Use of Monte Carlo simulation to evaluate the efficacy of tigecycline and minocycline for the treatment of pneumonia due to carbapenemase-producing Klebsiella pneumoniae

Background: Pneumonia caused by carbapenemase-producing Klebsiella pneumoniae (CP-KP) are increasingly encountered in hospitals worldwide, causing high mortality due to lack of treatment options. The goal of this study was to assess the efficacy of tigecycline and minocycline for CP-KP hospital-acquired pneumonia (HAP) by using Monte Carlo simulation. Methods: A total of 164 non-duplicated CP-KP strains were collected from sputum or blood in patients with HAP. The MICs for antimicrobials were determined by the agar dilution method. A 10,000-patient Monte Carlo Simulation based on a PK/PD model incorporating the MICs and population pharmacokinetic parameters were conducted to calculate probability of target attainment (PTA) at each MIC value and total cumulative fraction of response (CFR). Results: The susceptibility rate of tigecycline and minocycline were 79.9% and 41.5%, respectively. At recommended doses, an optimal PTA of 90% was obtained for treating HAP caused by CP-KP with MICs of tigecycline ≤0.5 mg/L or minocycline ≤4 mg/L. The CFR of tigecycline at the recommended dose and double dose (100 mg q12h) were 71.2% and 90.2%, respectively. The CFR of minocycline at recommended dose and double dose (200 mg q12h) was 53.4% and 77.2%, respectively. Conclusions: The findings of this study suggest that the recommended dose of tigecycline was not effective in HAP caused by CP-KP, and a higher CFR indicating a better clinical efficacy can be gained by doubling the dose (100 mg q12h). minocycline (200 mg q12h) might be a potential alternative of tigecycline to against strains with MICs ≤ 8 mg/L.

Keywords: Hospital-acquired pneumonia; Tigecycline; Minocycline; Monte Carlo simulation; Carbapenemase-producing Klebsiella pneumoniae

Carbapenems are the most commonly used first-line antibiotics for empirical treatment of infections caused by multi-drug resistant (MDR) Enterobacteriaceae. However, over the past decade, resistance to carbapenems has increased globally among these organisms, particularly in Klebsiella pneumoniae strains [[1]]. The production of various carbapenemases is the common mechanism of resistance; the most clinically prevalent carbapenemases are the KPC type, which are class A enzymes, the metallo-β-lactamases, including NDM, VIM, and IMP types, and the OXA-48 type, which are class D enzymes [[2]]. In addition to carbapenemases, carbapenemase-producing K. pneumoniae (CP-KP) strains often harbour various drug-resistance genes, which make other antimicrobials such as aminoglycosides and fluoroquinolones ineffective [[2]].

At present, only a few drugs, such as polymyxins, tigecycline, some aminoglycosides, and ceftazidime–avibactam, still have favourable in vitro activity against CP-KP. The potential nephrotoxicity of polymyxins and aminoglycosides restricts their broad use, especially among elder patients with renal insufficiency. Ceftazidime–avibactam only inhibits the activity of class A enzymes [[3]]. Tigecycline, the 9-t-butylglycylamido derivative of minocycline, is regarded as the last resort for the management of difficult-to-treat infections [[4]]. It shows favourable in vitro activity against MDR gram-positive and gram-negative bacteria, including carbapenem-resistant strains. However, the role of tigecycline in the treatment of severe nosocomial infections remains controversial because of its large volume of distribution and low concentrations in blood, urine, and epithelial lining fluid of the lungs [[5]]. In 2013, the Food and Drug Administration (FDA) reported an increased risk of death after tigecycline was used in FDA-approved and non-approved indications [[6]]. Owing to the limited treatment options, tigecycline is primarily used for infections caused by CP-KP in ICUs in Europe, China, and Latin America.

Although several clinical studies have reported the use of tigecycline for the treatment of CP-KP pneumonia [[7], [8], [9]], its definite efficacy is unclear. In addition, some experts have suggested that the current dosage of 50 mg twice daily may not be sufficient, and that higher doses should be considered, but this is still under debate [[10]]. To clarify these issues and to provide more comprehensive preclinical information on the rational use of tigecycline, we incorporated a Monte Carlo simulation into a pharmacokinetic/pharmacodynamic (PK/PD) model to simulate the clinical efficacy of different dosage regimens of tigecycline and minocycline for the treatment of CP-KP pneumonia.

A total of 164 non-duplicate clinical strains of K. pneumoniae resistant to imipenem and meropenem were isolated from sputum or blood of patients with hospital-acquired pneumonia (HAP) in three tertiary hospitals in Beijing, China, between December 2013 and May 2016. We used an agar dilution method to determine the minimum inhibitory concentrations (MICs) for tigecycline, minocycline, and other antimicrobials. Determinations were performed three times in strict accordance with the Clinical and Laboratory Standards Institute (CLSI) guidelines [[11]]. Escherichia coli ATCC25922 was the quality control strain used in each batch of tests. The CLSI and The European Committee on Antimicrobial Susceptibility Testing (EUCAST) interpretive breakpoints were used in this study [[11]]. PCR tests were used to detect the carbapenemase genes (KPC, NDM, VIM, IMP, and OXA-48) in these strains, as described in a previous study [[13]]. In addition, the MIC distributions of K. pneumoniae for tigecycline and minocycline from national and international studies, such as resistance surveillance programs (Alexander, BSAC, ECO-SENS, MYSTIC, NORM, and SENTRY) and other sources, were obtained from the EUCAST MIC distribution website (https://mic.eucast.org/Eucast2, last accessed 15 July 2017).

The population pharmacokinetic parameters of tigecycline were obtained from two studies including patients with community-acquired pneumonia (CAP), and one study on patients with hospital-acquired pneumonia (HAP) [[14]]. The disposition of tigecycline was best described by a two-compartment model with linear elimination; the primary structural parameters were steady-state total clearance (CLt) (mean, 19.2 L/h; standard deviation [SD], 7.76 L/h) [[14]]. The pharmacokinetic parameters of minocycline were obtained from one study that included elderly subjects with bacterial respiratory infections, and the steady-state CLt (mean ± SD) was 1.14 ± 0.4 L/h [[15]]. The best PK/PD index for prediction of the efficacy of tetracyclines and tigecycline is the ratio of the 24-h area under the concentration-time curve and the MIC (AUC0–24h/MIC) [[16]]. The equation: fAUC0-24h= (f*dose/CLt) was used to calculate the steady-state fAUC0-24h (f represents the fraction of free drug in the plasma, f = 0.2 for tigecycline and f = 0.35 for minocycline was used in this study [[17]]). Previous clinical studies have shown that the PK/PD index of tigecycline for prediction of efficacy in HAP was fAUC0–24h/MIC >0.9 [[19]], and the PK/PD index of minocycline for prediction of efficacy in pneumonia was fAUC0–24h/MIC >8.75 [[16]].

The Monte Carlo simulation was performed with 10,000 subjects using Crystal Ball software (version 11.1; Oracle, Denver, CO). Because all pharmacokinetic parameters were obtained from previous studies, the assumed simulated population was the same as those from the previous studies (patients with pneumonia). The CLt were defined as a log-Gaussian distribution and the MICs were defined as a discrete distribution. The simulated dosing regimens for tigecycline were 50 mg q12h, 75mg q12h and 100mg q12h, respectively, and the simulated doses for minocycline were 100 mg q12h, 150mg q12h, and 200mg q12h, respectively. The PK/PD values were calculated with one fixed MIC value and different dosing regimens in each simulation. The corresponding probability of target attainment (PTA) at each MIC value was obtained based on the PK/PD index. The cumulative fraction of response (CFR), defined as the total PTA for a dose regimen against the whole population of microorganisms, was calculated as follows:

Graph

where PT Ai stands for each MIC value and Fi is the proportion of isolates with this MIC [[20]]. The CFR for CP-KP and for K. pneumoniae strains from the EUCAST MIC distribution website were calculated separately. In brief, usually, a PTA or CFR >90% indicates that a specific regimen is optimal to treat infections caused by the specific microorganism populations.

Table 1 shows the antibiotic susceptibilities of the 164 CP-KP strains according to the MIC breakpoints of EUCAST or CLSI. The susceptibility breakpoint was 1 mg/L for tigecycline in the EUCAST breakpoints and 4 mg/L for minocycline in the CLSI breakpoints. All isolates were resistant to imipenem, meropenem, ceftazidime, and cefotaxime, and the susceptibility rate to moxifloxacin and fosfomycin was below 10%, as shown in the current EUCAST breakpoints. Colistin showed quite good in vitro activity against CP-KP, with the susceptibility rate of 97%. Tigeycline and minocycline were the second most active compounds, with the susceptibility rate of 79.9% and 41.5%, respectively. Amikacin and gentamicin were the third most active compounds, but only inhibited approximately 30% of the strains. As shown in Figure 1, compared with the MICs reported for K. pneumonia on the EUCAST distribution website, the MICs for tigecycline and minocycline were much higher in the CP-KP strains.

Graph: Figure 1. MIC distributions of 164 carbapenemase-producing Klebsiella pneumoniae (CP-KP) strains for tigecycline and minocycline compared with those obtained for K. pneumoniae from the EUCAST MIC distribution website.

Table 1. Antibiotic susceptibilities of 164 carbapenemase-producing Klebsiella pneumoniae strains according to the MIC breakpoints of EUCAST or CLSI.

1 MIC: minimum inhibitory concentration; S: susceptibility; I: intermediate; R: resistance.

KPC enzymes were the most frequently occurring carbapenemases, detected in 145 (88.4%) of the 164 CP-KP isolates. NDM genes were detected in 13 (7.9%), VIM in two (1.2%), and IMP in four (2.4%) isolates. OXA-48-like enzymes were not detected. The MIC50 and MIC90 for tigecycline in KPC-producing isolates were 1 mg/L and 2 mg/L, respectively. The MIC50 and MIC90 values in all other isolates were both 1 mg/L.

The PTA for the different dose regimens of minocycline and tigecycline at each MIC is shown in Figure 2. The simulation results for tigecycline at dosages between 50 and 100 mg every 12 h, showed that corresponding PTAs ranged from 72.2% to 99.2% at MIC 1 mg/L and from 11.3% to 70.8% at MIC 2 mg/L at the PK/PD index of fAUC0–24h/MIC >0.9 for HAP. The PTA for minocycline 100 mg q12h was >90% for MICs ≤4 mg/L, but was zero for all simulated regimens at MICs ≥32 mg/L.

Graph: Figure 2. Probability of target attainment (PTA) in 10,000 simulated patients given tigecycline and minocycline at different dosages.

The CFR of the different dose schemes of tigecycline and minocycline is shown in Table 2. The CFR of the recommended dose (50 mg q12h) and of the double dose (100 mg q12h) was 71.2% and 90.2%, respectively. The CFR of the minocycline recommended dose (100 mg q12h) and of the double dose (200 mg q12h) was 53.4% and 77.2%, respectively. A significant increase in CFRs were noted for isolates reported on the EUCAST website, with a value of >80% at the tigecycline or minocycline recommended doses.

Table 2. The cumulative fraction of response (CFR) for achieving PK/PD index with different antimicrobial regimens against carbapenemase-producing Klebsiella pneumoniae (CP-KP) and Klebsiella pneumoniae strains from the EUCAST MIC distribution website.

CP-KP are increasingly prevalent in many countries and have become endemic in some areas [[2]]. The most common carbapenemase found in this study was KPC enzymes, followed by metallo-β-lactamases, and the OXA type. In a recent European survey, KPC enzymes were detected in 42% of the carbapenemase-producing Enterobacteriaceae isolates, and the OXA-48-like enzymes, which were the second most frequent (38% of the isolates), were the dominant class of carbapenemases in eight countries [[21]]. These results suggested large variations in carbapenemase type in different regions. In another recent survey that included 18 European countries, the susceptibility rate to colistin and tigecycline in carbapenem-resistant Enterobacteriaceae (CRE) was 73.9% and 88.6%, respectively [[22]]. However, in our study, colistin was the most active agent with a susceptibility rate of 97%. This is probably because colistin is widely used for CRE infections in Europe, but not in China. Although carbapenemase cannot hydrolyze tigecycline, the tigecycline MICs of CP-KP isolates were much higher than those of the K. pneumoniae strains reported on the EUCAST website (most of them carbapenem-susceptible), which indicated a high level of resistance among CP-KP isolates.

A precise prediction of the efficacy of antibiotics against MDR bacteria is challenging, because these pathogens usually infect critically ill patients with various complications. No well-designed randomized controlled studies have established the role of tigecycline in the treatment of CRE infections. Our previous systematic review, which included 26 cohort studies, found that tigecycline might have a clinical efficacy similar to other antimicrobials [[23]]. However, we could not draw a definite conclusion because of the heterogeneity among the included studies and patients. Using Monte Carlo simulation in this study, we first conducted a PK/PD target attainment analysis to evaluate the efficacy of tigecycline and minocycline in CP-KP HAP. Our simulation showed that the CFR for tigecycline at the recommended dosage (50 mg q12h) for the treatment of HAP was 71.2%, which was similar to previously published clinical data [[7], [9]].

PK/PD-optimized dosing can improve the clinical cure rate, and the HAP guidelines suggest that antibiotic dosage should be determined using PK/PD data, rather than the manufacturer's prescribing information [[24]]. The PK/PD-guided antibiotic administration might be important for treatment of severe infections caused by highly resistant pathogens. Our simulation showed that the PTA and CFR of the recommended dosage of tigecycline for HAP caused by CP-KP was not optimal, while an increase of the dose to 100 mg q12h could achieve a more favourable CFR value (90.2%). The study of Xie et al. [[25]] also showed the standard tigecycline dosing regimen fails to achieve the best outcome in vivo for infections caused by K. pneumoniae. Two studies, which involved a limited number of patients, compared the outcomes of CRE infections treated with the approved dose of tigecycline and with high doses and found that mortality was significantly lower in patients on high doses [[26]]. In a phase 2 study of tigecycline for the treatment of HAP, the clinical cure rate with tigecycline 100 mg q12h (17/20, 85.0%) was numerically higher than with tigecycline 75 mg (16/23, 69.6%) and with imipenem/cilastatin (18/24, 75.0%) [[28]]. But the incidence of vomiting and nausea was the highest in the tigecycline 100 mg treatment group [[28]]. Therefore, 100 mg q12h may be an appropriate dose for tigecycline, but it should be noted that an increased dosage may increase the rate of adverse events.

Tigecycline was developed based on the structure of minocycline. Compared with other antimicrobial classes, minocycline still has comparatively good in vitro activity against MDR gram-negative bacteria [[29]]. In addition, it offers pharmacokinetic advantages over tigecycline as serum concentrations range from 2.1 to 6.6 µg/mL after a 200-mg intravenous dose [[30]]. Therefore, minocycline was reintroduced into the US market in 2009. The Monte Carlo simulation indicated that the CFR of minocycline at the recommended dose (100 mg q12h) and at the double dose (200 mg q12h) was 53.4% and 77.2% for the treatment of CP-KP pneumonia. For isolates with minocycline MICs ≤8 mg/L, the dose of 200 mg q12h had a PTA >90% using the PK/PD index of fAUC0–24h/MIC >8.75. These results suggested that minocycline might be a potential alternative to tigecycline to address the increasing drug resistance to current first-line agents.

There are several limitations to this study. First, the MICs of the CP-KP populations isolated from the three hospitals may not be representative of the MIC distributions in other regions. For example, the MIC50 of tigecycline in this study was 1 mg/L, but was 0.5 mg/mL in a recent European survey. Second, patients with CP-KP infections are usually severely sick with profound pathophysiological changes [[31]], such as hepatic or renal dysfunction, fluid derangements, and hypoproteinemia. Although we used PK data from infected patients, the PK/PD simulation could not consider all changes. Last, combination regimens are recommended for the treatment of CP-KP infections [[32]], but we did not consider combination regimens and merely predicted the efficacy of tigecycline and minocycline as monotherapy.

In summary, the paucity of effective regimens for CP-KP infections has posed a severe threat to public health. Tigecycline and minocycline are two of the few drugs with reasonably good in vitro activity against CP-KP. The PK/PD simulation showed that tigecycline (50 mg q12h) and minocycline (100 mg q12h) had a high probability of achieving adequate pharmacodynamic exposures in the treatment of infections caused by non-resistant isolates, but were suboptimal for infections caused by CP-KP. Considering the PK/PD characteristics of tigecycline, an increase of the dose to 100 mg q12h may improve clinical outcomes. Further well-designed clinical studies are urgently needed to verify these conclusions.

The authors report no conflicts of interest.