Influence of Strongyloides stercoralis Coinfection on the Presentation, Pathogenesis, and Outcome of Tuberculous Meningitis

Abstract Background Helminth infections may modulate the inflammatory response to Mycobacterium tuberculosis and influence disease presentation and outcome. Strongyloides stercoralis is common among populations with high tuberculosis prevalence. Our aim was to determine whether S. stercoralis coinfection influenced clinical presentation, cerebrospinal fluid (CSF) inflammation, and outcome from tuberculous meningitis (TBM). Methods From June 2017 to December 2019, 668 Vietnamese adults with TBM, enrolled in the ACT HIV or LAST ACT trials (NCT03092817 and NCT03100786), underwent pretreatment S. stercoralis testing by serology, stool microscopy, and/or stool polymerase chain reaction. Comparisons of pretreatment TBM severity, CSF inflammation (including cytokines), and 3-month clinical end points were performed in groups with or without active S. stercoralis infection. Results Overall, 9.4% participants (63 of 668) tested positive for S. stercoralis. Active S. stercoralis infection was significantly associated with reduced pretreatment CSF neutrophil counts (median [interquartile range], 3/μL [0–25/μL] vs 14 /μL [1–83/μL]; P = .04), and with reduced CSF interferon ɣ, interleukin 2, and tumor necrosis factor α concentrations (11.4 vs 56.0 pg/mL [P = .01], 33.1 vs 54.5 pg/mL [P = .03], and 4.5 vs 11.9 pg/mL [P = .02], respectively), compared with uninfected participants. Neurological complications by 3 months were significantly reduced in participants with active S. stercoralis infection compared with uninfected participants (3.8% [1 of 26] vs 30.0% [33 of 110], respectively; P = .01). Conclusions S. stercoralis coinfection may modulate the intracerebral inflammatory response to M. tuberculosis and improve TBM clinical outcomes.

The soil-transmitted helminth Strongyloides stercoralis causes strongyloidiasis, a neglected chronic parasitic disease of humans. Found throughout tropical and subtropical regions of the world, S. stercoralis infects an estimated 30-100 million individuals globally [1]. The geographic distribution of S. stercoralis overlaps with that of tuberculosis. Tuberculous meningitis (TBM) is the most severe form of tuberculosis, resulting in death in almost half of all cases, despite effective antituberculosis chemotherapy [2][3][4][5]. TBM is characterized by intracerebral inflammation, which can lead to fatal complications.
Helminth coinfection appears to modulate the host immune response to Mycobacterium tuberculosis infection and may increase susceptibility to developing disease (tuberculosis) and worsen its severity [6]. Helminth infections typically induce a T-helper (Th) 2 immune response, with an immunoglobulin E antibody class switch, production and activation of eosinophils, mast cell degranulation [7], and marked elevation of interleukin 4, 5, and 13 (IL-4, IL-5, and IL-13) [6]. Th2 responses appear to be cross-inhibitory with the proinflammatory Th1 immune responses associated with tuberculosis [8,9]. In a case-control study of 40 individuals with pulmonary tuberculosis, significantly lower blood interferon (IFN) ɣ levels and a nonsignificant trend toward more severe disease were found in helminth-coinfected individuals compared with helminth uninfected controls [10]. A study of proinflammatory cytokines in patients with pulmonary tuberculosis (n = 88; 42 of 88 coinfected with S. stercoralis) and latent tuberculosis (n = 88; 44 of 88 coinfected with S. stercoralis) found significantly lower plasma tumor necrosis factor (TNF) α, IFN-ɣ, and interleukin 2 (IL-2) in S. stercoralis-coinfected individuals, compared with a tuberculosis-only control group [11]. In addition plasma concentrations of anti-inflammatory cytokines interleukin 10 (IL-10), IL-4, IL-5, and IL-13 were significantly elevated in individuals with latent tuberculosis and S. stercoralis compared with those with latent tuberculosis alone.
The intracerebral inflammation of TBM is poorly understood. A Th1 immune response is typical, with phagocytosis, intracellular killing of microbes [7], and elevated cerebrospinal fluid (CSF) concentrations of proinflammatory cytokines [12][13][14], such as TNF-α and IFN-γ. However, previous studies have shown substantial heterogeneity in the response, with poor outcomes associated with both excessive and attenuated inflammatory responses [15][16][17]. The determinants of this heterogeneity are uncertain; host genetic variation in leukotriene A4 hydrolase (LTA4H) may play a role in some populations [18,19], but other determinants are likely. Here, we examine the hypothesis that helminth coinfection modulates the intracerebral inflammatory response to M. tuberculosis and thus influences the clinical presentation and outcomes of TBM.

Participants
We performed a prospective study in Vietnamese adults with TBM to evaluate the frequency and effect of S. stercoralis coinfection on presenting clinical phenotype, CSF inflammatory parameters, CSF cytokine concentrations, and clinical end points. Participants were enrolled from 2 ongoing randomized placebo-controlled phase III trials of adjunctive corticosteroid therapy for human immunodeficiency virus (HIV)-coinfected and HIV-uninfected adults with TBM (ACT HIV [NCT03092817 [20]] and LAST ACT [NCT03100786 [21]]).
Participants were ≥18 years old, with a diagnosis of TBM based on consistent clinical and CSF findings, with or without HIV coinfection, and admitted to the Hospital for Tropical Diseases or Pham Ngoc Thach Hospital for Tuberculosis and Lung Disease, both in Ho Chi Minh City, Vietnam. Patients were excluded if an additional brain infection to TBM was suspected, if they received >6 consecutive days of antituberculosis chemotherapy or systemic corticosteroids, or if corticosteroids were mandatory or contraindicated.
Written informed consent was obtained from all participants or from a relative if the participant was incapacitated. Ethical approvals for ACT HIV and LAST ACT were obtained from the Oxford Tropical Research Ethics Committee (nos. 36-16 and 52-16, respectively), the ethical committees of the Hospital for Tropical Diseases (nos. 14/HDDD and 37/HDDD, respectively) and Pham Ngoc Thach Hospital for Tuberculosis and Lung Disease (nos. 1033/HDDD-PNT and 460/HDDD-PNT, respectively), and from the Vietnam Ministry of Health (nos. 108/CN-BDGDD and 151/CN-BDGDD, respectively).

Clinical Data
Demographic data (age, sex), baseline Modified Research Council (MRC) TBM severity grade [22], and HIV status were recorded. Study participants were followed up for 3 months. Death and neurological complications by 3 months were recorded. Neurological complications were defined as a fall in Glasgow coma score of ≥2 points for ≥48 hours, a focal neurological sign, seizure, cerebellar signs, coma, or cerebral herniation.

Laboratory Testing
All participants enrolled in this study underwent ≥1 test for S. stercoralis infection. S. stercoralis serology (NovaTec Immunodiagnostica) was performed in participants at baseline (date of signing informed consent). Routine wet preparation stool microscopy was performed in participants within 7 days of baseline. Stool S. stercoralis PCR testing was performed in a subgroup of participants (those testing positive for S. stercoralis at serology or stool microscopy [allowing comparison of diagnostic tests] and in consecutively enrolled participants until a total of 200 PCR tests had been performed). Blood eosinophil count was measured at baseline in all participants.

Treatment
All participants received antituberculosis chemotherapy following national guidelines. Rifampicin, isoniazid, pyrazinamide, and ethambutol were given for at least the first 2 months, if drug resistance was not suspected or proved. Pyrazinamide was stopped after 2 months. At least 12 months of antituberculosis chemotherapy was received in total. Antituberculosis chemotherapy regimens are further described in the Supplementary Material. Participants with a positive result of stool microscopy or a PCR test for S. stercoralis received oral ivermectin (200 µg/kg/d for 10-14 days), with repeated stool microscopy required to demonstrate absence of S. stercoralis larvae. Participants with positive results of S. stercoralis serology were treated on a case-by-case basis. In addition, all participants were randomized to dexamethasone or placebo (termed "study drug"), a double-blinded allocation following 1:1 randomization (except LTA4H TT-genotype HIV-uninfected participants from the LAST ACT trial [approximately 7% of total participants] who all received open-label dexamethasone). The study drug was administered over 6-8 weeks, following a tapering course, with weekly reductions (Supplementary Table 1). The ACT HIV and LAST ACT trials are ongoing, and treatment allocations remain blinded. Permission to publish these study data was obtained by the respective trial steering committees.

Statistical Analysis
Primary analysis populations were selected based on clinical categories of S. stercoralis infection. The S. stercoralis "uninfected" group consisted of participants tested with S. stercoralis serology, stool microscopy, and stool PCR, with all results negative. This approach gave the highest certainty of a S. stercoralisuninfected status. A "past infection" group consisted of participants with positive results of S. stercoralis serology, with no positive stool result (but with stool microscopy and/or stool PCR performed). An "active infection" group consisted of participants with positive S. stercoralis stool microscopy or stool PCR results, regardless of other testing.
Secondary analyses were performed on 2 additional subpopulations; participants who had serology performed, and those who had both serology and stool microscopy performed (divided into groups A-C) (Supplementary Tables 2 and 3). Secondary analyses compared baseline TBM severity, CSF inflammatory parameters, and clinical end points between participants with or without positive S. stercoralis test results, for each subpopulation. CSF cytokine analysis was performed only for primary analysis populations.
Where CSF cytokine concentrations were undetected, either the lowest limit of extrapolation divided by 2, or the lowest limit of detection divided by 2, was used, whichever was lowest. CSF cytokine testing was performed across two 96-well plates, and value extrapolation was plate specific. Rarely, where cytokine concentrations were too high for quantification, the upper limit of detection multiplied by 2 was used; samples and testing kits were unavailable for sample dilution and repeated testing. Log 2 calculations of CSF cytokine concentration were performed. Given the unknown magnitude of S. stercoralis immunomodulation of CSF cytokine concentrations, sample size calculation was not possible. The number of CSF samples undergoing CSF cytokine analysis was based on availability of sample and testing kits and was therefore exploratory. Clinical data proportions were compared using χ2 tests, and CSF cytokine concentrations using Wilcoxon rank sum tests. Multivariate analysis (with odds ratios and 95% confidence intervals) was performed to evaluate whether age, MRC TBM grade, HIV coinfection, and active S. stercoralis infection predicted neurological complications by 3 months. Data were analyzed using R software (version 3.6).

Influence of S. stercoralis Infection on TBM Presentation and Routine CSF Parameters
A comparison of baseline TBM severity and routine CSF parameters between primary analysis populations is shown in Table 1 [26]. Baseline blood eosinophil counts were significantly elevated in active S. stercoralis infection compared with S. stercoralis-uninfected participants (median [IQR], 0.10 [0-0.38] ×10 9 /L vs [0-0.10] 0 ×10 9 /L), respectively (P = .02). The median (IQR) CSF neutrophil count and neutrophil percentage were reduced in active S. stercoralis infection compared with uninfected participants (neutrophil count, 3 Tables 2 and 3 show results of baseline TBM severity and CSF inflammatory parameter analyses in subpopulations of participants in whom serology or both serology and stool microscopy were performed. In participants with S. stercoralis serology performed, HIV coinfection was less common in those with positive than in those with negative serological results (20.8% [11 of 53 participants] vs 46.2% [280 of 606], respectively; P = .001).

Baseline CSF Cytokine Concentrations in S. stercoralis Coinfection
We hypothesized that in participants with active S. stercoralis infection, CSF concentrations of the proinflammatory cytokines IFN-ɣ, IL-2, and TNF-α would be reduced and CSF concentrations of the regulatory cytokines IL-4, IL-5, IL-10, and IL-13 would be increased, compared with S. stercoralis-uninfected participants. These cytokines, in addition to IL-1β, IL-6,    [26]. One participant in the uninfected group did not have CSF parameters available and could not be classified as definite, probable, or possible TBM. One in the past infection group scored <6 points for the TBM diagnostic score [26]. Both cases were considered to be TBM by the treating clinician and were treated as such. c Reference standard (final diagnosis or grade against which comparison was made). infection, compared with uninfected participants, CSF concentrations of IL-13 were reduced (7.5 vs 23.9 pg/mL; P = .03) and CSF concentrations of IL-5 were increased (0.37 vs 0.37 pg/mL; P = .02). Median cytokine concentrations for S. stercoralis-uninfected, past infection, and active infection groups, together with ratio of change and statistical comparison between groups, are shown in Supplementary Table 5. IL-6 concentrations showed the greatest ratio of reduction, approximately a 54-fold reduction in active S. stercoralis infection compared with uninfected participants. CSF concentrations of TNF-α, IFN-γ, IL-1β, IL-2, IL-4, IL-6, and IL-10 were significantly reduced in participants with active S. stercoralis infection, compared with uninfected participants, in an HIV-coinfected subgroup; however, these significant differences were not seen in an HIV-uninfected subgroup (Supplementary Table 6).  Table 9). Additional secondary subpopulation comparisons of neurological complications and death by 3 months, in participants with or without positive S. stercoralis results, are shown Supplementary Tables 10 and 11. In participants with serology   well recognized; however, immunomodulation of the intracerebral inflammatory responses associated with TBM has not previously been described to our knowledge. In our study of 668 Vietnamese adults with TBM, active S. stercoralis infection was associated with reduced intracerebral inflammation and reduced neurological events by 3 months, compared with S. stercoralis-uninfected participants. This association was strongest in HIV-coinfected participants.

S. stercoralis Coinfection and Outcome from TBM
In TBM, intracerebral inflammation manifests as abnormal routine CSF parameters (elevated total WBC counts, neutrophil counts, and protein levels and reduced glucose levels) and elevated proinflammatory CSF cytokine concentrations [12,13,27]. In our study, active S. stercoralis infection was associated with significant reductions in absolute CSF neutrophil count and neutrophil proportion and nonsignificant reductions in CSF total WBC, CSF protein, and an increase in CSF/blood glucose ratio. The reduced inflammatory CSF profile in active S. stercoralis infection was consistent with the trend toward reduced grade 3 TBM disease in this group, compared with S. stercoralis-uninfected participants.
Significantly reduced "definite" TBM cases (which require microbiological confirmation of M. tuberculosis) and positive GeneXpert MTB/RIF assay results in active S. stercoralis infection suggest reduced mycobacterial burden in these participants. We speculate that these findings may reflect better host immunological control of TBM disease in the context of S. stercoralis infection.
The CSF cytokine analysis further supports a model of reduced intracerebral inflammation in TBM in active S. stercoralis coinfection. Pretreatment CSF cytokine concentration analysis showed significantly reduced concentrations of the proinflammatory cytokines IFN-γ, IL-2, and TNF-α, in active S. stercoralis coinfection. S. stercoralis coinfection in TBM was also associated with significantly reduced CSF IL-4 and IL-10, cytokines associated with a Th2 immune response. The suppression of these cytokines does not fit our prior hypothesis, indicating more work to understand the mechanisms of S. stercoralis immunomodulation is needed. Previous data in fact show IL-10 levels to be elevated in TBM, decreasing after antituberculosis chemotherapy [12,13]. Neutrophils highly express IL-4 and IL-10 in M. tuberculosis infection [28]; therefore, a reduction in CSF IL-4 and IL-10 concentrations in S. stercoralis-coinfected TBM be may mediated through reduced CSF neutrophils. Interestingly, CSF cytokine suppression was greater in HIV-coinfected than in HIV-uninfected participants. HIV coinfection is associated with globally increased CSF cytokines in TBM [19], but why helminth coinfection would control CSF cytokines more in the context of HIV coinfection is unknown and a topic for future research.
Our data showed a significant reduction in neurological complications within 3 months in active S. stercoralis infection, compared with S. stercoralis-uninfected participants. This finding is consistent with the associations observed between S. stercoralis coinfection, reduced bacterial burden, and reduced intracerebral inflammation. In our multivariate analysis, reduced neurological complications could not be explained by differences in age or HIV coinfection between groups. Elevated CSF neutrophil counts have been linked to neurological immune reconstitution inflammatory syndrome in TBM HIV coinfection and to death in HIV-negative TBM disease [16,29]. Given the known detrimental consequences of excessive intracerebral inflammation to TBM outcomes [19], it is plausible that reduction of neuroinflammation secondary to helminth down-regulation of proinflammatory TBM immune responses reduces neurological complications. Indeed, therapies in severe TBM often attempt to suppress excessive host immune responses.
This study has limitations. The rate of true S. stercoralis coinfection in our study population may be higher than reported, given that not all participants were assessed with serology, stool microscopy and stool PCR. This resulted in the creation of subpopulations for analysis. In addition, the performances of diagnostic tests for S. stercoralis are suboptimal. The sensitivity of stool microscopy is low (<30%) [30] owing to intermittent larval shedding. Stool PCR is more sensitive (approximately 65% sensitive) [31], yet some S. stercoralis coinfection will still be missed. S. stercoralis serological tests are affected by reduced sensitivity in advanced immunosuppression [32,33] or persistence of serological positivity despite parasite clearance [31].
In our subpopulation in which all participants underwent S. stercoralis serology, serological results were less likely to be positive in HIV coinfection, possibly reflecting false-negative results in this group. Follow-up in our study was limited to 3 months; longer-term impact on neurological complications or death therefore cannot be assessed. In addition, repeated CSF cytokine analysis, to assess immune responses after S. stercoralis eradication, was not performed. Finally, the study drug allocation (dexamethasone or placebo) of the trial participants remains unknown. This will not influence baseline phenotype or pretreatment CSF analyses; all of which represent data or sampling before study drug administration. Given the randomized study drug allocation (1:1), dexamethasone and placebo are expected to be evenly distributed within each individual analysis population.
The strengths of the current study are that it is large and prospective, with careful clinical characterization of TBM and S. stercoralis coinfection. It is part of 2 clinical trials with precise treatment protocols and standardized testing and data collection procedures. In this study of TBM, CSF is used for routine parameter and cytokine measurement, allowing a study of inflammation at the site of the disease instead of using blood inflammatory changes to assess intracerebral inflammation.
In conclusion, in our study active S. stercoralis coinfection in TBM was associated with reduced intracerebral inflammation and reduced neurological events. Further understanding of these immunomodulatory processes may aid the development of novel host-directed therapies to manage excessive and damaging inflammation of TBM.