Impact of pH on the activity of novel cephalosporin cefiderocol in human urine

Abstract

Background

Antimicrobial activity of antibiotics can be impacted by pH, enhancing or reducing their bactericidal properties. Cefiderocol, a novel cephalosporin antibiotic that is among others indicated for the treatment of complicated urinary tract infections (cUTIs), lacks data on activity in urine.

Methods

Pooled human urine (iron levels ∼0.05 mg/L/24 h), CAMHB and iron-depleted CAMHB (ID-CAMHB) at pH 5, 7 and 8 served as media. MIC testing was done according to EUCAST with the broth microdilution method for 17 clinical isolates of Escherichia coli and ATCC 25922 (including isolates with ESBL activity), 17 clinical isolates of Klebsiella pneumoniae and ATCC 700603 (also with ESBL), and 6 clinical isolates of Pseudomonas aeruginosa and ATCC 27853. Time–kill curves (TKCs) were performed for selected strains at pH 5, 7 and 8 in urine.

Results

MIC values in urine, CAMHB and ID-CAMHB exhibited isolate-specific variations when assessed under identical pH conditions, ranging from a 1-fold dilution to changes of up to 4-fold dilutions in either direction. Median MICs of cefiderocol were up to 50-fold higher in pH 5 than in pH 7 for P. aeruginosa isolates and 32-fold higher in E. coli and K. pneumoniae isolates. TKCs with 650 and 1300 mg/L cefiderocol in urine showed that ATCC strains were efficiently eradicated despite the pH set.

Conclusions

Acidic pH had a significant negative impact on cefiderocol activity. Yet, after a recommended IV administration of 2 g cefiderocol every 8 h, a concentration of approximately 1300 mg/L can be achieved in urine, suggesting that efficient killing of all tested pathogens could have been possible even under acidic conditions in vivo.

Introduction

Cefiderocol is a novel injectable siderophore cephalosporin that like other β-lactam antibiotics inhibits bacterial cell-wall synthesis by targeting PBPs. The novelty of this antibiotic is that cefiderocol uptake differs from other β-lactams by binding to ferric iron via its catechol moiety, forming a chelating complex, which allows cefiderocol to be actively transported into the periplasmic space through siderophore uptake systems.¹ Cefiderocol received US FDA approval in October 2019 for the treatment of complicated urinary tract infections (cUTIs).² In 2020, authorization for hospital-acquired pneumonia and ventilator-associated bacterial pneumonia followed in the USA. The authorization in the EU for treatment of infections caused by aerobic Gram-negative bacteria when other treatment options seem unlikely to succeed followed as well in April 2020.¹

Since then, studies have evaluated the pharmacological and clinical profile of cefiderocol against Gram-negative pathogens and encourage the use of cefiderocol against bacteria that are associated with cUTI, such as Enterobacteriaceae and Pseudomonas aeruginosa.^3–8

Even though cefiderocol is excreted to an high extent unchanged in urine (60% to 90% unchanged drug, ∼1300 mg/L), achieving concentrations far above the EUACST breakpoint for systemic infections of >2 mg/L for Enterobacterales and P. aeruginosa, it should be considered that the pH of human urine can substantially vary from acidic (pH 4.5) to alkaline (pH 8) conditions.^1,6,7,9 Studies have already proven that pH can impact antibiotic activity of different antibiotics enhancing or reducing their bactericidal properties.^10–13

Thus, to better link in vitro experiments to in vivo conditions, we set out to evaluate the impact of pH 5, 7 and 8 on cefiderocol activity against clinical isolates of Escherichia coli, Klebsiella pneumoniae and P. aeruginosa at pH 5–8, in pooled human urine (iron levels ∼0.05 mg/L/24 h), CAMHB and iron-depleted CAMHB (ID-CAMHB).

Materials and methods

Bacterial strains

The 17 uropathogenic clinical isolates of E. coli, 17 uropathogenic clinical isolates of K. pneumoniae and 6 uropathogenic clinical isolates of P. aeruginosa [including isolates with ESBL activity: E. coli (n = 11), K. pneumoniae (n = 4)] reported in Table S2 (available as Supplementary data at JAC Online) were provided by the Department of Microbiology of the General Hospital in Vienna. ESBL strains were confirmed by a double-disc synergy test.

As quality control, reference strains from ATCC were used: E. coli ATCC 25922, K. pneumoniae ATCC 700603 and P. aeruginosa ATCC 27853.

Antibiotics

For broth microdilutions and time–kill curves (TKCs), cefiderocol from Shionogi B.V., London, UK (1 g sterile powder, sodium drug product) dissolved in NaCl (0.9%) as described in the manufacturer’s sheet, was used.

Growth media

Urine: midstream urine was obtained from young, male, healthy volunteers, frozen, thawed, pooled, sterile filtered (2 µm) and frozen again until usage to guarantee the usage of one single pooled batch of urine for all tests. The iron content of pooled urine, as determined in the central laboratory of the General Hospital of Vienna, was ∼ 0.05 mg/L/24 h.

CAMHB: CAMHB (Sigma–Aldrich, Steinheim, Germany) containing 17.5 g/L casein acid hydrolysate, 3 g/L beef extract and 1.5 g/L starch, 20–25 mg/L calcium, 10–12.5 mg/L magnesium and iron content ∼0.05 mg/L served as an additional reference medium for MIC testing as iron concentrations are similar to the physiological concentrations found in urine.

ID-CAMHB: the broth was produced as recommended by the EUCAST guidance document on cefiderocol BMD (latest version, December 2020) and as described in detail by Hackel et al.¹⁴ In short, ID-CAMHB was prepared by adding 100 g of Chelex^® 100 resin (Bio-Rad Laboratories, Hercules, CA, USA) to 1 L of freshly prepared CAMHB. The suspension was stirred for 2 h at room temperature (22°C), filtered using a 0.2 μm filter and cation adjusted as recommended.

The pH was set for all tested media (pooled urine, CAMHB and ID-CAMHB) with HCl or NaOH directly before the start of the experiments to values of 5, 7 and 8.

Broth microdilution

MIC testing was conducted in accordance with the performance standards for antimicrobial susceptibility testing (AST) established by EUCAST using the broth microdilution method. MIC determinations were performed for a comprehensive set of strains, including ATCC reference strains (E. coli ATCC 25922, K. pneumoniae ATCC 700603 and P. aeruginosa ATCC 27853) as well as clinical isolates of E. coli, K. pneumoniae and P. aeruginosa, with multiple determinations (ranging from at least 3 to 18 replicates per strain).

Table 1 provides an overview of the strains tested, the media used and the set pH value. MIC determinations were carried out for all strains in reference media ID-CAMHB, CAMHB and pooled urine, all adjusted to pH 7. Furthermore, the ATCC reference strains and three clinical isolates each of E. coli, K. pneumoniae and P. aeruginosa were additionally tested at pH 5, 7 and 8 in the reference medium ID-CAMHB.

Table 1.

This table summarizes in which media at which pH setting the specific strains were tested

Strains	pH in:
	ID-CAMHB	CAMHB	Pooled urine
ATCC strains	5, 7, 8	5, 7, 8	5, 7, 8
Isolates 1–3	5, 7, 8	5, 7, 8	5, 7, 8
All isolates	7	5, 7, 8	5, 7, 8

ATCC strains include E. coli ATCC 25922, K. pneumoniae ATCC 700603 and P. aeruginosa ATCC 27853. Isolates 1–3 include clinical isolates 1–3 of E. coli, clinical isolates 1–3 of K. pneumoniae and clinical isolates 1–3 of P. aeruginosa. All isolates includes all reference strains and all clinical isolates of E. coli, K. pneumoniae and P. aeruginosa.

To simulate physiologically relevant iron concentrations in MIC testing for subsequent TKCs in pooled urine, all strains were also tested in CAMHB and pooled urine at pH 5, 7 and 8.

TKCs

TKCs at pH 5, 7 and 8 in pooled human urine were done with the aforementioned ATCC strains in triplicate, challenged with the concentrations 1×MIC of the respective strain, 650 mg/L and 1300 mg/L, which resemble concentrations achieved in patients with normal renal function, after an IV administration of 2 g every 8 h. In analogy, growth controls (GCs) without antibiotic were conducted in duplicate. All TKC analyses and GCs were performed over 24 h in a shaking water bath (amplitude 22 mm, 150 amplitudes/min) at 37°C under aerobic conditions. The bacterial suspension was adjusted to 1.5 × 108 cells/mL in NaCl (0.9%), corresponding to a McFarland standard of 0.5 and was added to the test tubes at a final concentration of 1.5 × 106 cells/mL.

Samples of 100 µL were taken at timepoint 0 (before the addition of antibiotics) and then at 3, 7 and 24 h and pipetted in the first row of the 96-well microtitre plate. Subsequently, seven serial dilution steps with a volume of 20 µL were carried out in the 96-well microtitre plates filled with 180 µL of NaCl (0.9%) in rows two to seven. Aliquots of 20 µL of each concentration were dropped onto Columbia blood agar plates and incubated at 37°C under aerobic conditions for 24 h. After incubation of the agar plates, cfu were counted and the cfu/mL values were calculated by taking the dilution steps into consideration. This was done using the following equation: number of cfu multiplied by 5 × 10ⁿ, with ‘n’ representing the dilution number.

Statistical analysis

The statistical analysis was done in IBM SPSS 24.04 Win 7/8/10. The MIC results were tested with the Wilcoxon signed rank test to evaluate differences between acidic and neutral pH.

Ethics

In accordance with guidelines of the department for non-invasive waste samples, no EC approval was required for collection of urine.

Results

MICs

The median MIC values for ATCC strains and clinical isolates in ID-CAMHB at all three pH values were compared with MIC values in CAMHB, which are shown in Tables S1, S2 and S3. Between ID-CAMHB and CAMHB, MIC values varied for some only slightly, e.g. for E. coli ATCC 25922 and the respective isolates, the median ratio of CAMHB/ID-CAMHB at pH 5, 7 and 8 was 0.375, 1 and 1, respectively. For other isolates also, an MIC ratio of 4 was calculated, e.g. for P. aeruginosa isolate #3, the median ratio of CAMHB/ID-CAMHB at pH 5, 7 and 8 was 2, 1 and 4, respectively. Values above 1 indicate higher MICs of cefiderocol in CAMHB compared with ID-CAMHB, and values below 1 lower MICs. There was no clear trend either for the media or for the strains to show elevated or lowered MICs. Thus, these results indicate an isolate-specific variation.

In Table 2, the median MIC values obtained in CAMHB and in urine are shown. Variations in MIC values between urine or CAMHB were only found sporadically with maximum deviation of a 1-fold dilution up or down. The median MIC values of cefiderocol were up to 50-fold higher in pH 5 than in pH 7 for P. aeruginosa isolates (rising from 0.15 to 8 mg/L). For E. coli and K. pneumoniae isolates, MICs were up to 32-fold higher (rising from 0.125 to 4 mg/L and from 0.5 to 16 mg/L, respectively). On the contrary, for pH 8, no significant change in the MIC could be seen for all strains compared with values obtained at pH 7.

Table 2.

Median MIC values of all isolates and median MIC values of the ATCC strains for pH 5, 7 and 8 in CAMHB and in urine

	CAMHB					Urine
	Median MIC (mg/L)			Ratio of median MIC		Median MIC (mg/L)			Ratio of median MIC
Isolates	pH 5	pH 7	pH 8	pH 5/pH 7	pH 8/pH 7	pH 5	pH 7	pH 8	pH 5/pH 7	pH 8/pH 7
P. aeruginosa
Median MIC of 6 isolates	8 (2–16)	0.15 (0.03–1)	0.14** (0.015–0.75)	53*	0.93	7** (4–32)	0.14 (0.03–1)	0.08** (0.015–0.5)	50*	0.57
ATCC 27853	8	0.25	0.25	32*	1	8	0.25	0.125	32*	0.5
E. coli
Median MIC of 17 isolates	4 (0.25–32)	0.125 (0.03–1)	0.19** (0.03–1)	32*	1.25	4 (0.25–32)	0.125 (0.03–1)	0.09** (0.025–1)	32*	0.72
ATCC 25922	1	0.06	0.125	8*	2.08	1.5**	0.125	0.06	12*	0.48
K. pneumoniae
Median MIC of 17 isolates	16 (1–64)	0.5 (0.025–4)	0.31** (0.03–4)	32*	0.62	16 (1–64)	0.5 (0.03–4)	0.31** (0.03–2)	32*	0.62
ATCC 700603	8	0.25	0.125	32*	0.5	8	0.5	0.25	32*	0.5

Numbers in brackets resemble the range of MIC values. Additionally, the ratio between the median MIC values of pH 5/pH 7 and pH 8/pH 7 have been calculated. * indicates P < 0.005 (Wilcoxon signed rank test); ** indicates odd dilutions due to calculation (the median was formed from average single MIC values of the repeated MIC determinations).

TKCs

A compilation of the conducted TKCs with E. coli ATCC 25922, K. pneumoniae ATCC 700603 and P. aeruginosa ATCC 27853 is given in Figure 1, each panel representing the TKC of the respective bacterial strain (a, b and c) at pH 5, 7 and 8.

Figure 1.

Counts (cfu/mL) over 24 h with cefiderocol of E. coli ATCC 25922 (a), K. pneumoniae ATCC 700603 (b) and P. aeruginosa ATCC 27853 (c) are shown on the left at pH 5 (red), at pH 7 (black) and at pH 8 (blue). Filled symbols represent the tested antibiotic concentrations and open star symbols the GC. All values are represented with SDs (partly overlaid by symbols). This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Within TKC experiments, growth and bacterial eradication were evaluated up to 24 h. Growth of E. coli ATCC 25922 and K. pneumoniae ATCC 700603 were not impacted by acidic pH. On the contrary, P. aeruginosa ATCC 27853 showed stagnated growth up to 8 h in the acidic environment but reached a cfu count of 1.5 × 108 cfu/mL after 24 h, which was comparable to growth in pH 7.

Impact on bacterial killing at a pH set to 5 was most prominently seen with reference strain E. coli ATCC 25922 at a concentration of 1×MIC as only an inhibition of growth could be achieved leading to a cfu count of 1.5 × 107 cfu/mL after 24 h. In media with pH 7 and 8 a reduction of up to 3 log10 cfu/mL could be reached after 7 h and after 24 h bacteria were either completely eradicated in the pH 7 setting or at least reduced to a cfu count of 1.5 × 102 cfu/mL at pH 8.

In the acidic environment, the K. pneumoniae count of the reference strain stayed constantly at 1.5 × 106 cfu/mL when challenged with a concentration of 1 × MIC. Similarly to E. coli, up to 7 h in the pH 7 and 8 setting, a 3 log₁₀ reduction in bacterial count could be detected. At pH 7, regrowth to the initial inoculum of 1.5 × 10⁶ cfu/mL was observed after 24 h, in contrast to pH 8, where the bacterial count decreased to 1.5 × 10² cfu/mL. For P. aeruginosa, the concentration of 1 × MIC was least effective in the acidic environment up to the 7 h timepoint, yet no difference in bacterial killing was observed after 24 h between the three pH settings.

Even though acidic pH clearly impacts cefiderocol activity at concentrations close to the MIC, TKC results of all ATCC strains at concentrations of 650 and 1300 mg/L could efficiently eradicate the bacteria already after 3 h despite the pH set in urine.

Discussion

Within the last years several studies have shown that pH can impact antibiotic activity of different antibiotics by enhancing or reducing bactericidal properties.^10–13,15 Such behaviour is not described for β-lactams, but as cefiderocol is a siderophore cephalosporin and differs in its uptake into the bacterial cell, pH might be a limiting factor.¹ Moreover, the pH of human urine varies substantially across and within patients and thus, we set out to investigate if pH might impact cefiderocol activity, as it is indicated for the treatment of cUTI.^10,16,17

In routine AST, reference media are set to pH 7 and growth media such as CAMHB are considered to guarantee optimal growth of bacteria. Moreover, for cefiderocol it is recommended by the guidance document from EUCAST on cefiderocol BMD (latest version, December 2020) to test in ID-CAMHB (iron levels ≤ 0.03 mg/L). To better link in vitro experiments to in vivo conditions, pooled human urine (iron levels ∼0.05 mg/L/24 h) and CAMHB (iron levels ∼0.05 mg/L), were used as media to test cefiderocol activity against pathogens associated with cUTI, e.g. Enterobacteriaceae and P. aeruginosa at pH 5–8.

Testing in all three media showed no significant difference in median MIC values when compared within the same pH range (Wilcoxon signed rank test with P > 0.05).

To better understand the impact on bacterial killing within a time course of 24 h, TKC experiments in urine at pH 5, 7 and 8 were done. Growth of E. coli ATCC 29522 and K. pneumoniae ATCC 700603 was comparable between pH 5, 7 and 8, as seen in other studies,^11,13 indicating that pH is not a limiting factor for the growth of these pathogens. On the contrary, P. aeruginosa ATCC 27853 showed stagnated growth in pH 5 up to 7 h compared with pH 7. In the study by Ahmed et. al.,¹⁸ they propose that certain P. aeruginosa strains respond to acidic pH with reduced growth and the formation of resistant slow-growing subpopulations. Thus, reduced growth in the beginning of our experiments might relate to a stress response of P. aeruginosa. Nevertheless, after 24 h, growth in pH 5, 7 and 8 was comparable.

In the broth microdilution assay we could demonstrate that acidic pH reduced cefiderocol activity. Median MIC values were increased up to 50-fold (from 0.15 to 8 mg/L) in the case of P. aeruginosa isolates, followed by up to 32-fold higher MIC values for E. coli and K. pneumoniae isolates (median MIC values ranging from 0.19 to 4 mg/L and 0.31 to 16 mg/L, respectively). Moreover, within TKC experiments in an acidic environment, we could reveal that concentrations of 1×MIC led to stagnated growth or a maximum of 1 log₁₀ reduction in the bacterial load of all tested bacterial strains, while for TKCs at pH 7 at least a 3 log₁₀ reduction in the bacterial count was seen in the first 7 h.

Regrowth after 24 h, achieving a bacterial load comparable to the starting inoculum (1.5 × 106 cfu/mL), was only seen in TKC experiments with K. pneumoniae ATCC 700603 at a concentration of 1 × MIC. This could be explained by the ability of K. pneumoniae ATCC 700603 to produce an enzyme called SHV-1849, which belongs to the β-lactamases and is causing potential resistant bacterial subpopulations.¹⁹ Even though cefiderocol is resistant to a wide spectrum of β-lactamases, an accumulation of these in the test tube cannot be excluded and thus, a high amount of β-lactamases might decrease cefiderocol activity.

As urine and pH did not impact the overall bacterial growth in our experiments, one potential explanation of reduced cefiderocol activity could be that the acidic environment impacts cefiderocol stability. However, considering the final pH value of cefiderocol powder reconstituted in 0.9% sodium chloride, which ranges from 5.2 to 5.8, it seems unlikely that its stability is lowered in acidic milieu.¹

Another important aspect to consider could be reduced cefiderocol uptake in the bacterial cells. In the case of ciprofloxacin, a fluoroquinolone antibiotic, it has been shown that radioactive labelled [C¹⁴]-ciprofloxacin faced a reduced uptake into bacterial cells in acidic pH.¹¹ Ciprofloxacin has zwitterionic properties, and pH-dependent reversible changes in its ionic charges have been reported.

Yet, such a behaviour has not been described for β-lactams, but cefiderocol is a siderophore cephalosporin and differs in its uptake into the bacterial cell compared with other antibiotics in the β-lactam family. In detail, the mechanism of action of cefiderocol is binding to ferric iron via its catechol moiety, forming a chelating complex.¹ This allows cefiderocol to be actively transported into the periplasmic space through siderophore uptake systems in addition to passive diffusion through outer membrane porin channels.¹ In the publication by Lau et. al.,²⁰ they discuss the maintenance of iron homeostasis within the bacterial cell, and how bacteria evolved various types of iron acquisition systems. Ferric iron (Fe³⁺) is the dominant species in an oxygenated environment, while ferrous iron (Fe²⁺) is more abundant under anaerobic conditions or at low pH.²⁰ Thus, the acidic pH in our experiments might alter the availability of free ferric iron, which is required for optimal uptake into the cell.

Lastly, pH 5 might also affect porin conformation or even the genetic expression or inhibition of certain porins of the bacterial cells and thus, might enable the desired mechanism of action.^21,22

To link our in vitro experiments to the clinical in vivo situation we investigated if target site concentrations of cefiderocol in urine might also be impacted by an acidic environment. Studies have shown that cefiderocol is excreted to a high extent unchanged at the target site (60% to 90% of the dose).^1,6,7

Taking the pharmacokinetic formula for renal clearance of Katsube et. al., together with cefiderocol PK data of patients with normal renal clearance into account, concentrations of 1300 mg/L or more can be achieved in urine, after a recommended IV administration of 2 g cefiderocol every 8 h.^1,6,7 Within our in vitro TKC, conducted with concentrations of 1300 and 650 mg/L, efficient killing of all tested pathogens could be shown, even under acidic conditions.

Thus, even though MICs at pH 5 were ≥4 mg/L, which therefore were above the EUCAST breakpoint for Enterobacterales and P. aeruginosa of >2 mg/L, our results suggest sufficient concentrations at the target site in vivo.

However, our experiments face the limitations of in vitro studies, such as limited nutrient supply, missing host factors and short antibiotic exposure over 24 h (neglects development of antibiotic resistance). Moreover, fluctuating concentrations of cefiderocol due to the filling and emptying of the bladder cannot be considered in our static experiments. Thus, a dynamic bladder model would be necessary to better imitate pharmacokinetic/pharmacodynamic conditions.²³

Moreover, additional stability testing of cefiderocol during the experiments with HPLC would have been favourable.

In conclusion, this in vitro study demonstrates that the activity of cefiderocol is comparable in urine and CAMHB and is negatively impacted by acidic pH. Due to the high concentrations achieved in urine, a negative clinical outcome caused by low urine pH seems unlikely if current breakpoints are respected.

Funding

This study was supported by internal funding and the substance cefiderocol was thankfully provided by Shionogi Pharma & Co.

Transparency declarations

M.Z. has been consultant for Shionogi and PI- or investigator-initiated studies have been supported by Shionogi. The other authors declare that they have no conflicts of interest.

Supplementary data

Tables S1 to S3 are available as Supplementary data at JAC Online.