Overdoses with metformin are relatively uncommon, but may have serious consequences. In a five-year review of toxic exposures reported to U.S. poison control centers, only 4072 out of nearly 11 million exposures involved metformin, corresponding to less than one in 2500.3
There were a total of 9 deaths (0.2% of all metformin-related exposures), 32 cases with life-threatening signs or symptoms and/or residual disability (0.8%), and 187 cases with moderate clinical effects (4.6%). Cases of serious toxicity from metformin overdose are rare enough that single case reports and small case series continue to be published, which often describe extracorporeal methods of managing the consequent severe lactic acidosis.5–13,15
Lactic acidosis may occur either with therapeutic metformin dosing or in overdose; 0.03 cases of lactic acidosis per 1000 patient-years occur with therapeutic dosing, and most of these cases occur among patients who have contraindications to metformin, such as renal insufficiency.2
In overdose situations, lactic acidosis is seen much more commonly, although the exact incidence is unclear. Lactic acidosis was seen in 1.6% of metformin exposures reported to poison control centers, but only 10% of these exposures were due to intentional overdoses.3
In a review of poison control center inquiries from Germany, the incidence of metformin-associated lactic acidosis was 12.8% (14 of 109 calls); however, the incidence of attempted suicide was 56.9% in this group, suggesting that higher metformin doses were involved.11
Case reports and series of severe toxicity from metformin are typically associated with severe lactic acidosis,5–13,15,17,18
and many of these are due to intentional overdose.5–10,12,13,15,18
Such publications, however, form an obviously biased group, and the actual incidence of lactic acidosis with intentional metformin overdose remains unknown.
The prognosis in cases of undifferentiated lactic acidosis is poor, with an expected case-fatality rate of 30–50%.2,19
In cases of metformin-associated lactic acidosis, the serum lactate level does not correlate with prognosis, even with lactate levels as high as 35.5 mmol/L.19
In cases where metformin levels have been measured, there is a poorer prognosis with lower metformin levels.19
This unusual finding suggests that patients developing lactic acidosis at lower metformin levels probably have concurrent underlying disease, placing them at higher risk both for accumulating lactate and for death.
The most striking feature of the case reported here, however, is the profound and progressive hyperglycemia, which has not been previously reported. Metformin generally does not cause significant alterations in serum glucose levels, even in overdose situations. Unlike the sulfonylurea and meglitinide classes of diabetes drugs, which stimulate insulin release from the pancreas and therefore lower the glucose level, the biguanide drugs have more complex effects on glucose homeostasis that tend to reduce hyperglycemia without inducing hypoglycemia. These effects include: inhibition of hepatic gluconeogenesis (primarily through inhibiting hepatic lactate uptake), improving peripheral insulin sensitivity, inhibition of fatty acid oxidation, and possibly reducing intestinal glucose absorption.2,19
Metformin also increases production of lactate by the intestinal mucosa and suppresses pyruvate carboxylase, which impairs lactate clearance. Thus, lactic acidosis may occur in patients who have overdosed on metformin, or those with renal insufficiency, whose lactate clearance is already impaired.
Although metformin does not directly lower glucose levels, metformin-associated hypoglycemia has been reported, both with therapeutic dosing and with overdoses.3,14,16
Hypoglycemia was reported in 2.8% of metformin exposures reported to a poison control center.3
In many of these cases the patients were receiving concurrent treatment with sulfonylurea drugs, which likely contributed in causing hypoglycemia. Some instances of drug interactions with metformin have been associated with hypoglycemia (e.g, with enalapril, which increases insulin sensitivity)20
, although hypoglycemia with isolated metformin exposure is very rare.
Hyperglycemia associated with metformin overdose has occasionally been reported, but is even less common than hypoglycemia. In their review of 4072 metformin exposures, Spiller and Quadrani found only 18 cases associated with hyperglycemia (versus 112 with hypoglycemia), which they attributed as most likely due to the patients’ underlying diabetes.3
Nearly 40% of cases with hyperglycemia (seven of 18) occurred in patients who died, and seven of the nine overall fatalities had hyperglycemia, suggesting that it may be associated with particularly severe metformin toxicity.
Hyperglycemia has been related to acute pancreatitis in a few cases of metformin toxicity from both intentional overdose11,18
and therapeutic dosing.17
In the previously published cases of pancreatitis with metformin toxicity, the degree of hyperglycemia was considerably lower than in our patient: 162 mg/dL17
, 345 mg/dL11
, and 450 mg/dL (this final patient presented with hypoglycemia [23 mg/dL] and only became hyperglycemic after treatment with parenteral glucose).18
Similarly, our patient’s peak serum glucose exceeded the highest level (698 mg/dL) reported among diabetic patients taking metformin therapeutically who went on to develop lactic acidosis.21
Since our patient did not have a history of diabetes, his extreme hyperglycemia is all the more unexpected. Also, we could not identify any previously published cases where the hyperglycemia progressively worsened, in the absence of administration of glucose, as occurred in our patient early in his clinical course.
The potential mechanism for the severe hyperglycemia in our patient is not clear. Nothing among metformin’s known mechanisms would logically explain the progressive and severe hyperglycemia, especially since these mechanisms should tend to limit the glucose level. However, if one considers what might occur if the patient could no longer secrete enough insulin, as may occur with pancreatitis, then a potential explanation arises. With a lack of circulating insulin, metformin’s ability to enhance peripheral insulin sensitivity would count for nothing. Glucose would accumulate in the serum, since it would not be able to enter the tissues. It is also possible that a counter-regulatory hormone surge (epinephrine ± glucagon) from the acute physiologic stress of the overdose contributed to the hyperglycemia. This mechanism is conjectural, since pancreatitis was not confirmed in our patient with serum amylase and lipase levels, nor with any radiographic study showing pancreatic inflammation. Similarly, no circulating insulin levels were measured, which would have to be low for this mechanism to work. Nevertheless, pancreatitis remains a promising potential mechanism, as our patient’s clinical presentation with complaints of vomiting and abdominal pain is consistent with previously reported cases of metformin-associated pancreatitis.11,17,18
Alternate toxicologic causes for this patient’s hyperglycemia seem unlikely. Only a few toxins are routinely associated with hyperglycemia, including calcium channel-blockers and the ingestion of agents that specifically poison pancreatic beta-islet cells, such as alloxan, nitrophenolurea (an obsolete rodenticide), and streptozocin (a chemotherapy agent for pancreatic islet cell tumors). Calcium channel-blockers can impair insulin release, often resulting in modest hyperglycemia. Our patient had no known access to these drugs and did not present with hypotension or bradycardia. Similarly, our patient had no known access to beta-islet cell poisons, and the resultant hyperglycemia from these agents occurs in association diabetic ketoacidosis within a few days after significant exposure.
Our patient also ingested ethanol, which may have contributed to his severe toxicity. Ethanol itself may have increased the serum lactate level or induced pancreatitis. Also, the interaction of ethanol with biguanide drugs has been shown to increase lactate levels both in animal and human studies.4
Another potential complicating factor was the elevated osmolal gap in our patient, unaccounted for by the serum ethanol level. The presence of a highly elevated osmolal gap (>20 mOsm/kg) is most frequently caused by exposure to toxic alcohols,22
and our patient had a progressively worsening metabolic acidosis, which would be consistent with toxic alcohol poisoning. However, our patient denied ingesting anything but ethanol and metformin, and his parents confirmed that they saw no evidence at home of exposure to products containing isopropanol, methanol, or ethylene glycol. An unexplained osmolal gap may be seen in several severe disease states, including shock or sepsis,22
and was also seen in four out of nine cases of fatal metformin toxicity,3
providing an alternate explanation for its presence. We believe our patient’s severe acidosis and osmolal gap is highly unlikely to have occurred from toxic alcohol exposure for several reasons. Firstly, there was no historical evidence supporting the ingestion of any alcohol other than ethanol. Secondly, the patient did not exhibit additional clinical signs of end-organ damage, such as visual complaints from methanol toxicity, nor calcium oxalate crystalluria from ethylene glycol. The patient did have an elevated initial serum creatinine of 2.1 mg/dL, but this did not worsen as would be expected from ethylene glycol poisoning. Also, he had no ketonuria as would occur with isopropanol ingestion. Ingestion of propylene glycol might result in presentation with lactic acidosis and renal insufficiency, but such exposures are rare and severe acidosis should not occur in the presence of a supra-“therapeutic” ethanol level. Even if our patient had had severe toxic alcohol poisoning, the treatment would have included alcohol dehydrogenase inhibition, which was already effected by the presence of ethanol, and hemodialysis, which was already planned to treat his severe metformin toxicity.