Treating cognitive impairments in primary central nervous system infections: A systematic review of pharmacological interventions

Background: This research synthesized scientific evidence on the use of pharmacotherapy as intervention to reduce cognitive impairments in adult patients with primary central nervous system (CNS) infections. Methods: We searched for experimental studies published in English prior to October 2021 in MEDLINE, Embase and Cochrane databases. We included non-randomized studies (NRS) and randomized control trials (RCT) of pharmacotherapy versus placebo, drug, or a combination of drugs in adults with primary CNS infection. The certainty of the evidence was rated according to GRADE guidelines. Results: We included 8 RCTs and 1 NRS, involving a total of 805 patients (50.77% male patients; mean age 42.67 ± 10.58) with Lyme disease (LD), herpes simplex virus type 1 (HSV-1), or Creutzfeldt–Jakob disease (CJD) studying the efficacy of antibiotics, antiviral, and non-opioid analgesic drugs, respectively. In patients with LD, antibiotics alone or in combination with other drugs enhanced certain cognitive domains relative to placebo. In patients with HSV-1, the results were inconsistent. In patients with CJD, flupirtine maleate enhanced baseline cognitive scores. The quality of RCT studies was low, and the quality of NRS of intervention was very low, suggesting low and very low certainty in the reported results. Conclusion: There is limited evidence and low certainty regarding the efficacy of antimicrobials and analgesics in reducing cognitive impairments in patients with LD, HSV-1, and CJD. Future efforts must be aimed at enhancing attention to clinical trial methodology and reporting, as well as reaching a consensus on outcome measures and the endpoint of clinical trials relevant to patients.


Introduction
Infections of the central nervous system (CNS) are a heterogeneous group of disorders with respect to etiology and the effect on the nervous system. An infection refers to "the invasion of an organism's body tissues by disease-causing agents, their multiplication, and the reaction of host tissues to the infectious agents and the toxins they produce." [1] Many infectious agents (e.g., viruses, bacteria, fungi, prions, protozoa, and helminths) [2] may impose damage via direct invasion of the nervous system and early activation of the immune system with subsequent brain tissue damage, causing primary CNS infections, [3] and leaving the patient with cognitive impairments that may or may not follow a progressive neurological course.

Several longitudinal studies have reported a relationship between CNS infections during mid-life and cognitive
This research had no external funding source. During the work on this project, the senior author (TM) was supported by the Canada Research Chair for Neurological Disorders and Brain Health and the Alzheimer's Association fellowship grant (AARF-16442937). The funders had no role in study design, data collection, decision to publish, or preparation of the manuscript.
The corresponding protocol was registered before the initiation of the current work with the International prospective register of systematic reviews (PROSPERO; registration ID CRD42022330851). No amendments to the registered protocol were made.
The authors have no conflicts of interest to disclose. impairment later in life, [4] when the patient is no longer affected by the infection, indicating the long-term effects infection can have on cognition. [4,5] Long-term cognitive impairments within one or several domains of cognitive function such as memory, orientation, comprehension, learning, and language due to CNS infections have been noted to affect 20% to 40% of patients. [5] Research has shown that patients with CNS infections are also at increased risk of developing deficits in auditory differentiation and fine motor function. [6,7] The pathological processes through which CNS infections may influence cognitive function are generally divided into early functional (i.e., relating to regulation and maintenance of cerebral blood flow) and structural changes in the brain (i.e., increased white matter lesions, cerebral atrophy) in response to inflammation. [3,5] The "seeding" hypothesis [8] suggests that infectious agents invade the CNS and stimulate the microglia to induce an immune reaction that changes the levels of enzymes (i.e., neprilysin and insulin-degrading enzymes via inflammatory cytokines) that trigger production of amyloid proteins. Failure to clear amyloid proteins raises inflammation and expedites amyloid accumulation, which is a mechanism recognized in dementia. [8] These processes can be attenuated via drug interventions targeting infectious agents and/or the inflammation they cause. [9] For example, antibiotics (i.e., antibacterial, antiviral, and antiparasitic) either inhibit growth or kill infectious agents that cause CNS infection, invade neurons and cause damage. [10,11] Polyphenols have immunomodulatory and anti-inflammatory properties. [12,13] Many other medications with anti-inflammatory or neuroprotective properties [14,15] may also be beneficial in protecting neurons by slowing down or stopping microglial activation and, consequently, preventing or reversing cognitive deterioration. Hence, an evaluation of the role of pharmacotherapy in diminishing cognitive impairments stemming from primary CNS infection is warranted. We conducted a systematic review on the use of pharmacotherapy as intervention to reduce cognitive impairments in adult patients with primary CNS infections, with objectives to: provide a summary of the evidence on pharmacological interventions used in the treatment of cognitive impairments; evaluate whether these interventions improve cognitive outcomes; and report on specific intervention characteristics that may have an impact on results of interventions.

Methods
This systematic review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.

Eligibility criteria
Peer-reviewed intervention studies published in English, conducted in adult patients (≥18 years old) with primary CNS infections that measured cognition through clinical diagnosis or standardized measurement tools, regardless of the setting, were considered eligible. Case reports, pediatric studies, dissertations, and articles with no primary data were excluded. Eligible studies were selected according to the inclusion criteria by the PICOS strategy: • Participants (P): adult participants aged 18 years and above with primary CNS infection in any geographic region worldwide for whom intervention was prescribed to support cognitive recovery. • Study design (S): experimental studies, i.e., randomized controlled trials (RCT) and nonrandomized studies (NRS) of interventions were considered.

Information sources and search strategy
Researchers in collaboration with the information specialist at the large rehabilitation research teaching hospital developed a subject-specific MEDLINE search strategy and modified it several times to retrieve all relevant studies. The  A senior review author (T.M.) monitored the quality of the process and acted as quality assurer and arbiter if variances concerning study eligibility were noted between the reviewers. Disagreements were resolved through group discussion.

Data collection process and data items process.
Two independent reviewers (J.S. and S.S.) independently conducted a full-text review of relevant articles, extracted, and collected data using a standardized data extraction sheet, developed by the senior review author (T.M.), which was piloted in the first 3 studies in the review, for study characteristics and outcome data. Data extracted from the studies that met the inclusion criteria included: • Study identifiers (i.e., author names, publication year, setting, country). • Study characterizations (type of infectious agent, objectives, groups, type of intervention, design, sample size, inclusion and exclusion criteria). • Methodological qualities (i.e., study design, attrition rate, inclusion/exclusion criteria, sample size). • Participant characteristics (i.e., mean age, sex, range, duration of illness, comparison at baseline). • Intervention characteristics (i.e., dosage, frequency, mode of administration, duration of administration, change over time). • Outcomes, tools used to measure cognitive function, cognitive domains assessed (i.e., complex attention; executive functioning; learning and memory; language; perceptual-motor ability; and social cognition, [16] statistical methods and key findings (pre-intervention, post-intervention, change from pre to post and statistical results).
Two review authors (S.R. and J.S.) independently checked each study's characteristics for accuracy of data extraction against the study report. A senior review author (T.M.) monitored the quality of the process and double-checked that data were entered correctly.

Assessment of risk of bias in included studies
2.4.1. Study risk of bias assessment. For data from the NRS intervention study, [17] we used the Risk Of Bias in Nonrandomized Studies of Interventions tool, version of 2016, [18] assessing the following 7 domains: bias due to confounding, bias in the selection of patients into the study, bias in classification of interventions, bias due to deviations from the intended intervention, bias due to missing data, bias in the measurement of outcomes, and bias in the selection of the reported results. The risk of bias for RCT studies was assessed using version 2 of the Cochrane risk-of-bias tool for randomized trials (RoB 2). The RoB 2 considers the risk of bias in 5 domains which include the randomization process, deviations from the intended intervention, missing outcome data, measurement of the outcome, and selection of the reported results; each domain was judged as having "low," "some concerns," or "high" risk of bias based on an algorithm. [19] The overall risk of bias judgment was made for each study -a study was considered "low" when all domains were designated "low" risk of bias; "some concerns" when at least 1 domain was designated as having "some concerns" but where no domains had a "high" risk of bias; and "high" when at least one domain was designated "high" risk of bias or when multiple domains had "some concerns" in a way that lowered the assessor's confidence in the results. [19] Each study was independently assessed by 2 researchers (J.S. and S.R., with T.M. overseeing the process), who then compared their final assessments. In the case of a disagreement, the study in question was independently assessed by the senior author (T.M.) and all group members had a discussion to reach a consensus.

Effect measures.
The continuous outcome of change in cognitive function from pre-to post-intervention was presented as mean difference, group-time mean difference, or difference in change, as the data was provided in each study. The results were stratified by type of infectious agent, type and dose of drug, duration of drug administration (where possible), and cognitive domain and sub-domains assessed. When analyzing RCT studies, our interpretations were based on the principle of intention to treat (i.e., we analyzed the results of patients in their randomized group). A threshold of P < .05 designated statistical significance.

Dealing with missing data
We contacted study investigators to verify key study characteristics, aiming to obtain missing data where possible. Because of no response, we considered the impact of missing data in studies on the overall risk of bias assessed.

Synthesis methods
The Cochrane recommendations for conducting meta-analyses were used to determine if a meta-analysis was possible based on the assessment of clinical heterogeneity. [20] Study characteristics were tabulated into a file to evaluate similarities between studies (Table S2, Supplemental Digital Content, http://links. lww.com/MD/J200). The heterogeneity observed at all levels across PICOS criteria (i.e., population, intervention type, dose and duration, comparison, outcome measures, and study design), precluded a meta-analysis in its classical form. We carried out a best evidence synthesis to synthesize the results. [21] The studies were grouped based on the type of intervention they utilized, the infectious agent that caused CNS infection, and the cognitive outcome(s) measured used in clinical trials. Data conversions were not feasible as all studies reported changes in different domains and/or subdomains of cognition and used different measures (Table S2, Supplemental Digital Content, http://links.lww.com/MD/J200). Changes in scores from pre-to post-intervention in each of the reported domains and sub-domains were reported as follows: (i) "+" indicating improvement; (ii) "−" indicating decline; and (iii) "=" indicating no change. Aggregated changes reported in each study, by cognitive domain and sub-domains, were presented visually. To explore the extent and causes of heterogeneity, the study population, intervention dose(s), and duration of application were narratively synthesized. Studies with extensive missing data were noted, and methods used within those studies to handle missing data were reported.

Reporting biases and certainty assessment
The certainty of evidence was assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system. [22] The GRADE system downgrades the evidence by evaluating the extent of study limitations, including risk of bias, inconsistency of effects, imprecision, indirectness, and publication bias. [22] The certainty of evidence can also be upgraded by evaluating the extent of large effects, dose responses, and explanation of plausible confounding effects. [22] The certainty of the evidence was judged as high, moderate, low, or very low.

Sensitivity analysis
We carried out sensitivity analyses to test whether critical methodological factors or decisions affected the results. We grouped the main results by study design (RCTs or NRS) and compared the results of studies that focused on the same outcome, stratified by infectious agent, and intervention characteristics (Table  S2, Supplemental Digital Content, http://links.lww.com/MD/ J200). We then organized results based on overall certainty, considering risk of bias, inconsistency, indirectness, imprecision, publication bias, and sample size ( Table 5).

Ethical review
This study did not involve primary data collection; therefore, ethical approval was not sought.

Sample characteristics
Sample sizes of the studies varied from 23 [17] to 280 [23] patients. The mean age of participants (in years) within the 9 studies varied from 28 [25] to 59, [34] with a mean age (standard deviation (SD)) of 42.67 (10.5) years across the included studies. The percentage of male patients varied from 30% [17] to 72%, [25] with a mean percentage of male patients of 50.7% across the 9 studies. The duration (in years) of primary CNS infections varied from 1.8 [17] to 9, [29] with a mean (SD) duration of 4.69 (3.16) across the 9 studies. Two studies did not report the duration of CNS infections, [32,34] while one study reported the median duration of the disease. [23]

Interventions in patients with LD
Five studies involving patients with Lyme disease (LD) utilized antibiotics to reduce cognitive impairments. [17,23,29,31,32] Three of these studies used ceftriaxone (e.g., a broad-spectrum third-generation cephalosporin antibiotic that has a long halflife and broader and stronger gram-negative coverage than first or second-generation cephalosporins) through intravenous (IV) administration in comparison to IV placebo. [29,31,32] Fallon et al [29] utilized 15 outcome measures to evaluate 5 cognitive domains (i.e., complex attention, executive functioning, learning and memory, language, perceptual-motor ability) at baseline and then at a follow-up at 12 and 24 weeks. The treatment, placebo, and healthy participant groups were similar in age, gender, premorbid intelligence quotient (IQ), and the treatment and placebo groups were of similar duration of LD. There were noticeable differences between treatment and placebo group compared to healthy participants in the Wechsler Memory Scale-3rd Edition [41] scores for immediate memory and delayed (general) memory. Improved cognitive scores favoring treatment (i.e., IV ceftriaxone) as compared to the placebo and healthy groups were reported at week 12. At week 24, there were no significant differences between the treatment, placebo and healthy control groups.
Kaplan et al [31] investigated the efficacy of IV ceftriaxone followed by oral doxycycline (e.g., a broad-spectrum tetracycline-class antibiotic, long half-life, used in the treatment of infections caused by bacteria and certain parasites with an anti-inflammatory action). Before receiving treatment, baseline assessment of cognitive function was performed via for the neuropsychological tests; follow-up assessment took place 90 and 180 days (~13 and 26 weeks) after treatment implementation. Cognitive function was measured using 4 outcome measures that covered 4 cognitive domains (i.e., complex attention, executive functioning, learning and memory, and perceptual-motor ability). There was no significant difference between the treatment and placebo groups at baseline in age, sex ratio, scores for the neuropsychological tests, and duration of the disease. Comparing cognitive scores at follow-up showed no differences between the treatment group and the placebo.
Krupp et al [32] investigated the efficacy of IV ceftriaxone in patients with LD on cognitive function and fatigue. Mental speed was assessed using the Alpha-arithmetic Test [42] at baseline and then at a follow-up assessment after 6 months (~36 weeks).  Table 1 Participant demographic and study intervention characteristics.
Author Inclusion and continuation in the study, patients had to achieve a score of at least 50% in 2 of 12 subtests of the dementia tests (ADAS-Cog, GoeCJDDT). Medicine Table 2 Studies that have examined the efficacy of drugs in patients with cognitive deficits stemming from primary infection of the central nervous system: summary of study.
Authors (     Baseline characteristics, including sample size, age, sex ratio, and LD symptoms, did not differ between treatment, placebo, and healthy controls groups. No group differences at follow-up assessment between treatment and placebo were reported. Mental speed processing was lower in treatment and placebo groups than in healthy controls. Berende et al [23] utilized the IV ceftriaxone and doxycycline or IV ceftriaxone and a combination of clarithromycin with hydroxychloroquine (e.g., disease-modifying antirheumatic drugs decreasing the activity of the immune system, used as antimalarials) in comparison to IV ceftriaxone and placebo. Changes in cognitive function from baseline to follow-up assessment at 14, 26, and 40 weeks were assessed using 6 outcome measures which measured 4 different cognitive domains (i.e., complex attention, executive functioning, learning and memory, perceptual-motor ability). Baseline characteristics, including sample size, age, sex ratio, and LD symptoms, did not differ between the treatment and placebo groups. The study reported improvements in several cognitive domains at 14, 26, and 40 weeks; this improvement was not specific to the treatment group.
The NRS of intervention by Fallon et al [17] utilized various antibiotics administered orally, IV, or intramuscularly compared to the untreated persons with LD. Cognitive function was measured using 3 outcome measures covering 6 cognitive domains (i.e., complex attention, executive functioning, learning and memory, language, perceptual-motor ability, and social cognition) at the baseline and after 4 months (~16 weeks). The groups did not differ in baseline cognitive performance, anxiety, and depression scores. Compared to the placebo group, the group administered antibiotics showed significant overall and individual cognitive performance improvement at follow-up; IV administered antibiotics showed the greatest cognitive improvement. No improvement was observed in the non-treated persons.

Intervention in patients with HSV-1
Three studies reported on the use of antiviral drugs to target cognitive impairments in persons with herpes simplex virus type 1 (HSV-1). [24,26,36] One study [26] studied the efficacy of 16 weeks of valacyclovir administration compared to placebo in HSV-1 positive and HSV-1 negative groups. Two studies [24,36] used valacyclovir (e.g., an antiviral drug that inhibits viral DNA polymerase, incorporates into and terminates the growing viral DNA chain, and inactivates the viral DNA polymerase) in combination with antipsychotics and compared results to placebo in combination with antipsychotics. Bhatia et al [24] reported on specifics of antipsychotics used (i.e., trihexyphenidyl), whereas Prasad et al [43] did not specify the antipsychotic drug.
Bhatia et al [24] assessed cognitive functioning using 2 outcome measures (i.e., Penn Computerized Neurocognitive Battery (PennCNB) [43] and Emotion Identification & Discrimination test [44] which measured 6 different cognitive domains at baseline and then at a follow-up assessment after 4 and 5 months (~20 weeks). Baseline characteristics, including clinical variables, demographics, and cognitive status, except spatial ability (i.e., better in the treatment group than placebo), did not differ between the treatment and placebo groups. Comparison of cognitive function scores showed no group differences in the PennCNB at follow-up assessment between treatment and placebo groups, whereas only patients that received the treatment significantly improved the Emotion Identification & Discrimination accuracy index scores.
The study by Prasad et al [43] evaluated change in 6 cognitive domains (i.e., complex attention, executive functioning, learning and memory, language, perceptual-motor ability, language, and social cognition) using the PennCNB. Cognitive performance was assessed at baseline and then at a follow-up 18 weeks later. The treatment and placebo groups were similar in age, and duration of illness with respect to baseline demographics. There were differences between the groups in Positive and Negative Syndrome Scale scores, with the placebo group scoring higher. Improvement in verbal memory, visual object learning, and working memory favoring treatment valacyclovir group compared to the placebo was reported at follow-up.
Breier et al [26] assessed 4 cognitive domains (i.e., executive functioning, learning and memory, perceptual-motor ability, and language using the Measurement and Treatment Research to Improve Cognition in Schizophrenia Consensus Cognitive Battery. [45] The study reported no significant differences in cognitive performance on any cognitive domains for the treatment group at follow-up compared to baseline. [26] 3. 6

. Interventions in patients with CJD
One study tested the analgesic drug flupirtine maleate's efficacy compared to the placebo in patients with LD. [34] Five cognitive domains (i.e., complex attention, executive functioning, learning, and memory, language, and perceptual-motor ability) were assessed using the Alzheimer's Disease Assessment Scale-Cognitive Subscale, [46] the Mini-Mental State Examination, [47] and the Goettingen Creutzfeldt Jakob disease (CJD) Dementia Test [34] at baseline and then at a follow-up assessment 2, 4, 8, 12, 16, and 20 weeks after treatment administration. There were no significant differences between the treatment and placebo groups in age, CJD clinical signs, and cerebrospinal fluid at the baseline. The treatment group showed greater improvement in all studied domains of cognition at follow-up as compared to the placebo.

Response to intervention by cognitive domain
Summarized data results of changes in different cognitive domains across the included studies are reported in Table 4. For detailed data results of changes in cognitive function from baseline to follow-up(s) in different cognitive domains across the included studies please refer to (Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202).
Of 5 RCTs in persons with LD that assessed complex attention, [17,23,29,31,32] two [31,32] assessed both attention and speed of information processing subdomains, 2 studies [17,29] assessed Medicine Table 3 Summary of the cognitive outcome measures utilized across the studies and the cognitive domains they measure.

Table 5
Evaluation of the quality of evidence using the GRADE guidelines. attention sub-domain and one [23] study assessed speed of information processing. Across studies that assessed the sub-domain of attention, the results were inconsistent: 2 studies [17,32] reported no change in response to treatment and 2 studies [29,31] reported change in response to treatment. Results across the attention processing sub-domain were also inconsistent: 2 studies [23,31] reported improvement and one study [32] reported no change in response to intervention (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202). In the 3 studies on patients with HSV-1, one study [24] assessed complex attention and attention sub-domains. Researchers reported no significant change in scores in response to treatment in patients with HSV-1 disease compared to their baseline scores (Table 4; Table S4, Supplemental Digital Content, http://links. lww.com/MD/J202).
Two out of the 5 studies on persons with LD assessed working memory sub-domain. [23,29] The results were as follows: one study [29] reported an improvement in working memory from baseline to follow-up at 10 weeks and one study [23] reported improvement from baseline to follow-up at weeks 26 and 40 in the ceftriaxone followed by doxycycline treatment group but not in the treatment group that received ceftriaxone followed by clarithromycin and hydroxychloroquine. The latter study also assessed the executive function sub-domain and reported no change across both treatment groups from baseline to follow-up [23] (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202).
Of the 3 studies in persons with HSV-1 that assessed executive function, [24,25,36] all assessed the sub-domain of working memory. Two studies reported no change in the working memory sub-domain from baseline to last follow-up at 22 weeks and 16 weeks, respectively, [24,25] and one study [36] reported improvement in working memory from baseline to follow-up at 18 weeks. One study [24] that assessed abstraction and mental flexibility sub-domain reported no change from baseline in response to treatment (Table 4; Table S4, Supplemental Digital Content, http://links. lww.com/MD/J202).
In the single study of persons with CJD [34] only the sub-domain of executive function was assessed. Researchers reported improvement from baseline in response to flupirtine maleate (Table 4; Table S4, Supplemental Digital Content, http://links. lww.com/MD/J202).
Four studies among the 5 on patients with LD evaluated the learning and memory sub-domains. [17,23,29,31] General memory was the only sub-domain that was commonly assessed between the 2 studies. [17,31] Both studies reported improvements in general memory from baseline to follow-up at 17 weeks and 13 weeks, respectively; [17,31] no further change from 13 weeks to 26 weeks was observed. [31] The former study [29] also assessed the overall composite score and reported improvement. An improvement in delayed memory sub-domain but not visual memory sub-domain was reported in one study. [17] One RCT assessed the episodic memory sub-domain and reported improvement from baseline to follow-up at week 26 but no change from baseline to follow-up in weeks 14 and 40 in the ceftriaxone followed by doxycycline treatment group. [23] In the treatment group that received ceftriaxone followed by clarithromycin, the study reported improvement from baseline to follow-up at weeks 14 and 26 but no change in cognition from baseline to week 40 (Table 4; Table S4, Supplemental Digital Content, http://links. lww.com/MD/J202).
Among the 3 studies in persons with HSV-1, 2 assessed learning and memory, [24,25] of which one assessed the sub-domain of face memory [24] and one assessed overall composite memory. [25] Both studies reported no changes in response to intervention (Table 4; Table S4, Supplemental Digital Content, http://links. lww.com/MD/J202).
In the single study of persons with CJD, only the overall composite cognitive score was assessed. [34] Study results indicate improvement in cognition in patients after treatment compared to baseline (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202).
All 5 studies on persons with LD assessed language dom ain. [17,23,29,31,32] Four of the 5 studies assessed the sub-domain of verbal fluency. [17,29,31,32] Two studies [29,31] showed improvement in verbal fluency from baseline to follow-up measurement, one study [17] reported no change, and one study [23] reported no change from baseline to 14 weeks, but improvement from baseline to follow-up at weeks 26 and 40. Two studies [17,32] assessed the verbal memory sub-domain: one study [17] reported improvement from baseline to 17 weeks and one study [32] reported no change from baseline to follow-up at 26 weeks (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202).
Of the two among the 3 studies on patients with HSV-1 that assessed language, [25,36] one [25] that assessed the letter-number sequencing reported no change from baseline to follow-up and one study [36] that assessed verbal memory reported change in response to treatment (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202).
In the study of persons with CJD, [34] verbal fluency and language were assessed. The study results indicated an improvement in language and no change in verbal fluency at follow-up compared to baseline (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202).
Among the 5 studies that included patients with LD, one study [29] assessed perceptual-motor ability, specifically the motor and psychomotor sub-domains, and reported improvement across both sub-domains at follow-up compared to baseline assessment (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202).
In the 3 studies on patients with HSV-1, 2 studies evaluated perceptual-motor ability. [24,25] One study [24] assessed spatial memory and sensorimotor ability, and one [25] assessed visuospatial memory and spatial ability. Both studies [24,25] reported no change in follow-up compared to baseline across all sub-domains within the perceptual-motor ability cognitive domain (Table 4; Table S4, Supplemental Digital Content, http://links. lww.com/MD/J202).
Only the spatial ability sub-domain was assessed in the study of persons with CJD. [34] Study results indicated that spatial ability performance improved in patients after treatment compared to baseline assessment (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202). Medicine 3.8.6. Social cognition. The efficacy of pharmacotherapy on social cognition was assessed in one study. [24] Researchers evaluated the sub-domain of emotion within the social cognition domain and reported an improvement in emotional recognition scores at follow-up compared to a baseline assessment in response to pharmacotherapy (Table 4; Table S4, Supplemental Digital Content, http://links.lww.com/MD/J202).

Risk of bias in studies and certainty of the evidence
One study had an overall "low" risk of bias score [29] and the rest studies scored either "some concerns" [23,24,32,36] or a "high" risk of bias score [25,31,34] (Table S5, Supplemental Digital Content, http://links.lww.com/MD/J203). Across the 5 domains assessed, there was substantial agreement between reviewers (kappa = 0.89). The studies were unclear regarding randomization, allocation, and adherence according to the study's protocol. [22] A NRS of intervention [17] was of very low quality. According to the GRADE guidelines, the quality of the selected RCT studies was low, and the quality of the NRS of intervention was very low, suggesting low and very low certainty in the reported results (Table 5). [22]

Loss to follow-up
In the 3 studies that reported loss to follow-up, [25,31,32] results show that with the intervention completion percentages of 83%, [25] 98%, [31] and 88% [32] the majority of participants completed their treatment protocols. Prasad et al [36] mentioned that loss to follow-up was measured, however, they did not report the information. The 5 remaining studies included no information regarding the completion of the intervention. [17,23,24,29,34]

Discussion
This systematic review provides a comprehensive evidence synthesis of intervention studies testing the efficacy of pharmacotherapy for reducing cognitive impairments in patients with primary CNS infections, a relatively unexplored topic. Nine studies comprised of 8 RCTs and one NRS of intervention described the efficacy of antibiotics, antiviral, and non-opioid analgesic drug in adult patients with LD, HSV-1, and CJD, respectively. The present review highlights the limited breadth of existing evidence and sources of variation in response to intervention even with the application of a similar class of medication.
While the available evidence suggests that antibiotics, antivirals, and muscle relaxant might improve certain domains/ sub-domains of cognition in patients with LD, HSV-1, or CJD at least temporarily, of the available evidence, only one study [29] had "low" risk of bias. Due to a lack of methodologically high-quality studies, variations in the timing of administration, and heterogeneity in outcome measures used to evaluate cognition, we found it difficult to draw generalizable conclusions about the efficacy of pharmacotherapy on the cognitive outcomes of patients with primary CNS infections. Further investigations are required to develop scientific hypotheses concerning optimal treatment approaches, the length and timing of administration of pharmacotherapy during the course of CNS infection, as well as qualitative insights from people about the functional meaningfulness of improvement in scores on cognitive measures in response to interventions.

Patient population
The lack of consistency between studies, even relating to the same infectious disease of the CNS, can be attributed to a variety of factors, the individual impacts of each a challenge to interpret. To determine whether differences in response to interventions between study samples were attributable to patients' characteristics such as age, general and brain health, comorbid disorders, or the immune response due to interaction of the infectious agent with immune system of a patient, time since infection, level of impairment at the time of treatment initiation, or a combination of all, was a complex task. At this time the nature and implications of the results have not been adequately explored given limited number of studies and information available within them. Nevertheless, considering the extensive evidence on the complexity of the molecular mechanisms used by pathogens to colonize, invade, infect, and disrupt patients' brain structures involved in cognitive function, [49,50] the response to a variety of therapeutic strategies described in this review is likely to be influenced by all of the above. Consequently, future research needs to take these considerations into account at the design phase of clinical trials.

Interventions
Cognitive impairments in patients with primary CNS infection may result from damage that occurs through a variety of mechanisms within the brain tissue (i.e., excitotoxic, inflammatory, immune, necrosis, and apoptosis). [51,52] The mechanisms targeted by the studies included in this review focused on reduction of the toxicity by an infectious agent and reducing the extent of secondary injury. These secondary mechanisms evolve over variable periods of time following exposure to an infectious agent, thus providing an opportunity for pharmacological interventions to minimize the long-term effect of the primary CNS infection. Timing of treatment initiation after CNS infection was variable across the studies. Most of the studies [23][24][25]29,31,36] applied the treatment years after exposure to the infectious agent. The remaining few studies [32,34] did not report such information or applied intervention to patients with variable durations of illness. [17] The duration of the interventions was also variable across studies, depending on the type of CNS infection and a class of pharmacotherapy, with a range between 4 weeks in patients with LD treated with antibiotics [32] and 18 weeks in patients with HSV-1 treated with an antiviral drug. [36] Given the current uncertainty regarding whether cognitive improvement can be sustained in the long term after treatment discontinuation, there is a pressing need for further studies to investigate this important question.
The dosage of similar classes of drugs within the same type of CNS infection appeared to be relatively consistent across the studies. An oral dose between 1000 and 1500 mg of valacyclovir twice a day was the preferred dosage for HSV-1 patients; an intravenous 2000 mg of ceftriaxone alone or followed by an oral dose of doxycycline (100-200 mg once or twice daily) was a preferred dosage for LD patients. A combination of different antibiotic types and the mode of administration in patients with LD was performed in several studies [17,23,31,32] to increase drug tolerability and expand the spectrum of action. In a sole study of patients with CJD, a flupirtine maleate was administered in the dose of 100 mg daily, followed by an increase to 100 g 3 to 4 times daily, with medication stopped when patients no longer fulfilled the inclusion criteria.
The dose utilized in older versus younger patients seems not to differ, and the dose administered in male and female patients was the same. The existent research has not yet brought forth a discussion to address drug efficacy in the youngest and the oldest adult patients with known differences in absorption, distribution, and elimination, as well as the effects of drug interactions when multiple drugs are used concomitantly. Likewise, considering the sex-linked dimorphism of human physiology and gender-linked pathology experience, reflected in drug intake, metabolism, pharmacokinetics, pharmacodynamics, and bioavailability, the activity of pharmaceutical therapy and its effectiveness is expected to be sex-and age-specific. These directions of research remain to be initiated.
In addition, over the past decades, the genome of humans has been decoded, opening the door to the study of the genetic variation that causes differences in drug response due to variability in the activity of a particular enzyme responsible for the elimination of the drug. [53] Finally, adoption studies have shown that the risk of acquiring infections is heritable, and in recent years there is a growing interest in genetic variation in the immune system related to susceptibility and severity of infection and associated impairments. Future studies designing pharmacological interventions in patients with primary CNS infections expect to report on hypothetical correspondence between targets of pharmacological agents and cognitive function.

Outcome measures
The included 9 studies used 29 different cognitive outcome measures, evaluating one or more cognitive domains and sub-domains (i.e., complex attention, executive functioning, learning and memory, language, perceptual-motor ability, and social cognition). Only 2 outcome measures, the PennCNB [43] and the WAIS, [48] utilized in 3 studies, [17,24,36] assessed all 6 cognitive domains. Currently, knowledge of the psychometric properties relevant to an evaluation of change over time (i.e., test-retest reliability, responsiveness to change) of the utilized 29 outcome measures in the population of interest is lacking, which brings up the question of whether the observed improvement was the result of practice effect or the true effect of an intervention. The way forward seems to be agreement on a group of outcome measures with established evaluative properties suitable for clinical trials in patients with primary CNS infections. Consistent use of outcome measures can aid future comparisons between trials, combining data from different sources to arrive at a precise estimate of the effectiveness of interventions. Additionally, there is a need to consider the utility of outcome measures that evaluate changes in cognitive capacity rather than impairment.

Safety and adverse effects
The patients' adherence to antimicrobial therapy was not reported in the included studies, a significant limitation as tolerability is an important factor for patients and clinicians when making treatment choices. [54,55] Reporting safety concerns is equally important and should not be omitted from consideration in future trials. Some studies have shown that broad-spectrum antimicrobials can adversely affect bodily functions, even after short-term use, in special groups of patients, such as children, the elderly, and patients with degenerative and medical disorders (e.g., cardiovascular, renal, and hepatic diseases), [56] causing confusion, delirium, and decline in attention. Reporting individual patient changes in response to intervention and safety parameters, rather than group reporting, is likely the most effective approach to evaluate changes in cognitive scores and address the cause of variability in response to pharmacotherapy across different patients.

Study strengths and limitations
The strengths of our study include comprehensive coverage of current research findings, careful appraisal of study quality, a focus on clinically relevant endpoints, and comprehensive analyses allowing comparisons between studies.
There are several limitations to this research we wish to acknowledge. The focus of this evidence synthesis was the efficacy of pharmacotherapy in patients with primary CNS infections. Despite attempts to include all clinical trials on the topic, it is possible studies were missed. We excluded studies that focused on cognitive impairments stemming from exposure to chronic infectious agents whose primary targets are not neurotropic tissue, such as human immunodeficiency virus (HIV), hepatitis C virus, and emerging pathogens like severe acute respiratory syndrome coronavirus 2 among others. The significance of studying cognitive impairment stemming from these infectious agents cannot be overstated.
Further limitations include a focus on peer-reviewed Englishlanguage studies only, omitting potentially relevant results published in other languages, and non-peer-reviewed works. As such, there is a possibility for publication and results' generalizability biases. There are also limitations to the presented data on cognitive outcome measures used in the reviewed studies. The evaluation properties of the outcome measures were not reported.
We were not in the position to utilize advanced statistical techniques such as meta-analytic principal component analysis and sparse meta-analytic principal component analysis, which have been suggested to aid in identification patterns and reduce complexity of clinical trial data) [57][58][59][60] for several reasons. The high heterogeneity of the data in terms of study design, intervention, population, and outcome measures in principle violates the assumptions of homogeneity in the covariance organization across studies included in this evidence synthesis, as require for principal component analysis. In addition, incomplete and inconsistent data coming from different studies measuring change in similar cognitive domain via different measures requires significant human expertise to transform data into a multivariate format but limited published procedure protocols are available to do so. In the future, with additional research on the topic, it may be feasible to conduct these types of analyses for high-dimensional data for pattern recognition and vizualization of complex clinical data. This would involve careful consideration of assumptions, limitations, and appropriate validation and sensitivity analyses to ensure that the results are reliable and valid.
Finally, we acknowledge that missing data in studies included in this review may have impacted the reliability of our review findings, reflected in the risk of bias and certainty of evidence ratings. We also chose not to exclude the studies with missing data, as we believe in the importance of synthesizing all available evidence with clear critical appraisal and certainty assessment reporting. To mitigate the concern and assess the impact of missing data on our findings, we performed a proxy of qualitative sensitivity analysis. This involved re-grouping the data using different methods to determine the similarities and differences between studies included in this review within different domains of cognition, infectious agents, medication class, and duration of intervention. However, this approach should be interpreted with caution, as it is not a substitute for statistical analyses.

Quality of evidence
We conducted the review using a robust methodology, with at least 2 review authors independently assessing studies for eligibility, extracting data, and carrying out risk of bias assessment. For the evaluation of quality of the evidence, where some of the methodologies were not reported in sufficient detail to enable us to make a judgment, we contacted authors for clarification.
We also contacted authors of primary studies when it appeared that similar groups of patients had been described in more than one publication, in order to avoid double counting participants. [39,40] Although the majority of the studies were classified as "randomized," concerns around the randomization process, deviations from the intended intervention, missing outcome data, measurement of the outcome, and selection bias of the reported results were not consistently addressed across studies. [17,[23][24][25]29,31,32,34,36] The study protocols were not available as publications for any of the included studies except for one study. [23] This increases the risk of reporting bias and impacts the study quality. The Centre for Reviews and Dissemination Guidelines and the Cochrane Reviewers' Handbook suggest that "quality" is an important consideration when performing reviews, with "quality" relating to the extent to which a study minimizes bias and maximizes internal and external validity. Because there was a limited number of studies to address our research objectives, the decision was made to show all the evidence available, even in light of the high risk of bias and low certainty.

Implications for clinical practice and research
The evidence synthesis results demonstrated only low certainty evidence for the efficacy of pharmacological interventions for dealing with cognitive impairments in adult patients with LD, HSV-1, or CJD. The existing evidence refers only to 3 specific classes of pharmaceuticals: antibiotics (alone or in combination with antipsychotics), antiviral drugs, and non-opioid analgesics, used in patients with LD, HSV-1, and CJD, respectively. Placebobased comparisons of antibiotics and antivirals demonstrated inconsistent efficacy, utilizing various cognitive outcome measures in patients with LD and HSV-1. The NRS of intervention utilizing non-opioid analgesics in patients with CJD reported improvement, although this was a single study of very low quality. Also worth noting is that one third of included studies came from outside North America -one from India, [24] one from Germany, [34] and one from the Netherlands. [23] While we did not observe patterns based on country of origin, the role of culture in the presentation of and coping approaches to cognitive impairments stemming from CNS infections cannot be fully discarded. These are particularly relevant because the psychometric properties of the outcome measures utilized in the studies included in this review have not been described in cross-cultural communities.

Conclusions
The limited number of studies on each medication class and infectious agent, high heterogeneity, and inconsistency in the data, including the outcome measures utilized, preclude definite conclusions on the ability of antimicrobials and non-opioid analgesics in reducing cognitive impairments in patients with primary CNS infection. More translational research is needed to characterize cognitive impairment and its early identification in patients with CNS infection and to develop new prognostic indicators of cognitive impairment course, adverse effects, and safety indicators in patients of various ages, sexes, baseline cognitive status, and disease severity. Efforts must also be aimed at assessing the determinants of specific domains of cognitive function that do respond to therapy, to define whether they relate to the characteristics of the patient or the infectious agent, or the timing, the dose, of mode of administration of the intervention, or treatment adherence.
To promote the advancement of clinical trial methodologies, researchers should reach a consensus on outcome measures in clinical trials and on the core endpoints of clinical trials that are relevant to patients. The development and utilization of advanced statistical methodologies that enable digital transformation of heterogeneous data and machine learning techniques to identify hidden patterns in data present promising opportunities for detecting distinct subtypes in patients, leading to the realization of personalized medicine and tailored therapeutic approaches.