Metformin-associated lactic acidosis: reinforcing learning points

Abstract

Metformin-associated lactic acidosis (MALA) carries a high mortality rate. It is seen in patients with type 2 diabetes on metformin or patients who attempt suicide with metformin overdose. We present the case of a man in his early 20s with type 2 diabetes, hypertension and hypothyroidism who presented with agitation, abdominal pain and vomiting after ingesting 50–60 g of metformin; he developed severe lactic acidosis (blood pH 6.93, bicarbonate 7.8 mEq/L, lactate 28.0 mEq/L). He was managed with intravenous 8.4% bicarbonate infusion and continuous venovenous haemodiafiltration. He also developed acute renal failure (ARF) requiring intermittent haemodialysis and continuous haemodiafiltration. MALA is uncommon and causes changes in different vital organs and even death. The primary goals of therapy are restoration of acid-base status and removal of metformin. Early renal replacement therapy for ARF can result in rapid reversal of the acidosis and good recovery, even with levels of lactate normally considered to be incompatible with survival.

Background

Metformin is an anti-hyperglycaemic agent belonging to the class of biguanides. Lactic acidosis can occur with metformin use especially when taken in large amounts such as in an overdose. Metformin-associated lactic acidosis (MALA) refers to lactic acidosis associated with metformin overdose or in association with other comorbidities such as renal or hepatic insufficiency or acute infection. Clinically MALA usually presents with gastrointestinal symptoms (nausea, vomiting and diarrhoea), altered mental status, hypotension and hypothermia.¹ MALA is a rare entity, mostly reported to occur in 0.03–0.1 cases per 1000 patient-years² but has a high mortality rate, reported to be around 50%.³ Cases of MALA involving deliberate self-harm from metformin overdose are even more sparse in the literature.⁴ We present a case of MALA in a patient who presented to the emergency department following a deliberate mixed overdose of levothyroxine and a large amount of metformin. We also aim to provide a summary of the underlying pathophysiology and evidence-based management considerations.

Case presentation

A man in his early-20s, with a background of type 2 diabetes, hypertension and hypothyroidism, presented to the emergency department with agitation, severe abdominal pain and vomiting an hour following a mixed overdose of 50–60 g of metformin and an unclear amount of levothyroxine. Aside from metformin (500 mg daily) and levothyroxine (225 µg daily), his other regular medications included amlodipine (2.5 mg daily), telmisartan (20 mg) and hydrochlorothiazide (6.25 mg) daily as a combined pill.

On assessment, his blood pressure was 86/41 mm Hg. His blood biochemistry results revealed an acute kidney injury (AKL) (urea 3.1 mmol/L, creatinine 146 µmol/L, estimated glomerular filtration rate (eGFR) 58 mL/min); previous renal function was normal. Venous blood gas showed a compensated metabolic acidosis (lactate 8 mmol/L, bicarbonate 16.7 mEq/L, base excess (BE) −6.8, pH 7.37). Clotting was deranged with prothrombin time (PT) 17.2 s and activated partial thromboplastin time (APTT) 46.1 s but liver enzymes were normal. The patient was started on intravenous fluid resuscitation with crystalloids and admitted to the acute medical unit. His blood pressure improved but repeat biochemistry a few hours later showed worsening lactic acidosis (lactate 18 mmol/L, BE −17.5, pH 7.27) and renal function (creatinine 258 µmol/L, eGFR 29 mL/min). He was becoming increasingly agitated and confused and was transferred to the intensive care unit 10 hours after presentation where he was immediately intubated and ventilated and central venous access secured. At this time his pH was 6.89 and lactate was 28 mmol/L. His clotting had also worsened (PT 19.8 s, APTT 64.5 s). He was commenced on continuous venovenous haemofiltration (CVVH) and an intravenous infusion of bicarbonate. His blood pressure was supported by 0.1 µg/kg/min norepinephrine.

Despite of 3.5 hours CVVH, his lactate remained persistently raised (29 mmol/L) and following discussion with the national centre for toxicology and poisons, he was switched from CVVH to continuous venovenous haemodialysis (CVVHD) with a dialysate flow rate of 6000 mL/hour pending commencement on intermittent haemodialysis (IHD). This was done with the aim of improving clearance of metformin and lactate, as well as ongoing correction of his acid-base disturbance. Our patient went on to have two sessions of IHD lasting 3 hours over a period of 16 hours with continuous CVVHD in between. He was also given vitamin K and N-acetylcysteine in view of his coagulopathy and potential hepatic impairment in the context of multi-organ dysfunction syndrome.

Renal replacement therapy (RRT), including both continuous and intermittent modalities, was discontinued after 34 hours; at this time pH was 7.51, serum lactate and clotting had normalised and renal function had significantly improved (creatinine 114). His recovery was complicated by ventilator-associated pneumonia.

Outcome and follow-up

The patient remained well after the admission and had mental health team input as an outpatient. However, the patient has now been lost to our UK National Health Service system.

Discussion

Metformin is a small molecule (165 Da) and has an oral bioavailability of 55%. The mean initial half-life of metformin is ~5 hours and terminal half-life is ~20 hours. Eighty per cent of the drug is eliminated unchanged through the kidneys. Metformin is not plasma protein-bound and has a large volume of distribution in the range of 1–5 L/kg.⁵

Mechanism of lactic acidosis with metformin use

Metformin inhibits the conversion of pyruvate to oxaloacetate by pyruvate carboxylase during gluconeogenesis thus leading to increased conversion of pyruvate to lactate and also preventing the conversion of lactate back to pyruvate.

Metformin inhibits complex I of the mitochondrial respiration chain hence increasing lactate production.⁶ It also enhances intestinal glucose utilisation with a subsequent increase in lactate production through anaerobic metabolism.⁷ Metformin has not been shown to increase the release of lactate from muscles.⁸ Another proposed hypothesis is that metformin accumulates in the intestine, increasing the production of lactate within the liver and decreasing lactate metabolism by pyruvate carboxylase suppression.⁹

However, metformin has been shown to either have no significant effect^{9 10} or to only produce a limited increase on the fasting plasma lactate concentration in patients with non-insulin dependent diabetes taking the drug in therapeutic doses.¹¹ It is therefore believed that when metformin is taken in therapeutic doses, significant lactic acidosis would only be expected to occur in the presence of comorbid conditions that are themselves risk factors for lactic acidosis (eg, states of hypoperfusion, sepsis, heart failure, liver cirrhosis, etc). ¹² The literature does contain reports of lactic acidosis in the absence of known risk factors in the context of metformin overdose.^{13 14} In these cases, lactic acidosis develops as a result of the large amount of the drug acutely overwhelming clearing mechanisms.¹⁵

Regardless of the mechanism of MALA, the low pH induced by lactic acidosis further impairs the activity of pyruvate carboxylase further preventing lactate uptake for hepatic gluconeogenesis.^{16 17} The negative inotropic effect of metformin also reduces cardiac output and hepatic blood flow, further compounding the poor lactate clearance.¹⁷

AKI and MALA

AKI is usually seen with MALA both in chronic therapeutic use and in deliberate overdose of metformin, even in those with previous normal renal function (as in our patient).¹⁸ Metformin has been shown to induce mitochondrial function impairment in animal models,¹⁹ but is not known to be nephrotoxic in humans. A multicentre analysis of patients presenting with lactic acidosis secondary to a biguanide found that AKI was present in all their patients, but likely precipitated by other causes such as dehydration secondary to diarrhoea or vomiting²⁰ which are common clinical manifestations of metformin toxicity. In our patient, the AKI was probably, at least in part, contributed to by his angiotensin II receptor antagonist and thiazide diuretic, and potential dehydration from vomiting.

A large cohort study shows that the risk of lactic acidosis in metformin users increases with AKI, and the plasma lactate concentration also increases with increasing AKI stage.²¹ In our patient, the developing renal impairment is likely to have contributed to the worsening hyperlactatemia and acidosis.

Management of MALA and a comparison of different modalities of extracorporeal treatment

Cases of MALA with associated metabolic acidosis (from either chronic ingestion or acute intoxication) reported in the literature have been managed with intravenous bicarbonate therapy and/or RRT when supportive management. The primary aim of management is to treat the lactic acidosis and eliminate metformin. In an animal experiment using Wistar rats, Almani et al found a significant decrease in lactic acid in diabetic rats which were co-treated with metformin and berberine with potential additive effects.²² However this has yet to be translated into human trials for treatment of acute cases. Intravenous sodium bicarbonate is often used as part of the initial supportive management of MALA to counteract acidaemia, but its role in correcting lactic acidosis is debatable and has not convincingly shown a clinical benefit.^{23 24} Its use also comes with potential disadvantages including increased carbon dioxide production, calcium disturbances leading to impaired myocardial contractility, left shift of the oxyhaemoglobin dissociation curve,²⁵ excess sodium load, rebound metabolic alkalosis and reflex vasodilation.²⁶

There have been concerns regarding the management of MALA with RRT noted in the literature. It has been reported that both metformin level and absolute lactate concentration do not appear to be closely associated with prognosis.^{27 28} It has also been proposed that treating the absolute lactate value may not incur any benefit to the patient while exposing them to the risks associated with continuous renal replacement therapy (CRRT), and in fact the amount of lactate removed by RRT is negligible compared with overall plasma lactate clearance.²⁹ However, under certain strained physiological conditions the kidneys and other tissues have a more important role in lactate metabolism. The liver receives 25% of cardiac output with 50%–60% of its oxygen supply coming from the hepatic portal vein. Once hepatic blood flow is reduced to 25% or less, there is a reduction in the delivery and hence metabolism and clearance of lactate. This phenomenon is exacerbated in shocked states, as seen in our patient, where there is inhibition of gluconeogenesis secondary to intracellular acidosis and saturation of lactate uptake by monocarboxylate transporter systems. Paradoxically the liver as an organ in anaerobic conditions becomes a lactate-producing system.³⁰ In situations of hyperlactaemia renal excretion of lactate (with its threshold of 6–10 mmol/L) becomes significant with renal excretion, gluconeogenesis and oxidation of lactate.³¹ In patients such as ours, with MALA and AKI, worsening hyperlactaemia and metabolic acidosis are thus profound. The principle benefit to using RRT in this cohort of patient is thus likely to be as a supportive measure to buffer metabolic acidosis and control volaemic status. Stacpoole et al reported that lactate concentrations >5 mmol/L in patients with severe acidosis pH <7.35 or base deficit greater than six carries a mortality of 80%. Fluid accumulation is also associated with adverse outcomes in critically ill patients. A study by Lalau et al demonstrated that in patients with AKI, fluid overload was independently associated with mortality.⁶ While appropriate fluid resuscitation is critical in the resuscitation of critically unwell patients especially those with sepsis, fluid not enhancing perfusion may exacerbate oedema in the lungs and other tissues. The Fluids and Catheters Treatment Trial Study found that in patients with acute lung injury, a conservative versus liberal fluid strategy improved lung function and shortened the duration of mechanical ventilation and intensive care without increasing non-pulmonary organ failures and no significant difference in mortality.⁷

It is also well reported that AKI and indeed its severity, including the need for RRT, are independent predictors of critical care and hospital mortality.^{8 9}

A comprehensive review by Calello et al has made recommendations pertaining to the use of RRT in MALA, recommending its use in severe metformin poisoning.¹⁸ Suggested indications for extracorporeal treatment include severe acidaemia and high lactate levels, as well as comorbid conditions such as shock, impaired renal function, liver failure and decreased level of consciousness. The rationale behind the efficiency of RRT is that the low molecular weight of metformin and its minimal protein-binding property make it highly diffusible through haemodialysers and haemofilters. The limiting factor is its large volume of distribution.³²

IHD with a bicarbonate buffer has been recommended as first-line RRT.¹⁸ The relatively low molecular weight of metformin makes it more amenable to clearance using diffusive techniques such as IHD and CVVHD, compared with continuous modalities of RRT using convective principles, such as CVVH. IHD has been shown to be potentially better at restoring acid-base status and removing lactate³³ than continuous convective modalities. However, IHD may be less favourable in haemodynamically unstable patients. Keller et al presented arguments in favour of CRRT in their case series.³⁴ CRRT provides gradual removal of solutes and prolonged physiological steady state. They also highlighted the advantage of extended duration of CRRT in the clearance of metformin, which, due to its large volume of distribution, would require several hours of IHD to be restored to therapeutic levels. Regardless of the modality of RRT used, Arroyo et al and Barrueto et al purport that the focus should be on high-efficiency techniques as higher mortality rates have been observed when lower efficiency techniques have been used.^{35 36} A case report by Schopman et al describes the use of two CVVH machines to enhance the effluent rate when their patient with MALA failed to respond adequately with filtration using one CVVH machine.³⁷

In our patient, we started off with CVVH, this is the standard modality used at the time for CRRT within our critical care department. He was changed to CVVHD (initially with no anticoagulant given the coagulopathy and then with prostacyclin) with a dialysate flow rate of 6000 mL/hour. This therapy uses diffusion as the physical principle for solute removal and is primarily effective at removing small-sized and medium-sized molecules and correcting acid base disturbance, electrolyte imbalance and uraemia. The patient also received two sessions of 3 hours of IHD in the initial 16 hours to encourage clearance of metformin. CVVHD was continuous around and following the IHD sessions. Continuous venovenous haemodiafiltration was also suggested as a modality, with the ability to deliver a diffusive process with an additional convective component to facilitate solute removal (small, medium and larger molecular weight molecules) but this was not a therapy used within critical care at the time. A cohort study comparing continuous haemofiltration and haemodiafiltration in critically ill patients for the management of AKI showed that serum lactate and creatinine levels were higher in the haemofiltration group for the first 48 hours of CRRT, but this failed to achieve statistical significance and no difference occurred in 30-day mortality.³⁸ Larger robust studies comparing these two CRRT modalities are lacking in the literature. Of interest, our critical care standard RRT therapy is now CVVHD with citrate anticoagulation. This would have been possible with this patient but would need very strict monitoring for the development of citrate toxicity and also the use of amended protocols in relation to the significant liver impairment.

It is reasonable to conclude in this case study that the utilisation of a combination of CVVHD and sessions of IHD promoted metformin clearance and resolution of the acidosis. It is also our view that early use of RRT in this patient, with AKI and acid-base markers of poorest MALA prognosis, contributed to his survival. On review of the case, we concluded that a direct critical care referral and admission from the emergency department, based on the initial very significant metabolic acidosis and hyperlactaemia, rather than subsequently from the medical unit, would have been more appropriate.

Prognostic factors in MALA (acute deliberate overdose versus incidental toxicity)

The high mortality in MALA due to deliberate metformin overdose appears to be directly related to the severity of lactic acidosis. In a recent systematic review of case reports and case series on acute metformin overdose, it was found that lower serum pH and higher serum lactate concentrations correlated with increased mortality. In this review, all 16 survivors had a serum pH above 6.9 and a serum lactate below 25 mmol/L, while five of the six patients with a pH below 6.9 or a lactate above 25 mmol/L died.⁴ Despite the presence of these latter factors in our patient, he survived.

In contrast, in a study of 56 cases the mortality in MALA associated with incidental metformin toxicity (at therapeutic doses) appeared to be related not to arterial pH or the degree of lactic acidosis but to the presence of coexistent comorbidities.³⁹

Another study of 42 patients has shown that multi-organ dysfunction, in particular acute liver dysfunction, carried the highest risk of mortality.⁴⁰ It also suggests that the risk of death in MALA secondary to deliberate metformin overdose is low if appropriately managed in the intensive care unit. There appears to be consensus that metformin level is not a good predictor of mortality.^{24 27}

And he remained intubated for a further 6 days. He otherwise made an uneventful recovery.

Despite the history of thyroxine overdose free T4 level was 7.3 pmol/L (below normal range) and thyroid-stimulating hormone >100 mU/L. It transpired that he had been non-compliant with levothyroxine. Free T4 remained below or within normal range over the following days and he did not show any signs of thyroxine intoxication.

Learning points.

• Metformin-associated lactic acidosis (MALA) is a complex entity, associated with risks of multi-organ dysfunction and a high mortality rate.

• An understanding of metformin overdosage’s pathophysiology and approach to management is important to physicians.

• The key to successful treatment of MALA lies in early recognition, timely transfer to intensive treatment unit in patients with adverse prognostic features at presentation and early initiation of high efficiency renal replacement therapy if appropriate.