https://orcid.org/0000-0002-2859-8417
Abstract
While the treatment of ESBL-producing Enterobacterales osteomyelitis relies on carbapenems, the optimal regimen for OXA48 types remains unclear. We evaluated the efficacy of ceftazidime/avibactam in different combinations in an experimental model of OXA-48-/ESBL-producing Escherichia coli osteomyelitis.
E. coli pACYC184 is a clinical strain harbouring blaOXA-48 and blaCTX-M-15 inserts, with ‘increased exposure susceptibility’ to imipenem (MIC, 2 mg/L), gentamicin (MIC, 0.5 mg/L), colistin (MIC, 0.25 mg/L), ceftazidime/avibactam (MIC, 0.094 mg/L) and fosfomycin (MIC, 1 mg/L), and resistance to ceftazidime (MIC, 16 mg/L). Osteomyelitis was induced in rabbits by tibial injection of 2 × 108 cfu of OXA-48/ESBL E. coli. Treatment started 14 days later for 7 days in six groups: (1) control, (2) colistin 150.000 IU/kg subcutaneously (SC) q8h, (3) ceftazidime/avibactam 100/25 mg/kg SC q8h, (4) ceftazidime/avibactam + colistin, (5) ceftazidime/avibactam + fosfomycin 150 mg/kg SC q12h, (6) ceftazidime/avibactam + gentamicin 15 mg/kg intramuscularly (IM) q24h. Treatment was evaluated at Day 24 according to bone cultures.
In vitro, time–kill curves of ceftazidime/avibactam in combination showed a synergistic effect. In vivo, compared with controls, rabbits treated with colistin alone had similar bone bacterial density (P = 0.50), whereas ceftazidime/avibactam alone or in combinations significantly decreased bone bacterial densities (P = 0.004 and P < 0.0002, respectively). Bone sterilization was achieved using ceftazidime/avibactam in combination with colistin (91%) or fosfomycin (100%) or gentamicin (100%) (P < 0.0001), whereas single therapies were not different from controls. No ceftazidime/avibactam-resistant strains emerged in rabbits treated, regardless of the combination.
In our model of E. coli OXA-48/ESBL osteomyelitis, ceftazidime/avibactam in combination was more effective than any single therapy, whatever the companion drug used (gentamicin or colistin or fosfomycin).
Introduction
There are numerous mechanisms of antimicrobial resistance in Gram-negative bacteria that require appropriate antimicrobial therapies, posing a challenge that needs to be addressed. The dramatic worldwide spread of carbapenemase-producing Enterobacterales (CPE) is a constant threat and has been identified by the WHO1 as a top priority as the development of new effective drugs is urgently needed, as recently emphasized by IDSA in 2019.2
Historically, the main reservoirs of OXA-48-producing Enterobacterales were located in North Africa and in Turkey, where the first case was described in 2001.3 During the 2000s, Klebsiella pneumoniae was the primary host for resistance.4 Since then, we have observed a widespread occurrence of the blaOXA-48 gene in many European countries, including France, which has been involved in numerous hospital-acquired outbreaks.5 Moreover, as evidenced by the situation in France, the incidence of OXA-48-type isolates in Escherichia coli has risen dramatically and, by 2017, had equalled that of Klebsiella. It is noteworthy to mention that nearly 78% of OXA-48-type strains were also found to produce ESBLs.6 One concern about these superbugs is their spread in the community, as previously described in the Mediterranean region, with a high prevalence of OXA-48 producers at admission (13%) associated with ESBL producers (42.5%, mainly CTX-M-15).7 Moreover, these ESBL-producing Enterobacterales are very likely to be imported and subsequently spread during international travel, especially from southern Asia and sub-Saharan Africa.8,9
It should be noted that OXA-48 enzymes partially hydrolyse carbapenems but spare broad-spectrum cephalosporins, and are mainly found in K. pneumoniae rather than E. coli isolates.10 Nevertheless, physicians should be aware that blaOXA-48 genes spread extremely easily and quickly among Enterobacterales within the microbiota due to a high transfer efficiency of its plasmid, rather than the expansion of a single clone or gain of virulence.11
Furthermore, while Gram-negative bacteria are rarely involved in osteomyelitis (around 10% of cases),12 cases of such complex infections, especially those caused by CPE, have been increasingly described13–15 and do require expert advice to find the most appropriate combination. In addition, we recently showed in an experimental model of K. pneumoniae carbapenemase (KPC) K. pneumoniae osteomyelitis that the combination of ceftazidime/avibactam with colistin or gentamicin, but not fosfomycin, allowed significant bone sterilization in infected rabbits.16 As ceftazidime/avibactam restores the activity of ceftazidime against ESBL-producing Enterobacterales, but is also effective against OXA-48 producers,17 we decided to study its effectiveness for the treatment of osteomyelitis due to OXA-48/ESBL E. coli, deemed a difficult-to-treat infection.
Materials and methods
Bacterial strain
E. coli pACYC184-OXA-48-CTX-M-15 is an isogenic derivative of a urinary clinical strain of E. coli provided by The National Reference Center for Emerging Antibiotic Resistance (NARA), Fribourg, Switzerland. This isolate is a double mutant that co-expresses two genes of resistance: blaOXA-48, an OXA-48-type β-lactamase, and blaCTX-M-15, responsible for the ESBL activity.
This strain has ‘increased exposure susceptibility’ to imipenem (MIC, 2 mg/L) and full susceptibility to gentamicin (MIC, 0.5 mg/L), colistin (MIC, 0.25 mg/L), ceftazidime/avibactam (MIC, 0.094 mg/L) and fosfomycin (MIC, 1 mg/L), but is resistant to ceftazidime (MIC, 16 mg/L).
Before being inoculated into the animal, the double-mutant strain was cultured in casein hydrolysate-yeast extract-containing (CCY) broth medium at 37°C during 18 h. After centrifugation, the pellets were washed twice and resuspended in PBS. All inocula were assessed by plating serial dilutions on tryptic soy agar (bioMérieux, La Balme-les-Grottes, France).
Time–kill curve studies
Cultures were diluted in 10 mL of fresh Mueller–Hinton broth to yield an inoculum of 105 cfu/mL. The bactericidal activities of ceftazidime/avibactam, colistin, gentamicin, and fosfomycin, alone or in combination, were determined in triplicate. Antibiotic concentrations used were equivalent to four times the MIC to more closely reflect the optimal dosage of antibiotics. After 0, 1, 3, 6 and 24 h of incubation in a shaking water bath at 37°C, serial dilutions of 0.1 mL samples were cultured on Mueller–Hinton agar plates (Bio-Rad, Marnes la Coquettes, France) before determining the number of viable microorganisms using cfu. Synergistic effect was defined as a decrease of ≥ 2 log10 cfu/mL for the combination compared with that of its most active component. Bactericidal activity was defined by a reduction of ≥ 3 log10 cfu/mL compared with the initial inoculum.
Selection of antibiotic doses in rabbits
Plasma antibiotic concentrations were measured in uninfected animals to select doses leading to plasma concentrations equivalent to those observed in humans, as previously described in our previous experiment using the same drugs.16 Of note in the present experiment, gentamicin dose was halved, and administration routes of fosfomycin and colistin were switched from intramuscular (IM) to subcutaneous (SC) in accordance with the animal welfare committee and supported by pharmacokinetics data in veterinary medicine.18–20
Rabbit osteomyelitis model
Norden’s method was used to induce osteomyelitis in female New Zealand white rabbits, each weighing 2 to 3 kg, as previously described.21 The rabbits were housed in individual cages and received food and water ad libitum according to the French legislation on animal experimentation. The study was approved by the Animal Use Committee of Maisons-Alfort Veterinary School. The animals were anaesthetized by IM injection of ketamine, 25 mg/kg (Vibrac, Carros, France) and 2% xylazine, 25 mg/kg (Elanco, Cuxhaven, Germany). To ensure post-anaesthesia analgesia, an SC injection of buprenorphine, 30 µg/kg (Indivior, France) was carried out. Infection was induced by tibial intramedullary injection of a sclerosing agent, 0.1 mL of 3% sodium tetradecyl sulphate (Kreussler Pharma, La Chaussée-Saint-Victor, France), followed by 0.2 mL of the OXA-48/ESBL E. coli inoculum (109 cfu/mL) and 0.1 mL of saline. A high inoculum was preferred based on our previous experiments using a Gram-negative bacterium to achieve persistent osteomyelitis in the infected animals.22 Analgesia using a fentanyl patch of 12.5 µg/h (Mylan, Lyon, France) was applied and pethidine hydrochloride (Renaudin, Itxassou, France), 100 mg in 500 mL of water, was given in the rabbit bottles for 7 days following the procedure.
Treatment and its evaluation
Rabbits were randomly assigned to one of the following groups of rabbits (n = 12): (1) control; (2) colistin 150 000 IU/kg q8h SC; (3) ceftazidime/avibactam 100/25 mg/kg SC q8h; (4) ceftazidime/avibactam + gentamicin 15 mg/kg IM q24h; (5) ceftazidime/avibactam + colistin 150 000 IU/kg q8h SC; (6) ceftazidime/avibactam + fosfomycin 150 mg/kg q12h SC. Each regimen was initiated 14 days after bacterial inoculation on animals, and administered for a total duration of 7 days (until Day 21 of the procedure).
Rabbits (infected and controls) were euthanized by IV injection of pentobarbital 3 days after the end of antimicrobial therapy (Day 24) to avoid a carry-over effect of the antibiotics and allow regrowth of any potential mutants after a washout period. During animal autopsy, the upper third of the tibia (3 cm long), including compact bone and marrow, was isolated, weighted and split with bone crushing in an autopulverizer with 10 mL of sterile saline (Spex 6700; Freezer/Mill Industries Inc., Metuchen, USA). Thereafter, serial dilutions were performed and plated on tryptic soy agar for incubation at 37°C for 24 h before cfu were enumerated. The results were expressed as the median (IQR) log10 cfu/g of bone and as the percentage of rabbits with sterile bones, as previously defined.22
In vivo selection of mutants
Each undiluted bone homogenate (100 μL) of untreated and treated rabbits was plated onto Mueller–Hinton II agar and onto Mueller–Hinton II agar supplemented with ceftazidime/avibactam (0.125, 0.25 or 4 µg/mL) to detect resistant mutants after 24 h of incubation at 37°C. When bacterial growth was observed, E. coli identification was performed by MALDI-TOF mass spectrometry (MALDI Biotyper, Bruker Daltonique, Wissembourg, France), and specific enzymes of resistance were confirmed using the β LACTA test for ESBL production (Laboratoire Bio-Rad, 92430, Marnes-La-Coquette, France) and the NG-test CARBA-5 for OXA-48 (Laboratoire NG-Biotech, 35480, Guipry, France). The MIC values of ceftazidime/avibactam were determined by the ETEST method (bioMérieux, 69280, Marcy l’Étoile, France). Resistant mutants were defined by an MIC value of ≥4 µg/mL according to EUCAST.23
Data analysis
Numerical variables were represented by the median and IQR, whereas categorical variables were represented by numbers and percentages. Because of the relatively limited sample size, comparisons between groups were performed using Fisher’s exact test for categorical variables and the Kruskal–Wallis non-parametric method for numerical variables. The effect of the antibiotics and their 95% CIs on log10 cfu criteria were calculated by the Hodges–Lehmann estimator and the Bauer algorithm. The effects of antibiotics on the sterility criteria were estimated by the difference in the proportion of sterile bones and the 95% exact CI. A P value of <0.05 was considered as statistically significant. All analyses were realized with R software, using coin and exact2 × 2 packages.
Results
In vitro bactericidal effect
Time–kill curves obtained at 4× MIC (Figure 1) showed that ceftazidime/avibactam alone was rapidly bactericidal within 3 h compared with fosfomycin alone within 24 h. Conversely, colistin or gentamicin alone had no bactericidal activity.
In vitro killing curves for OXA-48-/ESBL-producing E. coli using ceftazidime/avibactam alone or in combination with various antibiotics using: (a) colistin; (b) gentamicin; or (c) fosfomycin at concentrations equivalent to 4× MIC. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
For the combination of ceftazidime/avibactam and colistin or gentamicin, fast and complete bactericidal and synergistic effects were observed within 6 h. When fosfomycin was added to ceftazidime/avibactam, slower bactericidal activity was observed compared with other companion drugs, with a synergistic effect after 24 h.
Furthermore, the use of a dual antibiotic regimen prevented regrowth, in contrast to the use of ceftazidime/avibactam alone.
Therapeutic studies
In control rabbits at Day 24, the median (IQR) of bacterial counts was 5.4 (4.7–5.7) log10 cfu/g of bone.
Compared with those of controls (Figure 2a and Figure 3), bacterial counts in rabbits treated with colistin were similar with a median (IQR) of 4.7 (3.4–5.4) log10 cfu/g of bone (P = 0.50), with no sterile bone observed.
Effect of ceftazidime/avibactam alone or in combination with various antibiotics on log10 cfu/g of bone (a) and on percentage of sterile bones (b) in animals with OXA-48-/ESBL-producing E. coli experimental osteomyelitis.
Bacterial density (log10 cfu/g of bone) of ceftazidime/avibactam alone or in combination with various antibiotics for the treatment of OXA-48-/ESBL-producing E. coli experimental osteomyelitis. Whiskers are the minimum and the maximum values and the horizontal line in each box plot, which covers the IQR, is the median.
Compared with those of controls, ceftazidime/avibactam alone significantly decreased bacterial counts with a median (IQR) of 3.5 (2.0–4.5) log10 cfu/g of bone (P = 0.004). However, the percentage of sterile bones (n = 3/11; 27%) at Day 24 was comparable to controls (P = 0.21) (Figure 2b). When used in combination, ceftazidime/avibactam + colistin or fosfomycin or gentamicin significantly decreased bone bacterial concentrations compared with controls (P < 0.0001, P = 0.0001 and P = 0.0002, respectively). Furthermore, those combinations also significantly decreased bone bacterial density compared with ceftazidime/avibactam alone (see Table 1) and promoted bone sterilization from 91% to 100% (P < 0.0001) compared with controls (Figure 2b).
Pairwise comparisons of ceftazidime/avibactam in combination with various antibiotics in carbapenemase (OXA-48)- and ESBL-producing E. coli experimental osteomyelitis
| Antibiotics | Difference in log10 cfu (95% CI) | P value |
|---|---|---|
| Ceftazidime/avibactam + gentamicin compared with | ||
| ȃColistin alone | −3.54 (−5.27 to −1.16) | 0.0004 |
| ȃCeftazidime/avibactam alone | −2.32 (−3.77 to −0.12) | 0.002 |
| ȃCeftazidime/avibactam + colistin | −0.01 (−0.13 to 0.07) | 0.51 |
| ȃCeftazidime/avibactam + fosfomycin | −0.04 (−0.12 to 0.08) | 0.44 |
| Ceftazidime/avibactam + fosfomycin compared with | ||
| ȃColistin alone | −3.50 (−4.38 to −2.10) | 0.0003 |
| ȃCeftazidime/avibactam alone | −2.30 (−3.73 to −0.18) | 0.007 |
| ȃCeftazidime/avibactam + colistin | 0.01 (−0.09 to 0.10) | 0.97 |
| Ceftazidime/avibactam + colistin compared with | ||
| ȃColistin alone | −3.51 (−4.29 to −1.85) | 0.0003 |
| ȃCeftazidime/avibactam alone | −2.23 (−2.88 to −0.13) | 0.004 |
| Ceftazidime/avibactam compared with | ||
| ȃColistin alone | −1.31 (−3.40 to 0.36) | 0.15 |
| Antibiotics | Difference in log10 cfu (95% CI) | P value |
|---|---|---|
| Ceftazidime/avibactam + gentamicin compared with | ||
| ȃColistin alone | −3.54 (−5.27 to −1.16) | 0.0004 |
| ȃCeftazidime/avibactam alone | −2.32 (−3.77 to −0.12) | 0.002 |
| ȃCeftazidime/avibactam + colistin | −0.01 (−0.13 to 0.07) | 0.51 |
| ȃCeftazidime/avibactam + fosfomycin | −0.04 (−0.12 to 0.08) | 0.44 |
| Ceftazidime/avibactam + fosfomycin compared with | ||
| ȃColistin alone | −3.50 (−4.38 to −2.10) | 0.0003 |
| ȃCeftazidime/avibactam alone | −2.30 (−3.73 to −0.18) | 0.007 |
| ȃCeftazidime/avibactam + colistin | 0.01 (−0.09 to 0.10) | 0.97 |
| Ceftazidime/avibactam + colistin compared with | ||
| ȃColistin alone | −3.51 (−4.29 to −1.85) | 0.0003 |
| ȃCeftazidime/avibactam alone | −2.23 (−2.88 to −0.13) | 0.004 |
| Ceftazidime/avibactam compared with | ||
| ȃColistin alone | −1.31 (−3.40 to 0.36) | 0.15 |
Pairwise comparisons of ceftazidime/avibactam in combination with various antibiotics in carbapenemase (OXA-48)- and ESBL-producing E. coli experimental osteomyelitis
| Antibiotics | Difference in log10 cfu (95% CI) | P value |
|---|---|---|
| Ceftazidime/avibactam + gentamicin compared with | ||
| ȃColistin alone | −3.54 (−5.27 to −1.16) | 0.0004 |
| ȃCeftazidime/avibactam alone | −2.32 (−3.77 to −0.12) | 0.002 |
| ȃCeftazidime/avibactam + colistin | −0.01 (−0.13 to 0.07) | 0.51 |
| ȃCeftazidime/avibactam + fosfomycin | −0.04 (−0.12 to 0.08) | 0.44 |
| Ceftazidime/avibactam + fosfomycin compared with | ||
| ȃColistin alone | −3.50 (−4.38 to −2.10) | 0.0003 |
| ȃCeftazidime/avibactam alone | −2.30 (−3.73 to −0.18) | 0.007 |
| ȃCeftazidime/avibactam + colistin | 0.01 (−0.09 to 0.10) | 0.97 |
| Ceftazidime/avibactam + colistin compared with | ||
| ȃColistin alone | −3.51 (−4.29 to −1.85) | 0.0003 |
| ȃCeftazidime/avibactam alone | −2.23 (−2.88 to −0.13) | 0.004 |
| Ceftazidime/avibactam compared with | ||
| ȃColistin alone | −1.31 (−3.40 to 0.36) | 0.15 |
| Antibiotics | Difference in log10 cfu (95% CI) | P value |
|---|---|---|
| Ceftazidime/avibactam + gentamicin compared with | ||
| ȃColistin alone | −3.54 (−5.27 to −1.16) | 0.0004 |
| ȃCeftazidime/avibactam alone | −2.32 (−3.77 to −0.12) | 0.002 |
| ȃCeftazidime/avibactam + colistin | −0.01 (−0.13 to 0.07) | 0.51 |
| ȃCeftazidime/avibactam + fosfomycin | −0.04 (−0.12 to 0.08) | 0.44 |
| Ceftazidime/avibactam + fosfomycin compared with | ||
| ȃColistin alone | −3.50 (−4.38 to −2.10) | 0.0003 |
| ȃCeftazidime/avibactam alone | −2.30 (−3.73 to −0.18) | 0.007 |
| ȃCeftazidime/avibactam + colistin | 0.01 (−0.09 to 0.10) | 0.97 |
| Ceftazidime/avibactam + colistin compared with | ||
| ȃColistin alone | −3.51 (−4.29 to −1.85) | 0.0003 |
| ȃCeftazidime/avibactam alone | −2.23 (−2.88 to −0.13) | 0.004 |
| Ceftazidime/avibactam compared with | ||
| ȃColistin alone | −1.31 (−3.40 to 0.36) | 0.15 |
Of note, in colistin and ceftazidime/avibactam + gentamicin groups, nine rabbits were analysed, knowing that in each group two animals died prematurely of sepsis and one was euthanized for suffering unbearable pain.
In accordance with time–kill curves, there was no difference in terms of bone bacterial density between the combinations, whether ceftazidime/avibactam was associated with colistin, fosfomycin or gentamicin (Table 1).
Ceftazidime/avibactam-resistant mutants in non-sterile animals
No ceftazidime/avibactam-resistant mutants emerged in rabbits treated with this regimen in single (n = 8) or in combination therapy with colistin (n = 1), with an MIC maintained at ≤0.25 mg/L. Likewise, no ceftazidime/avibactam-resistant mutants emerged in rabbits treated with colistin alone (n = 9).
Discussion
Our work emphasized the high efficacy of a 7 day course of ceftazidime/avibactam in dual therapy, in a subacute model of OXA-48-/ESBL-producing E. coli osteomyelitis. It should be remembered that in human clinical practice, osteomyelitis usually requires 6 to 12 weeks of treatment to achieve infection control.24 One strength of our experiment is its originality, aiming to address the issue of bone and joint infection by combining two mechanisms of resistance (co-expression of carbapenemase and ESBL enzymes), considered as a game-changer.
It is reassuring to observe that despite being a complex infection, it remains eligible for new drugs with an unexpected favourable outcome. Moreover, our findings are congruent with our previous work regarding KPC-producing K. pneumoniae,16 which highlighted the interest of such a regimen based on ceftazidime/avibactam to achieve bone sterilization.
Interestingly while ceftazidime/avibactam alone reduced bone bacterial density, it failed to obtain significant bone sterilization, indicating that medical treatment of such complex osteomyelitis should rely on combined therapy. Moreover, while all the combinations based on ceftazidime/avibactam were comparable, it is obvious that in clinical practice the use of colistin, fosfomycin or gentamicin should be balanced with the drug toxicity of these last-resort antibiotics, especially when prescribed for prolonged therapy such as for osteomyelitis.25 Indeed, in humans it is acknowledged that colistin and gentamicin can cause acute kidney failure when used at high doses, while fosfomycin is an additional sodium overload, often poorly tolerated in elderly patients suffering from underlying heart failure. Our findings are supported by the fact that ceftazidime/avibactam remains highly effective in vitro against OXA-48-producing E. coli (n = 9/9), as recently described in a large analysis of samples of CPE (n = 276) in the USA.26 On the other hand, there is a real lack of clinical data about ceftazidime/avibactam for the treatment of infections caused by OXA-48 carbapenemase-producing E. coli. Indeed, a recent publication in real-life conditions about ceftazidime/avibactam used as a salvage therapy for CPE showed that it was prescribed in association with other antibiotics in 65.8% of cases, with 73.7% clinical success and/or microbiological cure. This study concerned 12 patients infected by OXA-48 producers, with only one case being an E.coli.27 Likewise, in India ceftazidime/avibactam has been reported in clinical practice to be highly effective against OXA-48-producing Enterobacterales (82.35%; n = 14/17), including patients in ICU.28 In contrast to our work, those isolates were mainly composed of K. pneumoniae (84.2%) rather than E. coli. Moreover, our findings revealed no emergence of mutants within the limits of our sample size, even when using ceftazidime/avibactam in single therapy. Yet, resistance has already been described in K. pneumoniae isolates carrying blaOXA-48 genes after single therapy with ceftazidime/avibactam.29 Therefore, we cannot recommend its use alone, especially in difficult-to-treat infections such as osteomyelitis, which often require a new aggressive surgical procedure in case of failure, particularly on prosthetic materials.
It is important to note that a recent alert has also been reported regarding the risk of emergence of resistance in vitro for aztreonam/avibactam (another combination based on avibactam), against OXA-48 carbapenemase-producing E. coli in Germany.30 This argues for cautious use of these new drugs and encourages their use in combination with other drugs where they are susceptible, as resistance may develop if they are inappropriately prescribed on a larger scale.
There is a lack of literature regarding OXA-48-type osteomyelitis, therefore there is no recommended regimen for this particular entity. To the best of our knowledge, there are only two case series that address bone and joint infections due to OXA-48-producing Enterobacterales.15,31
Concerning bone and joint infections caused by ESBL-producing Enterobacterales, they have been mentioned in four original publications to date. Davido et al.32 reported the largest series with exclusively ESBL infections occurring on native bone and joints (n = 29), with a primary control of infection in 62.1% of cases, in contrast to ESBL infections on orthopaedic implants, where Pfang et al.31 reported infection control in 77.3% (n = 17/22) after implant removal, and Martinez-Pastor et al.33 in 42.8% (n = 3/7) when the implant was retained. Similarly, Papadopoulous et al.14 addressed the topic of MDR Gram-negative prosthetic joint infections in a large study including ESBL producers (n = 94), as well as MDR Pseudomonas spp. (n = 33) and some CPE (n = 12), and mentioned a pooled efficacy of 66.7% of cases (n = 72/108). Our findings regarding the interest of a combination therapy for the treatment of complex MDR organisms are concordant with Papadopoulous et al.14 showing that colistin was useful in combination in a model of prosthetic joint infections.
Unexpectedly, our findings show that fosfomycin combined with ceftazidime/avibactam was as effective as other agents, whereas this combination had been shown to be comparable to ceftazidime/avibactam alone in our previous work regarding KPC osteomyelitis.16 These results support the need for repeated experiments with different microorganisms expressing different mechanisms of resistance, in order to obtain the best data applicable in clinical practice.
Overall, in a subacute model of OXA-48/ESBL E. coli osteomyelitis, our results confirm the efficacy of ceftazidime/avibactam in combination, regardless of the associated agent. These data might suggest it would benefit from off-label use, if necessary, in expert centres.
Nevertheless, our study has some limitations. First, as we used 12 rabbits per treatment group and considering some animals became prematurely deceased, using 9 animals in the colistin group might have underestimated the efficacy of this drug. However, it should be noted that in our previous in vivo work, colistin monotherapy had no impact on bone sterilization, despite using 12 animals.16,22 Similarly, gentamicin was eventually used in combination with ceftazidime/avibactam on a limited number of animals due to unexpected fatal sepsis or killing, according to the animal wellness and welfare committee. However, the ultimate finding was unchanged considering 100% of the animals were sterile. Second, our experiments focused on a single strain of E. coli co-expressing ESBL and OXA-48. Hence, our findings may not be extrapolated to any CPE, including other microorganisms such as Klebsiella species or OXA-48-like-producing isolates. However, these findings are concordant with our previous experiment based on a KPC-99YC strain16 showing the necessity of combining drugs based on ceftazidime/avibactam to ensure bone sterilization. Third, we did not evaluate the interest of carbapenems (imipenem had increased exposure susceptibility), which are the antibiotics of choice for treating infections caused by ESBL-producing Enterobacterales. This choice was made considering our previous work relying on high-dose carbapenems,22 which caused severe diarrhoea in animals and could have led to a fatal outcome in terms of mortality with this particular virulent strain. Moreover, the increased use of carbapenems has led to a vicious cycle with increased antibiotic resistance and the emergence of CPE, and may not have been the best comparator in this experiment at a time where carbapenem-sparing is being encouraged.
In conclusion, in our model ceftazidime/avibactam in combination with colistin, fosfomycin or gentamicin was more effective than any other single therapy against OXA-48-/ESBL-producing E. coli.
Ceftazidime/avibactam should not be prescribed alone for the treatment of difficult-to-treat infections, due to its partial effectiveness in sterilizing deep tissues, such as bones in this case. Experimental stringent models of complex infections, including multiple mechanisms of resistance, may improve our knowledge to prescribe the optimal treatment in the event of a deadlock situation to ensure the best clinical outcomes.
Acknowledgements
We would like to thank Beatrice Gadaleta from l’Institut de Médecine et d’Epidémiologie Appliquée (IMEA) in Paris for her logistical assistance, the microbiological department of Raymond-Poincaré teaching hospital and the staff of the Plateforme de Recherche Biomédicale (PRBM) at Maisons-Alfort veterinary school for their technical support.
Funding
This study was supported by a grant from Pfizer (#52321751).
Transparency declarations
Authors declare no conflicts of interest.
