Comparison of clinical and biological characteristics of HIV-infected patients presenting Cryptococcus neoformans versus C. curvatus/C. laurentii meningitis

Background Cryptococcal meningitis is mainly caused by Cryptococcus neoformans/C. gattii complex. We compared the clinical, biological, and antifungal susceptibility profiles of isolates from HIV-Infected Patients (HIVIP) with C. neoformans (Cn) versus C. curvatus/C. laurentii (Cc/Cl) meningitis. Methods Comparative analytical study were conducted. Apart from patients’ clinical data, the following analysis were performed and the results were compared in both groups: biochemical examination, cryptococcal antigen test, India ink staining, and culture on Cerebral Spinal Fluid (CSF), strains identification by mass spectrometry, ITS sequencing, PCR serotyping and antifungal susceptibility. The main outcome variable was the “species of Cryptococcus identified”, which was compared to other variables of the same type using the Pearson Chi-square test or the Fisher exact test. Results A total of 23 (79.3%) Cn meningitis cases versus 6 (20.7%) Cc/Cl meningitis were retained. Cn meningitis was more frequently associated with headache (100% vs 50%, p = 0.005) than Cc/Cl meningitis and meningeal signs were more frequent in Cn infected patients. Biologically, hypoglycorrhachia and low CD4 count were more observed in Cn group (90% vs 20% of patients, p = 0.01; 45.6 vs 129.8 cells/µL, p = 0.02, respectively). A higher proportion of Cn strains (91.3%) showed a low Minimum Inhibitory Concentration (MIC) (< 8 mg/L) for fluconazole compared to Cc/Cl strains (66.7%). Also, Cc/Cl strains resistant to 5-flucytosine and amphotericin B were found in 16.7% of cases for each of the two antifungal agents. Cryptococcus detection by routine analysis (India ink, culture, and antigens) was better for Cn samples than Cc/Cl. Except ITS sequencing, which identified all strains of both groups, mass spectrometry and serotyping PCR identified Cn strains better than Cc/Cl (100% vs 80%, p = 0.1; 100% vs 0%, p < 0.0001, respectively). After treatment with amphotericin B, 5-flucytosine, and fluconazole in both groups, the outcome was similar. Conclusions Clinical presentation of Cn meningitis is certainly more severe than that of Cc/Cl meningitis, but Cc/Cl infection should be considered in the management of HIVIP with meningeal syndrome because of the diagnostic difficulty and the high MICs of antifungal agents required for the treatment of meningitis due to these cryptococcal species.

Zono et al. BMC Infectious Diseases (2021) 21:1157 Background In clinical pathology, two species of Cryptococcus spp. are mainly involved in Meningeal Cryptococcosis (MC), namely Cryptococcus neoformans (Cn) and C. gattii (Cg) [1]. The interest of Cryptococcus spp. identification down to species level is based on the fact that some species such as C. gattii cause infections that require a much more intensive therapeutic approach than those caused by C. neoformans [2]. Non-neoformans and non-gattii species have long been considered as saprophytes and non-pathogenic to humans. However, the prevalence of opportunistic infections due to these species (involving the skin, lungs, bloodstream, and central nervous system) has been increasing all over the world in recent years. Among these species, C. laurentii and C. albidus are implicated in 80% of cases [3,4]. Also in rare cases, C. curvatus has been incriminated in peritoneal and myeloradicular infections in hospital settings [5][6][7]. Few studies are comparing different characteristics of infections caused by the C. neoformans/C. gattii complex versus those due to non-neoformans and non-gattii species in HIV-Infected Patients (HIVIP).
We hypothesized that the clinical and biological characteristics of HIVIP with meningitis due to Cn could be different from those of patients infected by C. curvatus/C. laurentii (Cc/Cl). The objective of this study was to compare the clinical, biological, and therapeutic characteristics, antifungal susceptibility profile of strains isolated from HIVIP with Cn versus Cc/Cl meningitis. In addition, the molecular identification of the strains was also compared.

Study design, patients, and samples
This is a comparative analytical study. The patients were drawn from a cross-sectional study conducted in Kinshasa (Democratic Republic of Congo) from 1st February 2019 to 29th February 2020, having included HIVIP with the meningeal syndrome in the hospitals with expertise in the management of advanced HIV-infection, namely: Centre Hospitalier Mère et Enfant de NGABA (CHME NGABA), Centre Médical et Evangélique Révérend LUYINDU (CME LUYINDU) and Centre Hospitalier Roi Baudouin 1 er (CHRB 1 er ). The overall data from this study have not yet been published. In these public hospitals supported by Doctors without Borders-Belgium (MSF), lumbar puncture in HIVIP has specific indications. It was performed in each patient presenting meningeal signs or neurological symptoms including headache, epilepsy, lethargy, or cognitive deficits and/or in all HIVIP with a CD 4 LT number < 100 cells/µL with cryptococcal antigenemia. Diagnostic confirmation of cryptococcosis is based on the blood cryptococcal antigen detection for serum cryptococcosis or the Cerebral Spinal Fluid (CSF) cryptococcal antigen detection for MC. In the present study, routine diagnostic confirmation was extended. Thus, MC diagnosis was based on the detection in the patient CSF, of cryptococcal antigens and/or the yeasts by India ink staining and/or positive culture.

Initial analysis
The detection of Cryptococcal Antigens (CrAg) was carried out in the CSF of each patient, using the CrAg LFA IMMY test (Immuno-mycologic, Norman, OK, USA). Direct staining with India ink on the CSF was also carried out to reveal the cryptococcal capsule and the CSF was cultured on Sabouraud Dextrose Agar-Chloramphenicol medium (SDA-C, bioMérieux, France) at 30 °C for 48 to 72 h. Proteinorachia was determined by the Pandy test, a previously described qualitative test [8].

Identification by MALDI-TOF MS
The MALDI-TOF MS (Matrix-Assisted Laser Desorption Ionization Time-Of-Flight Mass Spectrometry, Bruker Daltonics GmbH, Germany), was used for the identification of all fungal strains at CHU de Liège (Belgium). From the culture on SDA-C, an extended direct deposit was performed by adding 1 μL of 70% formic acid to the sample on a MALDI target plate (MSP 96 BC ground steel target; Bruker Daltonics). Then, 1 μL of saturated cyano-4-hydroxycinnamic acid solution (HCCA matrix; Bruker Daltonics) was added. A Bruker Bacterial Test Standard (BTS255343; Bruker Daltonics) was used for instrument calibration. Each microorganism tested was spotted twice on the same target slide. Measurement was performed with MALDI Flex control V3.4 (Bruker Daltonics) following the settings suggested by the manufacturer using automated collecting spectra. The spectra of each duplicated spot were compared with those in the reference library (BD 8326 or version V 9.0) [9]. The following score was considered for the identification of the fungal species: MS Score ≥ 1.5 and the three first results difficulty and the high MICs of antifungal agents required for the treatment of meningitis due to these cryptococcal species.
Keywords: Cryptococcus neoformans, Cryptococcus curvatus, Cryptococcus laurentii, Meningitis, HIV, Clinical characteristic, Biological characterization, Kinshasa, DRC identical and consistent with the appearance of the colonies on agar.

Molecular analysis DNA extraction
Genomic DNA was extracted from the fresh 24-h cultures on SDA-C using the NucleoSpin blood quick pure kit (Macherey-Nagel, Duren, Germany). Two preliminary steps were added to the manufacturer's protocol, namely bead-beating, and thermal shock. In a 2 mL tube containing 0.5 mm glass beads (Roche Diagnostics GmbH, Penzberg, Germany), colonies were mixed with 350 µL lysis buffer (Promega Corporation, USA). The mixture was vortexed five times at 6000 vibrations per minute for 40 s (bead-beating). Between each pass, the tube was cooled between − 20 and 1 °C for 30 s in a Nalgene microtube cooler container (Dutscher, France) (thermal shock).

ITS sequencing
The rRNA ITS2 region was amplified using the ITS86 forward primer 5′GTG AAT CAT CGA ATC TTT GAA 3′ and ITS4 reverse primer 5′TCC TCC GCT TAT TGA TAT GC 3′ [10]. PCR was done on a classical thermocycler (Thermo Hybaid, Thermo Scientific). Purification of PCR products was then performed using the Exosap IT technique (Amersham, GE Healthcare Europe GmbH, Belgium). Bidirectional sequence data were generated after purification using the BigDye terminator sequencing kit (Applied Biosystems, Life Technologies, Belgium). Sequenced products were finally purified using the kit clean Seq Agencourt (Beckman Coulter Life Science). The sequencing was done on the automate ABI 3500/3500XL (Applied Biosystem, Life Technologies). Sequences were edited using the ABI Sequence Scanner V.1.0 software (Applied Biosystems, Life Technologies). Sequences generated by the software were then compared to the CBS database by using BIOLOMICS-NET software (http:// www. cbs. knaw. nl/ colle ctions/ Biolo MICSS equen ces. aspx), which comprises several databases including Genbank. Only results that repeated the same identification at least three times and had a similarity score greater than 95% were considered valid.

Serotyping PCR
A classical serotyping PCR, which has been designed for Cn/Cg species complex, was performed according to the protocol described by Ito-Kuwa et al. [11] Two primer pairs of the LAC1 gene and one primer pair of the CAP64 gene were used.

Antifungal susceptibility testing
Determination of Minimal Inhibitory Concentration (MIC) was done according to the EUCAST E.Def 7.3.1 procedure [12]. Inoculum suspensions of 0.5 McFarland standard were prepared and diluted 1:10 with sterile distilled water (Sensititre tm demineralized water, Thermo Scientific, USA). The final concentration range was 0.008-8 mg/L for amphotericin B (after inoculation) and 0.06-64 mg/L for 5-flucytosine and fluconazole. The reading of the MIC50 value (drug concentration resulting in inhibition of 50% of microorganisms) for 5-flucytosine and fluconazole, and MIC90 for amphotericin B, was done according to the described recommendations using a visual and automated reading at 405 nm with a Multiscan FC spectrophotometer (Thermo Scientific, MA, USA). Candida parapsilosis ATCC 22019 and Candida krusei ATCC 6258 were used as the quality control strains for the tests.  [13,14].

Statistical analysis
The analysis were carried out using R-Cmdr version 2.6-1 (R Foundation for Statistical Computing, Vienna, Austria). Missing data were considered completely random and the available data were analyzed. The continuous variables were summarised as mean ± standard deviation and compared using Student's t-test. The proportions and their respective 95% confidence intervals were calculated for the categorical data. The main outcome variable was the "species of Cryptococcus identified", which was compared to other variables of the same type using the Pearson Chi-square test or the Fisher exact test if the expected values were less than 5. All tests were two-tailed and a p < 0.05 was considered statistically significant.

Study population and clinical data
Of the 29 patients included for cryptococcal meningitis with positive culture on CSF, six (20.7%; 95% CI 6.9-34.5%) were infected by non-neoformans and non-gattii Cryptococcus species. In five cases, it was C. curvatus, and in one case C. laurentii.
The demographic and clinical characteristics of patients infected by each group species are presented in Table 1

HIV biological data of the patients
The HIV biological analysis results of the patients are presented in Table 2. The mean CD 4 cell count was 45.6 cells/µL in the Cn infected patients, which was significantly lower than in the Cc/Cl group (129.8 cells/µL, p = 0.02). While 50% of Cn meningitis patients had a moderately high viral load (10 3 -10 5 copies/mL), the only viral load available for the patients infected with Cc/Cl was very high (> 10 5 copies/mL).

CSF analysis characteristics CSF analysis characteristics patients
All patients infected with Cc/Cl had clear CSF versus 13% of cloudy appearance in Cn group patients. In both groups, the majority of patients had a very high opening pressure during a lumbar puncture (75% vs 66.6%, Cn and Cc/Cl respectively). Approximately 90% of patients with Cn meningitis had significantly low glycorrachia (< 50 mg/dL) compared to 20% of Cc/Cl infected patients (p = 0.01). Accordingly, the mean value of glycorrachia was lower in the Cn infected group (46.7 ± 11.2 mg/dL) than in the Cc/Cl group patients (66.1 ± 22.5 mg/dL, p = 0.04). Sixty per cent of patients infected with Cc/Cl had a positive qualitative proteinorachia (Pandy test), while 56.2% of patients infected with Cn had a negative test (p = 0.6). No significant difference was noted between the mean cytorrachia in the Cn patients group versus those infected with Cc/Cl. Nearly 33.3% of Cn infected patients had a predominantly lymphocytic   Table 3.

Species identification
All strains (100%) of Cn were identified by ITS sequencing and then by mass spectrometry. In comparison, 20% of the Cc/Cl strains were not identified by mass spectrometry although they were all identified by ITS sequencing. Between the same species, slight differences in ITS sequences were noted among both Cn and Cc isolates. Besides Cl isolate which was unidentified by MALDI-TOF MS, one Cc isolate was also unidentified, and the same isolate had a slightly different post multiplex PCR electrophoretic profile than others of the same species, namely the absence of the 1100 pb band (lane 5). The difference in MALDI-TOF MS identification between the two groups was not significant; although the spectral profiles were very different. By serotyping multiplex PCR, the Cc/Cl strains were not identifiable, whereas the Cn strains were all identified as serotype A. The Cc/Cl strains had an electrophoretic profile that was not comparable to the reference profile of the Cn species commonly serotyped by this technique (Fig. 2). Apart from this, four out of five Cc isolates had a similar electrophoretic profile and one was slightly different from the others. The difference in serotype identification was statistically significant in both groups (p < 0.0001). The comparison of identification results is presented in Table 4.  of the three antifungal agents tested was not statistically different in the two groups (Fig. 3). Table 5 shows the outcome of Cryptococcus-infected patients after treatment by antifungal agents. Treatment with liposomal amphotericin B (3 mg/kg/day for 7 days) combined with 5-flucytosine (100 mg/kg taken four times daily for 7 days) in the induction phase, followed by fluconazole (800 mg/day for 8 weeks) for the consolidation phase was administered to 95.7% of patients who were hospitalized after MC diagnosis. One (33.3%) of the Cc/ Cl infected patients was treated with fluconazole alone (800 mg/days) due to lack of meningeal location confirmation. Patients infected with Cc/Cl remained hospitalized for 12.4 vs 9.9 days in the Cn group (not significant). Thirteen out of 23 Cn group patients (56.5%) had a pejorative clinical course (death, status quo, and discharge against medical advice or transfer due to complications) versus 16.7% in the Cc/Cl group (not significant).

Discussion
Cryptococcal meningitis has been described as an opportunistic infection in HIVIP, mainly caused by the Cn/ Cg species complex [15,16]. However, non-neoformans and non-gattii Cryptococcus species have recently been isolated in cryptococcosis both in immunosuppressed and immunocompetent patients. C. curvatus and C. laurentii are among these species [3,6,17]. While C. laurentii has already been associated with bloodstream,  neuromeningeal, pulmonary, cutaneous, and peritoneal infections [3,4,[18][19][20], C. curvatus has very rarely been implicated in human infections, mainly in peritoneal and myeloradicular infections [5][6][7]. Of the 29 Cryptococcus spp. meningitis patients with positive culture included in the present study, six (20.7%, 95% CI 6.9-34.5%) had non-neoformans and non-gattii Cryptococcus species, including five cases of C. curvatus and one case of C. laurentii. To our best knowledge, this study is the first to compare the biological and clinical characteristics of Cryptococcus spp. meningitis induced by these two species groups in the sub-Saharan African region. Globally, 79.3% were identified as Cn and 20.7% as Cc/Cl. A similar prevalence of non-neoformans and non-gattii Cryptococcus infections was reported in three states of the United States of America (USA) in 2020 and during the same study, C. laurentii, C. liquefaciens, and C. magnus were considered as pathogenic [21].
Patients with Cn were most susceptible to headache on admission than patients with Cc/Cl and most neuromeningeal signs were also found preferentially in Cn patients. The distribution of these signs could suggest a more virulent trait of Cn strains compared to Cc/Cl strains. The main virulence factors of the Cryptococcus genus i.e. presence of the polysaccharide capsule, resistance to high temperatures (> 37 °C), and the activity of the laccase enzyme, have been identified for the Cn complex as well as for the non-neoformans and non-gattii Cryptococcus species. However, the laccase enzyme activity was described as lower for non-neoformans and non-gattii Cryptococcus species compared to that of the Cn [7]. This could explain the more severe clinical presentation of the Cn group patients compared to Cc/Cl group, as described also by Cano et al. [21].
As mentioned above, most species of the genus Cryptococcus possess a polysaccharide capsule and share the same antigenic determinants with minor differences, and are therefore likely to be detected by India ink staining and/or have antigenic activity detectable by diagnostic tests [22]. Nevertheless, identification of Cryptococcus species still requires strain culture and PCR assays. As described in the literature [7], the positivity rate of India ink detection in the present study was higher in the Cn group than in the Cc/Cl group, similarly for CSF  detection of cryptococcal antigens which had a much higher positivity rate in the Cn group. The partial expression of some virulence enzymatic factors in non-neoformans and non-gattii Cryptococcus species as described by K. Ferreira-Paim et al., and mentioned earlier in this manuscript could explain the low reactivity of Cc/Cl capsular antigens to the CrAg assay, which was developed for the detection of Cn/Cg species complex antigens. Given the increase in meningitis cases due to non-neoformans and non-gattii species worldwide, it is useful that the reactivity of these species to commercially available CrAg assays be established in subsequent larger studies [23]. For culture-positive samples with negative CrAg results, the antigen excess zone (prozone phenomenon) may be the explanation. Dilution of the tested sample could have resolved this discordance.
While Cn strains were easily identified by MALDI-TOF MS, ITS sequencing, and multiplex PCR serotyping used during the study, only four Cc/Cl strains (out of six) were identified by MALDI-TOF MS and all of them had a different profile from the reference agarose gel profile after multiplex PCR. The results of the MALDI-TOF MS identification were only 80% conclusive for the Cc/Cl strains. In the remaining cases, it was either an identification with a bad score or an outright wrong identification. As the spectrum generated during sample analysis is compared with the spectra in the manufacturer's database to establish a match, a limited panel of spectra or the absence of spectra of a microbial species in the database could result in a failed species identification. As both species are present in the database used, BD 8326 Bruker, an extraction process using ethanol, formic acid, and acetonitrile might have been necessary to improve the results because of the "big shell" of Cryptococcus spp. The non-neoformans/ non-gattii shell could be more refractory than that of Cn. For its part, the determination of serotypes of Cc/Cl strains by classical multiplex PCR targeting LAC1 and CAP64 genes was not possible. Based on its initial application, this PCR was designed to characterize strains of the Cn/Cg species complex. Given the superiority of the laccase enzyme activity of the Cn/Cg complex over that of non-neoformans and non-gattii species as described above, these results could be partially explained [24].
Although some peritonitis due to non-neoformans and non-gattii Cryptococcus species has been cured by early removal of the catheter without antifungal treatment, others require more intensive treatment because of the fluconazole and 5-flucytosine resistance associated with these isolates [3,25].
In the present study, 33.3% of the Cc/Cl strains had moderately high MICs to fluconazole (16-32 mg/L) compared to 8.7% in the Cn group. The proportion of Cc/Cl strains resistant to 5-flucytosine and fluconazole is evaluated between 50 and 100% in other studies and higher than in Cn/Cg complex, which is more marked than what we could observe [4,7,20]. One strain (16.7%) of the Cc/Cl group, C. laurentii, was resistant to amphotericin B which is known as the most effective antifungal agent in the management of Cryptococcus infections by all species [3,26]. This same strain was also resistant to 5-flucytosine (16.7%).
All patients with Cn meningitis received antifungal agents according to WHO recommendations and guidelines, MSF protocol [27]. In contrast, one patient in the Cc/Cl group was treated with fluconazole alone although the strain was susceptible in vitro. Despite this, the patients' therapeutic outcome was not significantly different in the two groups. Patients were selected in MSF-supported clinics where the whole management process is codified. Thus, patients for whom the Cryptococcus identification was provided by the study a few weeks after sample collection were not treated based on this identification. Consequently, 4.3% of patients in the Cn group versus 33.3% of patients in the Cc/Cl group were not treated with antifungals. The mean length of hospital stay for patients was 12.4 versus 9.9 days with the usual tri-antifungals, Cc/Cl, and Cn groups respectively. Longer hospital stays (60 days) have been described for C. albicans meningitis on amphotericin B [28].

Conclusions
Cn meningitis is clinically more severe than that caused by the Cc/Cl complex due to the high virulence of this species. Accordingly, some biological parameters are more altered in Cn infection than Cc/Cl, on one side. The routine biological detection and molecular identification of Cc/Cl strains are more delicate than Cn. Plus, Cc/Cl strains require high antifungal MICs than Cn in vitro, on the other side. That is why Cc/Cl complex should be considered in HIVIP management with meningitis.