This is a retrospective chart review conducted at Makassed General Hospital, a 200-bed university hospital located in Beirut, Lebanon, with a 13-bed critical care unit, between January 2012 and December 2013. It included 153 adult patients who received tigecycline for more than 72 h. The hospital's Institutional Review Board committee approved this study. All patients received the standard FDA-approved dosage of tigecycline: a loading dose of 100 mg, followed by 50 mg administered intravenously every 12 h.

We recorded baseline clinical and demographic characteristics, microbiological findings, indication for tigecycline therapy, concomitant use of other antibiotics, clinical and microbiological outcome, and mortality. FDA-approved indications were CAP, cSSTI, and cIAI. Off-label indications were HAP, VAP, UTI, diabetic ulcers, sepsis, bacteremia, and febrile neutropenia. Severity of underlying disease was assessed with sequential organ failure assessment (SOFA) score for critically ill patients.

Infections were defined according to clinical diagnostic criteria established by the Center of Disease Control and Prevention and other international guidelines (Garner et al., 1988; Centers for Disease Control Prevention, 2012; Dellinger et al., 2013). Infections were classified as nosocomial acquired when onset occurred more than 48 h after hospital admission, and the infection was not present or incubating at that time (Garner et al., 1988). Infections occurring within 48 h of admission to the hospital were considered community-acquired, unless the patient had been transferred directly from another healthcare facility or discharged within the 30 days preceding hospital admission (Garner et al., 1988).

Empiric use of tigecycline was defined as its administration to a patient with signs and symptoms of infection without an identified source or a specific microbiological isolate (Bassetti et al., 2010). Targeted therapy was defined as tigecycline administration in the presence of an identified isolate (Bassetti et al., 2010).

As recommended by the World health Organization (WHO), tigecycline consumption was reported in Defined Daily Doses (DDD)/100 Bed Days (BD), a standardized figure that provides a degree of comparison among in-patients in different hospitals (World Health Organization, 2003). DDD corresponds to the assumed average daily dose of a specific drug for its main indication in adults (World Health Organization, 2003). Number of BD = Number of beds × occupancy index × Number of days (during that period). A bed-day corresponds to one occupied hospital bed during 1 day (World Health Organization, 2003).

Bacterial identification was performed according to standard microbiological procedures. Antibiotic susceptibility was performed using the disc diffusion method as recommended by the Clinical and Laboratory Standards Institute (CLSI) for all organisms (CLSI, 2016). Quality-control strains used in the testing of Acinetobacter species and P. aeruginosa, and the Enterobacteriaceae were E. coli American Type Culture Collection (ATCC) 25922 and P. aeruginosa ATCC 27853. Acinetobacter species susceptibility was tested against all antimicrobials enlisted in CLSI guidelines (CLSI, 2016) except for ampicillin-sulbactam since it is non-available in Lebanon. Extensive drug resistance in A. baumannii was defined as non-susceptibility to at least one agent in all but susceptible to only one or two categories (Magiorakos et al., 2012). In our center, isolates were considered XDR if they were susceptible to tigecycline and colistin only and resistant to the following categories: aminoglycosides, antipseudomonal carbapenems, antipseudomonal FQs, extended-spectrum cephalosporins, antipseudomonal penicillins plus beta-lactamase inhibitors, monobactams and trimethoprim/sulfamethoxazole (TMP/SMX).

A 15-μg disc of tigecycline (Oxoid Ltd, Cambridge, UK) was used to determine susceptibility, and the breakpoints suggested by the FDA for Enterobacteriaceae (susceptible ≥19 mm; intermediate 15–18 mm, and resistant ≤ 14 mm), and gram-positive microorganisms (susceptible ≥19 mm) were used (Tygacil package insert, June 2005, Wyeth Pharmaceuticals, Inc., Philadelphia, PA). Tigecycline clinical breakpoints using the disc-diffusion method against A. baumanii have not been established by the CLSI (2016) or by the European Committee on Antimicrobial Susceptibility Testing (The European Committee on Antimicrobial Susceptibility Testing, 2016). Therefore, we applied the clinical breakpoints suggested by Jones et al. (2007) (susceptible ≥16 mm, intermediate 13–15 mm, and resistant ≤ 12 mm). Our medical laboratory department does not routinely perform broth microdilution method and a reference laboratory is not available in our country. However, during the study period, 11 randomly selected strains of A. baumanii were sent to another clinical microbiology laboratory (Faculty of Medicine and Medical Sciences, University of Balamand, Amioun, Lebanon) at a rate of one strain per 45 days when available, where broth microdilution method was performed for evaluating colistin and tigecycline susceptibility. We found 100% correlation between results of disc diffusion and broth microdilution (Supplementary File 1).

Clinical success was defined as a positive response (partial or complete improvement of signs/symptoms of infection) and the lack of need to use a new antibiotic for the initial infection for 72 h after the discontinuation of tigecycline (Kuo et al., 2011; Montravers et al., 2014). Persistence or deterioration of the initial infection symptoms/signs requiring a change of antibiotic therapy, an infection-related death occurring later than 48 h after the start of tigecycline, and/or premature treatment discontinuation because of a tigecycline-related adverse event were considered clinical failures (Montravers et al., 2014; Kuo et al., 2011). Cases of insufficient data, death not directly related to the initial infection or occurring within the first 48 h of treatment, or the addition of an antibacterial agent for another infection (Montravers et al., 2014) were considered to be an undetermined clinical response.

Microbiological success was defined as the eradication/sterile culture results during or after the course of antibiotic therapy (Montravers et al., 2014). The persistent identification of the same organism for 72 h after initiation of antibiotic treatment was considered a microbiological failure (Montravers et al., 2014). Superinfection was defined as a new isolate obtained during the treatment of the original infection with development of symptoms and signs of clinical infection (Kuo et al., 2011). A specimen that was not available for estimation of eradication was considered an unavailable response (Montravers et al., 2014). Because our study was retrospective, microbiological success rates were calculated when follow-up cultures were available.

Seven days post-treatment mortality, a surrogate marker for treatment failure, was defined as death occurring within 48 h up to 7 days after the end of tigecycline therapy. In-hospital mortality, a surrogate marker of the general patient condition, was defined as the death occurring between 7 and 30 days after finishing tigecycline therapy. Total mortality was the sum of both 7 days post-treatment mortality and in-hospital mortality.

The Statistical Package for Social Sciences program (IBM SPSS Statistics for Windows, Version 23.0, IBM Corp., Armonk, NY, USA) was used for data entry, management, and analyses. Descriptive analysis was carried out for categorical variables by number and percent. We assessed through univariate analysis the significance of baseline characteristics, comorbidities, types of infections, treatment strategy, outcome and mortality in patients who received tigecycline inside vs. outside critical care, and those with SOFA score <7 vs. those with SOFA score ≥7 inside ICU. Univariate analysis for the groups was carried out using the Chi square test for categorical variables, and independent t-test for continuous ones. This was followed by multivariate logistic regression for outcome and mortality to assess further significance between the groups, while controlling for potentially confounding variables. Parameters with P < 0.05 at the univariate level and other clinically significant variables were included in the multivariate model. Results were presented as odds ratio (OR) and 95% confidence interval (CI). We also performed univariate and multivariate analyses for outcome and mortality in FDA-approved indications vs. off-label indications.

In total, 153 patients received tigecycline therapy within our healthcare facility between January 2012 and December 2013. The clinical characteristics of the patients are reported in Table 1. Most of them were elderly, with an age above 65 years (53%). This population suffered from multiple comorbidities, including cardiovascular disease (59%), diabetes (30%), acute renal failure (23%), and hemodynamic insufficiency (14%). One third of our patients (30.7%) were critically ill, thus justifying the use of intravenous catecholamines (24%) and mechanical ventilation (23%).

Critically ill patients had a significantly higher incidence of cardiovascular disease (77%, P = 0.003) compared to those outside critical care area. Also, the need for vasopressors and mechanical ventilation was equally higher in this category (38%, P = 0.01 and 36%, P = 0.01, respectively) compared with the other group. In the critically ill group, the incidence of cardiovascular disease, acute renal injury, as well as the need for vasopressors was significantly higher in the sicker category of these patients, i.e., those with SOFA ≥7 (87%, P = 0.03; 37%, P = 0.02 and 53%, P = 0.05, respectively) in comparison to those with a SOFA score <7.

During 2012 and 2013, respectively, tigecycline consumption reached 2.6 DDD/100BD and 4 DDD/100 BD in our facility. The majority of the patients received tigecycline for a duration of between 72 h and 10 days (64%), with a median duration of 8 days (Table 2). According to each indication, the median duration of therapy varied between 6.5 and 9.5 days. The range varied from 3 to 30 days. The longest duration of tigecycline therapy was seen in cases with febrile neutropenia (median 7 days; range 3–30 days; Table 2).

During the study period, the number of patient days was 89,800 in the hospital and 2777 in the critical care unit. Tigecycline was prescribed much more outside the critical care unit than inside it (106 patients vs. 47 patients, respectively; Table 1). However, on adjusting these values to patient days, the number of tigecycline cycles was 16/1,000 patient days inside the critical care unit and 1.2/1,000 patient days outside it. Inside the critical care area, tigecycline was mostly prescribed to patients with a SOFA score ≥7 (30/47 patients, 64% of the critically ill category). These patients have been at a higher risk of contracting multidrug-resistant A. baumannii infection compared with patients. According to our infection control data, it has been endemic in the critical care area of our facility.

Tigecycline was prescribed more in off-label indications (124/153 patients, 81%) than in FDA-approved indications (29/153 patients, 19%; Table 1).

Tigecycline was used as targeted therapy in 56% of the cases (86/153 patients) and as empiric therapy in 44% (67/153 patients; Table 1). Empiric use was significantly higher outside critical care (49%, P = 0.049), unlike targeted therapy, which was significantly higher inside critical care area (68%, P = 0.049).

Tigecycline was combined with other antibiotics in 89% (136/153) of the cases without statistically significant difference between groups. Overall, tigecycline was most frequently combined with carbapenems (51/153, 46%) followed by colistin (47/153 patients, 30.7%) and piperacillin/tazobactam (38/153 patients, 24.8%).

Because of the retrospective nature of the study, clinical success rates were calculated in patients with a known clinical outcome. Clinical success was achieved in 43.4% of the whole patient population (59/136 patients; Table 3). It was significantly higher in patients treated outside critical care compared to those inside ICU at the univariate level (49.5 vs. 31%, P = 0.04). Yet, the adjusted multivariate model failed to confirm this finding, where there was an insignificant difference between the two groups upon controlling for the following potentially confounding variables (OR = 1.07 [0.45–2.57], P = 0.87; Table 3).

In the critically ill category, clinical success was significantly higher in patients with a SOFA score <7 compared with those with a SOFA score ≥7 using both univariate (62.5 vs. 13.8%, P = 0.003) and multivariate (OR = 12.51 [4.29–36.51], P < 0.0001) models (Table 4).

With respect to the indication, clinical success was significantly higher in the FDA-approved group compared with the off-label group through univariate analysis (64 vs. 38.7%, P = 0.02). However, this finding was not confirmed by multivariate analysis (Table 5).

CAP showed the highest success rate within the FDA-approved category (85.7%), whereas febrile neutropenia showed the best clinical outcome in the off-label category (84.2%). Septic patients receiving tigecycline showed the lowest clinical success rate (1/14 patients, 7.1%; Table 2).

Microbiological success rate was determined in cases in which follow-up cultures were available from patients' charts. Microbiological eradication of infection was achieved in 43.3% of patients who received tigecycline therapy (Table 3).

Yet, the univariate and multivariate models failed to show a significant difference in this outcome between patients inside vs. outside critical care area, those with SOFA score <7 vs. ≥7, and in FDA-approved vs. off-label indications (Tables 3, 4, 5).

Total mortality rate reached 45% in this study including 33% 7 day post-treatment mortality and 12% in hospital mortality. Findings of univariate and multivariate analysis show that there was not any statistically significant difference in these three types of mortality between patients inside and outside critical care (Table 3).

However, upon comparing patients with SOFA score ≥7 and those with SOFA score <7, only total mortality and 7 day post-treatment mortality were significantly higher in patients in the former group at the univariate (P < 0.0001) and multivariate levels (OR = 5.17 [2.43–11.01], P < 0.0001; OR = 6.41 [3.04–13.52], P < 0.0001, respectively; Table 4).

With respect to the indication, total mortality, and 7 days post-treatment mortality rates were significantly higher in the off-label indication group compared to the FDA-approved group, using univariate analysis (51.6 vs. 17.2%, P = 0.001; 38.7 vs. 10.3%, P = 0.004, respectively). However, the adjusted multivariate model confirmed this result for total mortality only (OR = 4.00 [1.30–12.31], P = 0.01; Table 5).

Distribution of gram-positive and gram-negative bacteria isolated from clinical specimens during the study period is shown in Figure 1. Gram-negative bacteria represented the majority of the clinical isolates (304/374 isolates, 81%; Supplementary File 2). Susceptibility to tigecycline reached 89% (194/217 isolates) in the gram-negative species and 91% (20/22 isolates) in the gram-positive species, as determined by the disc-diffusion method. A. baumannii (105/374 isolates, 28%) was predominant among the gram-negative species, of which 85% (87/102 isolates) were resistant to carbapenems. Yet, 82% (71/87 isolates) of the carbapenem-resistant Acinetobacter species were susceptible to tigecycline (XDR-AB). Pseudomonas species susceptibility to tigecycline was not tested (Table 4).

E. coli and Klebsiella species represented 16% (60/374 isolates) and 13% (48/374 isolates) of the total isolates, respectively. No carbapenem resistance was detected in either species. Overall, 98% (54/55 isolates) of the E. coli isolates were susceptible to tigecycline. Further, 77% of the E. coli isolates (46/60 isolates) were third-generation cephalosporin (3GC) resistant, of which 98% (43/44 isolates) were susceptible to tigecycline. With respect to Klebsiella species, 87% (40/46 isolates) were susceptible to tigecycline. Approximately 39% (18/48 isolates) were resistant to 3GCs, of which 78% (14/18 isolates) were tigecycline sensitive. P. aeruginosa was present in 12% (45/374 isolates) of the cultured bacteria, where 9% (4/45 isolates) were ceftazidime resistant and 44% (20/45 isolates) were carbapenem resistant (Table 4).

Among gram-positive bacteria, Enterococci represented 5% (19/374) of the total isolates, where none was vancomycin resistant. Tigecycline susceptibility was 83% (10/12) in this group. Staphylococcus aureus represented 3% (11/374) of the total isolates, where 55% (6/11) were methicillin resistant. All S. aureus strains were susceptible to tigecycline.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.