine), and colleagues from Yale and UCSF which again demonstrated increased particularly in the brainstem, as well reported a cross-sectional study of antineural immunoreactivity compared as lower mitochondrial redox state in immune responses in CSF collected with controls. These important findings both frontal cortex and brainstem. Al- support that the neurologic conse- though SIV RNA concentrations in CSF Anti-SARS-CoV-2 anti- quences of SARS-CoV-2 infection may and blood did not change following di- bodies were profiled and result from CNS-specific immune re- methyl fumarate, the investigators did sponses, including from the humoral not quantify SIV RNA or DNA in brain monoclonal antibodies immune response. One limitation of tissue. were derived, resulting the project is that it did not include biospecimens from people with SARS- Findings on the Effects of SARS- in the identification of CoV-2 infection who did not have neu- CoV-2 or Other Infections on the 2 antispike antibodies rologic complications. Brain Using an in vitro model, Clough, from blood and 1 from Mahajan, and colleagues examined the SARS-CoV-2 Infection CSF. The monoclonal effects of SARS-CoV-2 on the blood- Accumulating evidence shows that in- antibodies were then brain barrier (Abstract 346). Human fection with SARS-CoV-2 can lead to brain microvascular endothelial cells neurologic complications. In an invited incubated with mouse were treated with recombinant SARS- oral presentation, Benedict Michael dis- brain sections, resulting CoV-2 spike protein and heat-inacti- cussed ongoing work on neuroCOVID vated SARS-CoV-2. Compared with con- in the United Kingdom (Abstract 51). in 4 CSF antibodies trols, cells treated with SARS-CoV-2 had He first contextualized SARS-CoV-2 in that had antineural increases in hypoxia inducible factor relation to other that cause immunoreactivity (HIF)1/2, nitric oxide synthase, the neurologic disease, including herpes- NLRP3 inflammasome, and several viruses, West Nile virus, virus, from 6 hospitalized patients with SARS- cytokines including TNF-α and IL-6. and Japanese encephalitis virus. He dis- CoV-2 infection and 11 healthy controls ACE2 (necessary for SARS-CoV-2 cell cussed the fact that viruses do not nec- (Abstract 165). All 6 hospitalized pa- entry) was upregulated in treated brain essarily need to be neuroinvasive to tients had some degree of neurologic microvascular endothelial cells, and cause neurologic disease, and that some symptoms, including , blood–brain barrier integrity was de- viruses cause harm to the CNS by para- headache, and seizures. None had ele- creased by 30% in the SARS-CoV-2 infectious and autoimmune pathways. vated leukocytes or detectable SARS- treated samples. Lastly, levels of 4 dif- This led to the initiation of numerous CoV-2 RNA in CSF, although some had ferent tight junction proteins decreased neuroCOVID studies, including a large elevated total protein in CSF. Single-cell in the SARS-CoV-2 treated samples surveillance system to monitor neuro- mRNA sequencing was performed, compared with controls. Based on this COVID cases, which had identified 511 showing upregulation of pathways from study, SARS-CoV-2 appears to have a cases to date in the United Kingdom. both CD4+ and CD8+ T cells in CSF (in- detrimental effect on the blood–brain More than half (54%) involved a cere- cluding IL-1 and IL-12 responses) that barrier, which may be responsible for brovascular event, which appears to be were not upregulated in blood. Quanti- some of its neurologic sequelae. more common in older patients, and fication of cytokines confirmed increase neuropsychiatric syndromes are more Human Polyomavirus 2 Encephalitis in IL-1β and IL-12 p70 in CSF, which common in younger patients. However, again differed from blood (other bio- Progressive multifocal leukoencepha- cerebrovascular cases were generally markers such as CCL2 and IL-8 were lopathy (PML), caused by human poly- younger than non-COVID-19 stroke elevated in blood). Based on distinct omavirus 2 (also known as JC virus), cases collected from other surveillance CSF plasma cell clusters that were found still occurs in people with HIV, even systems. Patients with cerebrovascular in the SARS-CoV-2-infected patients, in some individuals with CD4+ T-cell events presented around the same time anti-SARS-CoV-2 antibodies were pro- counts greater than 200/µL. Pinnetti as the onset of respiratory symptoms, filed and monoclonal antibodies were and colleagues (Abstract 336) present- and patients with other neurologic derived, resulting in the identification ed a small open-label study of 5 people symptoms more frequently presented of 2 antispike antibodies from blood with HIV with PML who were taking about 2 weeks after onset of respira- and one from CSF. The monoclonal ART and who were treated with the tory symptoms. The most significant antibodies were then incubated with PD-1 blocker, pembrolizumab. They ob- predictors of poor outcome overall mouse brain tissue, resulting in 4 CSF served a decrease in PD-1 expression were older age and frailty. This surveil- antibodies (including the antispike an- on CD4+ and CD8+ T cells in blood lance system could serve as a model tibody) that had antineural immuno- and CSF in all participants, as well as

342 CROI 2021: Neurologic Complications Volume 29 Issue 2 April/May 2021

a decrease in human polyomavirus 2 phase were more likely to start ART (P=.001) species was significantly asso- DNA in CSF and an increase in hu- and less likely to be lost to follow up ciated with peripheral neuropathy pain man polyomavirus 2-specific T cells in than those in the clinician-directed symptoms in people with HIV, raising blood. However, 2 participants died phase. There was also a trend toward the possibility that manipulation of the after a relatively short period and 1 more frequent diagnosis of cryptococcal microbiome could be an approach to participant experienced immune re- meningitis in the point-of-care phase. peripheral neuropathy in future studies constitution inflammatory syndrome Thus, point-of-care CD4+ T cell count of people with HIV. (IRIS). Although the changes in disease and cryptococcus testing may be ben- indicators is promising, more research eficial in settings where low CD4+ T cell All cited abstracts appear in the is needed on pembrolizumab and other counts are common. CROI 2021 Abstract eBook, therapies for PML in people with HIV to available online at evaluate safety and efficacy. Peripheral Neuropathy www.CROIconference.org

Cryptococcosis Complications of the peripheral ner- Financial affiliations in the past 12 months: vous system remain common despite Dr Ances has no relevant financial affiliations Cryptococcal meningitis continues to ART, although their severity is markedly to disclose. (Updated April 11, 2019) Dr An- be a devastating opportunistic infec- reduced akin to the pattern observed derson has no relevant financial affiliations to tion in people with HIV with low CD4+ with cognitive impairment in the mod- disclose. (Updated May 7, 2021) Dr Letendre cell counts, particularly in sub-Saharan ern treatment era. Ellis and colleagues has had travel paid for by ViiV Healthcare. Africa. Drain and colleagues reported (Abstract 348) evaluated the relation- (Updated March 26, 2019) a prospective laboratory screening ship between peripheral neuropathy study for cryptococcal antigen in Um- pain symptoms and gut microbiota Additional References Cited lazi Township, South Africa (Abstract in 373 adults (72% people with HIV, in Text 564). This study (n=908) occurred in 3 90% of whom were virologically sup- chronologic phases. Clinician-directed pressed). Peripheral neuropathy pain 1. National Institute of Mental Health. Re- central laboratory cryptococcal anti- symptoms were more common and search Domain Criteria. 2021. 2. Bowman NM, Joseph SB, Kincer LP, et al. gen testing occurred in the first phase, more severe in people with HIV (P< HIV compartmentalization in the CNS is followed by reflex central laboratory .02). More severe pain was associated associated with neurocognitive impair- ment [CROI Abstract 401]. In Special Is- cryptococcal antigen testing based on with lower gut microbial diversity (de- sue: Abstracts From the 2016 Conference CD4+ T cell count in the second phase, termined by 16S rRNA sequencing) on Retroviruses and Opportunistic Infec- and then followed by point-of-care in people with HIV but not in people tions. Top Antiv Med. 2016;24(e-1):153. CD4+ T cell count and cryptococcal an- without HIV. Specifically, change to tigen testing by lateral flow in the third Lachnospira species from either Ru- Top Antivir Med. 2021;29(2):334-343. phase. Participants in the point-of-care minococcus (P=.007) or Streptococcus ©2021, IAS–USA. All rights reserved

343