Diabetes Risks from Prescription and Nonprescription Drugs. Samuel Dagogo-Jack

Austin Bradford Hill⁵ and later elaborated by others, Hill’s criteria form the basis of establishing scientifically valid causal connections between potential disease agents and the many diseases that afflict humankind. The criteria are as follows: 1) temporal relationship, an essential requirement that exposure to the agent must necessarily always precede the occurrence of the disease; 2) strength of association; 3) dose response, a requirement that exposure to increasing amounts of the agent should result in increasing severity of disease; 4) consistency; 5) biological plausibility, with regard to currently accepted understanding of disease mechanisms; 6) consideration of alternate explanations; 7) experimental insight, a requirement that the observed association be reproducible and modifiable through deliberate experimentation; 8) specificity of the association; and 9) coherence, which requires that the association be compatible with existing theory and knowledge.

      Applying Hill’s criteria,⁵ only a few exceptions (e.g., pancreatic toxins and steroid-induced diabetes), among the myriad of medications that have been reported to alter blood glucose levels, meet the scientific threshold for establishing direct causality. For the majority of medications, the occurrence of diabetes is observed only in a minority of patients exposed, and the reported link to diabetes suffers from the paucity of data from randomized controlled studies, weak association, inconsistent pattern, indeterminate temporal relationship (resulting from lack of baseline glucose data in many cases), unclear mechanisms, and uncertain dose-response relationship.

      Despite these caveats, it behooves clinicians to keep an open mind and maintain vigilance, because the lack of gold-standard evidence of causation may not be absolute evidence of the lack of a possible causal relationship. A case in point is the relationship between smoking and diseases of the lung and heart. No randomized controlled trials, in which subjects are allocated to smoke cigarettes or placebo “cigarettes” and are followed for many years until the development of lung or heart disease, can ever be conceived let alone conducted for obvious ethical and moral reasons. As a result, the epidemiological and association studies have held sway, such that their results demonstrating strong, consistent, and dose-response relationships between smoking and several cardiopulmonary disorders have informed the promulgation of far-reaching antismoking measures.⁶

      Besides a direct, causal mechanism, possible explanations for treatment-emergent diabetes could include preexisting undiagnosed diabetes; increased susceptibility to diabetes from underlying genetic or environmental risk factors; an indirect effect mediated by known risk factors, such as weight gain; a coincidental finding; or an idiosyncratic reaction in susceptible persons that is inherently unpredictable.

      Drugs Associated with Type 1 Diabetes

      Type 1 diabetes (T1D) is a disease of absolute or severe insulin deficiency. The vast majority of patients with T1D have autoimmune destruction of the pancreatic islet β-cells as the underlying mechanism for the insulin deficiency. In a small minority of patients with T1D, evidence for autoimmunity is lacking and the etiology of islet destruction is unclear. Drug-induced T1D is rare in clinical practice. Experimental insulin-deficient diabetes can be induced by treating rodents with the pancreatic toxins streptozotocin or alloxan. Ingestion of the rat poison Vacor (N-3 pyridylmethyl-N’ 4 nitrophenyl urea), which is structurally related to alloxan and streptozotocin, has been associated with human cases of diabetes.^7,8 The diabetes induced by Vacor poisoning can have a delayed onset (≥1 week), but it usually is severe and often presents with ketoacidosis. Prophylactic treatment with the antidote nicotinamide should be started as soon as possible following ingestion of Vacor, even in euglycemic persons.^7,9

      The antiprotozoal drug pentamidine, widely used for the treatment of refractory Pneumocystis jirovecii pneumonia (PJP; formerly known as Pneumocystis carinii pneumonia [PCP]) in HIV-infected patients and transplant recipients, can induce acute insulinopenic diabetes in some patients through destruction of the pancreatic β-cells.^10,11 An initial phase of hypoglycemia (reflecting β-cell degranulation and transient hyperinsulinemia) may precede the development of pentamidine-induced diabetes.¹²

      The risk factors and exact mechanisms for β-cell destruction by pentamidine are unknown, and the resultant diabetes tends to persist after withdrawal of pentamidine.¹⁰ In vitro studies suggest that pentamidine might be a substrate for the organic cation transporter 1 (OCT1).¹³ The OCTs modulate the availability of cationic drugs, such as metformin.¹⁴ Interestingly, cimetidine (an inhibitor of OCTs) has been shown to augment pentamidine-induced hyperglycemia in a rat model.¹⁵ The latter finding suggests that hyperglycemia following pentamidine exposure is related to drug availability and could be exacerbated by coadministration of cimetidine and other drugs that inhibit OCT1.

      Alcoholism and chronic pancreatitis have been associated with the development of insulin-deficient diabetes.¹⁶ More than 500 individual drugs have been associated with the risk of acute pancreatitis as an adverse effect, yet in only a small percentage of drugs has causality be established.¹⁷ Additionally, it is chronic exocrine pancreatic inflammation and damage rather than acute pancreatitis that is more likely to be linked etiologically to diabetes risk. The subject of alcohol and diabetes risk is discussed further in Chapter 10. It is possible that other as yet unidentified environmental toxins may be involved in the pathogenesis of the rare nonautoimmune cases of T1D.

      Drugs Associated with Type 2 Diabetes

      In contrast to the rarity of drug-associated typical T1D, several commonly used medications have been associated with the development of phenotypical type 2 diabetes (T2D). The latter classification is based loosely on the absence of absolute insulinopenia together with evidence for insulin resistance or related mechanisms for the disruption of glucoregulation. Depending on the severity and acuteness of the perturbation, some patients with drug-related diabetes may present with diabetic ketoacidosis (DKA). The fact that DKA is more characteristic of T1D than T2D may lead physicians to classify such presentations as T1D. Note, however, that ~25% of patients with T2D in the general population present with DKA.^18,19 After initial stabilization with insulin therapy and fluid repletion, the majority of such patients respond to oral antidiabetic agents, as is typical of T2D. Therefore, it is more important to stabilize the patient with drug-related hyperglycemic crisis than to be distracted by a quest for an exact classification in the acute setting. In fact, in many instances, the exact mechanism of the medication-related hyperglycemia remains obscure, and the condition is best assigned to the other category.

      According to the Centers for Disease Control and Prevention (CDC), the age-adjusted percentage of adults ≥18 years old with diagnosed diabetes who reported having hypertension was 57.1% in 2009.²⁰ Other chronic comorbidities frequently found in people with diabetes include dyslipidemia, degenerative joint disease, chronic obstructive pulmonary disease, sleep apnea, congestive heart failure, affective disorders, and peptic ulcer disease. People with diabetes also are at risk for infections. These various conditions often require chronic or recurrent treatment with a wide array of medications, some of which could affect insulin sensitivity, β-cell function, or other aspects of glucoregulation. Whenever feasible, preference should be given to those agents that are either neutral or beneficial in their effects on carbohydrate and lipid metabolism.

      In the chapters that follow, different classes of medications will be discussed with regard to their impact on diabetes risk. These medication classes were selected for discussion based either on 1) their historical association with dysglycemia in clinical practice, 2) extensive utilization for the management of comorbid conditions (e.g., hypertension, dyslipidemia) in patients with diabetes, or 3) existing or emerging reports of possible association with dysglycemia.

      References

      1. Comi RJ. Drug-induced diabetes mellitus. In Diabetes Mellitus. 2nd ed. LeRoith, Taylor, Olefsky, Eds. Lippincott Williams & Wilkins, Philadelphia, 2000, p. 582–588

      2. Zillich AJ, Garg J, Basu S, Bakris GL, Carter BL. Thiazide diuretics, potassium, and the development of diabetes: a quantitative review. Hypertension 2006;48:219–224

      3. Boursi B, Mamtani R, Haynes K, Yang YX. The effect of past antibiotic exposure on diabetes risk. Eur J Endocrinol 2015;172:639–648