C: PHARMACEUTICALS

HISTORY AND EPIDEMIOLOGY

During the last century there were several outbreaks of toxicity in the United States associated with pharmaceutical additives (Chap. 2). The 1937 Massengill sulfanilamide disaster is the most notorious of these epidemics. Diethylene glycol, an excellent solvent and potent nephrotoxin, was substituted for the additives propylene glycol and glycerin in the liquid formulation of a new sulfanilamide antibiotic because of lower cost.^27,64,73 As a result, more than 100 people died from acute kidney failure.²⁷ Outbreaks of acute kidney failure occurred when diethylene glycol was used to solubilize acetaminophen in South Africa, Bangladesh, Nigeria, and Haiti, cough syrup in Panama, and teething powder in Nigera.^{22,30,75,87,133,143}

In December 1983, a new parenteral vitamin E formulation (E-Ferol) was introduced. It contained 25 units/mL of α-tocopherol acetate, 9% polysorbate 80, 1% polysorbate 20, and water for injection. At the time, no premarketing testing was required for new formulations of an already approved drug. Several months after its release, a fatal syndrome in low-birth-weight infants, characterized by thrombocytopenia, acute kidney injury, cholestasis, hepatomegaly, and ascites, was described.^1,119 Thirty-eight deaths and 43 cases of severe symptoms were attributed to E-Ferol. Vitamin E was thought to be the cause and E-Ferol was recalled from the market 4 months after its release. It is now believed that the polysorbate emulsifiers were responsible.¹

Although these additive-related occurrences are rare, relative to the frequency of pharmaceutical additive use, they illustrate the potential of pharmaceutical additive toxicity.

Pharmaceuticals are labeled specifically to focus attention on the active ingredient(s) of a product, thus giving the misimpression that additive ingredients are inert and unimportant. Additives, or excipients as they are more properly termed, are necessary to act as vehicles, add color, improve taste, provide consistency, enhance stability and solubility, and impart antimicrobial properties to medicinal formulations. Although it is true that most cases of excipient toxicity involve exposure to large quantities, or to prolonged or improper use, these adverse events are nonetheless related to the toxicologic properties of the excipient.

Prior to selecting the specific additives and quantity necessary for a drug formulation, the drug manufacturer must consider several factors, including the active ingredient’s physical form, its solubility and stability, the desired final dosage form and route of administration, and compatibility with the dispensing container materials. Often, the same active ingredient requires different excipients to impart appropriate pharmacokinetic characteristics to different dosage forms, such as in long-acting and immediate-release formulations. Similarly, multiple-dose injection vials containing the same active ingredients as single-dose vials specifically require the addition of a bacteriostatic xenobiotic not necessary for single-dose vials.

Unlike requirements for active ingredients, there is no specific US Food and Drug Administration (FDA) approval system for pharmaceutical excipients. As such, the FDA determines the amount and type of data necessary to support the use of a specific excipient on a case-by-case basis. Under current practice, only excipients that were previously permitted for use in foods or pharmaceuticals are defined as generally recognized as safe (GRAS), or “GRAS listed.” All components of a pharmaceutical product, including excipients, must be produced in accordance with current good manufacturing practice standards to ensure purity. The Safety Committee of the International Pharmaceutical Excipients Council developed guidelines for the toxicologic testing of new excipients.¹⁷² Because of patent protection laws, it was not until 1985 that manufacturers were required to provide a list of inactive ingredients contained in all pharmaceutical products. Although it is becoming easier to identify pharmaceutical additives in products, information on their effects and the mechanisms by which they cause adverse responses are often unknown or difficult to obtain.

This chapter summarizes the available literature on commonly used additives associated with direct toxicities (Table 46–1). Data on pharmacokinetics and mechanism of toxicity are presented when data are available. Although many additives are associated with hypersensitivity reactions, including anaphylaxis, these are not discussed because of their nonpharmacologic basis. However, excipients should always be considered as possibly causative in patients who develop hypersensitivity reactions.

BENZALKONIUM CHLORIDE

Benzalkonium chloride (BAC, BAK), or alkyldimethyl (phenylmethyl) ammonium chloride, is a quaternary ammonium cationic surfactant composed of a mixture of alkyl benzyl dimethyl ammonium chlorides. Although it is the most widely used ophthalmic preservative in the United States, it is also considered the most cytotoxic (Table 46–2).^96,104 Benzalkonium chloride is also used in otic and nasal formulations, and in some small-volume parenteral preparations. The antimicrobial activity of BAC includes gram-positive and gram-negative bacteria, and some viruses, fungi, and protozoa. Because of its rapid onset of action, good tissue penetration, and long duration of action, BAC is preferred over other preservatives. The concentration of BAC in ophthalmic medications usually ranges from 0.004% to 0.01%.¹⁰⁴ Strong BAC solutions (>0.1%) can be caustic (Chap. 103).

Ophthalmic Toxicity

Corneal epithelial cells harvested from human cadavers within 12 hours of death were exposed to a medium containing 0.01% BAC.¹⁶⁸ The surfactant properties of BAC resulted in intracellular matrix dissolution and loss of epithelial superficial layers. Following exposure to the medium, mitotic activity ceased and degenerative changes to corneal epithelium were noted. During a 24-hour observation period, epithelial cell cytokinetic or mitotic activity did not occur. Patients with compromised corneal epithelia are at increased risk for the adverse effects of BAC.¹⁶⁸

Two case reports demonstrate the potential toxicity of BAC and highlight the difficulty of diagnosing BAC toxicity. A 36-year-old woman complained of decreased vision when she inadvertently switched from Lensrins, a contact lens cleaning solution, to Dacriose, an isotonic boric acid solution preserved with BAC. After 3 days, she had inflammation, pain, and decreased visual acuity. Examination of the cornea revealed many superficial punctate erosions of the epithelium. An in vitro experiment identified significant binding of BAC to soft contact lenses.⁵⁹ In the second case, a 56-year-old man diagnosed with keratoconjunctivitis sicca was treated with topical antibiotics and artificial tears containing BAC. Following one year of continual use, the patient developed intractable pain, photophobia, and extensive breakdown of the corneal epithelium. Not suspecting the BAC-containing products, the patient continued to use the artificial tears solution for another 9 years despite continued pain and decreasing visual acuity. Replacement with a preservative-free saline solution resulted in resolution of pain, photophobia, and corneal changes.¹⁰⁴ Furthermore, there are newer ophthalmic medication preservatives available (eg, polyquaternium-1, stabilized oxychloride complex, sodium perborate, Sof Zia), which in studies are less toxic than BAC.⁴

A case series of corneal endothelial injury following the inadvertent intraocular use of balanced salt solution (BSS) preserved with BAC instead of preservative-free BSS in 12 patients undergoing phacoemulsification, a surgical technique to remove cataract lenses. The BSS was instilled in the anterior chamber. The operating room had run out of preservative-free BSS and, unbeknownst to the surgeon, it was replaced with the BAC-containing BSS, which contained 0.013% BAC. This is in excess of recommended concentration for intraocular use and is associated with corneal endothelial injury and edema. Within 48 hours of instillation of the BSS, the visual acuity in all 12 patients was limited to only being able to count fingers at 2 feet. This persisted in 11 of the patients at a 6-month follow-up evaluation after the instillation of the BSS. One patient had improvement at 6 months of visual acuity to 20/120 and 20/30 without and with pinholes, respectively.¹⁰⁶

Nasopharyngeal and Oropharyngeal Toxicity

Human adenoidal tissue was exposed to oxymetazoline nasal spray preserved with BAC at concentrations ranging from 0.005 to 0.15 mg/mL for 1 to 30 minutes.¹⁶ Irregular and fractured epithelial cells occurred at all concentrations; however, these findings developed earlier and more frequently with the higher concentrations. The number of beating ciliary bodies also decreased as the duration and the concentrations increased. Benzalkonium chloride decreases the viscosity of the normal protective mucous lining of the naso- and oropharynx, resulting in cytotoxicity.

Administration of one of 3 nasal corticosteroid sprays (beclomethasone dipropionate, flunisolide, budesonide) preserved with either 0.031% or 0.022% BAC in the right nostril of rats twice daily for 21 days caused squamous cell metaplasia and a decrease in the number of goblet cells, cilia, and mucus.¹⁷ No histologic changes occurred in rats receiving any of the preservative-free steroids or in tissue exposed to 0.9% sodium chloride solution administered into the left nostril as the control. Similarly, in another study, epithelial desquamation, inflammation, and edema occurred when 0.05% and 0.10% BAC was applied hourly to the nostrils of rats for 8 hours.⁹⁹ No lesions developed in the nostrils of rats receiving 0.01% BAC.

In an in vitro study, cultured human nasal epithelial cells were exposed to varying concentrations of BAC compared with another preservative, potassium sorbate (PS), with phosphate-buffered saline (PBS) as a control. Cell viability was greatly reduced at the higher concentrations of BAC compared with no decrease in cell viability in the PS or PBS groups. Additionally, at concentrations used clinically, loss of microvilli, destruction of cell membranes, and poor cytoskeletal alignment demonstrated by electron microscopy occurred.⁸⁴

An in vitro study of human nasal mucosa exposed mucosa to either fluticasone or mometasone preserved with either BAC or PS at various concentrations with subsequent measure of ciliary beat frequency. Although PS did not affect ciliary beat frequency at any concentration, BAC adversely affected ciliary beat frequency. At lower concentrations, BAC slowed ciliary beat frequency and brought it to standstill at higher concentrations.⁸⁵ Another in vitro study in human nasal epithelial comparing budesonide, fluticasone propionate, azelastine hydrochloride, or levocabastine hydrochloride preserved with either BAC or PS at various concentrations showed similar results.⁹²

Polyquaternium-1 (PQ), an alternative detergent preservative to BAC, is reported to be less toxic to various cell types. In an in vitro study of cultured human corneal and conjunctival cells exposed to various topical glaucoma medications preserved with either BAC with concentrations ranging from 0.001% to 0.005% or PQ 0.04% with other arms using BSS as “live” control and 70% methanol and 0.2% saponin as “dead” controls, PQ had less cytotoxicity to both cell types compared to BAC.⁵ In an in vivo murine model comparing BAC concentrations 0.1% and 0.5% and PQ 0.1% and 0.5% to control BSS showed decreased tear production, increased corneal punctate lesions, increased in conjunctival injection, corneal neovascularization, and stromal inflammation in BAC compared to PQ and BSS.¹⁰⁰ It is important to note that both of these studies^5,100 involving PQ were funded by the manufacturer.

BENZYL ALCOHOL

Benzyl alcohol (benzene methanol) is a colorless, oily liquid with a faint aromatic odor that is most commonly added to pharmaceuticals as a bacteriostatic (Table 46–3). In 1982, a “gasping” syndrome, which included hypotension, bradycardia, gasping respirations, hypotonia, progressive metabolic acidosis, seizures, cardiovascular collapse, and death, was first described in low-birth-weight neonates in intensive care units.^23,66 All the infants had received either bacteriostatic water or sodium chloride solution containing 0.9% benzyl alcohol to flush intravenous catheters or in parenteral medications reconstituted with bacteriostatic water or saline.^23,66 The syndrome occurred in infants who had received more than 99 mg/kg of benzyl alcohol (range, 99–234 mg/kg).⁶⁶ The World Health Organization (WHO) currently estimates the acceptable daily intake of benzyl alcohol to be not more than 5 mg/kg body weight.²⁵

Pharmacokinetics

In adults, benzyl alcohol is oxidized to benzoic acid, conjugated in the liver with glycine, and excreted in the urine as hippuric acid. Preterm babies have a greater ability to metabolize benzyl alcohol to benzoic acid than do term babies, but are unable to convert benzoic acid to hippuric acid, probably because of glycine deficiency. This results in the accumulation of benzoic acid (Fig. 46–1).⁶⁶ A fatal case of metabolic acidosis was reported in a 5-year-old girl who received 2.4 mg/kg/h diazepam preserved with benzyl alcohol for 36 hours to control status epilepticus. Elevated benzoic acid concentrations were identified in serum and urine samples. The estimated daily dosage of benzyl alcohol was 180 mg/kg.^66,107

Neurologic Toxicity

Benzyl alcohol is believed to have a role in the increased frequency of cerebral intraventricular hemorrhages and mortality reported in very-low-birth-weight (VLBW) infants (weight <1,000 g) who received flush solutions preserved with benzyl alcohol.⁸³ An increased incidence of developmental delay and cerebral palsy was also noted in the same VLBW patients, suggesting a secondary damaging effect of benzyl alcohol.¹⁴

There are several case reports of transient paraplegia following the intrathecal or epidural administration of chemotherapeutics or analgesics containing benzyl alcohol as a preservative.^10,41,74,149 The local anesthetic effects are most likely responsible for the immediate paraparesis and limited duration of effects, rather than actual demyelination of nerve roots. In a rat study, lumbosacral dorsal root action potential amplitudes were measured after exposure to 0.9% or 1.5% benzyl alcohol solutions in either 0.9% sodium chloride solution or distilled water.⁷⁴ Rats exposed to all benzyl alcohol solutions for less than one minute had inhibited dorsal root action potentials. This was attributed to the local anesthetic effects of benzyl alcohol as function was 50% to 90% restored after rinsing the nerves with 0.9% sodium chloride solution. Chronic intrathecal exposure to benzyl alcohol 0.9% over 7 days resulted in scattered areas of demyelination and early remyelination. The 1.5% benzyl alcohol solution–exposed dorsal nerve roots showed greater changes, with widespread areas of demyelination and fatty degeneration of nerve fibers.

CHLOROBUTANOL

Chlorobutanol, or chlorbutol (1,1,1-trichloro-2-methyl-2-propanol), is available as volatile, white crystals with an odor of camphor. Chlorobutanol has antibacterial and antifungal properties and is widely used as a preservative in injectable, ophthalmic, otic, and cosmetic preparations at concentrations up to 0.5% (Table 46–4). Chlorobutanol also has mild sedative and local anesthetic properties and was formerly used therapeutically as a sedative–hypnotic.²⁰ The lethal human chlorobutanol dose is estimated to be 50 to 500 mg/kg.¹²⁸

Central Nervous System Toxicity

Chlorobutanol has a chemical structure similar to trichloroethanol (Fig. 72–1), the active metabolite of chloral hydrate, and is believed to exhibit similar pharmacologic properties. Central nervous system depression was reported in a 40-year-old alcoholic man who chronically abused Seducaps (formerly available in Australia and several other countries), a nonprescription hypnotic containing chlorobutanol as the active ingredient.²⁰ On admission to the emergency department, he had drowsiness, dysarthria, slurred speech, and occasional episodes of myoclonic movements. His peak serum chlorobutanol concentration was 100 mcg/mL, decreasing to 48 mcg/mL over 2 weeks, with a half-life of 3 days based on serial declining concentrations. This is similar to human volunteer pharmacokinetic data following oral administration of chlorobutanol demonstrating an elimination half-life of 10.3 ± 1.3 days.¹⁷⁰ His speech abnormality resolved after 4 weeks. Only chlorobutanol was detected in the patient’s urine or serum. In a second case, a possible central nervous system depressant effect from chlorobutanol was suggested in a 19-year-old woman treated with high doses of intravenous morphine preserved with chlorobutanol.⁴⁷ She received approximately 90 mg/h of chlorobutanol for several days. Her peak serum chlorobutanol concentration was 83 mcg/mL, a concentration similar to that in the previous case report;²⁰ however, the coadministration of morphine precludes the effects being attributed to chlorobutanol alone.

Ketamine is neurotoxic when administered intrathecally to animals.^116,117 The potential neurotoxic effects of chlorobutanol as a preservative in ketamine compared with preservative-free ketamine was studied in rabbits.¹¹⁷ Forty rabbits were given 0.3 mL intrathecally of either 1% preservative-free ketamine, 1% ketamine, 0.05% chlorobutanol, or 1% lidocaine as control. The rabbits were observed and hemodynamically monitored for 8 days and then euthanized. Histologic evaluation of the spinal cord as well as for blood–brain barrier lesions was performed. Seven of the 10 rabbits given intrathecal chlorobutanol showed both white and gray matter histologic changes as well as diffuse blood–brain barrier injury. No histologic changes were seen in either ketamine groups or the lidocaine group, and only one rabbit in each ketamine group had blood–brain barrier injury. These results suggest that chlorobutanol should not be administered intrathecally.¹¹⁷

A case series of 5 patients were given intra-arterial papaverine preserved with 0.5% chlorobutanol,¹⁵⁵ which is used to prevent cerebral vasospasm in patients with subarachnoid hemorrhage. Immediately after administration of papaverine in either the left, right, or bilateral anterior cerebral arteries, patients had an acute deterioration in neurologic status. Subsequent brain magnetic resonance imaging identified selective gray matter toxicity in the territories treated with papaverine. Postmortem brain histology in one patient also identified gray matter changes. The authors state the absence of white matter changes is not consistent with ischemic infarction but suggest direct toxic effect of either the papaverine or chlorobutanol. The manufacturer of the papaverine stated that no other reports had been made and the papaverine used came from 2 different lots; therefore, it is unclear if an unidentified independent variable caused these effects, but the authors caution using intra-arterial papaverine in patients with subarachnoid hemorrhage.¹⁵⁵

Ophthalmic Toxicity

Chlorobutanol is a commonly used preservative in ophthalmic preparations and is less toxic to the eye than benzalkonium chloride.¹³² Chlorobutanol increases the permeability of cells by impairing cell membrane structure.¹⁶⁸ An in vitro experiment using corneal epithelial cells harvested from human cadavers demonstrated arrested mitotic activity following chlorobutanol exposure.¹⁶⁸ At the commonly formulated concentration of 0.5%, chlorobutanol can cause eye irritation, most likely due to cellular contraction of epithelial microfilaments, cessation of normal cytokinesis, cell movement, and mitotic activity.⁵⁰ Degeneration of human corneal epithelial cells specifically manifested as membranous blebs, cytoplasmic swelling, and occasional breaks in the external cell membrane has also occurred at this concentration.⁵⁰

LIPIDS

In general, there are 3 types of commercial intravenous lipid drug-delivery systems available: lipid emulsion, liposomal, and lipid complex (Table 46–5). Lipid emulsions are immiscible lipid droplets dispersed in an aqueous phase stabilized by an emulsifier (eg, egg, soy lecithin). Liposomes differ from emulsion lipid droplets in that they are vesicles composed of one or more concentric phospholipid bilayers surrounding an aqueous core. Lipophilic drugs are formulated for intravenous administration by partitioning them into the lipid phase of either an emulsion or liposome. Liposomes are capable of encapsulating hydrophilic xenobiotics within their aqueous core to exploit lipid pharmacokinetic properties.¹⁶² Attaching a therapeutic drug to a lipid to form a lipid complex is another way to take advantage of lipid pharmacokinetics.

Lipid drug-delivery systems are biocompatible because of their similarity to endogenous cell membranes. They are created with stable lipid membranes resistant to hydrolysis or oxidation, to decrease toxicity, and to enhance therapeutic efficacy by altering drug pharmacokinetic and pharmacodynamic parameters. The biodistribution, and the rate of release and metabolism of a drug incorporated in a lipid drug-delivery system, is regulated by the type and concentration of oil and emulsifier used, pH, drug concentration dispersed in the medium, the size of the lipid particle, and the manufacturing process.^134,162 Intravenous formulations are usually isotonic and have a pH of 7 to 8.¹⁶²

The rate of clearance of a lipid drug-delivery system from the blood depends on its physicochemical properties and the molecular weight of the emulsifier. Electrically charged lipid carriers are removed more rapidly than neutral particles.^26,134 Smaller lipid particle size and high-molecular-weight emulsifiers decrease clearance. “Stealth” liposome formulations incorporate a polyethylene glycol coating that prevents rapid detection and clearance of liposomes by the reticuloendothelial system, prolonging circulation time.¹³⁴ Active drug targeting is achieved by conjugating antibodies or vectors to side chains on the emulsifier.^26,134 For a therapeutic drug available in more than one lipid drug-delivery systems (eg, amphotericin B), it is important to note that any change in the lipid formulation can alter the pharmacokinetic, pharmacodynamic, and safety parameters of the drug; consequently, they are not equivalent dosage formulations (Chap. 54).

The physicochemical properties of lipid emulsions not only affect the therapeutic drugs carried by them, but the lipids themselves also have direct pharmacologic effects on the central nervous¹⁸² and immune systems.¹⁰³ Lipid fatty acid mediators affect the membrane receptor channels of N-methyl-D-aspartate (NMDA) receptors, potentiating synaptic transmission.^{120,122,136,165} Dogs given a medium-chain triglyceride emulsion intravenous infusion developed dose-related central nervous system metabolic and neurologic effects, accompanied by electroencephalographic changes consistent with encephalopathy observed when serum octanoate concentration reached 0.5 to 0.9 mM.¹²⁰ In an in vitro model, 3 of 9 lipid emulsions tested (Abbolipid, 20% soya and safflower oil; Intra-lipid, 20% soya oil; and Structolipid, 20% structured triglycerides) demonstrated a dose-related activation of cortical neuronal NMDA receptor channels.¹⁸² The lipid source for all but one (Omegaven, 10% fish oil) of the emulsions tested was made up solely or partially by soya oil. The authors could not explain why the other 6 lipid emulsions did not induce membrane currents. Adequate control for the nonlipid constituent contribution of these emulsions is lacking. In another in vitro study, the same authors found that NMDA-induced neuronal currents are reduced by an unknown factor in the aqueous portion of Abbolipid.¹⁸³ This suggests that lipid emulsions pharmacologically enhance the anesthetic effect of hypnotics such as propofol. The clinical relevance of these studies remains to be determined.

Triglycerides in parenteral nutrition emulsions are implicated in altering the immune system, leading to an increased susceptibility to infection,^58,180 and altering lung function and hemodynamics in patients with acute respiratory distress syndrome.¹⁰³ Phospholipid activation of phospholipase A₂ may be an initiating cause.^58,103,180 However, it is not clear if these immunologic effects are a consequence of factors other than the lipid in the emulsion.

PARABENS

The parabens, or parahydroxybenzoic acids, are a group of compounds widely employed as preservatives in cosmetics, food, and pharmaceuticals because of their bacteriostatic, fungistatic, and antioxidant properties (Table 46–6).¹⁵⁰ A survey conducted by the FDA identified the parabens as the second most common ingredients in cosmetic formulations, with water being the most common.¹⁰⁸ Parabens are often used in combination, because the presence of 2 or more parabens are synergistic.¹⁰⁷ Methylparabens and propylparabens are most commonly used.¹⁵⁰ Pharmaceutical paraben concentrations usually range from 0.1% to 0.3%.¹⁴⁴

Widespread usage of parabens since the 1920s demonstrates a relatively low order of toxicity.¹⁰⁸ However, because of their allergenic potential they are currently considered less suitable for injectable and ophthalmic preparations.¹⁴⁴ Based on long-term animal studies, the WHO has set the total acceptable daily intake of ethylparabens, methylparabens, and propylparabens to be 10 mg/kg body weight.¹⁴⁴ In addition to allergic reactions, parabens have the potential to cause other adverse effects. Bilirubin displacement from albumin-binding sites occurred with administration of methylparaben- and propylparaben-preserved gentamicin when serum parabens concentrations were 3 to 15 mcg/mL.⁴³ Gentamicin alone has no effect on bilirubin displacement.¹⁰⁹ Spermicidal activity was demonstrated in an in vitro study of human semen specimens exposed to local paraben concentrations of 1 to 8 mg/mL.¹⁶⁰ Possible interference with conception and potential adverse effects on fertility were not investigated in this paper. However, other investigators show a direct toxic effect of parabens on sperm mitochondrial function.¹⁶⁶ Sperm require an enormous amount of ATP to adequately fertilize ova.

Concerns have arisen regarding the potential estrogenic and anti-androgenic effects of the parabens and their common metabolite, p-hydroxybenzoic acid. Xenobiotics with these effects are commonly referred to as endocrine disrupting substances. It has been suggested that methylparaben is less toxic than butyl and benzyl-parabens with regard to oxidative stress and resultant cytotoxicity.^44,153 The clinical significance of these effects is not elucidated.^{34,45,141,142,175} There is a focus on the role of parabens and the development of breast cancer due to the estrogenic effects of parabens. In a study of breast tissue following mastectomies of women with primary breast cancer, concentrations of parabens were highest in the axillary region compared to other more medial regions. These authors suggest there may be a role of using cosmetics and deodorants under the arm increasing local estrogenic effects on breast tissue.^12,77

PHENOL

Phenol (carbolic acid, hydroxybenzene, phenylic acid, phenylic alcohol) is a commonly used preservative in injectable medications (Table 46–7). Phenol is a colorless to light pink, caustic liquid with a characteristic odor. When exposed to air and light, phenol turns a red or brown color.⁴⁰ Phenol exerts antimicrobial activity against a wide variety of microorganisms, bacteria, mycobacteria, and some fungi and viruses.⁴⁰ Phenol is well absorbed from the gastrointestinal tract, skin, and mucous membranes and is excreted in the urine as phenyl glucuronide and phenyl sulfate metabolites.⁴⁰ Although there are numerous reports of phenol toxicity following intentional ingestions or unintentional dermal exposures (Chap. 103), adverse reactions to its use as a pharmaceutical excipient are uncommon, most likely because of the small quantities used.⁴⁰

Cutaneous Absorption

Systemic toxicity from cutaneous absorption of phenol is reported. Ventricular tachycardia was observed in an 11-year-old boy following application of a chemical peel solution containing 88% phenol in water and liquid soap. The solution was applied to 15% of his body surface area for the treatment of xeroderma pigmentosum. Immediately following the onset of the ventricular tachycardia, the phenol-treated areas were irrigated, an infusion of 0.9% sodium chloride solution was begun, and 2 intravenous lidocaine boluses were given followed by a lidocaine infusion. The dysrhythmia persisted for 3 hours. The urinary phenol concentration the following day was 58.9 mg/dL.¹⁷³ In a similar case, multifocal premature ventricular contractions were observed in a 10-year-old boy after application of a chemical peeling solution of 40% phenol, 0.8% croton oil in hexachlorophene soap, and water for the treatment of a giant hairy nevus.¹⁸¹ The premature ventricular contractions were refractory to intravenous lidocaine but resolved with intravenous bretylium. No phenol concentrations were obtained to confirm systemic absorption. In a case series of 181 patients undergoing chemical face peeling with phenol-based solutions, 12 demonstrated cardiac dysrhythmias. This occurred more commonly in patients with comorbid diabetes mellitus, hypertension, or treatment with antidepressants.¹⁰¹ Both lidocaine and propranolol treatment and pretreatment are reported to abate these dysrhythmias.¹⁰¹ However, these are anecdotal reports and have not been formally studied sufficiently to support recommending their use.^21,70,169 Despite the risk of toxicity, phenol chemical peels are still used in some cases because phenol penetrates deep into tissues, providing both long-lasting and good cosmetic results. Most commonly, phenol chemical peels are used for deep acne scars and other severe skin disorders.

A 9-year-old girl sustained partial-thickness burns over 17% of her body following the topical administration of Creolin, which contains phenolic compounds. The Creolin was applied by her mother to delouse the patient. Creolin is not intended for human use and is generally used as deodorant cleanser for bathrooms, kennels, and barns. The child had 8 ounces poured onto her hair, which then flowed onto her neck, chest, back, left shoulder, and upper arm. Within minutes she became obtunded, requiring endotracheal intubation by paramedics. In the emergency department, she had sinus tachycardia with brief runs of ventricular tachycardia. She was decontaminated with soap and water, low-molecular-weight polyethylene glycol and ethanol. The patient experienced a mild elevation of hepatic aminotransferases and had dark-green–black urine while in the intensive care unit. This resolved and the patient was extubated and discharged to home on hospital day 4.¹⁷⁷

Drowsiness, respiratory depression, and blue-colored urine were noted in a 6-month-old infant 12 hours after topical application of magenta paint over most of the body for seborrheic eczema.¹⁴⁶ Magenta paint (also known as Castellani paint) was widely used for seborrheic eczema and contained 4% phenol, magenta, boric acid, resorcinol, acetone, and ethanol. Further investigation found that phenol was detected in urine samples of 4 of 16 other infants with seborrheic eczema who had approximately 11% to 15% of their body surface area painted with magenta paint for 2 days.

POLYETHYLENE GLYCOL

Polyethylene glycols (PEGs; Carbowax, Macrogol) include several compounds with varying molecular weights (200–40,000 Da).¹⁴⁰ They are typically available as mixtures designated by a number denoting their average molecular weight. Polyethylene glycols are stable, hydrophilic substances, making them useful excipients for cosmetics and pharmaceuticals of all routes of administration (Table 46–8). Pegylation, a process that modifies the pharmacokinetics of therapeutic liposomes and proteins (eg, peginterferon-α), is the most recent application of PEG. At room temperature, PEGs with molecular weights less than 600 Da are clear, viscous liquids with a slight characteristic odor and bitter taste. Polyethylene glycols with molecular weights higher than 1,000 Da are soluble solids and range in consistency from pastes and waxy flakes to powders.¹⁴⁰ Commercially available products used for bowel-cleansing preparations and whole-bowel irrigation are solutions of PEG 3350 sometimes combined with electrolytes and known as PEG electrolyte lavage solution (Antidotes in Depth: A2).

The solid, high-molecular-weight PEGs are essentially nontoxic. Conversely, low-molecular-weight PEG exposures cause adverse effects similar to the chemically related toxic alcohols ethylene and diethylene glycol (Special Considerations: SC9).³⁰

Pharmacokinetics

High-molecular-weight PEGs (>1,000 Da) are not significantly absorbed from the gastrointestinal tract, but low-molecular-weight PEGs are absorbed when taken orally.^49,156,157 Topical absorption also occurs when PEGs are applied to damaged skin.^24,163 Once in the systemic circulation, PEGs are mainly excreted unchanged in the urine;⁴⁹ however, low-molecular-weight PEGs (eg, PEG 300, PEG 400) are partially metabolized by alcohol dehydrogenase to hydroxyacid and diacid metabolites. The pharmacokinetics of intravenously administered PEG 3350 Da has not been studied; however, it did not appear to have any systemic effects when unintentionally given by this route.¹⁴⁵

Nephrotoxicity

In rats fed various PEGs (200, 300, and 400) in their drinking water for 90 days, a solution of 8% PEG 200 produced renal tubular necrosis in all of the animals, followed by death within 15 days; however, a 4% PEG 200 solution resulted in only 2 of 9 rats dying within 80 days. A 16% PEG 400 solution killed all animals within 13 days; however, both 8% and 4% PEG 400 solutions had no observable effect except for a decrease in kidney weight when compared to control animals.¹⁵⁸ A canine study administering daily high-dose PEG 400 intravenously, up to 8.45 g/kg/day, failed to show serious renal toxicity. However, edema of kidney cells and increased glomerular volume occurred at these dosages, but were reversible.¹⁰⁵ Acute tubular necrosis was reported with oliguria, azotemia, and an anion gap metabolic acidosis following oral and topical exposures to low-molecular-weight PEGs (200 and 300). Acute kidney failure occurred in a 65-year-old man with a history of alcohol abuse and a seizure disorder after ingestion of the contents of a lava lamp containing 13% PEG 200.⁵¹ About 48 hours after admission (approximately 50–72 hours post ingestion), the patient became oliguric with an anion gap metabolic acidosis and acute kidney failure. Blood sampling confirmed traces of the lava lamp fluid; none was detected in the urine. After clinical complications from ethanol withdrawal and aspiration pneumonitis, the patient was discharged 3 months later with residual kidney dysfunction attributed to the PEG component of the lamp contents. Acute tubular necrosis was noted at autopsy of 6 burn patients treated with a topical antibiotic cream in a PEG 300 base.^24,163 Mass spectrometry detected hydroxyacid and diacid metabolites in serum and urine samples. Oxalate crystals were seen in the urine of 2 cases. These effects were reproduced with the topical application of PEG for 7 days to rabbits with full-thickness skin defects.¹⁶³

Neurotoxicity

There are reports of neurologic complications, such as paraplegia and transient bladder paralysis, following intrathecal corticosteroid injections containing 3% PEG as a vehicle.^15,19 In an in vitro experiment, rabbit vagus nerves were exposed to concentrations of PEG 3350 ranging from 3% to 40% for one hour.¹⁵ A total of 3% and 10% PEG had no effect on nerve action potential amplitude or conduction velocity. Doses of 20% and 30% PEG significantly slowed nerve conduction and had varying effects on the amplitudes of action potentials. Forty percent PEG completely abolished action potentials. These changes were reversible and thought to be related to PEG-induced osmotic effects. The administration of PEG 1800 is a potential therapy for spinal cord injury by repairing damaged axons through cellular fusion of damaged cells following the short-term (2 minutes) application, in a guinea pig model. This increases compound action potential conduction.¹⁵⁴ Interestingly, administering a similar dose of PEG continuously for 25 minutes decreased compound action potentials approximately 64% in both damaged and nondamaged isolated mammalian spinal cords, suggesting dose-response toxicity.³⁸

Another study in an ex vivo experiment of isolated injured spinal cords of Wistar rats showed combining hyperthermia (40°C) with various PEGs (400, 1000, and 2000) greatly improved compound action potential recovery compared to these PEGs at 25°C and 37°C. The lower-molecular-weight PEG 400 was the most effective. This occurs through PEG rapidly sealing damaged neuronal membranes.⁹⁵

Fluid, Electrolyte, and Acid–Base Disturbances

Hyperosmolality was reported in 3 patients with burn surface areas ranging from 20% to 56% following repeated applications of Furacin, a topical antibiotic dressing containing 63% PEG 300, 32% PEG 4000, and 5% PEG 1000.²⁴ Polyethylene glycol produces an osmotic effect that is greater than expected for its molecular weight.¹⁵¹ It is theorized that PEG increases osmolality by sequestering water through hydrogen binding, which reduces the availability of water to interact with solutes, thus increasing the chemical and osmotic activity of the solute. Hyperosmolality following the administration of a PEG-containing substance supports systemic PEG absorption.

Two cases of metabolic acidosis were reported following administration of therapeutic dosages of an intravenous nitrofurantoin solution containing PEG 300.¹⁶⁴ Similarly, an otherwise unexplained increased anion gap was reported in 3 patients being treated with a topical PEG-based burn cream.²⁴ Metabolism of the lower-molecular-weight PEGs by alcohol dehydrogenase to hydroxyacid and diacid metabolites explained the metabolic acidosis.⁸⁰

PROPYLENE GLYCOL

Propylene glycol (PG), or 1,2-propanediol, is a clear, colorless, odorless, sweet, viscous liquid employed in numerous pharmaceuticals (Table 46–9), foods, and cosmetics. Propylene glycol is used as a solvent and preservative with antiseptic properties similar to ethanol. The WHO has set the daily allowable intake of PG at a maximum of 25 mg/kg,¹⁸⁴ or 1.75 g/day for a 70-kg person.

Pharmacokinetics

Propylene glycol is rapidly absorbed from the gastrointestinal tract following oral administration and has a volume of distribution of approximately 0.6 L/kg.^118,160 When applied to intact epidermis, the absorption of PG is minimal. Percutaneous absorption occurs following application to damaged skin (eg, extensive burn surface areas). Approximately 12% to 45% of PG is excreted unchanged in the urine,⁴⁸ the remainder is hepatically metabolized sequentially by alcohol dehydrogenase to lactaldehyde, which is metabolized further by aldehyde dehydrogenase to lactic acid. Lactic acid is also formed by another metabolite, methylglyoxal.¹³¹ Lactic acid is reduced to pyruvate catalyzed by lactate dehydrogenase.¹³¹ The terminal half-life of propylene glycol is reported to be between 1.4 and 5.6 hours in adults and as long as 16.9 hours in neonates.^48,162

Cardiovascular Toxicity

Intravenous preparations of phenytoin contain 40% PG to facilitate the dissolution of phenytoin. Nine years after intravenous phenytoin became available, several deaths were attributed to the rapid administration of phenytoin used for the treatment of cardiac dysrhythmias.^65,171,195

Cardiovascular effects reported in these cases included hypotension, bradycardia, widening of the QRS complex, increased amplitude of T waves with occasional inversions, and transient ST segment elevations. Studies in cats¹¹⁰ and calves⁷¹ confirmed PG as the cardiotoxin. Bradycardia and depression of atrial conduction were not observed in cats pretreated with atropine, or in those with vagotomy following rapid intravenous infusion of PG, suggesting that these effects are vagally mediated.¹¹⁰ Amplification of the QRS complex was noted in these same pretreated cats, also suggesting a direct cardiotoxic effect of PG. Similar results were reported in calves pretreated with atropine that received oxytetracycline in a PG vehicle.⁷¹

Neurotoxicity

Infants appear to have a decreased ability to clear PG when compared with older children and adults.¹¹¹ An increased frequency of seizures was reported in low-birth-weight infants who received 3 grams of PG daily in a parenteral multivitamin preparation.¹¹¹ Seizures developed in an 11-year-old boy receiving long-term oral therapy with vitamin D dissolved in PG.⁸ Serum, electrolytes, and blood glucose were normal. Seizures abated after the product was discontinued. Propylene glycol possesses inebriating properties similar to ethanol. Central nervous system depression was reported following an intentional oral ingestion of a PG-containing product.¹¹⁸

A formerly available oral solution protease inhibitor amprenavir (Agenerase) had a black-box warning added to the product information for its high PG (550 mg/mL) vehicle content.¹⁴⁸ The recommended daily dosage of amprenavir supplied 1,650 mg/kg/day of PG. A 61-year-old man experienced visual hallucinations, disorientation, tinnitus, and vertigo after receiving a 750-mg dose (474 mg/kg PG) of amprenavir solution.⁹¹

Ototoxicity

Otic preparations contain up to 94% PG in solutions and 10% in suspensions as part of their vehicles.⁵⁴ In animal studies, application of high concentrations of PG (>10%) to the middle ear produced hearing impairment^124,125,178 and morphologic changes, including tympanic membrane perforation, middle ear adhesions, and cholesteatoma.^124,125,190 Although the effects of PG in the human middle ear have not been studied, all medications applied to the external ear canal are contraindicated in patients with perforated tympanic membranes.

Fluid, Electrolyte, and Acid–Base Disturbances

Patients receiving continuous or large intermittent quantities of medications containing PG develop high PG concentrations, particularly those with renal or hepatic insufficiency.^29,48 Propylene glycol–induced electrolyte and metabolic disturbances are evidenced by hyperosmolarity, and an elevated osmolar gap attributed to the osmotically active properties of PG. In most cases, both an elevated anion gap and elevated lactate concentration are present. Metabolic acidosis and hyperlactatemia result from PG metabolism.²⁸ These adverse effects are typically reported with intravenous preparations such as lorazepam,^7,76,86,192 diazepam,¹⁸⁶ etomidate,¹⁷⁴ nitroglycerin,⁴⁸ pentobarbital,¹²¹ pediatric multivitamins,⁶⁷ and topical silver sulfadiazine.^13,55,98,185

Systemic absorption of PG from topical application of silver sulfadiazine cream⁵⁵ resulted in hyperosmolality in patients with burn surface areas greater than 35% of their body.^13,55,98,185 In one study, 9 of 15 burn patients had osmolar gaps (>12) after application of the cream.⁹⁸ Similarly, a 3-year-old boy with extensive second- and third-degree burns over 60% body surface area developed a persistent elevated osmolar gap and metabolic acidosis with elevated lactate for approximately 20 days following application of silver sulfadiazine cream. Other potential etiologies were excluded. The osmolar gap and metabolic acidosis both resolved within 24 hours of discontinuing silver sulfadiazine.¹⁸⁵

Hyperosmolarity occurred in 5 infants receiving a parenteral multivitamin that provided a daily PG dose of 3 g.⁶⁷ After 12 days, one premature infant had a PG concentration of 930 mg/dL and an osmolar gap of 136. Anion gap and lactic acid concentrations were normal. Eleven intubated children aged 1 to 15 months who were receiving continuous lorazepam infusions over 3 to 14 days accumulated serum PG concentrations of 17 to 226 mg/dL; however, increases in osmolar gap or serum lactate concentrations were not significantly different from baseline.³⁵ This was attributed to normal renal function and the low cumulative PG doses received (mean, 60 g).

A 72-year-old man died following the ingestion of the contents of a cold/hot therapy gel pack, which contained more than 99% of propylene glycol. Six hours after ingestion, he had an elevated osmolar gap accompanied by acute kidney injury. At 12 hours, he became comatose and an anion gap acidosis with a normal L-lactate concentration. He was found to have a very elevated D-lactate concentration, which was attributed to the PG ingestion. This patient had an extensive medical history and the course was complicated by multiorgan failure and death.⁹⁴

Several small studies have found a strong correlation between elevated PG concentrations and increased osmolar gap measurements in critically ill patients receiving intravenous lorazepam and/or diazepam.^{7,187,191,192} An osmolar gap greater than 10 is suggested as a marker for potential PG toxicity.¹⁹¹ An osmolar gap of 20 corresponds to a serum PG concentration of approximately 48 mg/dL.⁷ These authors provide a formula to estimate PG concentrations based on the osmolar gap: [PG] = (–82.1 + [osmol gap × 6.5]). This equation should be used cautiously, as larger, more comprehensive studies are needed to validate it. In addition, elevated anion gap measurements and lactate concentrations occur. As PG toxicity mimics sepsis in these critically ill patients, sepsis should be excluded as the etiology of increased lactate, hypotension, and worsening renal function when considering PG toxicity. Hemodialysis is effective in removing PG and correcting acidemia. The administration of fomepizole should be considered with elevated PG concentrations to prevent further metabolism to lactic acid.^135,194

Nephrotoxicity

Human proximal tubular cells exposed in vitro to PG concentrations of 500 to 2,000 mg/dL exhibited significant cellular injury and membrane damage within 15 minutes of exposure.¹²⁷ Repeated exposure for up to 6 days produced dose-dependent toxic effects at lower concentrations (76, 190, and 380 mg/dL).¹²⁶

The chronic administration of PG contributes to proximal tubular cell damage and subsequent decreased kidney function. In a retrospective study of 8 patients who developed elevations in serum creatinine concentration while receiving continuous lorazepam infusions, serum creatinine rose within 3 to 60 days (median, 9 days).¹⁹² The magnitude of serum creatinine rise was found to correlate with the serum PG concentration and duration of infusion. Serum creatinine decreased within 3 days of discontinuing the infusion. Patients with chronic kidney disease are at greater risk for accumulating PG because 45% of PG is eliminated unchanged by the kidneys;⁴⁸ the remainder is metabolized by the liver. Caution should be used when prolonged administration of a PG-containing medication is necessary in the presence of renal or hepatic dysfunction.¹²⁷

There are reports of propylene glycol–induced renal tubular necrosis. Daily PG vehicle dosages of 11 to 90 g/day over 14 days was associated with rising serum creatinine concentrations (0.7–2.1 mg/dL), elevated serum lactate concentrations, osmolar and anion gaps, and a serum PG concentration of 21 mg/dL.¹⁹³ Urine sediment analysis revealed numerous granular, muddy-brown-colored casts and no eosinophils, suggesting an acute renal tubular necrosis. Kidney biopsy and electron microscopy showed extensive dilation of the proximal renal tubules, with swollen epithelial cells and mitochondria. Numerous vacuoles containing debris were also noted. A kidney biopsy of another case with a serum PG concentration of 30 mg/dL showed disrupted brush borders of the proximal renal tubules after a sudden rise in serum creatinine concentration (3.1 mg/dL), nonoliguric kidney failure, and metabolic acidosis. This was attributed to an average daily PG dose of 70 g for 17 days.⁷⁶

SORBITOL

Sorbitol (D-glucitol) is widely used in the pharmaceutical industry as a sweetener, moistening agent, and as a diluent (Table 46–10). Sorbitol occurs naturally in the ripe berries of many fruits, trees, and plants, and was first isolated in 1872 from the berries of the European mountain ash (Sorbus aucuparia).¹²⁹ It is particularly useful in chewable tablets because of its pleasant taste. In addition, it is widely used by the food industry in chewing gums, dietetic candies, foods, and enteral nutrition formulations. Sorbitol is approximately 50% to 60% as sweet as sucrose.¹²⁹

Pharmacokinetics

Unlike sucrose, sorbitol is not readily fermented by oral microorganisms and is poorly absorbed from the gastrointestinal tract. Any absorbed sorbitol is metabolized in the liver to fructose and glucose.¹²⁹ Sorbitol has a caloric value of 4 kcal/g and is better tolerated by diabetics than sucrose; however, because some of it is metabolized to glucose, it is not unconditionally safe for people with diabetes and is obviously not “dietetic.”¹²⁹

There is a concern of potentially fatal toxicity for individuals with hereditary fructose intolerance (HFI) receiving sorbitol-containing xenobiotics.⁵⁶ Hereditary fructose intolerance is an autosomal recessive disorder caused by a deficiency of fructose-1,6-bisphosphonate aldolase in the liver, kidney, cortex, and small intestine.⁹³ This results in the accumulation of fructose-1-phosphate, which prevents glycogen breakdown and glucose synthesis causing hypoglycemia. The prevalence of HFI is most commonly reported to be one in 20,000 persons, but can range between one in 11,000 and one in 100,000.^2,90,93

There are reports of death following the prolonged administration of sorbitol, fructose, or sucrose in individuals with HFI.^39,152 Dietary exclusion of fructose, sucrose, and sorbitol prevents the adverse effects. This condition should not be confused with the more common disorder of dietary fructose intolerance, which is caused by a defect in the glucose-transport protein 5 system. This defect leads to the breakdown of fructose to carbon dioxide, hydrogen, and short-chain fatty acids by colonic bacteria, resulting in abdominal pain and bloating.¹⁰² This latter syndrome is a potential cause of chronic abdominal pain in childhood.⁶⁹ Dietary fructose intolerance symptoms are minimized by limiting sorbitol, fructose, and sucrose in the diet.

Gastrointestinal Toxicity

In large dosages, sorbitol causes abdominal cramping, bloating, flatulence, vomiting, and diarrhea. Sorbitol exerts its cathartic effects by its osmotic properties, resulting in fluid shifts within the gastrointestinal tract. Iatrogenic osmotic diarrhea is reported following administration of many different liquid medication formulations containing sorbitol.^82,112 In a human volunteer study, 42 healthy adults ingested 10 g of a sorbitol solution. Sorbitol intolerance was detected in up to 55% of participants.⁸⁹ One theoretical explanation for why all participants did not experience the gastrointestinal adverse effects is unrecognized dietary fructose intolerance. Diarrhea resulting from sorbitol-containing medications is common and often overlooked as a possible etiology.^56,82 Ingestion of large quantities of sorbitol (>20 g/day in adults) is not recommended (Antidotes in Depth: A2).¹²⁹

THIMEROSAL

Thimerosal (Merthiolate, Mercurothiolate), or sodium ethylmercurithiosalicylate, is an organic mercury compound that is approximately 49% elemental mercury (Hg°) by weight.^147,179 It is metabolized to ethylmercury and thiosalicylate. Thimerosal has a wide spectrum of antibacterial activity at concentrations ranging from 0.01% to 0.1%; however, higher concentrations are sometimes also used topically as an antiseptic.^123,147 Thimerosal has been widely used as a preservative since the 1930s in contact lens solutions, biologics, and vaccines, particularly those in multidose containers (Table 46–11). The use of thimerosal, which is necessary for the production process of some vaccines (eg, pertussis, influenza), leaves trace amounts in the final product.¹¹ High-dose thimerosal exposure resulted in neurotoxicity and nephrotoxicity. Although concerns exist regarding infant exposure to low-dose thimerosal through vaccinations and its effects on neurodevelopment, including possible links to causes of autism,¹⁸ these concerns are unfounded (Chap. 95).⁴⁶

Because specific guidelines for ethylmercury exposure have not been developed, regulatory guidelines for dietary methylmercury exposure were applied to monitor ethylmercury exposure from injected thimerosal-containing vaccines. Methylmercury is a similar, but more toxic, organic mercury compound (Chap. 95). Maximum daily recommended methylmercury exposures range from 0.1 mcg Hg/kg (US Environmental Protection Agency {EPA}) to 0.47 mcg Hg/kg (WHO).^3,32,37

An FDA review of thimerosal-containing vaccines revealed that some infants, depending on the immunization schedule, vaccine formulations, and infant’s weight, might be exceeding the EPA exposure limit of 0.1 mcg Hg/kg/day for methylmercury. Over the first 6 months of life, a total cumulative dose of up to 187.5 mcg Hg total from thimerosal-containing vaccines was possible. The US Public Health Service and the American Academy of Pediatrics jointly responded by recommending the preemptive reduction or removal of thimerosal from vaccines wherever possible.^3,31 The WHO and European regulatory bodies have similar recommendations.⁵⁷ To date, thimerosal has been removed from most US-licensed immunoglobulin products. All vaccines routinely recommended for children younger than 7 years are either thimerosal-free or contain only trace amounts (<0.5 mcg Hg/dose), with the exception of some inactivated influenza vaccines. Multidose vials requiring thimerosal preservative remain important for immunization programs in developing countries because of a lack of ability to consistently refrigerate vaccines. Although efforts continue to eliminate all sources of mercury exposure, complete elimination of thimerosal from all vaccines is unlikely in the near future.¹¹ When a thimerosal-containing vaccine is the only alternative, the benefits of vaccination far exceed any theoretical risk of mercury toxicity.¹³⁰

Prior to thimerosal use in pharmaceuticals, evidence for its safety and effectiveness was provided in several animal species and in 22 humans.¹³⁹ Only limited data exist on infant mercury exposure from thimerosal-containing vaccines. Clinical studies that assess the effects of thimerosal exposure on neurodevelopment and renal and immunologic function are lacking. Based on a comprehensive review of epidemiologic data from the United States,^{36,60,63,167,179} Denmark,^114,115 Sweden,¹⁶¹ and the United Kingdom,^6,81 the Institute of Medicine’s Immunization Safety Review Committee,¹³⁰ the Global Advisory Committee on Vaccine Safety,¹⁸⁹ and the European Agency for the Evaluation of Medicinal Products⁵² have all concluded that no causal relationship exists between thimerosal-containing vaccines and autism. Continued surveillance of autistic spectrum disorders as thimerosal use declines will be conducted to evaluate any associated trends.

Pharmacokinetics

Limited pharmacokinetic data exist for thimerosal and ethylmercury. Once absorbed, thimerosal breaks down to form ethylmercury and thiosalicylate. Some ethylmercury further decomposes into inorganic mercury in the blood, and the remainder distributes into kidney and, to a lesser extent, brain tissue.^114,115 Because of its longer organic chain, ethylmercury is less stable and decomposes more rapidly than methylmercury, leaving less ethylmercury available to enter kidney and brain tissue.¹¹⁴ Ethylmercury crosses the blood–brain barrier by passive diffusion.¹¹⁵ Intracellular ethylmercury decomposes to inorganic mercury, which accumulates in kidney and brain tissues.¹¹⁵ The half-life of thimerosal is estimated to be about 18 days.¹¹⁵ Thimerosal is eliminated in the feces as inorganic mercury (Chap. 95).¹³⁸

Mercury or Thimerosal Toxicity

Oral Administration

A case report described a 44-year-old man who ingested 5 g (83 mg/kg) of thimerosal in a suicide attempt; within 15 minutes, he began vomiting spontaneously. Gastric lavage was performed and chelation therapy begun with dimercaptopropane sulfonate. Gastroscopy revealed a hemorrhagic gastritis. Acute kidney injury was noted on the day of admission and persisted for 40 days. Four days after admission, the patient developed fever and a maculopapular exanthem attributed to thimerosal. The patient also developed an autonomic and ascending peripheral polyneuropathy that persisted for 13 days. Chelation therapy was continued for a total of 50 days with dimercaptopropane sulfonate followed by succimer. Elevated blood and urine mercury concentrations persisted for more than 140 days. The patient was discharged 148 days following the ingestion with only sensory defects in his toes. No other neurologic sequelae were noted.¹³⁷

Oral absorption of thimerosal resulted in the fatal poisoning of an 18-month-old girl from the intra-otic instillation of a solution containing 0.1% thimerosal and 0.14% sodium borate. Tympanostomy tubes placed one year earlier allowed the irrigation solution to flow through the auditory tube into the nasopharynx, and subsequently to be swallowed and absorbed through the oral mucosa and gastrointestinal tract. A total of 1.2 L of solution (500 mg Hg) was instilled over a 4-week period, resulting in severe mercury poisoning. Four days after admission, the serum mercury concentration was 163 mcg/dL. The patient also received 1.7 g of boric acid. It is unclear what contribution, if any, the boric acid made to the serum mercury concentration. Chelation therapy with N-acetyl-D-penicillamine was initiated on day 51. Despite increased urinary mercury concentrations following administration of the N-acetyl-D-penicillamine, her neurologic function and blood mercury concentrations remained unchanged. The child died 3 months after admission. An autopsy was not performed.¹⁴⁷

Intramuscular Administration

Urine mercury concentrations of 26 patients with hypogammaglobinemia, who received weekly intramuscular immunoglobulin G (IgG) replacement therapy preserved with 0.01% thimerosal, were studied. The dosages of IgG ranged from 25 to 50 mg/kg, containing 0.6 to 1.2 mg of mercury per dose.⁷² The total estimated dose of mercury administered ranged from 4 to 734 mg over a period of 6 months to 17 years. Urine mercury concentrations were elevated in 19 patients, ranging from 31 to 75 mcg/L; however, no patients had clinical evidence of chronic mercury toxicity.⁷²

Six cases of severe mercury poisoning resulting in 4 deaths were reported following the intramuscular administration of chloramphenicol preserved with thimerosal. A manufacturing error produced vials containing 510 mg of thimerosal (250 mg Hg) instead of 0.51 mg per vial. Two adults received 4 and 5.5 g of mercury each and 4 children received 0.2 to 1.8 g each. All 6 patients had extensive tissue necrosis at the site of injection. Fever, altered mental status, slurred speech, and ataxia were noted. Autopsy identified widespread degeneration and necrosis of the renal tubules; however, creatine kinase concentrations were not reported, so pigment-induced nephrotoxicity cannot be excluded. Elevated mercury concentrations were found in the injection site tissues, and in the kidneys, livers, and brains.⁹

Topical Administration

Thirteen infants were exposed to 9 to 48 topical applications of a 0.1% thimerosal tincture for the treatment of exomphalos. Analysis for elevated mercury concentrations was performed in 10 of 13 infants who unexpectedly died. Mercury concentrations were determined in various tissues from 6 of the infants. Mean tissue concentrations in fresh samples of liver, kidney, spleen, and heart ranged from 5,152 to 11,330 ppb, suggesting percutaneous absorption from these repeated topical applications.⁵³

Ophthalmic Administration

Nine patients undergoing keratoplasty were exposed to a contact lens stored in a solution containing 0.002% thimerosal.¹⁸⁸ After 4 hours, the lens was removed and mercury concentrations of the aqueous humor and excised corneal tissues were determined. Mercury concentrations were elevated in both aqueous humor (range, 20–46 ng/mL higher) and corneal tissues (range, 0.6–14 ng/mL higher) as compared with eyes that had not been fitted with contact lenses. Only residual amounts of mercury remained on the contact lenses after 4 hours of wear. The authors noted that although the aqueous humor concentrations were in the same range as those measured in 10 patients with vision loss from systemic mercury poisoning (11–104 ng/mL), adverse effects did not occur.

A possible drug interaction between orally administered tetracyclines and thimerosal was reported to result in acute, varying degrees of eye irritation in contact lens wearers using thimerosal-containing contact lens solutions who started treatment with tetracycline.⁴²

SUMMARY

• The benefits of pharmaceutical excipients include improved xenobiotic solubility, stability, and palatability, antimicrobial activity, the availability of various dosage forms, the provision of products with long-term storage, and the availability of multiple-dose packaging. Although excipients are essential and effective, they are suggested to possess no pharmacologic or toxicologic properties but are actually responsible for severe—and sometimes fatal—adverse effects.

• The toxicity of pharmaceutical excipients should be evaluated in patients requiring high doses or prolonged administration of any medication containing excipients, particularly those additives known to have toxicities.

• Under circumstances in which there is no option but to continue treating a patient with a particular xenobiotic, switching to a preservative-free product, or to another brand without the offending excipient, should obviate the need for discontinuation of an effective xenobiotic. In addition to inherent toxicities, many excipients are also responsible for allergic reactions.

• In the majority of cases, pharmaceutical excipients are safe and effective, and their benefits far exceed their potential for adverse effects when properly administered.

REFERENCES

INTRODUCTION

HISTORY AND EPIDEMIOLOGY

Insulin first became available for use in 1922 after Banting and Best successfully treated diabetic patients using pancreatic extracts.¹³ In an attempt to more closely simulate physiologic conditions, additional “designer” insulins with unique kinetic properties have been developed, including a rapid-acting insulin that mimics baseline (before meals) insulin secretion, known as lispro.^81,157 The hypoglycemic activity of a sulfonamide derivative used for typhoid fever was noted during World War II, and verified later in animals.⁹⁰ The sulfonylureas in use today are chemical modifications of that original sulfonamide compound. In the mid-1960s, the first-generation sulfonylureas were widely used. Newer second-generation drugs differ primarily in their potency.

Although insulin is widely used for treating diabetes mellitus, oral hypoglycemic exposures are more commonly reported to poison control centers than are insulin exposures, based on 15 years of data from 2001 to 2015 (Chap. 130). In an older review of 1,418 medication-related cases of hypoglycemia, sulfonylureas (especially the long-acting chlorpropamide and glyburide) alone or with a second hypoglycemic accounted for the largest percentage of cases (63%).¹⁴⁷ Only 18 of the sulfonylurea cases in this series involved intentional overdose. However, hypoglycemia is reported in as many as 20% of patients using sulfonylureas.⁷³ In a study of 99,628 emergency hospitalizations for adverse drug events in adults older than 65 years, 14% were due to insulin and 11% were due to oral hypoglycemics. The majority (95%) of the hospitalizations related to insulin and oral hypoglycemics were due to hypoglycemia.²⁴ Other causes of hypoglycemia are listed in Table 47–1.

The biguanides metformin and phenformin were developed as derivatives of Galega officinalis, the French lilac, recognized in medieval Europe as a treatment for diabetes mellitus.¹¹ Phenformin was removed from the US market in 1977 because of its association with life-threatening metabolic acidosis with an elevated lactate concentration.

Development of the α-glucosidase inhibitors began in the 1960s when an α-amylase inhibitor was isolated from wheat flour.¹⁴³ Acarbose was discovered more than 10 years later and approved for use in the United States in 1995. Troglitazone and repaglinide were approved for use in the United States in 1997. The US Food and Drug Administration subsequently directed the manufacturer of troglitazone to withdraw the product from the US market in 2000 because of associated liver toxicity. Exenatide, a synthetic form of a compound found in the saliva of the Gila monster, is an incretin mimetic. Incretin is a hormone that stimulates insulin secretion in response to meals. Liraglutide, dulaglutide, and albiglutide are synthetic analogs of human incretin. Other newer xenobiotics include the gliptins, the amylin analog pramlintide, and the sodium glucose co-transporter 2 (SGLT2) inhibitors.

PHARMACOLOGY

Insulin is synthesized as a precursor polypeptide in the β islet cells of the pancreas. Proteolytic processing results in the formation of proinsulin, which is cleaved, giving rise to C-peptide and insulin itself, a double-chain molecule containing 51 amino acid residues. Glucose concentration plays a major role in the regulation of insulin release.¹³² Glucose is phosphorylated after transport into the β islet cell of the pancreas. Further metabolism of glucose-6-phosphate results in the formation of ATP. Adenosine triphosphate inhibition of the K⁺ channel results in cell depolarization, inward calcium flux, and insulin release. After release, insulin binds to specific receptors on cell surfaces in insulin-sensitive tissues, particularly the hepatic, muscle, and adipose cells. The action of insulin on these cells involves various phosphorylation and dephosphorylation reactions.

Figure 47–1 depicts the chemical structures of noninsulin antidiabetics. The sulfonylureas stimulate the β cells of the pancreas to release insulin and are often referred to as insulin secretagogues. They are ineffective in type 1 diabetes mellitus that results from islet cell destruction (Fig. 47–2). This stimulatory effect diminishes with chronic therapy. All the sulfonylureas bind to high-affinity receptor sensors on the pancreatic β cell membrane, resulting in closure of K⁺ channels.^49,60,61 Inhibition of potassium efflux mimics the effect of naturally elevated intracellular ATP and results in insulin release. High-affinity sulfonylurea receptors are also present within pancreatic β cells and are postulated to be either located on granular membranes or part of a regulatory exocytosis kinase. Binding to these receptors promotes exocytosis by direct interaction with secretory machinery not involving closure of the plasma membrane K⁺ channels.^49,60,61

The linkage of 2 guanidine molecules forms the biguanides. Metformin is an oral biguanide approved for treatment of type 2 diabetes mellitus. It acts by several mechanisms, the most important of which involves the inhibition of mitochondrial complex I, which subsequently leads to activation of adenosine monophosphate activated protein kinase (AMPK). Adenosine monophosphate activated protein kinase activation is postulated to mediate antihyperglycemic effects, although alternative signaling pathways may be involved.²²

Metformin has other actions that are independent of AMPK activation and result in beneficial as well as undesirable side effects.⁷¹ These include induction of mitochondrial stress, suppression of glucagon signaling, and activation of autophagy. Metformin inhibits glycerophosphate dehydrogenase, which is important in the glycerophosphate shuttle and results in inhibition of gluconeogenesis. Alteration of intestinal microbiota by metformin occurs and plays a role in the uptake and utilization of glucose by the intestines. Metformin also increases levels of glucagonlike peptide-1 (GLP-1) and induces GLP-1 receptors. Glucagonlike peptide-1 is an incretin that is released in response to an oral glucose load; it enhances the release of insulin, delays gastric emptying, and reduces food intake. Glucose lowering by metformin can be summarized to occur by a combination of effects, which include inhibition of gluconeogenesis, decreased hepatic glucose output, enhanced peripheral glucose uptake, decreased fatty acid oxidation, and increased intestinal use of glucose.

Acarbose and miglitol are oligosaccharides that inhibit α-glucosidase enzymes such as glucoamylase, sucrase, and maltase in the brush border of the small intestine. As a result, postprandial elevations in blood glucose concentrations are blunted.¹⁷¹ Delayed gastric emptying is another proposed mechanism for the antihyperglycemic effect of these oligosaccharides.¹³⁶

Insulin resistance in patients with type 2 diabetes mellitus occurs because of secretion of biologically defective insulin molecules, circulating insulin antibodies, or target tissue defects in insulin action.¹²⁰ The thiazolidinedione derivatives decrease insulin resistance by potentiating insulin sensitivity in the liver, adipose tissue, and skeletal muscle. Uptake of glucose into adipose tissue and skeletal muscle is enhanced, whereas hepatic glucose production is reduced.^21,72

Repaglinide and nateglinide are representatives of the meglitinide class that also bind to K⁺ channels on pancreatic cells, resulting in increased insulin secretion.¹³⁹ Compared to the sulfonylureas, the hypoglycemic effects of the meglitinides are shorter in duration.

Exenatide, liraglutide, dulaglutide, and albiglutide are structurally similar to glucagonlike peptide-1 (GLP-1), an incretin that is released in response to an oral glucose load. Glucagonlike peptide-1 enhances the release of insulin, delays gastric emptying, and reduces food intake. Glucagonlike ­peptide-1 is metabolized very rapidly, rendering it therapeutically ineffective. The GLP-1 analogs have much longer half-lives, rendering them useful in the treatment of type 2 diabetes mellitus.¹⁶⁵ Sitagliptin, saxagliptin, linagliptin, and alogliptin inhibit dipeptidyl peptidase-4 (DPP-4), the enzyme responsible for the inactivation of GLP-1.³ Pramlintide is an amylin analog. Amylin is produced in the pancreatic β cell and acts in conjunction with insulin to inhibit gastric emptying, decrease postprandial glucagon secretion, and promote satiety.⁹¹

The kidneys contribute significantly to glucose homeostasis. The SGLT2 inhibitors are located mainly in the proximal tubule, and account for up to 90% of glucose absorption. The SGLT2 inhibitors, such as canagliflozin, block the reabsorption of glucose, with subsequent excretion into the urine.⁵⁰

PHARMACOKINETICS AND TOXICOKINETICS

Pharmacokinetic parameters of the hypoglycemics are given in Tables 47–2 and 47–3. The onset and duration of action in therapeutic doses vary considerably among preparations. Insulin overdose usually occurs after administration by the subcutaneous or intramuscular route. As might be predicted based on slow onset and prolonged duration of action of some of the preparations, insulin overdose may result in delayed and prolonged hypoglycemia. However, hypoglycemia occurs with short-acting forms as well because of some unusual toxicokinetic features. Some of these unpredicted responses are caused by a depot effect following intramuscular or subcutaneous administration, and poor absorption is further potentiated by the poor perfusion that occurs during periods of hypoglycemia.^112,156 Because there are a finite number of insulin receptors, insulin overdoses of varying amounts probably are equivalent in terms of the degree of resultant hypoglycemia once receptor saturation occurs, but not in terms of its duration. A comparison can be made with the current treatment of diabetic ketoacidosis, in which lower doses of insulin are as effective as the higher doses used in the past.⁸⁰

Many of the sulfonylureas have long durations of action, which explains the unusually long period of hypoglycemia that occurs in both therapeutic use and overdose. Second-generation sulfonylureas (glimepiride, glipizide, glyburide) have half-lives that approach 24 hours and are characterized by substantial fecal excretion of the parent molecule. The sulfonylureas frequently cause hypoglycemia (Table 47–2). As with insulin, the sulfonylureas cause delayed onset of hypoglycemia following overdose.^121,134 The reason for the potential delayed onset of effects with sulfonylureas cannot be simply explained by known kinetic principles but could be related to counterregulatory mechanisms that fail over time.

Metformin metabolism is negligible, and the majority of an absorbed dose is actively secreted in the urine unchanged. With renal dysfunction, the clearance of metformin declines in proportion to the decline of creatinine clearance.⁶⁴ Plasma protein binding is negligible.¹² Repaglinide and nateglinide are prandial glucose regulators characterized by rapid onset and short duration of action. Overdose experience and toxicokinetic data for repaglinide are lacking. It is not clear whether hypoglycemia would be prolonged or delayed in onset following overdose. In one case of nateglinide overdose, hypoglycemia occurred early and was short lived.¹¹⁴ Exenatide is available in an extended-release form, and dulaglutide and albiglutide are formulated such that these drugs are administered once weekly. The gliptins have long half-lives and cause pronounced inhibition of DPP-4, resulting in a long duration of action (approximately 24 hours) after therapeutic doses.⁸⁸ Pramlintide has a half-life of approximately 40 minutes when used in therapeutic doses.³² The SGLT2 inhibitors have durations of action that allow for once-daily dosing.⁴³ The GLP-1 analogs and amylin analogs are large polypeptides; they represent the 2 classes of noninsulin antidiabetics that are only administered parenterally (subcutaneously).

PATHOPHYSIOLOGY OF HYPOGLYCEMIA

To varying degrees, the antidiabetics all produce a nearly identical clinical condition of hypoglycemia. The etiologies of hypoglycemia are divided into 3 general categories:⁵⁴ physiologic or pathophysiologic conditions (Table 47–1), direct effects of various hypoglycemics (Tables 47–2 and 47–3), and potentiation of hypoglycemics by interactions with other xenobiotics (Table 47–4).

Hypoglycemia usually results in decreased insulin secretion, with production of alternate fuels, particularly ketones. Ketone production occurs as a result of fatty acid metabolism.⁸⁵ Nonketotic hypoglycemia can occur with hyperinsulinemia such as from an insulinoma.¹²³

Central nervous system (CNS) symptoms predominate in hypoglycemia because the brain relies almost entirely on glucose as an energy source. However, during prolonged starvation, the brain can utilize ketones derived from free fatty acids. In contrast to the brain, other major organs such as the heart, liver, and skeletal muscle often function during hypoglycemia because they can use various fuel sources, particularly free fatty acids.¹⁴⁹

Emphasis on tighter glucose control as a means of preventing microvascular effects carries with it an increased risk for hypoglycemia.^39,40 Regulation of glucose control to near-normal glucose concentrations, the characteristics of each individual’s awareness of hypoglycemia, and the individual counterregulatory mechanisms define the frequency and intensity of hypoglycemia.¹⁴⁸ The Diabetic Control and Complications Trial research group reported 62 episodes of blood glucose concentration less than 50 mg/dL with CNS manifestations requiring assistance for every 100 patient-years in patients undergoing an intensive insulin therapy regimen. This was in comparison to a conventional therapy group, which had 19 such episodes per 100 patient-years.^39,40 The intensive therapy group received 3 or more insulin injections per day or used a pump in an effort to achieve a glucose concentration as close to normal as possible, whereas the conventional therapy group received one or 2 daily insulin injections.

There has also been an emphasis in tighter control of glucose in critically ill patients, even in the absence of known diabetes mellitus. Hyperglycemia occurs in critically ill patients due to several mechanisms and is associated with increased mortality in patients with a variety of medical and surgical diagnoses.⁸² In a study of critically ill surgical patients, tight control of glucose was associated with decreased morbidity and mortality.¹⁶⁹ However, such benefits in other studies are not always clearly replicated, and significant hypoglycemia is reported.^52,168

The issue of tight glycemic control has to be analyzed with respect to various factors, such as old versus young age, outpatient versus critically ill (including surgical vs nonsurgical patients), microvascular versus macrovascular complications, and diabetic versus nondiabetic patients. Although benefits of tight glycemic control are demonstrable, these are often outweighed by the deleterious effects of hypoglycemia.^{1,56,95,96,103}

The autonomic nervous system regulates glucagon and insulin secretion, glycogenolysis, lipolysis, and gluconeogenesis. β-Adrenergic antagonists affect all of these mechanisms and can result in hypoglycemia. β-Adrenergic antagonist–induced hypoglycemia is a particular risk⁶⁵ secondary to increased insulin half-life and reduced renal gluconeogenesis in the presence of renal dysfunction.¹²⁴ In addition, the clinical presentation of hypoglycemia may be muted when β-adrenergic antagonists are present because the expected autonomic responses of tachycardia, diaphoresis, and anxiety may not occur. Although this is assumed to be true, an adverse effect on hypoglycemic awareness could not be demonstrated in healthy volunteers given metoprolol, atenolol, and propranolol.⁷⁸

The concept of hypoglycemia-associated autonomic failure in diabetes mellitus is well described.³⁷ Recurrent episodes of hypoglycemia result in autonomic failure by causing defective compensatory glucose counterregulation and subsequent hypoglycemic unawareness. As glucose concentrations fall, normal sensing mechanisms result in decreased insulin secretion and increased glucagon and epinephrine secretion. These counterregulatory defenses against hypoglycemia are defective in most people with type 1 diabetes mellitus and in many with type 2 diabetes mellitus. A comparison study was conducted in diabetes patients with hypoglycemia awareness and hypoglycemia unawareness, using positron emission tomography with radiolabeled fluorodeoxyglucose to measure changes in brain glucose kinetics.⁴⁷ Alterations in cortical responses to hypoglycemia were found in those with unawareness of hypoglycemia. This group also had significantly less epinephrine output in response to hypoglycemia.

Although various nonantidiabetic xenobiotics can cause hypoglycemia (Table 47–1), salicylates and ethanol are particularly notable for their unintended hypoglycemic effects. The mechanism of ethanol-induced hypoglycemia is discussed in Chap. 76. Salicylate inhibition of prostaglandin synthesis in the β cell of the pancreas is postulated to result in enhanced insulin secretion.¹⁴ Salicylates may also cause hypoglycemia by mechanisms that do not involve enhanced insulin secretion. Hypoglycorrhachia in the presence of peripheral euglycemia was demonstrated in a mouse study,¹⁶⁴ and it correlates with reports in humans of reversal of euglycemic delirium after administration of dextrose.⁸⁴

CLINICAL MANIFESTATIONS

Hypoglycemia and its secondary effects on the CNS (neuroglycopenia) are the most common adverse effects related to insulin and the sulfonylureas. It is essential to remember that hypoglycemia is primarily a clinical, not a numerical, disorder. Clinical hypoglycemia is the failure to maintain a CSF glucose concentration that prevents signs or symptoms of glucose deficiency. The clinical presentations of patients with hypoglycemia are extremely variable and it must be excluded as an etiology of any neuropsychiatric abnormality, whether persistent or transient, focal or generalized. The cerebral cortex usually is most severely affected. These findings are categorized below:¹³¹

• Delirium with subdued, confused, or agitated behavior.

• Coma with multifocal brain stem abnormalities, including decerebrate spasms and respiratory abnormalities, with preservation of the oculocephalic (doll’s eyes), oculovestibular (cold-caloric), and pupillary responses.

• Focal neurologic deficits simulating a cerebrovascular accident (CVA) with or without the presence of coma. There are numerous reports^7,153 and series^145,170 of patients with focal neurologic deficits.

• Seizure activity (single, multiple, or status epilepticus).

These neuropsychiatric symptoms are usually reversible if the hypoglycemia is corrected promptly. The morbidity resulting from undiagnosed hypoglycemia is related partly to the etiology and partly to the duration and severity of the hypoglycemia. The etiologies of hypoglycemia encompass both severe diseases such as fulminant hepatic failure and benign problems such as a missed meal by an insulin-requiring diabetes patient or a patient on an oral antidiabetic. The literature with regard to outcome is difficult to interpret. Although a study of 125 emergency department (ED) cases of symptomatic hypoglycemia reported an 11% mortality rate,⁹⁹ only one death (0.8%) was attributed directly to hypoglycemia. In that same study, 9 patients (7.2%) presented with seizures (focal in one case), 3 patients (2.4%) presented with hemiparesis, and 4 survivors (3.2%) suffered residual neurologic deficits. In one tertiary care medical center, 1.2% of all admitted patients had numerical hypoglycemia (defined as a glucose concentration <50 mg/dL). The overall mortality was 27% for this group of 94 patients.⁵⁴ The longer and more profound the hypoglycemic episode, the more likely permanent CNS damage will occur.¹⁰

No absolute criteria available from the physical examination or history distinguish one form of metabolic coma from another. Moreover, all of the findings classically associated with hypoglycemia, such as tremor, sweating, tachycardia, confusion, coma, and seizures, are frequently absent.⁶⁹ The glycemic threshold is the glucose concentration below which clinical manifestations develop, a threshold that is host variable. In one study, the mean glycemic threshold for hypoglycemic symptoms was 78 mg/dL in patients with poorly controlled type 1 diabetes compared to 53 mg/dL in those ­without the disease.¹⁷

Many patients with well-controlled type 1 diabetes are unaware of hypoglycemia. It appears that even in the presence of numerical hypoglycemia, diabetes patients with near-normal glycosylated hemoglobin concentrations maintain near-normal glucose uptake by the brain, thereby preserving cerebral metabolism and limiting the response of counterregulatory hormones. The result of this limited response is unawareness of hypoglycemia.^17,19 A threshold is likely achieved below which the glucose concentration is inadequate, but this may be a concentration so close to that causing serious neuroglycopenia that patients have limited opportunity for corrective action.¹⁹ Hypoglycemia unawareness is most likely in diabetes patients with chronic use of hypoglycemics because of hypoglycemia-associated autonomic failure.³⁷ Acute ingestion of hypoglycemics in nondiabetic patients typically would cause more classic signs and symptoms.

Sinus tachycardia, atrial fibrillation, and ventricular premature contractions are the most common dysrhythmias associated with hypoglycemia.^89,119 An outpouring of catecholamines in a response to hypoglycemia, transient electrolyte abnormalities, and underlying heart disease appear to be the most likely etiologies. Based on their mechanisms of action, both insulin and the sulfonylureas are expected to promote the shift of potassium into cells, and hypokalemia after insulin overdose is well documented.^9,156 Other cardiovascular manifestations including acute coronary syndromes rarely are the sole manifestations of hypoglycemia.^16,46,128 Increased release of catecholamines during hypoglycemia increases myocardial oxygen demand and decreases supply by causing coronary vasoconstriction, especially in patients with diabetes who are at increased risk of atherosclerosis. Hypoglycemia leads to ­catecholamine release which also causes cardiac repolarization abnormalities that contribute to atrial and ventricular dysrhythmias.¹⁶⁷

Hypothermia often occurs in hypoglycemic patients.^55,77,158 If present, hypothermia usually is mild (90°F–95°F {32°C –35°C}), unless coexisting conditions such as environmental exposure, infection, head injury, or hypothyroidism are present. In a study comparing 2 groups of patients with depressed mental status, hypothermia was almost exclusively limited to the hypoglycemic patients; of these patients, 53% with demonstrated hypoglycemia showed hypothermia.¹⁵⁸ The central hypothalamic response to hypoglycemia stimulated by the sympathetic nervous system infrequently “overshoots” normal temperatures, resulting in hyperthermia following recovery.³³

Besides decreasing glucose concentrations, the hypoglycemics can produce a number of adverse effects, both in overdose and in therapeutic doses. Older sulfonylureas, predominantly chlorpropamide, cause a syndrome of inappropriate antidiuretic hormone secretion⁷⁵ and disulfiram-ethanol reactions.¹²⁹ These adverse effects are exceedingly uncommon with the newer second-generation sulfonylureas.

Hypoglycemia is at times delayed until 18 hours after lente insulin overdose,¹¹² persist for up to 53 hours after subcutaneous insulin glargine overdose,²³ and is reported to persist up to 6 days after ultralente insulin overdose.¹⁰⁰ Death after insulin overdose cannot be correlated directly with either the dose or preparation. Some patients have died with doses estimated in the hundreds of units, whereas others have survived doses in the thousands of units.¹⁴⁴ Mortality and morbidity likely correlate better with delay in recognition of the problem, duration of symptoms, onset of therapy, and type of complications, as opposed to the absolute degree of hypoglycemia or persistence of elevated insulin concentrations. In a retrospective study of insulin overdose, 7 of 17 cases (41%) developed recurrent hypoglycemia between 5 and 39 hours after overdose despite oral feeding and intravenous dextrose infusion ranging from 5 to 17 g of dextrose per hour.¹⁵⁶ A significant correlation was noted between the amount of insulin injected and the total amount of dextrose used for treatment. This is surprising because one would expect that the amount of dextrose infused would be independent of the dose of insulin following saturation of insulin receptors. Another study of intentional insulin overdose in 25 patients found no correlation between the insulin dose and the amount of administered dextrose.¹⁰⁹

Insulin is a large polypeptide structure. Although one would expect significant degradation by gut enzymes if ingested, there is a report of hypoglycemia to as low as 25 mg/dL after ingestion of 3,000 units of a combination of aspart, lispro, and glargine.¹⁵⁹

In a retrospective review of 40 patients with sulfonylurea overdoses, the time from ingestion to the onset of hypoglycemia, when known, was variable.¹²¹ The longest delay was 21 hours after ingestion of glyburide and 48 hours after ingestion of chlorpropamide. In a retrospective poison control center review of 93 cases of sulfonylurea exposures in children, 25 patients (27%) developed hypoglycemia, with a time of onset ranging from 0.5 to 16 hours and a mean of 4.3 hours.¹³⁴ In a prospective poison control center study of sulfonylurea exposures in children, 56 of 185 (30%) patients developed hypoglycemia, with a time of onset ranging from 1 to 21 hours and a mean of 5.3 hours.¹⁵² Single-tablet ingestions of chlorpropamide 250 mg, glipizide 5 mg, and glyburide 2.5 mg result in hypoglycemia in young children,¹³⁴ and the hypoglycemia is often delayed.¹⁶⁰ Hypoglycemia did not occur until 45 hours after ingestion of a 10-mg extended-release glipizide tablet in a 6-year-old child.¹²⁵

Hypoglycemia is reported in a few cases of metformin overdose. In a 2-case series, metabolic acidosis with an elevated lactate concentration was evident on initial presentation.¹⁶² Hypoglycemia was present initially in one of the cases but did not develop until 7 hours later in the second case. There is no mention of co-ingestants in these cases. Hypoglycemia is reported in a 15-year-old who presented 2 hours after ingestion of metformin and quetiapine.⁴ Extensive laboratory investigation excluded insulin and sulfonylureas as causal. Hypoglycemia and metabolic acidosis with an elevated lactate concentration are reported in a case of overdose with metformin, atenolol, and diclofenac.⁶⁶ In a series of 36 metformin overdose cases, 2 patients developed hypoglycemia, and co-ingestants were not felt to be contributory.¹⁰⁷ Insufficient evidence supports the concept that metformin-associated hypoglycemia can develop in a patient who does not have metabolic acidosis. Because many patients receiving metformin also take sulfonylureas, hypoglycemia should be anticipated after overdose.

Despite limited overdose data, acarbose and miglitol are not likely to cause hypoglycemia based on their mechanism of action of inhibiting α-glucosidase. The most common adverse effects associated with therapeutic use of these xenobiotics are gastrointestinal, including nausea, bloating, abdominal pain, flatulence, and diarrhea. Elevated aminotransferase concentrations were noted after use of acarbose in clinical trials.⁷⁰ Most patients were asymptomatic, and the aminotransferase concentrations returned to normal after the drug was discontinued. The therapeutic use of acarbose in some cases reportedly led to hepatotoxicity that resolved after the drug was discontinued.^8,30

Hypoglycemia would not be expected after thiazolidinedione overdose, despite limited overdose data. The most serious adverse effect of troglitazone is the development of liver toxicity with therapeutic doses, which in some cases was severe enough to require liver transplantation.^62,117 Liver toxicity related to therapeutic use of rosiglitazone^6,57 and pioglitazone is also reported.^98,101 Therapeutic use of pioglitazone and rosiglitazone may precipitate fluid retention in patients with congestive heart failure.¹¹⁶ A meta-analysis concluded that rosiglitazone therapy is associated with an increased risk of myocardial infarction and death from cardiovascular causes.¹¹⁸

Hypoglycemia should be anticipated after repaglinide and nateglinide ingestion. Hypoglycemia is reported after nateglinide overdose,¹¹⁴ and a case of intentionally self-induced hypoglycemia secondary to repaglinide is reported.⁶⁸ Hypoglycemia developed 14 hours postingestion in a case of repaglinide overdose.⁵¹

The GLP-1 analogs and gliptins are reported to have pharmacologic effects that are glucose-dependent.⁵⁸ This does not necessarily apply to the overdose situation, and hypoglycemia is uncommonly reported in such settings. “Severe hypoglycemia” was reported in a phase III clinical trial after inadvertent administration of 10 times the normal dose of exenatide. The specific glucose concentration is not noted in the report.²⁷ Injection of 90 mcg of exenatide in a 40-year-old and 1,800 mcg in a 46-year-old resulted in gastrointestinal symptoms but no hypoglycemia.^35,83 As a result of a transcription error, a 67-year-old received 18 mg liraglutide daily over a 7-month period. He had several hospitalizations for evaluation of nausea, vomiting, and diarrhea but never developed hypoglycemia.¹⁵ A 33-year-old injected 72 mg of liraglutide and developed nausea and vomiting, but no hypoglycemia.¹¹³

An 86-year-old ingested 1,700 mg of sitagliptin in a suicide attempt and did not develop hypoglycemia during an overnight hospitalization.⁵⁹ A multicenter poison control center study reviewed 650 cases of gliptin exposure, with exclusion of co-ingestant cases.¹⁴¹ Three of these reportedly developed hypoglycemia, although the description of these cases is not clear with respect to associated neuroglycopenia. The glucose concentrations were based on bedside testing, not on specimens sent to the laboratory.

There are limited overdose data with regard to amylin analogs. Pramlintide is used therapeutically in conjunction with insulin, and hypoglycemia in this setting is more likely than with insulin use alone.⁹¹

Therapeutic use of SGLT2 inhibitors results in euglycemic ketoacidosis.¹⁶¹ This is one of the newest classes of drugs used for diabetes, with limited overdose data. Based on their pharmacology, inhibition of renal glucose reabsorption would not be expected to occur to a degree that would result in hypoglycemia.³⁸

DIAGNOSTIC TESTING

Suspicion of hypoglycemia, particularly neuroglycopenia, is important in any patient with an abnormal neurologic examination. The most frequent reasons for failure to diagnose hypoglycemia and mismanaging patients are the erroneous conclusions that the patient is not hypoglycemic but rather is psychotic, epileptic, experiencing a CVA, or intoxicated because of an “odor of alcohol” on the breath (Chap. 76). Compounding the problem of misdiagnosis is the erroneous assumption that a single bolus of 0.5 to 1 g/kg of hypertonic dextrose will always be sufficient.

Plasma glucose concentrations, although generally accurate, are time consuming, so treatment should not be delayed pending the results of laboratory testing. Glucose reagent strip testing can be performed at the bedside. The sensitivity of these tests for detecting hypoglycemia is excellent, but these tests are not perfect. Several interfering substances may cause false elevations of bedside glucose reagent strip concentrations, including maltodextrin, acetaminophen, bilirubin, triglyceride, and uric acid.^48,79 Bedside glucose testing is discussed in more detail in Antidotes in Depth: A8.

Diagnostic studies other than determinations of glucose concentrations are often indicated, depending on the clinical situation. In some instances, determination of serum ethanol concentration is helpful in confirming alcohol as a contributing or sole etiologic factor. Analysis of BUN, creatinine, and eGFR (estimated glomerular filtration rate) indicates the presence of renal dysfunction as a causative factor of hypoglycemia. This commonly occurs in patients with diabetes taking insulin, who often develop renal dysfunction after they have had the disease for several years, with insulin half-life increasing as eGFR decreases. Measures of hepatic function will be a clue to liver disease as a cause of hypoglycemia, although liver disease will also be evident on physical examination. Seizures are commonly associated with hypoglycemia, but other studies, such as electrolytes, and imaging of the brain, are indicated if doubt about the etiology exists.

In the majority of overdose cases, laboratory testing for specific antidiabetics is not helpful. Exceptions might include malicious, surreptitious, or unintentional overdoses (discussed in the next section). Metformin concentrations vary and do not necessarily correlate with the clinical condition.^5,86,87

For known diabetic patients in whom intentional, unintentional or malicious overdose is not suspected, the clinician must search diligently for the cause of hypoglycemia. Sometimes it is as simple as a missed meal in an insulin or oral antidiabetic user or an unusually strenuous exercise routine, but in many cases the cause cannot be clearly defined. Numerous medical conditions, as well as a variety of xenobiotics, are commonly involved (Table 47–1), and diagnostic testing must be individualized for each episode depending on the clinical suspicion. Diagnosing the etiology as “idiopathic” is never acceptable.

EVALUATION OF MALICIOUS, SURREPTITIOUS, OR UNINTENTIONAL INSULIN OVERDOSE

The physical examination will often provide helpful clues to the evaluation of a suspected malicious, surreptitious, or unintentional insulin overdose. A meticulous search will reveal a site that is erythematous, hemorrhagic, atypically boggy in nature, or even painful if the subcutaneous or intramuscular injection of insulin was particularly large. A simple unexplained needle puncture mark in the appropriate clinical setting suggests insulin injection.

An understanding of how the β cells of the pancreas secrete insulin in response to glucose concentrations in the blood is essential to understanding the investigation of fasting hypoglycemia.³⁶ When the plasma glucose concentration is less than 45 mg/dL, insulin secretion should be almost completely suppressed, so plasma insulin concentrations should be minimal or absent.¹³⁰ Moreover, insulin is secreted as proinsulin, which is cleaved in vivo to form insulin (a double-stranded peptide), and C-peptide, which are released into the blood in equimolar quantities. Insulin is biologically active, whereas proinsulin has limited activity, and C-peptide has no activity. Although insulin is normally cleared during hepatic transit, C-peptide is not. For this reason, C-peptide can be utilized as a quantitative marker of endogenous insulin secretion. In contrast, commercially available exogenous human insulin does not contain C-peptide fragments (Table 47–5). When plasma glucose concentration falls to hypoglycemic concentrations (usually <60 mg/dL), insulin concentration should fall to less than 6 µU/mL. If hypoglycemia is caused by exogenous insulin administration, plasma C-peptide concentrations should be less than 0.2 nmol/L in the presence of insulin concentrations that are substantially higher than insulin concentrations resulting from an insulinoma. With insulinoma, insulin concentrations generally are greater than 6 µU/mL in the presence of hypoglycemia. Insulinoma results in elevations of both C-peptide and insulin concentrations. Sulfonylurea overdose is expected to have similar effects, but concentrations in reported cases of sulfonylurea-induced hypoglycemia vary considerably.⁴⁴ In the face of uncertainty, sulfonylurea concentrations are available from reference laboratories. Animal insulin can be distinguished from human insulin by high-performance liquid chromatography.⁶³ However, this technique has limited use because of the virtually exclusive use of recombinant insulin at present.

In summary, patients with exogenous insulin-induced hypoglycemia will have high insulin concentrations, the presence of insulin-binding antibodies (if chronic insulin users),⁵³ and low C-peptide concentrations. Those who have taken sulfonylureas will have high insulin concentrations, absent insulin-binding antibodies, high C-peptide concentrations, and the presence of urinary sulfonylurea metabolites (Table 47–5). The issues of evidence collection that are appropriate to document malicious or surreptitious use of insulin successfully are described (Chap. 138).⁹⁴

MANAGEMENT

Treatment centers on the correction of hypoglycemia and the anticipation that hypoglycemia may recur. Symptomatic patients with hypoglycemia require immediate treatment with 0.5 to 1 g/kg hypertonic intravenous dextrose in the form of D₅₀W in adults, D₂₅W in children, and D₁₀W in neonates. Occasionally, patients require a larger dose to achieve an initial response. If hypoglycemia is suspected but not confirmed, as in the absence of rapid reagent strip availability or when such readings are “borderline,” dextrose should be administered. Theoretical risks are associated with use of hypertonic dextrose in the setting of acute cerebral ischemia, but failure to rapidly correct hypoglycemia will lead to deleterious neurologic effects. Appropriate emergency and toxicologic uses of hypertonic dextrose are covered in detail in Antidotes in Depth: A8.

We recommend not to use glucagon as an antihypoglycemic except in the uncommon situation where intravenous access cannot be obtained. Glucagon has a delay to onset of action and is ineffective in patients with depleted glycogen stores, as in the malnourished elderly, cancer patients, or alcoholics. Glucagon also stimulates insulin release from the pancreas, which is reported to lead to prolonged hypoglycemia in settings such as sulfonylurea ingestion and insulinoma.¹⁶³

Numerous studies have evaluated approaches for treating insulin reactions with carbohydrates in tablet, solution, or gel forms in a well-defined diabetic population.¹⁵⁰ None of these forms is appropriate for the undifferentiated, possibly hypoglycemic patient if intravenous access is available, but should be carried by all diabetic patients.

A common occurrence involves symptomatic hypoglycemic patients who receive intravenous dextrose in the prehospital setting and subsequently refuse transport to the hospital. The authors of a retrospective review of 571 paramedic runs involving hypoglycemic patients concluded that out-of-hospital treatment of hypoglycemic diabetic patients is safe and effective even when transport is refused.¹⁵¹ However, of the 159 patients who agreed to hospital transport, 40% were admitted. The admitted group was older than those released from the ED. The admission rate for transported patients on oral hypoglycemics was higher than those on insulin. The reasons for admission are not otherwise detailed. The authors of a prospective study involving 132 hypoglycemic diabetic patients who refused transport after therapy concluded that most such patients have good short-term outcome, but they still encouraged transport because of the risk of recurrent hypoglycemia.¹⁰⁸ One patient died in each of these 2 studies. A prospective study in 35 patients with 38 hypoglycemic events related to insulin use concluded that most patients were successfully treated in the prehospital setting without transport.⁹² However, 2 patients developed recurrent hypoglycemia that they treated themselves, and one of these patients required placement in a long-term care facility for posthypoglycemic encephalopathy. We therefore recommend that all hypoglycemic patients should be transported to the ED.

Emesis, lavage, and catharsis are of limited benefit in the management of patients who overdose on hypoglycemics. The extensive affinity between chlorpropamide, tolazamide, tolbutamide, glyburide, glipizide, and activated charcoal is demonstrated in vitro.⁷⁶ The affinities ranged from 0.45 to 0.52 g/g activated charcoal at pH 7.5 and were higher at pH 4.9. Single-dose activated charcoal should be beneficial in the management of these overdoses. Although affinity studies are lacking for the other oral hypoglycemics, their chemical characteristics are such that single-dose activated charcoal is expected to be beneficial for these overdoses as well.

In patients who overdose on insulin, case reports describe the use of surgical excision of the injection site.^29,93,104 However, this technique has not been studied in a systematic fashion and we recommend against excision. Needle aspiration of a depot site is less invasive and can be performed, but intravenous dextrose should be sufficient in most cases.

MAINTAINING EUGLYCEMIA AFTER INITIAL CONTROL

After the patient is awake and alert, further therapy depends on the pharmacokinetics of the xenobiotic involved and pancreatic islet cell function. Some patients, particularly those with prolonged hypoglycemia, have persistent altered mental status despite euglycemia. Whether the event was unintentional or intentional with suicidal or homicidal intent must be determined. One problem associated with dextrose administration occurs in individuals who can produce insulin via glucose-stimulated insulin release (nondiabetic patients and those with type 2 diabetes mellitus), placing them at substantial risk for recurrent hypoglycemia. This complication occurs with insulin overdose but is particularly problematic with overdoses of sulfonylurea or meglitinide because these hypoglycemics stimulate insulin release. Treatment with hypertonic dextrose solutions can be expected to result in dramatic yet only transient increases in glucose concentrations, with a subsequent fall in plasma glucose concentration, possibly back to hypoglycemic concentrations.

For patients with diabetes who unintentionally inject an excessive amount of insulin and are not neuroglycopenic, feeding (carbohydrates and proteins) should be initiated and intravenous access maintained while avoiding routine dextrose infusion. In the event of recurrent symptomatic hypoglycemia, a hypertonic intravenous dextrose bolus followed by an infusion of D₅W is recommended. Overdose in the setting of suicidal or homicidal intent likely involves significant quantities of insulin. Nondiabetic patients are particularly prone to significant hypoglycemia because they lack insulin resistance. Feeding is recommended with a target of maintaining glucose concentrations in the 100- to 150-mg/dL range using a hypertonic dextrose infusion (D₁₀W or greater) as needed based on serial measurements of plasma glucose concentrations.

Central venous lines should be used when a D₂₀W or greater infusion is instituted, because hypertonic dextrose solutions are venous irritants and sclerosing to the vasculature. The presence of glycosuria is not an adequate indicator of euglycemia; frequent serial blood or reagent strip glucose concentrations should be obtained. The appropriate timing of glucose monitoring varies depending on the clinical situation. Mental status must be observed. As a rough guide, glucose monitoring every 1 to 2 hours after initial control is reasonable, with subsequent spacing of the intervals to once every 4 to 6 hours. Phosphate concentrations should be monitored because dextrose loading may lead to hypophosphatemia.¹¹⁰ Potassium concentrations should be monitored because dextrose administration leads to hypokalemia in nondiabetic patients and rarely hyperkalemia in patients with impaired insulin secretion.³⁴ The duration of serial sampling depends on the stability of the patient, the underlying metabolic disorders, the extent of overdose, and the rate of improvement. When the patient begins to eat an adequate diet and the initial hypoglycemia is controlled, the plasma glucose concentration will rise, and the concentration and rate of dextrose infusion can be tapered. Many patients may actually develop significant hyperglycemia.

The therapeutic approach differs for patients who overdose on sulfonylureas or meglitinides. After initial control of hypoglycemia with hypertonic dextrose, feeding is again recommended. Intravenous access is necessary, but routine continuous dextrose infusion should be avoided. As with insulin overdose, frequent monitoring of glucose concentrations and mental status is critical. We recommend use of octreotide for recurrent hypoglycemia in this setting because of the significant risk of glucose-stimulated insulin release.

Octreotide, a semisynthetic long-acting analog of somatostatin with an intravenous half-life of 72 minutes, inhibits glucose-stimulated β cell insulin release via receptors coupled to G proteins on β islet cells.¹⁸ Somatostatin is present in diverse tissues such as the hypothalamus, pancreas, and gastrointestinal tract. It alters the secretion of growth hormone and thyroid-stimulating hormone, gastrointestinal secretions, and the endocrine pancreas (glucagon and insulin).^137,138 Octreotide was compared to intravenous hypertonic dextrose and to diazoxide and concomitant dextrose in normal subjects brought to hypoglycemia using glipizide.¹⁸ Fewer episodes of recurrent hypoglycemia occurred after octreotide therapy, and overall dextrose requirements were lower than in the dextrose-alone and dextrose-plus-diazoxide groups. Several successful clinical experiences with octreotide are reported with quinine-induced hypoglycemia resulting from malaria therapy,¹²⁷ insulinoma,⁶⁷ nesidioblastosis of infancy,⁴¹ hypoglycemia related to therapeutic use of gliclazide,²⁰ and tolbutamide overdose.¹⁸ In a retrospective study of 9 patients with hypoglycemia resulting from either glyburide or glipizide, octreotide effectively reduced the risk of recurrent hypoglycemia.¹⁰⁵ Despite limited evidence, octreotide would be expected to be effective in the treatment of recurrent hypoglycemia following overdose of all classes of noninsulin antidiabetics.

Octreotide appears to be relatively free of serious side effects. The most likely adverse effects are injection-site discomfort if it is administered subcutaneously and gastrointestinal symptoms such as nausea, bloating, diarrhea, and constipation.⁹⁷ Our suggested adult octreotide dose is 50 mcg subcutaneously every 6 hours (Antidotes in Depth: A9). The patient should be monitored for 12 to 24 hours after the last dose of octreotide. This observation will ensure that recurrence of hypoglycemia does not occur. Like octreotide, diazoxide is reportedly effective in patients with refractory sulfonylurea-induced hypoglycemia.^74,121 However, octreotide is the preferred therapy because diazoxide has the potential to cause hypotension, cerebrovascular compromise and myocardial infarction.¹⁰⁵

ADMITTING PATIENTS TO THE HOSPITAL

The decision to admit a patient is complex, but several guidelines can be followed. Admission is usually required for hypoglycemia related to sulfonylureas, ethanol, starvation, hepatic failure, and renal dysfunction, and for hypoglycemia of unknown etiology. The decision to admit a patient often depends on finding an etiology for hypoglycemia, particularly in the setting of insulin use. In most cases, if a diabetic patient on therapeutic doses of insulin develops hypoglycemia after a missed meal, the patient can be discharged after a 4- to 6-hour observation period during which the individual eats a meal and remains asymptomatic with no evidence of hypoglycemia. Patients receiving therapeutic doses of insulin usually require inpatient evaluation of recurrent and unexplained hypoglycemic episodes. Patients with hypoglycemia after unintentional overdose with long-acting insulin should generally be admitted. Hospitalization is usually recommended after unintentional overdose with ultrashort-acting, short-acting, or intermediate-acting insulin if hypoglycemia is persistent or recurrent during a 4- to 6-hour observation period in the ED. Many factors are responsible for unintentional insulin overdose, such as patient error because of impaired vision, syringe structure, and prescription error when undiagnosed hospital admission is recommended. Admission is usually indicated for any patient, regardless of serum glucose concentration or presence or absence of symptoms, who intentionally overdoses on a sulfonylurea or any form of injected insulin, because delayed, profound, and protracted hypoglycemia is expected. Although intravenous insulin overdose is expected to result in more immediate symptoms, experience with this scenario is limited. Admission in this setting is generally advised unless short-acting insulin is involved. Glargine is expected to act like regular insulin when administered intravenously. It is long-acting when administered subcutaneously because of the formation of microprecipitates that are absorbed slowly. Hypoglycemia related to sulfonylurea use in any setting including a missed meal or a viral illness typically necessitates hospitalization.²⁵

Patients with possible intentionally self-induced hypoglycemia should generally be admitted. Intentionally self-induced hypoglycemia is most commonly recognized by members of the medical profession. Administration of insulin to a nondiabetic child is a form of child abuse or an attempt at homicide⁴⁵ (Chap. 31). Children who have been given an inappropriate dose of insulin, as well as any patient who may be a victim of attempted homicide, should generally be admitted.

A 4- to 6-hour observation period is recommended after metformin overdose. Further observation or hospital admission is usually not required for patients who remain asymptomatic during this period with no evidence of metabolic acidosis, elevated lactate concentration, renal dysfunction, emesis, or hypoglycemia. Patients who overdose on α-glucosidase inhibitors are not expected to have delayed or serious systemic toxicity, and routine medical admission is unnecessary. There are limited data regarding the risk of hypoglycemia and other adverse events after thiazolidinedione ingestion. Based on the mechanism of action and existing clinical experience, hypoglycemia is possible but uncommon after unintentional thiazolidinedione overdose. Delayed onset of hypoglycemia or other serious clinical manifestations is unlikely. A 4- to 6-hour observation period after thiazolidinedione overdose is recommended. Meglitinides are expected to behave pharmacologically like sulfonylureas. For this reason alone, hospital admission after meglitinide overdose is generally advisable, even when the patient is asymptomatic. A 4- to 6-hour observation period is recommended after overdose of GLP-1 analogs, gliptins, amylin analogs, and SGLT2 inhibitors. Delayed onset of clinical manifestations is unlikely.

Children who unintentionally ingest one or more sulfonylurea tablets should generally be hospitalized for 24 hours. Although this recommendation may be controversial and some authors suggest shorter observation periods²⁶ or even home monitoring in some cases,¹⁴⁰ we believe that delayed hypoglycemic effects of sulfonylurea ingestion in children are well documented^134,152,160 and convincing enough to support admission in most cases. Asymptomatic children with single-tablet exposures to sulfonylureas are best managed without prophylactic intravenous dextrose, which could contribute to delayed onset of hypoglycemia.²⁶ Elevations in glucose concentrations stimulate insulin release by the pancreas. Such patients instead are best managed by early feeding, frequent checks of glucose concentrations, and observation of mental status.

METABOLIC ACIDOSIS WITH AN ELEVATED LACTATE CONCENTRATION

Throughout this chapter, the term metabolic acidosis with an elevated lactate concentration (MALA) is used rather than metformin-associated lactic acidosis or metformin-induced metabolic acidosis. The biochemical and pathophysiologic processes involving lactate are complex, but a few points are worth summarizing. An elevated lactate concentration occurs in various diseases and can be present in the absence of acidosis. The production of lactic acid does not result in a net increase in hydrogen ion concentration unless there is associated impairment of oxidative metabolism. Impaired oxidative metabolism leads to an increase in hydrogen ion production through the hydrolysis of ATP.¹¹¹ In a pig overdose model, metformin caused metabolic acidosis with an elevated lactate concentration, mitochondrial dysfunction, and inhibition of oxygen consumption. Infusion of lactic acid alone did not cause inhibition of oxygen consumption (Fig. 47-3).¹³³

The biguanides are uniquely associated with the occurrence of MALA. Phenformin causes lactic acid production by several mechanisms, including interference with cellular aerobic metabolism and subsequent enhanced anaerobic metabolism. Phenformin suppresses hepatic gluconeogenesis from pyruvate and causes a decrease in hepatocellular pH, resulting in decreased lactate consumption and decreased hepatic lactate uptake. Metformin-associated metabolic acidosis with an elevated lactate concentration occurs 20 times less commonly than that occurring with phenformin. In isolated perfused rat liver, metformin inhibits both hepatic lactate uptake and conversion of lactate to glucose.¹³⁵ Metabolic acidosis with an elevated lactate concentration related to metformin usually occurs in the presence of an underlying condition, particularly renal dysfunction.³¹ In this setting, increased tissue burden of metformin, which is renally eliminated, probably occurs. Other risk factors include cardiorespiratory insufficiency, septicemia, liver disease, a history of MALA, advanced age, alcohol abuse, and use of radiologic contrast media.^12,31 Iodinated contrast material induces acute renal dysfunction, leading to accumulation of metformin and subsequent risk of development of MALA. However, the risk of developing MALA after contrast administration is low in patients taking metformin who have normal renal function and no other risk factors.^102,115 The American College of Radiology recommends and we concur on the temporary discontinuation of metformin for 48 hours after administration of intravenous contrast media if the eGFR is less than 30 mL/min/1.73 m².²

Metabolic acidosis with an elevated lactate concentration occurs after acute metformin overdose^{87,106,122,162} but is uncommon. In one case,¹⁰⁶ metabolic acidosis was not diagnosed until 14 hours after metformin overdose. The patient had early symptoms of repeated vomiting at 1 hour postingestion. The presence of early gastrointestinal symptoms prompted prolonged observation, allowing for the diagnosis of delayed-onset metabolic acidosis. Metabolic acidosis with an elevated lactate concentration occurred in 2 of 65 adult metformin exposures reported to a poison control center¹⁵⁴ and was not reported in a poison control center series of 55 pediatric metformin exposures.¹⁵⁵ In a retrospective review of 398 cases of acute metformin exposures from 2 poison control centers, MALA in 9.1% of single product overdose and in 0.7% of polypharmacy overdoses.¹⁷² In a retrospective review, MALA was reported in 11 of 36 metformin overdose cases.¹⁰⁷ Acid–base data are detailed in only a few of these cases described as having severe toxicity; the lowest blood pH was 7.16.

A systematic review from the Cochrane Library concluded that therapeutic use of metformin is not associated with an increased risk of MALA compared with other antidiabetic treatments if no contraindications are present.¹⁴² This erroneous conclusion was based on a review of prospective comparative trials and observational cohort studies. However, the risk of MALA in the setting of overdose or renal dysfunction was not assessed. Although MALA after overdose is not common, it does occur with sufficient frequency to require vigilance on the part of the treating physician. In this setting, metformin is clearly responsible for causing metabolic.²⁸ Case reports were not used in the Cochrane review, and a few cases of MALA in the setting of therapeutic use with no underlying risk factors are reported.^31,126,166

It is difficult to predict outcome after metformin overdose. A retrospective literature review found no deaths in cases with a nadir serum pH greater than 6.9, a peak serum lactate concentration less than 25 mmol/L, or a peak serum metformin concentration less than 50 mcg/mL.⁴² However, a retrospective review of intensive care unit patients with MALA found no association between metformin concentrations in survivors and nonsurvivors.¹⁴⁶ There was an association between mortality and lactate concentrations and pH.

Metformin-associated MALA is a potentially lethal condition. Recognition and awareness of this disorder are important. Symptoms are nonspecific and include abdominal pain, nausea, vomiting, malaise, myalgia, and dizziness. However, gastrointestinal symptoms are common adverse effects associated with therapeutic use of metformin and do not necessarily require discontinuation of the drug. More severe clinical manifestations of metformin-associated MALA include confusion, blindness, mental status depression, hypothermia, respiratory insufficiency, and hypotension. Serum metformin concentrations can be obtained as a diagnostic aid, but these not necessity correlate with the clinical condition in both the acute overdose setting and in the setting of therapeutic metformin use.^86,87

Aggressive airway management and vasopressor therapy should be given as clinically indicated. Indications for use of intravenous sodium bicarbonate in critically ill patients with metabolic acidosis with an elevated lactate concentration of various etiologies are poorly defined and controversial. Rather than using an arterial pH cutoff, we recommend using sodium bicarbonate given evidence of impaired buffering capacity based on a serum bicarbonate threshold concentration of less than 5 mEq/L.

Metformin would be expected to be dialyzable based on its small size and limited protein binding. However, it has a large volume of distribution. A literature review concluded that metformin is moderately dialyzable, and extracorporeal treatment is recommended if either the blood lactate concentration is greater than 20 mmol/L, or the pH is less than or equal to 7.00, or standard therapies, including sodium bicarbonate, are not effective.²⁸ Even if metformin removal is inadequate, clinical improvement may occur due to correction of acid–base status.

SUMMARY

• Numerous xenobiotics and medical conditions may cause hypoglycemia.

• Hypoglycemia is the predominant adverse effect related to therapeutic use and overdose of many of the drugs used for treatment of diabetes mellitus.

• Various clinical manifestations, particularly neuropsychiatric, may occur and can be confused with conditions such as ethanol intoxication, psychosis, epilepsy, and cerebrovascular accidents.

• The potential for delayed and prolonged hypoglycemia must be recognized in overdose situations.

• Although several treatment options exist, rapid intravenous administration of dextrose is the most important measure.

• Octreotide is useful for patients with recurrent hypoglycemia following sulfonylurea or meglitinide overdose, and would be expected to be useful for the treatment of recurrent hypoglycemia following overdose with all other noninsulin antidiabetics.

REFERENCES