This was a prospective observational study on patients enrolled in the Thiamine as Adjunctive Therapy in Diabetic Ketoacidosis Trial (DKAT) (NCT03717896) [15]. Patients were admitted with DKA after presenting to the Emergency Department (ED) of Beth Israel Deaconess Medical Center (BIDMC), an urban tertiary care center in Boston, Massachusetts, USA, between December 2018 and March 2023. Healthy controls were enrolled at the same time as the DKAT patients. Further details regarding the clinical trial can be found elsewhere [15]. This observational study utilized blood samples included at time of enrollment in the parent trial, before any study drug was administered. The study was approved by the local institutional review board (IRB) (protocol no. 2018P000475), and written informed consent was obtained from each patient prior to enrollment.

Adult patients (age ≥ 18 years) who presented to the ED and were diagnosed with DKA (defined by bicarbonate ≤ 15 mEq/L, anion gap > 12 mEq/L, pH ≤ 7.24, and urine or serum ketones ≥ 3 mmol/L) were included and enrolled within six hours of presentation [15]. Patients who received thiamine supplementation prior to enrollment; had a competing cause of severe acidosis (i.e., seizures, carbon monoxide poisoning, cyanide toxicity, or cardiac arrest); had a known allergy to thiamine; had contraindications to thiamine as determined by the clinical team; were a member of a research-protected population; and those who arrived with a do-not-resuscitate status were excluded. The control group included healthy individuals without any acute illnesses. Controls and DKA patients were enrolled concurrently to allow for simultaneous running of the mitochondrial test and minimization of measurement error. Therefore, many of the controls consisted of emergency department staff and providers and a control individual was allowed to be enrolled more than once at different time points.

Baseline demographic data and laboratory values were collected by trained research assistants and entered into a secure online database [16]. Data items included demographics, medical history, and laboratory values. Blood samples were collected upon enrollment into the study prior to study-drug administration. Fresh whole blood was used to isolate PBMCs to measure the cellular OCR.

PBMCs were isolated from fresh whole blood samples using the following method. Plasma was separated by centrifugation at 800 g for 15 min at 4 °C and saved without disturbing the buffy coat. Plasma was then replaced with the same volume of Roswell Park Memorial Institute Medium (RPMI), and the cell pellets were mixed by gently pipetting up and down several times to disperse the cells. PBMCs were then isolated from the RPMI substituted blood samples using a density gradient separation method according to the manufacture’s instruction (Ficoll-Paque premium; GE Healthcare Bio-Science Corporation, Piscataway, New Jersey, USA). The PBMCs were divided into four groups: (1) no treatment, (2) treatment with thiamine (0.5 μg/mL), (3) treatment with ubiquinol (reduced Coenzyme Q10, (CoQ10) (1 µg/mL), and (4) treatment with a combination of both thiamine and CoQ10. The dosing of thiamine and CoQ10, as well as the details of the experimental design, were based on our previous studies [17, 18]. The mitochondrial profile was measured using an XFe96 Extracellular Flux Analyzer and XF Cell Mito Stress Test Kit (Seahorse Bioscience, North Billerica, Massachusetts, USA). The XF Cell Mito Stress Test uses modulators of respiration that target components of the electron transport chain to measure key parameters of metabolic function. Oligomycin, carbonyl (cyanide-4-trifluoromethoxy) phenylhydrazone (FCCP), and a mix of rotenone and antimycin A were sequentially injected to inhibit specific components of the electron transport chain and provide isolated measurements of OCR for basal, maximal, ATP-linked, and non-mitochondrial respiration. From these measurements, spare respiratory capacity and proton-leaked respiration were calculated. The reported OCRs are in pmol/min/μg of protein. The protein amount was measured using the bicinchoninic acid (BCA) assay after lysing the cells. The experiment protocol followed the recommendations of the instrumentation company. Additional details regarding the instrument have been described elsewhere [17,18,19].

Statistical analyses were performed using R software (R version 4.1.1) [20]. Baseline characteristics of categorical variables are presented as counts and proportions. For continuous variables, medians and interquartile ranges or means and standard deviations were used as appropriate based on the distribution of the data. The outcome OCR variables were summarized for each group and treatment using raw medians and interquartile ranges. To assess median differences, linear quantile mixed models (LQMM, Geraci 2014) [21] were used with a compound symmetry structure to account for within-patient clustering (as some samples originated from the same control enrolled at different times). We assessed (1) the median differences between DKA patients and healthy controls at baseline (without treatment), (2) the relative median differences between no treatment and treatment within each group, and (3) the effects of treatment in the DKA group compared to the same effect in the control group. For the first assessment, the LQMM included a single term for the group as the predictor variable, while for the latter two assessments, the LQMM included an interaction term between the group and treatment. In all models, the effect of age was controlled for by adding the logarithm of age as a covariate, given the potential effect of age on OCR. The relative median differences in OCR values for a given treatment were expressed as a proportion of the value at baseline (no treatment). Statistical significance was a priori set at p < 0.05.