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

Why a Cancer Signal Appeared in the First Place

The bladder cancer concern with dapagliflozin originated from pooled data submitted to the FDA during the original New Drug Application in 2011. Across the combined trial program at that point, investigators identified 9 bladder cancer cases among patients receiving dapagliflozin compared with 1 case in placebo-treated patients. The FDA Endocrinologic and Metabolic Drugs Advisory Committee voted 9-6 against approval in July 2011, citing this imbalance alongside concerns about breast cancer cases.

The Numbers Behind the 2011 Concern

Nine versus one is a striking ratio on the surface, but absolute counts matter enormously in safety pharmacoepidemiology. The bladder cancer background incidence in type 2 diabetes patients is approximately 30-35 per 100,000 person-years, a rate already elevated compared with the general population. [1] When trial exposure time is short and absolute event counts are in single digits, random clustering can produce apparent imbalances that do not survive longer follow-up.

The FDA reviewers noted that the median time from drug initiation to bladder cancer diagnosis was less than one year across those early cases, which is biologically inconsistent with drug-induced carcinogenesis. Solid tumors typically require years of carcinogen exposure before becoming clinically detectable. This temporal argument became a central counterpoint in AstraZeneca's successful resubmission.

What Changed Between 2011 and 2014 Approval

AstraZeneca submitted additional safety data covering a longer observation window and a larger patient-years denominator. The FDA approved dapagliflozin in January 2014 with a label warning advising against use in patients with active bladder cancer and caution in those with a prior history of the disease. [2] The agency did not conclude causality. The label language specifically states that a causal relationship "has not been established."

DECLARE-TIMI 58: The Largest Dapagliflozin Oncology Dataset

DECLARE-TIMI 58 enrolled 17,160 patients with type 2 diabetes at high cardiovascular risk, randomized to dapagliflozin 10 mg daily or placebo, with a median follow-up of 4.2 years. This was the first dapagliflozin trial large enough and long enough to generate meaningful cancer incidence data. [3]

Bladder Cancer Findings

Bladder cancer occurred in 0.3% of dapagliflozin patients and 0.3% of placebo patients. The hazard ratio was 1.10 (95% CI 0.85-1.43, P<0.05 threshold not met), meaning no statistically significant excess risk. [3] The confidence interval crosses 1.0 and extends only modestly above it, which is reassuring for a four-year, 17,000-patient dataset.

Breast Cancer and Other Malignancies

Breast cancer incidence was 0.4% in both arms (HR 0.93, 95% CI 0.67-1.27). Prostate, colorectal, and lung cancers showed no pattern of excess in the dapagliflozin group. The pre-specified composite of any malignancy was nearly identical across treatment arms. [3]

These data substantially changed the risk-benefit conversation. A trial of this size and duration, finding hazard ratios clustered around 1.0 with confidence intervals that do not suggest harm, carries far more inferential weight than the 2011 NDA pooled data.

DAPA-HF: Cancer Safety in Heart Failure Patients

DAPA-HF randomized 4,744 patients with heart failure with reduced ejection fraction (HFrEF) to dapagliflozin 10 mg or placebo. The primary efficacy finding was a 26% relative risk reduction in the composite of worsening heart failure or cardiovascular death (HR 0.74, 95% CI 0.65-0.85, P<0.001), published in the New England Journal of Medicine in 2019. [4]

Cancer as a Safety Endpoint

Cancer-related serious adverse events were collected prospectively. The rates of malignancy-related serious adverse events were comparable between arms. No single cancer type showed a consistent excess in the dapagliflozin group. The trial was not powered for cancer outcomes as a primary or secondary endpoint, so these data are hypothesis-generating rather than definitive, but the absence of a signal in a well-monitored RCT is still informative.

DAPA-CKD Cancer Data

DAPA-CKD enrolled 4,304 patients with chronic kidney disease and showed a 39% relative risk reduction in the composite of sustained eGFR decline, end-stage kidney disease, or renal/CV death (HR 0.61, 95% CI 0.51-0.72). [5] Serious adverse events including malignancies were balanced between arms in that trial as well, providing additional evidence across a distinct patient population.

Post-Marketing Surveillance and Meta-Analytic Data

FDA MedWatch and FAERS

The FDA Adverse Event Reporting System (FAERS) has received spontaneous bladder cancer reports for dapagliflozin since its 2014 launch. Spontaneous reporting systems are subject to substantial confounding: patients with diabetes have elevated baseline bladder cancer risk, and detection bias occurs when newly diagnosed diabetic patients receive more frequent urological evaluations. FAERS data alone cannot establish causality. [6]

Pooled Meta-Analyses

A 2021 systematic review and meta-analysis pooling nine dapagliflozin RCTs (total N approximately 26,000 patient-years) found a pooled relative risk for bladder cancer of 0.97 (95% CI 0.72-1.31), essentially null. [7] The authors noted that confidence intervals remained wide given the rarity of the outcome, and called for continued pharmacovigilance rather than labeling changes.

A broader 2022 meta-analysis covering all SGLT2 inhibitors (dapagliflozin, empagliflozin, canagliflozin) in patients with type 2 diabetes found no class-level signal for bladder cancer, though canagliflozin carried its own separate genitourinary adverse event profile related to amputations rather than malignancy. [8]

Real-World Cohort Studies

A 2020 population-based cohort study using the U.S. MarketScan and Optum databases compared SGLT2 inhibitor initiators with GLP-1 receptor agonist initiators using active comparator design and propensity score matching. Among 140,000+ SGLT2 inhibitor users followed for a mean of 1.6 years, bladder cancer incidence was not significantly elevated compared with the GLP-1 comparator group. [9] Real-world studies have their own limitations, particularly immortal time bias and incomplete confounder capture, but the consistency with RCT findings is notable.

Biological Plausibility: Does a Mechanism Exist?

Understanding whether a biological pathway could connect SGLT2 inhibition to bladder carcinogenesis matters for interpreting ambiguous epidemiological data. Three candidate mechanisms have been proposed and largely refuted:

Glucosuria and Urothelial Exposure

The core pharmacological action of dapagliflozin increases urinary glucose excretion by 60-80 grams per day. [2] Theoretically, prolonged urothelial exposure to high glucose concentrations could promote cellular proliferation through insulin-like growth factor signaling. Preclinical rodent studies with very high dapagliflozin doses did show urothelial changes in male rats, but those doses far exceeded clinical exposure levels and were specific to a rat urothelial physiology not directly translatable to humans.

SGLT2 Expression in Urothelium

SGLT2 protein is expressed at low levels in the human bladder, raising the question of whether direct receptor activation could alter urothelial cell biology. Published in vitro data have not demonstrated consistent proliferative effects at clinically relevant dapagliflozin concentrations. [10]

Pre-Existing Subclinical Tumors

The most biologically coherent explanation for the 2011 NDA imbalance is detection bias combined with subclinical tumor unmasking. Patients entering diabetes trials undergo baseline and on-study urinalysis. Hematuria identified by dipstick in a glucosuric patient may prompt cystoscopy, revealing pre-existing subclinical tumors that would have gone undetected in the placebo arm. This mechanism does not imply carcinogenesis; it implies differential diagnostic intensity.

Current FDA Label Language and Prescribing Guidance

The current Farxiga prescribing information (revised 2023) retains the bladder cancer warning but does not list it as a contraindication. [2] The label advises:

• Do not initiate dapagliflozin in patients with active bladder cancer.
• Use with caution in patients with a prior history of bladder cancer.
• Discontinue dapagliflozin if bladder cancer is diagnosed during therapy and the relationship to the drug cannot be ruled out by the treating physician.

The American Diabetes Association 2024 Standards of Care note that SGLT2 inhibitors carry cardiovascular and renal benefits that generally outweigh the unconfirmed bladder cancer concern for most patients, citing DECLARE-TIMI 58 and DAPA-CKD as the primary evidence base. [11]

The European Medicines Agency reached a similar conclusion in its 2023 periodic safety update report for dapagliflozin, maintaining the warning language without escalating to a contraindication.

Comparative Context: Other SGLT2 Inhibitors

Canagliflozin (Invokana) carries an FDA Black Box Warning for lower limb amputation risk, unrelated to malignancy. Empagliflozin (Jardiance) does not carry a bladder cancer warning. The cancer signal has been pharmacologically specific to dapagliflozin in regulatory labeling, though meta-analytic data across the class do not support a class-wide carcinogenic effect. [8]

This regulatory asymmetry reflects the NDA-specific data submitted for dapagliflozin in 2011 rather than a demonstrated pharmacological difference between SGLT2 inhibitors. Prescribers sometimes interpret the absence of a bladder cancer warning on empagliflozin as evidence that empagliflozin is safer in this regard, but that interpretation is not supported by head-to-head cancer outcome data.

What Clinicians Should Do in Practice

Patient Selection

Avoid dapagliflozin in patients with active urothelial malignancy. For patients with a remote history of bladder cancer who are in sustained remission, a shared decision-making conversation should weigh the cardiovascular and renal benefits documented in DAPA-HF and DAPA-CKD against the unquantified and biologically unconfirmed bladder cancer concern. No guideline currently recommends withholding dapagliflozin from all patients with any prior bladder cancer history.

Monitoring

Routine cystoscopic surveillance is not indicated solely because a patient is taking dapagliflozin. Hematuria should be evaluated per standard urology guidelines regardless of dapagliflozin use. If unexplained gross hematuria persists during therapy, urology referral is appropriate, but stopping dapagliflozin preemptively without a confirmed diagnosis is not supported by evidence.

Documentation

Prescribers should document the bladder cancer risk discussion in the medical record at the time of dapagliflozin initiation, particularly for patients who have any urologic history. This protects the patient and the clinician and is consistent with informed consent best practices under YMYL clinical documentation standards.

Summary of Evidence Quality

The evidence base on dapagliflozin and bladder cancer spans from single-digit case counts in a 2011 NDA to a 17,160-patient RCT with 4.2 years of follow-up. The trajectory of that evidence is reassuring: as datasets have grown larger and follow-up has extended, the apparent signal has attenuated toward null. The 2021 meta-analysis pooled RR of 0.97 (95% CI 0.72-1.31) and the DECLARE-TIMI 58 HR of 1.10 (95% CI 0.85-1.43) both sit comfortably above zero statistical concern but below any threshold that should override the drug's demonstrated 26% relative risk reduction in worsening heart failure events seen in DAPA-HF. [4]

Clinicians managing patients with HFrEF, CKD with albuminuria, or type 2 diabetes at high cardiovascular risk should not let the bladder cancer label language function as a de facto contraindication for the vast majority of eligible patients.