Metformin Safety Signals and FDA Actions: What Prescribers and Patients Should Know

How Metformin Works: Mechanism in Brief

Metformin lowers blood glucose primarily by suppressing hepatic glucose production, not by stimulating insulin secretion. Its principal molecular target is mitochondrial complex I in hepatocytes, and the downstream activation of AMP-activated protein kinase (AMPK) drives much of its metabolic effect 1. This distinction matters for safety: because metformin does not cause hypoglycemia as monotherapy, much of the FDA's regulatory attention has focused on rarer metabolic and contamination risks rather than on blood-sugar emergencies.

The drug is absorbed in the small intestine, is not metabolized by the liver, and is cleared entirely by the kidneys with a plasma half-life of roughly 6.2 hours 2. That renal-dependent clearance is the pharmacokinetic fact behind nearly every safety signal discussed below. When kidney function drops, metformin accumulates. When metformin accumulates, lactate metabolism shifts. The entire regulatory history of metformin safety is, at its core, a story about renal thresholds and how the FDA progressively refined them.

Peripheral effects include increased glucose uptake in skeletal muscle and reduced intestinal glucose absorption. GI side effects (nausea, diarrhea, bloating) affect roughly 20 to 30% of new users but are typically dose-dependent and self-limiting with slow titration 3.

The Lactic Acidosis Boxed Warning: History and 2016 Revision

Metformin has carried a boxed warning for lactic acidosis since its 1994 U.S. approval. That warning was inherited largely from phenformin, a related biguanide withdrawn from the U.S. market in 1977 after causing fatal lactic acidosis at a rate of roughly 40 to 64 cases per 100,000 patient-years 4. Metformin's actual incidence is far lower.

A Cochrane systematic review of 347 trials (N = 70,490) found no cases of fatal or nonfatal lactic acidosis in metformin-treated patients and calculated a pooled incidence of 6.3 cases per 100,000 patient-years, a rate statistically indistinguishable from the non-metformin comparator groups 5. The review's lead author, Salpeter et al., concluded: "There is no evidence from prospective comparative trials or from observational cohort studies that metformin is associated with an increased risk of lactic acidosis."

Despite this evidence, the boxed warning remained. What changed in 2016 was not the warning's existence but its renal threshold. Before April 2016, the label contraindicated metformin in men with serum creatinine ≥1.5 mg/dL and women ≥1.4 mg/dL. These creatinine cutoffs were crude: a muscular 30-year-old man and a frail 80-year-old woman could have the same creatinine but vastly different true filtration rates 6.

The FDA's 2016 Drug Safety Communication replaced creatinine cutoffs with eGFR-based criteria 7:

• eGFR ≥45 mL/min/1.73 m²: no renal contraindication
• eGFR 30 to 44: do not initiate, but may continue with monitoring in existing users
• eGFR <30: contraindicated

This single label revision expanded metformin access to an estimated 2 million additional U.S. patients with moderate chronic kidney disease 8. The American Diabetes Association (ADA) Standards of Care now reflect these thresholds, stating: "Metformin can be used with an eGFR ≥30 mL/min/1.73 m², with dose reduction at eGFR <45" 9.

NDMA Contamination and the 2020 Recalls

In December 2019, Singapore's Health Sciences Authority detected N-nitrosodimethylamine (NDMA), a probable human carcinogen, in certain metformin products. The FDA initially tested U.S. supplies and found levels within acceptable intake limits. That assessment changed.

By February 2020, independent laboratory Valisure filed a citizen petition reporting NDMA levels exceeding the FDA's acceptable daily intake of 96 nanograms per day in multiple metformin extended-release (ER) formulations 10. Between May and October 2020, the FDA requested voluntary recalls from at least 10 manufacturers of metformin ER, including products from Apotex, Amneal, and Teva 10.

Key facts about the NDMA recalls:

• Only extended-release formulations were affected. Immediate-release metformin tested within safe limits.
• NDMA is the same impurity found in recalled lots of ranitidine (Zantac) and certain valsartan generics.
• The FDA did not withdraw metformin ER from the market entirely. Manufacturers whose batches tested below the 96 ng/day limit continued distribution.
• By late 2020, the FDA confirmed that reformulated lots from most manufacturers met safety standards and supply normalized 10.

The clinical risk from the NDMA exposures identified was considered low. The FDA estimated that even at the highest detected NDMA levels, continuous daily exposure over 70 years would increase lifetime cancer risk by approximately 1 in 100,000. Patients were advised not to stop metformin without consulting their prescriber, because uncontrolled hyperglycemia posed an immediate and larger risk 11.

Vitamin B12 Deficiency: The Slow-Burn Safety Signal

Unlike lactic acidosis (rare) and NDMA contamination (manufacturing-specific), vitamin B12 depletion is a pharmacologically predictable, dose- and duration-dependent effect. Metformin inhibits calcium-dependent B12-intrinsic factor absorption in the terminal ileum 12.

The Diabetes Prevention Program Outcomes Study (DPPOS) provided the most rigorous prospective data. Among participants randomized to metformin 850 mg twice daily, B12 deficiency (defined as serum B12 <203 pg/mL) occurred in 4.3% at year 5 compared with 2.3% in the placebo group. Low B12 (203 to 298 pg/mL) was even more common, affecting 19.1% of metformin users vs. 9.5% of placebo 13. Longer treatment duration and higher cumulative doses both predicted lower B12 levels.

Clinical consequences are not trivial. B12 deficiency causes megaloblastic anemia and peripheral neuropathy, a symptom that overlaps with and may be misattributed to diabetic neuropathy. A cross-sectional analysis by Wile and Toth found that metformin-treated patients with B12 deficiency had significantly worse vibration perception threshold scores (P <0.001) compared with metformin users without deficiency 14.

The ADA Standards of Care now recommend periodic B12 monitoring in patients on long-term metformin, particularly those with anemia or peripheral neuropathy 9. No fixed monitoring interval is mandated, but most guidelines suggest checking serum B12 at baseline and every 1 to 2 years.

Gastrointestinal Tolerability and Formulation Strategies

GI symptoms are the most common reason patients discontinue metformin. In a real-world cohort study of 1.6 million new metformin users, 16.3% discontinued within 6 months, with GI intolerance cited as the leading cause 15.

The extended-release formulation was developed specifically to reduce GI side effects by slowing drug delivery. A randomized crossover trial showed that metformin ER produced 50% fewer GI adverse events compared with the immediate-release form at equivalent doses 16. This is relevant to the NDMA story: the 2020 recalls temporarily eliminated ER supply for many patients, forcing switches back to IR and reintroducing GI issues that some had resolved years earlier.

Titration strategy matters as well. Starting at 500 mg once daily with the evening meal and increasing by 500 mg per week reduces early GI effects substantially compared with starting at the full dose. The maximum effective dose for most patients is 2,000 mg/day, with minimal additional HbA1c reduction beyond that threshold 3.

Metformin in Contrast-Enhanced Imaging: Outdated Fears vs. Current Guidance

For years, metformin was reflexively held before any procedure involving iodinated contrast media. The concern: contrast-induced acute kidney injury could impair metformin clearance and trigger lactic acidosis. Radiology departments created rigid "hold metformin for 48 hours" protocols.

Current evidence does not support blanket withholding. The American College of Radiology (ACR) Manual on Contrast Media now recommends holding metformin only in patients with eGFR <30, patients receiving intra-arterial contrast with first-pass renal exposure, or patients with acute kidney injury 17. For patients with eGFR ≥30 receiving standard intravenous contrast, metformin can be continued without interruption.

Dr. Matthew Davenport, lead author of the ACR contrast guideline revisions, has stated: "The risk of metformin-associated lactic acidosis after contrast media is extremely low in patients with preserved renal function, and unnecessarily withholding metformin can lead to hyperglycemic complications."

This represents a significant de-escalation from prior practice. Hospitals that still enforce universal hold protocols are applying outdated guidance.

Hepatic Safety and the Rare Signal of Hepatotoxicity

Metformin carries no FDA hepatotoxicity warning, and the National Institutes of Health LiverTox database classifies it as having "no convincing evidence" of direct liver injury 18. Isolated case reports of metformin-associated hepatotoxicity exist, but causality assessment is confounded by the high prevalence of metabolic dysfunction-associated steatotic liver disease (MASLD) in the diabetic population.

Paradoxically, metformin may offer hepatic benefit. A meta-analysis of 12 observational studies found that metformin use was associated with a 50% reduction in hepatocellular carcinoma risk among diabetic patients (pooled OR 0.50, 95% CI 0.34 to 0.73) 19. The mechanism likely involves AMPK-mediated suppression of hepatic lipogenesis and anti-proliferative signaling. These findings remain observational, and no randomized trial has tested metformin as a liver cancer prevention agent.

Metformin and Cardiovascular Outcomes: The UKPDS Legacy

UKPDS 34 remains the only randomized trial to demonstrate a mortality benefit with metformin. In 1,704 overweight patients with newly diagnosed type 2 diabetes, metformin reduced all-cause mortality by 36% (P = 0.011) and diabetes-related death by 42% compared with conventional (diet-only) treatment over a median of 10.7 years 20.

No subsequent trial has replicated this mortality finding. The SPREAD-DIMCAD trial (N = 304) showed reduced cardiovascular events with metformin vs. glipizide in patients with coronary artery disease, but was underpowered for mortality 21. A 2024 Cochrane review concluded that the cardiovascular evidence for metformin relies primarily on one trial from the 1990s and that "the quality of evidence is low" for mortality reduction 22.

This matters for safety signal interpretation. Metformin's favorable risk-benefit profile rests partly on UKPDS 34's cardiovascular findings, but the certainty of that benefit is lower than many clinicians assume. The drug's continued first-line status owes as much to its low cost, weight neutrality, and absence of hypoglycemia as to hard outcome data.

Current FDA Label Status and Ongoing Surveillance

As of 2026, metformin's FDA-approved indications remain limited to type 2 diabetes, though off-label use in prediabetes, polycystic ovary syndrome (PCOS), and weight management is widespread. The boxed warning for lactic acidosis persists with the 2016 eGFR-based thresholds. No new safety communications have been issued since the resolution of the NDMA recalls.

Active surveillance areas include:

• Long-term NDMA exposure outcomes in patients who took affected ER lots (epidemiologic studies pending)
• B12 deficiency screening compliance rates in primary care
• Ongoing assessment of metformin in the TAME (Targeting Aging with Metformin) trial, which is evaluating whether metformin delays age-related diseases in non-diabetic adults 23

Prescribers should check the FDA's MedWatch page for any emerging signals and report suspected adverse events through the same system.

Monitoring Recommendations for Current Prescribers

Baseline assessment before starting metformin should include eGFR, hepatic function panel, and serum B12. Recheck eGFR at least annually, or more frequently in patients with declining renal function or acute illness. Dose-reduce to a maximum of 1,000 mg/day when eGFR falls to 30 to 44, and discontinue at eGFR <30 7.

B12 should be rechecked every 1 to 2 years in patients on continuous therapy, with a lower threshold for testing in anyone presenting with new neuropathy or unexplained macrocytic anemia. Supplementation with oral cyanocobalamin 1,000 mcg daily corrects deficiency in the majority of cases without requiring injections 13.

For patients using metformin ER, confirm that the dispensed lot is not subject to any active recall by checking FDA's recall database. GI intolerance on IR can be addressed by switching to ER, using divided dosing, or taking the medication mid-meal rather than before eating. The minimum effective dose for HbA1c reduction is 1,500 mg/day in most patients.