Rhino-Orbital-Cerebral Mucormycosis, Arriving in DKA With a Clock Already Running
A single patient, in diabetic ketoacidosis with a fungal infection that won't wait for her glucose to normalize. The disagreement is how many hours of correction is enough before the clock on the infection outruns the clock on the anesthesia.
L.S., a 44-year-old woman, works nights as a respiratory therapist herself, which is part of why she recognized something was seriously wrong before most patients would have: a malfunctioning insulin pump overnight had let her blood glucose climb unchecked for hours, and by the time she arrived in her own hospital's emergency department she was in frank diabetic ketoacidosis, with a new facial pain and swelling over her left cheek that had started that same morning. On exam, a black eschar had already formed on the hard palate — a finding specific enough in a patient this acidotic that biopsy was sent before imaging even returned, and it confirmed broad, ribbon-like aseptate hyphae consistent with mucormycosis.
What makes today's decision genuinely hard is that three real clocks are running at once, and none of them can simply wait for the others. Mucormycosis is angioinvasive and spreads by the hour, not the day — case-series data on delayed debridement describes exactly the kind of tissue and vision loss her team is trying to prevent by acting fast. But general anesthesia in active, uncorrected DKA carries its own independently described mortality risk, one her endocrinology team sees cross their threshold regularly in exactly this population. And the antifungal dosing question sitting underneath both of those turns on a distinction the evidence itself makes badly: the ECMM global guideline names 5mg/kg as the standard dose but 10mg/kg specifically where the central nervous system is involved, and her imaging shows sinus and orbital disease with intracranial extension not yet ruled out — she sits on the boundary of that carve-out rather than inside or outside it. The support underneath the higher number is thinner than its guideline placement suggests: Lanternier's AmBizygo study, the only prospective trial of 10mg/kg in mucormycosis, was a single-arm pilot of forty patients with no 5mg/kg comparator at all, and its 45% week-12 response sat close to what historical 5mg/kg cohorts had already reported — while 40% of those patients doubled their creatinine. Cornely's AmBiLoad trial is the randomized 3-versus-10mg/kg comparison, and it found no benefit at the higher dose with more toxicity, but 97% of its patients had aspergillosis, not mucormycosis. Her creatinine is 0.9 and her baseline renal function is normal, which is the one thing that would make the higher dose survivable if the room decided her orbit was close enough to her brain to count.
In the resuscitation bay, three clocks running at once
This disease spreads angioinvasively by the hour. Every case series I know of on delayed debridement describes tissue, vision, or intracranial containment lost to exactly the kind of wait we're discussing right now. I want her in the OR as soon as she can physically survive anesthesia, not once her glucose is textbook-normal.
I'm not asking for textbook-normal — general anesthesia in active, uncorrected DKA carries its own real, independently described mortality risk, and I see that crossover go wrong when it's rushed. Give me two to three hours of aggressive insulin and fluid correction to get her pH and anion gap to a genuinely safer threshold, not full resolution. That's a real difference from "wait for perfect control."
Treating any delay as equally costly conflates a few hours of targeted correction with the days-long delays the case-series data you're citing actually describes.
Given the sinus and orbital involvement already on imaging, with intracranial extension not yet ruled out, I'd start liposomal amphotericin B at 10mg/kg/day rather than the standard dose. This isn't case-series impression — the ECMM global guideline names 10mg/kg explicitly for CNS involvement, and the AmBisome label says the same thing for brain involvement. The orbit is not the brain, but it is the anatomic corridor to it, and I'd rather be inside that recommendation than argue afterward about whether she qualified for it.
The guideline says what you say it says — I'd just push back on how much is actually underneath it. Lanternier's AmBizygo study is the only prospective trial of 10mg/kg in mucormycosis, and it was a single-arm pilot of forty patients with no 5mg/kg arm to compare against; its week-12 response rate landed close to what historical 5mg/kg cohorts reported, and 40% of the patients doubled their creatinine. The one genuinely randomized 3-versus-10mg/kg comparison, Cornely's AmBiLoad trial, found no benefit and more toxicity — but 97% of those patients had aspergillosis, so it doesn't cleanly transfer either. Tashiro's nationwide series in mucormycosis specifically found no survival difference above 5mg/kg.
Which leaves the guideline's CNS number resting on pharmacokinetic reasoning rather than a comparative outcome — and she doesn't yet have confirmed CNS disease, only an orbit that might become one.
I'd start 5mg/kg/day today, plan isavuconazole step-down once she's stable and tolerating oral intake, and treat the MRI and the debridement findings as the trigger to escalate to 10mg/kg — which her creatinine of 0.9 gives us real room to do. That's a different proposal from refusing the higher dose; it's declining to pay its renal cost before we know she's in the population it was written for.
Agreed: two to three hours of targeted DKA correction to a defined, safer anesthesia threshold rather than full resolution, debridement proceeding as soon as that threshold is met, and standard-dose liposomal amphotericin B started immediately regardless of the surgical timeline.
Not agreed, on two separate axes. The head and neck surgeon remains on record as uncomfortable with any delay at all, however short, and documented that preference explicitly rather than being talked out of it — the compromise held because the endocrinologist's threshold was concrete and time-bound, not because the surgeon's underlying urgency was resolved. Separately, the infectious disease physician did not accept the pharmacologist's reading of the dosing evidence: he holds that a guideline recommendation naming a dose for CNS involvement should be followed when the orbit is involved and the brain is not yet cleared, rather than after. Both agreed the MRI result would settle it either way within hours, which is the only reason the disagreement could be left open rather than decided.