Journal of Antimicrobial Chemotherapy Advance Access published October 25, 2014
J Antimicrob Chemother doi:10.1093/jac/dku432
Gentamicin therapy for sepsis due to carbapenem-resistant and colistin-resistant Klebsiella pneumoniae Marcelino Gonzalez-Padilla1†, Julia´n Torre-Cisneros1,2*†, Francisco Rivera-Espinar3, Antonio Pontes-Moreno3, Lorena Lo´pez-Cerero2,4,5, Alvaro Pascual2,4,5, Clara Natera1, Marina Rodrı´guez3, Inmaculada Salcedo6, Fernando Rodrı´guez-Lo´pez2,7, Antonio Rivero1 and Jesu´s Rodrı´guez-Ban˜o2,4,5 1
*Corresponding author. Unidad Clı´nica de Enfermedades Infecciosas, Hospital Universitario Reina Sofı´a-IMIBIC-Universidad de Co´rdoba, Avenida Menendez Pidal s/n, 14004-Co´rdoba, Spain. Tel: +34957011636; Fax: +34957011636; E-mail: [email protected] †M. G.-P. and J. T.-C. contributed equally to this work.
Received 24 June 2014; returned 4 August 2014; revised 11 September 2014; accepted 29 September 2014 Objectives: Antimicrobial therapy for sepsis caused by carbapenem- and colistin-resistant Klebsiella pneumoniae is not well established. We hypothesized that the early use of gentamicin in cases due to susceptible organisms would decrease the crude mortality rate of this infection. Methods: This retrospective cohort study examined 50 cases of sepsis caused by carbapenem-resistant K. pneumoniae occurring between June 2012 and February 2013 during an outbreak of K. pneumoniae ST512 producing KPC-3, SHV-11 and TEM-1. Survival curves categorized by the use of gentamicin were constructed using the Kaplan – Meier method and compared using the log-rank test. Eight multivariate models using Cox regression were designed to study the risk factors for mortality and test the hypothesis. Results: The 30 day crude mortality rate was 38%. The use of targeted gentamicin was associated with reduced mortality (20.7% versus 61.9%, P ¼ 0.02). In all multivariate regression models, the use of gentamicin was independently associated with lower mortality until Day 30 (HR 0.17 – 0.29, P ¼ 0.03 – 0.002 depending on the model) after controlling for other potential confounding variables such as age, optimal treatment, renal function, severity of infection, underlying disease, use of tigecycline and previous hospitalization. Conclusions: Gentamicin reduced the mortality from sepsis caused by this K. pneumoniae ST512 clone producing KPC-3, SHV-11 and TEM-1. Keywords: K. pneumoniae, carbapenem resistance, mortality
Introduction Infections caused by carbapenemase-producing Klebsiella pneumoniae isolates (KPC-Kp) are an emerging problem worldwide.1 Due to its high spreading capacity, outbreaks are increasingly frequent in hospitals, and this microorganism is now considered endemic in several countries.2 The reported crude mortality rate of invasive infections due to KPC-Kp is very high (39% – 69%)3 – 9 and frequently higher than that of K. pneumoniae strains that do not produce these
enzymes.5,10,11 A significant problem in managing KPC-Kp infections is to find the appropriate antibiotic regimen as carbapenem resistance is often accompanied by resistance to other families of first-line antibiotics, such as cephalosporins, b-lactamase inhibitors or quinolones.12,13 This makes it necessary to use second-line antibiotics such as tigecycline, aminoglycosides, colistin or fosfomycin.14,15 Some studies have reported an association between lower mortality and the use of carbapenems in combination with other active agents, particularly when the strains have shown low levels of in vitro resistance to those antimicrobials.6,8
# The Author 2014. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]
1 of 9
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
Clinic Unit of Infectious Diseases, Hospital Universitario Reina Sofı´a-IMIBIC-Universidad de Co´rdoba, Co´rdoba, Spain; 2Spanish Network for Research in Infectious Diseases (REIPI RD12/0015), Instituto de Salud Carlos III, Madrid, Spain; 3Intensive Care Unit, Hospital Universitario Reina Sofı´a, Co´rdoba, Spain; 4Clinic Unit of Infectious Diseases, Microbiology and Preventive Medicine, Hospitales Universitarios Virgen Macarena y Virgen del Rocı´o, Seville, Spain; 5Departamento de Medicina, Universidad de Sevilla, Seville, Spain; 6Clinic Unit of Preventive Medicine, Hospital Universitario Reina Sofı´a, Co´rdoba, Spain; 7Clinic Unit of Microbiology, Hospital Universitario Reina Sofı´a, Universidad de Co´rdoba, Co´rdoba, Spain
Gonzalez-Padilla et al.
However, some case series have reported good results with regimens that do not include carbapenems 16 or active aminoglycosides.14,15,17 However, the best therapeutic regimen for the treatment of KPC-Kp infections and the specific effectiveness of aminoglycosides in these cases is not well defined. The aim of this study was to analyse the impact of gentamicin therapy on mortality from sepsis caused by KPC-Kp during an outbreak due to a highly resistant clone.
Methods Study design
Variables The main outcome variable of the study was crude mortality within 30 days of KPC-Kp isolation. The explanatory variables were gender, age, disease severity using the Charlson Comorbidity Index,18 hospitalization in the previous 3 months, admission to ICU, invasive procedures in the week prior to the diagnosis of infection (a need for mechanical ventilation, use of a central venous catheter or a urinary catheter), renal function on the day of diagnosis of the infection, site of infection, severity of presentation (sepsis, severe sepsis or septic shock), presence of bacteraemia, and empirical and targeted antibiotic treatment. Data were collected from clinical records using a specific questionnaire.
Microbiological variables and antibiotic susceptibility studies Selected K. pneumoniae isolates from the outbreak had previously been characterized as producing KPC-3.22 Briefly, susceptibility to antimicrobials was studied by microdilution, which showed that the isolates were resistant to penicillins, cephalosporins, aztreonam, fluoroquinolones, amikacin, tobramycin, co-trimoxazole, chloramphenicol, ertapenem, imipenem and meropenem (.32 mg/L). Some isolates were susceptible to tigecycline, fosfomycin and gentamicin according to the EUCAST breakpoints. Carbapenemase genes were identified by PCR methods using specific primers for groups A, B and D and further sequencing. Other bla genes were also detected by PCR methods. All isolates harboured blaKPC-3, blaSHV-11 and blaTEM-1 genes. All isolates showed 99.5% similarity and were assigned to clone ST512 by MLST. Additional isolates included in this study were considered to belong to the epidemic clone if they showed the resistance pattern described above. For these isolates, identification and susceptibility testing was performed by MicroScan (Dade MicroScan, Sacramento, CA, USA). MICs were classified according to EUCAST breakpoints. MDR, XDR and PDR K. pneumoniae were defined according to a recent consensus document.23
Statistical analysis Definitions The definitions were established prior to data collection and analysis. The criteria proposed by the Centres for Disease Control and Prevention19 and interpreted by clinical experts were used to establish the diagnostic criteria for infection and infection types. Sepsis was defined as the presence of both infection and a systemic inflammatory response according to the criteria of the International Sepsis Definitions Conference.20 Severe sepsis was defined as sepsis complicated by organ dysfunction with reversible hypotension after volume infusion.20 Septic shock was defined as sepsis with organ dysfunction and persistent hypotension despite volume infusion.20 The day of onset of infection was defined as the day the index culture was collected and the microorganism was isolated. The Cockroft – Gault formula21 was used to calculate CLCR. Renal failure was defined as CLCR , 60 mL/min. An antibiotic regimen was considered ‘active’ when it included at least one antibiotic to which the strain was susceptible or exhibited intermediate susceptibility in vitro according to EUCAST breakpoints. Treatment regimens including at least one drug to which the isolate was fully
2 of 9
Patients who died within the first 30 days were compared with survivors in terms of exposure to the studied variables. The results are expressed as mean+SD, median (range) (quantitative variables) or percentages of the group to which they belonged (qualitative variables). Differences between patients treated with and without gentamicin were analysed by univariate logistic regression. The OR was calculated with 95% CI. Survival curves were constructed using the Kaplan – Meier method and were compared using the log-rank test. To identify risk factors for mortality, univariate and multivariate analysis was performed using Cox regression and calculation of HRs with 95% CIs. For the multivariate analysis, the hypothesis that the use of gentamicin was associated with reduced mortality was considered. The limited number of patients did not allow us to develop a single, comprehensive multivariate model that included all potential variables influencing mortality because it would have been oversaturated. Alternatively, we developed different multivariate models using a forward stepwise strategy, in which the association between our variable of interest (therapy with gentamicin) and mortality was adjusted by triples or pairs of potential confounders.
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
An observational, retrospective cohort study was conducted in a 950 bed teaching hospital with an active transplant programme. From 1 June 2012 to 15 February 2013 all KPC-Kp isolates were collected from the database of the Microbiology Service of the Hospital Universitario Reina Sofia-IMIBIC in the context of an infectious outbreak caused by this microorganism. The clinical records of patients with KPC-Kp isolates were reviewed. All cases that met the following criteria were included in the study: (i) clinical evidence of infection with criteria for sepsis, severe sepsis or septic shock (see the definitions below); and (ii) the administration of antibiotic therapy (empirical and/or targeted) for at least 48 h with ≥1 active antibiotic in vitro (including cases of intermediate susceptibility). Meropenem was considered active in XDR and pandrug-resistant (PDR) isolates, following the results of Tumbarello et al.8 Only those cases in which empirical treatment was initiated within 12 h after the onset of symptoms and cases in which targeted treatment was initiated within 5 days after obtaining cultures were included. Recurrent infections were excluded.
susceptible according to EUCAST breakpoints were considered ‘optimal’. Treatments with gentamicin were considered optimal when the isolate had an MIC ≤4 mg/L. Treatment regimens in which no fully active drug was included but included at least one drug with intermediate activity (or resistance in the case of meropenem) were considered ‘suboptimal’. The time to the initiation of optimal treatment was also recorded. Regardless of the in vitro activity of the drugs, treatment regimens were classified as monotherapy (single drug) or combination therapy (antibiotics active only against Gram-positive organisms, such as glycopeptides, daptomycin or linezolid, and those active against anaerobes, such as metronidazole, were not considered). Given that antibiotic therapies in these patients often vary, thus making it difficult to assign them to a specific regimen, patients were assigned to a regimen only if it was initiated during the first 5 days after diagnosis and maintained for at least 5 days (or at least 48 h if the patient died within 5 days of treatment). Patients who died before receiving 48 h of treatment with any antibiotic regimen were excluded from the analysis. The study was approved by the Ethics Committee of the Hospital Universitario Reina Sofı´a, which waived the need to obtain written informed consent. All data collected were anonymized.
JAC
Gentamicin in carbapenem resistance
Table 1. Baseline characteristics of 50 patients with severe infection caused by carbapenem-resistant and colistin-resistant K. pneumoniae: univariate analysis of factors associated with crude mortality at 30 days Number (%) of patients (unless otherwise stated)
Demographic variables age (years), median (range) male
total (n ¼50)
no survivors (n¼19)
survivors (n¼31)
60.5 (19– 86) 32 (64.0)
67 (41– 86) 12 (63.2)
55 (19–85) 20 (64.5)
0.046 1.03 (1.00–1.06) 0.971 0.98 (0.38–2.49)
P
HR (95% CI)
4 (0–11) 16 (32.0)
4 (0–11) 10 (52.6)
3 (0 –8) 6 (19.4)
0.178 1.13 (0.95–1.35) 0.008 3.44 (1.39–8.54)
Previous hospitalization (3 previous months)
16 (32.0)
10 (52.6)
6 (19.4)
0.022 2.88 (1.16–7.14)
Admission to the ICU
22 (44.0)
8 (42.1)
14 (45.2)
0.671 1.16 (0.59–2.59)
Invasive procedures (in previous week) mechanical ventilation central venous catheter urinary catheter
26 (52.0) 36 (72.0) 46 (92.0)
10 (52.6) 11 (57.9) 17 (89.5)
16 (51.6) 25 (80.6) 29 (93.5)
0.644 1.24 (0.49–3.16) 0.349 0.62 (0.23–1.68) 0.893 0.90 (0.21–3.92)
Prior antibiotic therapy (in the previous month) quinolones amoxicillin/clavulanic acid meropenem cephalosporins piperacillin/tazobactam
21 (42.0) 14 (28.0) 23 (46.0) 12 (24.0) 13 (26.0)
12 (63.2) 3 (15.8) 9 (47.4) 7 (36.8) 6 (31.6)
9 (29.0) 11 (35.5) 14 (45.2) 5 (16.1) 7 (22.6)
0.043 0.132 0.764 0.071 0.461
Type of infection pneumonia purulent tracheobronchitis urinary tract infection surgical wound infection intra-abdominal infection infection of skin and soft tissue endocarditis primary or catheter-related bacteraemia infection of the CNS
24 (48.0) 4 (8.0) 10 (20.0) 4 (8.0) 1 (2.0) 1 (2.0) 1 (2.0) 4 (8.0) 1 (2.0)
8 (42.1) 1 (5.3) 5 (26.3) 1 (5.3) 1 (5.3) 0 (0) 1 (5.3) 2 (10.5) 0
16 (51.6) 3 (9.7) 5 (16.1) 3 (9.7) 0 (0) 1 (3.2) 0 2 (6.5) 1 (3.2)
0.356 1.07 (0.93–1.23)
Bacteraemia
18 (36.0)
7 (36.8)
11 (35.5)
0.866 1.08 (0.43–2.57)
Severe sepsis/septic shock
30 (60.0)
18 (94.7)
12 (38.7)
0.006 16.6 (2.21–125.1)
CLCR at start of antibiotic treatment (mL/min), mean+SD
96.2+ + 53.2
69.4+ + 38.0
Active empirical treatment
2.63 (1.03–6.71) 0.42 (0.12–1.43) 1.14 (0.46–2.82) 2.36 (0.93–6.02) 1.44 (0.55–3.79)
112.6+ + 55.0
0.005 0.98 (0.97–0.99)
6 (12.0)
2 (10.5)
4 (12.9)
0.857 0.87 (0.20–3.78)
Time to initiation of optimal targeted treatment (days), mean (range)
2.1 (0–5)
1.7 (0–5)
2.2 (0 –5)
0.405 0.86 (0.61–1.22)
Optimal targeted treatment monotherapy tigecycline gentamicin combination therapy tigecycline+gentamicin
37 (74.0) 16 (32.0) 8 (16.0) 8 (16.0) 21 (42.0) 21 (42.0)
9 (47.4) 4 (21.1) 3 (15.8) 1 (5.3) 5 (26.3) 5 (26.3)
28 (90.3) 12 (38.7) 5 (16.1) 7 (22.6) 16 (51.6) 16 (51.6)
0.001 0.18 (0.07–0.45) 0.258 0.53(0.18–1.60)
Optimal targeted treatment with tigecycline
29 (58.0)
8 (42.1)
21 (67.7)
0.059 0.41 (0.16–1.03)
Optimal targeted treatment with high-dose tigecycline
10 (20.0)
1 (5.3)
9 (29.0)
0.098 0.18 (0.20–1.37)
0.058 0.37 (0.13–1.03)
Continued
3 of 9
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
Comorbidities Charlson index, median (range) renal failurea
Gonzalez-Padilla et al.
Table 1. Continued Number (%) of patients (unless otherwise stated)
total (n ¼50)
no survivors (n¼19)
survivors (n¼31)
Targeted treatment with meropenem
11 (22.0)
9 (47.4)
2 (6.4)
,0.001 6.02 (2.37–15.28)
Optimal targeted treatment with gentamicin MIC ≤2 mg/L MIC .2 to ≤4 mg/L
29 (58.0) 13 (26.0) 16 (32.0)
6 (31.6) 1 (5.3) 5 (26.3)
23 (74.2) 12 (38.7) 11 (35.5)
0.002 0.21 (0.08–0.57) 0.009 0.05 (0.01–0.47) 0.133 0.42 (0.14–1.30)
P
HR (95% CI)
Results Clinical characteristics of the patients During the study period, KPC-Kp were isolated from 78 patients, of whom 67 had active infection and 50 met the criteria for sepsis as well as the other criteria for inclusion in this study. The 30 day crude mortality rate following the diagnosis of infection was 38% (19/50 patients). The overall characteristics of the cohort and subgroups who died and survived are shown in Table 1. Most patients had underlying diseases, had undergone invasive procedures, and had been admitted to ICUs or medical services. The most common types of infection were pneumonia and urinary tract infections, and around one-third of patients had bacteraemia. Four patients were suffering from purulent tracheobronchitis associated with mechanical ventilation.
Microbiological characteristics Tigecycline and fosfomycin were active against 36 (72.0%) and 10 (20.0%) isolates, respectively. With regard to gentamicin, 18 isolates (36.0%) were susceptible (MIC ≤2 mg/L), 31 (62.0%) showed intermediate susceptibility (MIC .2 to ≤4 mg/L) and 1 (2.0%) was resistant (MIC .4 mg/L). All isolates showed resistance to other antibiotics including colistin and meropenem (MIC .32 mg/L for all isolates). The strains isolated from the patients were classified as MDR for 5 patients (10.0%), XDR for 35 patients (70.0%) and PDR for 10 patients (20.0%). All 10 patients with PDR isolates were included in the suboptimal treatment group.
Antibiotic treatment All patients received empirical antibiotic treatment, but optimal empirical treatments were used in only six patients (12%). Targeted treatment was optimal in 37 patients (74%). Tables 1 and 2 show the antibiotic regimens used. Only one patient was infected with a gentamicin-resistant KPC-Kp strain. However, this antibiotic was targeted in only 29 patients (58%). The dose administered was 4 – 5 mg/kg/24 h and corrected following the monitoring of blood levels to obtain a pharmacokinetic goal of a Cmax of 15 – 20 mg/L and a Cmin of ≤1 mg/L. The antibiotic was administered as monotherapy in 8 patients (4 urinary tract infections, 1 pneumonia, 1 purulent
4 of 9
tracheobronchitis, 1 non-catheter-related bacteraemia, and 1 surgical wound infection) and in combination with another drug in 21 patients. Gentamicin was initiated 3.5+1.0 days after diagnosis. Gentamicin was not used in 21 patients (42%), one of whom was the patient with a resistant strain. The remaining 20 patients had susceptible or intermediate strains. Patients who did not receive gentamicin received treatment for 7.1+4.5 days (range 2 –14 days). Tables 3 and 4 show the main characteristics of the patients who received gentamicin. In patients who received gentamicin, renal failure (17.2% versus 52.4%), hospitalization in the previous 3 months (17.2% versus 52.4%) and use of cephalosporins in the previous month (6.9% versus 47.6%) were less frequent than among patients who did not receive gentamicin. The concomitant use of targeted tigecycline was more frequent in these patients (72.4% versus 38.1%). A loading dose of tigecycline was administered followed by 100 mg/12 h (200 mg loading dose) in 12 cases: 7 cases of nosocomial pneumonia (38.9% of the 18 cases treated with tigecycline), 2 cases of bacteraemia, 2 surgical wound infections, and 1 infection of the CNS. A conventional dose of 50 mg/12 h (100 mg loading dose) was administered in the remaining cases. Meropenem was administered as an extended infusion (≥3 h) at a dose of 1 –2 g every 6 –8 h.
Risk factors for 30 day crude mortality Univariate analysis showed differences between patients who died and those who survived (Table 1). The group who died were older (median 67 versus 55 years of age) and the most frequent underlying disease was cancer (42.1% versus 9.7%). Renal failure (52.6% versus 19.4%), hospitalization in the previous 3 months (52.6% versus 19.4%), use of quinolones in the previous month (63.2% versus 29.0%) and severe sepsis/septic shock (94.7% versus 40.0%) were more frequent. Optimal targeted treatment (47.4% versus 90.3%) and use of gentamicin (31.6% versus 74.2%) were less frequent. The use of targeted meropenem was associated with higher mortality (Table 1). The use of targeted treatment with gentamicin was associated with lower mortality compared with the use of targeted treatment without gentamicin (optimal or suboptimal) (20.7% versus 61.9%, P ¼ 0.02) (Table 4). Mortality was significantly reduced when the strain was susceptible to gentamicin (7.7% versus 80.0%, P¼ 0.008) and lower when the strain showed intermediate
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
Variables with a statistically significant different distribution between survivors and non-survivors are shown in bold. a CLCR calculated using the Cockroft–Gault formula.
JAC
Gentamicin in carbapenem resistance
Table 2. Antibiotics used in 50 patients with severe infection caused by carbapenem-resistant and colistin-resistant K. pneumoniae Number (%) of patients mortality
suboptimal
mortality
total
mortality
Empirical treatment tigecycline tigecycline + gentamicin fosfomycin meropenem piperacillin/tazobactam amoxicillin/clavulanic acid others
6 (8.0) 4 (66.6) 2 (33.3) 0 0 0 0 0
2 (33.3) 2 (50.0) 0 0 0 0 0 0
44 (92.0) 0 0 1 (2.2) 18 (40.9) 7 (15.9) 2 (4.5) 16 (36.4)
17 (38.6) 0 0 0 8 (44.4) 0 0 9 (56.2)
50 4 (8.0) 2 (4.0) 1 (2.0) 18 (36.0) 7 (14.0) 2 (4.0) 16 (32.0)
19 (38.0) 2 (50.0) 0 0 8 (44.4) 0 0 9 (56.2)
Targeted treatment monotherapy tigecycline high-dose tigecycline gentamicin meropenem
37 (74.0) 16 (43.2) 8 (21.6) 3 (8.1) 8 (21.6) 0
9 (24.3) 4 (25.0) 3 (37.5) 0 1 (12.5) 0
13 (26.0) 6 (46.2) 1 (7.7) 0 0 5 (38.5)
10 (76.9) 5 (83.3) 0 0 0 5 (100)
50 22 (44.0) 9 (18.0) 3 (6.0) 8 (16.0) 5 (10.0)
19 (38.0) 9 (40.9) 3 (33.0) 0 1 (12.5) 5 (100)
21 (56.7) 21 (56.7) 7 (18.9) 0 0 0 0
5 (23.8) 5 (23.8) 1 (14.3) 0 0 0 0
7 (53.8) 0 0 1 (7.7) 1 (7.7) 1 (7.7) 5 (38.5)
5 (71.4) 0 0 1 (100) 1 (100) 1 (100) 3 (60.0)
28 (56.0) 21 (42.0) 7 (14.0) 1 (2.0) 1 (2.0) 1 (2.0) 5 (10.0)
10 (35.7) 5 (23.8) 1 (14.3) 1 (100) 1 (100) 1 (100) 3 (60.0)
combination therapy tigecycline +gentamicin high-dose tigecycline meropenem +fosfomycin tigecycline +colistin high-dose tigecycline meropenem +colistin+fosfomycin
susceptibility (31.2% versus 53.3%, P ¼ 0.133). In patients treated with gentamicin, mortality was lower if the strain was susceptible (MIC ≤ 2) rather than had intermediate susceptibility (7.7% versus 31.2%, P ¼0.183). Mortality in the group of patients with optimal targeted treatment containing gentamicin was 20.7%, compared with 37.5% of patients with optimal targeted treatment without gentamicin (P ¼ 0.21). The survival analysis performed using Kaplan – Meier curves showed that patients treated with gentamicin had higher survival rates at 30 days after diagnosis (Figure 1, log-rank test 11.9, P ¼ 0.001). Figure 2 shows the different impact on mortality of suboptimal targeted treatment and optimal targeted treatment with or without gentamicin (log-rank test 17.3, P,0.001). Table 5 shows the eight models obtained from the multivariate analysis to test the study hypothesis. In all the models, the use of gentamicin was found to provide protection against mortality independently of other potential confounding variables, such as any optimal targeted treatment (model 1), optimal empirical treatment (model 2), having severe sepsis/septic shock (models 1 and 3), age (models 1 and 4), urinary tract infection (model 5), CLCR (models 2 – 8), neoplasia as underlying disease (model 6), treatment with tigecycline in the first week (model 7), and hospitalization in the previous 3 months (model 8).
Discussion The mortality of the patients included in our study (38%) confirms the high mortality of infections caused by carbapenemase-
producing K. pneumoniae, particularly KPC-Kp (39% – 69%). However, studies published to date do not yield uniform results. The wide variability observed in mortality rates, which were generally higher than in our study, is probably due to differences in the characteristics of the patients studied (age, comorbidities, severity and types of infection), the specific type of carbapenemase and the resistance profile of the isolates.3 – 10,24 – 26 Nonetheless, it is essential to determine the most effective therapies in treating these infections, with a view to reducing mortality. Due to the lack of prospective randomized studies with new or existing active antibiotics, it is necessary to perform observational analyses of the therapeutic strategies used in clinical practice. Most studies on the treatment of KPC-Kp have been performed in patients with bacteraemia6 – 9 and lack information on other types of infections that are underrepresented in those studies, or have included patients with sepsis but without bacteraemia. Our study included patients with various infections, but with sepsis criteria in all cases. In addition, other studies have often included strains with relatively low MICs of carbapenems and/or susceptibility to colistin,6 – 8,16 while our study provides information about highly carbapenem-resistant and colistinresistant strains that are susceptible to gentamicin. Moreover, criteria for assigning patients to specific antibiotic regimens were not clearly specified in some studies, which is key in these infections in which regimens are frequently changed during the course of the disease. In our study, only 12% of patients received optimal empirical therapy, which may have contributed to the high mortality rate.
5 of 9
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
optimal
Gonzalez-Padilla et al.
Table 3. Characteristics of patients with severe infection caused by carbapenem-resistant and colistin-resistant K. pneumoniae who were treated and not treated with targeted gentamicin Comparison of patients treated with gentamicin against other treatment regimens
Patients treated with gentamicin, N¼ 29 (58.0%) monotherapy (n ¼8)
combination therapy (n ¼ 21)
Patients with other treatment regimens, N¼ 21 (42.0%)
P
OR (95% CI)
Demographic variables age (years), mean+SD male, n (%)
60.9+14.0 21 (72.4)
61.0+19.0 5 (62.5)
60.8+12.2 16 (76.2)
56.4+24.3 11 (52.4)
0.483 0.104
1.01 (0.98 –1.04) 2.72 (0.81 –9.15)
Comorbidities Charlson index, mean+SD renal failurea, n (%)
4.2+2.6 5 (17.2)
4.0+2.5 2 (25.0)
4.3+2.6 3 (14.3)
3.1+1.7 11 (52.4)
0.116 0.014
1.25 (0.95 –1.65) 0.19 (0.05– 0.72)
Previous hospitalization (3 previous months), n (%)
5 (17.2)
3 (37.5)
2 (9.5)
11 (52.4)
0.014
0.19 (0.05– 0.72)
Admission to ICU, n (%)
12 (41.4)
0
12 (57.1)
10 (47.6)
0.952
1.03 (0.45 –2.35)
Invasive procedures (in previous week), n (%) mechanical ventilation central venous catheter urinary catheter
14 (48.3) 20 (69.0) 26 (89.7)
1 (12.5) 4 (50.0) 8 (100)
13 (61.9) 16 (76.2) 18 (85.7)
12 (57.1) 16 (76.2) 20 (95.2)
0.494 0.316 0.463
0.66 (0.21 –2.13) 0.47 (0.10 –2.06) 0.42 (0.04 –4.32)
Prior antibiotic therapy (in the previous month), n (%) quinolones amoxicillin/clavulanic acid meropenem cephalosporins piperacillin/tazobactam
11 (37.9) 10 (34.5) 11 (37.9) 2 (6.9) 7 (24.1)
1 (12.5) 2 (25.0) 3 (37.5) 0 2 (25.0)
10 (47.6) 8 (38.1) 8 (38.1) 2 (9.5) 5 (23.8)
10 (47.6) 4 (19.0) 12 (57.1) 10 (47.6) 6 (28.6)
0.403 0.207 0.218 0.004 0.779
0.61 (0.19 –1.94) 2.36 (0.62 –8.98) 0.48 (0.15 –1.53) 0.08 (0.01– 0.45) 0.83 (0.23 –2.98)
Type of infection, n (%) pneumonia purulent tracheobronchitis urinary tract infection surgical wound infection intra-abdominal infection non-focal or catheter-related bacteraemia infection of the CNS
14 (48.3) 3 (10.3) 7 (24.1) 2 (6.9) 1 (3.4) 1 (3.4) 1 (3.4)
1 (12.5) 1 (12.5) 4 (50.0) 1 (12.5) 0 0 0
13 (61.9) 2 (9.5) 3 (14.3) 1 (4.8) 1 (4.8) 1 (4.8) 1 (4.8)
10 (47.6) 1 (4.8) 3 (14.3) 2 (9.5) 0 3 (14.3) 0
0.509 — 0.205 — — — —
1.46 (0.47 –4.57)
Severity of infection, n (%) association with bacteraemia severe sepsis/septic shock
10 (34.5) 15 (51.7)
3 (37.5) 3 (37.5)
7 (33.3) 12 (57.1)
8 (38.1) 15 (71.4)
0.864 0.208
0.90 (0.28 –2.91) (0.14 –1.54)
CLCR at start of antibiotic treatment (mL/min), mean+SD
108.0+47.7
104.5+48.1
109.4+48.6
79.8+57.2
0.070
1.01 (0.99 –1.02)
Optimal empirical treatment , n (%)
4 (13.8)
0
4 (19.0)
2 (9.5)
0.648
1.52 (0.25 –9.18)
Optimal targeted treatment, n (%)
29 (100)
8 (100)
21 (100)
8 (38.1)
J Antimicrob Chemother doi:10.1093/jac/dku432
Gentamicin therapy for sepsis due to carbapenem-resistant and colistin-resistant Klebsiella pneumoniae Marcelino Gonzalez-Padilla1†, Julia´n Torre-Cisneros1,2*†, Francisco Rivera-Espinar3, Antonio Pontes-Moreno3, Lorena Lo´pez-Cerero2,4,5, Alvaro Pascual2,4,5, Clara Natera1, Marina Rodrı´guez3, Inmaculada Salcedo6, Fernando Rodrı´guez-Lo´pez2,7, Antonio Rivero1 and Jesu´s Rodrı´guez-Ban˜o2,4,5 1
*Corresponding author. Unidad Clı´nica de Enfermedades Infecciosas, Hospital Universitario Reina Sofı´a-IMIBIC-Universidad de Co´rdoba, Avenida Menendez Pidal s/n, 14004-Co´rdoba, Spain. Tel: +34957011636; Fax: +34957011636; E-mail: [email protected] †M. G.-P. and J. T.-C. contributed equally to this work.
Received 24 June 2014; returned 4 August 2014; revised 11 September 2014; accepted 29 September 2014 Objectives: Antimicrobial therapy for sepsis caused by carbapenem- and colistin-resistant Klebsiella pneumoniae is not well established. We hypothesized that the early use of gentamicin in cases due to susceptible organisms would decrease the crude mortality rate of this infection. Methods: This retrospective cohort study examined 50 cases of sepsis caused by carbapenem-resistant K. pneumoniae occurring between June 2012 and February 2013 during an outbreak of K. pneumoniae ST512 producing KPC-3, SHV-11 and TEM-1. Survival curves categorized by the use of gentamicin were constructed using the Kaplan – Meier method and compared using the log-rank test. Eight multivariate models using Cox regression were designed to study the risk factors for mortality and test the hypothesis. Results: The 30 day crude mortality rate was 38%. The use of targeted gentamicin was associated with reduced mortality (20.7% versus 61.9%, P ¼ 0.02). In all multivariate regression models, the use of gentamicin was independently associated with lower mortality until Day 30 (HR 0.17 – 0.29, P ¼ 0.03 – 0.002 depending on the model) after controlling for other potential confounding variables such as age, optimal treatment, renal function, severity of infection, underlying disease, use of tigecycline and previous hospitalization. Conclusions: Gentamicin reduced the mortality from sepsis caused by this K. pneumoniae ST512 clone producing KPC-3, SHV-11 and TEM-1. Keywords: K. pneumoniae, carbapenem resistance, mortality
Introduction Infections caused by carbapenemase-producing Klebsiella pneumoniae isolates (KPC-Kp) are an emerging problem worldwide.1 Due to its high spreading capacity, outbreaks are increasingly frequent in hospitals, and this microorganism is now considered endemic in several countries.2 The reported crude mortality rate of invasive infections due to KPC-Kp is very high (39% – 69%)3 – 9 and frequently higher than that of K. pneumoniae strains that do not produce these
enzymes.5,10,11 A significant problem in managing KPC-Kp infections is to find the appropriate antibiotic regimen as carbapenem resistance is often accompanied by resistance to other families of first-line antibiotics, such as cephalosporins, b-lactamase inhibitors or quinolones.12,13 This makes it necessary to use second-line antibiotics such as tigecycline, aminoglycosides, colistin or fosfomycin.14,15 Some studies have reported an association between lower mortality and the use of carbapenems in combination with other active agents, particularly when the strains have shown low levels of in vitro resistance to those antimicrobials.6,8
# The Author 2014. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]
1 of 9
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
Clinic Unit of Infectious Diseases, Hospital Universitario Reina Sofı´a-IMIBIC-Universidad de Co´rdoba, Co´rdoba, Spain; 2Spanish Network for Research in Infectious Diseases (REIPI RD12/0015), Instituto de Salud Carlos III, Madrid, Spain; 3Intensive Care Unit, Hospital Universitario Reina Sofı´a, Co´rdoba, Spain; 4Clinic Unit of Infectious Diseases, Microbiology and Preventive Medicine, Hospitales Universitarios Virgen Macarena y Virgen del Rocı´o, Seville, Spain; 5Departamento de Medicina, Universidad de Sevilla, Seville, Spain; 6Clinic Unit of Preventive Medicine, Hospital Universitario Reina Sofı´a, Co´rdoba, Spain; 7Clinic Unit of Microbiology, Hospital Universitario Reina Sofı´a, Universidad de Co´rdoba, Co´rdoba, Spain
Gonzalez-Padilla et al.
However, some case series have reported good results with regimens that do not include carbapenems 16 or active aminoglycosides.14,15,17 However, the best therapeutic regimen for the treatment of KPC-Kp infections and the specific effectiveness of aminoglycosides in these cases is not well defined. The aim of this study was to analyse the impact of gentamicin therapy on mortality from sepsis caused by KPC-Kp during an outbreak due to a highly resistant clone.
Methods Study design
Variables The main outcome variable of the study was crude mortality within 30 days of KPC-Kp isolation. The explanatory variables were gender, age, disease severity using the Charlson Comorbidity Index,18 hospitalization in the previous 3 months, admission to ICU, invasive procedures in the week prior to the diagnosis of infection (a need for mechanical ventilation, use of a central venous catheter or a urinary catheter), renal function on the day of diagnosis of the infection, site of infection, severity of presentation (sepsis, severe sepsis or septic shock), presence of bacteraemia, and empirical and targeted antibiotic treatment. Data were collected from clinical records using a specific questionnaire.
Microbiological variables and antibiotic susceptibility studies Selected K. pneumoniae isolates from the outbreak had previously been characterized as producing KPC-3.22 Briefly, susceptibility to antimicrobials was studied by microdilution, which showed that the isolates were resistant to penicillins, cephalosporins, aztreonam, fluoroquinolones, amikacin, tobramycin, co-trimoxazole, chloramphenicol, ertapenem, imipenem and meropenem (.32 mg/L). Some isolates were susceptible to tigecycline, fosfomycin and gentamicin according to the EUCAST breakpoints. Carbapenemase genes were identified by PCR methods using specific primers for groups A, B and D and further sequencing. Other bla genes were also detected by PCR methods. All isolates harboured blaKPC-3, blaSHV-11 and blaTEM-1 genes. All isolates showed 99.5% similarity and were assigned to clone ST512 by MLST. Additional isolates included in this study were considered to belong to the epidemic clone if they showed the resistance pattern described above. For these isolates, identification and susceptibility testing was performed by MicroScan (Dade MicroScan, Sacramento, CA, USA). MICs were classified according to EUCAST breakpoints. MDR, XDR and PDR K. pneumoniae were defined according to a recent consensus document.23
Statistical analysis Definitions The definitions were established prior to data collection and analysis. The criteria proposed by the Centres for Disease Control and Prevention19 and interpreted by clinical experts were used to establish the diagnostic criteria for infection and infection types. Sepsis was defined as the presence of both infection and a systemic inflammatory response according to the criteria of the International Sepsis Definitions Conference.20 Severe sepsis was defined as sepsis complicated by organ dysfunction with reversible hypotension after volume infusion.20 Septic shock was defined as sepsis with organ dysfunction and persistent hypotension despite volume infusion.20 The day of onset of infection was defined as the day the index culture was collected and the microorganism was isolated. The Cockroft – Gault formula21 was used to calculate CLCR. Renal failure was defined as CLCR , 60 mL/min. An antibiotic regimen was considered ‘active’ when it included at least one antibiotic to which the strain was susceptible or exhibited intermediate susceptibility in vitro according to EUCAST breakpoints. Treatment regimens including at least one drug to which the isolate was fully
2 of 9
Patients who died within the first 30 days were compared with survivors in terms of exposure to the studied variables. The results are expressed as mean+SD, median (range) (quantitative variables) or percentages of the group to which they belonged (qualitative variables). Differences between patients treated with and without gentamicin were analysed by univariate logistic regression. The OR was calculated with 95% CI. Survival curves were constructed using the Kaplan – Meier method and were compared using the log-rank test. To identify risk factors for mortality, univariate and multivariate analysis was performed using Cox regression and calculation of HRs with 95% CIs. For the multivariate analysis, the hypothesis that the use of gentamicin was associated with reduced mortality was considered. The limited number of patients did not allow us to develop a single, comprehensive multivariate model that included all potential variables influencing mortality because it would have been oversaturated. Alternatively, we developed different multivariate models using a forward stepwise strategy, in which the association between our variable of interest (therapy with gentamicin) and mortality was adjusted by triples or pairs of potential confounders.
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
An observational, retrospective cohort study was conducted in a 950 bed teaching hospital with an active transplant programme. From 1 June 2012 to 15 February 2013 all KPC-Kp isolates were collected from the database of the Microbiology Service of the Hospital Universitario Reina Sofia-IMIBIC in the context of an infectious outbreak caused by this microorganism. The clinical records of patients with KPC-Kp isolates were reviewed. All cases that met the following criteria were included in the study: (i) clinical evidence of infection with criteria for sepsis, severe sepsis or septic shock (see the definitions below); and (ii) the administration of antibiotic therapy (empirical and/or targeted) for at least 48 h with ≥1 active antibiotic in vitro (including cases of intermediate susceptibility). Meropenem was considered active in XDR and pandrug-resistant (PDR) isolates, following the results of Tumbarello et al.8 Only those cases in which empirical treatment was initiated within 12 h after the onset of symptoms and cases in which targeted treatment was initiated within 5 days after obtaining cultures were included. Recurrent infections were excluded.
susceptible according to EUCAST breakpoints were considered ‘optimal’. Treatments with gentamicin were considered optimal when the isolate had an MIC ≤4 mg/L. Treatment regimens in which no fully active drug was included but included at least one drug with intermediate activity (or resistance in the case of meropenem) were considered ‘suboptimal’. The time to the initiation of optimal treatment was also recorded. Regardless of the in vitro activity of the drugs, treatment regimens were classified as monotherapy (single drug) or combination therapy (antibiotics active only against Gram-positive organisms, such as glycopeptides, daptomycin or linezolid, and those active against anaerobes, such as metronidazole, were not considered). Given that antibiotic therapies in these patients often vary, thus making it difficult to assign them to a specific regimen, patients were assigned to a regimen only if it was initiated during the first 5 days after diagnosis and maintained for at least 5 days (or at least 48 h if the patient died within 5 days of treatment). Patients who died before receiving 48 h of treatment with any antibiotic regimen were excluded from the analysis. The study was approved by the Ethics Committee of the Hospital Universitario Reina Sofı´a, which waived the need to obtain written informed consent. All data collected were anonymized.
JAC
Gentamicin in carbapenem resistance
Table 1. Baseline characteristics of 50 patients with severe infection caused by carbapenem-resistant and colistin-resistant K. pneumoniae: univariate analysis of factors associated with crude mortality at 30 days Number (%) of patients (unless otherwise stated)
Demographic variables age (years), median (range) male
total (n ¼50)
no survivors (n¼19)
survivors (n¼31)
60.5 (19– 86) 32 (64.0)
67 (41– 86) 12 (63.2)
55 (19–85) 20 (64.5)
0.046 1.03 (1.00–1.06) 0.971 0.98 (0.38–2.49)
P
HR (95% CI)
4 (0–11) 16 (32.0)
4 (0–11) 10 (52.6)
3 (0 –8) 6 (19.4)
0.178 1.13 (0.95–1.35) 0.008 3.44 (1.39–8.54)
Previous hospitalization (3 previous months)
16 (32.0)
10 (52.6)
6 (19.4)
0.022 2.88 (1.16–7.14)
Admission to the ICU
22 (44.0)
8 (42.1)
14 (45.2)
0.671 1.16 (0.59–2.59)
Invasive procedures (in previous week) mechanical ventilation central venous catheter urinary catheter
26 (52.0) 36 (72.0) 46 (92.0)
10 (52.6) 11 (57.9) 17 (89.5)
16 (51.6) 25 (80.6) 29 (93.5)
0.644 1.24 (0.49–3.16) 0.349 0.62 (0.23–1.68) 0.893 0.90 (0.21–3.92)
Prior antibiotic therapy (in the previous month) quinolones amoxicillin/clavulanic acid meropenem cephalosporins piperacillin/tazobactam
21 (42.0) 14 (28.0) 23 (46.0) 12 (24.0) 13 (26.0)
12 (63.2) 3 (15.8) 9 (47.4) 7 (36.8) 6 (31.6)
9 (29.0) 11 (35.5) 14 (45.2) 5 (16.1) 7 (22.6)
0.043 0.132 0.764 0.071 0.461
Type of infection pneumonia purulent tracheobronchitis urinary tract infection surgical wound infection intra-abdominal infection infection of skin and soft tissue endocarditis primary or catheter-related bacteraemia infection of the CNS
24 (48.0) 4 (8.0) 10 (20.0) 4 (8.0) 1 (2.0) 1 (2.0) 1 (2.0) 4 (8.0) 1 (2.0)
8 (42.1) 1 (5.3) 5 (26.3) 1 (5.3) 1 (5.3) 0 (0) 1 (5.3) 2 (10.5) 0
16 (51.6) 3 (9.7) 5 (16.1) 3 (9.7) 0 (0) 1 (3.2) 0 2 (6.5) 1 (3.2)
0.356 1.07 (0.93–1.23)
Bacteraemia
18 (36.0)
7 (36.8)
11 (35.5)
0.866 1.08 (0.43–2.57)
Severe sepsis/septic shock
30 (60.0)
18 (94.7)
12 (38.7)
0.006 16.6 (2.21–125.1)
CLCR at start of antibiotic treatment (mL/min), mean+SD
96.2+ + 53.2
69.4+ + 38.0
Active empirical treatment
2.63 (1.03–6.71) 0.42 (0.12–1.43) 1.14 (0.46–2.82) 2.36 (0.93–6.02) 1.44 (0.55–3.79)
112.6+ + 55.0
0.005 0.98 (0.97–0.99)
6 (12.0)
2 (10.5)
4 (12.9)
0.857 0.87 (0.20–3.78)
Time to initiation of optimal targeted treatment (days), mean (range)
2.1 (0–5)
1.7 (0–5)
2.2 (0 –5)
0.405 0.86 (0.61–1.22)
Optimal targeted treatment monotherapy tigecycline gentamicin combination therapy tigecycline+gentamicin
37 (74.0) 16 (32.0) 8 (16.0) 8 (16.0) 21 (42.0) 21 (42.0)
9 (47.4) 4 (21.1) 3 (15.8) 1 (5.3) 5 (26.3) 5 (26.3)
28 (90.3) 12 (38.7) 5 (16.1) 7 (22.6) 16 (51.6) 16 (51.6)
0.001 0.18 (0.07–0.45) 0.258 0.53(0.18–1.60)
Optimal targeted treatment with tigecycline
29 (58.0)
8 (42.1)
21 (67.7)
0.059 0.41 (0.16–1.03)
Optimal targeted treatment with high-dose tigecycline
10 (20.0)
1 (5.3)
9 (29.0)
0.098 0.18 (0.20–1.37)
0.058 0.37 (0.13–1.03)
Continued
3 of 9
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
Comorbidities Charlson index, median (range) renal failurea
Gonzalez-Padilla et al.
Table 1. Continued Number (%) of patients (unless otherwise stated)
total (n ¼50)
no survivors (n¼19)
survivors (n¼31)
Targeted treatment with meropenem
11 (22.0)
9 (47.4)
2 (6.4)
,0.001 6.02 (2.37–15.28)
Optimal targeted treatment with gentamicin MIC ≤2 mg/L MIC .2 to ≤4 mg/L
29 (58.0) 13 (26.0) 16 (32.0)
6 (31.6) 1 (5.3) 5 (26.3)
23 (74.2) 12 (38.7) 11 (35.5)
0.002 0.21 (0.08–0.57) 0.009 0.05 (0.01–0.47) 0.133 0.42 (0.14–1.30)
P
HR (95% CI)
Results Clinical characteristics of the patients During the study period, KPC-Kp were isolated from 78 patients, of whom 67 had active infection and 50 met the criteria for sepsis as well as the other criteria for inclusion in this study. The 30 day crude mortality rate following the diagnosis of infection was 38% (19/50 patients). The overall characteristics of the cohort and subgroups who died and survived are shown in Table 1. Most patients had underlying diseases, had undergone invasive procedures, and had been admitted to ICUs or medical services. The most common types of infection were pneumonia and urinary tract infections, and around one-third of patients had bacteraemia. Four patients were suffering from purulent tracheobronchitis associated with mechanical ventilation.
Microbiological characteristics Tigecycline and fosfomycin were active against 36 (72.0%) and 10 (20.0%) isolates, respectively. With regard to gentamicin, 18 isolates (36.0%) were susceptible (MIC ≤2 mg/L), 31 (62.0%) showed intermediate susceptibility (MIC .2 to ≤4 mg/L) and 1 (2.0%) was resistant (MIC .4 mg/L). All isolates showed resistance to other antibiotics including colistin and meropenem (MIC .32 mg/L for all isolates). The strains isolated from the patients were classified as MDR for 5 patients (10.0%), XDR for 35 patients (70.0%) and PDR for 10 patients (20.0%). All 10 patients with PDR isolates were included in the suboptimal treatment group.
Antibiotic treatment All patients received empirical antibiotic treatment, but optimal empirical treatments were used in only six patients (12%). Targeted treatment was optimal in 37 patients (74%). Tables 1 and 2 show the antibiotic regimens used. Only one patient was infected with a gentamicin-resistant KPC-Kp strain. However, this antibiotic was targeted in only 29 patients (58%). The dose administered was 4 – 5 mg/kg/24 h and corrected following the monitoring of blood levels to obtain a pharmacokinetic goal of a Cmax of 15 – 20 mg/L and a Cmin of ≤1 mg/L. The antibiotic was administered as monotherapy in 8 patients (4 urinary tract infections, 1 pneumonia, 1 purulent
4 of 9
tracheobronchitis, 1 non-catheter-related bacteraemia, and 1 surgical wound infection) and in combination with another drug in 21 patients. Gentamicin was initiated 3.5+1.0 days after diagnosis. Gentamicin was not used in 21 patients (42%), one of whom was the patient with a resistant strain. The remaining 20 patients had susceptible or intermediate strains. Patients who did not receive gentamicin received treatment for 7.1+4.5 days (range 2 –14 days). Tables 3 and 4 show the main characteristics of the patients who received gentamicin. In patients who received gentamicin, renal failure (17.2% versus 52.4%), hospitalization in the previous 3 months (17.2% versus 52.4%) and use of cephalosporins in the previous month (6.9% versus 47.6%) were less frequent than among patients who did not receive gentamicin. The concomitant use of targeted tigecycline was more frequent in these patients (72.4% versus 38.1%). A loading dose of tigecycline was administered followed by 100 mg/12 h (200 mg loading dose) in 12 cases: 7 cases of nosocomial pneumonia (38.9% of the 18 cases treated with tigecycline), 2 cases of bacteraemia, 2 surgical wound infections, and 1 infection of the CNS. A conventional dose of 50 mg/12 h (100 mg loading dose) was administered in the remaining cases. Meropenem was administered as an extended infusion (≥3 h) at a dose of 1 –2 g every 6 –8 h.
Risk factors for 30 day crude mortality Univariate analysis showed differences between patients who died and those who survived (Table 1). The group who died were older (median 67 versus 55 years of age) and the most frequent underlying disease was cancer (42.1% versus 9.7%). Renal failure (52.6% versus 19.4%), hospitalization in the previous 3 months (52.6% versus 19.4%), use of quinolones in the previous month (63.2% versus 29.0%) and severe sepsis/septic shock (94.7% versus 40.0%) were more frequent. Optimal targeted treatment (47.4% versus 90.3%) and use of gentamicin (31.6% versus 74.2%) were less frequent. The use of targeted meropenem was associated with higher mortality (Table 1). The use of targeted treatment with gentamicin was associated with lower mortality compared with the use of targeted treatment without gentamicin (optimal or suboptimal) (20.7% versus 61.9%, P ¼ 0.02) (Table 4). Mortality was significantly reduced when the strain was susceptible to gentamicin (7.7% versus 80.0%, P¼ 0.008) and lower when the strain showed intermediate
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
Variables with a statistically significant different distribution between survivors and non-survivors are shown in bold. a CLCR calculated using the Cockroft–Gault formula.
JAC
Gentamicin in carbapenem resistance
Table 2. Antibiotics used in 50 patients with severe infection caused by carbapenem-resistant and colistin-resistant K. pneumoniae Number (%) of patients mortality
suboptimal
mortality
total
mortality
Empirical treatment tigecycline tigecycline + gentamicin fosfomycin meropenem piperacillin/tazobactam amoxicillin/clavulanic acid others
6 (8.0) 4 (66.6) 2 (33.3) 0 0 0 0 0
2 (33.3) 2 (50.0) 0 0 0 0 0 0
44 (92.0) 0 0 1 (2.2) 18 (40.9) 7 (15.9) 2 (4.5) 16 (36.4)
17 (38.6) 0 0 0 8 (44.4) 0 0 9 (56.2)
50 4 (8.0) 2 (4.0) 1 (2.0) 18 (36.0) 7 (14.0) 2 (4.0) 16 (32.0)
19 (38.0) 2 (50.0) 0 0 8 (44.4) 0 0 9 (56.2)
Targeted treatment monotherapy tigecycline high-dose tigecycline gentamicin meropenem
37 (74.0) 16 (43.2) 8 (21.6) 3 (8.1) 8 (21.6) 0
9 (24.3) 4 (25.0) 3 (37.5) 0 1 (12.5) 0
13 (26.0) 6 (46.2) 1 (7.7) 0 0 5 (38.5)
10 (76.9) 5 (83.3) 0 0 0 5 (100)
50 22 (44.0) 9 (18.0) 3 (6.0) 8 (16.0) 5 (10.0)
19 (38.0) 9 (40.9) 3 (33.0) 0 1 (12.5) 5 (100)
21 (56.7) 21 (56.7) 7 (18.9) 0 0 0 0
5 (23.8) 5 (23.8) 1 (14.3) 0 0 0 0
7 (53.8) 0 0 1 (7.7) 1 (7.7) 1 (7.7) 5 (38.5)
5 (71.4) 0 0 1 (100) 1 (100) 1 (100) 3 (60.0)
28 (56.0) 21 (42.0) 7 (14.0) 1 (2.0) 1 (2.0) 1 (2.0) 5 (10.0)
10 (35.7) 5 (23.8) 1 (14.3) 1 (100) 1 (100) 1 (100) 3 (60.0)
combination therapy tigecycline +gentamicin high-dose tigecycline meropenem +fosfomycin tigecycline +colistin high-dose tigecycline meropenem +colistin+fosfomycin
susceptibility (31.2% versus 53.3%, P ¼ 0.133). In patients treated with gentamicin, mortality was lower if the strain was susceptible (MIC ≤ 2) rather than had intermediate susceptibility (7.7% versus 31.2%, P ¼0.183). Mortality in the group of patients with optimal targeted treatment containing gentamicin was 20.7%, compared with 37.5% of patients with optimal targeted treatment without gentamicin (P ¼ 0.21). The survival analysis performed using Kaplan – Meier curves showed that patients treated with gentamicin had higher survival rates at 30 days after diagnosis (Figure 1, log-rank test 11.9, P ¼ 0.001). Figure 2 shows the different impact on mortality of suboptimal targeted treatment and optimal targeted treatment with or without gentamicin (log-rank test 17.3, P,0.001). Table 5 shows the eight models obtained from the multivariate analysis to test the study hypothesis. In all the models, the use of gentamicin was found to provide protection against mortality independently of other potential confounding variables, such as any optimal targeted treatment (model 1), optimal empirical treatment (model 2), having severe sepsis/septic shock (models 1 and 3), age (models 1 and 4), urinary tract infection (model 5), CLCR (models 2 – 8), neoplasia as underlying disease (model 6), treatment with tigecycline in the first week (model 7), and hospitalization in the previous 3 months (model 8).
Discussion The mortality of the patients included in our study (38%) confirms the high mortality of infections caused by carbapenemase-
producing K. pneumoniae, particularly KPC-Kp (39% – 69%). However, studies published to date do not yield uniform results. The wide variability observed in mortality rates, which were generally higher than in our study, is probably due to differences in the characteristics of the patients studied (age, comorbidities, severity and types of infection), the specific type of carbapenemase and the resistance profile of the isolates.3 – 10,24 – 26 Nonetheless, it is essential to determine the most effective therapies in treating these infections, with a view to reducing mortality. Due to the lack of prospective randomized studies with new or existing active antibiotics, it is necessary to perform observational analyses of the therapeutic strategies used in clinical practice. Most studies on the treatment of KPC-Kp have been performed in patients with bacteraemia6 – 9 and lack information on other types of infections that are underrepresented in those studies, or have included patients with sepsis but without bacteraemia. Our study included patients with various infections, but with sepsis criteria in all cases. In addition, other studies have often included strains with relatively low MICs of carbapenems and/or susceptibility to colistin,6 – 8,16 while our study provides information about highly carbapenem-resistant and colistinresistant strains that are susceptible to gentamicin. Moreover, criteria for assigning patients to specific antibiotic regimens were not clearly specified in some studies, which is key in these infections in which regimens are frequently changed during the course of the disease. In our study, only 12% of patients received optimal empirical therapy, which may have contributed to the high mortality rate.
5 of 9
Downloaded from http://jac.oxfordjournals.org/ at University of Pittsburgh on January 17, 2015
optimal
Gonzalez-Padilla et al.
Table 3. Characteristics of patients with severe infection caused by carbapenem-resistant and colistin-resistant K. pneumoniae who were treated and not treated with targeted gentamicin Comparison of patients treated with gentamicin against other treatment regimens
Patients treated with gentamicin, N¼ 29 (58.0%) monotherapy (n ¼8)
combination therapy (n ¼ 21)
Patients with other treatment regimens, N¼ 21 (42.0%)
P
OR (95% CI)
Demographic variables age (years), mean+SD male, n (%)
60.9+14.0 21 (72.4)
61.0+19.0 5 (62.5)
60.8+12.2 16 (76.2)
56.4+24.3 11 (52.4)
0.483 0.104
1.01 (0.98 –1.04) 2.72 (0.81 –9.15)
Comorbidities Charlson index, mean+SD renal failurea, n (%)
4.2+2.6 5 (17.2)
4.0+2.5 2 (25.0)
4.3+2.6 3 (14.3)
3.1+1.7 11 (52.4)
0.116 0.014
1.25 (0.95 –1.65) 0.19 (0.05– 0.72)
Previous hospitalization (3 previous months), n (%)
5 (17.2)
3 (37.5)
2 (9.5)
11 (52.4)
0.014
0.19 (0.05– 0.72)
Admission to ICU, n (%)
12 (41.4)
0
12 (57.1)
10 (47.6)
0.952
1.03 (0.45 –2.35)
Invasive procedures (in previous week), n (%) mechanical ventilation central venous catheter urinary catheter
14 (48.3) 20 (69.0) 26 (89.7)
1 (12.5) 4 (50.0) 8 (100)
13 (61.9) 16 (76.2) 18 (85.7)
12 (57.1) 16 (76.2) 20 (95.2)
0.494 0.316 0.463
0.66 (0.21 –2.13) 0.47 (0.10 –2.06) 0.42 (0.04 –4.32)
Prior antibiotic therapy (in the previous month), n (%) quinolones amoxicillin/clavulanic acid meropenem cephalosporins piperacillin/tazobactam
11 (37.9) 10 (34.5) 11 (37.9) 2 (6.9) 7 (24.1)
1 (12.5) 2 (25.0) 3 (37.5) 0 2 (25.0)
10 (47.6) 8 (38.1) 8 (38.1) 2 (9.5) 5 (23.8)
10 (47.6) 4 (19.0) 12 (57.1) 10 (47.6) 6 (28.6)
0.403 0.207 0.218 0.004 0.779
0.61 (0.19 –1.94) 2.36 (0.62 –8.98) 0.48 (0.15 –1.53) 0.08 (0.01– 0.45) 0.83 (0.23 –2.98)
Type of infection, n (%) pneumonia purulent tracheobronchitis urinary tract infection surgical wound infection intra-abdominal infection non-focal or catheter-related bacteraemia infection of the CNS
14 (48.3) 3 (10.3) 7 (24.1) 2 (6.9) 1 (3.4) 1 (3.4) 1 (3.4)
1 (12.5) 1 (12.5) 4 (50.0) 1 (12.5) 0 0 0
13 (61.9) 2 (9.5) 3 (14.3) 1 (4.8) 1 (4.8) 1 (4.8) 1 (4.8)
10 (47.6) 1 (4.8) 3 (14.3) 2 (9.5) 0 3 (14.3) 0
0.509 — 0.205 — — — —
1.46 (0.47 –4.57)
Severity of infection, n (%) association with bacteraemia severe sepsis/septic shock
10 (34.5) 15 (51.7)
3 (37.5) 3 (37.5)
7 (33.3) 12 (57.1)
8 (38.1) 15 (71.4)
0.864 0.208
0.90 (0.28 –2.91) (0.14 –1.54)
CLCR at start of antibiotic treatment (mL/min), mean+SD
108.0+47.7
104.5+48.1
109.4+48.6
79.8+57.2
0.070
1.01 (0.99 –1.02)
Optimal empirical treatment , n (%)
4 (13.8)
0
4 (19.0)
2 (9.5)
0.648
1.52 (0.25 –9.18)
Optimal targeted treatment, n (%)
29 (100)
8 (100)
21 (100)
8 (38.1)
