The New England Journal of Medicine published two rather baffling papers this month: one entitled “Omadacycline for Community-Acquired Bacterial Pneumonia,” and the other “Omadacycline for Acute Bacterial Skin and Skin-Structure Infections.” Reading them, I found myself full of questions. Questions like: How did this get published in NEJM? Did this study need to be done? Why is this how we do R&D for new antibiotics?
To be clear, I don’t mean to cast aspersions on the researchers who designed this drug and then ran the trials to jump it through the necessary hoops to get it into the hands of patients. These ladies and gentleman have certainly done more for humanity than I have, and more power to them. But can we talk for a moment about why and whether any of this work really needed to be done?
Let’s start with a brief word on the papers:
In the first article, the authors randomized 774 patients with community-acquired bacterial pneumonia to receive either omadacycline or moxifloxacin. When I say the authors, it’s worth mentioning that Paratek Pharmaceuticals designed and conducted the trial, designed and conducted the statistical analysis, and paid a medical writer to handle the first draft of the manuscript. It would be impertinent for me to ask why anybody other than Paratek’s staff was listed as an author rather than a collaborator, so I won’t do that. Keep in mind, though, that a Cochrane review indicates that industry-sponsored clinical trials are more likely to report the efficacy of the tested product and more likely to make favorable conclusions about the product regardless of the trial results.
Anyway, the patients started with IV therapy and could switch to oral versions of the same drug after three days, with each continued for a total 7-14 days. This is my first beef – we treat CAP with five days of therapy, often switching to orals by day 2 or 3, and this trial doesn’t really tell me how well omadacycline will perform under those conditions. Or am I supposed to be prescribing two weeks for CAP when I want to use omadacycline? If so: nah, pass, I’ll take amox-clav and azithromycin.
The primary endpoint of the trial was early clinical response, assessed by a symptom score at day 3-5, and the study used a non-inferiority design with a margin of 10%. The study enrolled patients with moderate to severe pneumonia (85% were PSI class III or IV); the two arms of the study were well matched, and >40% of participants were over age 65. Rates of early response between omadacycline and moxifloxacin were similar (81% vs 83%) and met noninferiority criteria. The two drugs also produced similar rates of response at the post-treatment visit (88% vs 85%). In subgroup analyses, one concerning signal was a lower rate of post-treatment clinical response with omadacycline in patients who had bacteremic pneumonia (11/15 or 73% with omadacycline vs 15/18 or 83% with moxifloxacin). This isn’t surprising given the tetracycline class’s poor persistence in blood, and unfortunately makes us recall tigecycline’s black box warning for pneumonia – something underscored by the “higher” (2% vs 1%) rate of mortality with omadacycline in the study. On the bright side, omadacycline was associated with less diarrhea than moxifloxacin, and not a single case of CDI was observed in the omadacycline arm, suggesting that omadacycline will continue the tradition of tetracyclines being neutral or protective again C.difficile infection.
In the second trial (similarly authored by Paratek Pharmaceuticals), investigators with randomized adults with eligible soft tissue infections (cellulitis, wound infections, and abscesses, not associated with a chronic skin lesion, ulcer or wound) 1:1 to receive 7-14 days of either omadacycline or linezolid. Again, these durations seem a bit long, but in this case linezolid’s FDA indication for soft tissue infection is a 7-14 day course, so the comparison makes sense. Again, the primary endpoint was early clinical response, which in this setting meant survival plus at least a 20% reduction in the size of the lesion within 72 hours of initiating antibiotic therapy. Also again, the study used a noninferiority design with a 10% margin. A total 655 patients were randomized and 627 were included in the modified intention-to-treat population. The clinical characteristics of patients in the two study arms were similar; 70% had an identified gram-positive pathogen, and about a quarter had MRSA. Early response rates were similar between the omadacycline and linezolid recipients (85% vs 86%, meeting criteria for noninferiority), as were the rates of lesion resolution. Again, while the numbers were small, there was a signal towards poorer outcomes with omadacycline in patients with bacteremia – in this subset, only 9/11 (82%) of omadacycline recipients experienced a clinical response at the end of therapy, versus 9/9 (100%) of linezolid recipeints. Adverse event rates were similar.
So, to summarize, we have a paper saying that omadacycline usually works as well as moxifloxacin when each are given for far too long for community-acquired pneumonia (a condition that can be managed by cheaper and narrower-spectrum agents with better and longer track records of safety than either of those drugs), and we have another paper saying that omadacycline works as well as linezolid for skin and soft tissue infection, an indication for which omadacycline has no clear benefit over trimethoprim-sulfa or doxycycline or minocycline, all cheaper agents with decades more clinical experience behind them.
…cool, I guess?
This brings me to what I actually want to say about these two clinical trials, which is: the current FDA approval process for new antibiotics doesn’t make sense to me. On its website, the FDA tells the public that “FDA reviewers analyze the condition or illness for which the drug is intended and evaluate the current treatment landscape, which provide the context for weighing the drug’s risks and benefits.” When it comes to antibiotics, however, I’m not sure this is the case.
The root problem is that the FDA evaluates and approves antibiotics for indications proposed by drug manufacturers, even when that has little to no bearing on how the drug is actually going to be used. As a result, antibiotic manufacturers are incentivized to pursue indications that come with the easiest, fastest, and most likely to be positive trials. That means trials in community-acquired pneumonia, soft tissue infection, and UTI. These trials become necessary because they fulfill a federal regulatory requirement and get the drug on the market, but they may totally fail to inform patient care, or at best fail to answer the most pressing question: how will this drug fare under the circumstances in which it’s actually going to be used?
Take omadacycline. This is an MDR drug like tigecycline, active against a variety of highly resistant Gram-negative organisms, and as of this writing its unique virtue is its oral formulation. We should expect to see omadacycline being used as primary and stepdown therapy for invasive CRE and other MDR-GNR infections (e.g. line-associated bacteremias, complicated intraabdominal and wound infections, ventilator-associated pneumonias) in real life. In particular, we should expect to see omadacycline being used in addition to or in place of standard therapies for those infections, like colistin, tobramycin, and cephalosporin and carbapenem plus beta-lactamase inhibitor combination drugs. So, why doesn’t the FDA demand a trial of omadacycline under those circumstances, and why doesn’t it compare its efficacy and side effect profile to the best alternative therapies? Perhaps such a trial would be too difficult to run – too slow to recruit or too confounded by complex comorbidities to interpret. But in that case, what are really learning about omadacycline from a phase 3 trial done for cellulitis that we couldn’t have gotten from a large phase 2 trial enrolling a variety of infections? The latter would still provide us some evidence of the drug’s adverse event profile, and might provide a more complete assessment of the type of infections for which omadacycline produces or fails to produce a good outcome. In fact, for a drug like omadacycline, I’d much rather have a large phase II trial providing evidence for the drug’s safety profile and general performance across a number of infections, plus a focused animal study more exactly mimicking the situations in which I’d genuinely use the drug (e.g. a neutropenic pig model of MDR-GNR CRBSI and ventilator-associated pneumonia), than a phase 3 trial for an indication I’ll never use.
There’s a lot in the pharmaceutical industry and pharmaceutical development that needs fixing. But the FDA’s approach to antibiotic approval is certainly on the list.
And now, on to the medical literature:
Antimicrobial agents research this month included the problem of shifting goalposts in serial non-inferiority trials of antibiotic duration, minocycline’s breakpoints for Acinetobacter being too generous, limited but promising data for fosfomycin in MDR gram-negative bacterial prostatitis, and repurposing antabuse for NTM and clofazamine for MDR tuberculosis.
ID diagnostics research covered the microbiology of recurrent diabetic foot osteomyelitis, clinical risk scores for PJI, procalcitonin to stop antibitoics in COPD exacerbation, and the increased microbiologic yield of parapneumonic fluid testing with PCR.
General ID research included the utility of the influenza vaccine in a “bad vaccine year,” why we ought to use the high-dose vaccine in patients over 65, detection of asymptomatic simian malaria in people who live around monkeys, and the US Air Force’s inability to appropriately order and interpret Lyme serologies (which, to be fair, no one else can do either).
HIV and STDs papers of interest this month included data on mediocre outcomes with integrase inhibitors in HIV-2 infections, gonorrhea resistance in women being mediated by menstruation via vaginal pH, predictors of doravirine resistance in NNRTI-experienced patients, and the superiority of dolutegravir over lopinavir-ritonavir as second-line therapy in resource limited settings.
Onc and transplant ID research include transplant-and-treat for HCV-infected heart donors, a feared complication of S.maltophilia bacteremia, ciprofloxacin’s failure to prevent BK viremia, and the prevalence of invasive fungal infection in lung transplant recipients.
Antimicrobial stewardship and infection control research included the utility of C.difficile colonization as a predictor of CDI during hospitalization, MRSA decolonization to prevent subsequent infection in patients being discharged from the hospital, and the role of early followup in OPAT to prevent re-admissions.