I NTRAPULMONARY AND S YSTEMIC P HARMACOKINETICS OF C OLISTIN

C

M

(CMS)

C

A

0.5 MIU

CMS

C

I

P

(S

I)

Cette partie de notre travail présente les résultats de COLIPOP Aero LBA faible dose, deuxième étude ancillaire du PHRC inter-régional COLIPOP. Elle a fait l’objet d’une soumission présentée ci-après selon le format (références, numérotation figures et tableaux) original de la revue Journal of Antimicrobial Chemotherapy. Seule une traduction en français de l’abstract a été ajoutée.

Intrapulmonary and Systemic

Pharmacokinetics of Colistin

Methanesulfonate (CMS) and Colistin

after Aerosol Delivery of 0.5 MIU of

CMS in Critically Ill Patients

Matthieu Boisson1,2,3, Nicolas Grégoire1,2, Marielle Cormier3, Patrice Gobin1,4, Sandrine Marchand1,2,4, William Couet1,2,4, Olivier Mimoz1,2,5

1_{Inserm U1070, Pôle Biologie Santé, 1 rue Georges Bonnet, 86000 Poitiers, France}

2_{Université de Poitiers, UFR Médecine-Pharmacie, 6 rue de la milétrie, 86000 Poitiers, France}

3_{Département d’Anesthésie-Réanimation, CHU de Poitiers, 2 rue de la milétrie, 86000}

Poitiers, France

4_{Service de Toxicologie-Pharmacocinétique, CHU de Poitiers, 2 rue de la milétrie, 86000}

Poitiers, France

5_{Service des Urgences – SAMU 86 – SMUR, CHU de Poitiers, 2 rue de la milétrie, 86000}

Poitiers, France

RESUME

Objectifs La posologie optimale pour la nébulisation du colistiméthate sodique (CMS), prodrogue de la colistine, n’est pas connue. Nous avons décrit les caractéristiques pharmacocinétiques intra-pulmonaires et plasmatiques du CMS et de la colistine après la nébulisation de 0,5 million d’unités internationales (MUI) de CMS chez des patients présentant une pneumopathie acquise sous ventilation mécanique (PAVM).

Méthodes Douze (12) patients de réanimation ont reçu un aérosol de 0,5 MUI de CMS sur 30 min toutes les 8 h. Des prélèvements sanguins ont été réalisés juste avant et à 0,5, 1, 2, 3, 5 et 8 h après la première nébulisation ; des mini-lavages bronchoalvéolaires (mini-LBA) bronchoalveolar lavages (mini-BAL) ont été réalisés à 1 et 5 h ou 3 et 8 h (6 patients de chaque) après la nébulisation. La colistine et le CMS ont été dosés dans le plasma et dans le LBA par chromatographie liquide couplée à la spectrométrie de masse avec une limite de quantification (LDQ) de 0,0097 mg/L dans les deux liquides biologiques. Afin d’estimer la concentration dans l’ELF, les concentrations dans le LBA ont été corrigées par un facteur de dilution estimé par la méthode de l’urée. L’analyse pharmacocinétique a été réalisée pour les concentrations plasmatiques de CMS et de colistine en utilisant une méthode non- compartimentale.

Résultats-Discussion Après nébulisation, les concentrations de CMS et de colistine dans l’ELF ([LDQ-70,8 mg/L] et [LDQ-20,1 mg/L] respectivement) étaient plus élevées (170 à 1 073 fois) que celles dans le plasma. Les concentrations de la colistine dans l’ELF doivent être considérées avec précaution du fait d’une importante fixation de la colistine au matériel de prélèvements des mini-LBA pour des concentrations inférieures à 6 mg/L dans le LBA. Néanmoins, les concentrations de CMS et de colistine dans l’ELF étaient beaucoup plus faibles qu’attendues avec les précédents résultats utilisant une dose de 2 MIU. Avec les caractéristiques pharmacocinétiques plasmatiques du CMS et de la colistine, il a été montré que la biodisponibilité du CMS était légèrement diminuée (environ 25 %) pour 0,5 MUI comparée à 2 MUI. Les différences de concentrations dans l’ELF entre les doses pourraient être dues aux differences de diffusion pulmonaires et/ou à une absorption plus rapide à 0,5 MUI.

Conclusions Cette étude montre que les concentrations de colistine et de CMS sont plus élevées dans l’ELF que dans le plasma après un aerosol de 0,5 MIU de CMS mais plus faibles

qu’attendues avec les précédents résultats utilisant une dose de 2 MIU. Ainsi, dans l’attente d’une nouvelle évaluation pharmacocinétique et pharmacodynamique du traitement des PAVM avec du CMS nébulisé, la dose de 2 MUI doit être préférée.

ABSTRACT

Objectives The optimal dosing for nebulized colistin methanesulphonate (CMS), the prodrug of colistin, is unknown. We described the pulmonary and systemic pharmacokinetics of CMS and colistin following the nebulization of 0.5 million International Unit (MIU) of CMS in patients with ventilator-associated pneumonia (VAP).

Methods Twelve (12) critically ill patients received 0.5 MIU doses of CMS administered every 8 h as 30-min nebulizations. Blood samples were collected immediately before and at 0.5, 1, 2, 3, 5 and 8h after the first nebulization; mini-bronchoalveolar lavages (mini-BAL) were performed at 1 and 5 h or 3 and 8 h (6 patients each) postdose. Colistin and CMS were assayed in plasma and BAL by chromatography-tandem mass spectrometry with a limit of quantification (LOQ) of 0.0097 mg/L in both matrices. In order to estimate concentrations in ELF, concentrations in BAL were corrected for dilution by the urea method. The pharmacokinetic analysis was performed for CMS and colistin plasma concentrations using a non-compartmental method.

Results-Discussion After aerosol delivery, CMS and colistin concentrations in ELF (ranges=[LOQ-70.8 mg/L] and [LOQ-20.1 mg/L] respectively) were much higher (170 to 1073-fold) than those in plasma. Concentrations of colistin in ELF should be considered with caution because when lower than 6 mg/L in BAL colistin bound to the mini-BAL devices. Nevertheless, CMS and colistin concentrations in ELF were much lower than expected from previous results with a 2 MUI dose. From CMS and colistin plasma pharmacokinetics it was shown that the CMS systemic bioavailability was only slightly decreased (about 25 %) for the 0.5 MIU dose compared to 2 MIU. The pronounced differences of concentrations in ELF between doses might be due to different location of lung deposition and/or a faster absorption at 0.5 MIU than 2 MIU but this should be further assessed.

Conclusions This study shows that colistin and CMS concentrations were much higher in ELF than in plasma after a 0.5 MIU aerosol of CMS but much lower than expected from previous results with a 2 MIU dose. Therefore, until new pharmacokinetic and pharmacodynamic assessments of VAP treatment with nebulized CMS, the 2 MIU dose should be preferred to the 0.5 MIU dose.

INTRODUCTION

Ventilator-associated pneumonia (VAP) is the most common complication due to prolonged mechanical ventilation. In intensive care unit (ICU), VAP represents near half of the healthcare-associated infections, affecting 18% of ventilated patients on average1,2. Responsible for a high mortality, going from 24 to 50% according to the pathogens involved, VAP prognosis is linked to the efficacy of initial antimicrobial treatment3. Gram-negative bacteria (Pseudomonas aeruginosa, Acinetobacter baumannii, Klebsiella pneumoniae…) are pathogens commonly responsible for VAP. For the last decades, bacterial resistance, in particular with Gram-negative bacilli (GNB), has increased while, at the same time, the development of new antibiotics has collapsed4.

In the absence of new antibiotics, colistin appears as a salvage therapy to multi-resistant GNB-induced VAP5. Colistin is an old antibiotic of the polymyxins family. Usually administrated as an inactive prodrug, colistin methanesulfonate (CMS), colistin is bactericidal against most GNB by acting on their membrane6,7. Developed in the 1950’s, it was quickly abandoned because of toxicity. For the last decade, a better knowledge of its pharmacokinetics properties allowed safer use and colistin has proved to be less toxic than the aminoglycosides8,9. Until recently, acquired resistance to colistin was uncommon. But, with the renewed interest in colistin use, description of cases of acquired resistance is increasing10. One of the possible explanations is the difficulty to reach the target concentrations with usual doses of CMS9.

Thus colistin is commonly used by inhalation in cystic fibrosis patients. Its efficacy has been proven for decades to treat infections or colonization with multidrug-resistant GNB11. In critically ill patients, notwithstanding paucity of level of evidence, aerosolized colistin in association with parenteral antibiotics (colistin or another) improved outcome of patients with VAP12.

A previous study in ICU patients with VAP showed that aerosol delivery of 2 million International Unit (MIU) of CMS allowed for very high concentrations of colistin into the epithelial lining fluid (9.53 to 1,137 mg/L)13. Based of these findings, we assumed that aerosol delivery of 0.5 MIU of CMS would be enough to achieve sufficient intrapulmonary colistin concentrations while improving patients’ safety.

So the main aim of this study was to assess pulmonary and systemic concentrations of CMS and colistin following nebulization of 0.5 MIU of CMS in ICU patients with VAP.

MATERIALS AND METHODS

Study population

The study was conducted between July 2014 and April 2015 in 12 adult patients hospitalized in the surgical ICU of the University Hospital of Poitiers, France, having developed VAP caused by GNB. Patients were eligible if they were between 18 and 85 years of age and developed VAP caused by GNB susceptible to colistin. Patients were excluded if they had received colistin during the 7 days preceding the study, had creatinine clearance < 30 ml/min, or had personal or family history of myasthenia.

At study onset, the following data were collected: age, sex, weight, height, diagnosis on admission, serum urea, serum creatinine, simplified acute physiology score (SAPS II)14 and sequential organ failure assessment score (SOFA)15. Creatinine clearance was calculated according to the Modification of Diet in Renal Disease (MDRD) formula16.

The study protocol was approved by the local ethics committee (Comité de Protection des Personnes Ouest III, approval number 2009009578-28) and registered on ClinicalTrials.gov (number NCT01060891). Written informed consent was obtained in all patients from their nearest relatives prior to initiation of the study.

CMS administration

Patients received 0.5 MIU of CMS (Colimycine®, Sanofi, Paris, France) dissolved in 5 mL of saline solution by nebulization over 30 min using a vibrating-mesh nebulizer (Aeroneb Pro®, Aerogen, Galway, France) every 8 hours. Solutions were prepared extemporaneously.

During aerosol delivery, all patients were sedated and received volume-controlled mechanical ventilation with a tidal volume of 7 to 8 mL per kilogram of predicted body weight and respiratory rates of 12 to 15 cycles/min. The humidifier was removed and the nebulizer was inserted near the T-piece connector on the inspiratory arm.

Sampling procedures

Blood samples

Blood samples were collected immediately before and at 0.5, 1, 2, 3, 5 and 8 h after the first nebulization of CMS. They were immediately centrifuged (3,000 × g for 10 min) at 4°C and the plasma was stored at – 80°C until analysis.

BAL fluid samples

Mini-bonchoalveolar lavage (mini-BAL) was performed as previously described13. Briefly, mini-BAL were performed with 16-French (Fr) double sterile catheters (BAL, KimVent®, Kimberly-Clark, Roswell, GA) inserted through the endotracheal tube. Two 20-mL aliquots of saline solution were instilled and then immediately aspirated with a syringe; these two BAL fluid samples were pooled and rapidly centrifuged (3,000 × g for 10 min), supernates were stored at – 80°C until analysis.

For the first six patients, mini-BAL were performed at 1 h and 3 h after initiating the first aerosolization. For last six patients, mini-BAL were performed at 5 h and 8 h after initiating of the first aerosolization.

Colistin and CMS assay in plasma and BAL fluid

A previously described liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay was used for CMS and colistin concentration measurements in plasma17 and BAL fluid18 samples. The limit of quantification (LOQ) of colistin was 0.0097 mg/L in both media.

Urea analysis in plasma and BAL fluid

A previously described liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay was used for urea concentrations measurements in BAL fluid samples13,19. The LOQ of the assay was 1 mg/L in BAL fluid. Urea concentrations in plasma were measured by photometric detection using an automatic analyzer (Modular automatic analyzer; Roche, France).

Estimation of CMS and colistin concentrations in epithelial lining fluid

Actual ELF concentrations of CMS or colistin (CELF) were obtained from measured BAL

fluid concentrations after correction for dilution18, according to the following equation: CELF=CBAL(Ureaplasma/UreaBAL)

where: CBAL corresponds to the CMS or colistin concentration measured in BAL fluid, and

Ureaplasma and UreaBAL correspond to the concentrations of urea determined in plasma and

BAL fluid, respectively.

Determination of the adsorption of CMS and colistin to inner surfaces of mini-

BAL devices

An in vitro study was performed in triplicate within a range of concentrations of CMS or colistin of 0.06, 0.6 and 6 mg/L. Concentrations in saline solutions were determinated before and after handling with the same double sterile catheter (BAL, KimVent®, Kimberly-Clark, Roswell, GA), syringes and vials that were used for ELF sampling.

Pharmacokinetic analysis

Individual pharmacokinetic analysis was performed for CMS and colistin plasma concentrations by a non-compartmental method with Phoenix® WinNonlin 6.3 (Certara L.P., St. Louis, Missouri). The following CMS and colistin pharmacokinetic parameters were derived for each subject as explained: the maximum plasma concentration (Cmax) and the time

taken to reach Cmax (tmax) were obtained directly from the concentration-time data; areas under

the concentration-time curve in plasma from time zero (predose) to the last quantifiable concentration (AUClast) were calculated using a linear trapezoidal method. Dose-normalized

pharmacokinetic parameters were statistically compared between 0.5 and 2 MIU dosage regimen with a linear model by using R software (v 3.1.0, function lm)

RESULTS

Patients

A total of 3 women and 9 men were enrolled. Their demographic, clinical and biological data are shown in Table 1. Patient 4 was excluded from analysis because of aerosol delivery defect.

Colistin absorption to inner surfaces of mini-BAL devices

In saline solution, in vitro study showed no major absorption of CMS to mini-BAL devices whatever the concentration. For colistin, the non-specific absorption was not significant for concentrations equal to 6 mg/L but at 0.6 mg/L about 80 % of colistin stuck non-specifically to the mini-BAL device and at 0.06 mg/L concentrations were measured below the limit of quantification (0.0097 mg/L).

Pulmonary pharmacokinetics

CMS and colistin concentrations estimated in ELF after 0.5 MIU nebulization of CMS were below limit of quantification (BLQ) for 6 and 7 of the 22 BAL fluid assays, respectively. Median CMS and colistin concentrations ranged from 1 to 20 mg/L between 1 and 8 h post- dose. CMS and colistin concentrations were much higher (approximately 100 to 1 000-fold) in the ELF than those in plasma but still with a high variability. However colistin ELF concentrations should be considered with caution because of non-specific absorption to mini- BAL devices as discussed later and for that reason only ELF concentrations of CMS are shown on Figure 1 together with concentrations measured after a 2 MIU dose after normalization to a 0.5 MIU dose13 for comparison purpose.

Plasma pharmacokinetics

CMS and colistin concentrations measured in plasma were BLQ for 7 and 6 of the 44 plasma assays in plasma, respectively. Plasma concentration-time profiles of CMS and colistin after 0.5 MIU nebulization of CMS are presented on Figure 2. CMS and colistin Tmax were

respectively 0.53 h and 2.97 h with corresponding Cmax at 0.04 mg/L and 0.03 mg/L The

systemic bioavailability was around 25 % lower after nebulization of 0.5 MIU of CMS than after 2 MIU. However none of dose-normalized AUCs (nor Cmax) was statistically different

between the two dosages (p > 0.05) (Table 2) but CMS Tmax was shorter after 0.5 MIU than

DISCUSSION

A previous study has shown that after nebulization of 2 MIU of CMS to ICU patients, colistin concentrations in ELF were very high, greater than 100 mg/L in most of the cases, whereas plasma concentrations were low (lower than 1 mg/L)13. Therefore it was hypothesized that aerosol dosage of CMS could be advantageous for the treatment of pulmonary infections but that a 2 MIU dose is probably unnecessarily high. The aim of the present study was therefore to assess the systemic and pulmonary pharmacokinetics of CMS and colistin after nebulization of CMS 0.5 MIU.

CMS and colistin concentrations in the ELF were obtained by mini-bronchalveolar lavage (BAL), demonstrating large inter-individual variability. The necessity of correcting measured BAL concentrations by urea to estimate ELF concentrations increases the experimental uncertainty. However BAL is achieved without visual control and samples may come from different lung segments, thus contributing as well to the large observed inter-individual variability. Yet previous studies have shown that this technique was a reliable method in comparison with bronchoscopic BAL20. Furthermore, with its protected extremity, the catheter should be minimally contaminated by colistin deposited on trachea and stem bronchi. Noticeably the experimental conditions were similar between the previous study using a 2 MIU dose of CMS13 and this new one conducted with 0.5 MIU, including the procedure and equipment used for BAL, but also the analytical assay for CMS and colistin allowing comparisons between the two. The only difference was that the volume of nebulized CMS solution in sodium chloride 0.9 % was 5 mL in this new study and 10 mL in the previous one13 and the CMS concentration was half of the previous one in this new study. Actually, deep lung deposition depends on droplet size which is influenced by physiochemical properties of nebulized solution21. This difference of CMS concentration might impact the location of lung deposition.

After nebulization of the 0.5 MIU dosage, CMS and colistin concentrations were much greater (100 to 700-fold) in ELF than in plasma, confirming the previous reports of limited systemic absorption of CMS and colistin after aerosol delivery13,19,22,23.

Whereas from previous results with a dose of 2 MIU13, concentrations of colistin in ELF greater than 25 mg/L were expected, most of concentration were measured below 5 mg/L (17

of 22 ELF concentrations). Therefore the colistin ELF concentrations difference is apparently greater that the 4 fold difference between doses. Colistin binds to synthetic membranes in a concentration depended manner24,25 which may contribute to this discrepancy. This was confirmed by in vitro data showing that at concentrations equal or less than 6 mg/L in saline solution, a large fraction of colistin sticks to the mini-BAL device (about 80 % at 0.6 mg/L) whereas the adsorption was negligible at a concentration of 6 mg/L, suggesting that previously reported data after nebulization with 2 MIU should be minimally affected by this problem. But colistin concentrations measured in BAL during this new study are not reliable and therefore are not reported. This non specific binding may possibly explain why Imberti et al. could not measure colistin in BAL fluid after IV CMS administration26. But as another consequence colistin concentrations measured in ELF after the 0.5 MIU dose of CMS are difficult to compare with those following the 2 MIU dose.

But colistin plasma concentrations are not affected by this non-specific adsorption on plastic tubing and can be compared between studies after being normalized to the same dose. Interestingly colistin plasma concentrations measured in the present study after 0.5 MIU of nebulized CMS are consistent with those previously reported after 2 MIU of nebulized CMS13 after normalization, as can be observed on Figure 2. For unexplained reasons the inter-patients variability seems lower after administration of the lowest dose. As far as dose normalized partial AUCs (from time 0 to 8 h) may be used for comparing CMS bioavailability after nebulization with 0.5 and 2 MIU, the consistency between these concentrations suggests that the fraction of the dose eventually absorbed is not affected by the dose. However colistin plasma concentrations depend not only on the fraction of the CMS dose absorbed but also on the pre-systemic and systemic conversion of CMS into colistin. Plasma concentrations of the parent drug (CMS) should rather be used to investigate bioavailability. In fact as observed with colistin, comparisons of normalized plasma concentrations versus time profiles of CMS suggest that the nebulized dose (0.5 versus 2 MIU) has no major effect if any, on the fraction bioavailable that was previously estimated to 9 % on average13.

But interestingly because CMS does not bind significantly to synthetic membranes, ELF concentrations of CMS may be compared between studies and appear to be considerably less than proportional to the dose. In fact it can be observed on Figure 1 that even after normalization to the dose, ELF concentrations of CMS are roughly 10 fold lower after nebulization of 0.5 MIU (ranging between 1 and 100 mg/L with only two exceptions) than after nebulization of 2 MIU (ranging between 10 and 1,000 mg/L with only one exception).

The most likely explanation for this apparent discrepancy between plasma and ELF data is that CMS should be absorbed faster after nebulization of the lower dose. Consistent with that, CMS time to peak (tmax) is shorter after nebulization of 0.5 MIU than after 2 MIU (0.5 h vs

1.5 h). This can also be observed on Figure 2 which also suggests that following the 0.5 MIU