Drug Information Rounds

The Safety of Metformin in Heart Failure Felicia Roberts and Gina J Ryan

Request Is metformin safe to use in patients with heart failure?

Response BACKGROUND

OBJECTIVE: TO examine the evidence regarding Ihe safely of metformin in heart failure. DATA SOURCES: Searches in MEDLINE and International Pharmaceutical Abstracts were performed (1966-February 2007). Search terms included metformin, heart failure, lactic acidosis, clinical trials, and insulin resistance. STUDY SELECTION AND DATA EXTFIACTION: Published studies and case reports

that evaluated the causal link between metformin and lactic acidosis in patients with heart failure were selected for review.

Chronic heart failure (CHF) is among DATA SYNTHESIS: There were no case reports of patients who had metforminthe nation's leading causes of hospitalassociated lactic acidosis when heart failure was the only contraindication. Two large retrospective studies showed that metformin does not increase the risk of izations for individuals 65 years of age lactic acidosis in patients with heart failure. However, these retrospective and older, affecting nearly 5 million US analyses did not account for many important confounding variables. A reduction residents. In people with diabetes; the in mortality rates in metformin users with New York Heart Association Class III risk of developing CHF nearly doubles and IV heart failure was observed in one small (N = 94) prospective trial. as the population ages.' Although the exCONCLUSIONS: Results from 3 trials suggest that metformin may be safe to use in act etiology of CHF is not fully underheart failure. Large prospective trials are needed to provide conclusive evidence stood, studies support the theory of inregarding metformin's safety. Until then, use of metformin in heart failure patients sulin resistance as a predictive risk facshould not be recommended routinely. If it is used in patients with heart failure, tor.1.3 If such a role exists, aggressive they should be monitored closely for signs of lactic acidosis. treatment with drugs that lower insulin KEY WORDS: heart failure lactic acidosis, metformin, resistance would be of benefit in manAnn Pharmacothcr 2007;41:642-6. agementof thesecomorbidconditions. Published Online, 20 Mar 2007, www.theannals.com, DOI 10.1345/aph.1H523 Metformin,a biguanide,lowers blood glucose levels by increasing hepatic intial risk for development of lactic acidosis. This rare but sulin sensitivity, increasingmuscle cell insulin sensitivity, potentially fatal metabolic condition results from severe and decreasing intestinal glucose absorption.' Metformin tissue hypoperfusion, which causes an increased producalso improves cardiovascular risk factors through its eftion and decreased metabolism of lactate. Serum lactate fects on glycemic control,lipid metabolism, inflammation, levels elevate to dangerously high levels (>45.0 mg/dL), vascular endothelial function, and plasminogen activator blood pH decreases «7.35), and there is an increased aninhibitor levels.5.' Metformin is the only oral antihyperion gap. Signs and symptoms of lacticacidosis are nonspeglycemic agent shown to reduce the risks of cardiovascucific and include nausea, vomiting, fatigue, altered conlar-related deathsin patients with type 2 diabetes," sciousness, abdominal pain,and thirst.' Despite these benefits, the use of metformin is conBiguanideshave been reported to cause lactic acidosis. traindicated in patients with heart failuredue to the potenPhenformin, an earlier biguanide, was removed from the Author information provided at the endof the text.

US market in 1976 after 306 cases of lactic acidosis were reported," Phenforminhas a long lipophilic side chain that

642 • The AnnalsofPharmacotherapy • 2007April, Volume 41

wwwiheannalscom

causes it to bind to mitochondrial membranes,which impairs the electron transport system. This side chain also causes metformin to be hydroxylated by the liver. This action inhibits aerobic metabolism, resulting in decreased lactate oxidation and increased lactate release from skeletal muscle.10,11 Although phenformin and metformin are pharmacologically related, the incidence of lactic acidosis is 10-20 times less with metformin,"Metformin is bisubstituted with 2 smallgroups that decrease its lipophilicity. As a result, it is poorly bound to mitochondrial membranes and does not increase lactate production in skeletal muscle.10,1I Animal data indicate that metformin does increase lactate production in the extrahepatic splanchnic bed of the small intestine. However, at recommended dosages,the magnitude of this effect is minimal. The true incidence of metformin-associated lactic acidosis (MALA) remains unknown." The average rate is estimated to be 0.03cases per 1000patient-years, with50% of cases being fatal.14 However, evidence linking elevated lactic acid levels to metformin use or metformin levelsis poor.13,1S,16 For example, a 2003 Cochrane Database SystematicReview showed no difference in the net change from baseline in lactate levelsin patients treated with metformin compared with patients treated with non-biguanlde agents." A more recently published Cochrane Database Systematic Review of 206comparative trials and cohortstudies revealed no cases offatal or nonfatal lactic acidosis in 47846 patient-years of metformin use," Additionally, results from a one yeartrial showed no observed difference in meanplasmalactate levels in patients treated withmetformin (n = 7227)compared with those receiving antidiabetic agents other than metformin (n = 1505).18 No cases of lactic acidosis were reported However, since metformin is associated with a 50% mortality rate," minimizing the risk of developing MALA is critically important.Conditions such as renal dysfunction, CHF,liver disease,and tissue hypoxia interfere with normal metabolism and excretion of lactic acid and increase the risk of developing MALA.4.13,19 Some suspect that high-riskconcomitantconditionsare responsiblefor mortality rates, rather than the use of metformin.19.20 In CHF, cardiac output is decreased, tissue perfusion is reduced, and the risk of lactic acidosis is increased. Despite the manufacturer's warning" against the potential risk of l~tic acidosis, manyphysicians prescribe metformin in patients withexisting CHF.IS In onereport, 73% of metformin Users had at leastone contraindication to its use at hospitalization, with 25% havingCHF.II Interestingly, 41% ofpatients werecontinued on metformin therapy despite hospital admissions to cardiology and general medicine services." Perhaps their prescribers agree with data that show that metformin improves cardiovascular outcomes in patients withheartfailure.s", This article examines evidence regarding the safety of metformin use in heartfailure. wwwzheannalscom

LITERATURE REVIEW

Searches of MEDLINE and International Pharmaceutical Abstracts (1966-February 2007) were performed to identify literaturethat evaluated the effects of metformin on clinical outcomes in patients with type 2 diabetes and CHF. Published case reports wereincluded if CHF was the onlyreported contraindication to metformin use.Tworetrospective studies and one prospective trialthatmet thesecri~eria ,:ere identified. Therewere no case reports of MALA ill which the only contraindication to metformin was CHF. In a retrospective cohortanalysis by Masoudi et al.,' the Nati~nal Heart Care Projectdatabase was used to identify ~edicare patients with diabetes who had recently been discharged from the hospital with a primary diagnosis of CHF. The primaryoutcomewas to establish a correlation between prescriptions for antihyperglycemic agents and CHF outcomes. The analysis included 16417 patients aged 65 years or older who were treated with CHF drugs (eg, angiotensin-converting enzymeinhibitors, ~-blockers, statins, aspirin) duringone year. Subjects were excluded if they died while in the hospital, were less than 65 years of age, data on vital status or readmission were unavailable, or they weredischarged to a hospice. Patients were divided into the following treatment groups: metformin (n 1861),thiazolidinedione (TZO) (n =2226),and eitherinsulin or sulfonylurea (n =12069). Of thosetreated with metformin, the mean age ± SO was75.8 ± 6.8 years,57% werefemale, 13%werenonwhite, and 26% hadglucose values greater thanor equal to 140mg/dL. In the 1ZD group,the mean age was 75.9 ± 6.8 years,61% were females, 12% werenonwhite, 41% haddiabetes-related complications, and 20%had glucose values greater thanor equal to 140 mg/dL. Themeanagein thegroup treated with insulin or sulfonylurea was77.0± 7.1 years, with 58%being female, 16% being nonwhite, and 22% having a glucose value greater thanorequal to 140mg/dL.' Multivariable Cox proportional hazard models were usedto assess the independent relationship between insulin sensitizerprescriptions and healthoutcomes. Adjustments usinghazard ratios were madeto account for differences in confounding variables between patient, physician, andhospital characteristics. Patient characteristics thatweretaken into account wereleft ventricular function, cardiachistory, diabetes complications, history of cerebrovascular accident, discharge medication, chronic lungdisease, anddementia.' Results of the study showed a lower unadjusted one year mortality rate in patients treated with metformin compared with those treated with a sulfonylurea or insulin (24.7% and 36.0%,respectively; p < 0.0001). Afteradjustments,metformin remained associated with a significantly lower risk of death from all causes (HR 0.86; 95% CI 0.78 to 0.97). Patients on metformin also had lower unadjusted rates of hospital readmissions for CHF (58.6%; p