A novel approach for evaluating the microbiological efficacy of tigecycline in patients with complicated skin and skin-structure infections

Abstract number: 1134_02_382

Meagher A., Ambrose P., Passarell J., Cirincione B., Babinchak T., Ellis-Grosse E.J.

Objectives:

Tigecycline (T) is a glycylcycline in development for the treatment of patients (pts) with serious infections, including complicated skin and skin-structure infections (cSSSI). While cSSSI can be caused by a mixture of Gram-positive and -negative bacteria, Staphylococcus aureus and streptococci are the predominant pathogens. Previous analyses combining all pathogens have failed to identify an exposure-response (ER) relationship. A method was developed to create more homogenous pt populations for the microbiological (M) ER analysis of T in the treatment of cSSSI.

Methods:

Pts from 3 cSSSI clinical trials (one Phase 2 & two Phase 3) with T pharmacokinetic data and classified as both clinically and M evaluable, were pooled for analysis. Pts received 100 mg loading dose (LD)/50 mg q12h (100/50) or 50 mg LD/25 mg q12h (50/25). At the test of cure visit, M (eradication or persistence) response was evaluated. Indeterminate responses were excluded. Non pathogenic baseline isolates were excluded. Five homogeneous pt cohorts (C) were created based on baseline pathogens: S. aureus only (C1); S. aureus or streptococci (C2); 2 Gram-positive pathogens (C3); polymicrobial (C4); other monomicrobial infections (C5). Prospective step-wise procedures for combining C to increase sample size were used. Logistic regression was used to evaluate steady-state 24 hr area under the concentration-time curve (AUC) to MIC ratio (AUC/MIC) to predict response.

Results:

The dataset included 58 pts with 88 observations. C1 (n = 20) and C2 (n = 9) could not be evaluated due to small sample size. Analysis began with pooled C2 + C3. Continuous AUC/MIC ratio was marginally significant (p = 0.1130); a pt was 5.1% more likely to have successful response for every one-unit increase in AUC/MIC. Adding C4, including pathogens with MIC values up to 16 mcg/mL, decreased AUC/MIC, added cures to the lower end of the distribution, and added significant noise to the analysis. Adding C5 increased sample size and further decreased the ability to detect a relationship.

Conclusion:

Analysis of all pathogens combined could not identify an ER relationship. Polymicrobial infections with Gram-negative and anaerobic pathogens, associated with high MIC values, added noise to the analysis and decreased the predictive capability of the model. The approach of creating homogenous populations based on two key pathogens in cSSSI, S. aureus and streptococci, was critical for identifying significant ER relationships.