In this work we investigated tigecycline interactions with various antimicrobials by a two-step approach, involving preliminary chequerboard screening and subsequent time-kill assays. The chequerboard is an easy to perform, high-throughput method which provides single time-point evidence of bacterial growth inhibition, and generally results in an overestimate of synergistic interactions . For these reasons, all effective combinations inferred from chequerboard analysis were reassessed by time-kill assays. Although time consuming and cumbersome, the time-kill assays provide a dynamic picture of antibiotic action over time . Hence, only combinations showing synergy in both assays were interpreted as authentic synergistic interactions.
While chequerboard screening provided synergistic results for the combinations tigecycline/levofloxacin, tigecycline/amikacin, tigecycline/imipenem and tigecycline/colistin in 7 out of 24 isolates, time-kill kinetics confirmed synergism in only 5 out of 7 isolates, 4 of which were resistant to carbapenems and belonged to type 1 (related to the European clonal lineage II) . The different synergistic activities observed in A. baumannii isolates sharing the same epidemiological type  probably reflect the variable expression of different resistance determinants. This poses the need to test synergistic interactions even in case of clonal isolates characterized by identical genetic fingerprint and resistance profile.
Three out of five tigecycline synergistic concentrations observed in this study with time-kill assays exceed the maximum plasmatic concentration of tigecycline (0.38 mg/L) achievable with a standard dosage . However, thanks to its pharmacodynamic properties, tigecycline is rapidly distributed into tissues resulting in up to 78-fold higher tissue concentrations, compared to plasma [42, 43]. These considerations suggest a clinical usefulness for some of the synergistic combinations here detected for tigecycline.
We identified one isolate (study code 71) showing tigecycline/amikacin synergistic interaction at 24 h (1 and 64 mg/L for tigecycline and amikacin, respectively). This strain was resistant to amikacin (MIC = 256 mg/L) and showed an intermediate resistance to tigecycline (MIC = 4 mg/L). Although the synergistic concentration for amikacin (64 mg/L) is significantly above the threshold achievable in clinical treatments with a multi daily dosing regimen (20–30 mg/L), higher concentrations (65–75 mg/L) can be achieved with a single amikacin daily dose .
In vitro synergistic interactions between tigecycline and colistin have previously been demonstrated by time-kill assays in Klebsiella pneumoniae  and in vivo for the treatment of a severe case of MDR Pseudomonas aeruginosa osteomyelitis . Although several studies have reported clinical efficacy of colistin [7, 8], the synergistic effect of tigecycline/colistin combination has never been demonstrated in A. baumannii by time-kill analysis. Here, we showed for one A. baumannii isolate (study code 75) a synergistic effect at 3 h of incubation for tigecycline/colistin combination (2 and 0.25 mg/L for tigecycline and colistin, respectively), with a subsequent re-growth within 24 h. This strain was susceptible to colistin (MIC = 0.5 mg/L) and intermediate resistant to tigecycline (MIC = 4 mg/L). Notably, the colistin synergistic concentration is significantly below the serum concentration achievable after standard dosing regimen (5–6 mg/L) [44, 47]. As noted by various authors [48–50], colistin causes permeabilisation of the bacterial outer membrane, which would allow enhanced penetration by and activity of the other antibiotic in combination. The tigecycline/colistin synergistic interaction could therefore have an impact in clinical practice by reducing the therapeutic dosage of colistin, and hence the risk of collateral effects which currently represent a major limitation to its clinical use [7, 8].
We also demonstrated a synergistic interaction for the combination tigecycline/imipenem (0.5 and 16 mg/L for tigecycline and imipenem, respectively) in one A. baumannii isolate (study code 80), belonging to the epidemic type 1, and carrying the blaOXA-58 gene . It is of note that the serum concentration achievable during imipenem treatment is 20 mg/L . Thus, the synergistic interaction tigecycline/imipenem, which has never been described before for A. baumannii, could represent a valid therapeutic option to combat the increasingly frequent A. baumannii isolates resistant to both these drugs.
Resistance to quinolones is widespread among MDR A. baumannii strains . In this study, a high percentage of A. baumannii isolates were resistant to levofloxacin as a single agent. Here we report for the first time a synergistic interaction between tigecycline and levofloxacin (0.25 and 4 mg/L respectively) for 2 A. baumannii isolates, at 6 h of incubation. These strains showed full resistance to levofloxacin (MIC = 16 mg/L) and intermediate resistance to tigecycline (MIC = 4 mg/L). Also in this instance, the levofloxacin synergistic concentration is below the maximum serum concentration (5.9 mg/L) .
Even if no undesirable antagonistic combinations were confirmed in this study by time-kill assay, we detected a decreased antimicrobial efficacy for the tigecycline/piperacillin-tazobactam combination, compared to the antimicrobial efficacy of piperacillin-tazobactam alone (data not shown). This result is worrying considering that tigecycline/piperacillin-tazobactam combination therapy is often given empirically, without the support of in vitro interaction assays.
The molecular mechanisms of synergy between tigecycline and the various antibiotics deserve further investigation. Overexpression of the AdeABC efflux pump has been demonstrated in tigecycline resistant A. baumannii isolates , and our results indicate that all A. baumannii isolates tested carry the adeABC/adeIJK genes, suggesting that their variable expression level – but not their presence per se – could contribute to the extent of resistance. We also showed that adeDE is not present in A. baumannii, in agreement with previous studies [38, 39].
The regrowth after 24 h observed in time-kill experiments for all confirmed synergistic combination, except for tigecycline/amikacin, could reflect the labile nature in solution of tigecycline due to oxidative degradation (Wyeth Research, unpublished data) and/or the tendency of A. baumannii strains to induce resistance on exposure to antimicrobial agents, especially at sub-MIC concentrations. At present, we are unable to check the tigecycline levels and therefore we cannot determine if tigecycline was degraded, at least partially, during the experimental time course.
Determination of mutation frequencies for resistance to levofloxacin, amikacin, imipenem, colistin, and tigecycline at four-fold the MIC failed to detect any hypermutator phenotype for all isolates showing synergy in time-kill assays, irrespective of regrowth. Moreover, the mutation frequency toward resistance to tigecycline (~5 × 10-8) or other antibiotics (<10-8) is incompatible with the observed regrowth kinetics (Figure 1). Hence, we can only speculate that regrowth was due to different response of isolates to antibiotic-induced overexpression of broad-specificity multidrug efflux systems, like AdeABC and AdeIJK [16, 37–39], rather than hypermutability. According to this hypothesis, the tigecycline/amikacin interaction may have prevented the expression of efflux-based resistance by a still undefined mechanism, ultimately resulting in more effective synergism. In fact, a recent study on tigecycline/amikacin synergistic interactions in A. baumannii demonstrated the suppression of regrowth at 24 h for this particular antibiotic combination, in full agreement with our findings .
Further studies are needed to elucidate the molecular mechanisms responsible for synergistic interactions with tigecycline and to explore their therapeutic potential. It will also be necessary to combine in vitro findings with additional pharmacokinetic and pharmacodynamic data in order to provide more meaningful prediction of the in vivo efficacy of synergistic combinations in clinical practice. Lastly, in vitro synergy testing of tigecycline combinations is recommended prior to starting any combined therapy for treatment of infections sustained by MDR and pan-resistant A. baumannii.