FDA reports for the two feline SGLT2 inhibitors, Senvelgo and Bexacat: ketosis and DKA dominate, deaths cluster early, and Bexacat has a higher serious-report rate.

Bexacat (bexagliflozin tablets) and Senvelgo (velagliflozin oral solution) are the two oral SGLT2 inhibitors the FDA has approved for feline diabetes — Bexacat on December 8, 2022, the first oral diabetes drug and the first SGLT2 inhibitor approved in any animal species, and Senvelgo on August 10, 2023. Both carry the same boxed warning, and both reframed how newly diagnosed diabetic cats are treated: for an appropriate candidate, a once-daily oral medication can replace twice-daily insulin injections. The question that follows every new drug class is how it behaves once it leaves the field trial and enters general practice.

To put real numbers on that, VetMedGuide analyzed the FDA Center for Veterinary Medicine (CVM) public adverse-event database, a June 2026 extract of 1.34 million reports. We identified reports naming either SGLT2 inhibitor and tabulated their reactions, outcomes, and reporting history. Three findings shape the picture, and all three are consistent with the boxed warning the drugs already carry: ketosis and diabetic ketoacidosis (DKA) dominate the reported reactions, deaths cluster in the period when the label says ketoacidosis risk is highest, and Bexacat carries a higher serious-report rate than Senvelgo.

A note on what spontaneous-report data can and cannot say before the numbers: the FDA explicitly cautions that these reports do not establish causation or true event rates. A report "involving" a drug means the drug was being given when the event was recognized — not that the drug caused it. Every cat in this dataset is a diabetic cat, already at baseline risk for ketoacidosis. The value of the analysis is in the pattern of what gets flagged, not in any single count.

How we identified reports (and why the count is conservative)

The FDA's public database masks brand names — every product is recorded as "MSK" — so a drug has to be identified by its active ingredient. Velagliflozin is unique to Senvelgo and bexagliflozin is unique to Bexacat, so we matched reports whose active-ingredient field named one or the other. We worked from the event-level file (one record per unique adverse-event report) rather than the per-reaction rollup, so each report is counted once even when it lists many reactions.

By that method, the June 2026 extract holds 4,088 reports naming velagliflozin (Senvelgo) and 3,521 naming bexagliflozin (Bexacat) — 7,609 combined. The species breakdown is overwhelmingly feline: 3,812 of the Senvelgo reports (93%) and 3,462 of the Bexacat reports (98%) list the affected animal as a cat, which is what you would expect for two cat-only drugs.

The dominant signal is ketosis — exactly what the boxed warning predicts

The single most-reported reaction for both drugs is ketosis, and it is not close. Across the two products, ketosis appears in 2,792 reports — 1,352 for Senvelgo and 1,440 for Bexacat — meaning roughly one in three reports for each drug mentions ketosis. Diabetic ketoacidosis (DKA) is named in a further 412 reports (156 Senvelgo, 256 Bexacat), and ketonuria in 475 Senvelgo reports on its own.

This is the class signature, not a surprise — and it is a species-specific one. SGLT2 inhibitors lower blood glucose by dumping it into the urine, and in a cat that still needs insulin, the liver keeps making ketones even while blood glucose looks controlled — the mechanism behind euglycemic DKA (eDKA), the life-threatening variant the FDA's Dear Veterinarian letters specifically warn can be missed because blood glucose is not high. In people, where the same drug class has been used for type 2 diabetes since 2013, ketoacidosis is rare enough (estimated below 0.15% of patients) that it took years to recognize as a class signal. In cats the ketoacidosis rate runs in the single digits even in well-selected patients and far higher when the wrong cat is started on the drug — which is why a third of the spontaneous reports naming one of these products mention ketosis. The velagliflozin field trial found DKA or eDKA in roughly 5% of newly diagnosed cats and in 18% of cats that had previously been treated with insulin, which is precisely why the label contraindicates the drug in insulin-pretreated cats and why about 86% of ketosis and ketoacidosis episodes occurred within the first two weeks (30 of 35 cases in the trial).

After ketosis, the reported reactions read like the known profile of a diabetic cat that is not doing well: weight loss (1,801 combined), lack of efficacy (1,379 combined), diarrhea (1,302), vomiting (964), lethargy, anorexia, and elevated BUN and liver enzymes. Bexacat shows more hypoglycemia reports (381) than Senvelgo, and more glucosuria; both are consistent with the mechanism. None of this should change a clinician's understanding of the drugs — it confirms, at population scale, what the labels and field trials already said.

Bexacat has a higher serious-report rate — and an earlier reporting curve

The FDA flags each report as serious or not (serious meaning death, life-threatening event, or an outcome requiring intervention). The two drugs differ here in a way worth noting. Senvelgo: 646 of 4,088 reports marked serious (15.8%). Bexacat: 779 of 3,521 reports marked serious (22.1%). Whether that gap reflects a real difference in risk, a difference in the populations selected, or simply reporting patterns is something spontaneous data cannot resolve — but it is the largest single difference between the two in this dataset, and it favors Senvelgo.

The reporting-over-time curves reflect launch dates. Bexacat, approved first, shows 887 reports received in 2023, 1,444 in 2024, and 1,190 in the first part of 2025. Senvelgo, approved in August 2023 and reaching clinics that fall, shows 231 reports in 2023, 1,938 in 2024, and 1,919 in 2025. In other words, the two products converge on similar annual reporting volumes once both have been in the field a full year — which is what you would expect for two drugs competing in the same narrow indication.

Deaths, and the important caveat about them

Outcomes are recorded where known. Across the two drugs, the dataset records 476 reports ending in euthanasia and 129 ending in death — 605 reports total with a fatal outcome (251 euthanized / 67 died for Senvelgo; 225 euthanized / 62 died for Bexacat). Another 1,487 reports are still marked ongoing and a large block are unknown.

These numbers are the ones most easily misread. A fatal outcome in a report involving an SGLT2 inhibitor does not mean the drug killed the cat. These are diabetic cats — a population in which DKA, concurrent pancreatitis, and other complications are common causes of death and euthanasia independent of which therapy was chosen. The velagliflozin field trial, for context, reported a mortality rate around 7%, similar to the ~8% seen in the Caninsulin (insulin) comparison group. What the spontaneous data does corroborate is the label's central safety point: when ketosis or DKA develops on one of these drugs, it can be fatal, and the first days to two weeks of therapy are the window that demands ketone monitoring.

What to take into the exam room

The dataset does not change the label; it explains why the label is written the way it is. The practical points, all reinforced by the numbers above:

Patient selection is the whole game. The contraindication in insulin-pretreated cats exists because that population's ketoacidosis rate is roughly triple. The reports pile up where selection was wrong.
Ketones, not glucose, are the safety metric in the first weeks. Euglycemic DKA is the failure mode that hides from a glucometer. A cat that goes off food, loses weight, or becomes lethargic on either drug needs ketones checked — blood or urine — regardless of what the glucose reads.
Screen before starting, and stop and switch if ketones rise. Both labels require ruling out ketonuria, ketonemia, pancreatitis, and significant renal or hepatic disease before the first dose.
Both products carry the same boxed warning. The serious-rate difference in the spontaneous data is a signal to watch, not a verdict. A cat that is not an appropriate SGLT2 candidate should be on insulin.