Abstract
Ceftaroline is a cephalosporin with broad-spectrum in vitro activity against pathogens commonly associated with acute bacterial skin and skin structure infections (ABSSSI), including methicillin-resistant Staphylococcus aureus. Ceftaroline fosamil, the prodrug of ceftaroline, is approved for the treatment of patients with ABSSSI. Using data from the microbiologically evaluable population from two phase 2 and two phase 3 randomized, multicenter, double-blind studies of patients with ABSSSI, an analysis examining the relationship between drug exposure, as measured by the percentage of time during the dosing interval that free-drug steady-state concentrations remain above the MIC (f%T>MIC), and clinical and microbiological responses was undertaken. The analysis population included 526 patients, of whom 423 had infections associated with S. aureus. Clinical and microbiological success percentages were 94.7 and 94.5%, respectively, among all of the patients and 95.3 and 95.7%, respectively, among those with S. aureus infections. Univariable analysis based on data from all of the patients and those with S. aureus infections demonstrated significant relationships between f%T>MIC and microbiological response (P < 0.001 and P = 0.026, respectively). Multivariable logistic regression analyses demonstrated other patient factors in addition to f%T>MIC to be significant predictors of microbiological response, including age and infection type for all of the patients evaluated and age, infection type, and the presence of diabetes mellitus for patients with S. aureus infections. Results of these analyses confirm that a ceftaroline fosamil dosing regimen of 600 mg every 12 h provides exposures associated with the upper plateau of the pharmacokinetic-pharmacodynamic relationship for efficacy.
INTRODUCTION
Pharmacokinetic-pharmacodynamic (PK-PD) analyses of phase 2 and 3 data allow for a better understanding of the relationships between exposure and efficacy and/or safety endpoints for intended patient populations. Results from analyses of these types provide an opportunity to close the loop between knowledge gained during early- and late-stage development and thus confirm early-stage predictions made for dose selection (1). The evaluation of such information also allows for a better understanding of how to translate data from preclinical models to those from clinical studies (2).
Ceftaroline, the active form of the prodrug ceftaroline fosamil, is a cephalosporin with broad-spectrum in vitro activity against Gram-positive and -negative bacteria commonly associated with skin and skin structure infections, including methicillin-resistant Staphylococcus aureus (MRSA). This agent was approved by the United States Food and Drug Administration (FDA) in October 2010 for the treatment of acute bacterial skin and skin structure infections (ABSSSI) and community-acquired bacterial pneumonia (3) and approved in Europe for similar indications (4). In an effort to evaluate decisions about dose selection during the last stage of drug development, the data from four clinical studies (5–9) were pooled and PK-PD relationships for the efficacy of ceftaroline fosamil in patients with ABSSSI were evaluated. The results of these analyses are described herein.
MATERIALS AND METHODS
Patient population.
Data for this analysis were obtained from four multicenter, randomized, comparative clinical studies, two phase 2 and two phase 3 studies, each evaluating the efficacy and safety of ceftaroline fosamil in adult patients with ABSSSI. A brief summary of the design of each study is provided below. Additional details regarding the designs of these studies are provided elsewhere (5–9).
The first phase 2 study evaluated adult patients with ABSSSI following treatment with intravenous (i.v.) ceftaroline fosamil versus standard therapy for 7 to 14 days (phase 2 i.v. study) (6). Standard therapy was vancomycin or a penicillinase-resistant penicillin for Gram-positive organisms other than MRSA. The second phase 2 study evaluated adult patients with ABSSSI following treatment with intramuscular (i.m.) ceftaroline fosamil versus i.v. linezolid for 5 to 14 days (phase 2 i.m. study, NCT00633152) (6). The two phase 3 studies, CANVAS 1 and 2 (NCT00424190 and NCT00423657, respectively), evaluated adult patients with ABSSSI following treatment with i.v. ceftaroline fosamil versus i.v. vancomycin plus aztreonam for 5 to 14 days (5, 6, 9). All of these studies were approved by the institutional review board or ethics committee, were conducted in compliance with Good Clinical Practice, and required that written informed consent be obtained from each patient. For a summary of the main inclusion and exclusion criteria, see the supplemental material. Additional details regarding the inclusion and exclusion criteria for the four studies are provided elsewhere (5–9).
Drug doses and administration.
Patients in the phase 2 studies received 600 mg ceftaroline fosamil i.v. infused over 60 min every 12 h (q12h) for 7 to 14 days (5) or 600 mg ceftaroline fosamil i.m. q12h for 5 to 14 days. If the creatinine clearance (CLCR) was ≤50 ml/min, the i.m. dose of ceftaroline fosamil was reduced to 400 mg q12h but readjusted to 600 mg q12h if renal function improved (i.e., if the CLCR improved to >50 ml/min) (6).
In the two phase 3 studies, patients received 600 mg ceftaroline fosamil i.v. infused over 60 min q12h for 5 to 14 days. The dose was reduced to 400 mg over 60 min q12h for those patients with moderate renal impairment whose calculated CLCR was ≤50 ml/min (but greater than the required 30 ml/min) (7, 8).
Pharmacokinetic sample collection.
Blood samples were collected in each study for the determination of ceftaroline fosamil and ceftaroline concentrations in plasma. Following i.v. infusion of ceftaroline fosamil in the phase 2 and 3 studies (5, 7–9), up to four PK samples were to be collected (within 15 min prior to the administration of a dose, as well as within 5 min, 1 to 3 h, and 4 to 8 h following the end of infusion) on day 3 of therapy from a subset of patients who participated in each of these studies. Following i.m. injection of ceftaroline fosamil in the phase 2 study (6), up to five PK samples were to be collected (within 15 min prior to the administration of a dose, as well as 5 min to 2 h, 2 to 4 h, 4 to 8 h, and 8 to 12 h after injection) from all of the patients on day 3 of therapy.
Efficacy endpoints.
Patients enrolled in the phase 2 i.v. study were evaluated for efficacy on days 7 to 14 (end of therapy [EOT]), days 8 to 14 (test of cure [TOC]), and days 21 to 28 posttherapy (late follow-up [LFU]) (5). Patients enrolled in the phase 2 i.m. study and the two phase 3 studies were evaluated for treatment efficacy on days 5 to 14 (EOT), days 8 to 15 (TOC), and days 21 to 35 posttherapy (LFU) (6–9).
Clinical responses were categorized as cure, failure, or indeterminate. A clinical response of cure (subsequently referred to as clinical success) was assigned to a patient who had a total resolution of all of the signs and symptoms of the ABSSSI under study or showed improvement to the extent that no additional antibiotic therapy was necessary. A clinical response of failure was assigned to a patient who required continued antibiotic therapy because of the persistence, incomplete resolution, or worsening of signs and symptoms, required surgical intervention as an adjunct or follow-up therapy because of the study drug failure, or died as a result of the ABSSSI. An outcome of indeterminate was assigned to a patient whose study data were not available for evaluation, who was lost to follow-up, or who died of something other than the ABSSSI.
Microbiological responses were categorized as eradication, presumed eradication, persistence, presumed persistence, or indeterminate. Patients with microbiological success had baseline pathogens that were eradicated or presumed eradicated. Patients with microbiological failure had baseline pathogens that were persistent or presumed persistent.
Determination of PK-PD index.
The PK-PD index of interest in this evaluation was the percentage of time the free-drug concentration remained above the MIC (f%T>MIC), the measure that has been demonstrated to be most predictive of ceftaroline efficacy on the basis of nonclinical data (10). The estimation of f%T>MIC values was based on the MICs for baseline pathogens and a previously developed population PK model (11).
The process used to identify the MIC to use for the calculation of f%T>MIC for each patient who had multiple baseline pathogens was conducted by using the following steps. When S. aureus and S. pyogenes were both present in baseline cultures, the highest MIC for these two organisms was chosen. When S. aureus and/or S. pyogenes were present with other pathogens (either Gram-positive and/or Gram-negative organisms) at baseline, the MIC for S. aureus or S. pyogenes was chosen. When Gram-positive and/or Gram-negative organisms were present at baseline without S. aureus or S. pyogenes, the case was closely evaluated by the authors (G.L.D., P.G.A., and S.M.B.) and the most likely infecting pathogen was chosen. Lastly, when an organism was isolated from both blood and tissue and when the MICs differed, the higher MIC was chosen.
The previously conducted population PK analysis (11) demonstrated that a three-compartment model with zero-order i.v. input into the central compartment, or dual-phase first-order absorption from depot compartments into the central compartment following i.m. administration, and first-order elimination best described ceftaroline fosamil PK. A two-compartment disposition model with first-order conversion of the prodrug to ceftaroline and parallel linear and saturable elimination pathways best described ceftaroline PK. The population mean (based upon individual covariate values) and individual post hoc predicted plasma ceftaroline concentrations were unbiased and in good agreement with the observed data (r2 = 0.93 and 0.98, respectively) following either i.v. or i.m. dose administration. Since CLCR was found to be the primary determinant of ceftaroline exposure and the population PK model with the estimated covariate relationships reasonably predicted ceftaroline exposure in the absence of plasma concentration-time data for each individual, the predicted exposures based on the population mean PK parameters and patient-specific covariate information were used as the ceftaroline exposure measures for patients with ABSSSI in the phase 3 studies from whom PK samples were not collected. Using a baseline measure of serum creatinine, CLCR was calculated according to the method described by Cockcroft and Gault (12) and was normalized by body surface area (13).
Using the Bayesian PK parameter estimates obtained from patients with sufficient PK data based on the final population PK models for ceftaroline fosamil and ceftaroline, individual predicted ceftaroline concentrations were generated during the 12-h dosing interval at steady state for each patient. Assuming 20% protein binding (3), free-drug concentrations were determined by multiplying the individual predicted concentrations by 0.8. As described previously, for those patients from whom PK samples were not collected, population mean predicted ceftaroline concentrations were generated by using the population PK model and patient-specific covariate information. For such patients, population mean predicted f%T>MIC values were estimated and used instead of the individual predicted f%T>MIC values.
To assess the feasibility of the above-described approach, the bias and precision of f%T>MIC values were examined for all of the patients with plasma PK samples available by examining the distribution of the percent predicted error (PE%) and the absolute predicted error (|PE%|), respectively (14). PE% was calculated as the population mean predicted f%T>MIC minus the individual predicted f%T>MIC multiplied by 100 and then divided by the individual predicted f%T>MIC, while |PE%| was calculated as the absolute value of the PE%.
PK-PD analyses.
PK-PD analyses for two efficacy endpoints assessed at TOC, clinical response (success versus failure) and microbiological response (success versus failure), were conducted by using data from the microbiologically evaluable (ME) population and those patients in this population with S. aureus isolated at baseline. Patients who were designated clinical failures for reasons other than the study drug (e.g., for a treatment-limiting adverse event) were excluded from these analyses.
PK-PD analyses were performed with Systat software, version 11.0 (Systat Software, Inc., Richmond, CA) (15) and R 2.4.1 (16). Exploratory analyses included an examination of the relationship between the proportion of successful responses for each efficacy endpoint among groups of patients categorized by increasing ranges of f%T>MIC values. Classification and regression tree (CART) analysis was used to identify f%T>MIC thresholds that distinguish cohorts of patients with impressive response differences. The significance of the differences in the proportion of successful responses above and below such CART-derived thresholds was evaluated by using the Pearson chi-square test or Fisher's exact test, as appropriate. In addition, pairs of threshold values that split f%T>MIC into three groups were assessed to investigate potential nonlinearity, with an optimal pair chosen on the basis of that which achieved the greatest statistical significance when three groups of at least 10 patients each were compared.
Univariable analyses, which consisted of the chi-square or Fisher's exact test for categorical independent variables and logistic regression for continuous independent variables, were carried out to identify factors in addition to f%T>MIC that were associated with clinical and microbiological responses. The categorical independent variables considered included baseline patient demographic and disease-related characteristics and underlying comorbidities. The continuous independent variables considered included age, body mass index (BMI), disease severity score (17), MIC, and weight. The disease severity score was calculated on the basis of patient age, comorbid conditions, a physical examination, laboratory tests, and other characteristics. Continuous independent variables were evaluated as such and also as categorical variables in various forms as described above for f%T>MIC.
Multivariable logistic regression models were constructed with both backward elimination and forward inclusion of independent variables (α = 0.1) for clinical and microbiological responses. If the degree of concordance between these efficacy endpoints was high, multivariable analyses were limited to the endpoint manifesting the strongest univariable relationship with f%T>MIC. In order to reduce overparameterization and construct parsimonious multivariable models that contained the number of independent variables recommended by Hosmer and Lemeshow (18), a stringent inclusion and exclusion criterion (α = 0.05) was used in both forward and backward stepwise procedures. Candidate independent variables for model inclusion were chosen on the basis of significance of univariable associations with responses (P ≤ 0.20). Separate sets of models in which a given independent variable was considered continuously or categorically were evaluated. If the significance of univariable relationships for f%T>MIC was similar or more impressive than those for MIC, multivariable models based on f%T>MIC were developed. Interactions between f%T>MIC and independent variables retained in each multivariable model were considered to further refine multivariable models. In addition to the above-described considerations, discrimination among candidate final multivariable models was also based on clinical judgment, which involved assessment of the utility of clusters of independent variables and the form in which such variables were evaluated.
Average model-predicted probabilities of a successful response, based on final multivariable logistic regression models, were assessed relative to observed proportions of successful responses for cohorts of patients described by independent variables included in the final multivariable logistic regression models. For models with categorical independent variables, observed proportions and model-predicted probabilities of a successful response were summarized numerically and graphically for each cohort of patients characterized by a combination of variable categories in which there were observed patients. When models contained continuous independent variables, those variables were categorized by using the median value for the threshold for the purpose of computing observed proportions of successful responses within combinations of categories.
RESULTS
Of the 534 patients in the ME population, 8 were classified as clinical failures for reasons other than the lack of a therapeutic effect of the study drug and were thus excluded. Of the remaining 526 patients, 423 had S. aureus isolated at baseline. Summary statistics of the continuous and categorical baseline patient demographic and disease characteristics of all of the patients and those with S. aureus infections are presented in Tables 1 and 2, respectively. The percentages of patients with clinical and microbiological success were 94.7 and 94.5% among all of the ME patients and 95.3 and 95.7% among those with S. aureus infections, respectively. For a summary of the percentages of microbiological responses by clinical responses in all of the patients and those with S. aureus infections, see Table S1 in the supplemental material.
TABLE 1.
Summary statistics of the continuous baseline demographic and disease characteristics for all of the patients and those with S. aureus infections
| Parameter | All patients |
Patients with S. aureus infections |
||||
|---|---|---|---|---|---|---|
| n | Mean (CV%)a | Median (min, max) | n | Mean (CV%) | Median (min, max) | |
| Age (yr) | 526 | 45.9 (34.6) | 46.0 (18, 88) | 423 | 45.5 (35.1) | 46.0 (18, 88) |
| BMI (kg/m2) | 525 | 28.1 (24.5) | 26.8 (14.1, 74.1) | 422 | 28.1 (23.9) | 26.5 (16.0, 62.6) |
| CLCR (ml/min/1.73 m2) | 526 | 102.5 (33.4) | 99.7 (32.9, 266) | 423 | 102.9 (34.0) | 99.8 (32.9, 266) |
| Wt (kg) | 525 | 82.8 (25.1) | 80.0 (41.0, 227) | 422 | 82.6 (24.5) | 80.0 (41.0, 181) |
| Disease severity scoreb | 437 | 82.3 (38.3) | 77.0 (29.0, 198) | 352 | 81.8 (38.6) | 77.0 (29.0, 198) |
CV% = percent coefficient of variation.
The disease severity score was calculated on the basis of patient age, comorbid conditions, a physical examination, laboratory tests, and other characteristics (17).
TABLE 2.
Summary statistics of the categorical baseline patient demographic and disease characteristics for all of the patients and those with S. aureus infections
| Variable | All patients | Patients with S. aureus infections |
|---|---|---|
| BMI category | ||
| Underweight (<18.5 kg/m2) | 1.90 (10/525)a | 1.42 (6/422) |
| Normal wt (≥18.5 to <25.0 kg/m2) | 35.1 (184/525) | 36.0 (152/422) |
| Overweight (≥25.0 to <30.0 kg/m2) | 31.2 (164/525) | 31.3 (132/422) |
| Obese (≥30.0 to <40.0 kg/m2) | 26.3 (138/525) | 25.1 (106/422) |
| Morbidly obese (≥40.0 kg/m2) | 5.52 (29/525) | 6.16 (26/422) |
| Sex | ||
| Female | 34.8 (183/526) | 34.5 (146/423) |
| Male | 65.2 (343/526) | 65.5 (277/423) |
| Race | ||
| White | 75.5 (397/526) | 75.9 (321/423) |
| Black | 7.41 (39/526) | 7.57 (32/423) |
| Asian | 0.76 (4/526) | 0.95 (4/423) |
| American Indian/Alaska Native | 0.57 (3/526) | 0.47 (2/423) |
| Native Hawaiian or Other Pacific Islander | 0.57 (3/526) | 0.71 (3/423) |
| Multirace/other | 1.90 (10/526) | 1.89 (8/423) |
| Unknown | 13.3 (70/526) | 12.5 (53/423) |
| Ethnicity | ||
| Nonhispanic | 81.2 (427/526) | 81.6 (345/423) |
| Hispanic | 18.8 (99/526) | 18.4 (78/423) |
| Geographic region | ||
| Africa | 2.85 (15/526) | 2.13 (9/423) |
| Eastern Europe | 41.1 (216/526) | 42.1 (178/423) |
| Latin America | 7.60 (40/526) | 5.91 (25/423) |
| Western Europe | 8.17 (43/526) | 6.86 (29/423) |
| United States | 40.3 (212/526) | 43.0 (182/423) |
| Infection category | ||
| Infected wound | 14.1 (74/526) | 12.8 (54/423) |
| Major abscess | 36.7 (193/526) | 37.4 (158/423) |
| Infected ulcer | 9.13 (48/526) | 9.22 (39/423) |
| Infected burn | 4.56 (24/526) | 4.96 (21/423) |
| Infected bite | 1.52 (8/526) | 1.65 (7/423) |
| Deep/extensive cellulitis | 28.5 (150/526) | 29.6 (125/423) |
| Lower extremity ABSSSI with diabetes mellitus or peripheral vascular disease | 4.94 (26/526) | 4.02 (17/423) |
| Other | 0.57 (3/526) | 0.47 (2/423) |
| Infection location | ||
| Head and/or neck | 5.70 (30/526) | 6.15 (26/423) |
| Lower limb | 47.0 (247/526) | 46.3 (196/423) |
| Other location | 47.3 (249/526) | 47.5 (201/423) |
| Presence of bacteremia | ||
| No | 95.6 (497/520) | 95.5 (399/418) |
| Yes | 4.42 (23/520) | 4.55 (19/418) |
| Presence of MRSA at baseline | ||
| No | 66.7 (351/526) | 58.6 (248/423) |
| Yes | 33.3 (175/526) | 41.4 (175/423) |
| Presence of polymicrobial infection | ||
| No | 72.6 (382/526) | 72.8 (308/423) |
| Yes | 27.4 (144/526) | 27.2 (115/423) |
| Presence of Pseudomonas or Acinetobacter species at baseline | ||
| No | 96.2 (506/526) | 96.7 (409/423) |
| Yes | 3.80 (20/526) | 3.31 (14/423) |
| Disease severity score categoryb | ||
| I (<65) | 36.2 (158/437) | 36.7 (129/352) |
| II (65 to 82) | 19.9 (87/437) | 18.8 (66/352) |
| III (83 to 100) | 16.7 (73/437) | 17.9 (63/352) |
| IV (>100) | 27.2 (119/437) | 26.7 (94/352) |
| Presence of diabetes mellitus | ||
| No | 82.3 (433/526) | 83.2 (352/423) |
| Yes | 17.7 (93/526) | 16.8 (71/423) |
| Presence of peripheral vascular disease | ||
| No | 84.4 (444/526) | 85.6 (362/423) |
| Yes | 15.6 (82/526) | 14.4 (61/423) |
| Previous systemic antibacterial usec | ||
| No | 59.7 (314/526) | 60.5 (256/423) |
| Yes | 40.3 (212/526) | 39.5 (167/423) |
| Prior treatment failure | ||
| No | 91.1 (479/526) | 90.8 (384/423) |
| Yes | 8.94 (47/526) | 9.22 (39/423) |
| Route of drug administration | ||
| I.v. | 90.7 (477/526) | 90.1 (381/423) |
| I.m. | 9.32 (49/526) | 9.93 (42/423) |
Data are expressed as the percentage of patients (number of patients/total).
The disease severity score was calculated on the basis of patient age, comorbid conditions, a physical examination, laboratory tests, and other characteristics (17).
Receipt of an antibiotic within 96 hours of initiating study drug.
Approximately 27% of all of the patients and those with S. aureus infections had multiple organisms isolated at baseline. After following the process outlined to identify the baseline isolate used to calculate f%T>MIC values, the MIC50/MIC90 (min, max) values for these isolates among all of the patients were 0.25/0.5 (≤0.004, >32) mg/liter, respectively. For patients with S. aureus infections, these values were 0.25/0.5 (≤0.015, 2) mg/liter.
Among the 526 patients, 104 had plasma PK samples available for analysis. As shown in Table S2 in the supplemental material, the median (min, max) PE% and |PE%| for f%T>MIC values were 0 (−31.5, 96.2) and 0 (0, 96.2)%, respectively, with an r2 value of 0.727 from among the 54 patients with PK for whom population mean predicted and individual predicted f%T>MIC values could be determined. Examination of this subset of data demonstrated the population mean predicted f%T>MIC values in patients to be an unbiased and acceptably precise surrogate for individual predicted f%T>MIC values. In all patients, f%T>MIC values ranged from 0 to 100, with a median (25th percentile, 75th percentile) of 100 (84.2, 100). For patients with S. aureus infections, f%T>MIC values ranged from 38.3 to 100 and the median (25th percentile, 75th percentile) was 100 (82.5, 100).
PK-PD analyses. (i) Univariable analyses.
Microbiological responses by f%T>MIC ranges for all of the patients and those with S. aureus infections are shown in Table 3. As shown by these data, 97.3% (512/526) and 99.1% (419/423) of all of the patients and those with S. aureus infections, respectively, had f%T>MIC values of >50. CART-derived f%T>MIC thresholds of 54.2 and 55 were found to be significantly associated with microbiological response based on data for all of the patients and those with S. aureus infections (P = 0.001 and 0.023), respectively. The percentage of patients with microbiological success was 95.3% (486/510) for those with f%T>MIC values of ≥54.2 and 68.8% (11/16) for those with f%T>MIC values below this threshold. For patients with S. aureus infections, the percentage of patients with microbiological success was 96.2% (401/417) for those with f%T>MIC values of ≥55 and 66.7% (4/6) for those with f%T>MIC values below this threshold.
TABLE 3.
Microbiological response by ceftaroline f%T>MIC categories for all ME patients and those with S. aureus infections
| f%T>MIC | All patients |
Patients with S. aureus infections |
||||
|---|---|---|---|---|---|---|
| n | Success | Failure | n | Success | Failure | |
| ≤10 | 7 | 57.1 (4/7)a | 42.9 (3/7) | 0 | (0/0) | (0/0) |
| >10 to ≤50b | 7 | 85 (6/7) | 14.3 (1/7) | 4 | 75.0 (3/4) | 25.0 (1/4) |
| >50 to ≤60 | 12 | 8.3 (10/12) | 16.7 (2/12) | 10 | 80 (8/10) | 20.0 (2/10) |
| >60 to ≤70 | 36 | 94.4 (34/36) | 5.56 (2/36) | 34 | 97.1 (33/34) | 2.94 (1/34) |
| >70 to ≤80 | 46 | 93.5 (43/46) | 6.52 (3/46) | 44 | 93.2 (41/44) | 6.82 (3/44) |
| >80 to ≤90 | 52 | 90.4 (47/52) | 9.62 (5/52) | 50 | 92.0 (46/50) | 8.00 (4/50) |
| >90 to <100 | 45 | 97.8 (44/45) | 2.22 (1/45) | 43 | 100 (43/43) | 0 (0/43) |
| 100 | 321 | 96.3 (309/321) | 3.74 (12/321) | 238 | 97.1 (231/238) | 2.94 (7/238) |
| Total | 526 | 94.5 (497/526) | 5.51 (29/526) | 423 | 95.7 (405/423) | 4.27 (18/423) |
Data are expressed as the percentage of patients (number of patients/total).
The small number of patients with f%T>MIC values of <50 did not allow f%T>MIC values to be broken down into several categories within this range.
With a low number of observed failures in the analysis populations, there is an expected degree of uncertainty in the identification the CART-derived thresholds. This uncertainty was demonstrated by the assessment of 1,000 bootstrap samples to derive estimated 10th and 90th percentiles for the sampling distribution of the CART-derived f%T>MIC thresholds. For the f%T >MIC target of 54.2 based on data from all of the patients, the 10th and 90th percentiles were 26.7 and 88.3, respectively. For the f%T >MIC target of 55 based on data from patients with S. aureus infections, they were 55.0 and 90.8, respectively. Additionally, given that only 16 out of the 526 patients and 6 out of the 423 patients with S. aureus infections had f%T>MIC values below the CART-derived thresholds of 54.2 and 55, respectively, any interpretation of the data and model results for patients with f%T>MIC values below these thresholds should take into consideration the limited number of cases below these thresholds and, as a result, the high model uncertainty in these ranges.
Univariable logistic regression analysis results for microbiological response and f%T>MIC evaluated as a continuous variable demonstrated significant relationships based on data from all of the patients (P < 0.001) and those with S. aureus infections (P = 0.026). As shown in Fig. 1, these univariable relationships between microbiological response and f%T>MIC are shown for all of the ME patients (panel A) and those with S. aureus infections (panel B). Fitted functions for the univariable relationship between the probability of microbiological response and f%T>MIC are represented by the solid black lines, with 95% confidence bands derived from the standard error of the fitted logistic regression model parameters shown by the gray lines. The observed proportion of successful microbiological responses among patients with f%T>MIC values of <100 in three groups of the same size is shown by the dashed lines. In each of the images shown, the fitted univariable logistic regression functions are overlaid on a histogram showing the distribution of f%T>MIC values for that population. Wider 95% confidence intervals (CIs) around the logistic regression relationship at f%T>MIC values of <50 were likely a function of the limited number of patients with very high MICs (the MIC90 was 0.5 mg/liter) and, hence, low f%T>MIC values. Similar such univariable relationships between clinical response and f%T>MIC evaluated as a continuous variable were evident (P = 0.003 and 0.07, respectively).
FIG 1.
Univariable relationships between f%T>MIC values and microbiological response based on data for all of the patients (A) and those with S. aureus infections (B). The fitted logistic regression function for each univariable relationship is shown by the black line, with the 95% confidence bands around this function shown by the gray lines. The observed proportion of successful microbiological responses among patients with f%T>MIC values of <100 in three groups of the same size is shown by the dashed lines. Each fitted functions is overlaid on a histogram showing the population-specific distribution of f%T>MIC values.
Univariable relationships with f%T>MIC were relatively stronger for microbiological response than for a clinical response for all of the patients and those with S. aureus infections. As expected, given that microbiological response is frequently presumed from clinical response, the concordance between microbiological and clinical responses (see Table S1 in the supplemental material) was strong for all of the patients and those with S. aureus infections, with seven or fewer discordant outcomes among the individual comparisons. Given this concordance and the relatively stronger univariable relationships for microbiological response than for clinical response, multivariable analyses using data for all of the patients and those with S. aureus infections were limited to microbiological response. Table S3 in the supplemental material shows a summary of the results of the univariable analyses that were conducted on the basis of data for all of the patients and those with S. aureus infections. The independent variables listed represent those for which a univariable relationship was associated with a P value of ≤0.2 for at least one of the analysis populations. Independent variables for which P values were ≤0.2 were considered for multivariable analyses. Multiple forms of continuous independent variables, which were evaluated as such and as categorical variables on the basis of some grouping by percentiles and by two- and three-group splits to form two- and three-group independent variables, respectively, are shown for disease severity scores and MICs even when P values of the different forms exceeded 0.2. Separate multivariable models that evaluated different forms of independent variables with P value of ≤0.2 were considered.
A summary of univariable analysis results evaluating the relationships between microbiological response and f%T>MIC (evaluated in all forms) for the subset of 104 patients with PK data and those with S. aureus infections (n = 80) is shown in Table S4 in the supplemental material. Evaluations based on these data failed to demonstrate PK-PD relationships for efficacy.
(ii) Multivariable analyses.
Final multivariable logistic regression models for microbiological response based on data from all of the patients and those with S. aureus infections are shown in Tables 4 and 5, respectively. Each model contained patient factors that were significant predictors of microbiological response in addition to f%T>MIC. Patient age of <55 years and an infection type other than a wound infection (e.g., major abscess, an infected ulcer, an infected burn, an infected bite, deep/extensive cellulitis, a lower-extremity ABSSSI in patients with diabetes mellitus or peripheral vascular disease) were each predictive of microbiological success based on data from all of the patients. The absence of both diabetes mellitus and a wound infection (compared to the above-described other infection types) were each predictive of microbiological success in patients with S. aureus infections.
TABLE 4.
Multivariable logistic regression model for factors associated with microbiological success based on the data for all of the patientsa
| Independent variable | Parameter estimate | Odds ratio (95% CI) | P value |
|---|---|---|---|
| Age of <55 yearsb | 1.66 | 5.24 (2.29, 12.0) | <0.0001 |
| f%T>MICc | 0.311 | 1.36 (1.18, 1.58) | <0.0001 |
| Wound infectiond | −1.17 | 0.310 (0.129, 0.746) | 0.009 |
Included were all of the patients in the ME population (n = 526). The overall chi-square P value of the multivariable model was <0.000001.
Age was evaluated as a categorical variable based on the CART-derived threshold of 55 years; the reference group was patients ≥55 years old.
The parameter estimate and odds ratio for microbiological success relative to f%T>MIC, which was evaluated as a continuous variable, correspond to an increase per 10 percentage points of f%T>MIC.
Reference group: major abscess, infected ulcer, infected burn, infected bite, deep/extensive cellulitis, lower-extremity ABSSSI in patients with diabetes mellitus or peripheral vascular disease, or other.
TABLE 5.
Multivariable logistic regression model for factors associated with microbiological success based on data of all for the patients with S. aureus infectionsa
| Independent variable | Parameter estimate | Odds ratio (95% CI) | P value |
|---|---|---|---|
| Presence of diabetes mellitusb | −1.91 | 0.148 (0.054, 0.412) | 0.0002 |
| f%T>MICc | 0.299 | 1.35 (0.996, 1.83) | 0.05 |
| Wound infectiond | −1.45 | 0.234 (0.076, 0.718) | 0.01 |
Included were all of the patients in the ME population with S. aureus infections (n = 423). The overall chi-square P value of the multivariable model was 0.00002.
Reference group: patients without diabetes mellitus.
The parameter estimate and odds ratio for microbiological success relative to f%T>MIC, which was evaluated as a continuous variable, correspond to an increase per 10 percentage points of f%T>MIC.
Reference group: major abscess, infected ulcer, infected burn, infected bite, deep/extensive cellulitis, lower-extremity ABSSSI in patients with diabetes mellitus or peripheral vascular disease, or other.
Although evaluated, interactions among independent variables retained in each model were not apparent, suggesting that the nature of the relationship between microbiological response and f%T>MIC was not dependent on any of these independent variables. Further evaluation of the impact of independent variables on the relationships between f%T>MIC and microbiological response using CART within patient subsets defined by the presence or absence of an independent variable was also carried out for both analysis populations. The results of these additional analyses demonstrated the following. (i) Among the patients with diabetes mellitus, an age of ≥55 years, or a wound infection, too few were below the CART-derived f%T>MIC thresholds of 54.2 and 55 based on data from all of the patients and those with S. aureus infections, respectively, to allow precise assessments of this group. (ii) There was no evidence of a f%T>MIC threshold above 54.2 or 55 that appears beneficial in differentiating percent probabilities of microbiological success among patients with wound infections. (iii) For patients with diabetes mellitus in both analysis populations, a f%T>MIC threshold of 90.4, which was predictive of a higher percentage of microbiological success, was identified, whereas a f%T>MIC threshold of 30 was identified for all of the patients without diabetes mellitus (see Tables S5 and S6 in the supplemental material). (iv) Among all of the patients and those with S. aureus infections who were ≥55 years of age, f%T>MIC thresholds of 82.5 and 90.4, respectively, were identified. However, given that diabetes mellitus was present in 34.2% of the patients ≥55 years of age and 11.3% of the patients <55 years of age (and in 35.1 and 10%, respectively, of the patients with S. aureus infections), the effects of age and diabetes mellitus on the percent probability of microbiological response may be confounded. The observation of higher percentages of microbiological success in older patients with higher f%T>MIC values may be, in part, due to the higher percentage of diabetics in this older subpopulation. Given that the above-described subset evaluation was based on limited samples of patients and a limited range of f%T>MIC values, with the majority of these values being high, the actual thresholds identified are imprecise, making interpretation problematic. Further evaluation of this signal in diabetic patients with ABSSSI is required to substantiate these findings. In addition, given that diabetics were older and had worse renal function, as evidenced by lower CLCRs than nondiabetic patients (a median age of 57 versus 44 years and a median CLCR of 85.7 versus 101 ml/min/1.73 m2), it was not surprising that f%T>MIC for ceftaroline was >90 in more than 75% of the diabetic patients in both analysis populations. These data suggest that the standard dose of 600 mg q12h should provide adequate exposure for the majority of diabetics.
Figure 2A and B show the comparison of model-predicted probabilities and observed proportions of microbiological responses as a function of individual cohorts of patients described by independent variables included in each of the two final multivariable logistic regression models shown in Tables 4 and 5, respectively. For the assessment of each model, f%T>MIC was categorized by using the median value of 100 for the purpose of comparing model-predicted probabilities and observed proportions of successful microbiological responses by cohort. The agreement between model-predicted probabilities and observed proportions of responses was reasonable, even for cohorts of inadequate sample size, for which observed proportions have inherently large variability.
FIG 2.
Assessment of model performance for the multivariable logistic regression models for factors associated with microbiological success based on data for all of the patients (A) and those with S. aureus infections (B).
DISCUSSION
The objective of this analysis was to evaluate the PK-PD relationships for efficacy in patients with ABSSSI who had received ceftaroline fosamil by using data from two phase 2 and two phase 3 clinical studies (5–9). Often, such analyses are limited by the number of patients for whom PK data were collected. When PK-PD data sets are of limited sample size and include data for patients who received fixed dosing regimens (thus resulting in a narrow range of exposures) or for whom efficacy rates were high, the inferences that can be made about PK-PD relationships for efficacy are limited. The lack of homogeneity in the patient population arising from diverse inclusion criteria and/or polymicrobial infections further hinders the opportunity to identify PK-PD relationships for efficacy.
For the analysis described herein, the sample size of patients from whom PK data were collected was limited (104 of 526 ME patients) and evaluations based on this subset of patients failed to demonstrate PK-PD relationships for efficacy. Additionally, the percentages of clinical and microbiological success were high (and hence, the number of failures was low), the ceftaroline fosamil dosing regimen was fixed, the MIC distribution was narrow (hence, the narrow f%T>MIC range), and approximately 27% of all of the patients or those with S. aureus infections had polymicrobial infections, as determined by the number of pathogens isolated at baseline. Given the lack of findings of the subset analysis conducted with data from patients with sufficient PK data and the above-described factors, data from a larger number of patients were needed to increase the likelihood of identifying PK-PD relationships for efficacy.
One approach to increasing the number of patients to include in a PK-PD analysis is to use some reasonable surrogate measure of exposure so as to include all of the evaluable patients, including those for whom PK data were collected. The development of a population PK model incorporating individual patient covariates provides a mechanism for estimating population mean predicted exposures in all evaluable patients on the basis of their demographic characteristics. The PK variability of drugs that are predominantly renally cleared (e.g., aminoglycosides, β-lactams) can often be explained by patient covariates. In the case of ceftaroline, renal elimination is the major route of elimination along with conversion to the M-1 metabolite of ceftaroline (3). Therefore, it is not surprising that CLCR and age were statistically significant covariate effects in the population PK model for ceftaroline fosamil and active ceftaroline and thus helped explain a substantial proportion of the variability in the total clearance of ceftaroline (11).
Using the above-described population PK models (11) to provide population mean predicted f%T>MIC values for ME patients without PK data (n = 422) and combining these data with individual predicted f%T>MIC values for ME patients with PK data (n = 104), a data set containing of 526 patients (of whom 423 had S. aureus infections at baseline) was available for analysis. Univariable PK-PD analyses revealed significant relationships between the probability of microbiological success and f%T>MIC, whether evaluated continuously or categorically. The wider 95% CI around the function describing f%T>MIC evaluated as a continuous variable at values of <50 was indicative of poor agreement between observed and predicted responses. However, uncertainty of any mathematical function describing the PK-PD relationships in this region would be expected because of the limited number of patients with very high MICs (i.e., the MIC90 was 0.5 mg/liter) and, hence, low f%T>MIC values.
Findings for ceftaroline against S. aureus in a neutropenic-murine thigh infection model demonstrated that a median (min, max) f%T>MIC of 26 (15, 36) was associated with net bacterial stasis (10), an endpoint correlated with a high percentage of successful outcomes in patients with ABSSSI (2). Given the limited number of low f%T>MIC values and the above-described uncertainty of the mathematical function in this region, it is difficult to evaluate the concordance of these data and precisely estimate the probability of microbiological response at a f%T>MIC of 26. Nonetheless, the evaluation of the results of both sets of analyses serve to demonstrate that patients with ABSSSI on a ceftaroline fosamil dosing regimen of 600 mg q12h achieved f%T>MIC values associated with the upper plateau of the nonclinical PK-PD relationship for efficacy. Moreover, results of the above-described nonclinical PK-PD analysis for S. aureus also serve to provide an understanding of the lower bound of f%T>MIC values associated with efficacy. Such data are not only useful early in drug development to support dose selection but also useful to support the establishment of new and the reevaluation of existing in vitro susceptibility test interpretive criteria (19–22).
Multivariable analyses for microbiological response served to demonstrate that higher f%T>MIC values were significantly associated with an increased probability of microbiological success, even when such relationships were evaluated in the context of other patient factors. Significant independent variables retained in these models included age, the presence of diabetes mellitus, and the infection type. A patient age of <55 years, the absence of diabetes mellitus, and an infection type other than a wound infection were predictive of microbiological success. Given the underlying comorbidities in this patient population and the potential for complications when comorbidities are present, the impact of each of these patient factors on microbiological response was not unexpected (23–25). For patients with diabetes mellitus, the course of infection is often more complicated than that for other patients. Possible causes include defects in immunity among such patients. With regard to cellular innate immunity, studies have shown decreased functions (e.g., chemotaxis, phagocytosis, killing) of diabetic polymorphonuclear cells and diabetic monocytes/macrophages compared to the cells of controls. Additionally, a high-glucose environment is associated with greater organism virulence (25). Lastly, reduced blood flow to the extremities of diabetic patients may reduce drug exposure at the infection site, which would be expected to affect response. Indeed, as described herein, diabetic patients had higher percentages of microbiological success at higher f%T>MIC values. The finding that the presence of a major abscess, an infected ulcer, an infected burn, an infected bite, deep/extensive cellulitis, or a lower-extremity ABSSSI compared to wound infection was predictive of microbiological success can perhaps be explained by the fact that infection types other than wound infection have lighter bacterial burdens, can be drained, or are surgically debrided.
The results of the analyses described herein focus on efficacy endpoints assessed at TOC, endpoints that have been traditionally evaluated for agents to treat patients with ABSSSI. However, the U.S. FDA introduced draft guidance in 2010 that described the assessment of clinical response at 48 to 72 h (26). Clinical response was defined as cessation of the spread of the redness, edema, and/or induration of the lesion or a reduction in the size of the area showing redness, edema, and/or induration at 48 to 72 h after enrollment and resolution of fever (i.e., absence of fever). The definition of a clinical response at 48 to 72 h has since been modified to a reduction in lesion size of ≥20% from baseline (27).
The FDA's and sponsor's sensitivity analysis of clinical responders on day 3 among patients in the exploratory modified intent-to-treat (MITT) population, which included all of the randomized patients who received any study drug and with a lesion size of >75 cm2 and deep and/or extensive cellulitis, a major abscess, or an infected wound from among the two phase 3 studies, CANVAS 1 and 2, supported the noninferiority of ceftaroline fosamil to vancomycin with aztreonam. The noninferiority margin was less than 4% for both trials. The results of these analyses for a clinical response on day 3, which was defined as cessation of infection spread and absence of fever, demonstrated that 148/200 (74.0%) versus 135/209 (64.6%) and 148/200 (74.0%) versus 128/188 (68.1%) patients in the MITT population receiving ceftaroline fosamil versus vancomycin and aztreonam in CANVAS 1 and 2, respectively, achieved a clinical response on day 3 (28, 29). Additionally, the FDA's analysis demonstrated that 115/200 (57.5%) versus 106/209 (50.7%) and 120/200 (60%) versus 105/188 (55.9%) patients in the MITT population receiving ceftaroline fosamil versus vancomycin and aztreonam in CANVAS 1 and 2, respectively, achieved a 20% reduction from baseline in lesion size on day 3 (28). These post hoc early treatment effect analyses, which favored ceftaroline fosamil over vancomycin with aztreonam, suggest that ceftaroline fosamil monotherapy may provide a greater benefit than combination therapy with vancomycin with aztreonam. Clinical trials undertaken since the introduction of the FDA guidance for ABSSSI (26), which include the collection of serial lesion size measurements, provide the opportunity to conduct PK-PD analyses with more-sensitive efficacy endpoints and thus, the opportunity to better characterize PK-PD relationships for efficacy and identify exposures associated with favorable early and TOC endpoints (30).
Population PK and PK-PD analyses are conducted to confirm early preclinical and clinical predictions. Herein, relationships between ceftaroline exposure, as measured by f%T>MIC values, and patient response to therapy at TOC were identified. Univariable analyses for microbiological response demonstrated that a higher f%T>MIC was significantly associated with a higher probability of microbiological success. Multivariable analyses for microbiological response demonstrated that a higher f%T>MIC was significantly associated with an increased probability of microbiological success, even when such a relationship was evaluated in the context of other significant independent variables, including age, the presence of diabetes mellitus, and infection type. These analyses confirm that a ceftaroline fosamil dosing regimen of 600 mg q12h provides exposures associated with the upper plateau of the PK-PD relationship for efficacy.
Supplementary Material
ACKNOWLEDGMENT
This study was supported by Forest Laboratories, Inc. Forest Laboratories, Inc., was involved in the study design; the collection, analysis, and interpretation of data; and the decision to present these results. Scientific Therapeutics Information, Inc., provided editorial coordination, which was funded by Forest Research Institute, Inc.
We thank Kim Charpentier at the Institute for Clinical Pharmacodynamics, Latham, NY, for her assistance in the preparation of the manuscript.
Footnotes
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AAC.02531-14.
REFERENCES
- 1.Bonapace CR. 2004. In vitro/animal models to support dosage selection: FDA perspective. IDSA/ISAP/FDA Workshop. 16 April 2004. U.S. Food and Drug Administration, Rockville, MD: http://www.fda.gov/downloads/Drugs/DrugSafety/InformationbyDrugClass/UCM187288.ppt. [Google Scholar]
- 2.Ambrose PG, Bhavnani SM, Rubino CM, Louie A, Gumbo T, Forrest A, Drusano GL. 2007. Pharmacokinetics-pharmacodynamics of antimicrobial therapy: it's not just for mice anymore. Clin Infect Dis 44:79–86. doi: 10.1086/510079. [DOI] [PubMed] [Google Scholar]
- 3.Forest Pharmaceuticals, Inc. 2013. Teflaro package insert. Forest Pharmaceuticals, Inc; St. Louis, MO. [Google Scholar]
- 4.AstraZeneca AB. 2012. Zinforo summary of product characteristics. AstraZeneca AB, Sodertalje, Sweden. [Google Scholar]
- 5.Talbot GH, Thye D, Das A, Ge Y. 2007. Phase 2 study of ceftaroline versus standard therapy in treatment of complicated skin and skin structure infections. Antimicrob Agents Chemother 51:3612–3616. doi: 10.1128/AAC.00590-07. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Cerexa, Inc. 2012. Efficacy and safety of ceftaroline versus linezolid in subjects with complicated skin and skin structure infections. ClinicalTrials.gov, Bethesda, MD: http://clinicaltrials.gov/show/NCT00633152. [Google Scholar]
- 7.Corey GR, Wilcox MH, Talbot GH, Thye D, Friedland D, Baculik T. 2010. CANVAS 1: the first phase III, randomized, double-blind study evaluating ceftaroline fosamil for the treatment of patients with complicated skin and skin structure infections. J Antimicrob Chemother 65(Suppl 4):iv41–iv51. doi: 10.1093/jac/dkq254. [DOI] [PubMed] [Google Scholar]
- 8.Wilcox MH, Corey GR, Talbot GH, Thye D, Friedland HD, Baculik T. 2010. CANVAS 2: the second phase III, randomized, double-blind study evaluating ceftaroline fosamil for the treatment of patients with complicated skin and skin structure infections. J Antimicrob Chemother 65(Suppl 4):iv53–iv65. doi: 10.1093/jac/dkq255. [DOI] [PubMed] [Google Scholar]
- 9.Corey GR, Wilcox M, Talbot GH, Friedland HD, Baculik T, Witherell GW, Critchley I, Das AF, Thye D. 2010. Integrated analysis of CANVAS 1 and 2: phase 3, multicenter, randomized, double-blind studies to evaluate the safety and efficacy of ceftaroline versus vancomycin plus aztreonam in complicated skin and skin-structure infection. Clin Infect Dis 51:641–650. doi: 10.1086/655827. [DOI] [PubMed] [Google Scholar]
- 10.Andes D, Craig WA. 2006. Pharmacodynamics of a new cephalosporin, PPI-0903 (TAK-599), active against methicillin-resistant Staphylococcus aureus in murine thigh and lung infection models: identification of an in vivo pharmacokinetic-pharmacodynamic target. Antimicrob Agents Chemother 50:1376–1383. doi: 10.1128/AAC.50.4.1376-1383.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Van Wart SA, Forrest A, Khariton T, Rubino CM, Bhavnani SM, Riccobene TA, Ambrose PG. 2013. Population pharmacokinetics of ceftaroline in patients with acute bacterial skin and skin structure infections or community-acquired pneumonia. J Clin Pharmacol 53:1155–1167. doi: 10.1002/jcph.153. [DOI] [PubMed] [Google Scholar]
- 12.Cockcroft DW, Gault MH. 1976. Prediction of creatinine clearance from serum creatinine. Nephron 16:31–41. doi: 10.1159/000180580. [DOI] [PubMed] [Google Scholar]
- 13.DuBois D, DuBois DF. 1916. A formula to estimate the approximate surface area if height and weight be known. Arch Int Med 17:863–871. [Google Scholar]
- 14.Sheiner LB, Beal SL. 1981. Some suggestions for measuring predictive performance. J Pharmacokinet Biopharm 9:503–512. doi: 10.1007/BF01060893. [DOI] [PubMed] [Google Scholar]
- 15.Systat Software, Inc. 2004. Systat 11, statistics II, Systat Software, Inc., Richmond, CA. [Google Scholar]
- 16.R Development Core Team. 2010. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria: http://www.r-project.org. [Google Scholar]
- 17.Wilson SE, Solomkin JS, Le V, Cammarata SK, Bruss JB. 2003. A severity score for complicated skin and soft tissue infections derived from phase III studies of linezolid. Am J Surg 185:369–375. doi: 10.1016/S0002-9610(02)01411-3. [DOI] [PubMed] [Google Scholar]
- 18.Hosmer DW, Lemeshow S. 1989. Applied logistic regression. Wiley, New York, NY. [Google Scholar]
- 19.Dudley MN, Ambrose PG, Bhavnani SM, Craig WA, Ferraro MJ, Jones RN. 2013. Antimicrobial Susceptibility Testing Subcommittee of the Clinical and Laboratory Standards Institute. Background and rationale for revised Clinical and Laboratory Standards Institute interpretive criteria (breakpoints) for Enterobacteriaceae and Pseudomonas aeruginosa: I. Cephalosporins and aztreonam. Clin Infect Dis 56:1301–1309. doi: 10.1093/cid/cit017. [DOI] [PubMed] [Google Scholar]
- 20.Bhavnani SM, Hammel JP, Jones RN, Sader HS, Ambrose PG. 2011. Pharmacokinetic-pharmacodynamic evaluations in support of in vitro interpretive criteria for fluoroquinolones against salmonellae and Enterobacteriaceae, abstr A2-043 51st Interscience Conference on Antimicrobial Agents and Chemotherapy, Chicago, IL, 17–20 September 2011. [Google Scholar]
- 21.Ambrose PG, Meagher AK, Passarell JA, Van Wart SA, Cirincione BB, Bhavnani SM, Ellis-Grosse E. 2009. Application of patient population-derived pharmacokinetic-pharmacodynamic relationships to tigecycline breakpoint determination for staphylococci and streptococci. Diagn Microbiol Infect Dis 63:155–159. doi: 10.1016/j.diagmicrobio.2008.10.011. [DOI] [PubMed] [Google Scholar]
- 22.Bhavnani SM, Dudley MN, Landersdorfer C, Drusano GL, Craig WA, Jones RN, Ambrose PG. 2010. Pharmacokinetic-pharmacodynamic basis for CLSI carbapenems susceptibility breakpoint changes, abstr A1-1382 50th Interscience Conference on Antimicrobial Agents and Chemotherapy Boston, MA, 12–15 September, 2010. [Google Scholar]
- 23.Bhavnani SM, Iyer V, Hammel J, Forrest A, Rubino CM, Van Wart SA, Prakhya A, McCollam J, Ambrose PG. 2008. Pharmacokinetic-pharmacodynamic oritavancin efficacy in patients with complicated skin and skin-structure infection, abstr A-3563 48th Interscience Conference on Antimicrobial Agents and Chemotherapy/46th Annual Meeting of the Infectious Diseases Society of America, Washington DC, 25–28 October 2008. [Google Scholar]
- 24.Lipsky BA, Itani KMF, Weigelt JA, Joseph W, Paap CM, Reisman A, Myers DE, Huang DB. 2011. The role of diabetes mellitus in the treatment of skin and skin structure infections caused by methicillin-resistant Staphylococcus aureus: results from three randomized controlled trials. Int J Infect Dis 15:e140–e146. doi: 10.1016/j.ijid.2010.10.003. [DOI] [PubMed] [Google Scholar]
- 25.Geerlings SE, Hoepelman AIM. 1999. Immune dysfunction in patients with diabetes mellitus (DM). FEMS Immunol Med Microbiol 26:259–265. doi: 10.1111/j.1574-695X.1999.tb01397.x. [DOI] [PubMed] [Google Scholar]
- 26.U.S. Food and Drug Administration. 2010. Guidance for industry (draft). Acute bacterial skin and skin structure infections: developing drugs for treatment. U.S. Food and Drug Administration, Rockville, MD. [Google Scholar]
- 27.U.S. Food and Drug Administration. 2013. Guidance for industry. Acute bacterial skin and skin structure infections: developing drugs for treatment. U.S. Food and Drug Administration, Rockville, MD. [Google Scholar]
- 28.U.S. Food and Drug Administration. 2010. Briefing document for Anti-Infective Drugs Advisory Committee Meeting. Ceftaroline fosamil for the treatment of community-acquired bacterial pneumonia and complicated skin and skin structure infections. U.S. Food and Drug Administration, Rockville, MD. [Google Scholar]
- 29.Friedland HD, O'Neal T, Biek D, Eckburg PB, Rank DR, Llorens L, Smith A, Witherell GW, Laudano JB, Thye D. 2012. CANVAS 1 and 2: analysis of clinical response at day 3 in two phase 3 trials of ceftaroline fosamil versus vancomycin plus aztreonam in treatment of acute bacterial skin and skin structure infections. Antimicrob Agents Chemother 56:2231–2236. doi: 10.1128/AAC.05738-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bhavnani SM, Hammel JP, Van Wart SA, Rubino CM, Davenport JM, McIntyre G, Ambrose PG. 2012. Pharmacokinetic-pharmacodynamic analysis for efficacy of JNJ-Q2 using phase 2 data from patients with acute bacterial skin and skin structure infections, abstr A-1962 52nd Interscience Conference on Antimicrobial Agents and Chemotherapy San Francisco, CA, 9–12 September, 2012. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.