Cerebrospinal Fluid Mononuclear Cell Phenotype and Activation Predictors of 2-week and 1-year survival among Persons with HIV-Associated Cryptococcal Meningitis

Abstract

Background:

Despite efforts to optimize therapy for HIV-associated cryptococcal meningitis (CM), survival outcomes remain poor. It is unclear how the cerebrospinal fluid (CSF) cellular immune phenotype and activation contribute to 2-week and 1-year survival following CM.

Methods:

We compared baseline CSF mononuclear cell phenotype and activation among adults with HIV-associated cryptococcal meningitis who died within 2-weeks of CM diagnosis to survivors who were alive at 1-year. The activated CSF T-lymphocytes, CD14⁺monocytes, and CD56⁺natural killer cells were determined from freshly collected CSF using Cytek Aurora Spectroflo cytometry. Quantitative CSF soluble cryptococcal antigen (CrAg) titer from frozen CSF at baseline and 1-year was determined using CrAg lateral flow assay. Data were analyzed using STATA v9.

Results:

Compared to survivors, participants who died within 2 weeks had significantly low absolute CSF CD8⁺T cells at baseline. For every 10% increase in PD-1 expression at baseline, the relative risk of 2-week mortality increased by 20–60%. CSF CD14⁺ monocytes among those who died demonstrated low HLA-DR⁺ and high CD163⁺ expression compared to survivors. We noted a significant reduction in the median CSF CrAg titer from 1:2560 at baseline to 1:5 at 1-year (p <0.0001) with 8/21 (38%) participants testing negative for CSF CrAg.

Conclusions:

Expression of CD163 on CD14⁺macrophages and immune exhaustion of CSF mononuclear cells at baseline are associated with an increased risk of early mortality in CM. After one year of treatment, approximately 4 in 10 patients with CM have a negative CSF CrAg.

Introduction

Cryptococcal meningitis (CM) caused by Cryptococcus spp. is a severe fungal disease of the central nervous system (CNS). Cryptococcus is classified by the World Health Organisation as a priority fungal pathogen [1]. Cryptococcal meningitis is considered an acquired immune deficiency syndrome (AIDS)-defining illness among people living with human immunodeficiency virus (HIV) [2,3]. In 2020, CM was estimated to cause 19% of AIDS-related deaths [4], with three quarters of these deaths occurring in sub-Saharan Africa [5]. In Uganda, CM accounts for 60% of adult meningitis cases[6], with a high unacceptable mortality [7] despite the rollout of new antifungal regimens [8]. The long term outcomes among patients who survive hospitalization are poor with only 50% reported to be alive at 1-year in clinical trial settings [9] and 25% after 3 years in routine settings [10].

Following infection with Cryptococcus spp., initially in the lungs of healthy individuals, the innate immune response leads to activation of the adaptive immune system which usually contains and/or clears the fungus from a host [11]. In advanced HIV disease, the fungus, through its virulence factors, evades and modifies the typical host-immune response resulting in dissemination [12,13] and establishment of the Cryptococcus spp.in the CNS compartment.

To date, the contribution of cerebrospinal fluid (CSF) cellular phenotype and activation to the high mortality in CM remains unclear. Previous studies [14,15] have examined the immune activation and CD4⁺ T-cell function in peripheral blood compartment rather than the CNS where the disease occurs. Moreover, the cytokine signature in the CNS compartment of patients with HIV-associated CM differs from that seen in peripheral blood [16]. This underscores the need to study the CSF cellular phenotype and activation at the site of disease to understand the contribution of the CNS host immune response to mortality in CM. This study aimed to compare the CSF mononuclear cell phenotype and activation in people living with HIV who died within two weeks of CM diagnosis to individuals who survived and were alive following one year of antifungal therapy. Additionally, we sought to describe changes in CSF characteristics, including CSF cryptococcal antigen (CrAg) titers, and the mononuclear cell profile at baseline and one year after initiation of CM treatment among survivors.

Methods

Study design and population

This was a case-control study, nested within a prospective cohort of adults with CM in the ‘Improving Diagnostics and Neurocognitive Outcomes in HIV/AIDS-related Meningitis’ (COAST) study at the Infectious Diseases Institute, Kampala, Uganda. Cases were adults who had died within two weeks following CM diagnosis (baseline). Controls were adults on antifungal therapy who were alive one year following CM diagnosis. Cases and controls were matched in a 1:1 ratio by sex. Participants without a known date of death or one-year outcome were excluded. At the time of the baseline CSF sample collection the participants had not received any antifungal therapy.

In the parent study, CM was diagnosed following a positive CSF cryptococcal antigen (CrAg) lateral flow test (Immy Inc., Norman, Oklahoma) among persons living with HIV. Participants in this study consented to have lumbar punctures performed at the time of meningitis diagnosis, on days 7, 14, or whenever they experienced symptoms of elevated intracranial pressure during CM treatment, and after one year of antifungal therapy. Corresponding participant’s age, clinical characteristics, coinfection (tuberculosis or pneumocystis jirovecii pneumonia) status, anti-retroviral therapy (ART) status, and CSF characteristics (opening pressure, protein, glucose, cryptococcus quantitative colony count) at baseline and 1 year were obtained from the Infectious Diseases Institute database using a unique identification number.

We utilized fresh CSF cell pellets which were processed within 2 hours of collection and stored CSF supernatant samples from patients diagnosed with CM and followed for one year.

Sample size

In this case-control nested study, a sample size of 36 participants, as shown in figure 2, was calculated using a two-tailed test with type I error of 5% and 80% power to detect differences between 18 cases who died within 2 weeks and 18 controls who had survived to 1 year (ratio of cases to controls 1:1). The reference population that was used to calculate the sample size for this case-control study was based on CSF CD8⁺HLA-DR⁺ expression of 88% among patients diagnosed with CM who survived to reach day 14 [17], and a hypothesized proportion of HLA-DR⁺ expression of 42% among patients who died within 2 weeks of CM diagnosis [14] (Supplementary 1).

Figure 2: Selection of study participants and CSF samples.

Forty-two participants (21 at baseline and 21 at 1 year) were used to determine the changes in CSF characteristics and mononuclear cell profile at baseline and at 1 year after initiation of CM treatment among survivors. We arrived at the sample size of 42 participants by assuming a two-tailed test with type I error of 5%, power of 80%, ratio of baseline to one year of 1:1 and effect size of 0.9 (Supplementary 2). We consecutively sampled stored flow cytometry standard files of processed CSF cell pellets and stored CSF supernatant of all eligible participants until the target sample size was attained using a storage data set.

Processing of CSF cell pellets for cell characterization

Following a lumbar puncture at baseline and at 1 year, CSF was centrifuged at 400xg for 5 minutes at 4°C. The CSF cell pellets were processed as summarized in supplementary 3 and stored as FCS files. The CSF supernatant was stored at −80°C.

The processed cell pellets were incubated in non-specific human Fc block prior to staining with commercial monoclonal antibodies i.e., CD16^V459 (clone 3G8, BD Horizon), CD4^BV605 (clone OKT4, BioLegend), PD-1^{BV 750} (Clone 2D3, BioLegend), CD56^BV650 (clone 5.1H11, BioLegend), CD3^AF488 (clone UCHTI, BioLegend), HLA-DR^PE-CF594 (clone G46–6, BD Horizon), CD45^PerCP-Cy5.5 (clone 2D1, BioLegend), CD163^{PerCp-eFluor 710} (clone TNKUPJ, Thermoscientific), CD14^APC (clone M5E2, BioLegend), CD8^APC-Cy7 (clone SK1, BD Horizon) and CD57^BUV711 (clone QA17A04, BioLegend). Fluorescence minus one were used as controls to set gates. Data acquired on a Cytek flow cytometer was analyzed using FlowJo^™ V10.8 software (BD Life Sciences, Ashland, USA). The flow cytometry gating strategy is shown in Figure 1.

Figure 1:

Panel A: Flowcytometry analysis of thawed CSF samples.

Panel B: Comparative bar graphs between those who died in 2 weeks and survivors.

Panel C: Comparative bar graphs between baseline and 1-year of CSF samples among the CM survivors.

Processing of CSF supernatants for CrAg titers

CSF supernatant had been stored for a median of 6 months (IQR= 3–15). For CrAg titer determination, CSF supernatant retrieved from storage was thawed at room temperature for 1 hour. Serial dilutions of 1:5, the first dilution corresponding to a low titer, through 1:2560, the 10^th dilution corresponding to a very high titer, were performed per the manufacturer’s instructions (LFA; Immuno-Mycologics Inc., Norman, OK).

Statistical analysis

Continuous variables were expressed as median with interquartile range, and categorical variables were summarized as proportions. Data from the polychromatic flow cytometry were presented as side scatter plots with differential gating based on size and granularity. The CSF CD4⁺ T and CD8⁺ T cell lymphocytes, CD14⁺ monocytes and CD56⁺ natural killer cell phenotypes and activation were compared using Mann Whitney U test and measures of association were measured using binary logistic regression and modified Poisson logistic regression (Supplementaries 4 and 5). The changes in the CSF characteristics, including the CSF CrAg titer, and the CSF CD4⁺ T and CD8⁺ T cell lymphocytes, CD14⁺ monocytes and CD56⁺ natural killer cell phenotypes and immune activation from baseline to 1 year after initiation of CM treatment were analyzed using Wilcoxon paired signed rank sum test.

Ethical consideration

The study was approved by the Uganda National Council of Science & Technology and the School of Medicine Research and Ethics Committee (SOMREC 2024–873). The primary study was approved by the Mulago Hospital Research and Ethics committee (MHREC 1246).

Results

Selection of study participants

From January 2019 to April 2024, 1109 participants diagnosed with HIV-associated CM had been enrolled in the parent study of whom 46 had an annual visit recorded. Paired CSF samples i.e., cell pellet FCS files and stored supernatant were available for 909 and 28 participants at baseline and 1 year respectively. Of the 28 participants with 1 year FCS files and stored CSF supernatant samples, 24 also had a corresponding baseline CSF cell pellet FCS files and stored supernatant available. It is from these 24 participants that a pool of 21 participants (survivors) was created after consecutive sampling. From this pool of 21 participants, 18 participants were consecutively sampled to serve as controls in the case-control study. Of the 909 participants who had baseline CSF cell pellet FCS files and supernatant in storage, 278 were identified to have died from CM within 2 weeks of being diagnosed from whom 18 participants (cases) were consecutively sampled (Figure 2).

Participants’ characteristics

The median age of the participants was 35 years (IQR= 30 – 45), 23/39 (59.0%) were male. In the case-control, 20/36 (55.6%) participants were male, with a comparable median weight between those who died and survivors. Among those who died, 7/18 (38.9%) had tuberculosis/pneumocystis jirovecii pneumonia co-infections compared to 3/18 (16.7%) among survivors. Similarly, 7/18 (38.9%) of those who died were on ART at baseline compared to 4/18 (22.2%) survivors, although this difference was not statistically significant (p=0.47). The rest of the characteristics were comparable between the two groups as summarized in table 1.

Table 1: Baseline characteristics of study participants.

Variable	Died n=18	Survivors n=18	P-valuew
Male, n (%)	10 (55.6)	10 (55.6)
Weight (kg), median (IQR)	58 (50–60)	58 (51– 66)	>0.9
On ART, n (%)	7 (38.9)	4 (22.2)	0.47
Duration of CNS symptoms before CM diagnosis (days), n (IQR)	11 (7–16)	14 (13–29)	0.12
Peripheral CD4 T-cell count (cells/μL), median (IQR)	25 (11–34)	29 (14–49)	0.37
Coinfections: TB or PJP, n (%)	7 (38.9)	3 (16.7)	0.26
CSF opening pressure (mmHg), median (IQR)	230 (140–239)	255 (170–355)	0.32
CRP (mg/dL), median (IQR)	55 (14–133)	25 (5.1–81)	0.24
CSF glucose (mg/dL), median (IQR)	64 (36–82)	51 (41–67)	0.18
CSF lactate (mmol/L), median (IQR)	4.4 (2.7–7)	3.2 (2.5–3.7)	0.05
CSF protein (mg/dL), median (IQR)	80 (40–163)	80 (60–130)	0.88
Log10 QCC (CFU/ml), median (IQR)	5.92 (5.52– 6.12)	5.9 (4.56– 6.45)	0.86

CSF-Cerebrospinal fluid, QCC-quantitative cryptococcal culture, CFU-colony forming units, TB-Tuberculosis, PJP-pneumocystis jirovecii pneumonia, ART-antiretroviral therapy, CRP-C reactive protein, CM-cryptococcal meningitis, w-Mann Whitney U test.

Comparison of immune phenotype and activation in those who died within 2 weeks versus survivors

Overall, participants who died had lower median CD45⁺ cells of 1699 cells/ml (IQR= 221–3131) in CSF compared to the survivors, 4132 cells/ml (IQR= 1425–10330). However, this difference was not statistically significant (p=0.12).

Lymphocytes

Participants who died had significantly lower median CD8⁺ T cell counts compared with survivors (405 cells/ml (IQR=69–1173) versus 2691cells/ml (IQR= 782–7182), p=0.03). The median proportion of PD-1⁺ expression on both CD8⁺ and CD4⁺ T cells was significantly higher among those who died compared to survivors (CD8⁺PD-1⁺: 69% (IQR= 49–77) versus 16% (IQR= 14–33), p <0.0001, adjusted relative risk (aRR) 1.05 (95% CI 1.03–1.06) p<0.001; CD4⁺PD-1⁺: 56% (IQR= 53–67) versus 19% (IQR= 16–26), p<0.0001, aRR 1.04 (95% CI 1.02–1.06) p<0.001). There were no significant differences in the CD4⁺ T cell absolute count and HLA-DR⁺ expression on CD4⁺ and CD8⁺ T cells. This data is summarized in Table 2.

Table 2: Immune marker expression on CSF mononuclear cells among survivors and participants who died within 2 weeks.

Variable	Died n=18	Survivors n=18	P valueW
T Lymphocytes
CD8+ cell count/ml	405 (69–1773)	2691 (787–7182)	0.03*
CD8+ HLA-DR+ %	69 (60–80)	79 (69–84)	0.15
CD8+ PD-1+ %	69 (49–77)	16 (14–33)	<0.0001*
CD4+ cell count/ml	28(2– 267)	125 (36–258)	0.19
CD4 HLA-DR+ %	67 (37–79)	64 (56–72)	0.81
CD4 PD-1+ %	56 (53–67)	19 (16–26)	<0.0001*
CD4+: CD8+ ratio	0.05 (0.03–0.12)	0.04(0.02–0.15)	0.87
CD14+ Monocytes
CD14+ monocyte count/ml	6 (1–15)	27 (3–114)	0.14
Classical (CD14++CD16−) %	48 (38–75)	52 (37–81)	0.91
Intermediate (CD14++CD16+) monocyte %	8.1 (0.0–25)	0 (0–43)	0.85
Non classical (CD14+CD16++) monocyte %	0.0 (0–1.5)	0 (0–0.0)	0.05
All CD14+ monocytes HLA-DR+ %	21 (16–26)	77 (75–80)	<0.0001*
All alternatively activated (CD163) CD14+ monocytes %	66 (47–76)	25 (19–32)	0.0001*
All CD14+ monocytes PD-1+%	71 (64–78)	21(7.2–30)	<0.0001*
CD56+ NK cells
CD3−CD56+ cell count/ml	3 (1–11)	6 (3–12)	0.19
CD3−CD56+ bright %	16 (7.8–50)	37 (46–27)	0.12
CD3−CD56+ dim %	25 (3.4–49)	16 (13–25)	0.32
CD3−CD56neg %	20 (10–29)	9.6 (6.1–17)	0.04*
All NK, HLA-DR+ %	34 (25–41)	59 (52–66)	<0.0001*
All NK, PD1+ %	61 (50–82)	23 (16–32)	<0.0001*
All NK, 57+ %	65 (56–69)	16 (11–25)	<0.0001*

Values are median (IQR). w-Mann Whitney U test, IQR-Inter quartile range

^*P<0.05.

CD14⁺ Monocytes

Participants who died had lower monocyte counts compared to survivors though this difference was not statistically significant (6 cells/ml (IQR=1–15) versus 27 cells/ml (IQR=3–114), p=0.14). As shown in table 2, monocytes from participants who died demonstrated a shift towards an alternatively activated state (CD14⁺CD163⁺) and a had a high expression of PD-1 (CD14⁺PD-1⁺) compared with survivors (CD14⁺CD163⁺: 66% (IQR=47–76) versus 25% (IQR=19–32), p=0.0001, aRR 1.03 (1.02–1.05) p<0.001; CD14⁺PD-1⁺: 71% (IQR=64–78) versus 21% (IQR=7.2–30), p<0.0001, aRR 1.06 (95% CI 1.04-.1.07) p<0.001). There were no significant differences in the classical (CD14⁺⁺CD16⁻), intermediate (CD14⁺⁺CD16⁺) and non-classical (CD14⁺CD16⁺⁺) phenotypes of CD14⁺ monocytes between those who died and those who survived.

CD56⁺ NK cells

Participants who died had a lower absolute CD3⁻ NK cell count compared to survivors albeit not being significant (3 cells/ml (IQR=1–11) versus 6 cells/ml (IQR=3–12), p=0.19). As shown in table 2, those patients who died had higher proportions of CD3⁻CD56^neg NK cell phenotype compared to survivors (20% (IQR=10–29) versus 9.6% (IQR=6.1–17), p=0.04) respectively. Significantly, participants who died had a low absolute CD3⁺NK cells as well as a lower proportion of CD3⁺CD56^dim NK cells compared to survivors (CD3⁺NK cells:1 cells/ml (0–2) versus 4 cells/ml (2–8), p=0.001; CD3⁺CD56^dim NK cells: 15.8% (11.4–36.8) versus 50% (38.3–58.4) p<0.001 (supplementary 9). Likewise, those who died had lower HLA-DR⁺ expression, higher PD-1⁺ expression and a greater proportion of CD57⁺ NK cell phenotype compared to controls. This data is summarized in table 2.

Comparison of baseline versus 1 year for survivors.

Table 3 summaries the differences in CSF parameters and CSF immune phenotype and activation of innate cells at baseline and 1 year among the CM survivors. At 1 year after CM treatment, survivors had lower CSF opening pressure, CSF protein, CSF lactate and a higher CSF glucose compared to their baseline (at diagnosis). The CrAg titer at 1-year was significantly lower (Baseline versus 1 year, 1:2560 (IQR=1:960–1:10,240) versus 1:5 (negative-1:2560) at 1 year, p<0.0001, with 38.1% converting to CSF CrAg negative at 1-year of follow up. The CSF CrAg titer dilutions had significantly reduced from a median of 1:10 at baseline to median of 1:1 at 1 year (p<0.0001) (Supplementaries 6 and 8 ).

Table 3: Comparison of CSF parameters and immune marker expression at Baseline versus One year among survivors.

Variable	Baseline (at diagnosis)	One year	P valuer
CSF Opening pressure, mmH20.	270 (185–330)	110(75–135)	<0.0001*
CSF protein, mg/dL	80 (60–143)	20 (20–35)	<0.0001*
CSF lactate, mmol/L	3.4 (2.6–4.0)	2.2 (1.9–2.4)	0.0044*
CSF glucose, mg/dL	50 (41–67)	75 (68–81)	<0.0001*
CSF CrAg titer	1: 2560 (1: 960–1:10240)	1:5 (negative-1:25)	<0.0001*
CSF titer dilutions	10 (8.5–12)	1.0 (negative-3.0)	<0.0001*
CSF titer decline		2550 (950–10240)
CSF dilution decline		8.0 (5.0–9.0)
T Lymphocytes
CD8+ count/ml	3571 (969–9452)	642 (301– 4087)	0.04*
CD8+ HLA-DR+ %	79 (68–84)	50 (36–61)	<0.0001*
CD8 PD1+ %	16 (14–35)	20 (16–23)	0.59
CD4+ count/ml	138 (38–552)	299 (106–1572)	0.27
On ART	125 (45–309)	640 (319–3679)	0.12
Not on ART	145(38–1062)	129 (92–1539)	0.78
CD4+ HLA-DR+ %	64 (53–70)	17 (8.5–26)	<0.0001*
CD4+ PD-1+ %	20 (16–26)	31 (18–36)	0.06
CD4+:CD8+ratio	0.05(0.03–0.01)	0.37(0.27–0.66)	<0.0001*
CD14+ monocytes
CD14+monocyte count/ml	21 (4–123)	2 (1–17)	0.01*
Classical (CD14++CD16−) %	54 (35–75)	96 (72–100)	<0.0001*
Intermediate (CD14++CD16+) %	0.0 (0.0–44)	4.2 (0.0–26)	0.24
Non classical (CD14+CD16++) %	0.0 (0.0–0.0)	00 (0.0–00)	0.25
All CD14+ monocytes HLA-DR+ %	76 (74–80)	71 (65–79)	0.07
All CD14+ monocytes alternatively activated (CD163+) %	25 (17–30)	14 (8.0–24)	0.04*
All CD14+ monocytes PD1+ %	21 (10–30)	15 (11–26)	0.05
CD56+ NK cells
CD3−CD56+ cell count/ml	7(3–15)	2 (1–9)	0.16
CD3−CD56+ bright % (IQR)	37 (30–48)	39 (22–50)	>0.9
CD3−CD56+ dim % (IQR)	17 (14–29)	11 (0–25)	0.43
CD3−CD56 neg % (IQR)	8.8 (3.7–15)	4.5 (0–12)	0.05
All NK HLA-DR+ % (IQR)	59 (51–64)	74 (53–88)	0.15
All NK, PD1+ % (IQR)	27 (17–35)	27 (20–39)	0.81
All NK 57+ % (IQR)	17 (10–224)	25 (9.1–34)	0.27

Values are medians (IQR). IQR-Interquartile range, CSF-cerebrospinal fluid

^*P<0.05, r-Wilcoxon matched-pairs signed rank test

Using flow cytometry, the median CD45⁺ cell count at 1 year was 1382 cells/ml (IQR=575–7898) compared to the baseline of 592 cells/ml (IQR=1841–12187), p=0.09. The CD8⁺ lymphocyte count at 1 year was significantly lower than at diagnosis. The proportion of CD8⁺HLA-DR⁺ and CD4⁺HLA-DR⁺ T cells had significantly reduced at 1 year compared to baseline (CD8⁺HLA-DR⁺: 79% (IQR=68–84) versus 50% (IQR=36–16), P<0.0001; CD4⁺HLA-DR⁺: 64% (IQR=53–70) versus 17% (IQR=8.5–26), p<0.0001). There were no significant differences in the proportion of CD8⁺PD-1⁺ and CD4⁺PD-1⁺ T cells at baseline and one year after CM treatment (Supplementary 7).

The absolute monocyte counts after 1 year were significantly lower compared with the baseline (2 cells/ml (1–17) versus 21 cells/ml (4–123), p=0.01). At 1 year, CD163⁺ expression on the monocytes significantly reduced (25% (17–30) versus 14% (8–24), p=0.04).

The absolute count of NK cells at 1-year was lower compared with baseline though this difference was not statistically significant (CD3⁻CD56⁺NK cells: 2 cells/ml (1–9) versus 7 cells/ml (3–15), p=0.16; CD3⁺CD56⁺NK cells: 5 cells/ml (2–9) versus 1 cell/ml (0–2), p=0.05 (supplementary 10). The expression of HLA-DR⁺, PD-1⁺ and CD57⁺ on NK cells was not significantly different between baseline and one year after treatment of CM.

Discussion

Our study demonstrated that at CM diagnosis, participants who died within 2 weeks had significantly lower absolute CSF CD8⁺ T cells but increased proportions of CD4⁺PD-1⁺, CD8⁺PD-1⁺ T cells, CD14⁺PD-1⁺ monocytes and CD56⁺PD-1⁺ NK cells compared to survivors. There were lower proportions of CSF CD14⁺HLA-DR⁺ monocytes among those who died within 2 weeks and a higher proportion of alternatively activated CD⁺14 monocytes (CD14⁺CD163⁺) compared to survivors.

The association between absence of CSF pleocytosis in the presence of CNS infection and mortality has been reported previously in patients with CM [18]. Scriven and colleagues found that paucity of CSF T cellular infiltrate rather than alternative macrophage activation using CD206⁺ expression on CD14⁺ monocytes correlated with CM severe disease [19]. Having few CD8⁺ T cells at the time of Cryptococcus neoformans infection is problematic in itself and could partly contribute to the high mortality seen in patients with CM. Murine studies by Lindell et al demonstrated that even in the absence of CD4⁺ T cells, CD8⁺ T cells retained their capacity to traffic to the site of cryptococcal neoformans infection, undergo clonal expansion as well as produce interferon-gamma [20]. These CD4⁺ T cell independent effector functions of CD8⁺ T cells imply that the main source of cell mediated immune response in patients with HIV associated CM remains the intact CD8⁺ T cells.

In this study we noted that the CD8⁺ T cells among those who died in 2 weeks were exhausted as evidenced by higher PD-1 expression compared with those who survived leaving room for speculation that possibly these cells had poor quality effector immune responses against the invading fungus.

We hypothesize that the low expression of HLA-DR on CD14⁺ monocytes implies that these cells are unable to effectively present the processed Cryptococcus neoformans antigen to the cells responsible for the adaptive immune response and produce vital cytokines such as IL-15 [21]. Ma et al showed that IL-15 is required and sufficient to upregulate granulysin production and release from cytotoxic CD8⁺ T cells [11].

A higher expression of PD-1 on all the CSF mononuclear cells studied i.e. CD4⁺ T lymphocytes, CD8⁺ T lymphocytes, all CD14⁺ monocytes and all CD56⁺ NK cells among those who died within 2 weeks compared with those who survived to 1 year is somewhat concerning. PD-1 is commonly upregulated by IL-4 and IL-10 [22,23]. PD-1 through its ligands i.e. PD-L1, PD-L2 produces a state of immune hypo-responsiveness by inhibiting immune cell replication, production of crucial Th1 proinflammatory cytokines (IL-2, INF-γ and TNF-α) and reducing cell survival of the CSF mononuclear cells [22]. The proinflammatory Th1 cytokines enhance the cell mediated immune responses and murine studies have shown that they offer protection against fatal Cryptococcus neoformans infection [24]. This immune hypo-responsiveness following increased expression of PD-1 on the CSF mononuclear cells may predispose patients with HIV-associated CM to poor outcomes.

A high proportion of CSF CD56^neg NK cells among participants who died within two weeks of CM diagnosis compared with those who survived was an intriguing finding. This subtype of NK cells commonly seen in ART naïve patients with HIV is atypical and is associated with decreased quantities of granzyme B and perforin [25] through which NK cells exert their cytotoxic function. Worse still, unlike the CD56^bright and C56^dim NK cells, the cytotoxic activities of these CD56^neg NK cells are highly dependent on antibody stimulation via B cell lymphocytes [26], which is usually poor in these patients with chronic HIV infection [27].

To our knowledge, our reported CSF CrAg conversion after a year of antifungal therapy is novel. Prior studies observed CSF CrAg titer decline in day 7 and day 14 [28,29] but no prior reports in the literature have examined CrAg titer levels beyond these time points. Also, as expected, the CSF opening pressure, protein and lactate, cryptococcal antigen titer significantly reduce, and CSF glucose increases gradually over 1 year period among CM survivors. The observed increase in CSF glucose at 1 year can be attributed to the clearance of fungus from the CNS compartment following successful antifungal therapy. Glucose is the preferred carbon source for Cryptococcus ssp and is vital for both fungal development and virulence [30].

In addition to the known risk factors of 2-week mortality from HIV associated cryptococcal meningitis, this study demonstrated that immunological factors contribute to the 2-week mortality. Immune adjunctive therapy in cryptococcal meningitis has not been exhaustively evaluated to reduce mortality. Jarvis and colleagues showed that adjunctive interferon-gamma was safe and improved CSF fungal clearance in a phase 2 study [31]. Further studies are needed to evaluate the role of adjunctive anti-PD1 therapy in patients with cryptococcal meningitis to reduce mortality and increase survival.

Conclusion

Unfavorable CSF immunological factors at CM diagnosis such as low CD8⁺ T lymphocytes, abundance of alternatively activated monocytes, terminally differentiated NK cells and immune exhaustion of CD4⁺ T and CD8⁺ T lymphocytes, CD14⁺ monocytes and CD56⁺ NK cells were associated with 2 week mortality in CM. Approximately 60% of patients diagnosed with HIV associated CM will still have a positive CSF cryptococcal antigen test after 1 year which may present a clinical diagnostic conundrum. Understanding host immunological factors and mechanisms acting directly or indirectly with cryptococcal antigens may contribute to a better understanding of 2-week mortality risk and provide context for future modes of immune therapy to improve survival following HIV associated cryptococcal meningitis.

Strengths

In this study we interrogated the differences in the CSF cellular phenotype and activation among patients with CM who died within 2 weeks and those who survived to 1 year using samples collected from the site of infection. Secondly, we used fresh CSF cell pellets. We studied the markers of immune activation and cellular phenotype using flow cytometry, a method that is easily reproducible.

Limitations

We were unable to perform complimentary CSF cytokine studies, intracellular staining of the mononuclear cells and CSF HIV viral loads to correlate with our findings among those who died and the survivors. Potential confounders of our observations include the higher proportion of opportunistic coinfections i.e., TB and PJP, and ART exposure at baseline in those who died affecting immunological and survival outcomes. There was a selection bias for the participants in the prospective cohort study following the consecutive sampling. The selection of the survivors for the nested case-control study was susceptible to a survival bias.

Availability of data and materials

The authors confirm that all data underlying the findings are fully available without restriction in the manuscript and supplementary material.

Clinical summary.

At cryptococcal meningitis diagnosis, for every 10% increase in PD-1 expression on CSF mononuclear cells, the relative risk of 2-week mortality increased by 20–60%. Despite 1 year of antifungal treatment, 62% of survivors had persistent CSF cryptococcal antigenemia.