Luseogliflozin Trial Puts Ketosis Risk Under a 12-Week Microscope in Insulin-Treated Type 2 Diabetes
A 12-week randomized trial in Malaysia examined whether luseogliflozin could improve insulin resistance in adults with type 2 diabetes already using moderate-dose insulin, while also tracking the ketone-related adverse effects that have complicated the use of sodium-glucose cotransporter 2 inhibitors in practice. The design is straightforward, but the clinical question is not: adding an SGLT2 inhibitor to insulin can reduce glucose burden and potentially improve metabolic control, yet it can also push some patients toward ketonemia or ketonuria when insulin exposure is insufficient for the degree of glucosuria induced.
Forty participants were enrolled in this interventional phase 3 study, with randomization to either open-label luseogliflozin 2.5 mg daily for 12 weeks or no new medication beyond ongoing diabetes treatment. The cohort was limited to adults aged 30 to 65 years with body mass index above 23 kg/m2, HbA1c between 6.5% and 10.5%, and preserved kidney function defined by eGFR above 50 mL/min. Every participant was already receiving more than 0.5 U/kg/day of insulin and had to have remained on a stable dose for the previous 12 weeks, a detail that matters because unstable insulin dosing can confound both glycemic outcomes and ketosis risk.
The primary endpoints were tightly focused on metabolism and safety. Investigators measured ketonemia by blood ketone levels, ketonuria by urine ketone testing, and two surrogate markers of insulin resistance: HOMA-IR and the triglyceride glucose index. Secondary outcomes included change in HbA1c, body weight, and BMI. Those measures capture the physiologic tradeoff at the center of this trial: if the drug reduces glucose toxicity and lowers insulin resistance, downstream insulin requirements may ease; if relative insulin deficiency emerges, ketone production can rise even while glucose control improves.
Why Ketones Matter When an SGLT2 Inhibitor Is Added to Insulin
SGLT2 inhibition lowers plasma glucose by increasing urinary glucose excretion, a mechanism that can reduce the burden on beta-cell function and improve insulin sensitivity signals in some patients. The same glucose-lowering mechanism can also reduce circulating insulin levels and increase glucagon-to-insulin balance, setting up a metabolic environment in which lipolysis and hepatic ketogenesis become more active. In patients already taking insulin, the risk is not theoretical. When insulin doses are not adequately matched to the altered metabolic state, ketone accumulation can emerge despite a modestly improved glucose profile.
The trial’s primary safety focus on ketonemia and ketonuria reflects that concern directly. The choice is clinically sensible because blood ketone measurement offers a more immediate readout of ketone burden than urine testing alone, while urine ketones can lag behind acute physiologic change. In a patient population with established insulin treatment, even a short 12-week exposure can reveal whether the added drug meaningfully shifts the balance toward ketosis or whether stable background insulin is sufficient to blunt that effect.
The eligibility criteria also sharpen the clinical context. Participants with frequent hypoglycemia, recent hospitalization, severe heart failure, advanced renal dysfunction, active hepatic or inflammatory disease, malignancy, pregnancy, or breastfeeding were excluded. Those exclusions reduce noise in the safety readout, but they also narrow the population to a relatively stable outpatient cohort. The findings, when they become available, will most directly apply to adults with moderate obesity, preserved renal function, and moderate-to-high insulin requirements rather than to frailer or medically unstable patients.
The Metabolic Endpoints Were Chosen for Bedside Relevance
HOMA-IR and the triglyceride glucose index are surrogate markers, not clinical outcomes, but they remain useful when a study is trying to determine whether a therapy changes insulin resistance rather than only glucose numbers. HOMA-IR incorporates fasting insulin and fasting plasma glucose, while the triglyceride glucose index combines fasting triglycerides and fasting plasma glucose into a single metabolic signal. In a setting where glucosuria can lower glycemia without directly addressing insulin resistance, those indices help distinguish a true metabolic shift from a simple glucose-lowering effect.
HbA1c, weight, and BMI serve a different purpose. They translate the mechanistic question into practical bedside consequences: fewer grams of circulating glucose exposure, less weight burden, and potentially a lower insulin requirement over time. For clinicians managing type 2 diabetes on insulin, weight gain and intensification of insulin doses often reinforce each other, creating a cycle that is hard to break. A drug that lowers glycemia without further weight gain, or preferably with some weight reduction, would be clinically attractive if the ketone signal remains acceptable.
One detail stands out: the comparator arm received no new medication, which makes the trial easier to interpret but also means the study is testing drug addition against standard care rather than against another active glucose-lowering strategy. The open-label design creates another practical limitation because participants and investigators knew whether luseogliflozin had been started. In metabolic studies, that can influence behavior around diet, home glucose monitoring, and even reporting of symptoms that might trigger ketone testing.
The trial enrolled 40 participants, a modest sample for a phase 3 study. Small trials can still provide useful signals, especially when the endpoints are physiologic and the follow-up is short, but they rarely settle a safety question on their own. A rare adverse event such as clinically significant ketoacidosis may never appear in a cohort this size, which is precisely why the distinction between ketonemia, ketonuria, and overt diabetic ketoacidosis remains important in interpretation.
A Narrow but Clinically Useful Signal Could Still Emerge
The study began on 2022-05-01 and reached actual completion on 2024-11-30, with the record later updated and posted in October 2026. The protocol allowed only a 12-week treatment window, which is long enough to detect early metabolic changes but not long enough to answer questions about durability, adherence, or infrequent adverse events. A short duration also matters for insulin-treated diabetes, where changes in body weight, HbA1c, and insulin sensitivity may unfold gradually and may be influenced by seasonal changes in diet or activity.
The protocol’s focus on stable insulin dosing is another clue to its intent. Investigators were not trying to test luseogliflozin as a rescue therapy for brittle diabetes; they were trying to isolate the drug’s additive effect in a controlled outpatient population. If ketone markers remain low while HOMA-IR and the triglyceride glucose index improve, the study would support a more favorable metabolic balance in this subgroup. If ketones rise without meaningful improvement in insulin resistance, the safety signal would argue for caution in insulin-treated patients who resemble this cohort.
No results have been posted yet, so the trial remains a design-level signal rather than a practice-changing dataset. Even so, it reflects a clinically relevant question facing endocrinologists: whether SGLT2 inhibition can be layered onto established insulin therapy without tipping susceptible patients toward ketosis. The posted protocol is precise enough to answer that question within the trial’s narrow frame, but the broader answer will still depend on whether future data include symptomatic events, insulin dose adjustments, and longer-term follow-up beyond 12 weeks.