Omadacycline and linezolid produced similar overall outcomes in a patient-centered analysis of 1,347 participants from two randomized trials of acute bacterial skin and skin structure infections, according to a retrospective study, published in Open Forum Infectious Diseases.
The investigators retrospectively applied a desirability of outcome ranking (DOOR) endpoint developed by a multidisciplinary Antibacterial Resistance Leadership Group task force. DOOR is a patient-centered framework that ranks outcomes based on survival and three undesirable events: lack of clinical response, infectious complications and nonfatal serious adverse events. This approach allows treatment benefits and harms to be evaluated together. A DOOR probability of 50% means a randomly selected participant from either treatment group is equally likely to have the more desirable outcome; confidence intervals that include 50% indicate no statistically significant difference between groups.
“Using DOOR can provide clinicians and patients with a better understanding of the benefits and risks of new anti-infectives being studied to treat skin infections,” wrote the authors.
They applied the DOOR framework to two completed trials, OASIS-1 and OASIS-2. Both were phase 3, multicenter, double-blind, randomized noninferiority trials comparing omadacycline with linezolid for acute bacterial skin and skin structure infections (ABSSSI). OASIS-1 assessed intravenous-to-oral treatment, while OASIS-2 assessed oral treatment.
The DOOR analysis included 627 patients from OASIS-1 and 720 from OASIS-2 in the modified intention-to-treat populations. In OASIS-1, patients began intravenous omadacycline or linezolid, with the option to transition to oral treatment after 3 days. OASIS-2 compared oral omadacycline with oral linezolid. Therapy lasted 7 to 14 days in both studies.
DOOR assigned each patient in the trials a mutually exclusive rank based on survival and three undesirable events: absence of clinical response, infectious complications and nonfatal serious adverse events. Patients who were alive without any of those events received the most desirable rank 1, while patients who died received the least desirable rank 5. Patients who were alive with one, two or three events received intermediate ranks 2 through 4.
Two board-certified infectious diseases clinicians, blinded to treatment assignment, reviewed adverse events to determine whether they met criteria for infectious complications. Unresolved events were referred to the full task force.
About 80% of patients in each trial had the most desirable DOOR outcome: survival without any of the prespecified undesirable events. For both OASIS-1 and OASIS-2, the DOOR distribution between treatment arms was similar. The probability that a patient treated with omadacycline would have a more desirable outcome than a patient treated with linezolid was 50.6% (95% CI, 47.4%-53.7%) in OASIS-1 and 52.1% (95% CI, 49.2%-55.0%) in OASIS-2.
In OASIS-2, researchers also used changes from baseline in physical and mental health-related quality-of-life scores from the 36-Item Short Form Health Survey Version 2 (SF-36v2) to break ties between patients with the same DOOR rank. Incorporating either measure slightly shifted the DOOR probability in favor of omadacycline, but neither difference was statistically significant.
Overall, absence of clinical response, infectious complications, nonfatal serious adverse events and death were uncommon. The only infectious complication identified in either trial was progression of or new ABSSSI. The researchers found no significant treatment differences for any individual DOOR component, although absence of clinical response was numerically less frequent with omadacycline in both trials; the differences were not statistically significant.
A subgroup analyses examined patients according to diabetes status, infection type at enrollment, age, sex and identification of Staphylococcus aureus as a causative pathogen. OASIS-1 showed no significant treatment differences in the subgroups studied. In OASIS-2, the 45 individuals with diabetes had a 58.1% probability of a more desirable outcome with omadacycline than linezolid (95% CI, 50.0%-65.7%). However, the authors cautioned that the small subgroup made the finding hypothesis-generating.
Researchers also conducted partial-credit analyses in which hypothetical scoring systems assigned different values to intermediate DOOR ranks. These analyses showed no treatment differences. The authors noted that the scoring keys were intended to demonstrate feasibility and did not necessarily represent realistic patient preferences. Sensitivity analyses produced similar results when patients with missing or indeterminate outcomes were ranked above patients with clinical failure, classified as having clinical cure or excluded from the analysis.
The authors noted several limitations. Data were available from only two trials evaluating the same antibiotics and funded by the same pharmaceutical company. Because serious adverse events and deaths were rare, few patients fell into the three least desirable DOOR ranks, making it difficult to distinguish between treatments. The infectious complication review was conducted post hoc and relied on coded adverse event data that might not have captured complete event details.
The authors concluded that DOOR was feasible and called for future ABSSSI trials to prospectively incorporate DOOR and more specific health-related quality-of-life instruments. They also recommended studying whether prospective collection of additional adverse-event, symptom-duration and health-related quality-of-life data could improve DOOR’s ability to distinguish outcomes.
“DOOR allows clinicians and patients to have a more comprehensive understanding of the benefits and risks of choosing one treatment over another and provides a framework for sharing more information on the benefit-risk profile than is typically presented in clinical trials,” wrote the authors.
The study was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health. All authors reported funding support from the Antibacterial Resistance Leadership Group of the National Institutes of Health and the National Institute of Allergy and Infectious Diseases. Potential conflicts of interest also were reported, including grants, consulting fees, honoraria, royalties, travel support, advisory and data safety monitoring board roles, leadership positions, patents, stock or stock options, and other relationships with government, academic, professional and industry organizations.
Expert Insight
Corresponding author, Helen W. Boucher, MD of Tufts University School of Medicine in Boston, was invited to elaborate on the study's findings.
What data surprised you?
It was initially surprising that incorporating health-related quality-of-life scores as a tiebreaker in the DOOR analysis did not result in a meaningful change in the overall DOOR probability. This may be because the two study drugs truly produced similar patient outcomes, or it could reflect limitations in how healthy-related quality of life was measured. Quality of life was assessed using SF-36v2, which is a validated instrument but was not specifically designed for infectious diseases or ABSSSI. In addition, many SF-36v2 questions used a 4-week recall period, which may not align well with the timing of symptoms and recovery in patients with ABSSSI. Disease-specific patient-reported outcome measures may be more informative in future studies.
How might the findings influence clinical practice?
We believe that incorporating DOOR as a primary or secondary endpoint in ABSSSI clinical trials can provide clinicians and patients with a more complete understanding of the benefits and risks of new antibiotics. Although this study is unlikely to change day-to-day clinical practice immediately, it may help improve how future antibiotic trials are designed, analyzed and interpreted.
Is there anything else you'd like to say about this work?
We found that patients generally did very well in the original ABSSSI trials, with relatively few participants falling into the less desirable DOOR ranks (3 to 5). Future DOOR studies that prospectively collect more detailed patient-centered data, including common adverse effects and quality-of-life measures, may provide additional insights. In addition, DOOR may be particularly valuable in investigator-initiated and pragmatic clinical trials, which often enroll more clinically complex or acutely ill patients and may therefore provide greater opportunity to distinguish differences between treatment strategies.
