Farxiga Cancer Risk Signal Review: What the Clinical Evidence Actually Shows

Dapagliflozin, sold as Farxiga, is an SGLT2 (sodium-glucose cotransporter-2) inhibitor. It is FDA-approved for type 2 diabetes, heart failure with reduced ejection fraction, and chronic kidney disease. It is not chemically related to GLP-1 receptor agonists despite sometimes being discussed alongside them in diabetes care content. This article addresses one narrow question: does dapagliflozin cause bladder cancer, and how should the 2011 regulatory signal be weighed against later evidence.

The direct answer: an early, small numerical imbalance in bladder cancer cases (9 on dapagliflozin versus 1 on placebo) delayed FDA approval in 2011. Larger and longer trials completed since then, most notably DECLARE-TIMI 58, have not reproduced a statistically significant excess of bladder cancer. The current FDA label still carries a bladder cancer caution and states that a causal relationship has not been established. That combination, an unresolved label warning sitting on top of reassuring larger-trial data, is exactly what a reader needs to understand before treating the warning as a reason to avoid or stop the drug.

Why the signal exists at all

The bladder cancer question originated from pooled safety data submitted with dapagliflozin's original New Drug Application. Investigators reported more bladder cancer cases on dapagliflozin than on placebo across the early trial program, and the FDA's Endocrinologic and Metabolic Drugs Advisory Committee voted against approval in 2011, citing that imbalance along with breast cancer case counts. This is documented regulatory history, not a disputed claim, but the underlying case counts were small, which matters for how much weight the imbalance deserves.

Two features of that early data argued against a direct causal reading, and both were raised during the FDA's own review process:

• Absolute numbers were tiny. A shift from one case to nine cases in a large multi-arm trial program can reflect chance clustering rather than a true effect, especially over short follow-up.
• Timing was biologically inconsistent with drug-caused cancer. Solid urothelial tumors generally take years to become detectable. Cases surfacing within months of starting the drug are more consistent with a pre-existing, previously undiagnosed tumor than with a newly drug-induced one.

The FDA approved dapagliflozin in January 2014 after the manufacturer submitted a longer follow-up dataset. The agency did not conclude that dapagliflozin causes bladder cancer. It kept a caution in the label instead of clearing the question entirely, which is a middle position readers should not mistake for either exoneration or confirmation.

What the larger trials show

DECLARE-TIMI 58 is the largest and longest cardiovascular outcomes trial run on dapagliflozin, enrolling more than 17,000 people with type 2 diabetes and cardiovascular risk factors and following them for a multi-year period. Its published results, widely reported in cardiology and endocrinology literature, showed bladder cancer occurring at similar low rates in the dapagliflozin and placebo groups, with a hazard ratio close to 1.0 and a confidence interval that did not indicate a significant excess risk. Breast cancer and other malignancy rates were likewise similar between arms. Because this trial is an order of magnitude larger than the original 2011 NDA dataset and followed patients for years rather than months, it carries substantially more weight for judging long-term cancer risk. Readers or clinicians who want the exact hazard ratio and confidence interval should pull the original NEJM publication rather than rely on any secondary summary, including this one, since precise figures should be confirmed against the primary paper.

DAPA-HF (heart failure with reduced ejection fraction) and DAPA-CKD (chronic kidney disease) both collected cancer-related serious adverse events as routine trial safety monitoring, not as a primary or powered endpoint. In both trials, malignancy-related serious adverse event rates were reported as comparable between dapagliflozin and placebo arms. Because neither trial was designed or powered to detect a cancer signal, this is supportive, hypothesis-consistent safety data rather than a definitive answer. It adds to the overall picture without settling it on its own.

What post-marketing and observational data add, and their limits

The FDA's Adverse Event Reporting System (FAERS) has collected spontaneous bladder cancer reports for dapagliflozin since its 2014 launch. FAERS is useful for signal detection but cannot establish causation on its own: people with type 2 diabetes already have an elevated baseline bladder cancer risk, and newly diagnosed diabetic patients tend to receive more urological testing, which can surface pre-existing tumors that would otherwise go undetected (detection bias). A rise in reported cases in a spontaneous system does not by itself mean the drug is causing more cancer.

Systematic reviews pooling multiple dapagliflozin randomized trials have reported bladder cancer risk estimates clustering near 1.0 (no difference between drug and placebo), consistent with DECLARE-TIMI 58. Exact pooled point estimates and confidence intervals from any specific meta-analysis should be verified against the original publication before being quoted in a clinical conversation, since secondary summaries (including prior versions of this page) can misstate precise figures.

Is there a plausible biological mechanism, in either direction

Three explanations have been proposed for why dapagliflozin could theoretically affect bladder tissue, and none has strong support at clinically relevant exposure:

• Glucosuria and urothelial exposure. Dapagliflozin increases urinary glucose excretion, and higher sugar concentration in urine has been hypothesized to promote urothelial cell proliferation. Rodent studies at very high, supratherapeutic doses have shown urothelial changes, but those findings have not translated to human data at clinical doses, and rat urothelial biology does not map cleanly onto human tissue.
• Direct SGLT2 activity in bladder tissue. SGLT2 protein is expressed at low levels in human urothelium, but in vitro work has not shown a consistent proliferative effect at the drug concentrations achieved clinically.
• Detection bias from increased testing. Patients starting an SGLT2 inhibitor often get more urinalysis and follow-up testing because glucosuria and hematuria overlap clinically. This can unmask pre-existing, subclinical tumors sooner in the drug arm than in the placebo arm without the drug having caused anything. Many reviewers consider this the most likely explanation for the original 2011 imbalance.

Separately, a distinct and more recent research thread is examining whether SGLT2 inhibitors might have anticancer or cardioprotective properties in oncology patients rather than a cancer-causing effect. A 2025 narrative review discusses proposed anticancer mechanisms of SGLT2 inhibitors across preclinical and early clinical work.[1] A separate 2025 review examines the clinical and cardio-oncology implications of using SGLT2 inhibitors in cancer patients, largely for cardioprotective and metabolic reasons during cancer treatment.[2] These lines of research answer a different question than "does dapagliflozin cause bladder cancer," and they should not be read as resolving the bladder cancer question in either direction. They are included here because a reader evaluating cancer-related literature on this drug should know that the more active current research area is the opposite hypothesis, not confirmation of carcinogenicity, and this itself is a reason for caution about over-interpreting either direction from an evolving evidence base.

What is established, what is plausible, and what is not established

Established: The FDA reviewed a numerical bladder cancer imbalance in 2011, delayed approval on that basis, and later approved the drug in 2014 with a retained caution rather than a contraindication. The current label does not describe a proven causal relationship.

Plausible but not proven: A larger, longer trial base (led by DECLARE-TIMI 58) trends toward no meaningful excess bladder cancer risk. This is a reasonable, evidence-supported impression, not a formal declaration that the risk is zero, because bladder cancer is uncommon enough that even large trials have limited statistical power to rule out a small true effect.

Not established: That dapagliflozin causes bladder cancer through any of the proposed biological mechanisms. That the drug is safe for a patient with an active or recent bladder malignancy, which the label continues to caution against. That any specific numeric hazard ratio or confidence interval from a given trial or meta-analysis is precisely correct without checking the primary source, since secondary reporting (including prior versions of pages like this one) can misquote figures.

Current FDA and guideline posture (verify against the live label before prescribing)

As of the last FDA label revision on file, the prescribing information for Farxiga advises against initiating the drug in patients with active bladder cancer, recommends caution in patients with a prior history of bladder cancer, and states that a causal relationship between dapagliflozin and bladder cancer has not been established. Label language and revision dates change over time, so anyone making a prescribing decision should confirm the current label text directly through the FDA's Drugs@FDA database rather than relying on a cited date in this article.

Diabetes care guidelines, including the American Diabetes Association's annually updated Standards of Care, generally frame the cardiovascular and renal benefits of SGLT2 inhibitors as outweighing the unconfirmed bladder cancer concern for most eligible patients with type 2 diabetes, heart failure, or chronic kidney disease.[3] This is a guideline judgment made in the context of a benefit that is well documented (reduced heart failure hospitalization and cardiovascular death, reduced kidney disease progression) weighed against a risk that remains statistically unconfirmed.

A decision framework for the bladder cancer question

This framework does not replace an individualized clinical decision. It organizes the few facts that actually change what a patient and prescriber should do, since the label caution alone does not tell a reader what action follows from their specific situation.

Practical points for the office visit

Documenting a brief discussion of the bladder cancer label caution at the time dapagliflozin is started, especially for anyone with a urologic history, is consistent with standard informed consent practice for any drug carrying a label warning. Routine cystoscopic screening is not supported by current evidence solely because a patient takes dapagliflozin; hematuria should be worked up through standard urology pathways whether or not the patient is on the drug. Anyone with new visible blood in the urine, unexplained flank pain, or unintentional weight loss while on dapagliflozin should be evaluated by their prescriber promptly rather than making an independent decision to stop the medication.

What this page cannot tell you

This article cannot tell an individual patient whether their personal history makes dapagliflozin appropriate. It does not replace a conversation with the prescribing clinician, who can weigh personal cancer history, current disease indication, and alternative options. Anyone experiencing gross hematuria, unexplained pelvic pain, or other concerning urinary symptoms should seek prompt medical evaluation rather than self-diagnosing based on this or any general reference.

References

1. Narrative review of proposed anticancer mechanisms of SGLT2 inhibitors in oncology (2025). https://pubmed.ncbi.nlm.nih.gov/40551258/
2. Review of SGLT2 inhibitor use in cancer patients, cardio-oncology implications (2025). https://pubmed.ncbi.nlm.nih.gov/40429921/
3. American Diabetes Association Standards of Care, current annual issue. https://diabetesjournals.org/care/issue/47/Supplement_1
4. FDA Adverse Event Reporting System (FAERS) Public Dashboard, for post-marketing signal context. https://www.fda.gov/drugs/questions-and-answers-fdas-adverse-event-reporting-system-faers/fda-adverse-event-reporting-system-faers-public-dashboard

Note for editorial review: precise hazard ratios and confidence intervals for DECLARE-TIMI 58, DAPA-HF, DAPA-CKD, and any pooled meta-analysis cited in earlier drafts of this page should be re-verified against the original journal publications before republication, since the source PMIDs previously attached to these figures could not be confirmed as accurate for this draft. The FDA label revision date and specific document link should also be confirmed against the live Drugs@FDA record before publication.