Metformin, heart failure, and lactic acidosis: is metformin absolutely contraindicated?

Evidence from case reports

Most case reports of lactic acidosis in people taking metformin have failed to provide adequate information to permit assessment of causation, including lactic acid concentrations and pH.7 In a review of published case reports, Stades et al showed that plasma concentrations of metformin were not related to increased lactic acid concentration.7 In addition, increased concentrations of neither lactic acid nor metformin were associated with increased mortality risk.7 In contrast, acute cardiovascular events, liver cirrhosis, and sepsis were all associated with an increased mortality risk.7 Interestingly, all but one of the cases in this review had at least one risk factor (renal failure, cardiovascular events, pulmonary failure, hepatic failure, alcohol excess, or sepsis) for the development of lactic acidosis independent of metformin use. Most of the patients developed lactic acidosis in the presence of acute or worsening renal failure. Creatinine concentrations, however, did not correlate with lactic acid concentrations, metformin concentrations, or mortality.7

Similar results were described by Lalau et al, who showed that neither lactate nor metformin concentrations were prognostically related to mortality and that death seemed to be related to other hypoxic disease or underlying ill health.8 The median lactate concentrations were similar in patients who survived and those who died.8 In addition, the median plasma metformin concentration was three times higher in patients who survived, which suggests that accumulation of metformin may not be as important in lactic acidosis as has been thought.8 The lack of a relation between lactic acid/metformin concentrations and mortality and the absence of an association between metformin concentration and lactic acid concentration suggest that the association between lactic acidosis and metformin is coincidental,7 although causality cannot be ruled out completely.

Epidemiological data

Brown et al collected 41 426 person years of data for patients with type 2 diabetes in the era before the introduction of metformin and found a rate of 0.097-0.169 events of lactic acidosis per 1000 person years.9 This rate of lactic acidosis events is similar to that reported in patients with type 2 diabetes taking metformin,9^{w1 w11} which raises the possibility that the incidence of lactic acidosis in metformin treated patients might be related to type 2 diabetes rather than metformin treatment itself.9 A Cochrane review of 206 comparative trials and cohort studies in patients with type 2 diabetes who were treated with metformin and had no contraindications to its use, found no evidence of increased risk of developing fatal or non-fatal lactic acidosis in the subgroup of metformin treated patients.3 It also found no difference in lactate concentrations between patients treated with metformin or with non-biguanide drugs.3

Disregard of contraindications

Several reports found that physicians have increasingly ignored contraindications to prescribing metformin and yet the incidence of lactic acidosis has remained very low (table 2⇓).4 6 10 11 12 13 14^w7 Emslie-Smith et al, in a population based study in Scotland between January 1993 and June 1995, found that 24.5% of patients receiving metformin had contraindications to its use, including acute myocardial infarction, cardiac failure, renal impairment, or chronic renal disease. Despite this, only one episode of lactic acidosis occurred in 4600 patient years, and this was in a 72 year old patient with acute myocardial infarction complicated by acute renal failure.4 A cross sectional analysis, by Holstein et al, of 308 consecutive type 2 diabetes patients treated with metformin from 1 January 1995 to 31 May 1998 found that 73% of these patients had at least one contraindication to the use of metformin.6 None the less, no cases of lactic acidosis were seen.6 Contraindications in the study by Holstein et al included renal impairment (creatinine clearance <60 ml/min), hepatic impairment, chronic respiratory failure, heart failure (ejection fraction <50%, lung congestion on radiograph), advanced coronary heart disease conditions, chronic alcohol misuse, severe infections, pregnancy or breast feeding, intravenous administration of contrast agents, and operations under general anaesthesia. Of note, none of the patients in the UK prospective diabetes study developed lactic acidosis.1 The study protocol, however, would have ensured that metformin was not used in patients with contraindications.

Table 2⇑ summarises the studies that have shown increased disregard of contraindications to metformin and the contraindications used in each study. However, these studies are observational, so confounding factors, particularly confounding by indication, affecting the outcome could not be excluded. In addition, the percentage of patients taking metformin who have at least one contraindication varies considerably. The evidence from these reports reinforces the viewpoint that metformin is an extremely rare cause of lactic acidosis in patients with type 2 diabetes, even in the presence of contraindications including renal, hepatic, and cardiac failure.

Canadian population data

The Saskatchewan health database (in Canada) was used to examine population based mortality for new users of oral hypoglycaemic agents.19 The researchers identified patients with prescriptions for sulphonylurea or metformin in 1991-6. They followed prescription records prospectively for between one and nine years; patients with any insulin use were excluded. The different groups were controlled for age, sex, comorbidity, and the presence of coronary artery disease. Data from 8866 patients (3033 sulphonylurea monotherapy, 1150 metformin monotherapy, and 4683 combination treatment) were reported. One hundred and fifty nine (13.8%) deaths occurred in the metformin monotherapy group and 635 (13.6%) in the combination treatment group, compared with 750 (24.7%) deaths in the sulphonylurea monotherapy group. The numbers of cardiovascular related deaths were 80 (7.0%) in the metformin monotherapy group, 299 (6.4%) in the combination group, and 351 (11.6%) in the sulphonylurea monotherapy group.

Table 3⇓ summarises the results of this study, including the odds ratios and confidence intervals. Metformin, alone or in combination with sulphonylurea, was associated with reduced all cause and cardiovascular mortality compared with sulphonylurea monotherapy among new users of these agents. The main limitation of this study is that it is observational and based on an administrative database, so the data on drug prescription may not exactly reflect the drug consumption. The authors claim that their database is known for its quality.19 In addition, because of the observational nature of the study, confounding factors such as confounding by indication cannot be ruled out.

Canadian cardiac failure data

In a more recent study using the same database, Eurich et al compared the clinical outcomes of patients with type 2 diabetes who were known to have cardiac failure and were on metformin alone, sulphonylurea alone, or combination treatment.18 They identified 2793 patients with type 2 diabetes who were treated with oral hypoglycaemic agents and had a hospital admission with heart failure between 1991 and 1996. Patients who were on insulin or had had heart failure for more than three years before starting oral hypoglycaemic agents were excluded, and 1833 patients were eligible for the study. Patients were followed up for a period of between one and nine years, and the primary outcome was all cause mortality at one year (short term) and at the end of follow-up (long term). Secondary outcomes were all cause hospital admissions at one year and long term. The researchers also evaluated the effects of oral hypoglycaemic agents on composite outcomes, including all cause hospital admission and all cause mortality. At one year, compared with the 200 (26%) deaths in the sulphonylurea monotherapy group, 29 (14%) deaths occurred in the metformin monotherapy group (unadjusted hazard ration 0.52, 95% confidence interval 0.35 to 0.76) and 97 (11%) deaths (unadjusted hazard ratio 0.41, 0.32 to 0.52) in the metformin-sulphonylurea combination group. After controlling for age, sex, drugs known to affect outcomes of heart failure, and total physician visits before diagnosis of heart failure, the authors found that metformin alone (adjusted hazard ratio 0.66, 0.44 to 0.97) or in combination with other agents (0.54, 0.42 to 0.70) was associated with reduced one year all cause mortality compared with sulphonylurea monotherapy in patients with heart failure. The long term mortality and morbidity in patients treated with metformin, alone or in combination with other antidiabetic agents, was lower than that observed for patients treated with sulphonylurea only (52% for sulphonylurea monotherapy v 33% for metformin monotherapy v 31% for combination treatment).18

Table 3⇑ summarises the long term outcomes of this study, including hazard ratios and 95% confidence intervals. The all cause hospitalisation was significantly lower in the metformin monotherapy group compared with sulphonylurea monotherapy or combination treatment both at one year and long term. Similar to the previous study by Johnson et al, one of the main limitations is that this study is observational and based on a database, which means that drug prescription may not reflect exposure and that confounding by indication could not be ruled out. Another important limitation is that the investigators did not have any information about the severity of heart failure and the presence or absence of renal failure. The latter is particularly important, as renal failure is an independent predictor of poor prognosis in heart failure. If the metformin group had less renal failure, this would affect the validity of the results. This is unlikely, however, as renal dysfunction is very common in patients with heart failure,20 so a considerable proportion of patients in all groups are likely to have had renal impairment.

US cardiac failure data

Masoudi et al found similar results. They evaluated the impact of insulin sensitisers (metformin or thiazolidinediones) on outcomes in patients with type 2 diabetes and cardiac failure.5 Crude one year mortality was lower among patients treated with a thiazolidinedione (30.1%) or metformin (24.7%) compared with patients treated without insulin sensitising drugs (36.0%; P=0.0001 for both comparisons).5 Treatment with thiazolidinedione (hazard ratio 0.87, 0.80 to 0.94) or metformin (0.86, 0.78 to 0.97) was associated with significantly lower risks of death.5 No significant association was found between treatment with sulphonylurea (0.99, 0.91 to 1.08) or insulin (0.96, 0.88 to 1.05) and mortality.5 Admissions for all causes did not differ with either insulin sensitiser, although the risk of readmission for heart failure was higher in those receiving thiazolidinedione (1.06, 1.00 to 1.09) and lower with metformin treatment (0.92, 0.92 to 0.99).5 The study was an observational retrospective cohort, so the results should be interpreted with caution. Although the authors made adjustments for a wide range of variables, including markers of severity of heart failure and comorbidities, variations in the institution and clinician who treated the patients may influence the results. Consequently, these results could reflect the influences of unmeasured confounding factors, including confounding by indication.