The combination of tamoxifen with amphotericin B, but not with fluconazole, has synergistic activity against the majority of clinical isolates of Cryptococcus neoformans

Summary Background Cryptococcal meningitis has fatality rates of 40%‐70%, resulting in 200 000 deaths each year. The best outcomes are achieved with amphotericin combined with flucytosine but flucytosine is expensive and unavailable where most disease occurs. More effective and affordable treatments are needed. Tamoxifen, a selective oestrogen receptor modulator frequently indicated for breast cancer, has been found to have synergistic activity against the Cryptococcus neoformans type strain when combined with amphotericin or fluconazole. It is cheap, off‐licence, widely available and well‐tolerated, and thus a pragmatic potential treatment for cryptococcal disease. Objectives We wanted to determine the susceptibility of clinical isolates of C. neoformans to tamoxifen alone and in combination with other antifungals, to determine whether there is sufficient evidence of activity to justify a clinical trial. Methods We used the CLSI broth microdilution protocol to test the susceptibility of 30 randomly selected clinical isolates of C. neoformans to tamoxifen, in dual combination with amphotericin, fluconazole or flucytosine, and in triple combination with amphotericin and fluconazole. Evidence of drug interactions was assessed using the fractional inhibitory concentration index. Results The MIC50 and MIC90 of tamoxifen were 4 and 16 mg/L, respectively. The combination of tamoxifen and amphotericin suggested a synergistic interaction in 20 of 30 (67%) isolates. There was no interaction between tamoxifen and either fluconazole or flucytosine. Synergy was maintained in 3‐Dimensional chequerboard testing. There was no evidence of antagonism. Conclusions Tamoxifen may be a useful addition to treatment with amphotericin and fluconazole for cryptococcal meningitis; a trial is justified.


| INTRODUC TI ON
Every year, more than 200 000 people die from cryptococcal meningitis. 1 The vast majority of cases are due to Cryptococcus neoformans and occur in patients who have underlying immunosuppression.
Currently, the most frequent cause of immunosuppression is HIV infection, but iatrogenic causes, including therapy for connective tissue disorders, cancers and solid organ transplantation, are increasingly important as healthcare advances. 2 Cryptococcal meningitis also occurs in immunocompetent patients, where Cryptococcus gattii sensu lato is an important cause, particularly in the tropics, Western Canada and the Pacific Northwest of the USA. 2 In Southeast and East Asia, both C. neoformans and C. gattii sensu lato cause disease in apparently immunocompetent patients. 3 Treatment guidelines for cryptococcal meningitis are based upon the results of a number of randomised controlled trials, largely completed in patients with HIV infection. [4][5][6][7][8] The best clinical outcomes are achieved using induction treatment consisting of amphotericin combined with flucytosine. 6,9 This combination was first shown to deliver improved rates of sterilisation of cerebrospinal fluid (CSF) compared with amphotericin monotherapy in 1997; subsequently, a trial from Vietnam demonstrated that as well as delivering faster rates of clearance of yeast from CSF, it also resulted in better survival, with a 40% reduction in the risk of death by 10 weeks. 4,6 However, even on gold standard therapy, death rates 3 months after diagnosis remain high, at between 15 (USA) and 40% (Asia and Africa). 1 Recently, the key role of flucytosine in delivering the best outcomes has been underlined by the publication of the ACTA trial. 9 This study confirmed the role of amphotericin combined with flucytosine as the treatment of choice and also demonstrated that an oral regimen consisting of flucytosine combined with fluconazole delivered survival rates approaching that of amphotericin combined with flucytosine. However, despite having been off-patent for several decades, and the existence of an energetic and passionate advocacy campaign, there has been little progress in improving access to flucytosine. 10 In fact, prices have risen significantly in recent years. 11 Flucytosine is an unattractive prospect for generic manufacturers because of its limited indications outside cryptococcal meningitis, and because the vast burden of cryptococcal disease is in low-income countries. Both these factors reduce the possibility of generating significant profit, and this conundrum exists for any treatment specific for cryptococcal meningitis. The perfect drug for cryptococcal meningitis would be off-patent, manufactured by a number of different companies, and frequently indicated for some other, common, disease. These circumstances would confer commercial viability driven by high numbers of prescriptions rather than high per tablet profit margins, keep prices affordable through competition and increase the likelihood of licensing and availability where cryptococcal disease occurs. This paradigm has driven interest in the repurposing of generic drugs for neglected tropical and other diseases of poverty.
Tamoxifen, a selective oestrogen reuptake modulator, was first noted to have antifungal activity against Saccharomyces cerevisiae in 1989. 12 Soon afterwards, it was found to have antifungal action against Candida albicans, and against C. neoformans in 2009. 13,14 Tamoxifen has a wide range of effects in mammalian cells including anti-oxidant activity, alteration of cell membrane properties and induction of apoptosis. 14 In Cryptococcus, it has been shown to bind to calmodulin and a calmodulin-like protein, preventing the activation of calcineurin, which is involved in the yeast stress response. 14,15 Furthermore, when tested against the C. neoformans type strain H99, tamoxifen interacted synergistically in combination with both amphotericin and fluconazole, and had a fungicidal effect in combination with fluconazole in the mouse infection model. 15 Tamoxifen has high oral bioavailability, is lipophilic resulting in high brain concentrations, and is concentrated in macrophage phagosomes-a site of growth of C. neoformans. [15][16][17] Therefore, tamoxifen, which is offpatent, widely available and affordable, is a promising treatment to augment current antifungal therapy for cryptococcal meningitis. As a first step, we tested the susceptibility of clinical isolates of C. neoformans from our hospital in Vietnam to tamoxifen, amphotericin B, fluconazole and flucytosine, alone and in combination. Because we occasionally see disease in our patients due to C. gattii sensu lato, we also measured the susceptibility of a small number of isolates and control strains of these species to tamoxifen alone and in combination with amphotericin.

| Fungal isolates
We randomly selected thirty isolates of C. neoformans from the strain collection at our institute. The strain collection consists of isolates derived from the cerebrospinal fluid of patients at the point of diagnosis of cryptococcal meningitis. The patients were all enrolled into randomised controlled trials or prospective descriptive studies of cryptococcal meningitis (N = 299). 3,6 All 30 isolates were subjected to amphotericin and tamoxifen susceptibility testing; because testing revealed that evidence of drug interaction was rare between tamoxifen and either fluconazole or flucytosine, susceptibility testing to these drugs was limited to a subset of 20 isolates.
We also determined the susceptibility of four C. gattii sl isolates, derived from participants from the descriptive study of cryptococcal meningitis to amphotericin and tamoxifen.  for 72 hours. All plates were inspected by two observers, and the MIC estimated as the lowest concentration that completely inhibited visual growth for amphotericin and tamoxifen. For fluconazole and flucytosine, the endpoint was defined as the lowest drug concentration that resulted in an 80% reduction of visual growth compared with that of the drug-free growth control well.

| Chequerboard microdilution assays
Combinations of tamoxifen plus amphotericin, tamoxifen plus fluconazole, tamoxifen plus flucytosine and tamoxifen plus amphotericin plus fluconazole were tested as has been described previously. 19 For two-dimensional microplate preparation, a stock solution of each antifungal drug was made in RPMI at fourfold the final desired test concentrations. To test a combination of drug "A" with drug "B," 50 µL of drug "A" solution at fourfold the desired final concentration was combined with 50 µL of the chosen combination drug "B" at fourfold its desired final concentration to give a volume of 100 µL. One hundred µL of inoculum was added to this to give a final volume of 200 µL (and hence a fourfold dilution of the original stock drug solution). The final drug concentrations tested in combination were as follows: tamoxifen from 0.25 to 64 µg/mL; fluconazole and flucytosine each from 1 to 64 µg/mL, and amphotericin from 0.25 to 16 µg/mL. All plates had a drug-free growth control well and a sterility control well.
For three-dimensional microplate preparation, nine microplates were needed per isolate with each plate containing a fixed concentration of amphotericin in a twofold series, ranging from 0 to 8 µg/ mL. The concentration of tamoxifen ranged from 0.25 to 64 µg/mL on the x-axis, and the concentration of fluconazole varied from 1 to 32 µg/mL on the y-axis. As before, 50 µL of each drug was used at fourfold the desired final concentration, but only 50 µL of inoculum was used per well, at a higher density of 1.0-5.0 × 10 3 CFU/mL, to deliver a final reaction volume of 200 µL. Every plate included a drug-free growth control well and a sterility control well. The MIC of a drug combination was defined as the lowest concentration for which no growth was observed. Results were recorded 72 hours after incubation at 35°C, and all plates were assessed by two observers.

| Statistical analysis
Susceptibility was expressed as the minimum inhibitory concentration (MIC) of the particular drug at which growth was inhibited for 50% (MIC 50 ), and 90% (MIC 90 ) of all isolates, and as the geometric mean inhibitory concentration of each drug. The differences in MICs of amphotericin and fluconazole when incubated with and without tamoxifen were compared using the Wilcoxon rank-sum test. All analyses were done using r software version 3.1.2. 22

| Ethics
The authors confirm that the ethical policies of the journal, as noted on the journal's author guidelines page, have been adhered to. All clinical studies from which the isolates were derived had ethical approval from the Hospital for Tropical Diseases, and either Liverpool School of Tropical Medicine, UK or the Oxford Tropical Ethics Committee, UK. All participants gave informed consent.

| RE SULTS
The drug susceptibility (MIC 50 , MIC 90, ranges and geometric means) by species are summarised in Table 1 C. gattii species complex strains had similar susceptibilities to tamoxifen as C. neoformans, ranging from 2 to 8 µg/mL, with no clear difference by species (Table 1).

| The combination of tamoxifen with amphotericin B, but not with fluconazole or flucytosine, appears synergistic for the majority of clinical isolates of C. neoformans
The results of chequerboard testing are shown in Table 2 Similarly, there was apparent synergy between tamoxifen and amphotericin for six of eight C. gattii isolates (75%, three of four Vietnamese clinical strains and three of four control strains) (Table 3).
Interestingly, the two strains where the drug combination appeared to be only additive were both C. deuterogattii (BMD800 and control strain WM178).
When tamoxifen and amphotericin were combined, there was a statistically significant reduction in the geometric mean MIC of amphotericin B from 1.1 to 0.26 µg/mL (difference between the groups P < 0.001) ( Table 3) for the 30 C. neoformans isolates. While there was little evidence of a synergistic interaction between tamoxifen and fluconazole, there was a decrease in the geometric mean MIC of fluconazole from 9.19 to 2.30 µg/mL when the two drugs were combined (difference between the groups P = 0.005, Table S1). This was also seen for the combination of tamoxifen and flucytosine where the geometric mean MIC of flucytosine fell from 8.9 to 1.2 µg/mL (monotherapy versus combination therapy, difference between the groups P < 0.001, Table S2).
Of isolates that appeared to have reduced susceptibility to fluconazole, four of the nine had at least fourfold increases in susceptibility when combined with tamoxifen. All 14 strains with probable reduced susceptibility to flucytosine had eightfold or greater increases in susceptibility to the drug (to 1 μg/mL) when combined with tamoxifen.

| Synergy is preserved when tamoxifen, amphotericin and fluconazole are combined in triple combination
We tested the three drugs-tamoxifen, amphotericin and fluconazolein combination against four clinical C. neoformans isolates and the H99 type strain. The amphotericin-tamoxifen dual drug combination had appeared synergistic with two of the selected isolates (FICI = 0.5), but to have only additive effect with the other two (0.5 < FICI ≤ 4). In the triple combination testing (Table 4), there appeared to be a synergistic drug interaction for three of the four clinical isolates (FICI < 1), no evidence of antagonism for the remaining clinical isolate (FIC = 1) and a suggestion of antagonism for the H99 type strain (FICI > 1).

| D ISCUSS I ON
Cryptococcal meningitis, even with gold standard therapy, is a devastating illness with mortality rates in the order of 30% 10 weeks after diagnosis. 1,6,9 The best outcomes depend on induction treatment with amphotericin and flucytosine over the first 1-2 weeks. 4,6,9 Unfortunately, flucytosine is usually neither available nor affordable where the major burden of disease occurs; flucytosine-sparing therapies have significantly higher mortalities. 10 There is a clear need to develop treatment that is more effective, tolerable, available and affordable. However, new drugs developed specifically to treat cryptococcal disease are unlikely to fulfil the last two of these criteria in low-income settings.
Here, we have confirmed that tamoxifen, a selective oestrogen receptor modulator used most frequently to treat breast cancer, and which was shown to have an anti-cryptococcal effect against the type strain almost 10 years ago, has in vitro activity against clini- Moreover, tamoxifen is concentrated in lipid-rich tissues including the brain, the site of disease in cryptococcal meningitis. 17 Here, concentrations are 40-to 100-fold higher than in serum and thus likely to safely exceed the MIC in the majority of infections. Tamoxifen is also concentrated within macrophages, an important site of replication of C. neoformans. 15 Therefore, it is likely that oral administration of tamoxifen can achieve sufficient concentrations of tamoxifen at the site of disease to have therapeutic effect.
Furthermore, we confirmed that tamoxifen appears to have a synergistic effect when combined with amphotericin, as has been reported previously for the H99 type strain. 15 This was seen in the majority (67%) of our clinical isolates (and is similar to reported rates of synergy with amphotericin and flucytosine in clinical isolates 33 ).
Of note, we did not see evidence of synergy with C. deuterogattii, although only two strains were tested. However, in contrast with the previous report, we found the interaction between tamoxifen and fluconazole to be simply additive for almost all (95%) of our isolates.
Importantly, there was no evidence of antagonism between tamoxifen and any of amphotericin, fluconazole or flucytosine. When we combined the three drugs, we found that a synergistic interaction was preserved in strains where it had been seen for the amphotericin-tamoxifen dual combination. In addition, combining the three drugs also seemed to deliver a synergistic interaction in one of the clinical strains where synergy had not been apparent for the dual drug combination. Therefore, a triple drug combination has the potential to be synergistic in at least two-thirds of Vietnamese cases.
However, we also found that the combination of amphotericin, fluconazole and tamoxifen appeared antagonistic for the H99 type strain and were unable to replicate previously published data that tamoxifen was synergistic when combined with either amphotericin or tamoxifen for the type strain. The reasons for this are not clear.
However, microevolution and divergence of the H99 type strain in different laboratories are well-documented. 34 Synergy is an attractive property for an anti-microbial treatment, promising non-linear gains in efficacy, potentially with little additional side effect cost. There has been recent interest in the use of sertraline to treat cryptococcal meningitis. Sertraline also appears to work synergistically with either amphotericin or fluconazole against C. neoformans in vitro. 35,36 A (historically) controlled phase 2 trial of this drug in combination with amphotericin and fluconazole appeared promising, suggesting improved rates of clearance of yeast from cerebrospinal fluid, although this contrasted with a smaller controlled trial which found no difference. 37,38 Disappointingly, a recent phase 3 randomised placebo-controlled trial of sertraline boosted therapy, powered to survival, was stopped early due to futility after enrolling 460 patients. 39 No differences were found between the sertraline or placebo arms in patient survival or yeast clearance rates from cerebrospinal fluid.
Despite the fact that we can discern a range of susceptibilities to tamoxifen in vitro, there is no clear relationship between antifungal susceptibility in vitro and response to treatment in human cryptococcal disease-clinical breakpoints are not defined and outside of relapsed disease the value of susceptibility testing is unclear. 7,40 However, the rate of clearance of yeast from cerebrospinal fluidearly fungicidal activity (EFA)-does appear to correlate with survival in cryptococcal meningitis at the population level. 5,6,9 Therefore, as a next step, tamoxifen should be tested in a small randomised controlled trial, to generate initial efficacy (EFA) and safety data. A dose in the order of 300 mg/d would be needed to attain serum levels around the MIC 90 . While such a dose is likely to be less tolerable than that used in breast cancer, it must be considered in the context of the poor prognosis of cryptococcal meningitis, which compares unfavourably with glioblastoma, prostate and small cell lung cancer, where large doses of tamoxifen have been used. 41 Recognised side effects include thromboembolic disease and retinopathy (although the risk of these is low with short duration treatment) and QT prolongation of the cardiac cycle. 42 QT prolongation might be expected to be a particular risk as doses are increased, especially when combined with fluconazole, which can also cause this conduction abnormality, and disordered serum electrolytes, frequently seen in cryptococcal disease as a consequence of amphotericin therapy and intracranial pathology. [43][44][45] Any study would need to pay close attention to these risks, and a trial is underway in Vietnam (Clinicaltrials.gov NCT03112031).

This work was supported by a Wellcome Trust Intermediate
Fellowship grant number WT097147MA to JND.

CO N FLI C T S O F I NTE R E S T
None to declare. The funders had no role in study design, analysis or decision to publish.