Neurocryptococcosis is a subacute meningoencephalitis caused by the inhalation of the fungus Cryptococcus spp. After a latency period in the pulmonary lymph nodes, it spreads throughout the body with tropism for the central nervous system (Coelho et al., 2014). Neurocryptococcosis is more prevalent in HIV and other immunosuppressed patients and less commonly in individuals considered immunocompetent (Jarvis et al., 2010; Durski et al., 2013; Pyrgos et al., 2013; Rolfes et al., 2015; Williamson et al., 2016).

The treatment for neurocryptococcosis aims to sterilize the central nervous system and reduce intracranial pressure to values below 200 mmH₂O considering that pressures above 350 mmH₂O are associated with papilledema, decreased visual acuity, decreased hearing capacity, headaches, and confusion. Symptoms resulting from increased cerebrospinal fluid (CSF) pressure can be controlled by lumbar punctures to reduce pressure to levels below 200 mmH₂O (Redmond et al., 2007).

Amphotericin B (AmB; C₄₇H₇₃NO₁₇) is a polyene derived from Streptomyces nodosus, a compound discovered in the 1950s that remains the first line of treatment for invasive fungal infections, although new triazole antifungal drugs with a broad spectrum of action and good distribution to the central nervous system, such as voriconazole and posaconazole are also available (Black and Baden, 2007; Lestner et al., 2010).

AmB is mainly eliminated unchanged via biliary secretion. There are no known AmB metabolites described either in preclinical or clinical studies. Tissue accumulation accounts for most of the drug’s disposition. AmB can still be detected in the liver, spleen, and kidneys for up to 1 year after the end of therapy (Benson and Nahata, 1988). Based on experimental studies in mice, AmB is probably a substrate of P-glycoprotein (P-gp) (Wu et al., 2015) and breast cancer resistance protein (BCRP) (Stevens et al., 2015) drug transporters expressed in the brain-blood barrier (Cordon-Cardo et al., 1989; Schinkel et al., 1996).

The integrity of the brain-blood barrier and the composition of the CSF may change due to infectious and/or inflammatory diseases such as neurocryptococcosis (Shawahna, 2015) due to the high levels of pro-inflammatory cytokines and other inflammatory mediators (Williamson et al., 2016). In vitro studies indicate that the permeability of the brain-blood barrier for AmB is altered by TNF-α and lipopolysaccharide (Pyrgos et al., 2010). Considering that AmB is a Class 4 compound according to the Biopharmaceutics Drug Disposition Classification System (Shugarts and Benet, 2009; Hosey et al., 2016) and that central nervous system efflux and biliary secretion drug transporters play a significant role in the pharmacokinetics of this drug, adjustments to AmB dose may be necessary for neurocryptococcosis patients over the treatment evolution.

The first method for quantifying AmB in plasma dates from the late 1970s and reports the use of high-performance liquid chromatography (HPLC) with detection by UV or fluorescence (Nilsson-Ehle, 1977). Other methods applying HPLC or UHPLC (ultra-high-performance liquid chromatography) with detection by UV or fluorescence continue to be used over the years to date (Liu et al., 1995; Campanero et al., 1997; Echevarría et al., 1998; Groll et al., 2000; Lee et al., 2001; Wurthwein et al., 2005; Qu et al., 2017; Wirth et al., 2018).

The application of mass spectrometry (LC-MS/MS) to quantify was described for the first time by Lee et al. (Lee et al., 2001) in the assessment of four diffAmB erent biological matrices using 50 μl of plasma, urine, and human fecal homogenate with lowest limits of quantification (LLOQ) of 2000, 50 and 40 ng ml⁻¹, respectively, and 200 μl of human plasma ultrafiltrate with an LLOQ of 1 ng ml⁻¹. Other previous methods that apply LC-MS/MS (Lee et al., 2001; Deshpande et al., 2010; Strenger et al., 2014) have used larger volumes of biological matrices. In addition, these cited studies used the solid-phase extraction technique in the sample preparation, which makes the analysis of a large number of samples more expensive, a common scenario in clinical pharmacokinetics studies. LC-MS/MS was first used to analyze human CSF samples by Xiong et al. (2009), with 100 μl samples and an LLOQ of 0.5 ng ml⁻¹.

The present study aimed to develop and validate LC-MS/MS methods for quantifying AmB total concentrations in plasma, urine, and CSF, as well as the unbound concentration in plasma using plasma ultrafiltrate. The investigation of the pharmacokinetics of AmB in patients with neurocryptococcosis requires the development and validation of methods with sensitivity compatible with the low concentrations of AmB reported in the CSF, taking into consideration the low volumes available for this matrix (Leenders et al., 1997; Wurthwein et al., 2005; Xiong et al., 2009; Hamill et al., 2010; Strenger et al., 2014; Stone et al., 2016). The methods described in the present study were successfully applied to quantify AmB in all available biological fluids including CSF samples from one patient with neurocryptococcosis up to the 16th day of treatment with AmB lipid complex.

The initial experiments of AmB analysis in plasma and urine were performed using the Micromass Quattro Micro™ triple quadrupole mass spectrometer (Waters Corporation, Milford, MA, United States). However, the equipment presented a lower sensitivity considering the expected AmB concentrations in biological samples. Thus, the methods were development using the system XEVO TQ-S^® triple quadrupole mass spectrometer (Waters Corp., Milford, MA, United States) previously described. Figure 2 shows the protonated ([M + H]⁺) precursor ions and their respective product ions. AmB was monitored in the transition m/z 906 → 743. The IS piroxicam was chosen in the absence of a commercially available deuterated standard for AmB. Piroxicam was a suitable IS for a sample preparation method applying protein precipitation and presented retention times similar to AmB (differences of 0.10–0.19 min depending on the biological matrix, Figure 3). The sequential analysis in plasma, plasma ultrafiltrate, urine, and CSF are presented in Figure 3. The methods for quantifying of AmB in the four biological matrices were applied to the investigation of the pharmacokinetics of AmB in a male patient treated for neurocryptococcosis. The patient’s anthropometric, biochemical, and hematological parameters, as well as a complete list of medications in use by the time of phases 1 and 2, are presented in Table 1. The validation parameters for the methods of quantification of AmB in plasma, plasma ultrafiltrate, urine, and CSF can be found in Table 2. All validation parameters relative errors were within the range of ±15% and coefficients of variation below 1%. The pharmacokinetic profiles of AmB in each phase of the study are shown in Figure 4 for both plasma and urine. The unbound concentrations of AmB in plasma were determined in the last five samples comprising the terminal elimination phase and are presented in Figure 4 and Table 3, with mean values of 3.25 (±0.22) % in phase 1 and 2.99 (±0.27) % in phase 2. Two CSF samples were obtained in the time between phases 1 and 2. The concentration of AmB in these samples and the relationships between the concentration in CSF and plasma as total and unbound concentrations collected simultaneously are described in Table 4. The ratios of total plasma concentration by CSF concentration ratios were 30.90 and 19.53, respectively at the 8th and 15th days of treatment with AmB. In contrast, the ratios of unbound plasma concentration by CSF concentration were 0.77 and 0.44, respectively at the 8th and 15th days, demonstrating the increasing concentrations of AmB in the CSF throughout the treatment. An overview of AmB concentrations determined in plasma as total and unbound concentrations and in CSF are shown in a comprehensive timeline in Figure 5. Table 5 presents the pharmacokinetic parameters of AmB, the amount excreted into the urine in 24 h, and the renal and hepatic clearances. All datasets generated for this study are included in the article’s Supplementary Material.

The present study aimed to develop a method capable of quantifying AmB in CSF at any given time of the treatment in addition to plasma, plasma ultrafiltrate, and urine. The quantification of AmB in CSF has been a methodological challenge from the last decades to the present day, with studies reporting a large number of CSF samples in which AmB concentrations are undetectable or below the LLOQ in a fraction of or all samples, despite the observation of clinical efficacy (Leenders et al., 1997; Wurthwein et al., 2005; Hamill et al., 2010; Strenger et al., 2014). Thus, the method developed for quantification of AmB in human CSF combines the convenience of a smaller sample volume (100 μl) and lower LLOQ (0.1 ng ml⁻¹) compared to those reported in the literature, with sample volumes of 1,000 μl (Liu et al., 1995), 500 μl (Strenger et al., 2014), and 250 μl (Wirth et al., 2018) and LLOQ’s of 500 ng ml⁻¹ (Strenger et al., 2014), 1 ng ml⁻¹ (Liu et al., 1995), and 0.5 ng ml⁻¹ (Xiong et al., 2009).

The development of the chromatographic method was initially carried out following the method described by Su et al. (2018), which uses an Ascentis C18 reverse-phase column, 5 μm particles, 50 × 4.6 mm and applies a gradient of 0.2% formic acid and 5 mM ammonium acetate in water (solution A) and 0.2% formic acid in acetonitrile (solution B). During the parameters optimization step with an infusion of AmB, it was observed in our experiments that only 0.2% formic acid contributed to the increase in the signal of the analyte. The initial tests were performed using the reversed-phases columns Purospher RP-18e (123 × 3 mm), RP-Select B (125 × 4.6 mm), and RP-Select B (250 × 4.6 mm), all with 5 µm particles maintained at 40°C and protected by a guard column of the same specification. The columns were tested in different mobile phase combinations containing 0.1 and 0.2% formic acid in both water and acetonitrile in the proportions 70:30, 60:40, 50:50, 40:60 and 30:70 (v/v). The best chromatographic profile for the analysis of AmB in plasma and urine samples was observed in the RP-Select B reverse phase column (125 × 4.6 mm) with a mobile phase consisting of a mixture of 0.1% formic acid in water and 0.1% formic acid in acetonitrile in a 40:60 (v/v) ratio. However, analysis of AmB in CSF and plasma ultrafiltrate samples required changing the proportion of mobile phase constituents (50:50, v/v) to obtain better shaped and more symmetrical chromatographic peaks (Figure 3). Initial tests were performed with the column maintained at 40°C (Su et al., 2018), but considering that at 25°C the chromatographic profile remains constant, the lowest temperature was chosen.

Considering the absence of a commercially available deuterated standard for AmB, the search for an IS was based on previously published methods. Piroxicam, used in previous studies with spectrophotometric detection, proved to also be adequate for the protein precipitation process and mass spectrometry detection used in the present study (Campanero et al., 1997; Echevarría et al., 1998; Hope et al., 2012). It is noteworthy that piroxicam was monitored at transition m/z 332 → 95 in the analysis in plasma and urine, and at transition m/z 332 → 121 in the ana analysis in plasma ultrafiltrate and CSF to avoid interference with matrices components (Figure 3).

The methods for the chromatographic analysis of AmB in biological matrices use the solid-phase extraction technique, which makes the analysis of a large number of samples more expensive (Liu et al., 1995; Lee et al., 2001; Bekersky et al., 2002b; Bellmann et al., 2003; Su et al., 2018; Van Daele et al., 2021). Sample preparation was initially tested with the intention of replacing solid-phase extraction by protein precipitation, aiming for lower costs of analysis and shorter times of sample preparations, key aspects of pharmacokinetic studies in which large numbers of samples are analyzed using sensitive methods. The best result was obtained by precipitating plasma, plasma ultrafiltrate, urine, and CSF samples with a 0.1% formic acid solution in acetonitrile, followed by dilution of the respective supernatants with a 0.1% formic acid solution in water.

The preparation of plasma, plasma ultrafiltrate, urine, and CSF samples with protein precipitation followed by dilution of the supernatants resulted in the absence of any significant matrix effect, evaluated as the observation of the coefficients of variation of matrix factor values normalized by the IS less than 10% (Table 2).

The linearity of the methods for the analysis of the concentration of AmB in plasma (5–1,000 ng ml⁻¹), plasma ultrafiltrate (0.6–250 ng ml⁻¹), urine (5–1,000 ng ml⁻¹), and in CSF (0.1–250 ng ml⁻¹, Table 2) proved to be adequate for the application in clinical pharmacokinetics studies, with determination coefficients (r ²) greater than 0.98 for plasma ultrafiltrate and 0.99 for the other matrices. The wide linear range will enable the quantification of AmB in biological fluids from patients treated with different AmB formulations, such as deoxycholate or lipid formulations (Ayestarán et al., 1996). The upper limit of quantification (ULOQ) values of 1,000 ng ml⁻¹of plasma or urine and 250 ng ml⁻¹ of plasma ultrafiltrate and CSF were the highest within the linear ranges for which no carryover effect was observed. The methods developed and validated in this study using aliquots of 25 μl of human plasma or urine and 100 μl of human plasma ultrafiltrate and CSF, with LLOQ values of 5 ng ml⁻¹ for plasma or urine, 0.6 ng ml⁻¹ for plasma ultrafiltrate, and 0.1 ng ml⁻¹ for CSF can be considered the most sensitive ones described so far. Overall, the methods described here are 3.2–1,500 times more sensitive than the LC-MS/MS methods previously described in the literature (Lee et al., 2001; Bekersky et al., 2002a; Xiong et al., 2009; Deshpande et al., 2010; Qin et al., 2012; Al-Quadeib et al., 2014; Strenger et al., 2014; Su et al., 2018; Van Daele et al., 2021). Despite reaching the highest sensitivity in the literature, the methods could be further adjusted to use even lower sample volumes by adding a final concentration step, in which the supernatants are evaporated and reconstituted in lower volumes. Even though undesirable for clinical pharmacokinetic studies, since it adds a new time-consuming step, this modification would allow for the methods to be applied to animal models, such as mice.

The methods of analysis of total concentrations of AmB in plasma, urine, and CSF and the unbound concentration in plasma proved to be precise and accurate, with coefficients of variation and relative error values below 12% (Table 2). The freeze and thaw, short-term temperature, and post-preparative stabilities studies showed coefficients of variation and relative error values equal to or less than 12% when quantified with freshly prepared calibration curves (Table 2).

AmB is a highly lipophilic drug administered parenterally due to its low oral absorption. The administration of AmB in lipid formulation results in faster AmB accumulation in peripheral tissues, in lower plasma concentrations, and minimizes adverse reactions compared to the deoxycholate formulation (Ayestarán et al., 1996). The methods validated in the present study were applied to investigate the pharmacokinetics of AmB following infusions of 100 mg every 24 h for 16 days administered as a lipid complex throughout the treatment of a neurocryptococcosis patient.

All collected biological samples (Figure 1) were quantified considering the high sensitivity of the method. It is important to highlight that AmB was successfully quantified in CSF samples collected on the 8th and 15th days of the treatment (Figure 5).

Clinical pharmacokinetics of AmB lipid complex was previously reported in healthy volunteers (Kan et al., 1991) and in patients with systemic fungal infections (Adedoyin et al., 2000). The AmB pharmacokinetic parameters such as total clearance and volume of distribution at steady state are dose-dependent considering that the drug accumulates in the tissues. The cited authors observed total clearance values in healthy volunteers ranging from 0.07 to 0.09 L h⁻¹ kg⁻¹, values similar to those presented in Table 5 for the patient with neurocryptococcosis (0.133 L h⁻¹ kg⁻¹ on the 8th day to 0.076 L h⁻¹ kg⁻¹ on the 15th day). However, the administration of AmB lipid complex at 5 mg kg⁻¹ day⁻¹ to patients with fungal infections resulted in higher total clearance values (17.8 ± 5.2 L h⁻¹) compared to the values presented in Table 5 for the investigated patient with neurocryptococcosis treated with daily infusions of 100 mg (7.21 L h⁻¹ on the 8th day to 4.25 L h⁻¹ on the 15th day). However, when the different studies are dose-normalized, all the values are equivalent. Considering that in the present study AmB was also quantified in urine, it was possible to verify that total clearance values are very similar to hepatic clearance (0.077 L h⁻¹ kg⁻¹ on the 8th day and 0.047 L h⁻¹ kg⁻¹ on the 15th day) due to its elimination mainly by biliary secretion; Table 5.

AmB lipid complex is highly distributed. Volumes of distribution at steady state were also evaluated in healthy volunteers (1.7–3.9 L kg⁻¹) (Kan et al., 1991) and in patients with systemic fungal infections (865 ± 347 L) (Adedoyin et al., 2000). Volumes of distribution at steady state for the investigated patient were 5.59 (8th day) to 3.67 L kg⁻¹ (15th day) (Table 5); values close to the cited studies when dose-normalized.

AmB unbound fraction following the lipid formulation was evaluated in the investigated neurocryptococcosis patient. The data reported in Tables 3 and 4 show AmB unbound fraction values of 3.25 (4th day), 2.50 (8th day), 2.26 (15th day), and 2.99 (16th day). Although there are no reports in the literature about AmB unbound fraction following the lipid formulation, the administration of other formulations such as liposomal or deoxycholate also result in unbound fraction values of 4.5 and 20.6%, respectively (Bekersky et al., 2002b).

In the present study, total plasma concentration/CSF concentration ratios ranged from 30.90 (8th day) to 19.53 (15th day), whereas unbound plasma concentration/CSF concentration ratio ranged from 0.77 (8th day) to 0.44 (15th day), showing increasing concentrations of AmB in the CSF with the evolution of the treatment. Strenger et al. (2014) reported total plasma concentration/CSF concentration ratios ranging from 124 to 2,391 in haemato-oncological pediatric patients treated with liposomal AmB (3 mg kg⁻¹ day⁻¹) in samples evaluated from 0 to 48 h after drug infusion. However, total plasma concentration/CSF concentration and unbound plasma concentration/CSF concentration ratios were not found in the available literature following the administration of lipid formulation.

In conclusion, the present study described the most complete and sensitive methods for AmB analysis in plasma, plasma ultrafiltrate, urine, and CSF applied to a clinical pharmacokinetic study following the administration of the drug as a lipid complex in one patient with neurocryptococcosis. Previous clinical studies failed to fully assess all biological matrices due to low sensitivity and higher matrices volumes, especially for CSF. The method can be applied to investigate the pharmacokinetics of AmB in CSF at any given time of the treatment.