Introduction

Tigecycline is one of the scarce available compounds, with a broad-spectrum activity, effective against multidrug-resistant strains including Gram-positive, Gram-negative aerobic, anaerobic bacteria and atypical microorganisms. In randomised controlled trials (RCT), tigecycline was effective in the treatment of complicated skin and soft tissue infections (cSSTIs) [1–3], complicated intra-abdominal infections (cIAIs) [4–7] and community-acquired pneumonia (CAP) [8–10]. Additional studies showed that tigecycline was effective in serious infections caused by known resistant pathogens [11, 12].

However, the use of tigecycline in patients with severe underlying diseases is limited, and little is known about its efficacy [13–15]. To date, RCTs included few intensive care unit (ICU) patients. Few data are available for ICU patients with bacteraemia [16].

For these reasons, we carried out a prospective, observational study in the intensive care setting to describe tigecycline prescription patterns and outcomes in critically ill patients from French ICUs. Some of our results were recently published in a series of articles reporting the “real-life” practice gathering five European databases (German, Italian, two Spanish studies and the current cohort). These analyses focused on labelled indications [17–19], global microbiology results [20] and safety issues [21], but did not address several key points, such as off-label indications, bacteraemia, emergence of resistance, superinfections and long-term outcomes; this led us to consider this in-depth analysis.

Materials and methods

Study design and patients

This prospective, multicentre, national observational study included consecutive cases of adult ICU patients treated with tigecycline. The only inclusion criterion was the receipt of tigecycline therapy in any, approved or non-approved, indication as mono- or combination, empiric, documented or rescue therapy for a specific localised source of infection or a specific flora. There was no recommendation on dosage or needed indication for the study protocol.

In accordance with French law, approval of an ethics committee was not required. The protocol was approved by the institutional review board (CEERB, CHU Bichat, Paris). All patients were informed of the data collection and agreed to participate in the study. A scientific committee (the authors) independently designed the study and reviewed all the collected data.

Clinical and microbiological data

Data were collected at ICU admission, at the start and end of tigecycline treatment, and 7 days after the end of treatment (or at hospital discharge if earlier). Clinical data included demographics, underlying diseases [22] (including diabetes mellitus, chronic renal failure, and chronic liver failure assessed using the Child–Pugh score [23]), immunosuppression (defined as steroid therapy or cancer therapy), severity of illness (assessed using the simplified acute physiology score (SAPS) II at ICU admission [24] and the sequential organ failure assessment (SOFA) score at the start of tigecycline treatment [25]), and previous tigecycline therapy.

The tigecycline-treated infection was clinically and microbiologically characterised. cIAIs (localised, generalised peritonitis, etc.) and cSSTI (dermis-hypodermis, fascia, etc.) were detailed as assessed during the surgery. The infection site and hospital- or community-acquired settings were collected.

The results of direct examinations and cultures were recorded. Identification and in vitro sensitivity testing of the pathogens were performed in the microbiology laboratory of each hospital using routine methods. Isolates were classified as susceptible (S), intermediate (I) or resistant (R) to tigecycline according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) methodology. Data concerning persistent and emerging isolates, including Pseudomonas aeruginosa, were collected.

Treatment data

The reasons for choosing tigecycline and any anti-infective agents received in the previous 28 days were recorded. The tigecycline regimen (loading dose, maintenance dose and treatment duration) and the associated anti-infective agents were recorded.

Outcomes

Response to treatment was determined as clinical success/failure at the end of tigecycline treatment and 7 days later (or at hospital discharge if earlier). Success was defined by the lack of need to use a new antibiotic or a surgical treatment not initially planned for the initial infection. Criteria for failure were persistence of the initial infection signs requiring a change of antibiotic therapy or a surgical intervention, reappearance of the initial infection signs, infection-related death occurred later than 48 h after the start of tigecycline and/or premature treatment discontinuation due to a tigecycline-related adverse event. Response was classified as undetermined in case of insufficient data (e.g. de-escalation before the fourth day of treatment), death not directly related to the initial infection or occurred within the first 48 h of treatment, or addition of an antibacterial agent for another infection. Survival was recorded 28 days after the end of tigecycline treatment: data were retrieved from the French national epidemiology center for medical causes of death (CépiDc, Centre d’épidémiologie sur les causes médicales de décès, INSERM).

Statistics

Data were analysed using SAS® 8.2 (SAS Institute Inc., Cary, NC, USA). To describe a characteristic or an event with a 10 % frequency and an accuracy of ±5 % as assessed by the 95 % confidence interval (95 % CI), 150 patients were to be enrolled.

For the purpose of this study, patients’ characteristics were stratified for disease severity assessed by SOFA score (<7 or ≥7). Outcomes were stratified for SOFA score, body mass index (BMI ≤35 kg and >35 kg/m2), immunosuppression, and age groups (<70 and ≥70 years of age).

Variables were expressed as median values and ranges for numerical variables, and as frequencies and percentages for categorical variables. The 95 % CIs of the response rates were calculated. Groups were compared using Wilcoxon signed-rank test for numerical variables, and the Chi square or the Fisher’s exact test for categorical variables. Comparisons for success rates were carried out under the worse assumption, i.e. considering undetermined responses as failures. Factors independently associated with success to treatment at 7 days were identified by multivariate stepwise logistic regression among the factors that were statistically significant at the 10 % level in univariate regressions and taking into account significant interactions. The predictive performance of the final multivariate model was evaluated using receiver operating characteristic (ROC) analysis [26]. Survival, defined as the time from the first tigecycline intake to 28 days after last intake (or death), whichever occurs first, was assessed using the Kaplan–Meier method. A log-rank test was performed for subgroup analyses. As day-28 follow-up was not available for 18 patients, their survival time was censored at the time of the last visit in the study. A death recorded after day 28 was censored at the time of day 28. Statistical significance was accepted at the 5 % level.

Results

Patients’ clinical characteristics

A total of 156 patients from 26 ICUs were enrolled between September 2008 and April 2010, including 73 patients with a SOFA score <7 and 83 severe cases (SOFA ≥7) (Table 1). Immunosuppression was mainly due to cancer (n = 29) and steroid therapy (n = 16).

Table 1 Treatment with tigecycline: patients’ characteristics at baseline, types of infections treated and characteristics of treatment
Full size table

The majority of patients (145/156, 93 %) had received one or more anti-infective agents in the last 30 days before the start of tigecycline [including penicillins (60 %), cephalosporins (42 %), aminoglycosides (39 %), carbapenems (28 %), glycopeptides (26 %) or fluoroquinolones (26 %)].

Infections treated with tigecycline

Tigecycline was given for the treatment of cIAI in 88/156 (56 %) patients, cSSTI in 29 (19 %) and other infections in 56 (36 %) (mainly lung infections, n = 38, 24 %) (Table 1). Most of the treated infections were hospital-acquired (131/173, 76 %). A positive blood culture was observed in 17 (12 %) patients. Overall, 17/145 (12 %) patients had secondary bacteraemia, including 9/63 (12 %) patients with a SOFA score ≥7 (P = 0.963). Indications for tigecycline were similar in the less and the most severely ill patients.

Microbiological data

Overall, 146 (94 %) patients had at least one microbiological sample at the start of tigecycline [direct examination in 87 (56 %) patients, species identified in 127 (81 %) patients] (Table 2). Infection was polymicrobial in 29 cases. There were no marked differences between the less and the most severely ill patients (data not shown).

Table 2 Number (%) of baseline isolates by sensitivity to tigecycline
Full size table

Three microorganisms in three patients acquired resistance to tigecycline during the course of treatment (Escherichia coli, Enterobacter cloacae and Enterobacter aerogenes). Sixty-four microorganisms in 41 patients emerged during the course or at the end of treatment: 18 Gram-positive cocci (including 11 staphylococci), 39 Gram-negative bacilli (including 9 P. aeruginosa, 6 Enterobacter ssp., 3 Proteus ssp., and 1 Morganella morganii), 4 anaerobic germs and 3 yeasts. A total of 28 microorganisms in 23 patients persisted with no change in susceptibility (both emergence and persistence were observed in 13 of these 23 patients) during the course of tigecycline treatment: 11 Gram-positive cocci (5 enterococci and 6 staphylococci) and 16 Gram-negative bacilli (including 6 E. coli, 4 Klebsiella spp. and 3 other enterobacteria).

Treatment with tigecycline

Characteristics of treatment are provided in Table 1. The vast majority of patients were given the recommended loading and maintenance doses with an overall median treatment duration of 9 (1–78) days: 8 (1–78) days in the less severely ill, 9 (2–43) days in the most severely ill (P = 0.499) patients and 9 days (2–78) in patients alive at the end of treatment. Tigecycline was combined with other anti-infective agents in two-thirds of the patients (101/156, 65 %) (Fig. 1), without statistically significant difference between groups (64 vs. 65 % for the <70 and ≥70 years age groups, respectively, P = 0.906; and 63 vs. 66 % for the SOFA <7 and ≥7 groups, respectively, P = 0.671). The aminoglycosides used were amikacin (17 % of all patients) and gentamicin (8 %). The most frequently used penicillin was piperacillin (combined with tazobactam, 11 %).

Fig. 1
figure 1

Anti-infective agents combined with tigecycline

Full size image

Tigecycline treatment was prematurely stopped in 66 (42 %) patients, without statistically significant difference according to the illness severity (P = 0.774). The reasons were resistant strain included (n = 11), clinical failure (n = 12), de-escalation (n = 20), death (n = 14), new infection (n = 4), persistent fever of unknown origin (n = 1), unjustified antibacterial agent change (n = 1) and/or shock probably not of infectious origin (n = 1). In the less severely ill patients, the most common reasons were de-escalation (10/37, 33 %) and resistant strain (n = 8), whereas in the most severely ill patients they were death (12/36, 33 %) and de-escalation (n = 10).

Adverse events were reported in 16 and 29 % of the less and most severely ill patients, respectively (23 % of total patients). Three adverse events were considered as probably/definitely related to tigecycline: drug resistance (n = 1), drug inefficacy (death due to septic shock, n = 1) and acute renal failure (patient cured with tigecycline, n = 1). Irrespective of causality, serious adverse events (fatal or not) occurred in 10 and 23 % of the less and most severely ill patients, respectively (17 % of total patients).

Response to treatment

The overall success rate was 60 % [93/156, 95 % CI (51–67)] at the end of treatment, and 53 % [77/145, 95 % CI (45–61)] 7 days after the end of treatment (Table 3). The difference in success rates between the less and the most severely ill patients was significant at both time points (P = 0.005, P = 0.001, respectively). The success rate at the end of treatment for patients alive after the last tigecycline uptake was 65 % [92/141, 95 % CI (57–73)]. The causes of failure at the end of treatment are described in Table 3. Table 4 provides the success rates obtained in the patient subgroups of interest, 7 days after the end of treatment.

Table 3 Response to treatment with tigecycline
Full size table
Table 4 Success rate according to the major characteristics of patients, infections and tigecycline treatment
Full size table

The success rate at 7 days after treatment was statistically significantly higher when tigecycline treatment duration was longer. In the whole cohort, the success rate was 70 % with a treatment duration >9 days vs. 40 % with a treatment duration ≤9 days (P < 0.001). Similarly, it was 80 vs. 55 % (P = 0.097) respectively, in the less severely ill patients and 61 vs. 27 %, respectively, in the most severely ill patients (P = 0.008). Combination of another antibiotic with tigecycline did not markedly influence the success rate in the whole cohort, in the less and in the most severely ill patients (54 vs. 51 %; 67 vs. 64 %; 42 vs. 39 %, respectively).

A reduced rate, although not statistically significant, was observed with concomitant bacteraemia vs. without bacteraemia, in the less (62 vs. 74 %) and in the most severely ill patients (33 vs. 54 %).

Univariate regressions identified two factors associated with failure of treatment at 7 days: the SOFA score, either expressed as increasing SD units [OR = 1.72, 95 % CI (1.20–2.50), P = 0.003] or a score ≥7 [OR = 2.70, 95 % CI (1.39–5.26), P = 0.004], and BMI, either expressed as increasing SD units [OR = 9.09, 95 % CI (1.92–50), P = 0.005] or a BMI >35 kg/m2 [OR = 1.39, 95 % CI (0.96–2.00), P = 0.080]. The factors associated with failure and identified using the multivariate analysis were an increasing SOFA score [OR = 1.67, 95 % CI (1.12–2.44), P = 0.010] and a BMI >35 kg/m2 [OR = 8.33, 95 % CI (1.82–33.33), P = 0.007]. The sensitivity and specificity of this model were 94 % [95 % CI (0.88–1.00)] and 37 % [95 % CI (0.25–0.49)], respectively.

Survival

Global survival at 28 days was 85 % and no statistical difference was observed between age groups (87 vs. 80 % for the <70 and ≥70 years age groups, respectively, P = 0.408). Survival at day 28 was higher in patients with a BMI ≤35 kg/m2 than with a BMI >35 kg/m2 (87 vs. 63 %, P = 0.003) and statistically significantly higher in the less than in the most severely ill patients (96 vs. 75 %, P < 0.001) (electronic supplementary material). Moreover, patients receiving catecholamines treatment had a statistically significant lower survival rate than those not treated with catecholamines (75 vs. 94 %, P = 0.001). Survival at day 28 was 86, 93 and 80 % for the patients initially suffering from IAI, SSTI and other infections, respectively.

Discussion

The efficacy of tigecycline versus other antimicrobial agents for the treatment of cSSTI [1–3] and cIAI [4–7, 27] has been evidenced in several RCTs, demonstrating that tigecycline was as efficacious as the comparators in treating infections, with a comparable safety profile. Three recent meta-analyses evaluating the published data from available RCTs also found no statistically significant difference in treatment success between cIAI patients treated with tigecycline and those treated with comparators [28–30]. However, in both cSSTI and cIAI trials, severe cases were usually excluded from the protocol, leading to insignificant information about these particular cases.

Recently, the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) issued a warning regarding the risk of increased mortality in patients treated with tigecycline, observed in the clinical trials [31, 32]. In all phase III and IV (cSSTI and cIAI) studies, death occurred in 2.4 % (54/2,216) of patients receiving tigecycline and in 1.7 % (37/2,206) of patients receiving comparator drugs [33]. A number of meta-analyses also reported higher all-cause mortality in patients treated with tigecycline versus comparators in RCTs [28–30, 34, 35]. Taking into account these findings, the EMA Committee for Medicinal Products for Human Use concluded that tigecycline benefits continue to outweigh its risks, but recommended modifications in product information to ensure an appropriate use, by making prescribers aware of the observed increased mortality risk. In addition, they issued a recommendation stating that tigecycline should be used only when other antibiotics are known or suspected to be not suitable [33].

However, clinical trials concerning critically ill patients’ management are limited. A few large observational studies have been set up to determine the outcomes of ICU tigecycline-treated patients in clinical practice. The success rates obtained in the current study are in line with those reported in previous registries [13–15, 18] and in the European registry [17–19]. Interestingly, the success rate is also comparable to those reported in RCTs assessing the efficacy of other antibiotics in the ICU setting [36].

RCTs in contrast to registries generally do not reflect the clinical practice and do not represent real-life patients because of stringent selection criteria. This point is of interest particularly in ICU patients, for whom a limited number of trials are conducted only in selected indications. In addition, the evaluation endpoints used here were selected to ensure that the physician’s concerns were respected and obviously do not favour the product.

The tigecycline dosage used in this study stands in the conventional range (100 mg daily) in 93–94 % of the cases. Interestingly, a relatively satisfactory success rate was reported in pulmonary infection patients. Similarly to other antibiotic drugs [37], concerns have been raised about the best dose for tigecycline, especially for treatment of pulmonary infections. Recent publications suggest a potential benefit of high dose tigecycline (200 mg daily) in severe/difficult-to-treat infections [38]. Effectiveness and safety of this policy without adding new risk factors for potentially resistant bacteria remain to be confirmed [39].

The never-ending debate on bacteriostatic drugs use for treatment of serious infections was re-triggered with the launch of tigecycline [40]. In many ICUs across the world, physicians dare to prescribe this bacteriostatic agent without having the feeling of putting their patients in jeopardy [13–15]. This point is particularly relevant in patients receiving a monotherapy as a first-line treatment. We, like others, report good results in monotherapy in more than half of patients receiving tigecycline. In contrast to other countries, tigecycline seems to be used in France mainly in combination with agents directed particularly against Gram-negative microorganisms, and the preferred indications are those of the marketing authorisations. This finding was also reported in the European registries [17–19]. For instance, in a severe ICU population in Germany, tigecycline was also used mostly in combination (76 %), but the therapy frequently targeted Gram-positive cocci (including enterococci) possibly because of the high frequency of vancomycin-resistant strains. In Latin American countries, the tigecycline use in off-label indications appears to be important [41].

Finally, there were no serious adverse events requiring tigecycline discontinuation, and few tigecycline-related adverse events were reported. It is important to note that tigecycline administration in severe ICU patients, particularly in those with multiple organ failure, raised no safety concerns.

Like all observational studies, the limitations of our study included the lack of a control group and randomisation. However, our results contribute to the knowledge about tigecycline use in severely ill patients, a fragile population lacking clinical data. It is interesting that no increased mortality was observed in the cIAI and cSSTI tigecycline-treated patients, as observed in the five different European registries.

From our experience, tigecycline could be proposed as an empiric therapy in low severity cases, non-immunosuppressed or non-bacteraemic infections. In severe infections, immunosuppressed, bacteraemic or obese patients, its use should be cautiously considered, and restricted to documented therapy based on susceptibility testing in difficult-to-treat infections. Other options should be considered for suspected P. aeruginosa infections [42, 43].

Conclusions

In this ICU population treated with tigecycline, the success rates were comparable to those obtained in clinical studies analysing severe infections. In contrast to its use in other countries, tigecycline appears to be used mainly in combination with agents directed particularly against Gram-negative microorganisms.