(The following is a brief discussion of the OVIVA trial followed by a narrative review of the history of osteomyelitis treatment in children and adults. If you’re not interested in that, skip to the bottom of this post for links to this months’ articles)
If you follow clinical research in infectious diseases, you’ve probably heard rumors about OVIVA, or “Oral Versus IntraVenous Antibiotics for the treatment of bone and joint infection,” an open-label multicenter RTC in the UK comparing… well, oral versus IV antibiotics for bone and joint infections. The researchers published their study protocol in 2015, then presented preliminary results at ESCMIID 2017. They enrolled 1054 patients across 26 medical centers, and report that definitive treatment failure at one year occurred with similar frequency among the patients receiving oral versus IV antibiotic (13.2% vs 14.6%), meeting their prespecified 7.5% risk difference threshold for noninferiority. These data have yet to appear in the published literature, which leads me to imagine the study is trapped in revision hell over some statistical quibble.
When OVIVA’s results were first announced, I read several takes (and thought myself) that finally, here was definitive proof that oral antibiotics could be as effective as their IV counterparts for osteomyelitis. But this begs the question: where did we get the notion that bone and joint infections needed IV antibiotics in the first place? Nobody thinks cellulitis, cystitis, or intraabdominal infections must only be treated with IV antibiotics. Is it simply because it’s a “bad” infection? Are there any data to suggest that oral antibiotics are inferior to intravenous antibiotics? Is this just another thing we do for no reason? This month, I decided to search the literature on osteomyelitis, starting from the first published references, to understand how we’ve gotten where we are today.
The earliest paper on the antibiotic treatment of osteomyelitis I could find is this 1945 case series by Compere et al, which reports the outcome of penicillin treatment in twelve children with acute hematogenous osteomyelitis. At the time, penicillin treatment was given intramuscularly every three to four hours around the clock – by 1948, this practice would be supplanted by the intramuscular drip, and then at some point later by IV infusion. Anyway, in the introduction to the case series, the authors state that “the value of penicillin in treatment of infections of bones [in chronic osteomyelitis] has been well established,” but don’t provide a reference to support that assertion. They find that penicillin recipients required surgery less frequently than patients treated with no antibiotics or sulfonamides alone, and ended up with good functional outcomes. Compere et al conclude their series by suggesting various penicillin dosing schedules for osteomyelitis; as for the duration of treatment, they recommend that penicillin be continued for five days after the cessation of fever. Interesting that “days afebrile” never caught on as a means of individualizing duration of therapy.
The next reference I found also informs the appropriate duration of therapy for osteomyelitis. In 1946, Higgins, Brown, and Bodian compare the outcomes of penicillin treatment in acute (n=31) versus chronic (n=5) osteomyelitis in children; in both groups they found uniformly excellent outcomes with 1-4 weeks of therapy (mostly 10-14 day courses in both groups). I was unable to find any reasoning in later publications as to why 4-6 weeks of therapy for osteomyelitis subsequently became the norm (i.e. I didn’t find any papers reporting poor outcomes with these 2-week courses), but the 4-6 week duration had been universally adopted by the 1970s. I checked Mandell’s 8th edition chapter on osteomyelitis, which states that experimentally induced rabbit osteomyelitis studies by Norden et al (the references to which are not given in the text – come on, Mandell!) showed that 14 days of clindamycin was less effective than 28 days at sterilizing rabbit bone, so maybe these durations come from animal data?
One of the few references to support tetracyclines for osteomyelitis is actually from the pediatric literature – presumably before the risks of that drug class in children had been recognized. In 1962, Cullen and Hargadon report the outcomes of osteomyelitis in 55 children treated with penicillin (n=35) or tetracycline (n=20), each for 3-4 weeks. Complications (defined as formation of a soft tissue or bone abscess, sequestra – AKA dead bone - needing resection, or pathologic fracture) occurred in 54% of the penicillin recipients but only 15% of the tetracycline recipients. Children receiving tetracyclines were more likely to be discharged within 30 days (60% vs 26%), which isn’t surprising given that OPAT had yet to be invented. A caveat of this study is that, of the 33 patients with a microbiologic diagnosis, more than half had penicillin-resistant S. aureus; however, among the patients with penicillin-sensitive isolates treated with penicillin, the rate of complication was still about 60%.
By the time this 1965 review of antistaphylococcal drugs was published, penicillin-resistant S. aureus had become highly prevalent, the antistaphylococcal penicillins were the standard of care, cephalosporins had just been invented, and there were only a few reports of S. aureus with “intrinsic” resistance to penicillins not mediated by a beta-lactamase (i.e.; MRSA). I point out this review for two reasons. First, it’s the one of the earliest papers I’ve come across that expresses the old dictum that bacteriocidal antibiotics are more effective than bacteriostatic ones for severe infections (this isn’t biologically plausible given the standard laboratory definitions of ‘bacteriocidal’ and ‘bacteriostatic’; plus, there’s no evidence this is true and a reasonable amount it isn’t).
The second is because this paper mentions lincomycin, a predecessor to clindamycin approved the year prior, noting that “it is of interest that relatively high concentrations of lincomycin occur in the bone, and the effects of this have been borne out clinically in the remarkable responses exhibited by patients with acute and chronic osteomyelitis.” This is the earliest mention I found of antibiotic bone penetration being an important parameter of osteomyelitis treatment outcome. The criticality of bone penetration in osteomyelitis is another truism that seems self-evident but doesn’t have much data to support it. Interestingly, Dr. Robert Fitzgerald of the Mayo clinic (who did many of the original bone penetration studies, quantifying antibiotic concentrations in bone shavings taken from volunteers undergoing elective orthopedic procedures), writes in his 1984 review of the topic that “[a]ntibiotics are able to penetrate normal and osteomyelitic bone in a fashion that permits concentrations equivalent to those found in the serum following intravenous administration.” It’s unclear to me why he says this, as there are several studies reporting bone antibiotic levels a log or more under than those simultaneously achieved in serum. But in this quick tour of the literature I did note the marked absence of studies showing poorer outcomes with “low bone penetration” antibiotics. Perhaps the nature of osteomyelitis is such that antibiotic levels achieved in vascularized trabecular bone and newly formed osteoid matrix matter more than levels in calcified bone?
Aside from the tetracycline paper mentioned above, the first in-depth description of oral antibiotic treatment of osteomyelitis I was able to locate appears in The Lancet in 1968. Bell, an Australian, reports the outcomes of 19 patients with chronic staphylococcal osteomyelitis treated with oral cloxacillin – only five of whom were treated with IV antibiotics beforehand, and even then just for a week. Bell used high doses of cloxacillin (5 grams a day, plus probenecid!) for a minimum of six months, and left half of his patients on 1g cloxacillin daily maintenance therapy after clinical evidence of their infection improved. Still, he noted fair responses, with 14/19 patients free of any evidence of infection by six months, and all but one patient ultimately improving enough to return to work. He also reports “all 19 patients tolerated the high doses of the penicillins remarkably well and no toxic effects were observed in any of the patients.”
Sure, Dr. Bell. Whatever you say.
Several additional series of oral treatment of osteomyelitis, mostly using penicillins and early cephalosporins, appear in the literature over the next two decades – in an evening on Pubmed I found references from Walker in 1973, Teztlaff in 1978, Bryson in 1979, Dunkle in 1982, and Aranoff in 1986. These reports have several things in common: (1) the samples sizes were small (pooled N = 113), (2) the patients were all children with mostly acute osteomyelitis mostly caused by S. aureus, (3) initial antibiotic therapy was intravenous and continued until systemic symptoms improved, which was about a week in each case, (4) total duration of IV and oral antibiotic therapy ranged from 4-6 weeks, and (5) outcomes with were universally excellent, with rare or no relapsed infections or need for additional surgeries after antibiotic treatment.
Reading these papers, three things jumped out at me:
1) I’ve been taught that oral beta-lactams are poor choices for osteomyelitis on account of limited bioavailability and poor bone penetration. And yet, in these studies the (again, mostly PSSA and MSSA) osteomyelitis cure rates were fantastic with oral penicillin V, ampicillin, (di)cloxacillin, and cephalexin given at standard doses and for standard osteomyelitis durations.
2) While only a few children fell into the former group, kids with chronic osteomyelitis did just as well as those with acute osteomyelitis. So again, where did this idea that treatment outcomes data for acute osteomyelitis can’t be applied to chronic osteomyelitis come from? A key caveat is that in some cases, the patients with chronic osteomyelitis received several months of oral therapy to achieve their cures; but on the other hand, it’s not clear that these patients all received aggressive surgical debridement before antibiotic therapy, and we know now that inadequate debridement of chronic osteomyelitis predicts a poor outcome.
3) All of these reports came from the pediatric literature. Initially I thought this must just be because pediatricians were more determined than internists to get their patients out of the hospital and back to their day-to-day lives, but in fact studies of oral quinolone use in adults appear a decade later. So here’s my alternate hypothesis for why doctors might have been hesitant to try oral beta-lactams in adults: adults more often have chronic osteomyelitis, chronic osteomyelitis has a more diverse microbiology than acute osteo, and penicillins and early-gen cephalosporins have poorer activity versus gram-negative and anaerobic organisms – which might have driven (unpublished) unfavorable experiences with beta-lactams for osteomyelitis in adults. For what it’s worth, though, in the one adult study of oral antibiotics for osteomyelitis I could find from this period (Black et al JID 1987), the patients did just as well as those in the pediatric studies.
Everything changes in the late 1980s when fluoroquinolones hit the market. For all the hate they receive now, quinolones were miracle drugs when they first arrived: broad-spectrum (including activity versus Pseudomonas!), orally bioavailable, great bone penetration, and with less resistance than we have for carbapenems today. We see a slew of reports supporting the use of oral quinolone therapy for osteomyelitis published during this period: in my cursory search, I found papers by Greenberg, Greenberg again, Dellamonica, MacGregor, Dan, Gentry, and Gentry again. These are all reports looking at chronic osteomyelitis in adults, and among these infections the microbiology is more diverse, the courses of treatment longer, and the outcomes less reliably successful (50-70%) than described in previous reports in children. This isn’t too surprising given the now-recognized inadequacy of early generation quinolone (i.e.; FQs other than moxifloxacin and delafloxacin) monotherapy for infectious due to S. aureus, which was and is the most common cause of osteomyelitis in every form.
By the end of the 1990s, there is more published data to support the use of oral antibiotics for osteomyelitis than parenteral antibiotics, with the body of literature being most robust for the fluoroquinolones. Interestingly, though, oral beta-lactams still have their expert advocates. In 1997, Dr. Benjamin Lipsky, reviewing the treatment of osteomyelitis in Clinical Infectious Diseases, writes, “Parenteral therapy, preferably for 4-6 weeks, is traditionally recommended. This dictum is mainly based on data from animal experimental models and limited, often anecdotal experience…. Appropriate oral antibiotics include dicloxacillin, cephalexin, and clindamycin…[fluoroquinolones] achieve high tissue concentrations and have yielded success rates in treatment of osteomyelitis similar to those of parenteral agents.”
The 21st century provides additional clinical trial and retrospective cohort data on osteomyelitis, mostly in adults (any doubts about the adequacy of oral antibiotics for acute osteomyelitis in children having been thoroughly put to bed). I’ve summarized the trials I could find in the table below. I chose to exclude any studies that employed IV to oral stepdown therapy in the “oral” group unless it was something relatively nonsensical, like switching the near-perfectly bioavailable ciprofloxacin from IV to oral after a few days.
So after all of those readings, here are some concluding thoughts:
1. I’ve read OVIVA described as “merely the cherry on top of the cake” of extant studies supporting oral antibiotic therapy, but IMO that underappreciates how marginal the existing body of literature really is. Even if you put together all the comparative studies I mentioned above, OVIVA will increase the total sample size of patients in such trials by about 400%. There’s room to expand research in this niche of ID for sure.
2. Despite concerns about bioavailability and bone penetration, the early literature on oral beta-lactams for osteomyelitis reported relatively good outcomes in children. Perhaps it’s time for new studies to assess the relative efficacy of these drugs versus parenteral beta-lactams and/or alternative oral agents in adults – particularly for the beta-lactams with rather good oral bioavailability, like amoxicillin (75-90%), amoxicillin clavulanate (75%? - note that while reports of amox-clav’s bioavailability are variable, it seems to cure osteo as well as linezolid) cephalexin (90-100%), and cefadroxil (75-80%). As for other antimicrobials, the clinical outcomes data on trimethoprim-sulfa, quinolones, rifampin, and clindamycin in adult osteomyelitis is already good, and what data we have for linezolid and tetracyclines, while scant, looks promising.
3. I failed to find any data suggesting the superiority of IV to oral therapy for bone and joint infections. Expert recommendations for all-IV (and later IV-to-oral stepdown) therapy for osteomyelitis appear in early reviews of this topic seemingly from nowhere and are repeated thereafter, suggesting this practice reflects eminence-based rather than evidence-based medicine. Given that there seems to be more published experience with oral than parenteral antibiotics for bone infections, perhaps we adult doctors should take a page from our pediatric colleagues and start asking is there a reason I need to use IV antibiotics? rather than can I use oral antibiotics? for osteomyelitis.
4. I still have no idea why we moved from short, clinical response-guided durations of therapy for osteomyelitis to a 4-6 week standard between the 1940s and 1960s. We do have data suggesting that longer courses of antibiotic therapy truly are indicated for vertebral osteomyelitis when risk factors like undrained abscesses, ESRD, or MRSA are present. On the other hand, clinical trial data in children with acute hematogenous osteomyelitis suggests treatment for three weeks or fewer may be sufficient in that population. Finally, the historic assertion that the pathophysiologic differences between acute and chronic osteomyelitis make the approaches to their treatment noncomparable doesn’t have much evidence in the literature to support it. So, while certainly treatment approaches shown to work for osteomyelitis in children ought to be investigated in adults rather than adopted without question, I don’t think adult ID doctors have much of a leg to stand on in ignoring the pediatric osteomyelitis literature either. I’d love to see a clinical trial risk-stratifying adults with osteomyelitis and randomizing patients at low risk of failure to shorter antibiotic courses guided by clinical response – I bet we could safely spare a lot of patients a lot of antibiotic-days.
With all that out of the way, here are the links to this month’s literature reviews:
Antimicrobial Agents research topics this month included neutropenia associated with cefepime given by IV push, cefiderocol’s successful debut trial versus imipenem-cilastin for UTI, the relative susceptibility to ceftazidime-avibactam and ceftolozane-tazobactam among strains of MDR P. aeruginosa, and the association between rifaximin prophylaxis and carriage of rifampin-resistant staphylococci.
ID Diagnostics research covered a machine learning algorithm to improve the diagnosis of infection in the ED using routine bloodwork, a mathematical modeling of the clinical value of TEE for patients with S. aureus bacteremia, and an analysis of the diagnostic accuracy of current guidelines for which children with gastroenteritis should receive specific enteropathogen testing.
General ID papers this month included a new RTC comparing 7 versus 14 days of antibiotics for gram-negative bacteremia, aspirin plus a macrolide for the prevention of cardiovascular mortality after severe pneumonia, the duration of delay in effective antimicrobial therapy associated with poor clinical outcome in gram-negative bacteremia, the explosion of drug use-associated infective endocarditis in the United States, and the role of oral beta-lactams in the treatment of pyelonephritis.
HIV and STD literature this month covered the high rates of viral suppression in patients with risk factors for nonadherence initiating ART, good adherence to PrEP among MSM who use drugs, DRE for the detection of anal cancer in high-risk MSM, and better identification of transgender and non-binary patients using a more sophisticated assessment of gender on clinic intake forms.
Onc and Transplant ID research included a review of the inflammatory syndromes associated with cancer immunotherapies and the infectious import of retained cardiovascular hardware in patients undergoing heart transplantation.
Antimicrobial stewardship and infection control papers this month covered the utility of a urinary tract infection bundle including reduced micro laboratory reporting of culture data on appropriateness of UTI and ASB management, as well as a paper reviewing an infection control team’s (only partially successful) efforts to curtail a large outbreak of CRE in two wards of a UK hospital.