Case
A 50-year-old female was brought in by ambulance after her 17-year-old son found her acutely confused and incoherent. Initial Glasgow Coma Scale score was 11 (E3V3M5). She was hypothermic (29.9°C), hypotensive (96/49), and hyperglycaemic (blood glucose level HI, ketones 4.8 mmol/L by glucometer). Past medical history comprised only type 2 diabetes mellitus (T2DM) treated with metformin 2 g and sitagliptin 100 mg. Of note, she was not prescribed an SGLT2 inhibitor. Her son perceived her to have good medication compliance and good general health; she was a non-smoker with rare alcohol consumption and worked full-time in an administrative role. There was a maternal family history of T2DM, but no autoimmune disease. Her son described a two-day history of nausea and vomiting preceding her acute presentation.
Initial biochemistry (table 1) demonstrated a marked metabolic acidosis (pH 6.59, bicarbonate 2 mmol/L), ketosis and a glucose level of 44.0 mmol/L, with renal impairment and raised inflammatory markers. Examination demonstrated hypovolaemia and cool peripheries, with an otherwise benign abdominal, respiratory and cardiovascular examination. A small skin wound was noted on the posterior thorax without overlying cellulitis or clinical evidence of abscess.
She was treated for severe diabetic ketoacidosis (DKA) in ICU with fluid rehydration, insulin infusion and potassium replacement. Management for presumed cold sepsis included empirical broad-spectrum intravenous antibiotics after a septic screen was sent including a COVID swab. She required vasopressor support as her blood pressure deteriorated further to 73/54 with a heart rate of 66 bpm. Oliguria developed in the context of acute kidney injury necessitating continuous renal replacement therapy. An agitated delirium developed requiring physical restraints and haloperidol sedation.
After commencing urgent resuscitation, attention was directed to the precipitant of severe DKA, particularly a source of sepsis. The posterior thorax skin lesion, initially deemed not amenable to incision and drainage, was reviewed by the surgeons 15 hours after presentation. The assessment was difficult due to patient agitation and wound location. Nevertheless, debridement was suggested once she improved metabolically. Venous pH normalised to 7.36 within 18 hours of admission, with a glucose of 16.1 mmol/L. Upon further surgical review, it was thought unlikely the skin lesion could have caused such a severe illness, thus a CT abdomen and pelvis was organised to exclude an intra-abdominal source of sepsis. This occurred 20 hours after admission and was unremarkable, as was chest imaging and respiratory viral swabs, including for COVID.
The skin wound was therefore explored under anaesthesia 26 hours after admission. Upon incision, ‘dirty dishwasher fluid’ and copious amounts of pus were expressed with clear evidence of necrotising fasciitis. A 15 x 8 cm incision was required for adequate access, with dissection to the trapezius muscle, and the wound was left open. Antibiotics were escalated from piperacillin-tazobactam to meropenem, clindamycin and vancomycin. Hyperbaric oxygen therapy was administered on day 0, 1 and 2 post-operatively. Cultures from the wound grew methicillin-sensitive staphylococcus aureus, and antibiotic therapy was rationalised.
She had a prolonged stay lasting 23 days for further washouts and debridements, VAC dressing management, antibiotic therapy, transition from nasogastric feeds to oral intake, and transition from intravenous to subcutaneous insulin (see Figure 1). Her back wound required a split skin graft from her right anterolateral thigh and subsequent wound care. She required intermittent haemodialysis on discharge for oliguric renal failure, likely due to acute tubular necrosis. HbA1c was 13.6% and fasting C peptide was 592 pmol/L (260-1730). Anti-GAD and IA2 antibodies were negative. After initial stabilisation, ongoing glycaemia was managed with once-daily mixed insulin, as per the patient's request.
Discussion
DKA is typically associated with type 1 diabetes but may also occur in patients with T2DM. Indeed, a recent study showed 125 cases of DKA in patients with T2DM who were not taking SGLT2 inhibitors in Victorian public hospitals over a 26-month period (1).
Identifying the precipitating event of DKA enables specific treatment and prevention of recurrence. In many cases the cause is clear from the onset, with the most common precipitants being infection and inadequacy of insulin therapy (under-dosing or omission) (2, 3). Where the cause is unclear, however, the importance of a thorough physical examination is amplified, especially when the patient has a diminished GCS which limits the availability of history. The examination may also be revealing when considering other precipitants such as pancreatitis, myocardial infarction, or stroke.
In this case, however, we see the value of combining the examination with an index of suspicion for necrotising fasciitis (NF) as a cause of what may appear to be occult sepsis. This rapidly progressive soft tissue infection involving fascia and subcutaneous tissue initially spares the cutaneous layer leading to a paucity of physical findings even when there is systemic and life-threatening toxicity (4). The most common symptoms when present are swelling, pain and erythema, while signs in advanced stages include bullae, skin discolouration and crepitus (5), but these only occur only in 14-33% of patients (6). It is well established that glucose toxicity in patients with diabetes mellitus is associated with increased susceptibility to infection (7, 8). In a systematic review of 1463 individuals with NF, diabetes was a comorbidity in 45% (range 15.2-71%) (5). Diabetes mellitus has been associated with atypical milder presentations of NF featuring fewer complaints of pain (9), further contributing to misdiagnosis. CT imaging may show gas in soft tissues, but the diagnosis is made by direct visualisation of the fascia intraoperatively. Delays in surgery can result in increased mortality (4), and limb NF in the presence of diabetes has been associated with higher amputation rates (10).
Our patient was noted to have an apparently benign skin lesion that was identified on detailed examination but thought unlikely to be the source of sepsis driving life-threatening DKA, so surgical debridement was pursued only once all other possible causes were excluded. The operative findings were especially alarming in this context, combined with the impression of apparently reasonably controlled T2DM although this proved not to be the case. In this individual who was not insulin-requiring but developed severe life-threatening DKA with no other obvious precipitant, an index of suspicion for NF may have expedited surgical exploration and debridement, which may have improved her outcome.
Our patient required multiple operative procedures for debridement, VAC management and grafting, and multiple sessions of hyperbaric oxygen therapy. For her severe DKA and associated multiorgan sequelae, she required intubation, a prolonged ICU admission, renal replacement therapy with transition to intermittent haemodialysis, and ongoing close outpatient monitoring.
From this case we can be prompted to:
Table 1: Initial pathology results
Figure 1: Timeline of events