Cryptococcal meningitis and meningoencephalitis (MME) are major causes of global infectious morbidity and mortality and can affect patients regardless of immune status (Iyer et al., 2021; Lee et al., 1996; Sudan et al., 2013; Su et al., 2024). Mortality rates due to cryptococcal MME reportedly range from 24% to 47% at 10 weeks (Jarvis et al., 2022; Molloy et al., 2018). In 2022, the World Health Organization listed Cryptococcus neoformans as a top fungal priority pathogen (Chang et al., 2024; World Health Organization, 2022).

Fluconazole (FLC) has been widely used for over 3 decades in the treatment of fungal infections. However, its increased global use has been associated with rising minimum inhibitory concentrations (MICs) against a variety of fungal pathogens (Bicanic et al., 2007; Pfaller et al., 2009; Sar et al., 2004). Due to FLC’s relatively narrow antifungal spectrum and the gradual increase in resistance, alternative antifungal agents with broader activity have been developed over the past 2 decades to target clinically important fungal species (Su et al., 2024).

Infections caused by Cryptococcus species in the central nervous system (CNS) are not confined to the cerebrospinal fluid (CSF). Lee et al. reported that these infections not only result in severe meningitis due to disruption of the blood-CSF barrier, but also involve the brain parenchyma, leading to severe encephalitis (Lee et al., 1996). This is particularly common in immunocompromised hosts such as patients with chronic human immunodeficiency virus (HIV) infection, Hodgkin’s disease, liver cirrhosis, or those undergoing solid organ transplantation (SOT), etc. (Chang et al., 2024). Therefore, when selecting antifungal agents for the treatment of CNS cryptococcosis, it is essential to consider their penetration into both the CSF and brain tissue. FLC and voriconazole (VRC) have demonstrated effective CNS penetration (see below), even in the absence of meningeal inflammation—unlike amphotericin B (AMB) and posaconazole (POS) (Felton et al., 2014; Nau et al., 2010).

The 2024 guidelines for cryptococcosis treatment recommend all newer triazoles as alternative agents (Chang et al., 2024). Although the guidelines mention alternative triazoles, robust evidence on MIC distributions and pharmacokinetic (PK)/pharmacodynamic (PD) analyses for Cryptococcus are scarce. Comodeling both PK and PD data—where available—provides valuable insights into the importance of tissue concentrations for antifungal agents (Felton et al., 2014). This knowledge gap, therefore, prompted us to analyze the variability among these key antifungal agents from multiple perspectives relevant to the effective treatment of cryptococcal infections worldwide.

The variability of MIC measurements of microorganisms—including Cryptococcus isolates and other fungi—against a given antifungal is well recognized. This variability is generally attributed to biological variation (i.e., inter-strain differences), intra-laboratory assay variability, and inter-laboratory methodological differences (Espinel-Ingroff et al., 2009; Espinel-Ingroff et al., 2012a; Mouton et al., 2018). Nevertheless, at any site of infection, MIC data for a specific antimicrobial against pathogens are crucial for understanding exposure-response relationships. Additionally, when the PK-PD parameters are applied, it is also important to consider that the hysteresis phenomenon, which refers to a delayed distribution of an antimicrobial to the site of action or the involvement of active metabolites. This may result in a lag or persistence in the antimicrobial effect (liposomal AMB [LAmB] (Bekersky et al., 2000) and caspofungin), even when the plasma drug concentration has declined (Felton et al., 2014; Hope, 2012).

The epidemiological cutoff values (ECVs) of antifungal agents against Cryptococcus species are not predictive of in vivo efficacy (Pharkjaksu et al., 2020), but they are used to distinguish wild-type (WT) from non-WT (i.e., resistant) isolates (CLSI, 2018). Due to the complexity of interpreting susceptibility data, we referred to the ECVs of antifungals against Cryptococcus species documented in the PubMed database (Jean et al., 2024), rather than relying on a single MIC value—such as MIC₉₀ values of invasive Cryptococcus isolates for given antifungals—as a surrogate for further analysis.

Given the known PK variability of agents such as VRC, incorporating ECV-based strategies for dose optimization is methodologically sound—particularly in the absence of robust PD targets for Cryptococcus infections. In this context, PD targets established for Candida species (Li et al., 2010) are cautiously extrapolated to Cryptococcus isolates in this investigation. Furthermore, relatively few studies have estimated concentration-time profiles of antifungals in humans. Therefore, PK data for key antifungal agents—including data from laboratory animal models, where relevant and where corresponding human data are unavailable—were obtained from previously published sources (Jean et al., 2022). In this study, we applied the documented ranges of PK parameters, rather than relying solely on mean values, to evaluate the probability of PK/PD target attainment. The novelty of this study, which combines Antimicrobial Testing Leadership and Surveillance (ATLAS) MIC data with PK/PD analysis, lies in its indirect assessment of the appropriateness of various antifungal dosages for treating Cryptococcus infections, which often involve the CSF and cerebral tissue.

The strong negative correlations observed between FLC and VRC/POS MICs (Figure 1A) suggest that possible resistance mechanisms to FLC—such as mutations in ERG11 (encoding lanosterol 14α-demethylase [i.e., CYP51]) and efflux pump overexpression—may not confer cross-resistance to newer triazoles like VRC and POS. Moreover, chromosomal plasticity, particularly whole-chromosome aneuploidy (notably chromosome one disomy in C. neoformans), can emerge rapidly under triazole pressure and markedly increase FLC MICs, facilitating Cryptococcus adaptation to FLC and highlighting in-host evolution driven by genome plasticity (Sionov et al., 2010).

The parallel trend in VRC and POS MIC distributions (Figure 1B) indicates possible co-selection of resistance upon exposure, implying that empirical use of either agent may be less reliable without MIC-guided therapy.

Structurally, POS contains a rigid triazole core and a bulky hydrophobic side chain, whereas ISA features a bulky isobutyryloxy side arm attached to its core, which greatly enhances the orientation of its triazole ring and enables tighter binding to fungal CYP51 (Datta et al., 2013). Differences in CYP51 binding affinity between ISA and VRC/POS to Cryptococcus isolates suggest ISA’s ability to evade common resistance mechanisms of other azoles, such as overexpression of the ABC transporter AFR1 (Arastehfar et al., 2020), which actively extrudes many triazoles from the fungal cells, thereby lowering their intracellular concentrations; and point mutations like Y145F and G484S in the ERG11 gene (Sionov et al., 2012), which alter the CYP51 binding pocket and reduce the affinity of many triazoles, particularly for VRC and POS, to non-WT Cryptococcus isolates. As a result, cross-resistance between ISA and other triazoles cannot be reliably inferred.

These findings may have therapeutic implications. The distinct susceptibility profile of ISA suggests that it could serve as an alternative treatment option for non-WT Cryptococcus strains exhibiting elevated MICs to FLC, VRC, and POS. ISA or LAmB should be strongly considered for the treatment of systemic cryptococcosis in patients with recent exposure to VRC or POS.

In this study, we identified dosing strategies optimized for the effective treatment of invasive Cryptococcus infections by leveraging antifungal MIC data from the ATLAS database, documented ECVs, and published PK–PD parameters. Additionally, as illustrated in Figure 2, we cautiously excluded 3,486 studies that did not report the detailed methods to generate the PK data for antifungals, or clinical outcomes for patients with cryptococcosis receiving antifungal therapy.

The published ECVs for FLC against Cryptococcus species range from 4 mg/L (for C. neoformans and C. grubii) to 8 mg/L (for C. gattii) (Espinel-Ingroff et al., 2012a; Jean et al., 2024; Lockhart et al., 2012). Among the invasive Cryptococcus isolates evaluated, approximately 91% had FLC MICs below 4 mg/L and were classified as WT strains (Table 1).

For FLC, we screened 1,264 studies to investigate its PK profile. Based on the PK data for FLC (Arredondo et al., 1994; Buijk et al., 2001; Nau et al., 2010; Santos et al., 2016) (Table 5), the free-drug area under-the-concentration-time curve from zero to 24 h (AUC_0-24) is estimated to range from 191.7 μg·h/mL to 375.4 μg·h/mL in CSF, and from 157.6 μg·h/mL to 226.1 μg·h/mL in brain tissue for patients receiving 400 mg of FLC daily. Accordingly, regrowth of Cryptococcus isolates residing in brain tissue with an MIC of ≤4 mg/L to FLC could be effectively inhibited in patients with CNS cryptococcosis receiving FLC at 400 mg/day.

Similar to the global WT rate of 90.9% (Jean et al., 2024), approximately 91% of worldwide Cryptococcus isolates in the CSF exhibited FLC MICs ≤4 mg/L. Additionally, data from the 2010–2020 antifungal ATLAS database indicate that the MIC distribution of invasive Cryptococcus isolates against FLC was statistically shifted toward lower values in the Asia–Western Pacific region and Europe (MIC₉₀ = 4 mg/L), compared to higher values in North and Latin America (MIC₉₀ = 8 mg/L) (Jean et al., 2024).

Because FLC concentrations in blood and CSF correlate strongly with daily dose in humans (Schiave et al., 2018), increasing the maintenance dose of FLC to at least 800 mg/day may be necessary for effective treatment of patients with CNS cryptococcosis during the induction phase—particularly in regions outside the Americas. For invasive Cryptococcus infections in North and Latin America, and specific genotypes of C. gattii isolates in China and Colombia (Firacative and Escandón, 2021; Hagen et al., 2010; Xu et al., 2021), increasing the FLC maintenance dose up to 1,200 mg/day, adding 5-flucytosine (5-FC), or switching to VRC (see below) during the induction phase appears to be warranted. This dosage rationale is consistent with previous investigations (Schiave et al., 2018; Sudan et al., 2013).

For VRC, the published ECVs for Cryptococcus species reportedly ranged from 0.5 mg/L (for C. neoformans and C. grubii) to 1 mg/L (for Cryptococcus gatti) (Espinel-Ingroff et al., 2012a; Jean et al., 2024; Lockhart et al., 2012). The majority of Cryptococcus isolates collected in the 2010–2020 ATLAS project exhibited MIC ≤0.5 mg/L to VRC (100% for BSI isolates and 99.5% for CSF isolates) (Table 1).

For effective treatment of yeast infections, an AUC/MIC ratio greater than 25 is generally recommended for azoles (Li et al., 2010). If the MICs of Cryptococcus isolates to VRC approach the ECV (1 mg/L), achieving an adequate AUC of ≥25 μg h/mL at the infection site becomes critical. For VRC, we screened 1,130 studies to investigate its PK profile. Integrating the human PK profiles from important literature regarding administration of standard VRC dosage (Table 5) (Al-Saigh et al., 2012; Elter et al., 2006; Felton et al., 2014; Hope, 2012), VRC appears to be effective in inhibiting the growth of Cryptococcus isolates in the CNS, even in the absence of pronounced meningeal inflammation.

Anecdotal case reports have also demonstrated the efficacy of VRC after other antifungal therapies failed in the treatment of cryptococcosis (Bandettini et al., 2009; Carbonara et al., 2009; Chang et al., 2010; Shen et al., 2008). However, therapeutic drug monitoring (TDM) for VRC is warranted during its administration to avoid adverse effects (e.g., hepatotoxicity, visual disturbances, and neurotoxicity), as significant variations in serum levels can occur due to multiple influencing factors.

The published ECV for POS against major Cryptococcus species is 0.5 mg/L (Espinel-Ingroff et al., 2012a; Jean et al., 2024; Lockhart et al., 2012). Similar to VRC, 99.5% of invasive Cryptococcus isolates collected in the 2010–2020 ATLAS project were categorized as WT strains for POS (Table 1).

For POS, we screened 595 studies to investigate its PK profile. Although the 2024 global cryptococcosis treatment guidelines list POS as an optional alternative to FLC (Chang et al., 2024), the molecular weight of POS (approximately 700 g/mol) is significantly higher than that of other triazoles (Felton et al., 2014). This characteristic may partly explain the extremely low concentrations of POS in brain tissue and CSF (Felton et al., 2014). These PK limitations are consistent with the findings from a case series by Pitisuttithum et al., which enrolled 29 patients with cryptococcal meningitis receiving POS therapy (Pitisuttithum et al., 2005). The series reported a clinical failure rate of 51.7% (15/29). Among the 14 patients with successful outcomes, 6 (42.9%) still tested for positive India ink staining in the CSF, and one patient remained serum cryptococcal antigen-positive (Pitisuttithum et al., 2005). Accordingly, POS appears to be a suboptimal option for managing cryptococcosis involving the CNS, especially when compared to other triazoles (Arredondo et al., 1994). It is noteworthy that POS is predominantly metabolized by the CYP3A4 enzyme. TDM for immunosuppressants is essential to optimize their efficacies and minimize toxicities in immunocompromised hosts.

The published ECV for ISA against major Cryptococcus species is 0.25 mg/L (Espinel-Ingroff et al., 2015; Jean et al., 2024). Also similar to VRC and POS, 98.7% of the studied Cryptococcus isolates exhibited MIC below 0.25 mg/L for ISA (Table 1).

For ISA, we screened 234 studies to investigate its PK profile. ISA displays highly variable PK in human plasma (Desai et al., 2016). Despite these data, adjustment of the ISA maintenance dosage is usually not needed.

Notably, ISA achieves adequate concentrations in brain tissue (Wiederhold et al., 2016) but demonstrates poor penetration into the CSF, with plasma-to-CSF concentration ratios ranging from 0.02 to 0.04 in a rabbit model (Kovanda et al., 2019). This poses a significant challenge for its use in cryptococcal MME unless supported by another antifungal agent. Additionally, similar to POS, ISA is primarily metabolized by the CYP3A4 enzyme and, to a lesser extent, via glucuronidation pathways. TDM for key immunosuppressants is essential in immunocompromised hosts. Nevertheless, ISA may be preferred over VRC in elderly patients for antifungal treatment.

A small case series by Thompson et al. included five patients with cryptococcal meningitis treated with standard ISA dosing (372 mg every 8 h for six doses, followed by 372 mg once daily from day 3 onward) for durations ranging from 11 to 181 days. This study reported that there was complete success in only one case, partial success in three cases, and a fatal outcome in one case—despite sterile CSF cultures (Thompson et al., 2016). These findings further raise concerns regarding the clinical efficacy of ISA for the treatment of cryptococcosis, particularly in CNS infections.

A summary of key PK parameters—including mean total-drug plasma AUC_0–24, protein-binding rates, and ratios of plasma-to-CSF penetration—for the four triazoles is illustrated in Figure 3.

For AMB, the published ECVs for Cryptococcus species range from 0.5 mg/L for C. gattii to 1 mg/L for C. neoformans and C. grubii (Espinel-Ingroff et al., 2012b; Jean et al., 2024). Approximately 43% of Cryptococcus isolates had MICs ≤0.5 mg/L, and 99% had MICs ≤1 mg/L to AMB (Table 1). Additionally, 55% of the studied C. gattii isolates (6/11) were categorized as WT strains, in contrast to 99%–100% of the C. neoformans and C. grubii isolates, which were WT (Table 4). Notably, the potential to AMB deoxycholate to achieve adequate CSF concentrations is constrained by its toxicity (Bekersky et al., 2002).

Over the past 2 decades, LAmB has been increasingly favored for the treatment of cryptococcal MME due to its reduced toxicity profiles (Iyer et al., 2021; Jarvis et al., 2019; Jarvis et al., 2022). Additionally, LAmB shows non-inferior efficacy to AMB deoxycholate for treatment of high fungal burden cases, such as disseminated cryptococcosis or severe pulmonary disease (Jarvis et al., 2022; Kimuda et al., 2025).

For LAmB, we screened 287 studies to investigate its PK profile. The useful PK profiles were clearly documented in a few investigations (Jean et al., 2022; Lee et al., 1994; O'Connor et al., 2013). The optimal PD index of AMB against yeasts is the ratio of the maximal concentration (C_max) to MIC (C_max/MIC), with a target value of ≥4 at the site of infection (Li et al., 2010).

Jarvis et al. demonstrated that a single high-dose administration of 10 mg/kg LAmB on day 1 is required in combination with maintenance antifungals with good CSF penetration (e.g., FLC 1200 mg/day and 5-FC 100 mg/kg/day) to achieve optimal CSF clearance rates of Cryptococcus (−0.40 log₁₀ cfu/mL/day) in HIV patients with cryptococcal MME (Jarvis et al., 2022). Although data on the efficacy of a single high-dose LAmB in SOT recipients and other cryptococcal disease syndromes are lacking, a single 10 mg/kg LAmB therapy for cryptococcal BSI/MME is anticipated to significantly reduce the health-economic burden in low- and middle-income countries compared to a 3–4 mg/kg LAmB daily dosage for 7–14 days, and thus is worthy of broader consideration.

This survey of antifungal MIC distributions for Cryptococcus isolates has several limitations. First, the PD target used for Cryptococcus in this study was extrapolated from Candida studies, which may not accurately reflect cryptococcal pathophysiology. Second, LAmB and the newer triazoles remain largely unavailable in many low-income countries, limiting generalizability of the findings. Third, the number of C. gattii isolates in this study was too small to draw reliable conclusions. Fourth, even with early diagnosis, low MIC levels against Cryptococcus isolates and apparently optimal PK-PD profiles for a specific antifungal may not guarantee favorable outcomes, particularly in patients with severe systemic cryptococcosis and compromised immune conditions. Therefore, further clinical studies are needed to validate these observations.

In conclusion, all the newer triazoles (VRC, POS, ISA) demonstrated low MIC values against major Cryptococcus species. Both VRC and FLC achieve optimal CNS penetration. The combination of LAmB with FLC (1,200 mg daily) and 5-FC (100 mg/kg daily)–a cost-effective regimen–has been listed the standard induction–phase regimen. VRC exhibits superior PK characteristics for managing CNS cryptococcosis relative to ISA and POS, and appears to be a reasonable alternative to FLC during the induction treatment. However, due to substantial interindividual variability in PK profiles and notable drug-drug interactions, TDM for VRC is indicated for dose optimization. Additionally, ISA may represent a preferable option over POS during the consolidation phase of cryptococcosis treatment.