Neurocognitive dysfunction and brain FDG-PET/CT findings in HIV-infected hemophilia patients and HIV-infected non-hemophilia patients

This single-institution cross-sectional study aimed to grasp the prevalence and features of neurocognitive dysfunction in HIV-infected hemophilia patients in Japan. We conducted neuropsychological tests and medical examinations in 56 HIV-infected hemophilia patients who received outpatient treatment at the AIDS Clinical Center, National Center for Global Health and Medicine. A total of 388 HIV-infected non-hemophilia patients who received outpatient treatment at the same institution were included as a control group. To investigate sites responsible for neurocognitive dysfunction in HIV-infected hemophilia patients using brain FDG-PET/CT scans, the accumulation of FDG in each brain region was compared. Approximately 50% of HIV-infected hemophilia patients had neurocognitive dysfunction. The prevalence of asymptomatic neurocognitive impairment was high (34%). Neurocognitive dysfunction was associated with educational level in HIV-infected hemophilia patients. In the symptomatic group, hemophilic arthropathy and history of cerebrovascular disorders were associated with neurocognitive dysfunction. Left temporal lobe function was reduced in the symptomatic group.


Introduction
HIV-associated neurocognitive disorders (HAND) occur even in patients in whom viral suppression with anti-HIV drugs is favorable [1]. HAND is associated with poor quality of life [2]. with acute or subacute lesions detected on brain MRI that could affect cognitive function and patients with foreign materials that were unacceptable for an MRI scan were also excluded. When applicable patients visited the hospital for an outpatient visit, the coordinator provided a written explanation of the present study and obtained consent by signature. Thereafter, patient information was collected from medical records. Other necessary information was collected by interviews on the morning of the neuropsychological testing day. Medical examinations were performed by a psychiatrist. In the afternoon, the same neuropsychological tests were performed in one session by the same clinical psychologist as that in the J-HAND study (8).
The following assessment items were collected from medical records and interviews: age, sex, educational level, work experience, presence/absence of persons living with the patients, smoking history, alcohol intake, previous history of AIDS-defining illness, hypertension, diabetes, anemia, dyslipidemia, previous positive history for treponema pallidum hemagglutination test (TPHA), presence of hepatitis B virus/hepatitis C virus (HCV) coinfection, presence/ absence of history of HCV treatment, HCV-RNA levels, minimum CD4 count (Nadir CD4), current CD4 count/HIV-RNA levels at the most recent neuropsychological testing, state of introduction of anti-HIV drugs (non-nucleoside reverse transcriptase inhibitor, protease inhibitor, and integrase strand transfer inhibitor), presence/absence of upper-limb dysfunction, classification of coagulation disorders, presence/absence of regular infusion, history of inhibitors, presence/absence of hemophilic arthropathy, history of cerebrovascular disorders, presence/absence of previous virological treatment failure, and presence/absence of previous discontinuation of treatment. Previous virological treatment failure was defined as twice or more viral load > 200 copies/mL of HIV-RNA as tested > 1 month apart at 6 months after start of anti-HIV drugs.
For control group data from HIV-infected non-hemophilia patients, the data obtained at ACC from the J-HAND study was used [8]. Three time zones (9 a.m. to 10 a.m., 10 a.m. to 11 a.m., 11 a.m. and later) were set in each of three outpatient examination rooms. Patients were recruited who met the selection criteria in the order of visit time, and verbal consent was obtained from up to three patients per day in each room. The control group included 388 randomly extracted Japanese HIV-infected patients who received outpatient treatment at the ACC between July 2014 and July 2016 (route of infection: sexual transmission, n = 385; other and unknown, n = 3). Participant exclusion criteria were in accordance with DSM-5 [9].

Assessment of cognitive functions
Psychiatric symptoms and functional status of daily activities were assessed using the Mini International Neuropsychiatric Interview (M.I.N.I) [10,11], the Japanese edition of the brief profile of mood states (POMS) [12], the general health questionnaire mental health questionnaire (GHQ-28) [13], and the scale of the 7-item version of the International Classification of Functioning, Disability and Health (ICF) core set [14,15]. Cognitive functions were assessed with the Mini-Mental State Examination (MMSE) and the following 14 tests in eight cognitive domains [16]: language (language fluency test [category] and [letter]) [17], attention/working memory (recitation in normal order/reverse order) [18,19] [24]. Based on the results of age-standardized neuropsychological tests [17,19,[23][24][25][26], patients with scores lower than one or more standard deviations (SDs) were identified. The Frascati Criteria was used for making a diagnosis of neurocognitive dysfunction and determining its severity [27].
On standardized neuropsychological testing, ANI and MND are documented by performance of at least 1SD below the mean involving at least two domains. HAD is ascertained with at least two domains with 2SD or greater than the mean. ANI does not interfere with daily functioning, MND produces at least mild interference, and HAD produces marked interference.

Assessment of imaging
Brain MRI was performed on a clinical 3.0-Tesla MRI unit (MAGNETOM Verio; Siemens AG, Erlangen, Germany) and included three-dimensional T1-weighted images (magnetization-prepared 180˚radio-frequency pulses and rapid gradient-echo sampling, MPRAGE) used for the computed analysis of Voxel-based Specific Regional Analysis System for Alzheimer's Disease (VSRAD) [28], T2 � -weighted images used for evaluation of microbleeding, MR angiography used for evaluation of cerebrovascular disorder, diffusion-weighted images used for evaluation of acute or subacute cerebrovascular attack, T2-weighted images, and fluid-attenuated inversion recovery image (FLAIR) used for evaluation of general brain disorders. All MR images were evaluated by an observer (TN) with 10 years of experience in neuroradiology.
For brain FDG-PET/CT scans, 18 F-FDG was synthesized in the hospital using synthesizing equipment, cyclotron (F200, Sumitomo Heavy Industries, Ltd.) at the National Center for Global Health and Medicine. Participants fasted for over 6 hours. After resting for over 15 minutes with an eye mask in a decubitus position in a dark, quiet room, 18 F-FDG 5 MBq/kg (lower limit: 185 MBq, upper limit: 370 MBq) was intravenously injected. The condition of wearing an eye mask in a decubitus position in a dark, quiet room was maintained until imaging was performed. Imaging was started with PET/CT equipment (Biograph mCT S20: Siemens Medical Solutions) 45 minutes after 18 F-FDG administration. Blood glucose levels at scanning were confirmed to be � 200 mg/dL in all participants. The obtained PET data were analyzed using the data analysis software of SIEMENS, MI Neurology, in which a total of 18 regions were set: left/right frontal lobes, left/right temporal lobes, left/right parietal lobes, left/ right cingulate and paracingulate gyri, left/right central regions, left/right occipital lobes, left/ right basal ganglia, left/right mesial temporal lobes, and left/right cerebellum. The 18 F-FDG accumulation in each region (SUVmean) was measured, and the SDs of SUVmean were compared. The mean value of the accumulation in the left and right regions whose SD was the smallest was set as the value of the control region. The ratios of the values of the other 16 regions to the control value (SUVr) were calculated for assessment.

Statistical analysis
For statistical analysis, the statistical software IBM SPSS Statistics ver. 23 was used.

1) Assessment of cognitive functions
To examine differences in patient characteristics between HIV-infected hemophilia patients (hemophilia group) and HIV-infected non-hemophilia patients (non-hemophilia group), χ 2 test or Mann-Whitney U test was performed. To examine differences in the prevalence of neurocognitive dysfunction, cognitive domains, and neuropsychological test results between the groups, the χ 2 test was performed. Results were considered statistically significant if the P value (p) was < 0.05.
To identify factors associated with NDHH, the χ 2 test or Mann-Whitney U test was performed between the NDHH group (ANI, MND, and HAD) and normal group (without ANI, MND, or HAD) among HIV-infected hemophilia patients. The χ 2 test or Mann-Whitney U test was also performed between the symptomatic group (MND and HAD) and asymptomatic group (normal and ANI). Accordingly, factors that were different at a significance level of 5% as explanatory variables were used to perform binary logistic regression. The items from which information could not be obtained were considered missing values.

2) Assessment of imaging
The Mann-Whitney U test was performed to compare the NDHH group (ANI, MND, and HAD) and normal group (without ANI, MND, or HAD) as well as the symptomatic group (MND and HAD) and asymptomatic group (normal and ANI) among HIV-infected hemophilia patients. A P value (p) < 0.05 was considered statistically significant.

Results
In total, 82 HIV-infected hemophilia patients received outpatient treatment at the ACC between May 1, 2016 and February 28, 2018. Of these, eight patients satisfied exclusion criteria, nine patients refused to participate in the study, one patient withdrew consent for participation, and one patient did not undergo neuropsychological testing during the study period; thus, 63 patients underwent neuropsychological testing and medical examination by a psychiatrist. Ten patients were diagnosed with psychiatric disorder in a medical examination by a psychiatrist: schizophrenia (n = 1), bipolar disorder (n = 1), dysthymia (n = 2), developmental disorder (n = 2), alcoholism (n = 1), and sleep disorder (n = 3). Of these, three patients with psychiatric disorders that could affect cognitive function (schizophrenia, bipolar disorder, and alcoholism) and four patients with unavailable recent brain imaging findings were excluded. A final total of 56 patients were included in analysis; however, brain FDG-PET was performed only in 55 patients since 1 patient missed the scheduled examination date.
There were significant differences in most items with regard to patient characteristics between the hemophilia group and non-hemophilia group (Table 1). In the hemophilia group, all patients were male (100%, χ 2 (1, N = 444) = 3.341, p = 0.048) and had hypertension (43%,

Associated factors of NDHH
The results of the χ 2 test or Mann-Whitney U test for the analysis of associated factors in the NDHH group between the NDHH group and normal group are shown in Table 3. A binary We performed the χ 2 test or Mann-Whitney U test to analyze factors associated with symptoms of neurocognitive dysfunction between the symptomatic and asymptomatic groups (Table 4). Binary logistic regression revealed that the presence of smoking history (p = 0.041), hemophilic arthropathy (p = 0.041), and history of cerebrovascular disorder (p = 0.018), which were significantly different at a significance level of 5%, were used as explanatory variables. The presence of hemophilic arthropathy (OR, 11.998; 95% CI, 1.130-127.403; p = 0.039) and history of cerebrovascular disorder (OR, 10.993; 95% CI, 1.779-67.922; p = 0.010) were suggested to be associated risk factors in the group with symptoms of neurocognitive dysfunction (χ 2 test, p = 0.002; Nagelkerke R 2 , 0.352; sensitivity, 62.5%; specificity, 95.8%).

Brain FDG-PET/CT findings in HIV-infected hemophilia patients
One patient who missed the FDG-PET examination date fell under MND. Therefore, of the 55 patients analyzed, 29 were in normal, 19 in ANI, 6 in MND, and 1 in HAD group. The mean value of SUVmean in the left cerebellum (4.70 ± 0.91) and the right cerebellum (4.88 ± 0.95) was 4.79 ± 0.93, and these SDs were the smallest. There was no difference in FDG accumulation between the NDHH group and normal group. Compared to the asymptomatic group, the symptomatic group exhibited a significant decrease in accumulation in the left temporal lobe (p = 0.021) ( Table 5).

Discussion
The present study suggested that the prevalence of NDHH was higher than that of HIVinfected non-hemophilia patients. In particular, the prevalence of ANI which does not hamper daily activities was higher. At present, the level of NDHH is mild in most patients, but cerebrovascular aging eventually leads to cognitive decline [29], so it may gradually advance as patients become older. Intracerebral hemorrhage is a known factor that increases the risk of developing cognitive impairment [29]. In the symptomatic group, since hemophilic arthropathy and presence of a history of cerebrovascular disorder were identified as risk factors for neurocognitive dysfunction, lifestyle review and regular administration of blood products are also important, and medical staff/supporters are required to manage these patients. In contrast, educational level (university or higher) was an associated protective factor in the NDHH group in this study, and it has been reported that higher education has a protective effect against neurocognitive impairments even for HIV-infected patients [30]. In the future, nurturing professionals who can provide support for these associated factors as rapidly as possible and prepare an appropriate medical environment for neurocognitive dysfunction are required.
TMT tasks are reported to be very sensitive to differences between a mild Alzheimer's disease group and a control group [31]. In clinical settings, where the assessment of cognitive functions in eight cognitive domains (verbal/language, attention/working memory, abstraction/executive, memory, speed of information processing, sensory-perceptual, and motor skills) [27] is difficult due to the use of multiple neuropsychological tests (lack of manpower, limited time for medical examination, etc.), conducting TMT-B may be useful as an easy assessment given its high sensitivity/specificity for NDHH. Although the brain sites responsible were not clarified in the NDHH group, the present study suggested that left temporal lobe function was reduced in the symptomatic group. Reduced FDG-PET metabolism within the temporo-parietal and prefrontal cortex were reported in elderly healthy control subjects that developed mild cognitive impairment or Alzheimer's disease, and the combination of TMT-B and FDG-PET/CT are considered useful for prediction of the clinical progress to dementia [32]. Although there have been studies indicating a significant relationship between temporal lobe metabolism and the severity of AIDS dementia [33], the decreasing pattern of FDG uptake in patients with HAND was considered nonspecific [4]. Of 56 hemophilia patients, 14 (25%) patients had history of cerebrovascular disorder. However, we were unable to identify the relationship between sites and hypometabolism because some patients lack medical records of the site, and some sites were unclear in current MRI.
In studies comparing HIV-infected hemophilia patients and HIV-uninfected hemophilia patients, HIV-infected hemophilia patients were associated with a decline in neurocognitive performance [6,7]. In our study comparing HIV-infected hemophilia patients and HIVinfected non-hemophilia patients, the prevalence of NDHH was high, and attention was impaired. These results suggest that both HIV infection and hemophilia may be risk factors for neurocognitive dysfunction. In the future, it will be necessary to conduct a prospective cohort study while observing whether subjects develop dementia in combination with brain MRI and molecular imaging modalities.
This study has several limitations. Many participants lacked records of coagulation factor activity during examination. Similarly, accurate analysis of the age of onset of intracranial hemorrhage, site, and physical impediment was difficult, as past medical records did not exist and confirmation was unavailable. For hemophilic arthropathy, we referred to medical records by physicians and did not base diagnosis on evaluation standards such as the hemophilia joint health score (HJHS) or Pettersson score [34,35]. Since upper-limb dysfunction was judged by Binary logistic regression revealed that educational level (university or higher), and the presence of history of cerebrovascular disorder, which were significantly different at a significance level of 5%, were used as explanatory variables. This suggested that educational level was an associated protective factor in the NDHH group. patients in interviews and not by disabilities of the arm, shoulder, and hand (DASH) [36], judgments may have varied depending on patients' subjectivity. Thus, accurate assessments are required using objective data with regard to hemophilia-related items for the analysis of associated factors in the NDHH. In the present study, we assessed TMT and GP using international standard values as standardized Japanese data were not available, and we did not use tests standardized for education level. At present, neuropsychological tests standardized for sex and education level are limited in Japan; thus, advancement in the neuropsychological field is desirable. In addition, the J-HAND study demonstrated that age � 50 years and incomplete virological suppression were identified as risk factors for the symptomatic group, and current treatment with anti-HIV drugs was identified as a protective factor for neurocognitive dysfunction [8]. However, the present participants were a homogeneous population with regard to these factors. Comparison with HIV-uninfected hemophilia patients in Japan to analyze causes of neurocognitive dysfunction are required in the future.

Conclusion
In the present study, 48% of HIV-infected hemophilia patients had neurocognitive dysfunction that met the diagnostic criteria for HAND. In particular, the prevalence of ANI was high. The analysis of associated factors with neurocognitive dysfunction identified educational level in the NDHH group and hemophilic arthropathy and history of cerebrovascular disorder in the symptomatic group. TMT-B may be useful for the rapid screening of neurocognitive dysfunction in busy medical facilities. In addition, the present study suggested that left temporal lobe function was reduced in the symptomatic group. Thus, longitudinal follow-up of the asymptomatic group by brain FDG-PET/CT may improve understanding of the prognosis of neurocognitive dysfunction. In the future, for HIV-infected hemophilia patients to easily receive treatment and support when they manifest symptoms of neurocognitive dysfunction, professionals with the requisite knowledge and skills to appropriately assess and support patients are necessary, alongside a medical environment that promotes patient support and prevention of the initial onset and progression of neurocognitive dysfunction.

Acknowledgments
We would like to thank Editage (www.editage.jp) for English language editing. ・The SDs of the accumulation in the brain parenchyma (SUVmean) were compared, and the mean value of the accumulation in the left and right regions of the cerebellum, whose SD was the smallest, was used as the value of the control region. ・Significance test: Mann-Whitney U test ・Significance level: p < 0.05 There was no difference in 18-fluorine-fluorodeoxyglucose accumulation between the NDHH group and normal group. Compared to the asymptomatic group, the symptomatic group exhibited a significant decrease in accumulation in the left temporal lobe. Abbreviations: SUV, standardized uptake value; NDHH, neurocognitive dysfunction in HIV-infected hemophilia patients; ANI, asymptomatic neurocognitive impairment; MND, mild neurocognitive disorder; HAD, HIV-associated dementia; SD, standard deviation; L, left; R, right. https://doi.org/10.1371/journal.pone.0230292.t005