Introduction

HIV-associated cryptococcal meningitis (CM) remains a significant public health problem and a major cause of AIDS-related mortality, despite increasing access to antiretroviral therapy (ART) [1, 2]. There are an estimated 220,000 cases each year, with the greatest burden of disease found within resource-limited, high HIV prevalence settings. Across sub-Saharan Africa, CM remains the most common cause of meningitis in adults living with HIV [3]. As ART programmes were introduced, the number of new cases fell, but in recent years, and despite widened ART rollout, this number has plateaued with CM being responsible for more than 180,000 annual deaths and 15% of all AIDS-related mortality globally [4•].

This review will focus on new approaches to the treatment of CM, with a specific focus on the potential of short-course or oral combination antifungal treatment regimens in low- and middle-income countries (LMICs). We outline the limitations of the currently available treatment options, summarise recently published research that has changed treatment guidelines, and describe ongoing and future studies that may further inform this approach. We then contextualise these data and discuss current barriers to implementation that need to be overcome, particularly regarding access to medications in resource-limited settings.

The current treatment paradigm

Cryptococcal meningitis is uniformly fatal if left untreated [4•]. Treatment of CM is challenging, requiring intensive antifungal therapy, the identification and management of drug-related toxicities, strict control of raised intracranial pressure, and the safe promotion of immune recovery with ART. The median CD4 count of individuals diagnosed with CM is approximately 30 cells/μL [5••], and therefore, the management of these patients also involves the diagnosis and treatment of other complications of advanced HIV disease.

The unacceptably high mortality associated with CM is driven, at least in part, by the limited availability of effective antifungal medication in LMICs. The most widely used oral antifungal treatment is fluconazole which is affordable, widely available (often through donation), and has a good safety profile. When given in high doses (800–1200 mg per day) as monotherapy for CM in a clinical trial setting, fluconazole is associated with a 10-week mortality of 50–60% [6, 7]. The limited efficacy of fluconazole when used in monotherapy is attributed to its predominantly fungistatic activity, inhibiting growth of Cryptococcus to achieve a therapeutic effect, in contrast to other agents such as amphotericin which are fungicidal. Fluconazole remains predominantly fungistatic even at high doses; when maximally tolerated doses are given, clinical failure rates remain high in comparison to other regimens [8]. In addition, fluconazole susceptibility in Cryptococcus species appears to be decreasing. A systematic review in 2019 identified a rise in the median fluconazole minimum inhibitory concentration (MIC) from 4 μg/mL between 2000 and 2012 to 8 μg/mL globally between 2014 and 2018 [9]. Similar findings were observed in South Africa where the MICs of prospectively collected cryptococcal isolates demonstrated a doubling of the geometric mean MIC between 2007 and 2008 and 2017 from 2.08 to 4.11 μg/mL [10]. Furthermore, a recent study demonstrated that even in settings where primary genetic fluconazole resistance is rare, poor fungal clearance occurs due to the phenomenon of heteroresistance: small numbers of resistant subpopulations existing at levels low enough not to exert a significant effect on the MIC but, in the presence of fluconazole alone, susceptible subpopulations die and resistant clones dominate [11].

Given the marked limitations of fluconazole monotherapy, most international guidelines for the management of CM in resource-limited settings have recommended the addition of amphotericin B deoxycholate which, when administered for 2 weeks alongside high-dose fluconazole, is associated with roughly 30–40% mortality at 10 weeks [5••] in clinical trial settings. However, amphotericin B is notoriously toxic, and induction treatment for 2 weeks is commonly associated with thrombophlebitis, as well as an average drop in haemoglobin of 2.3 g/dL and an increase in creatinine of 73% [12] (Fig. 1). These toxicities require regular monitoring which may be difficult in settings with limited healthcare workers and laboratory capacity. When comparing outcome data from clinical trials with those collected through observational studies, these challenges appear to translate to higher mortality rates in routine care. Data collected from the national death registry in Botswana between 2012 and 2014, where amphotericin B and fluconazole was the standard of care, identified a 10-week mortality of 50% [13]. In comparison, randomised controlled trials of this regimen conducted in Vietnam and across multiple sites in Africa have demonstrated lower mortality rates of 30–40% [5, 14]. Whilst selection bias may influence the overall mortality rates in some clinical trials, closer monitoring for toxicities and enhanced clinical care likely confer a significant survival benefit.

Fig. 1
figure 1

Individual data points and fitted locally weighted scatterplot smoothing curves for haemoglobin (left) and creatinine (right) values over the first 14 days of antifungal therapy. (a) All patient receiving 14 days of amphotericin B based induction therapy. (b) Plot by amphotericin B duration (short-course versus standard treatment). (c) Plot by sex. The broken lines indicate an anaemia threshold of 6.5 g/dL (division of AIDS grade IV adverse event) and a creatinine of 220 μmol/L, indicating nephrotoxicity [12].

The combination of 2 weeks of intravenous treatment alongside the associated toxicity means that managing a CM patient with this regimen involves a minimum hospital stay of 14 days and regular blood tests. As a consequence, in Uganda, the hospitalisation costs are approximately USD 560 [15]. In Zimbabwe, costs are approximately USD 800–1000 [16], and in South Africa, the cost is estimated at USD 2453 [17]. Finally, amphotericin B must be refrigerated, and the maintenance of a cold chain poses logistical challenges in resource-limited settings.

Shorter-course amphotericin-based treatment

The major challenges of managing patients with two-week courses of amphotericin B in low-resource settings have prompted investigation into whether shorter courses could provide similar antifungal effects and reduce the associated cost and toxicity. Data from animal models have paved the way for future clinical trials. In particular, a 2013 study using murine and rabbit models showed that 3 days of amphotericin B as induction therapy was as effective as daily therapy for 14 days. These animal data also showed a favourable pharmacokinetic profile of the abbreviated amphotericin B regimen, with sustained antifungal activity and persistence in the CSF (Fig. 2) [18•].

Fig. 2
figure 2

Effects of various regimens of amphotericin B for experimental cryptococcal meningoencephalitis in rabbits. Amphotericin B at 1 mg/kg was administered daily (b), every third day (c), as an abbreviated regimen for 3 consecutive days (d), or once (e). The solid arrows show the times of drug administration. The data in panels (a) to (e) represent the time course of infection in the CSF. Each line represents the fungal burden in an individual rabbit. Panel (f) shows the fungal density in the cerebrum at the end of the experiment. Each bar represents the mean fungal density ± the standard error of the mean. There is no statistically significant difference in cerebral fungal density between rabbits that received daily therapy and those that received an abbreviated regimen of 1 mg/kg at 48, 72, and 96 h after inoculation. *, p = 1.00 (not significant); **, p = 0.006; ***, p < 0.001; ****, p < 0.001; *****, p = 1.00 (not significant) [18•].

Two phase II clinical studies, one conducted in Uganda [19] and another in Malawi [7], both found participants treated with shorter courses of amphotericin B had good clearance of Cryptococcus from the brain and favourable side-effect profiles. The Ugandan study enrolled 30 HIV-infected patients with first-episode CM and treated them for 5 days with amphotericin B alongside 2 weeks of high-dose fluconazole. Repeat lumbar punctures were performed on days 3, 7, and 14 to establish the early fungicidal activity (EFA), a calculation of the mean rate of clearance of Cryptococcus from the CSF which correlates with the clinical outcome. Although the study was underpowered to detect any difference in mortality, it demonstrated mono-exponential fungal clearance over the observed 14 days with the calculated EFA being the same at 7 (− 0.31 ± 0.14 log CFU/day) and 14 days (− 0.30 ± 0.11 log CFU/day), thereby supporting the animal model data showing persistent antifungal effect after cessation of amphotericin B. The Malawian study, through a two-step process, randomised patients to one of four treatment arms: either 2 weeks of fluconazole monotherapy with or without amphotericin for 7 days or 2 weeks of fluconazole and flucytosine with or without amphotericin for 7 days. Similar to the Ugandan study, there was no reduction in EFA in the second week of treatment after the course of amphotericin B had been completed. In addition, when compared with historical cohorts treated with 2 weeks of amphotericin B, both of these studies found similar rates of EFA. The triple-combination treatment arm in the Malawian study, containing fluconazole, amphotericin B, and flucytosine, had a significantly higher EFA (− 0.50 ± 0.15 log CFU/day) than the other treatment arms (amphotericin and fluconazole − 0.38 ± 0.19 log CFU/day; fluconazole and flucytosine − 0.28 ± 0.17 log CFU/day) supporting the rationale for phase III trials exploring both shorter-course amphotericin B treatment and the role of flucytosine.

Flucytosine and the potential for oral combination therapy

In addition to the potential benefits of shorter-course amphotericin-based regimens, the antifungal flucytosine has been tested both as an alternative and in addition to fluconazole. Flucytosine’s effectiveness in combination therapy was demonstrated in a 1997 trial showing flucytosine to increase the rate of CSF sterilisation. In this study, 60% of patients receiving amphotericin B and flucytosine had negative CSF cultures at 2 weeks compared to 51% of patients treated with amphotericin alone [20]. Subsequent trials reiterated this and also highlighted flucytosine as superior to fluconazole when used in combination with amphotericin B. In a randomised controlled trial that recruited 299 participants in Vietnam, combination therapy with flucytosine was shown to confer a survival benefit at 10 weeks when compared to amphotericin alone (30% versus 44%; HR 0.61; 95% CI 0.39–0.97; p = 0.04). Despite a lower absolute mortality rate, there was no statistically significant difference in mortality when this arm was compared to amphotericin B given with fluconazole (30% versus 33%; HR 0.87; 95% CI 0.53–1.42; p = 0.57) [14].

In the same Malawian study discussed above, flucytosine and fluconazole combination therapy had a higher EFA than fluconazole alone [7]. A second study in Malawi also found this effect on EFA in addition to improved mortality at 2 (10% versus 37%; HR 0.24; 95% CI 0.05–1.16; p = 0.05) and 10 weeks (43% versus 58%; HR 0.59; 95% CI 0.25–1.44; p = 0.25) [21]. One benefit of an oral combination antifungal treatment regimen is that it would not require any daily intravenous infusion which would eliminate the risk of treatment-associated thrombophlebitis. An oral treatment could also potentially reduce the length of hospital admission, enabling milder or asymptomatic cases to continue induction therapy at home and therefore reducing the costs associated with CM. It is unlikely that this approach would ever obviate the need for hospital admission as CM patients are often very unwell and can deteriorate rapidly.

The ACTA study

A highly influential study in this field was the Antifungal Combinations for Treatment of Cryptococcal Meningitis in Africa (ACTA) phase III randomised controlled trial. This clinical endpoint trial recruited participants in southern, eastern, and central Africa. HIV-infected patients with first-episode CM were randomised to one of five arms: 2 weeks of amphotericin (1 mg/kg/day) with either fluconazole (1200 mg/day) or flucytosine (100 mg/kg/day), 1 week of amphotericin with either fluconazole or flucytosine, or oral combination therapy with flucytosine and fluconazole. The trial recruited a total of 721 participants, and a pooled analysis of mortality at 10 weeks identified that, irrespective of partner drug, treatment for 7 days with amphotericin B was non-inferior to treatment for 14 days (HR 0.89; 95% CI 0.66–1.21) (Fig. 3). Those randomised to the arm receiving 7 days of amphotericin plus flucytosine experienced the lowest 10-week mortality rate (24.2%) when compared with all other regimes, including 2 weeks of amphotericin plus flucytosine (HR 0.56; 95% CI; 0.35–0.91; p = 0.001).

Fig. 3
figure 3

Kaplan-Meier curves from the ACTA study illustrating the all-cause mortality at 10 weeks according to treatment strategy. The trial recruited a total of 721 participants, and mortality in the oral regimen, 1-week amphotericin B, and 2-week amphotericin B groups was 35.1%, 36.2%, and 39.7% respectively. Irrespective of the partner drug, treatment with 7 days of amphotericin B was non-inferior to treatment for 14 days (HR 0.89; 95% CI 0.66–1.21) [5••].

As a partner drug to amphotericin B, flucytosine performed better than fluconazole (HR for death at 10 weeks with flucytosine vs. fluconazole 0.62; 95% CI 0.45–0.84; p = 0.002). All the flucytosine-containing regimens performed better, with the data showing that fluconazole as a partner drug to amphotericin B was associated with the highest 10-week mortality: 48.6% when given for 1 week (indicating that flucytosine is essential if short courses of amphotericin are used) and 41.3% when given for 2 weeks of amphotericin B. Despite having the slowest rate of clearance of cryptococcus from the CSF, the oral combination arm was more effective in averting mortality than both of the amphotericin B and fluconazole arms, and was the second best performing arm with a 10-week mortality rate of 35% (95% CI 28.9–41.3%). This amphotericin-sparing regimen was also associated with the smallest median change in both haemoglobin (− 0.4 g/dL IQR − 1.0–0.4) and creatinine (0 μmol/L IQR − 8.8–13.0).

Following the publication of the ACTA trial, a Cochrane review of treatment for CM [22] analysed 13 clinical trials, most of which were conducted in resource-limited settings. The review concluded that 1 week of amphotericin B with flucytosine probably resulted in a lower risk of death within 10 weeks, with shorter duration of amphotericin B therapy being associated with a lower risk of life-threatening toxicities. Furthermore, the review highlighted that combination oral therapy with flucytosine and fluconazole had a similar risk of death compared to the traditional 2 weeks of amphotericin B and flucytosine. Based on these findings, the World Health Organization updated their guidance on the management of CM in a resource-limited setting to recommend 1 week of amphotericin B with flucytosine as a first-line treatment in LMICs [23•]. In addition, where amphotericin B is not available, the oral combination therapy is recommended as a second-line regimen.

The role of liposomal amphotericin

The use of liposomal amphotericin (L-AmB, AmBisome, Gilead Sciences) has the potential to further improve outcomes in the management of CM. Despite being the mainstay of treatment in high-income settings [24, 25], L-AmB has not been widely used in LMICs due to high costs. L-AmB is however well-suited to short-course treatment of CM because high doses can be safely administered owing to lower rates of drug toxicity than amphotericin B, and because the drug has a long half-life in the brain [26, 27]. Animal model studies have shown efficacy with higher doses of L-AmB (10–20 mg/kg/day) [28, 29]. Studies to determine the shortest effective period of induction therapy with L-AmB in murine and rabbit models used high-performance liquid chromatography to measure plasma and brain concentrations of L-AmB, finding a dose-dependent decline in fungal burden in the brain of mice, with a near-maximal efficacy achieved with L-AmB at 10–20 mg/kg/day. Interestingly, the pharmacodynamics of a single dose of 20 mg/kg was the same as that of 20 mg/kg/day given for 2 weeks [29] (Fig. 4).

Fig. 4
figure 4

The pharmacokinetics of liposomal amphotericin B in murine plasma (red) and cerebrum (black) in cohorts of mice infected with Cryptococcus neoformans receiving 20 mg/kg single-dose intravenously. The terminal half-life in the plasma and cerebrum is circa 133 h [28•].

In addition, multiple human studies have also demonstrated safety when administering higher doses of L-AmB for the treatment of visceral leishmaniasis and invasive fungal infections in cancer patients [30]. A randomised controlled trial conducted in India found that a higher dose L-AmB (10 mg/kg) for the treatment of visceral leishmaniasis was not associated with an increase in adverse effects when compared with conventional therapy [31]. In this trial, a single dose of L-AmB (10 mg/kg) was compared with amphotericin B (1 mg/kg administered every other day for 29 days). The study demonstrated similar cure rates of 95.7% (95% CI 93.4–97.9) with L-AmB and 96.3% (95% CI 92.6–99.9) with conventional amphotericin B. A similar dose has also been used in studies of stem cell and liver transplant patients for antifungal prophylaxis, and both studies showed the dose was well-tolerated [32, 33]. The potential benefits of moving from daily intravenous amphotericin B to a single dose of L-AmB include fewer drug-induced toxicities, a shorter hospital stay, and a reduction in associated costs. In addition, L-AmB can be stored at an ambient temperature which would simplify distribution and storage of medication.

The AMBITION study

The AMBIsome Therapy Induction OptimisatiON (AMBITION) phase II trial was an open-label randomised non-inferiority trial to compare different short-course L-AmB regimens for the treatment of HIV-associated CM [34••]. The study had four arms: (1) L-AmB 10 mg/kg/day on day 1 (single dose), (2) L-AmB 10 mg/kg/day on day 1 and 5 mg/kg/day on day 3 (2 doses), (3) L-AmB 10 mg/kg/day on day 1 and 5 mg/kg/day on days 3 and 7 (3 doses), and (4) L-AmB 3 mg/kg/day for 14 days (control). All patients also received oral fluconazole 1200 mg/day for 14 days. The primary endpoint was EFA and a total of 80 participants were recruited from sites in Botswana and Tanzania. The study found that the EFA in all short-course regimens was comparable to, or greater than, the control at the pre-defined non-inferiority margin. There was no evidence for any dose-response effect with additional doses. There were only 10 grade 3 and two grade 4 serious adverse events that were attributed to study drugs. These included one grade 3 and one grade 4 anaemia which compared favourably to historical data from 368 patients treated in Africa with amphotericin B where 33% of patients reported a grade 3 or 4 anaemia [12]. Across all arms, the median drop in haemoglobin over the 2-week induction period was 0.9 g/dL compared to a median drop of 2.3 g/dL observed with amphotericin B [12]. Similarly, the percentage increase of baseline creatinine was 14% compared to 73% with amphotericin B (Fig. 5). Finally, the overall 10-week mortality was 29% which compared very favourably with mortality rates recorded elsewhere in the literature [14, 35,36,37].

Fig. 5
figure 5

(a) The median percentage change from baseline creatinine and (b) the median drop in haemoglobin in all treatment arms and each individual treatment arm in the AMBIsome Therapy Induction Optimisation (AMBITION) phase II trial [34••], compared with a pooled analysis of 368 patients treated with amphotericin B [12].

The single, high-dose regimen has been taken forward to a clinical endpoint phase III trial which is currently recruiting participants from eight hospitals across five countries in southern and eastern Africa [16]. The AMBITION phase III trial will recruit 850 individuals with HIV-associated cryptococcal meningitis and randomise them 1:1 to either (1) single-dose L-AmB (10 mg/kg) given alongside 14 days of fluconazole (1200 mg/day) and flucytosine (100 mg/kg/day); or (2) 7 days of amphotericin B (1 mg/kg/day) given alongside 7 days of flucytosine (100 mg/kg/day) and followed by 7 days of fluconazole (1200 mg/day). The primary endpoint is all-cause mortality at 10 weeks. The trial commenced recruitment in January 2018 and is expected to complete follow-up in early 2021.

Cryptococcal antigen screening

Novel short-course and oral therapies for CM may also be relevant for the management of cryptococcal antigenaemia. After individuals with advanced HIV are exposed to Cryptococcus through respiratory inhalation, and usually after a period of dormancy in the lungs, the organism enters the blood stream, causing disseminated cryptococcal disease (cryptococcosis) before then reaching end organs, most commonly the central nervous system [38]. Throughout the disease, Cryptococcus sheds its polysaccharide fungal capsule which can be detected in the blood by highly sensitive and specific cryptococcal antigen (CrAg) lateral flow assays [39, 40]. The prevalence of cryptococcal antigenaemia among individuals with a CD4 count < 100 cells/μL is approximately 6.0–6.5% [41, 42], and among those with a CD4 count 101–200 cells/μL is approximately 2.0% [42]. Cryptococcal antigenaemia has consistently been shown to be highly predictive of death, most commonly due to the development of CM, and the detection of CrAg in the blood provides an opportunity to intervene and give preemptive treatment [43, 44]. CrAg screening programmes have therefore been adopted in many high HIV prevalence settings. The current recommended initial treatment for cryptococcal antigenaemia is fluconazole 1200 mg/day for 2 weeks [45••], and there is consistent evidence that these programmes reduce the incidence of CM and related mortality [46].

Despite this, there remains a persistent association between CrAg positivity and death, even among those who receive preemptive fluconazole therapy [47,48,49]. This may be partially because some of these individuals have symptomatic CM that has not been adequately investigated or because of subclinical CM which has been found to occur in 25–40% of asymptomatic individuals with antigenaemia [50,51,52]. However, even for a subset of the remaining patients who do not have CM, it is likely that fluconazole is not adequate to prevent the development of CM. Given that these individuals are often well and are managed as outpatients, there may be a potential role for short-course intravenous or oral combination therapy in preventing the development of CM. Randomised controlled data on this subject are needed. A clinical trial exploring the potential role for a single, high dose of L-AmB is currently underway in Uganda [53], and a trial of oral combination therapy is also planned.

Implementation challenges

Despite the growing evidence to support both the use of flucytosine and liposomal amphotericin in effective short-course and/or oral treatment regimens for the treatment of CM in LMICs, there are significant implementation challenges to overcome before they can be widely used in the regions of the world with the highest burden of HIV-associated CM [54].

Flucytosine was developed in 1957 and is no longer on patent but remains expensive due to monopoly pricing by a limited number of manufacturers [55]. A 2-week treatment course in North America costs roughly 25,000 USD [56]. Flucytosine is also manufactured in Europe (Mylan, France) and costs 1.20 USD per tablet, meaning that a two-week treatment for a patient weighing 60 kg costs 200 USD. An economic model using outcome data from the ACTA trial found that, based on a price of 1.30 USD per tablet, the addition of 2 weeks of flucytosine to fluconazole in settings where fluconazole monotherapy is the mainstay of treatment would cost just 65 USD per life-year saved. Reducing the cost of a single tablet to 0.80 USD and 0.40 USD would see this incremental cost-effectiveness ratio reduce to 44 USD and 28 USD per life-year saved, respectively [57].

Another barrier to the implementation of flucytosine therapy is that until recently, it was not registered in any African countries. This is changing thanks to the efforts of stakeholders in individual countries, the incorporation of flucytosine into national treatment guidelines such as in South Africa [45••], and internationally coordinated initiatives aiming to widen access to flucytosine across countries with a high incidence of CM. In order to keep up with this burgeoning demand, there is also a need to increase manufacturing capacity.

Access to liposomal amphotericin is also impacted by the same economic factors. A single vial costs between 80 USD and 400 USD, and a single high-dose treatment for a patient weighing 60 kg requires 12 vials [58]. In 2018, the manufacturers, Gilead Sciences, announced that vials of AmBisome would become available at cost price as part of an expanded access programme for CM in 116 LMICs [59], as is already the case for visceral leishmaniasis. Dependent upon the results of the AMBITION trial, an economic evaluation will be conducted to determine the cost-effectiveness of giving a single high-dose [58].

Conclusion

The incidence and mortality associated with HIV-associated cryptococcal meningitis is still unacceptably high, despite widened access to ART. This is in part due to the suboptimal effects of fluconazole monotherapy, significant drug-induced toxicities associated with amphotericin B, and limited access to superior treatments. The recently completed ACTA trial has provided evidence to support the use of a short-course amphotericin B and oral combination antifungal treatment regimens for CM, and the ongoing AMBITION trial hopes to usher in a short-course, high-dose treatment with liposomal amphotericin. Significant implementation challenges must be overcome to ensure that these research findings contribute to a reduction in mortality from this devastating disease.