ABSTRACT
Our objective was to describe the prescribing practices, clinical characteristics, and outcomes of patients treated with ceftolozane-tazobactam (C/T) for multidrug-resistant (MDR) Gram-negative infections. This was a multicenter, retrospective, cohort study at eight U.S. medical centers (2015 to 2019). Inclusion criteria were age ≥18 years and receipt of C/T (≥72 hours) for suspected or confirmed MDR Gram-negative infection. The primary efficacy outcome, evaluated among patients with MDR Pseudomonas aeruginosa infections, was composite clinical failure, namely, 30-day all-cause mortality, 30-day recurrence, and/or failure to resolve or improve infection signs or symptoms after C/T treatment. In total, 259 patients were included, and P. aeruginosa was isolated in 236 (91.1%). The MDR and extremely drug-resistant phenotypes were detected in 95.8% and 37.7% of P. aeruginosa isolates, respectively. The most common infection source was the respiratory tract (62.9%). High-dose C/T was used in 71.2% of patients with a respiratory tract infection (RTI) overall but in only 39.6% of patients with an RTI who required C/T renal dose adjustment. In the primary efficacy population (n = 226), clinical failure and 30-day mortality occurred in 85 (37.6%) and 39 (17.3%) patients, respectively. New C/T MDR P. aeruginosa resistance was detected in 3 of 31 patients (9.7%) with follow-up cultures. Hospital-acquired infection and Acute Physiological and Chronic Health Evaluation II (APACHE II) score were independently associated with clinical failure (adjusted odds ratio [aOR], 2.472 and 95% confidence interval [CI], 1.322 to 4.625; and aOR, 1.068 and 95% CI, 1.031 to 1.106, respectively). Twenty-five (9.7%) patients experienced ≥1 adverse effect (9 acute kidney injury, 13 Clostridioides difficile infection, 1 hepatotoxicity, 2 encephalopathy, and 2 gastrointestinal intolerance). C/T addresses an unmet medical need in patients with MDR Gram-negative infections.
INTRODUCTION
Pseudomonas aeruginosa is a leading cause of health care-associated infections, particularly among critically ill and immunocompromised patients (1, 2). Treatment of these infections is challenging due to the pathogen’s diverse arsenal of virulence factors, intrinsic antimicrobial resistance, and ability to acquire a variety of resistance determinants (3). Furthermore, remaining antibiotics with preserved activity against multidrug-resistant (MDR) strains are limited by unfavorable pharmacokinetics and/or toxicity (4–6). The high morbidity and mortality associated with infections caused by MDR P. aeruginosa are due, in part, to the paucity of safe and effective treatment options and attest to the need for new therapeutic strategies (2, 7).
Ceftolozane-tazobactam (C/T) is a combination antibiotic consisting of a novel oxyimino-aminothiazolyl cephalosporin and a well-established beta-lactamase inhibitor (8, 9). It has in vitro antipseudomonal activity against isolates with the MDR phenotype (8, 10). It is also active against some extended-spectrum beta-lactamase (ESBL)-producing Enterobacterales; activity against other problem Enterobacterales (i.e., AmpC derepressed and carbapenemase producing) is more limited (11). Labeled indications include complicated intra-abdominal infections (cIAIs), complicated urinary tract infection (cUTIs), and, most recently, hospital-acquired or ventilator-associated bacterial pneumonia (HAP/VAP) (1). The clinical studies leading to the initial approval of C/T included few patients infected with MDR bacteria even though this is the population for whom it can fill an unmet medical need (12–14). Generalizing results from noninferiority studies conducted in patients infected with susceptible pathogens to patients with MDR infections is problematic (15). Those infected with resistant bacteria are typically older, have a higher burden of comorbidities, and are more critically ill (15, 16). These factors influence the effectiveness of antibiotics independent of microbiological activity, and recent history has reminded us that in vitro activity does not always reflect direct patient benefits (i.e., delafloxacin for uncomplicated gonorrhea, tigecycline for bacteremia/HAP, ceftobiprole for VAP, and daptomycin for pneumonia) (15, 17–20). In addition, the majority of patients in registry studies were recruited from sites in Eastern Europe where standards of care may differ from those in the United States (12, 13). Published data on the use of C/T for the treatment of MDR infections is slowly accumulating in the form of case reports, case series. and uncontrolled retrospective cohort studies (21–26). We sought to add to these data and describe the prescribing practices, clinical characteristics, microbiology, and outcomes of a large cohort of U.S. patients treated with C/T for confirmed or suspected MDR Gram-negative bacterial infections.
RESULTS
Patient characteristics.In total, 259 patients were included. A description of baseline demographic and clinical characteristics is shown in Table 1. Overall, the study cohort represented an elderly population (median age, 62; interquartile range [IQR], 49 to 72 years; ≥65 years, 40.9%) with high prevalences of diabetes mellitus (42.1%) and chronic obstructive pulmonary disease (COPD; 21.6%). The majority of patients (55.2%) had a history of colonization or infection with an MDR pathogen within the past year and 73.7% and 68.3% had a recent (90 day) hospitalization or systemic antibiotic exposure, respectively. Many patients had a high severity of illness at infection onset, with 50.6% residing in the intensive care unit (ICU) and a median (IQR) APACHE II score of 21 (12 to 27).
Demographic and clinical characteristicsa
Infection characteristics.The majority of infections (62.2%) were hospital acquired with the median (IQR) time from admission to infection onset of 5 (1 to 16) days. C/T was most commonly used to treat respiratory tract infections (62.9%), followed by skin and soft tissue (10.9%) and urinary tract (10.0%) infections (Table 2). Blood cultures were positive in eight (3.1%) patients (four with primary bacteremia, three with a respiratory tract infection, and one with a skin infection). A total of 384 isolates were cultured from 259 patients, including P. aeruginosa in 236 (91.1%) patients and Enterobacterales in 60 (23.2%). Over one-third (35.1%) of cultures were polymicrobial, while 19 (7.3%) patients had negative cultures or cultures were not obtained. All patients without a positive culture had a history of MDR or XDR P. aeruginosa infection(s). C/T susceptibility testing was performed on 168 (71.2%) P. aeruginosa isolates; 88.7% were susceptible. Among MDR (n = 167) and XDR (n = 74) strains tested, C/T susceptibility rates were 88.6% and 83.8%, respectively. A complete P. aeruginosa antibiogram is shown in Supplementary Appendix 1.
Infection characteristics
Infection management.A summary of infection management is shown in Table 3. Overall, 99.6% of patients received an infectious disease consult. C/T was initiated at a median (IQR) of 84 (18 to 164) hours after the infectious disease consult. Source control (e.g., abscess drainage, wound debridement, and line removal) was pursued in 73.8% of patients with infections potentially amendable to source control (n = 80). The median time from culture collection to C/T initiation was 87 (51 to 139) hours. Among patients with a positive culture (n = 250), 86 (34.4%) received in vitro-active antibiotic therapy prior to C/T, most commonly with an aminoglycoside (n = 31, 12.4%). The median time to active antibiotic therapy was 50 (7 to 94) hours. High-dose C/T was used in 165 (63.7%) patients, including 116 (71.2%) with a respiratory tract infection. The C/T dose was renally adjusted in 79 (30.5%) patients. Among patients with a respiratory tract infection who had their dose adjusted for impaired kidney function (n = 48), only 19 (39.6%) received renally adjusted high-dose C/T, while the remainder were potentially underdosed. Combination IV antibiotic therapy was used in 64 (24.7%) patients, most commonly with an aminoglycoside (18.1%). Among patients with a respiratory tract infection, 48 (29.4%) received adjuvant therapy with inhalation tobramycin or colistin. The median (IQR) duration of C/T was 10 (6 to 15) days.
Treatment information
Outcomes.Patient outcomes are displayed in Table 4. Overall, composite clinical failure and 30-day mortality occurred in 85 (37.6%) and 39 (17.3%) patients in the primary efficacy population (MDR P. aeruginosa infections), respectively. Among patients originally admitted from home (n = 104), 36.5% and 7.7% required new nursing home placement or inpatient rehabilitation following discharge, respectively. By source, the highest rates of clinical failure and 30-day mortality were recorded in patients with a respiratory tract infection (45.0% and 24.2%), while the lowest rates were in patients with a urinary tract infection (9.5% and 4.8%). Outcomes were similar in patients with MDR P. aeruginosa infections confirmed to be C/T susceptible (Table 4). On bivariate analysis, additional variables associated with higher clinical failure included (Supplementary Appendix 2 and 3) the following: older age, CrCl of ≤30 ml/min or on hemodialysis, chronic obstructive pulmonary disease, hospital-acquired infection, respiratory tract infection, ICU at infection onset, sequential organ failure assessment (SOFA) score, APACHE II score, and C/T renal dose adjustment. The impact of renal dose adjustment on clinical failure was most pronounced in patients with a respiratory tract infection (OR, 3.409; 95% CI, 1.627 to 7.142). Early active antibiotic therapy, early C/T, high-dose C/T, and combination antibiotic therapy did not impact clinical failure rates in the overall efficacy population or among patients with an MDR P. aeruginosa respiratory tract infection (Supplementary Appendix 2 and 3). The final multivariable logistic regression models for clinical failure in the primary efficacy population and in patients with MDR P. aeruginosa infections confirmed to be susceptible to C/T are shown in Table 5. Hospital-acquired infection and higher APACHE II score were the independent predictors of clinical failure in both models, while CrCl of <30 ml/min or receipt of hemodialysis remained an explanatory variable in the primary efficacy population, and older age was an additional independent predictor in the subgroup analysis restricted to patients with C/T-susceptible MDR P. aeruginosa infections.
Effectiveness outcomesa
Multivariable logistic regression models for clinical failurea
Follow-up cultures were available for 31 (20.9%) patients with MDR P. aeruginosa infections that were susceptible to C/T at baseline. Follow-up cultures were obtained more frequently in patients who experienced clinical failure than those who did not (29.5% versus 14.9%, P = 0.032). The development of new C/T resistance among MDR P. aeruginosa isolates was documented in 3 patients (9.7%) at 3, 7, and 8 days after C/T initiation. Two patients had VAP (one with bilateral necrotizing infection) and were treated with high-dose C/T. The third patient to develop resistance had necrotizing fasciitis and was treated with standard-dose C/T. No patients among this group received combination therapy. Two of the three patients had worsening signs and symptoms of infection at the time resistance was documented. Resistance was detected by disk diffusion in two patients and by Etest in one patient (initial C/T MIC, 3 mg/liter; day 3 MIC, 128 mg/liter).
With regard to safety, a total of 27 adverse events occurred in 25 patients (9.7%) (Table 6). Nine patients developed acute kidney injury (AKI) while receiving C/T; all of these patients were receiving concomitant nephrotoxic agents around the time of the event. In particular, 7 (12.7%) patients who received C/T combination therapy with an aminoglycoside or a polymyxin experienced AKI compared to 2 (1.0%) who did not receive either of these antibiotic classes (P < 0.001). Thirteen patients (5.0%) developed Clostridioides difficile-associated diarrhea (two received C/T combination therapy and eight received high-dose C/T). Two patients experienced possible C/T-associated encephalopathy (decreased mentation and confirmed by electroencephalogram). One of these patients received high-dose C/T (CrCl, 65 ml/min) and the other received standard-dose C/T adjusted for hemodialysis. Both patients had other contributing factors. Hepatotoxicity occurred in one patient and gastrointestinal (GI) intolerance (nausea and vomiting diarrhea) occurred in two.
Safety outcomes for total cohorta
DISCUSSION
Registration studies conducted for drug approval give information about the efficacy and safety of a drug under ideal conditions in patient populations that may be very different from those we care for in everyday clinical practice (27). Real-world studies are, therefore, necessary and complementary to ensure that the results seen in registration studies actually translate into benefits for our patients (27). This is particularly relevant for new antibiotics, such as C/T, that are brought to market under the FDA accelerated approval program. Under this program, antibiotics with a novel mechanism of action or structural alteration that confer an expanded spectrum of antimicrobial activity may be granted regulatory approval on the basis of noninferiority to the current standard of care in patients who have other treatment options (i.e., not restricted or required to include resistant isolates) (28, 29). Extrapolating results from noninferiority studies to patients in whom the control intervention would not be effective is difficult (15). In the present study, we, therefore, sought to augment data from these studies by evaluating C/T patterns of use, effectiveness, and adverse effects in routine clinical practice using a cohort of patients from a diverse range of academic and community medical centers across the United States. Patients enrolled in the present study represent a population that has been underrepresented in C/T randomized controlled trials (RCTs). Our patients were older, had multiple comorbidities, and had extensive prior health care and antibiotic exposures, and over 40% of them resided in the ICU at infection onset. P. aeruginosa was isolated from over 90% of patients, and the vast majority (95.8%) were MDR strains. Approximately one in four patients received combination antibiotic therapy, which was prohibited in RCTs, and many patients received doses that differed from the current FDA-approved doses.
With regard to the last point, approximately 70% of patients with a respiratory tract infection in our study received high-dose C/T. The FDA approval for the HAP/VAP indication, along with the modified dosing recommendations, came shortly after our study closed. We infer that the use of higher C/T doses was based on pharmacokinetic/pharmacodynamic data published in 2016 suggesting 3 grams every 8 hours may improve the probability of target attainment within epithelial lining fluid (30). The recommended C/T doses for HAP/VAP are also 2- and 3-fold greater than for the former indications in patients with CrCl of <50 ml/min and those receiving intermittent hemodialysis, respectively. However, the appropriate renally adjusted high-dose C/T was only used in 39.6% of patients with a respiratory tract infection who had their C/T dose renally adjusted. The higher rate of clinical failure observed in patients with renal impairment highlights the importance of dose optimization in this setting.
These observations also call attention to the potential tradeoffs of initial accelerated antibiotic approval based on indications that are not reflective of the medical need the antibiotic is targeted to meet. That is, antipseudomonal drugs are rarely indicated for cUTIs and cIAIs. On the other hand, P. aeruginosa is one of the primary pathogens responsible for HAP/VAP, and not surprisingly, respiratory tract infections have been the most common C/T indication in postapproval observational studies to date, including the present study (12, 13, 21, 23–26). The C/T HAP/VAP approval, along with a modified dosing recommendation, came nearly 5 years after initial approval (1).
Although C/T was mostly used for off-label indications in this study, it clearly addressed an unmet need in our patients. As noted previously, the vast majority of patients received C/T to treat P. aeruginosa infections, with 95.8% demonstrating the MDR phenotype. All patients without a positive culture also had a history of MDR or XDR P. aeruginosa infections. In agreement with previous surveillance data (8, 11), C/T was active against 88.7% of these highly resistant P. aeruginosa isolates. The only agent with greater in vitro activity was amikacin, which is not recommended for monotherapy in respiratory tract infections due to inferior clinical outcome versus beta-lactams, which is likely related to poor pulmonary penetration and diminished antibacterial activity in the acidic pneumonic airways (6, 31, 32). Aminoglycosides are also challenging to use in elderly patients, with a high burden of comorbidity and preexisting organ impairment, such as those in this study.
Although it is difficult to make comparisons across studies due to differences in study design and case mix, the rates of clinical failure (37.6%) and 30-day mortality (17.3%) in the present study are broadly comparable to previous observational studies describing the use of C/T for MDR P. aeruginosa infections and suggest meaningful progress for patients compared with historical controls (16, 21–25). Clinical outcomes among patients with a respiratory tract infection in this study were remarkably similar to those reported for the recently completed ASPECT-NP study, a randomized controlled phase 3 study comparing C/T to meropenem in patients with ventilated nosocomial pneumonia (14). In this RCT, 28-day mortality in the intention-to-treat population randomized to C/T was 24.0% (compared to 24.2% for 30-day mortality in the present study) (14). Although less than 5% of patients in the ASPECT-NP study were infected with MDR P. aeruginosa, the comparison suggests that patients in routine clinical practice are achieving expected outcomes. However, the fact that less than 40% of patients in our study originally admitted from home were discharged directly home again is a sobering reminder that there is still much work to be done to return patients to their baseline state of health after surviving a serious infection.
It is notable that in this study C/T was used earlier in the course of the infection (median, 85 h after infection onset) than earlier observational studies where C/T was often reserved for salvage therapy (21, 23). This may suggest that clinical laboratories are streamlining the C/T susceptibility testing process and is a positive signal considering that a number of studies have shown that the treatment of serious infections is time sensitive with negative consequences for delays in appropriate therapy (33–36).
P. aeruginosa is remarkable for its ability to acquire new resistance mechanisms under selective antibiotic pressure (3). Among patients with follow-up cultures in this study, three (9.7%) isolates developed new C/T resistance as early as 3 days after C/T initiation. Two of these patients showed signs and symptoms of worsening infection at the time resistance was detected, suggesting it does impact outcomes. Moving forward, it will be important to identify risk factors for resistance development and to determine what strategies, if any, are preventative. No patients who developed C/T resistance received combination therapy. Whether combination therapy could attenuate resistance development in some strains remains uncertain; however, it is clear that combination therapy carries definite risks of more adverse effects, as seen in this study and others (37, 38).
With regard to safety, our study provides important insights into potential C/T-related toxicities in real-world patients with multiple underlying risk factors. It is notable that all patients who experienced AKI on C/T were receiving other nephrotoxins and that the use of concomitant aminoglycoside or polymyxin therapy was significantly more common in these patients. This finding underscores the importance of limiting the administration of nephrotoxins whenever possible. Although we cannot exclude selection bias, combination antibiotic therapy was not associated with improved effectiveness, suggesting this common practice (approximately one in four patients in this study) should be reconsidered. The incidence of C. difficile infection in this study was over 16-fold higher than that in the C/T phase 3 cUTI and cIAI studies (3/1,015, 0.3% versus 13/259, 5.0%). It is not surprising that RCTs underestimate this adverse effect given the risk differences of the populations; in particular, 73% of patients in our study had recent antibiotic exposure, which is typical for patients with MDR infections (12, 13, 16). To the best of our knowledge, we are the first to report potential C/T-related encephalopathy in two patients. Neurotoxicity has been reported with virtually all cephalosporins and can range from mild headache and confusion to seizures (39). Consistent with the cases in this study, older age, higher doses, and renal impairment have been identified as risk factors (39). Data regarding C/T central nervous system (CNS) penetration have not yet been reported; however, the use of C/T for the treatment of meningitis has been described (40, 41). Although there are a wide spectrum of causes for CNS disturbances in patients with serious infections, our findings suggest that C/T-associated CNS disturbances should be considered in the differential for at-risk patients.
This study has important limitations. First, the study is subject to inherent biases and limitations with its retrospective design. Treatment-related factors, such as the time from infection onset to C/T initiation, C/T dose, and the use of combination therapy, were not assigned randomly, and it is, therefore, difficult to determine how these factors affected outcomes. Important information, such as the results of follow-up cultures, was available for only a minority of patients. Follow-up testing in this study is reflective of real-world practice. The fact that a greater proportion of patients who went on to experience clinical failure had follow-up cultures collected suggests that the incidence of resistance development found in this study (9.7%) may be an overestimate. Additionally, even though this represents the largest study to date evaluating the use of C/T for MDR infections, the sample size was still relatively small, limiting our ability to conduct meaningful subgroup analyses. We did not include a contemporary control group, and although our results suggest improvements over historical outcomes in patients with MDR P. aeruginosa infections, such comparisons are fraught with limitations due to changes over time in referral patterns, diagnostic modalities, ancillary care, and the underlying health of the population (42). Comparative outcome research of newer antibiotics, preferably in the form of prospective RCTs, designed, conducted, analyzed, and reported by independent groups without competing interests, is desperately needed.
In conclusion, this study adds considerably to the growing body of literature describing C/T treatment patterns and outcomes for MDR infections. Our study suggests that C/T can be an effective antibiotic for patients with limited treatment options. We describe hitherto unrecognized safety signals that may prompt increased vigilance and earlier detection. This study also identifies patient groups at higher risk for poor outcomes, such as those with renal impairment and critical illness, for whom continued advancement is needed.
MATERIALS AND METHODS
Study design and population.This was a multicenter, retrospective, noncomparative cohort study conducted at eight academic and community medical centers in the United States between 2015 and 2019. Inclusion criteria were the following: (i) age of ≥18 years and (ii) receipt of ≥72 hours of C/T for suspected or confirmed MDR Gram-negative infection. For each patient, only the initial eligible C/T treatment course during the study period was included.
Ethics.Approval was obtained from each medical center’s institutional review board with a waiver for informed consent.
Data collection and study definitions.Pharmacy records were screened for all patients who received at least one dose of C/T during the study period. Relevant demographic, clinical, microbiological, and treatment data were extracted from the electronic medical record and entered into a secure online data collection form (43). Comorbidity burden was quantified using the Charlson comorbidity score (44). The severity of illness at infection onset was assessed using the SOFA and APACHE II scores (28, 45). Infection onset was considered to be the time that the index culture was collected or, for patients that did not have cultures collected, when signs and symptoms were first documented. Sources of infection were based on available clinical, microbiological, and diagnostic data. The infection was considered hospital acquired if the index culture was obtained greater than 48 h after admission (31). Bacterial identification and antibiotic susceptibilities were performed at each center according to standard procedures. C/T susceptibility was determined using disk diffusion or gradient strips, when available. MDR P. aeruginosa was defined by nonsusceptibility to at least one antibiotic in at least three classes that are typically active against wild-type P. aeruginosa (29). Extensively drug resistant (XDR) was defined as nonsusceptible to at least one antibiotic in all but two classes (29). Carbapenem-resistant Enterobacterales (CRE) was defined by current U.S. Centers for Disease Control and Prevention (CDC) criteria (7). Standard- and high-doses were defined as 1.5 g of intravenous (i.v.) C/T every 8 hours and 3 g of i.v. C/T every 8 hours, respectively, with dose adjustments for renal impairment according to the manufacturer’s recommendations (1). For the purposes of this study, C/T combination therapy was defined as the receipt of a concomitant i.v. antipseudomonal antibiotic for ≥48 hours with C/T. Microbiological failure was defined as microbiologically confirmed recurrence after 7 days of C/T therapy to the end of follow-up plus signs and symptoms of infection. Patients were followed for 30 days after hospital discharge. Clinical failure was defined as a composite of all-cause 30-day mortality, microbiological failure, and/or failure to resolve or improve signs and symptoms of infections during C/T therapy. Acute kidney injury (AKI) was evaluated in patients not receiving hemodialysis at the start of C/T and was defined as a serum creatinine increase of ≥0.5 mg/dl or 50% from baseline on two consecutive measurements while on C/T and up to 72 h following the last dose. Hepatotoxicity was defined as previously described (14).
Statistical analysis.Baseline characteristics of the overall cohort and in the subgroup of patients with MDR P. aeruginosa infections were evaluated using descriptive statistics; categorical data were reported as counts and percentages, and continuous data were reported as medians and interquartile ranges (IQRs). The primary efficacy outcome was composite clinical failure in patients from whom MDR P. aeruginosa was isolated. An additional post hoc analysis was also conducted in patients with an MDR P. aeruginosa isolate confirmed to be susceptible to C/T. Multivariable logistic regression analysis was performed to identify independent predictors of clinical failure. Clinically relevant variables were selected for model entry based on bivariate comparisons (P < 0.2) and biological plausibility. Some variables were collapsed into single composite variables when the number of patients in subgroups was too small to allow for meaningful analysis. The selected model was simplified based on the Akaike information criterion (AIC) in backward fashion. Multicolinearity of candidate regression models was assessed via the variance inflation factor, with values less than three considered acceptable. Secondary outcomes of interest included individual components of the composite outcome, discharge disposition, emergence of C/T resistance, and hospital length of stay. Safety outcomes were evaluated in the total cohort and included AKI, dermatological reactions, gastrointestinal intolerance, cytopenias, central nervous system disturbances, and Clostridioides difficile-associated diarrhea.
All analyses were performed using SPSS Statistics version 25.0 (IBM Corp., Armonk, NY) and SAS 9.4 Statistical Software (SAS Institute Inc., Cary, NC). A two-tailed P value less than 0.05 was statistically significant.
ACKNOWLEDGMENTS
This study was funded by an investigator-initiated grant from Merck.
The following authors disclose financial or other relationships relevant to the study: M.J.R. (research support; consultant or speaker for Allergan, Melinta, Merck, Motif, Nabriva, Paratek, Tetraphase, and Shionogi), J.R.R. (consulting agreements or is on the speaker’s bureau with Allergan, Merck, Shinogi, Tetraphase, Melinta, and Paratek), S.L.D. (consultant for Allergan, Spero, and Tetraphase), and S.J.E. (employee of T2 Biosystems). All other authors have nothing to disclose.
FOOTNOTES
- Received 16 November 2019.
- Returned for modification 14 December 2019.
- Accepted 5 January 2020.
- Accepted manuscript posted online 13 January 2020.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
