Real-world use of ceftolozane/tazobactam: a systematic literature review

Antibacterial-resistant gram-negative infections are a serious risk to global public health. Resistant Enterobacterales and Pseudomonas aeruginosa are highly prevalent, particularly in healthcare settings, and there are limited effective treatment options. Patients with infections caused by resistant pathogens have considerably worse outcomes, and incur significantly higher costs, relative to patients with susceptible infections. Ceftolozane/tazobactam (C/T) has established efficacy in clinical trials. This review aimed to collate data on C/T use in clinical practice. This systematic literature review searched online biomedical databases for real-world studies of C/T for gram-negative infections up to June 2020. Relevant study, patient, and treatment characteristics, microbiology, and efficacy outcomes were captured. There were 83 studies comprising 3,701 patients were identified. The most common infections were respiratory infections (52.9% of reported infections), urinary tract infections (UTIs; 14.9%), and intra-abdominal infections (IAIs; 10.1%). Most patients included were seriously ill and had multiple comorbidities. The majority of patients had infections caused by P.aeruginosa (90.7%), of which 86.0% were antimicrobial-resistant. C/T was used as both a 1.5 g q8h and 3 g q8h dose, for a median duration of 7–56 days (varying between studies). Outcome rates were comparable between studies: clinical success rates ranged from 45.7 to 100.0%, with 27 studies (69%) reporting clinical success rates of > 70%; microbiological success rates ranged from 31 to 100%, with 14 studies (74%) reporting microbiological success rates of > 70%. Mortality rates ranged from 0 to 50%, with 31 studies (69%) reporting mortality rates of ≤ 20%. In comparative studies, C/T was as effective as aminoglycoside- or polymyxin-based regimens, and in some instances, significantly more effective. The studies identified in this review demonstrate that C/T is effective in clinical practice, despite the diverse group of seriously ill patients, different levels of resistance of the pathogens treated, and varying dosing regimens used. Furthermore, comparative studies suggest that C/T offers a successful alternative to standard of care (SoC).


Background
Antibacterial resistance is a serious risk to global public health. The problem of resistance is especially acute for gram-negative pathogens [1]. Enterobacterales and Pseudomonas aeruginosa are the most prevalent gram-negative hospital-acquired infections (HAIs), collectively accounting for 30% of all HAIs in the United States (US) [2]. Patients in intensive care units (ICUs) are particularly vulnerable to gram-negative infections and accounts for 70% of the HAIs acquired in ICUs [2][3][4].
The burden of infections caused by these pathogens are intensified because of limited effective treatment options. Pathogen susceptibility to many of the available gram-negative antibacterial agents have diminished Open Access *Correspondence: laura.puzniak@merck.com 1 Merck & Co., Inc., 2000 Galloping Hill Road, Kenilworth, NJ 07033, USA Full list of author information is available at the end of the article over time [5]. Patients with infections caused by resistant pathogens have considerably worse outcomes relative to their susceptible counterparts [6,7]. In a US national database study, patients with multidrug-resistant (MDR) P. aeruginosa respiratory infections had higher mortality, an approximately 7-day longer length of stay (LOS), $20,000 excess costs, higher readmission rates, and > $10,000 excess net loss per case for the hospital relative to those with non-MDR P. aeruginosa infections [7]. Further, when the infection is caused by resistant pathogens, it increases the likelihood for receipt of initial inappropriate antibacterial therapy, which has been shown to diminish clinical outcomes and increase costs [8,9].
The challenge of resistance and deleterious impact on outcomes is further compounded by the serious drugrelated toxicity associated with some of the current treatment options for resistant gram-negative pathogens. Aminoglycosides (e.g. gentamicin, tobramycin and amikacin) and polymyxins (e.g. colistin) are reported to cause nephrotoxicity and/or ototoxicity [10,11]. Although these antibacterial agents tend to have higher susceptibility to many gram-negative pathogen, they come at a cost of toxicity.
Due to this imminent threat of drug-resistant Enterobacterales and P. aeruginosa, and the limited treatment options and toxic effects of some antibacterial agents, the World Health Organization (WHO) in 2017 designated both Enterobacterales and P. aeruginosa as the highest 'critical' priority in need of new therapies to counteract this crisis [12].
Ceftolozane/tazobactam (C/T) is a β-lactam/βlactamase inhibitor antibacterial agent, consisting of a fixed (2:1) combination of an antipseudomonal cephalosporin, ceftolozane, and the well-established β-lactamase inhibitor, tazobactam [13]. C/T is approved in the US and Europe for clinical use in adults with complicated urinary tract infections (cUTIs), including pyelonephritis, complicated intra-abdominal infections (cIAIs), and hospital-acquired bacterial pneumonia/ventilator-associated bacterial pneumonia (HABP/VABP) [13,14]. The approval of C/T was supported by three multinational, randomized, double-blind, active comparator-controlled trials: ASPECT-cUTI, ASPECT-cIAI and ASEPCT-NP [15][16][17]. In the ASPECT trials, C/T demonstrated superiority over levofloxacin (ASPECT-cUTI), and noninferiority to meropenem (ASPECT-cIAI and -NP) [15][16][17]. Since launch in 2014, real-world evidence (RWE) for the use of C/T in clinical practice has been accumulating. The purpose of this systematic literature review (SLR) was to identify and collate published RWE to better understand the characteristics of patients treated with C/T and clinical outcomes.

Literature search
A search of the literature for C/T RWE, published between 1st January 2009 and 3rd June 2020, was conducted in the following biomedical and economic databases via the OVID platform: Embase, MEDLINE, PsycInfo, Econlit, and EBM Reviews (ACP Journal Club, Cochrane Database of Systematic Reviews, Cochrane Methodology Register, Database of Abstracts of Reviews of Effects, Health Technology Assessment, NHS Economic Evaluation Database, Cochrane Clinical Answers). The search was conducted in January 2019 with a 10-year time horizon, then re-ran to capture literature published between January-November 2019, and November 2019-June 2020. The time horizon was chosen to minimize erroneous data identification given that C/T was approved for use in 2014-using a longer horizon would capture any publications reporting on expanded access or compassionate use. The search was limited to English Language publications only.
Due to the heterogeneity of reporting of RWE, the search was designed to be broad to ensure relevant studies which may not be appropriately indexed were retrieved. Table 1 details the search strategy. A further search of internet-based sources relating to C/T RWE was also conducted (limited to English language only). This gray literature review involved searching conference proceedings of two conferences-European Congress of Clinical Microbiology and Infectious Diseases [ECCMID] and Infectious Disease Week (IDWeek)-two of the largest infectious disease conferences in Europe and the US. Conference proceedings, when published as part of an abstract book, were also identified during the OVID search.

Study selection
All screening (by title and abstract, and by full-text) was performed by two reviewers and any uncertainties were resolved by a third reviewer. Predetermined inclusion and exclusion criteria were used to assess the eligibility of identified abstracts and full-texts for inclusion. PICOS eligibility criteria for study inclusion included observational and non-controlled studies reporting on the use of C/T to treat adult patients (≥ 18 years of age) with gramnegative infections in real-world clinical practice. Only studies in English were included. Studies were excluded if they did not meet the PICOS criteria, such as randomized controlled trials (RCTs) or other randomized or controlled experimental studies. A complete description of the PICOS criteria is provided in Additional file 1: Table S1.

Data extraction and analysis
Relevant study, patient, and treatment characteristics, microbiology, and efficacy outcomes were extracted into a data extraction form by one reviewer and checked by a senior reviewer. Efficacy outcomes included clinical cure (typically defined as the resolution of signs or symptoms of infection following therapy and survival), microbiological cure (typically defined as large reduction or eradication in the number of pathogens following therapy), and mortality.

SLR results
A total of 1,222 records were identified from the database searches, and 23 records were identified from the gray literature search. This resulted in 874 non-duplicate records that were subject to title and abstract screening. A total of 730 records were excluded according to the PICOS criteria and 144 were included for full-text review. Of these, 83 studies were determined to be eligible for data extraction and qualitative synthesis. The results of the SLR and study selection processes are presented in Fig. 1.

Patient characteristics
Identified studies included a total 3701 distinct patients treated with C/T. Excluding the single-patient case reports, the median number of patients included was 30 (range: 2(100)-1490 (87)). Patient populations were heterogeneous, with a number of different sources of infections and pathogens reported. There were 3,735 total infections. Of these, there were 1807 infections where the source of infection was not reported (48.4%); excluding those publications, the most common source of infection(s) were pneumonia/respiratory tract infections (RTIs; 52.9% of reported infections), UTIs (14.9%), and IAIs (10.1%). There was also report of C/T use in SSTIs (7.1%), bone and joint infections (6.1%), and primary bacteremia (4.2%). Over time, the number of patients treated with C/T has grown, but the proportion of each infection type has remained relatively consistent (Fig. 2). The number of patients treated for RTIs was consistently high over the time period studied ( The patient population included in these RWE publications were often classified as seriously ill with multiple comorbidities. In total, 1,751 patients (47.3% of 3,701 patients reported) were admitted to the ICU. The literature review recorded three commonly used measures of patient illness severity-Acute Physiology and Chronic Health Evaluation (APACHE) II, Sequential Organ Failure Assessment (SOFA), and Charlson Comorbidity (CC) index. APACHE and SOFA are systems for predicting ICU mortality. Nine publications, comprising 794 patients treated with C/T, reported APACHE scores ranging from 13 to 40, with larger studies (> 50 patients) ranging from 18 to 40 [22,24,33,36,80,84,90,97,100]. Six publications, comprising 472 patients treated with C/T, reported SOFA scores ranging from 3 to 8 [22,26,27,30,38,39]. The CC index quantifies the comorbidity burden of included patients by predicting the mortality of patients with multiple comorbidities. Twenty-one publications, comprising 2930 patients, reported CC index scores ranging from 2 to 6 [18, 22, 24-28, 30-33, 39, 40, 80, 84, 86, 87, 90, 96, 97, 100]. These measures show the high severity of illness of patients included in the RWE of C/T treatment.
A total of 1,294 (35.0%) patients did not have a causative pathogen specified (note that the majority of these came from a single publication [87]. For publications that reported a causative pathogen, the majority of patients (90.7%; N = 2,184) had infections that were caused by P. aeruginosa, of which 14.0% were caused by non-drug resistant P. aeruginosa, or the level of resistance was not specified, 72.3% by MDR P. aeruginosa, 13.4% extensively-drug-resistant (XDR), and 0.2% pan-drug-resistant (PDR). Note that the level of resistance specified (MDR/ XDR/PDR) was recorded as described in the publication.          Resistant infections comprised the majority of infections treated in studies published in the first three-year period captured (2015-2017) vs. the second three-year period (Fig. 3).

Treatment characteristics
C/T is indicated for use at two doses: either 1.5 g q8h (for cIAI and cUTI) or 3 g q8h (for patients with HABP/ VABP). For patients with renal insufficiency, doses are reduced according to level of creatinine clearance. In studies that reported dosing information (N = 1,418 patients), C/T was used as a 1.5 g q8h dose in 619 (43.7%) patients, as a 3 g q8h dose in 621 (43.8%) patients, and as a creatinine clearance adjusted dose in 178 (12.6%) patients. Note, however, that reporting of dosing was inconsistent between studies and the specific dose by type of infection (i.e., 3 g q8h for respiratory) was not always delineated. Of studies that reported the timing of C/T treatment (N = 893 patients), C/T was administered empirically (i.e. prior to susceptibility results) in 222 (24.9%) patients and administered confirmed (i.e. following susceptibility results) in 671 (75.1%). There was little year-on-year change in the proportion of patients treated empirically or confirmed, or treated with a 1.5 g q8h or 3 g q8h regimen-despite the approval of the 3 g q8h dose in 2019.
There was large variation in the duration of C/T therapy reported, often different to the label dose of 4-14 (cIAI), 7 (cUTI), or 8-14 (HABP/VABP) days. In all studies, the median duration of C/T therapy ranged from 7 to 56 days, irrespective of dose. Median duration in larger studies (> 50 patients) ranged from 8 to 16.1 days, consistent with the indicated duration. Excluding single-patient case reports, 12 studies (231 patients) reported an average duration of C/T exceeding the label maximum dose of 14 days; with three studies (26 patients) reporting an average duration of > 28 days. Of these three studies, two included patients with osteomyelitis and MDR P. aeruginosa infection [24,29], and one included patients with severe infections caused by MDR or XDR P. aeruginosa [42]. Furthermore, 15 single-patient case reports reported C/T durations exceeding the maximum label dose, with 12 reporting a duration of > 28 days, and three reporting a duration of > 42 days. All three that reported a duration of > 42 days administered 8-week courses of C/T [56,63,67]. Two patients received C/T for XDR P. aeruginosa osteomyelitis [56,67], and one received C/T for MDR P. aeruginosa mycotic pseudoaneurysm [63]. All patients had these infections following surgery.
Outcomes were consistent in the 36 single-patient case reports-clinical cure was reported in 28 of 32 studies (87.5%), microbiological cure in 18 of 23 studies (78.3%), and mortality in 4 of 32 studies (11.4%).

Outcomes by treatment characteristics
Seven studies reported on the treatment characteristics that were risk factors for clinical outcomes [18,25,28,30,32,33,39]. Patient cohort size ranged from 21 to 205, with a median of 90. Five studies included patients with P. aeruginosa infections [25,28,30,32,33,39]; one included patients with Enterobacterales infections [18]. There was a diverse range of infection types included.
Five studies found mixed evidence that a delay in receipt of C/T led to worse outcomes [18,28,30,33,39]. Bassetti et al. 2020 found that a significantly higher proportion of patients who achieved clinical success received empiric C/T and had a significantly shorter latency between infection onset and C/T administration (both p < 0.001) [18]. Similarly, Gallagher et al. 2018 found that starting C/T less than four days after positive culture was associated with significantly higher clinical and microbiological cure rates, and that starting C/T more than four days after positive culture was associated with significantly higher mortality [33]. In contrast, three studies found no association between initiating C/T within 48 h of P. aeruginosa isolation, time to C/T, or type of treatment (empiric, semi-empiric, or confirmed) (all p > 0.05) [28,30,39]. These three studies were of smaller size (169 combined patients vs. 258 for the two previously mentioned studies), and, importantly, Rodriguez-Nunez et al. included some patients that were also reported in Diaz-Cañestro et al. 2018, effectively double-counting these patients and possibly giving them disproportionate influence over the conclusion drawn in this review [28,30].

Outcomes by PsA resistance subtype
Two studies were identified that conducted an analysis to understand whether P. aeruginosa resistance was a factor in clinical outcome [28,30]. In univariate analysis, Rodriguez-Nunez et al. found that similar proportions of survivors and non-survivors had XDR PsA infections [28]. Whereas, Diaz-Cañestro et al. found that resistance profile (the proportion of patients with MDR vs. XDR infections) was significantly different between patients who were clinical successes or failures (Table 3) [30].
In the three studies with aminoglycoside-/polymyxinbased comparators, all reported mortality rates [23,27,80], two reported clinical cure rates [23,27], and one reported microbiological cure rate [80]. In Pogue et al., patients treated with C/T had significantly higher clinical cure rate (p = 0.002), but there was no difference in in-hospital mortality [27]. In response, Vena et al. conducted a similar case-control study, but balanced the proportion of patients with pneumonia in each arm, ensured patients received a sufficient polymyxin dosage, and ensured that all included patients had an infectious disease consultation [23]. Results were comparable with Pogue et al.-patients treated with C/T had a numerically higher clinical cure rate and lower mortality rate than patients treated with aminoglycoside/polymyxin regimen, though this did not reach statistical significance [23]. Caffrey et al. showed that patients treated with C/T were significantly less likely to die as inpatients than patients treated with aminoglycoside/polymyxin-based regimens, although there was no difference in 30-day mortality rates or microbiological cure rates, and clinical cure rates were not reported [80].
In the two studies that compared patients treated with C/T with mixed SoC antibacterial agents, both reported clinical cure rates and mortality [26,83]. Both studies found that patients treated with C/T had numerically higher clinical cure rates than patients treated with other antibacterial agents. Fernández-Cruz et al. additionally found that patients treated with C/T had significantly lower mortality rates (p < 0.05) [26]; such a difference was not apparent in Mills et al. [83].

Discussion
The principal finding of this SLR was that there is a body of RWE that establishes the effectiveness of C/T in realworld clinical practice, including patients described as severely ill patients and/or with resistant infections. Considering the patient disease severity measures, publications reported APACHE scores ranging from 13 to 40, with larger studies (> 50 patients) ranging from 18 to 40 [22,24,33,36,80,84,90,97,100]. This is higher than the APACHE score reported in ASPECT-NP (median 17) [15], and significantly higher than reported in ASPECT-IAI (mean 6.2) [16]. Furthermore, inclusion of immunocompromised patients, typically excluded by clinical trials, offers valuable insights into C/T effectiveness in this underrepresented population. A key limitation of many clinical trials is the exclusion of these seriously ill patients, and the restriction of recruitment to only patients with a narrow range of infections. By filling this gap, the RWE therefore provides valuable data on the outcomes of these patients seen in clinical practice. Despite the heterogeneity in the patient population, outcomes of treatment with C/T were consistent with those found in the ASPECT clinical trials. In larger RWE studies (> 50 patients), clinical cure rates ranged from 56.7 to 83.7%, microbiological cure rates ranged from 31 to 75.3%, and mortality rates ranged from 5 to 29%. By way of descriptive comparison, C/T outcomes in the ASPECT trials were: ASPECT-cUTI, clinical cure = 92.0%, microbiological eradication = 80.4%, and mortality = 0.2% [17,101]; ASPECT-cIAI, clinical cure = 83.0%, microbiological cure = 85.3%, and mortality = 2.3% [16,102]; and ASPECT-NP, clinical cure = 54.4%, microbiological eradication = 73.1%, and 28-day mortality = 24.0% [15].
Treatment characteristics were broadly aligned with the approved use of C/T and both indicated doses of C/T were used approximately equally; however, it was unclear which dose was used for which indication and often the outcomes were not stratified by dose and indication. This result is concerning since the indicated dose for pneumonia is based on optimized pharmacokinetic and pharmacodynamic properties. C/T was more commonly used as confirmed therapy than as an empiric therapy (75.1% vs. 24.9%). This is consistent with the principles of antimicrobial stewardship, whereby broader-spectrum antibacterial agents are reserved for special clinical situations when other treatments have failed. However, there were two studies that suggested patients who were treated earlier; either empirically, or sooner after infection onset, had better clinical outcomes [18,33], Although a similar association was not found in three other studies [28,30,39], comparison of early vs. late use of C/T warrants further investigation. Late use of C/T may be indicative of initial inappropriate antibacterial therapy with other agents, which has been shown in the literature to have deleterious effects on outcomes [8,9]. Data from the comparative studies suggest that C/T is at least as effective as, and in several cases, significantly better than, aminoglycoside-or polymyxin-based regimens for serious, MDR infections [23,27,80]. Outside the scope of this review, though pertinent to clinicians, is the lower risk of nephrotoxicity with C/T compared to aminoglycosides or polymyxins. Both comparative studies that assessed safety found a significantly lower incidence of acute kidney injury with C/T than with aminoglycoside/polymyxin-based comparators [23,27]. This combination of comparable effectiveness and lower risk of nephrotoxicity means that C/T can be an alternative to these therapies, particularly in patients with decreased renal function.
This SLR highlights the inconsistent reporting that is common within published RWE. Due to differences in study design, objectives, outcome assessment and definitions, there were often incomplete data for the variables of interest, as set out in this SLR. This variability in turn imposes challenges in attributing outcomes to the exposure studied. The inclusion of conference proceedings, which are not subject to the same rigorous peer-review, may have affected evidence included within this review, and thus the conclusions drawn. As mentioned in the results, some studies included data that were reported in part by other studies-this may be more widespread than thought as some large database studies collected patients across hundreds of hospitals, possibly capturing patients reported in other studies. As this is a qualitative review, this double counting was not adjusted for. However, given the consistency of outcomes between studies conducted in different locations, in different years, and by different authors, it is likely that the outcomes reported approximate the true treatment effect.
As was to be expected, many studies had small sample sizes and did not include comparison groups for statistical inference purposes. In the comparative cohort studies that did, C/T had comparable efficacy to standard of care, and was significantly better in several outcomes. Furthermore, identified risk factors may have been subject to a reporting bias: with some studies only reporting multivariate analysis, it was difficult to recognize which risk factors were non-significant, and therefore excluded, in univariate analysis. Moreover, the vast majority of publications were of a retrospective design. This may lead to selection bias, as both exposure and outcome of patients are already known. Many studies had industry authors and/or were sponsored by grants from industry which may lead to publication bias; however, the results appeared consistent regardless of authorship or sponsorship. Further publication bias may have arisen due to potential nonpublication of negative results. Schumucker et al. found some evidence that meta-analyses which do not include unpublished or grey literature studies overestimate the treatment effect [103]. To mitigate this, this review included a search of recent ECCMID and IDWeek conference proceedings-two of the largest microbiology conferences in the US and Europe-to identify grey literature studies. However, this review did not include a comprehensive search of all relevant microbiology conferences or search for studies that were unpublished or preprints. Though these are pragmatic limitations associated with all literature reviews, there remains a possibility that the studies included in this review overestimate the treatment effect.
In conclusion, this SLR identified and summarized the published RWE on the use of C/T in clinical practice. These studies demonstrate the clinical effectiveness of C/T, despite the diverse group of seriously ill patients and level of resistance of the pathogens treated. The RWE body of literature provides additional insights into patient types that are commonly encountered in everyday practice and may have been excluded from the registration trials. Further studies are needed that evaluate homogenous patient sub-types and that account for other treatments that were received prior to C/T to properly attribute outcomes to the effectiveness of C/T.