Report of the Expert Committee on the Diagnosis and Classification Of

COMMITTEE REPORT

Report of the Expert Committee on the Diagnosis and Classiﬁcation of Diabetes Mellitus

THE EXPERT COMMITTEE ON THE DIAGNOSIS AND CLASSIFICATION OF DIABETES MELLITUS*

he current classification and diagno- nosis of diabetes were warranted. The in insulin secretion, insulin action, or sis of diabetes used in the U.S. was Committee met on multiple occasions both. The chronic hyperglycemia of dia- T developed by the National Diabetes and widely circulated a draft report of betes is associated with long-term dam- Data Group (NDDG) and published in their findings and preliminary recom- age, dysfunction, and failure of various 1979 (1). The impetus for the classifica- mendations to the international diabe- organs, especially the eyes, kidneys, tion and diagnosis scheme proposed then tes community. Based on the numerous nerves, heart, and blood vessels. holds true today. That is, comments and suggestions received, in- Several pathogenic processes are including the opportunity to review unpub- volved in the development of diabetes. the growth of knowledge regarding the eti- lished data in detail, the Committee These range from autoimmune destruc- ology and pathogenesis of diabetes has led discussed and revised numerous drafts of tion of the ␤-cells of the pancreas with many individuals and groups in the diabe- a manuscript that culminated in this pub- consequent insulin deficiency to abnor- tes community to express the need for a lished document. malities that result in resistance to insulin revision of the nomenclature, diagnostic This report is divided into four sec- action. The basis of the abnormalities in criteria, and classification of diabetes. As a tions: definition and description of dia- carbohydrate, fat, and protein metabo- consequence, it was deemed essential to develop an appropriate, uniform terminology betes, classification of the disease, lism in diabetes is deficient action of in- and a functional, working classification of diagnostic criteria, and testing for diabe- sulin on target tissues. Deficient insulin diabetes that reflects the current knowledge tes. The aim of this document is to define action results from inadequate insulin se- about the disease. (1) and describe diabetes as we know it to- cretion and/or diminished tissue re- day, present a classification scheme that sponses to insulin at one or more points in It is now considered to be particularly im- reflects its etiology and/or pathogenesis, the complex pathways of hormone action. portant to move away from a system that provide guidelines for the diagnosis of the Impairment of insulin secretion and de- appears to base the classification of the disease, develop recommendations for fects in insulin action frequently coexist disease, in large part, on the type of phar- testing that can help reduce the morbidity in the same patient, and it is often unclear macological treatment used in its manage- and mortality associated with diabetes, which abnormality, if either alone, is the ment toward a system based on disease and review the diagnosis of gestational di- primary cause of the hyperglycemia. etiology where possible. abetes. Symptoms of marked hyperglycemia An international Expert Committee, include polyuria, polydipsia, weight loss, working under the sponsorship of the DEFINITION AND sometimes with polyphagia, and blurred American Diabetes Association, was es- DESCRIPTION OF DIABETES vision. Impairment of growth and suscep- tablished in May 1995 to review the sci- MELLITUS — Diabetes mellitus is a tibility to certain infections may also ac- entific literature since 1979 and to decide group of metabolic diseases characterized company chronic hyperglycemia. Acute, if changes to the classification and diag- by hyperglycemia resulting from defects life-threatening consequences of diabetes ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● are hyperglycemia with ketoacidosis or From the American Diabetes Association, Alexandria, Virginia. Originally approved 1997. Modified in 1999 the nonketotic hyperosmolar syndrome. based on the Proceedings of the Fourth International Workshop-Conference on Gestational Diabetes Mel- Long-term complications of diabetes litus (Diabetes Care 21 [Suppl. 2]:B1–B167, 1998). include retinopathy with potential loss of *Members of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus: James R. Gavin vision; nephropathy leading to renal fail- III, MD, PhD (Chair), K.G.M.M. Alberti, MD, Mayer B. Davidson, MD, Ralph A. DeFronzo, MD, Allan Drash, ure; peripheral neuropathy with risk of MD, Steven G. Gabbe, MD, Saul Genuth, MD, Maureen I. Harris, PhD, MPH, Richard Kahn, PhD, Harry Keen, MD, FRCP, William C. Knowler, MD, DrPH, Harold Lebovitz, MD, Noel K. Maclaren, MD, Jerry P. foot ulcers, amputation, and Charcot Palmer, MD, Philip Raskin, MD, Robert A. Rizza, MD, and Michael P. Stern, MD. joints; and autonomic neuropathy caus- Abbreviations:ACOG, American College of Obstetricians and Gynecologists; FPG, fasting plasma glu- ing gastrointestinal, genitourinary, and cose; GCT, glucose challenge test; GDM, gestational diabetes mellitus; HNF, hepatocyte nuclear factor, IFG, cardiovascular symptoms and sexual dys- impaired fasting glucose; IGT, impaired glucose tolerance; MODY, maturity-onset diabetes of the young; NDDG, National Diabetes Data Group; NHANES III, Third National Health and Nutrition Examination function. Glycation of tissue proteins and Survey; OGTT, oral glucose tolerance test; PAI-1, plasminogen activator inhibitor-1; WHO, World Health other macromolecules and excess pro- Organization; 2-h PG, 2-h postload glucose. duction of polyol compounds from glu-

DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 S5 Committee Report cose are among the mechanisms thought of the NDDG (2). These groups recog- glucose tolerance (IGT), in which plasma to produce tissue damage from chronic nized two major forms of diabetes, which glucose levels during an OGTT were hyperglycemia. Patients with diabetes they termed insulin-dependent diabetes above normal but below those defined as have an increased incidence of atheroscle- mellitus (IDDM, type 1 diabetes) and diabetes. rotic cardiovascular, peripheral vascular, non-insulin-dependent diabetes mellitus The NDDG/WHO classification high- and cerebrovascular disease. Hyperten- (NIDDM, type 2 diabetes), but their clas- lighted the heterogeneity of the diabetic sion, abnormalities of lipoprotein metab- sification system went on to include evi- syndrome. Such heterogeneity has had olism, and periodontal disease are often dence that diabetes mellitus was an important implications not only for treat- found in people with diabetes. The emo- etiologically and clinically heterogeneous ment of patients with diabetes but also for tional and social impact of diabetes and group of disorders that share hyperglyce- biomedical research. This previous classi- the demands of therapy may cause signif- mia in common. The overwhelming evi- fication indicated that the disorders icant psychosocial dysfunction in patients dence in favor of this heterogeneity grouped together under the term diabetes and their families. included the following: differ markedly in pathogenesis, natural The vast majority of cases of diabetes history, response to therapy, and preven- fall into two broad etiopathogenetic cate- 1. There are several distinct disorders, tion. In addition, different genetic and en- gories (discussed in greater detail below). most of them rare, in which glucose vironmental factors can result in forms of In one category (type 1 diabetes), the intolerance is a feature. diabetes that appear phenotypically simi- cause is an absolute deficiency of insulin 2. There are large differences in the lar but may have different etiologies. secretion. Individuals at increased risk of prevalence of the major forms of di- The classification published in 1979 developing this type of diabetes can often abetes among various racial or eth- was based on knowledge of diabetes at be identified by serological evidence of an nic groups worldwide. that time and represented some compro- autoimmune pathologic process occur- 3. Patients with glucose intolerance mises among different points of view. It ring in the pancreatic islets and by genetic present with great phenotypic vari- was based on a combination of clinical markers. In the other, much more preva- ation; take, for example, the differ- manifestations or treatment requirements lent category (type 2 diabetes), the cause ences between thin, ketosis-prone, (e.g., insulin-dependent, non-insulin-de- is a combination of resistance to insulin insulin-dependent diabetes and pendent) and pathogenesis (e.g., malnu- action and an inadequate compensatory obese, nonketotic, insulin-resistant trition-related, “other types,” gestational). insulin secretory response. In the latter diabetes. It was anticipated, however, that as knowl- category, a degree of hyperglycemia suffi- 4. Evidence from genetic, immunolog- edge of diabetes continued to develop, the cient to cause pathologic and functional ical, and clinical studies shows that classification would need revision. When changes in various target tissues, but in western countries, the forms of the classification was prepared, a defini- without clinical symptoms, may be diabetes that have their onset pri- tive etiology had not been established for present for a long period of time before marily in youth are distinct from any of the diabetes subclasses, except for diabetes is detected. During this asymp- those that have their onset mainly in some of the “other types.” Few suscep- tomatic period, it is possible to demon- adulthood. tibility genes for diabetes had been dis- strate an abnormality in carbohydrate 5. A type of non-insulin-requiring di- covered, and an understanding of the metabolism by measurement of plasma abetes in young people, inherited in immunological basis for most type 1 dia- glucose in the fasting state or after a chal- an autosomal dominant fashion, is betes was just beginning. lenge with an oral glucose load. clearly different from the classic The current Expert Committee has acuteonset diabetes that typically carefully considered the data and ratio- CLASSIFICATION OF DIABETES occurs in children. nale for what was accepted in 1979, along MELLITUS AND OTHER 6. In tropical countries, several clinical with research findings of the last 18 years, CATEGORIES OF GLUCOSE presentations occur, including dia- and we are now proposing changes to the REGULATION — A major require- betes associated with fibrocalcific NDDG/WHO classification scheme (Ta- ment for epidemiological and clinical re- pancreatitis. ble 1). The main features of these changes search and for the clinical management of are as follows: diabetes is an appropriate system of clas- These and other lines of evidence sification that provides a framework were used to divide diabetes mellitus into 1. The terms insulin-dependent diabe- within which to identify and differentiate five distinct types (IDDM, NIDDM, gesta- tes mellitus and non-insulin- its various forms and stages. While there tional diabetes mellitus [GDM], malnutri- dependent diabetes mellitus and have been a number of sets of nomencla- tionrelated diabetes, and other types). their acronyms, IDDM and NIDDM, ture and diagnostic criteria proposed for The different clinical presentations and are eliminated. These terms have diabetes, no generally accepted system- genetic and environmental etiologic fac- been confusing and have frequently atic categorization existed until the tors of the five types permitted discrimi- resulted in classifying the patient NDDG classification system was pub- nation among them. All five types were based on treatment rather than eti- lished in 1979 (1). The World Health Or- characterized by either fasting hypergly- ology. ganization (WHO) Expert Committee on cemia or elevated levels of plasma glucose 2. The terms type 1 and type 2 diabe- Diabetes in 1980 and, later, the WHO during an oral glucose tolerance test tes are retained, with arabic numer- Study Group on Diabetes Mellitus en- (OGTT). In addition, the 1979 classifica- als being used rather than roman dorsed the substantive recommendations tion included the category of impaired numerals. We recommend adop-

S6 DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 Committee Report

Table 1—Etiologic classification of diabetes mellitus tion of arabic numerals in part because the roman numeral II can I. Type 1 diabetes* (␤-cell destruction, usually leading to absolute insulin deficiency) easily be confused by the public as A. Immune mediated B. Idiopathic the number 11. The class, or form, II. Type 2 diabetes* (may range from predominantly insulin resistance with relative insulin deficiency named type 1 diabetes encompasses to a predominantly secretory defect with insulin resistance) the vast majority of cases that are III. Other specific types primarily due to pancreatic islet A. Genetic defects of ␤-cell function ␤ 1. Chromosome 12, HNF-1␣ (MODY3) -cell destruction and that are 2. Chromosome 7, glucokinase (MODY2) prone to ketoacidosis. This form in- 3. Chromosome 20, HNF-4␣ (MODY1) cludes those cases currently ascrib- 4. Mitochondrial DNA able to an autoimmune process and 5. Others those for which an etiology is un- B. Genetic defects in insulin action known. It does not include those 1. Type A insulin resistance ␤ 2. Leprechaunism forms of -cell destruction or fail- 3. Rabson-Mendenhall syndrome ure for which non-autoimmune- 4. Lipoatrophic diabetes specific causes can be assigned (e.g., 5. Others cystic fibrosis). While most type 1 C. Diseases of the exocrine pancreas 1. Pancreatitis diabetes is characterized by the 2. Trauma/pancreatectomy presence of islet cell, GAD, IA-2, IA- 3. Neoplasia 2␤, or insulin autoantibodies that 4. Cystic fibrosis identify the autoimmune process 5. Hemochromatosis that leads to ␤-cell destruction, in 6. Fibrocalculous pancreatopathy 7. Others some subjects, no evidence of auto- D. Endocrinopathies immunity is present; these cases are 1. Acromegaly classified as type 1 idiopathic. 2. Cushing’s syndrome 3. The class, or form, named type 2 3. Glucagonoma diabetes includes the most preva- 4. Pheochromocytoma 5. Hyperthyroidism lent form of diabetes, which results 6. Somatostatinoma from insulin resistance with an in- 7. Aldosteronoma sulin secretory defect. 8. Others 4. A recent international meeting re- E. Drug- or chemical-induced 1. Vacor viewed the evidence for and charac- 2. Pentamidine teristics of malnutrition-related 3. Nicotinic acid diabetes (3). While it appears that 4. Glucocorticoids malnutrition may influence the ex- 5. Thyroid hormone pression of other types of diabetes, 6. Diazoxide 7. ␤-adrenergic agonists the evidence that diabetes can be di- 8. Thiazides rectly caused by protein deficiency 9. Dilantin is not convincing. Therefore, the 10. ␣-Interferon class termed malnutrition-related 11. Others diabetes mellitus has been eliminat- F. Infections 1. Congenital rubella ed. Fibrocalculous pancreatopathy 2. Cytomegalovirus (formerly a subtype of malnutri- 3. Others tion-related diabetes) has been re- G. Uncommon forms of immune-mediated diabetes classified as a disease of the exocrine 1. “Stiff-man” syndrome pancreas. 2. Anti-insulin receptor antibodies 3. Others 5. The stage termed impaired glucose H. Other genetic syndromes sometimes associated with diabetes tolerance (IGT) has been retained. 1. Down’s syndrome The analogous intermediate stage of 2. Klinefelter’s syndrome fasting glucose is named impaired 3. Turner’s syndrome fasting glucose (IFG). 4. Wolfram’s syndrome 5. Friedreich’s ataxia 6. The class termed gestational diabe- 6. Huntington’s chorea tes mellitus (GDM) is retained as de- 7. Laurence-Moon-Biedl syndrome fined by the WHO and NDDG, 8. Myotonic dystrophy respectively. Selective rather than 9. Porphyria 10. Prader-Willi syndrome universal screening for glucose in- 11. Others tolerance in pregnancy is now rec- IV. Gestational diabetes mellitus (GDM) ommended. *Patients with any form of diabetes may require insulin treatment at some stage of their disease. Such use of 7. The degree of hyperglycemia (if insulin does not, of itself, classify the patient. any) may change over time, de-

DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 S7 Committee Report

-Even after presenting in ketoacidosis, these patients can brieﬂy return to normoglyء .Figure 1—Disorders of glycemia: etiologic types and stages in rare instances, patients in these categories (e.g., Vacor toxicity, typeءء ;(cemia without requiring continuous therapy (i.e., “honeymoon” remission 1 diabetes presenting in pregnancy) may require insulin for survival.

pending on the extent of the under- class. For example, a person with immune destruction of the ␤-cells of the lying disease process (Fig. 1). A GDM may continue to be hypergly- pancreas (4). Markers of the immune de- disease process may be present but cemic after delivery and may be destruction of the ␤-cell include islet cell may not have progressed far enough termined to have, in fact, type 1 autoantibodies (ICAs), autoantibodies to to cause hyperglycemia. The same diabetes. Alternatively, a person insulin (IAAs), autoantibodies to glutamic disease process can cause IFG and/ who acquires diabetes because of acid decarboxylase (GAD65), and autoan- or IGT without fulfilling the criteria large doses of exogenous steroids tibodies to the tyrosine phosphatases IA-2 for the diagnosis of diabetes. In may become normoglycemic once and IA-2␤ (5–13). One and usually more some individuals with diabetes, the glucocorticoids are discontin- of these autoantibodies are present in 85– adequate glycemic control can be ued, but then may develop diabetes 90% of individuals when fasting hyper- achieved with weight reduction, ex- many years later after recurrent ep- glycemia is initially detected. Also, the ercise, and/or oral glucose-lowering isodes of pancreatitis. Another ex- disease has strong HLA associations, with agents. These individuals therefore ample would be a person treated linkage to the DQA and B genes, and it is do not require insulin. Other indi- with thiazides who develops diabe- influenced by the DRB genes (14,15). viduals, who have some residual tes years later. Because thiazides in These HLA-DR/DQ alleles can be either insulin secretion but require exoge- themselves seldom cause severe hy- predisposing or protective. nous insulin for adequate glycemic perglycemia, such individuals In this form of diabetes, the rate of control, can survive without it. In- probably have type 2 diabetes that is ␤-cell destruction is quite variable, being dividuals with extensive ␤-cell de- exacerbated by the drug. Thus, for rapid in some individuals (mainly infants struction and therefore no residual the clinician and patient, it is less and children) and slow in others (mainly insulin secretion require insulin for important to label the particular adults [16]). Some patients, particularly survival. The severity of the meta- type of diabetes than it is to under- children and adolescents, may present bolic abnormality can progress, re- stand the pathogenesis of the hyper- with ketoacidosis as the first manifesta- gress, or stay the same. Thus, the glycemia and to treat it effectively. tion of the disease. Others have modest degree of hyperglycemia reflects the fasting hyperglycemia that can rapidly severity of the underlying metabolic Type 1 diabetes (␤-cell destruction, change to severe hyperglycemia and/or process and its treatment more than usually leading to absolute insulin ketoacidosis in the presence of infection the nature of the process itself. deficiency) or other stress. Still others, particularly 8. Assigning a type of diabetes to an Immune-mediated diabetes. This form adults, may retain residual ␤-cell function individual often depends on the cir- of diabetes, previously encompassed by sufficient to prevent ketoacidosis for cumstances present at the time of the terms insulin-dependent diabetes, many years. Many such individuals with diagnosis, and many diabetic indi- type 1 diabetes, or juvenile-onset diabe- this form of type 1 diabetes eventually be- viduals do not easily fit into a single tes, results from a cellular-mediated auto- come dependent on insulin for survival

S8 DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 Committee Report and are at risk for ketoacidosis. At this definitive subclassification. Although the years). They are referred to as maturity- latter stage of the disease, there is little or specific etiologies of this form of diabetes onset diabetes of the young (MODY) and no insulin secretion, as manifested by low are not known, autoimmune destruction are characterized by impaired insulin se- or undetectable levels of plasma C- of ␤-cells does not occur, and patients do cretion with minimal or no defects in in- peptide. Immune-mediated diabetes not have any of the other causes of diabe- sulin action (44–46). They are inherited commonly occurs in childhood and ado- tes listed above or below. in an autosomal dominant pattern. Ab- lescence, but it can occur at any age, even Most patients with this form of diabe- normalities at three genetic loci on differ- in the 8th and 9th decades of life. tes are obese, and obesity itself causes ent chromosomes have been identified to Autoimmune destruction of ␤-cells some degree of insulin resistance (22,23). date. The most common form is associ- has multiple genetic predispositions and Patients who are not obese by traditional ated with mutations on chromosome 12 is also related to environmental factors weight criteria may have an increased per- in a hepatic transcription factor referred that are still poorly defined. Although pa- centage of body fat distributed predomi- to as hepatocyte nuclear factor (HNF)-1␣ tients are rarely obese when they present nantly in the abdominal region (24). (47,48). A second form is associated with with this type of diabetes, the presence of Ketoacidosis seldom occurs spontane- mutations in the glucokinase gene on obesity is not incompatible with the diag- ously in this type of diabetes; when seen, chromosome 7p and results in a defective nosis. These patients are also prone to it usually arises in association with the glucokinase molecule (49,50). Gluco- other autoimmune disorders such as stress of another illness such as infection kinase converts glucose to glucose-6- Graves’ disease, Hashimoto’s thyroiditis, (25–27). This form of diabetes frequently phosphate, the metabolism of which, in Addison’s disease, vitiligo, and pernicious goes undiagnosed for many years because turn, stimulates insulin secretion by the anemia. the hyperglycemia develops gradually ␤-cell. Thus, glucokinase serves as the Idiopathic diabetes. Some forms of type and at earlier stages is often not severe “glucose sensor” for the ␤-cell. Because of 1 diabetes have no known etiologies. enough for the patient to notice any of the defects in the glucokinase gene, increased Some of these patients have permanent classic symptoms of diabetes (28–30). plasma levels of glucose are necessary to insulinopenia and are prone to ketoacido- Nevertheless, such patients are at in- elicit normal levels of insulin secretion. A sis, but have no evidence of autoimmu- creased risk of developing macrovascular third form is associated with a mutation in nity. Although only a minority of patients and microvascular complications (30– the HNF-4␣ gene on chromosome 20q with type 1 diabetes fall into this category, 34). Whereas patients with this form of (51,52). HNF-4␣ is a transcription factor of those who do, most are of African or diabetes may have insulin levels that ap- involved in the regulation of the expres- Asian origin. Individuals with this form of pear normal or elevated, the higher blood sion of HNF-1␣. The specific genetic de- diabetes suffer from episodic ketoacidosis glucose levels in these diabetic patients fects in a substantial number of other and exhibit varying degrees of insulin de- would be expected to result in even individuals who have a similar clinical ficiency between episodes. This form of higher insulin values had their ␤-cell presentation are currently unknown. diabetes is strongly inherited, lacks im- function been normal (35). Thus, insulin Point mutations in mitochondrial munological evidence for ␤-cell autoim- secretion is defective in these patients and DNA have been found to be associated munity, and is not HLA associated. An insufficient to compensate for the insulin with diabetes mellitus and deafness (53– absolute requirement for insulin replace- resistance. Insulin resistance may im- 55). The most common mutation occurs ment therapy in affected patients may prove with weight reduction and/or phar- at position 3243 in the tRNA leucine come and go (17). macological treatment of hyperglycemia gene, leading to an A-to-G transition. An but is seldom restored to normal (36– identical lesion occurs in the MELAS syn- Type 2 diabetes (ranging from 40). The risk of developing this form of drome (mitochondrial myopathy, en- predominantly insulin resistance diabetes increases with age, obesity, and cephalopathy, lactic acidosis, and stroke- with relative insulin deficiency to lack of physical activity (29,41). It occurs like syndrome); however, diabetes is not predominantly an insulin secretory more frequently in women with prior part of this syndrome, suggesting differ- defect with insulin resistance) GDM and in individuals with hyperten- ent phenotypic expressions of this genetic This form of diabetes, previously referred sion or dyslipidemia, and its frequency lesion (56). to as non-insulin-dependent diabetes, varies in different racial/ethnic subgroups Genetic abnormalities that result in type 2 diabetes, or adult-onset diabetes, is (29,30,41). It is often associated with a the inability to convert proinsulin to in- a term used for individuals who have in- strong genetic predisposition, more so sulin have been identified in a few fami- sulin resistance and usually have relative than is the autoimmune form of type 1 lies, and such traits are inherited in an (rather than absolute) insulin deficiency diabetes (42,43). However, the genetics autosomal dominant pattern (57,58). The (18–21). At least initially, and often of this form of diabetes are complex and resultant glucose intolerance is mild. Sim- throughout their lifetime, these individu- not clearly defined. ilarly, the production of mutant insulin als do not need insulin treatment to sur- molecules with resultant impaired recep- vive. There are probably many different Other specific types of diabetes tor binding has also been identified in causes of this form of diabetes, and it is Genetic defects of the ␤-cell. Several a few families and is associated with an likely that the proportion of patients in forms of diabetes are associated with autosomal inheritance and only mildly this category will decrease in the future as monogenetic defects in ␤-cell function. impaired or even normal glucose metab- identification of specific pathogenic pro- These forms of diabetes are frequently olism (59–61). cesses and genetic defects permits better characterized by onset of hyperglycemia Genetic defects in insulin action. There differentiation among them and a more at an early age (generally before age 25 are unusual causes of diabetes that result

DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 S9 Committee Report from genetically determined abnormali- Somatostatinoma- and aldoster- sus and other autoimmune diseases (63). ties of insulin action. The metabolic ab- onoma-induced hypokalemia can cause As in other states of extreme insulin resis- normalities associated with mutations of the diabetes, at least in part, by inhibiting in- tance, patients with anti-insulin receptor insulin receptor may range from hyper- sulin secretion (75,76). Hyperglycemia antibodies often have acanthosis nigricans. insulinemia and modest hyperglycemia to generally resolves after successful re- In the past, this syndrome was termed severe diabetes (62,63). Some individuals moval of the tumor. type B insulin resistance. with these mutations may have acanthosis Drug- or chemical-induced diabetes. Other genetic syndromes sometimes nigricans. Women may be virilized and Many drugs can impair insulin secretion. associated with diabetes. Many genetic have enlarged, cystic ovaries (64,65). In These drugs may not cause diabetes by syndromes are accompanied by an in- the past, this syndrome was termed type A themselves, but they may precipitate creased incidence of diabetes mellitus insulin resistance (62). Leprechaunism diabetes in individuals with insulin re- (90). These include the chromosomal and the Rabson-Mendenhall syndrome sistance (77,78). In such cases, the classi- abnormalities of Down’s syndrome, are two pediatric syndromes that have fication is unclear because the sequence Kline-felter’s syndrome, and Turner’s mutations in the insulin receptor gene or relative importance of ␤-cell dysfunc- syndrome. Wolfram’s syndrome is an au- with subsequent alterations in insulin re- tion and insulin resistance is unknown. tosomal recessive disorder characterized ceptor function and extreme insulin resis- Certain toxins such as Vacor (a rat poison) by insulin-deficient diabetes and the ab- tance (63). The former has characteristic and intravenous pentamidine can per- sence of ␤-cells at autopsy (91). Addi- facial features and is usually fatal in in- manently destroy pancreatic ␤-cells (79– tional manifestations include diabetes fancy, while the latter is associated with 82). Such drug reactions fortunately are insipidus, hypogonadism, optic atrophy, abnormalities of teeth and nails and pi- rare. There are also many drugs and hor- and neural deafness. Other syndromes are neal gland hyperplasia. mones that can impair insulin action. Ex- listed in Table 1. Alterations in the structure and func- amples include nicotinic acid and tion of the insulin receptor cannot be glucocorticoids (77,78). Patients receiv- Gestational diabetes mellitus (GDM) demonstrated in patients with insulin- ing ␣-interferon have been reported to GDM is defined as any degree of glucose resistant lipoatrophic diabetes. Therefore, develop diabetes associated with islet cell intolerance with onset or first recognition it is assumed that the lesion(s) must reside antibodies and, in certain instances, se- during pregnancy. The definition applies in the postreceptor signal transduction vere insulin deficiency (83,84). The list regardless of whether insulin or only diet pathways. shown in Table 1 is not all-inclusive, but modification is used for treatment or Diseases of the exocrine pancreas. Any reflects the more commonly recognized whether the condition persists after preg- process that diffusely injures the pancreas drug-, hormone-, or toxin-induced forms nancy. It does not exclude the possibility can cause diabetes. Acquired processes of diabetes. that unrecognized glucose intolerance include pancreatitis, trauma, infection, Infections. Certain viruses have been as- may have antedated or begun concomi- pancreatectomy, and pancreatic carcino- sociated with ␤-cell destruction. Diabetes tantly with the pregnancy (92). Six weeks ma (66–68). With the exception of can- occurs in patients with congenital rubella or more after pregnancy ends, the woman cer, damage to the pancreas must be (85), although most of these patients have should be reclassified, as described below extensive for diabetes to occur. However, HLA and immune markers characteristic (see diagnostic criteria for diabetes melli- adenocarcinomas that involve only a small of type 1 diabetes. In addition, coxsackie- tus), into one of the following categories: portion of the pancreas have been associ- virus B, cytomegalovirus, adenovirus, and 1) diabetes, 2) IFG, 3) IGT, or 4) normo- ated with diabetes. This implies a mecha- mumps have been implicated in inducing glycemia. In the majority of cases of GDM, nism other than simple reduction in ␤-cell certain cases of the disease (86–88). glucose regulation will return to normal mass. If extensive enough, cystic fibrosis Uncommon forms of immune-mediat- after delivery. and hemochromatosis will also damage ed diabetes. In this category, there are GDM complicates ϳ4% of all preg- ␤-cells and impair insulin secretion twoknownconditions,andothersarelike- nancies in the U.S., resulting in ϳ135,000 (69,70). Fibrocalculous pancreatopathy ly to occur. The stiff-man syndrome is an cases annually (93). The prevalence may may be accompanied by abdominal pain autoimmune disorder of the central ner- range from 1 to 14% of pregnancies, de- radiating to the back and pancreatic cal- vous system characterized by stiffness of pending on the population studied (93). cifications on X ray (71). Pancreatic fibro- the axial muscles with painful spasms GDM represents nearly 90% of all preg- sis and calcium stones in the exocrine (89). Patients usually have high titers of nancies complicated by diabetes (94). ducts have been found at autopsy. the GAD autoantibodies and approxi- Clinical recognition of GDM is important Endocrinopathies. Several hormones mately one-third will develop diabetes. because therapy, including medical nutri- (e.g., growth hormone, cortisol, gluca- Anti-insulin receptor antibodies can tion therapy, insulin when necessary, and gon, epinephrine) antagonize insulin ac- cause diabetes by binding to the insulin antepartum fetal surveillance, can reduce tion. Excess amounts of these hormones receptor, thereby blocking the binding of the well-described GDM-associated peri- (e.g., acromegaly, Cushing’s syndrome, insulin to its receptor in target tissues natal morbidity and mortality (95). Ma- glucagonoma, pheochromocytoma) can (63). However, in some cases, these anti- ternal complications related to GDM also cause diabetes (72–75). This generally bodies can act as an insulin agonist after include an increased rate of cesarean de- occurs in individuals with preexisting de- binding to the receptor and can thereby livery and chronic hypertension (95–97). fects in insulin secretion, and hyperglyce- cause hypoglycemia. Anti-insulin recep- Although many patients diagnosed with mia typically resolves when the hormone tor antibodies are occasionally found in GDM will not develop diabetes later in excess is removed. patients with systemic lupus erythemato- life, others will be diagnosed many years

S10 DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 Committee Report

Table 2—Diagnosis of GDM with a 100-g or 75-g OGTT values in different popu- teristics should follow one of two ap- 75-g glucose load lations (109–111). This method has proaches: provided values for plasma glucose con- One-step approach: mg/dl mmol/l centrations that are similar to the Carpen- ter/Coustan extrapolations of the 100-g Perform a diagnostic OGTT with- 100-g Glucose load OGTT. out prior plasma or serum glucose Fasting 95 5.3 Recommendations from the Ameri- screening. The one-step approach 1-h 180 10.0 can Diabetes Association’s Fourth Inter- may be cost-effective in high-risk 2-h 155 8.6 national Workshop-Conference on patients or populations (e.g., some 3-h 140 7.8 Gestational Diabetes Mellitus held in Native-American groups). 75-g Glucose load March 1997 support the use of the Car- Two-step approach: Fasting 95 5.3 penter/Coustan diagnostic criteria as well Perform an initial screening by 1-h 180 10.0 as the alternative use of a diagnostic 75-g measuring the plasma or serum glu- 2-h 155 8.6 2-h OGTT (111a). These criteria are sum- cose concentration 1 h after a 50-g Two or more of the venous plasma concentrations marized below. oral glucose load (glucose challenge must be met or exceeded for a positive diagnosis. Testing for gestational diabetes. Previ- The test should be done in the morning after an test [GCT]) and perform a diagnos- overnight fast of between 8 and 14 h and after at least ous recommendations have been that tic OGTT on that subset of women 3 days of unrestricted diet (Ն150 g carbohydrate per screening for GDM be performed in all day) and unlimited physical activity. The subject exceeding the glucose threshold pregnancies. However, there are certain value on the GCT. When the two- should remain seated and should not smoke factors that place women at lower risk for throughout the test. step approach is employed, a glu- the development of glucose intolerance cose threshold value Ͼ140 mg/dl during pregnancy, and it is likely not cost- (7.8 mmol/l) identifies approxi- postpartum as having type 1 diabetes, effective to screen such patients. This low- type 2 diabetes, IFG, or IGT (98–103). mately 80% of women with GDM, risk group comprises women who are and the yield is further increased to Deterioration of glucose tolerance oc- Ͻ 25 years of age and of normal body 90% by using a cutoff of Ͼ130 curs normally during pregnancy, particu- weight, have no family history (i.e., first- larly in the 3rd trimester. The criteria for mg/dl (7.2 mmol/l). degree relative) of diabetes, have no his- abnormal glucose tolerance in pregnancy, tory of abnormal glucose metabolism or which are widely used in the U.S., were With either approach, the diagnosis poor obstetric outcome, and are not mem- proposed by O’Sullivan and Mahan (98) of GDM is based on an OGTT. Diagnostic bers of an ethnic/racial group with a high in 1964 and were based on data obtained criteria for the 100-g OGTT are derived prevalence of diabetes (e.g., Hispanic from OGTTs performed on 752 pregnant from the original work of O’Sullivan and American, Native American, Asian Amer- women. Abnormal glucose tolerance was Mahan, modified by Carpenter and Cous- defined as two or more blood glucose val- ican, African-American, Pacific Islander) tan, and are shown in the top of Table 2. ues out of four that were greater than or (112–114). Pregnant women who fulfill Alternatively, the diagnosis can be made equal to two standard deviations above all of these criteria need not be screened using a 75-g glucose load and the glucose the mean. These values were set based on for GDM. threshold values listed for fasting, 1 h, the prediction of diabetes developing later Risk assessment for GDM should be and 2 h (Table 2, bottom); however, this in life. undertaken at the first prenatal visit. test is not as well validated as the 100-g In 1979, the NDDG revised the Women with clinical characteristics con- OGTT. O’Sullivan and Mahan criteria, converting sistent with a high risk of GDM (marked the whole blood values to plasma values obesity, personal history of GDM, glyco- Impaired glucose tolerance (IGT) (1). These criteria were adopted by the suria, or a strong family history of diabe- and impaired fasting glucose (IFG) American Diabetes Association and the tes) should undergo glucose testing (see The terms IGT and IFG refer to a meta- American College of Obstetricians and below) as soon as feasible. If they are bolic stage intermediate between normal Gynecologists (ACOG) (104), but are at found not to have GDM at that initial glucose homeostasis and diabetes. This variance with WHO criteria. screening, they should be retested be- stage includes individuals who have IGT Carpenter and Coustan (105) sug- tween 24 and 28 weeks of gestation. and individuals with fasting glucose levels gested that the NDDG conversion of the Women of average risk should have test- Ն110 mg/dl (6.1 mmol/l) but Ͻ126 O’Sullivan and Mahan values from the ing undertaken at 24–28 weeks of gesta- mg/dl (7.0 mmol/l) (IFG). The term IFG original Somogyi-Nelson determinations tion. was coined by Charles et al. (115) to refer may have resulted in values that are too A fasting plasma glucose level Ͼ126 to a fasting plasma glucose (FPG) level high. They proposed cutoff values for mg/dl (7.0 mmol/l) or a casual plasma Ն110 mg/dl (6.1 mmol/l) but Ͻ140 plasma glucose that appear to represent glucose Ͼ200 mg/dl (11.1 mmol/l) meets mg/dl (7.8 mmol/l). We are using a simi- more accurately the original O’Sullivan the threshold for the diagnosis of diabe- lar definition, but with the upper end and Mahan determinations. In three stud- tes, if confirmed on a subsequent day, and lowered to correspond to the new diag- ies, these criteria identified more patients precludes the need for any glucose chal- nostic criteria for diabetes. A fasting glu- with GDM whose infants had perinatal lenge. In the absence of this degree of hy- cose concentration of 109 mg/dl (6.1 morbidity (106–108). Additional studies perglycemia, evaluation for GDM in mmol/l) has been chosen as the upper have been completed to define abnormal women with average or high-risk charac- limit of “normal.” Although it is recog-

DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 S11 Committee Report

Table 3—Criteria for the diagnosis of diabetes mellitus WHO (2). The revised criteria for the diagnosis of diabetes are shown in Table 3. 1. Symptoms of diabetes plus casual plasma glucose concentration Ն200 mg/dl (11.1 mmol/ Three ways to diagnose diabetes are pos- l). Casual is defined as any time of day without regard to time since last meal. The classic sible, and each must be confirmed, on a symptoms of diabetes include polyuria, polydipsia, and unexplained weight loss. subsequent day, by any one of the three methods given in Table 3. For example, or one instance of symptoms with casual 2. FPG Ն126 mg/dl (7.0 mmol/l). Fasting is defined as no caloric intake for at least 8 h. plasma glucose Ն200 mg/dl (11.1 mmol/ or l), confirmed on a subsequent day by 1) Ն Ն FPG 126 mg/dl (7.0 mmol/l), 2) an 3. 2-h PG 200 mg/dl (11.1 mmol/l) during an OGTT. The test should be performed as Ն described by WHO (2), using a glucose load containing the equivalent of 75-g anhydrous OGTT with the 2-h postload value 200 glucose dissolved in water. mg/dl (11.1 mmol/l), or 3) symptoms with a casual plasma glucose Ն200 mg/dl In the absence of unequivocal hyperglycemia with acute metabolic decompensation, these criteria should be confirmed by repeat testing on a different day. The third measure (OGTT) is not recommended for routine (11.1 mmol/l), warrants the diagnosis of clinical use. diabetes. For epidemiological studies, estimates of diabetes prevalence and inci- nized that this choice is somewhat arbi- cludes hypertriglyceridemia, which is dence should be based on an FPG Ն126 trary, it is near the level above which acute highly correlated with small dense LDL mg/dl (7.0 mmol/l). This recommenda- phase insulin secretion is lost in response and increased plasminogen activator in- tion is made in the interest of standardiza- to intravenous administration of glucose hibitor-1 (PAI-1) levels. The former is tion and also to facilitate field work, (116) and is associated with a progres- thought to have enhanced atherogenicity, particularly where the OGTT may be dif- sively greater risk of developing micro- perhaps as a result of its greater vulnera- ficult to perform and where the cost and and macrovascular complications (117– bility to oxidation than normal LDL. demands on participants’ time may be ex- 121). PAI-1 is a cardiovascular risk factor prob- cessive. This approach will lead to slightly Note that many individuals with IGT ably because it inhibits fibrinoloysis. lower estimates of prevalence than would are euglycemic in their daily lives (122) Thus, the insulin resistance syndrome be obtained from the combined use of the and may have normal or near normal contains many features that increase car- FPG and OGTT (Table 4). glycated hemoglobin levels (123). Indi- diovascular risk. IFG and IGT may not in The Expert Committee recognizes an viduals with IGT often manifest hypergly- themselves be directly involved in the intermediate group of subjects whose glu- cemia only when challenged with the oral pathogenesis of cardiovascular disease, cose levels, although not meeting criteria glucose load used in the standardized but rather may serve as statistical risk fac- for diabetes, are nevertheless too high to OGTT. tors by association because they correlate be considered altogether normal. This In the absence of pregnancy, IFG and with those elements of the insulin resis- group is defined as having FPG levels IGT are not clinical entities in their own tance syndrome that are cardiovascular Ն110 mg/dl (6.1 mmol/l) but Ͻ126 right but rather risk factors for future di- risk factors. mg/dl (7.0 mmol/l) or 2-h values in the abetes and cardiovascular disease (117). OGTT of Ն140 mg/dl (7.8 mmol/l) but They can be observed as intermediate DIAGNOSTIC CRITERIA FOR Ͻ200 mg/dl (11.1 mmol/l). Thus, the cat- stages in any of the disease processes DIABETES MELLITUS egories of FPG values are as follows: listed in Table 1. IFG and IGT are associated with the insulin resistance syndrome The new criteria ● FPG Ͻ110 mg/dl (6.1 mmol/l) ϭ nor- (also known as syndrome X or the met- The diagnostic criteria for diabetes melli- mal fasting glucose; abolic syndrome), which consists of tus have been modified from those previ- ● FPG Ն110 (6.1 mmol/l) and Ͻ126 insulin resistance, compensatory hyper- ously recommended by the NDDG (1) or mg/dl (7.0 mmol/l) ϭ IFG; insulinemia to maintain glucose homeostasis, obesity (especially abdominal or Table 4—Estimated prevalence of diabetes in the U.S. in individuals 40–74 years old using visceral obesity), dyslipidemia of the data from the NHANES III high-triglyceride and/or low-HDL type, and hypertension (124). Insulin resistance is directly involved in the pathogen- Prevalence (%) of diabetes by Total diabetes esis of type 2 diabetes. IFG and IGT glucose criteria without a prevalence appear as risk factors for this type of dia- Diabetes diagnostic criteria medical history of diabetes* (%)† betes at least in part because of their cor- Medical history of diabetes — 7.92 relation with insulin resistance. In WHO (2) criteria for diabetes: contrast, the explanation for why IFG and FPG Ն140 mg/dl (7.8 mmol/l) or 6.34 14.26 IGT are also risk factors for cardiovascular 2-h PG Ն200 mg/dl (11.1 mmol/l) disease is less clear. The insulin resistance FPG Ն126 mg/dl (7.0 mmol/l) 4.35 12.27 syndrome includes well-recognized car- Data are from K. Flegal, National Center for Health Statistics, personal communication. *Diabetes prevalence diovascular risk factors such as low HDL (by glucose criteria) in those without a medical history of diabetes ϫ (100%-prevalence of diabetes by levels and hypertension. In addition, it in- medical history); †first column of data plus 7.92.

S12 DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 Committee Report

● FPG Ն126 mg/dl (7.0 mmol/l) ϭ provisional diagnosis of diabetes (the diagnosis must be conﬁrmed, as described above).

The corresponding categories when the OGTT is used are the following:

● 2-h postload glucose (2-h PG) Ͻ140 mg/dl (7.8 mmol/l) ϭ normal glucose tolerance; ● 2-h PG Ն140 (7.8 mmol/l) and Ͻ200 mg/dl (11.1 mmol/l) ϭ IGT; ● 2-h PG Ն200 mg/dl (11.1 mmol/l) ϭ provisional diagnosis of diabetes (the diagnosis must be conﬁrmed, as described above).

Since the 2-h OGTT cutoff of 140 mg/dl (7.8 mmol/l) will identify more people as having impaired glucose homeostasis than will the fasting cutoff of 110 mg/d (6.1 mmol/l), it is essential that investigator always report which test was used.

Rationale for the revised criteria for diagnosing diabetes The revised criteria are still based on measures of hyperglycemia. Whereas many different diagnostic schemes have been used all have been based on some measurement of blood or urine glucose, as reviewed by McCance et al. (125). The metabolic defects underlying hyperglycemia, such as islet cell autoimmunity or insulin resistance, should be referred to independently from the diagnosis of diabetes, i.e., in the classiﬁcation of the disease. Determining the optimal diagnostic level of hyperglycemia depends on a bal- ance between the medical, social, and economic costs of making a diagnosis in someone who is not truly at substantial risk of the adverse effects of diabetes and those of failing to diagnose someone who is (126). Unfortunately, not all these data are available, so we relied primarily on medical data. Plasma glucose concentrations are distributed over a continuum, but there is an approximate threshold separating those subjects who are at substantially in- Figure 2—Prevalence of retinopathy by deciles of the distribution of FPG, 2-h PG, and HbA1c in creased risk for some adverse outcomes Pima Indians (A) described by McCance et al. (129), in Egyptians (B) described by Engelgau et al. caused by diabetes (e.g., microvascular (130), and in 40- to 74-year-old participants in NHANES III (C) (K. Flegal, National Center for Health Statistics, personal communication). The x-axis labels indicate the lower limit of each complications from those who are not. decile group. Note that these deciles and the prevalence rates of retinopathy differ considerably Based in part on estimates of the thresh- among the studies, especially the Egyptian study, in which diabetic subjects were oversampled. olds for microvascular disease, the previ- Retinopathy was ascertained by different methods in each study; therefore, the absolute prevalence ous WHO criteria deﬁned diabetes by rates are not comparable between studies, but their relationships with FPG, 2-h PG, and HbA1c are FPG Ն140 mg/dl (7.8 mmol/l), 2-h PG very similar within each population.

DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 S13 Committee Report

Table 5—FPG cutpoints equivalent to the WHO 2-h plasma glucose criterion of 200 mg/dl (130), although the relationship was strongest for 2-h PG. As in other studies, Study and reference Method Fasting plasma glucose* the prevalence rose dramatically in the highest decile of each variable, corre- Pima Indians (129) ROC curves† 123 mg/dl (6.8 mmol/l) sponding to FPG Ն120 mg/dl (6.7 mmol/ Pima Indians (129) Equal prevalence‡ 120 mg/dl (6.7 mmol/l) l), 2-h PG Ն195 mg/dl (10.8 mmol/l), Ն Several Pacific populations (134) Equal prevalence‡ 126 mg/dl (7.0 mmol/l) and HbA1c 6.2%. As in the Pima Indian NHANES III§ Equal prevalence‡ 121 mg/dl (6.7 mmol/l) (129) and Egyptian (130) studies, esti- *The results for the receiver-operating characteristics (ROC) curve analysis of the Pima Indian data and those mates of these “thresholds” for retinopa- from the Pacific populations appear in the cited publications (in millimoles per liter). The other results have thy are somewhat imprecise. More Ն not been published; †equivalent to the WHO criterion of 2-h PG 200 mg/dl (11.1 mmol/l) in sensitivity precision cannot easily be obtained by us- and specificity for retinopathy from analysis of ROC curves; ‡the method is described by Finch et al. (134); §NHANES III subjects ages 40–74 years, excluding users of insulin and oral hypoglycemic agents, weighted ing narrower glycemic intervals (e.g., 20 according to sampling plan (K. Flegal, National Center for Health Statistics, personal communication). instead of the 10 shown in Fig. 2) because of the limited numbers of cases of retinopathy in each sample (32 cases in the Pima Ն200 mg/dl (11.1 mmol/l) in the OGTT, that point the prevalence of the microvas- study, 146 in the Egyptian study, and 111 or both. These criteria effectively defined cular complications considered specific in NHANES III). There are no absolute diabetes by the 2-h PG alone because the for diabetes (i.e., retinopathy and ne- thresholds because some retinopathy oc- fasting and 2-h cutpoint values are not phropathy) increases dramatically. This curred at all glucose levels, presumably equivalent. Almost all individuals with property of the 2-h PG has been com- because of measurement or disease vari- FPG Ն140 mg/dl (7.8 mmol/l) have 2-h pared with the FPG in population studies ability and because of nondiabetic causes PG Ն200 mg/dl (11.1 mmol/l) if given an of the Pima Indians in the U.S., among of retinopathy. OGTT, whereas only about one-fourth of Egyptians, and in the Third National The associations between FPG and those with 2-h PG Ն200 mg/dl (11.1 Health and Nutrition Examination Survey 2-h PG and macrovascular disease have mmol/l) and without previously known (NHANES III) in the U.S. In other studies, been examined in adults without known diabetes have FPG Ն140 mg/dl (7.8 the relationships between glycemia and diabetes (131). The 2-h PG was somewhat mmol/l) (127). Thus, the cutpoint of FPG macrovascular disease have also been ex- more closely associated with major coro- Ն140 mg/dl (7.8 mmol/l) defined a amined. nary heart disease, but there was no sig- greater degree of hyperglycemia than did The relationships of FPG and 2-h PG nificant difference in the association of the the cutpoint of 2-h PG Ն200 mg/dl (11.1 to the development of retinopathy were FPG or the 2-h PG with other indexes of mmol/l). It is the consensus of the Expert evaluated in Pima Indians over a wide macrovascular disease. Similarly, the re- Committee that this discrepancy is un- range of plasma glucose cutpoints (Fig. 2A) lationship between glycemia and periph- warranted and that the cutpoint values for (129). Both variables were similarly asso- eral arterial disease was studied in 50- to both tests should reflect a similar degree ciated with retinopathy, indicating that 74-year-old Caucasians (132). The prev- of hyperglycemia and risk of adverse out- by this criterion, each could work equally alence of arterial disease was strongly re- comes. well for diagnosing diabetes. The au- lated to the FPG and 2-h PG. The Under the previous WHO and the thors concluded that both measures were associations appeared to be of the same NDDG criteria, the diagnosis of diabetes equivalent in terms of the properties pre- strength for both variables. is largely a function of which test is per- viously used to justify diagnostic criteria. In a recent analysis of the Paris Pro- formed. Many individuals who would These findings were confirmed in a spective Study, the incidence of fatal cor- have 2-h PG Ն200 mg/dl (11.1 mmol/l) similar study in Egypt, in which the FPG onary heart disease was related to both in an OGTT are not tested with an OGTT and 2-h PG were each strongly and FPG and 2-h PG determined at a baseline because they lack symptoms or because equally associated with retinopathy (Fig. examination (118). Incidence rates were they have an FPG Ͻ140 mg/dl (7.8 mmol/ 2B) (130). For both the FPG and the 2-h markedly increased at FPG Ն125 mg/dl l). Thus, if it is desired that all people with PG, the prevalence of retinopathy was (6.9 mmol/l) or 2-h PG Ն140 mg/dl (7.8 diabetes be diagnosed and the previous markedly higher above the point of inter- mmol/l). Similarly, the incidence of coro- criteria are followed, OGTTs must be per- section of the two components of the bi- nary artery disease and the all-cause mor- formed periodically in everyone. How- modal frequency distribution (FPG ϭ tality rates were predicted by the FPG in ever, in ordinary practice, not only is the 129 mg/dl [7.2 mmol/l] and 2-h PG ϭ the Baltimore Longitudinal Study of Ag- OGTT performed infrequently, but it is 207 mg/dl [11.5 mmol/l]). ing (R. Andres, C. Coleman, D. Elahi, J. usually not used even to confirm sus- In the NHANES III, 2,821 individuals Fleg, D.C. Muller, J.D. Sorkin, J.D. Tobin, pected cases (128). In summary, the diag- aged 40–74 years received an OGTT, a personal communication). The incidence nostic criteria are now revised to 1) avoid measurement of HbA1c, and an assess- rates of both these outcomes increased the discrepancy between the FPG and 2-h ment of retinopathy by fundus photogra- markedly and almost linearly above FPG PG cutpoint values and 2) facilitate and phy (K. Flegal, personal communication). levels in the range of 110–120 mg/dl encourage the use of a simpler and Figure 2C shows that all three measures of (6.1–6.7 mmol/l). In conclusion, both equally accurate test—fasting plasma glu- glycemia (FPG, 2-h PG, and HbA1c) are the FPG and 2-h PG provide important cose—for diagnosing diabetes. strongly associated with retinopathy, information regarding risk of both micro- The cutpoint for the 2-h PG has been which is similar to the relationships found and macrovascular disease, and the ap- justified largely because at approximately in the Pima Indians (129) and Egyptians proximate thresholds for increased risk

S14 DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 Committee Report correspond with those for retinopathy Table 6—Criteria for testing for diabetes in asymptomatic, undiagnosed individuals and with the revised diagnostic criteria. Reproducibility is another important 1. Testing for diabetes should be considered in all individuals at age 45 years and above and, property of a diagnostic test, a property if normal, it should be repeated at 3-year intervals. for which the FPG appears to be prefera- 2. Testing should be considered at a younger age or be carried out more frequently in ble. When OGTTs were repeated in adults individuals who: 2 during a 2- to 6-week interval, the intra- ● are overweight (BMI Ն25 kg/m ) individual coefficients of variation were ● have a first-degree relative with diabetes 6.4% for the FPG and 16.7% for the 2-h ● are members of a high-risk ethnic population (e.g., African-American, Hispanic Ameri- PG (133). can, Native American, Asian American, Pacific Islander) It is important to review the rationale ● have delivered a baby weighing Ͼ9 lb or have been diagnosed with GDM for retaining the diagnostic cutpoint of ● are hypertensive (Ն140/90) 200 mg/dl (11.1 mmol/l) for the 2-h PG. ● have an HDL cholesterol level Յ35 mg/dl (0.90 mmol/l) and/or a triglyceride level This cutpoint was originally adopted for Ն250 mg/dl (2.82 mmol/l). three reasons (1,2). First, 200 mg/dl (11.1 ● on previous testing, had IGT or IFG mmol/l) has been found to approximate The OGTT or FPG test may be used to diagnose diabetes; however, in clinical settings the FPG test is greatly the cutpoint separating the two compo- preferred because of ease of administration, convenience, acceptability to patients, and lower cost. nents of the bimodal distribution of 2-h PG. Second, in several studies, the preva- cutpoint of 200 mg/dl (11.1 mmol/l) are recommended for diagnosis of diabetes, lence of microvascular disease sharply in- also quite similar to the values of FPG 129 although some studies have shown that ϳ creased above 2-h PG levels of 200 mg/dl (7.2 mmol/l) and 2-h PG 207 mg/dl the frequency distributions for HbA1c mg/dl (11.1 mmol/l). Third, an enormous (11.5 mmol/l) that separated the compo- have characteristics similar to those of the body of clinical and epidemiological data nents of the bimodal frequency distribu- FPG and the 2-h PG. Moreover, these has been collected based on the 2-h PG tions and identified individuals with a studies have defined an HbA1c level above cutpoint of 200 mg/dl (11.1 mmol/l). high prevalence of retinopathy among which the likelihood of having or devel- Thus, this value has been retained for the Egyptians (130). Because the standard er- oping macro- or microvascular disease diagnosis of diabetes because it would be rors of these estimates are not known, the rises sharply (Fig. 2) (129–132). Further- very disruptive, and add little benefit, to small differences in the estimates shown more, HbA1c and FPG (in type 2 diabetes) alter the well-accepted 2-h PG diagnostic in Table 5 may be consistent with sam- have become the measurements of choice level of Ն200 mg/dl (11.1 mmol/l). pling variability. in monitoring the treatment of diabetes, Changing the diagnostic cutpoint for We chose a cutpoint at the upper end and decisions on when and how to imple- the FPG to 126 mg/dl (7.0 mmol/l) is of these estimates (FPG Ն126 mg/dl, 7.0 ment therapy are often made on the basis based on the belief that the cutpoints for mmol/l). This value is slightly higher than of HbA1c. These observations have led the FPG and 2-h PG should diagnose sim- most of the estimated cutpoints that some to recommend HbA1c measurement ilar conditions, given the equivalence of would give the same prevalence of diabe- as a diagnostic test (126,135). the FPG and the 2-h PG in their associa- tes as the criterion of 2-h PG Ն200 mg/dl On the other hand, there are many tions with vascular complications and (11.1 mmol/l). That is, slightly fewer peo- different methods for the measurement of their discrimination between two compo- ple will be diagnosed with diabetes if the HbA1c and other glycosylated proteins, nents of a bimodal frequency distribution new FPG criterion is used alone than if and nationwide standardization of the (129,130). McCance et al. (129) com- either the FPG or the OGTT is used and HbA1c test has just begun (136). Studies puted the FPG level equivalent (in sensi- interpreted by the previous WHO and of the utility of the test compared with the tivity and specificity for retinopathy) to NDDG criteria (Table 4). FPG and 2-h PG have used different as- the 1985 WHO criterion of the 2-h PG As noted above, although the OGTT says, thereby making it difficult to assign Ն200 mg/dl (11.1 mmol/l) and found it is an acceptable diagnostic test and has an appropriate cutpoint. Also, the FPG, Ն to be an FPG of 123 mg/dl (6.8 mmol/l) been an invaluable tool in research, it is 2-h PG, and HbA1c tests are imperfectly (Table 5). Finch et al. (134) approached not recommended for routine use. Be- correlated. In most clinical laboratories, a the problem in each of 13 Pacific popula- cause of its inconvenience to patients and “normal” HbA1c is usually based on a sta- tions surveyed with OGTTs by determin- the perception by many physicians that it tistical sampling of healthy, presumably ing the value in the FPG that, when used is unnecessary, the OGTT is already not nondiabetic individuals. In conclusion, alone as a diagnostic criterion, gave the widely used for diagnosing diabetes. In HbA1c remains a valuable tool for moni- same prevalence of diabetes as did 2-h PG addition, it is more costly and time- toring glycemia, but it is not currently rec- Ն200 mg/dl (11.1 mmol/l). The sum- consuming than the FPG, and the repeat ommended for the diagnosis of diabetes. mary estimate from all these populations test reproducibility of the 2-h PG is worse The revised criteria are for diagnosis was a cutpoint of 126 mg/dl (7.0 mmol/l). than that of the FPG (133). If the OGTT is and are not treatment criteria or goals of The same method was applied to data de- used, either for clinical or research pur- therapy. No change is made in the Amer- rived from the Pima Indians and resulted poses, the test procedure methods recom- ican Diabetes Association’s recommenda- in an FPG cutpoint of 120 mg/dl (6.7 mended by the WHO (2) and the tions of FPG Ͻ120 mg/dl (6.7 mmol/l) Ͻ mmol/l). In NHANES III, the correspond- diagnostic criterion in Table 3 should be and HbA1c 7% as treatment goals (137). ing cutpoint was 121 mg/dl (6.7 mmol/l) employed. The new diagnostic cutpoint (FPG Ն126 (Table 5). These values and the 2-h PG HbA1c measurement is not currently mg/dl [7.0 mmol/l]) is based on the ob-

DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 S15 Committee Report servation that this degree of hyperglyce- teria may, however, have a large impact may be of value to identify which newly mia usually reflects a serious metabolic on the number of people actually diag- diagnosed patients have immune-medi- abnormality that has been shown to be nosed with diabetes. Presently, about half ated type 1 diabetes in circumstances associated with serious complications. the adults with diabetes in the U.S. are where it is not obvious, particularly when The treatment of nonpregnant patients undiagnosed (127), but many might now therapies become available to preserve with hyperglycemia near the cutpoint be diagnosed if the simpler FPG test were ␤-cell mass. should begin with an individualized life- always used. Undiagnosed type 2 diabetes is com- style-modification regimen (i.e., meal mon in the U.S. As many as 50% of the planning and exercise). Initiation of phar- TESTING FOR DIABETES IN people with the disease, or about 8 mil- macological therapy in these patients has PRESUMABLY HEALTHY lion individuals, are undiagnosed (127). INDIVIDUALS not yet been shown to improve prognosis — Type 1 diabetes is Of concern, there is epidemiological evi- and may lead to an unacceptably high in- usually an autoimmune disease, charac- dence that retinopathy begins to develop cidence of hypoglycemic reactions with terized by the presence of a variety of au- at least 7 years before the clinical diagno- certain drugs (e.g., sulfonylureas, insu- toantibodies to protein epitopes on the lin). surface of or within the ␤-cells of the pan- sis of type 2 diabetes is made (142). Be- The new criteria have implications for creas. The presence of such markers because hyperglycemia in type 2 diabetes estimates of the prevalence of diabetes. fore the development of overt disease can causes microvascular disease and may Although an FPG Ն126 mg/dl (7.0 identify patients at risk (138). For exam- cause or contribute to macrovascular dis- mmol/l) and a 2-h PG Ն200 mg/dl (11.1 ple, those with more than one autoanti- ease, undiagnosed diabetes is a serious mmol/l) have similar predictive value for body (i.e., ICA, IAA, GAD, IA-2) are at condition. Patients with undiagnosed adverse outcomes, the two tests are not high risk (139–141). At this time, how- type 2 diabetes are at significantly in- perfectly correlated with each other. A ever, many reasons preclude the recom- creased risk for coronary heart disease, given person may have one glucose value mendation to test individuals routinely stroke, and peripheral vascular disease. In above one cutpoint and another value be- for the presence of any of the immune addition, they have a greater likelihood of low the other cutpoint. Thus, simulta- markers outside of a clinical trials setting. having dyslipidemia, hypertension, and neous measurement of both FPG and 2-h First, cutoff values for some of the assays obesity (143). PG will inevitably lead to some diagnostic for immune markers have not been com- Thus, early detection, and conse- discrepancies and dilemmas. Although pletely established for clinical settings. quently early treatment, might well re- diagnosing diabetes by either test will re- Second, there is no consensus yet as to duce the burden of type 2 diabetes and its sult in a similar number of “cases,” differ- what action should be taken when a pos- complications. However, to increase the ent individuals in different hyperglycemic itive autoantibody test is obtained. Thus, cost-effectiveness of testing undiagnosed, stages may be identified. (This situation autoantibody testing may identify people otherwise healthy individuals, testing would be even more complicated if a third at risk of developing type 1 diabetes with- should be considered in high-risk popu- diagnostic test, such as HbA1c, were out offering any proven measures that lations. Suggested criteria for testing are used.) However, according to the data re- might prevent or delay the clinical onset given in Table 6. Factors leading to these viewed above, there is no basis for con- of disease. Of note, however, is that there recommendations include: 1) the steep cluding that the 2-h PG is more reliable are a number of ongoing well-controlled than the FPG. Thus, the FPG alone should clinical studies testing various means of rise in the incidence of the disease after be used for estimating the comparative preventing type 1 diabetes. These studies age 45 years, 2) the negligible likelihood prevalence of diabetes in different popu- conducted in high-risk subjects may one of developing any of the complications of lations. day offer an effective means to prevent diabetes within a 3-year interval of a neg- Table 4 shows the effect of the new type 1 diabetes, in which case screening ative screening test, and 3) knowledge of diagnostic criteria on the estimated prev- may become appropriate. Last, because the well-documented risk factors for the alence of diabetes in the U.S. population the incidence of type 1 diabetes is low, disease. Although the OGTT and FPG are aged 40–74 years using data from routine testing of healthy children will both suitable tests, in clinical settings, the NHANES III. Diagnosing diabetes in identify only the small number (Ͻ0.5%) FPG is strongly recommended because it those without a medical history of diabe- who at that moment may be “prediabetic.” is easier and faster to perform, more con- tes by using only the FPG test would re- Thus, the cost-effectiveness of such venient and acceptable to patients, more sult in a lower prevalence of diabetes than screening is questionable, at least until an reproducible, and less expensive. would using WHO criteria (4.35 vs. effective therapy is available. For the 6.34%). The total prevalence of diabetes above reasons, the clinical testing of indi- (including those with a medical history) viduals for autoantibodies related to type Acknowledgments— We gratefully ac- would be 12.27%, or 14% lower than the 1 diabetes, outside of research studies, knowledge the invaluable assistance of Robert prevalence of 14.26% by the WHO crite- cannot be recommended at this time. Misbin, MD, in the development of the manu- ria. Of note, these prevalence estimates Similarly, antibody testing of high-risk in- script; Katherine Flegal, PhD, for her analysis refer to results of testing on one occasion. dividuals (e.g., siblings of type 1 patients) of the NHANES III data set; Reubin Andres, The prevalence of diabetes confirmed by a is also not recommended until the efficacy MD, for sharing unpublished data from the second test will be lower regardless of and safety of therapies to prevent or delay Baltimore Longitudinal Study of Aging; and which criteria are used. type 1 diabetes have been demonstrated. Michael Engelgau, MD, for providing the raw Widespread adoption of the new cri- On the other hand, the autoantibody tests data from the Egyptian Study (130).

S16 DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 Committee Report

Identification of a second transmem- Evans DF, Hartz AJ, Kalkhoff RK, Adams References brane protein tyrosine phosphatase, PW: Relationship of body fat distri- 1. National Diabetes Data Group: Classifi- IA-2␤, as an autoantigen in insulin-de- bution to metabolic complications of cation and diagnosis of diabetes mellitus pendent diabetes mellitus: precursor of obesity. J Clin Endocrinol Metab 54:254– and other categories of glucose intoler- the 37-kDa tryptic fragment. Proc Natl 260, 1982 ance. Diabetes 28:1039–1057, 1979 Acad Sci USA 93: 2307–2311, 1996 25. Butkiewicz EK, Leibson C, O’Brien PC, 2. World Health Organization: Diabetes 14. Cantor AB, Krischer JP, Cuthbertson Palumbo PJ, Rizza RA: Insulin therapy Mellitus: Report of a WHO Study Group. DD, Schatz DA, Riley WJ, Malone J, for diabetic ketoacidosis. Diabetes Care Geneva, World Health Org., 1985 Schwartz S, Quattrin T, Maclaren NK: 18:1187–1190, 1995 (Tech. Rep. Ser., no. 727) Age and family relationship accentuate 26. Banerji MA, Chaiken RL, Huey H, 3. Hoet JJ, Tripathy BB, Rao RH, Yajnik CS: the risk of IDDM in relatives of patients Tuomi T, Norin AJ, Mackay IR, Rowley Malnutrition and diabetes in the tropics. with insulin dependent diabetes. J Clin MJ, Zimmet P, Lebovitz H: GAD anti- Diabetes Care 19:1014–1017, 1996 Endocrinol Metab 80:3739–3743, 1995 body negative NIDDM in adult black 4. Atkinson MA, Maclaren NK: The patho- 15. Huang W, Connor E, DelaRosa T, Muir subjects with diabetic ketoacidosis and genesis of insulin dependent diabetes. N A, Schatz D, Silverstein J, Crockett S, increased frequency of human leukocyte Engl J Med 331:1428–1436, 1994 She JX, Maclaren NK: Although DR3- antigen DR3 and DR4. Diabetes 43:741– may be associated with multiple 745, 1994 ءBaekkeskov S, Neilsen JH, Marner B, DQB1 .5 Bilde T, Ludvigsson J, Lernmark A: Au- component diseases of the autoimmune 27. Umpierrez GE, Casals MMC, Gebhart toantibodies in newly diagnosed dia- polyglandular syndromes, the human SSP, Mizon PS, Clark WS, Phillips LS: betic children with immunoprecipitate leukocyte antigen DR4-DQB110302 Diabetic ketoacidosis in obese African- human pancreatic islet cell proteins. Na- haplotype is implicated only in beta cell Americans. Diabetes 44:79–85, 1995 ture 298:167–169, 1982 autoimmunity. J Clin Endocrinol Metab 28. Harris MI: Impaired glucose tolerance in 6. Atkinson MA, Maclaren NK, Riley WJ, 81:1–5, 1996 the U.S. population. Diabetes Care 12: Winter WE, Fisk DD, Spillar RP: Are in- 16. Zimmet PZ, Tuomi T, Mackay R, Rowley 464–474, 1989 sulin autoantibodies markers for insu- MJ, Knowles W, Cohen M, Lang DA: La- 29. Zimmet PZ: Kelly West Lecture 1991: lin-dependent mellitus? Diabetes 35: tent autoimmune diabetes mellitus in challenges in diabetes epidemiology: 894–898, 1986 adults (LADA): the role of antibodies to from west to the rest. Diabetes Care 15: 7. Kaufman D, Erlander M, Clare-Salzler glutamic acid decarboxylase in diagnosis 232–252, 1992 M, Atkinson M, Maclaren N, Tobin A: and prediction of insulin dependency. 30. Fujimoto WY, Leonetti DL, Kinyoun JL, Autoimmunity to two forms of gluta- Diabet Med 11:299–303, 1994 Shuman WP, Stolov WC, Wahl PW: mate decarboxylase in insulin-depen- 17. Banerji M, Lebovitz H: Insulin sensitive Prevalence of complications among sec- dent mellitus. J Clin Invest 89:283–292, and insulin resistant variants in IDDM. ond-generation Japanese-American men 1992 Diabetes 38:784–792, 1989 with diabetes, impaired glucose toler- 8. Christie MR, Tun RY, Lo SSS, Cassidy D, 18. Reaven GM, Bernstein R, Davis B, Olef- ance or normal glucose tolerance. Diabe- Brown TJ, Hollands J, Shattock M, Bot- sky JM: Nonketotic diabetes mellitus: in- tes 36:730–739, 1987 tazzo GF, Leslie RDG: Antibodies to sulin deficiency or insulin resistance? 31. Moss SE, Klein R, Klein BEK, Meuer MS: GAD and tryptic fragments of islet 64K Am J Med 60:80–88, 1976 The association of glycemia and cause- antigen as distinct markers for develop- 19. Olefsky JM, Kolterman OG, Scarlett, JA: specific mortality in a diabetic popula- ment of IDDM: studies with identical Insulin action and resistance in obesity tion. Arch Int Med 154:2473–2479, 1984 twins. Diabetes 41:782–787, 1992 and noninsulin-dependent type II diabe- 32. Kuusisto J, Mykknen L, Pyo¨ra¨la¨ K, 9. Schott M, Schatz D, Atkinson M, Kris- tes mellitus. Am J Physiol 243:E15–E30, Laakso M: NIDDM and its metabolic cher J, Mehta H, Vold B, Maclaren N: 1982 control predict coronary heart disease in GAD65 autoantibodies increase the pre- 20. DeFronzo R, Deibert D, Hendler R, Felig elderly subjects. Diabetes 43:960–967, dictability but not the sensitivity of islet P: Insulin sensitivity and insulin binding 1994 cell and insulin autoantibodies for devel- to monocytes in maturity-onset diabe- 33. Andersson DKG, Svaardsudd K: Long- oping insulin dependent diabetes melli- tes. J Clin Invest 63:939–946, 1979 term glycemic control relates to mortal- tus. J Autoimmunity 7:865–872, 1994 21. Turner RC, Holman, RR, Matthews D, ity in type II diabetes. Diabetes Care 18: 10. Schmidli RS, Colman PG, Harrison LC: Hockaday TDR, Peto J: Insulin defi- 1534–1543, 1995 Do glutamic acid decarboxylase anti- ciency and insulin resistance interaction 34. Uusitupaa MIJ, Niskanen LK, Siitonen bodies improve the prediction of IDDM in diabetes: estimation of their relative O, Voutilainen E, Pyo¨ra¨la¨ K: Ten year in first-degree relatives at risk for IDDM? contribution by feedback analysis from cardiovascular mortality in relation to J Autoimmunity 7:873–879, 1994 basal plasma insulin and glucose con- risk factors and abnormalities in li- 11. Myers MA, Rabin DU, Rowley MJ: Pan- centrations. Metabolism 28:1086–1096, poprotein composition in type 2 (non- creatic islet cell cytoplasmic antibody in 1979 insulin-dependent) diabetic and non- diabetes is represented by antibodies to 22. Kolterman OG, Gray RS, Griffin J, Bur- diabetic subjects. Diabetologia 11:1175– islet cell antigen 512 and glutamic acid stein P, Insel J, Scarlett JA, Olefsky JM: 1184, 1993 decarboxylase. Diabetes 44:1290–1295, Receptor and postreceptor defects con- 35. Polonsky KS, Sturis J, Bell GI: Non-insu- 1995 tribute to the insulin resistance in non- lin-dependent diabetes mellitus: a genet- 12. Lan MS, Wasserfall C, Maclaren NK, insulin-dependent diabetes mellitus. ically programmed failure of the beta cell Notkins AL: IA-2, a transmembrane pro- J Clin Invest 68:957–969, 1981 to compensate for insulin resistance. N tein of the protein tyrosine phosphatase 23. Bogardus C, Lillioja S, Mott DM, Hollen- Engl J Med 334:777–784, 1996 family, is a major autoantigen in insu- beck C, Reaven G: Relationship between 36. Scarlett JA, Gray RS, Griffin J, Olefsky lindependent diabetes mellitus. Proc degree of obesity and in vivo insulin ac- JM, Kolterman OG: Insulin treatment re- Natl Acad Sci USA 93:6367–6370, 1996 tion in man. Am J Physiol 248:E286– verses the insulin resistance of type II 13. Lu J, Li Q, Xie H, Chen Z, Borovitskaya E291, 1985 diabetes mellitus. Diabetes Care 5:353– AE, Maclaren NK, Notkins AL, Lan MS: 24. Kissebah AH, Vydelingum N, Murray R, 363, 1982

DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 S17 Committee Report

37. Firth RG, Bell PM, Rizza RM: Effects of Cox RD, Lathrop GM, Boriraj VV, Chen sulinemia: two cohorts secreting in- tolazamide and exogenous insulin on in- X, Cox NJ, Oda Y, Yano H, Le Beau MM, distinguishable type II intermediates of sulin action in patients with non-insu- Yamada S, Nishigori H, Takeda J, Fajans proinsulin conversion. J Clin Invest 73: lin-dependent diabetes mellitus. N Engl SS, Hattersley AT, Iwasaki N, Hansen T, 714–719, 1984 J Med 314:1280–1286, 1986 Pedersen O, Polonsky KS, Bell GI: Mu- 59. Tager H, Given B, Baldwin D, Mako M, 38. Simonson DC, Ferrannini E, Bevilacqua tations in the hepatocyte nuclear fac- Markese J, Rubenstein A, Olefsky J, S, Smith D, Barrett E, Carlson R, De- tor-1␣ gene in maturity-onset diabetes Kobayashi M, Kolterman O, Poucher R: Fronzo RA: Mechanism of improvement of the young (MODY 3). Nature 384: A structurally abnormal insulin causing in glucose metabolism after chronic gly- 455–458, 1996 human diabetes. Nature 281:122–125, buride therapy. Diabetes 33:838–845, 49. Froguel P, Vaxillaire M, Sun F, Velho G, 1979 1984 Zouali H, Butel MO, Lesage S, Vionnet 60. Haneda M, Chan SJ, Kwok SCM, Ruben- 39. Henry RR, Wallace P, Olefsky JM: Effects N, Clement K, Fougerousse F, et al.: stein AH, Steiner DF: Studies on mutant of weight loss on mechanisms of hyper- Close linkage of glucokinase locus on human insulin genes: identification and glycemia in obese non-insulin-depen- chromosome 7p to early-onset non-in- sequence analysis of a gene encoding dent diabetes mellitus. Diabetes 35:990– sulin-dependent diabetes mellitus. Na- [SerB24]insulin. Proc Natl Acad Sci USA 998, 1986 ture 356:162–164, 1992 80:6366–6370, 1983 40. Wing RR, Blair EH, Bononi P, Marcus 50. Vionnet N, Stoffel M, Takeda J, Yasuda 61. Given BD, Mako ME, Tager HS, Baldwin MD, Watanabe R, Bergman RN: Caloric K, Bell GI, Zouali H, Lesage S, Velho G, D, Markese J, Rubenstein AH, Olefsky J, restriction per se is a significant factor in Iris F, Passa P, et al.: Nonsense mutation Kobayashi M, Kolterman O, Poucher R: improvements in glycemic control and in the glucokinase gene causes early-on- Diabetes due to secretion of an abnormal insulin sensitivity during weight loss in set non-insulin-dependent diabetes mel- insulin. N Engl J Med 302:129–135, obese NIDDM patients. Diabetes Care litus. Nature 356:721–722, 1992 1980 17:30–36, 1994 51. Bell GI, Xiang K, Newman MV, Wu S, 62. Kahn CR, Flier JS, Bar RS, Archer JA, 41. Harris MI, Couric CC, Reiber G, Boyko Wright LG, Fajans SS, Spielman RS, Cox Gorden P, Martin MM, Roth J: The syn- E, Stern M, Bennett P (Eds.): Diabetes in NJ: Gene for non-insulin-dependent di- dromes of insulin resistance and acan- America. 2nd ed. Washington, DC, U.S. abetes mellitus (maturity-onset diabetes thosis nigricans. N Engl J Med 294:739– Govt. Printing Office, 1995 (NIH publ. of the young subtype) is linked to DNA 745, 1976 no. 95–1468) polymorphism on human chromosome 63. Taylor SI: Lilly Lecture: molecular 42. Newman B, Selby JV, Slemenda C, Fab- 20q. Proc Natl Acad Sci 88:1484–1488, mechanisms of insulin resistance: les- sitz R, Friedman GD: Concordance for 1991 sons from patients with mutations in type 2 (non-insulin-dependent) diabetes 52. Yamagata K, Furuta H, Oda N, Kaisaki the insulin-receptor gene. Diabetes 41: mellitus in male twins. Diabetologia 30: PJ, Menzel S, Cox NJ, Fajans SS, Sig- 1473–1490, 1992 763–738, 1987 norini S, Stoffel M, Bell GI: Mutations in 64. Ciaraldi TP, El-Roeiy A, Madar Z, 43. Barnett AH, Eff C, Leslie RDG, Pyke DA: the hepatocyte factor-4␣ gene in matu- Reichart D, Olefsky JM, Yen SSC: Cellu- Diabetes in identical twins. Diabetologia rity-onset diabetes of the young (MODY lar mechanisms of insulin resistance in 20:87–93, 1981 1). Nature 384:458–460, 1996 polycystic ovarian syndrome. J Clin En- 44. Herman WH, Fajans SS, Oritz FJ, Smith 53. Reardon W, Ross RJM, Sweeney MG, docrinol Metab 75:577–583, 1992 MJ, Sturis J, Bell GI, Polonsky KS, Halter Luxon LM, Pembrey ME, Harding AE, 65. Dunaif A, Segal KR, Shelley DR, Green JB: Abnormal insulin secretion, not in- Trembath RC: Diabetes mellitus associ- G, Dobrjansky A, Licholai T: Evidence sulin resistance, is the genetic or primary ated with a pathogenic point mutation in for distinctive and intrinsic defects in in- defect of MODY in the RW pedigree. Di- mitochondrial DNA. Lancet 340:1376– sulin action in polycystic ovary syn- abetes 43:40–46, 1994 1379, 1992 drome. Diabetes 41:1257–1266, 1992 45. Byrne MM, Sturis J, Menzel S, Yamagata 54. van den Ouwenland JMW, Lemkes 66. Schwartz SS, Zeidler A, Moossa AR, K, Fajans SS, Dronsfield MJ, Bain SC, HHPJ, Ruitenbeek W, Sandkuijl LA, de Kuku SF, Rubenstein AH: A prospec- Hattersley AT, Velho G, Froguel P, Bell Vijlder MF, Struyvenberg PAA, van de tive study of glucose tolerance, insulin, GI, Polonsky KS: Altered insulin secre- Kamp, Maassen JA: Mutation in mito- C-peptide, and glucagon responses in tory response to glucose in diabetic and chondrial tRNA (Leu(URR) gene in a patients with pancreatic carcinoma. Di- nondiabetic subjects with mutations in large pedigree with maternally transmit- gestive Dis 23:1107–1114, 1978 the diabetes susceptibility gene MODY3 ted type II diabetes mellitus and deaf- 67. Cersosimo E, Pister PWT, Pesola G, Mc- on chromosome 12. Diabetes 45:1503– ness. Nature Genet 1:368–371, 1992 Dermott K, Bajorunas D, Brennan MF: 1510, 1996 55. Kadowaki T, Kadowaki H, Mori Y, Tobe Insulin secretion and action in patients 46. Clement K, Pueyo ME, Vaxillaire M, Ra- K, Sakuta R, Suzuki Y, Tanabe Y, Sakura with pancreatic cancer. Cancer 67:486– kotoambinina B, Thuillier F, Passa P, H, Awata T, Goto Y, et al.: A subtype of 493, 1991 Froguel P, Roberts J, Velho G: Assess- diabetes mellitus associated with a mu- 68. Larsen S, Hilsted J, Tronier B, Worning ment of insulin sensitivity in glucoki- tation of mitochondrial DNA. N Engl H: Metabolic control and B cell function nase-deficient subjects. Diabetologia 39: J Med 330:962–968, 1994 in patients with insulin-dependent dia- 82–90, 1996 56. Johns DR: Mitochondrial DNA and dis- betes mellitus secondary to chronic pan- 47. Vaxillaire M, Boccio V, Philippi A, Vi- ease. N Engl J Med 333:638–644, 1995 creatitis. Metabolism 36:964–967, 1987 gouroux C, Terwilliger J, Passa P, Beck- 57. Gruppuso PA, Gorden P, Kahn CR, 69. Phelps G, Chapman I, Hall P, Braund W, man JS, Velho G, Lathrop GM, Froguel Cornblath M, Zeller WP, Schwartz R: Fa- Mackinnon M: Prevalence of genetic P: A gene for maturity onset diabetes of milial hyperproinsulinemia due to a pro- haemochromatosis among diabetic pa- the young (MODY) maps to chromo- posed defect in conversion of proinsulin tients. Lancet ii:233–234, 1989 some 12q. Nature Genet 9:418–423, to insulin. N Engl J Med 311:629–634, 70. Handwerger S, Roth J, Gorden P, Di 1995 1984 Sant’ Agnese P, Carpenter DF, Peter G: 48. Yamagata K, Oda N, Kaisaki PJ, Menzel 58. Robbins DC, Shoelson SE, Rubenstein Glucose intolerance in cystic fibrosis. S, Furuta H, Vaxillaire M, Southarm L, AH, Tager HS: Familial hyperproin- N Engl J Med 281:451–461, 1969

S18 DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 Committee Report

71. Yajnik CS, Shelgikar KM, Naik SS, Kan- High frequency of diabetes mellitus in 99. O’Sullivan JB: The Boston Gestational itkar SV, Orskov H, Alberti KGMM, young patients with congenital rubella; Diabetes Studies: review and perspec- Hockaday TDR: The ketoacidosis-re- Lancet ii:332–334, 1971 tive. In Carbohydrate Metabolism in Preg- sistance in fibro-calculous-pancreatic- 86. King ML, Bidwell D, Shaikh A, Voller A, nancy and the Newborn. Sutherland HW, diabetes. Diabetes Res Clin Pract 15:149– Banatvala JE: Coxsackie-B-virus-specific Stowers JM, Pearson DWM, Eds. Lon- 156, 1992 IgM responses in children with insulin- don, Springer-Verlag, 1989, p. 187–294 72. Soffer LJ, Iannaccone A, Gabrilove JL: dependent (juvenile-onset; type I) dia- 100. O’Sullivan JB: Diabetes after GDM. Dia- Cushing’s syndrome. Am J Med 30:129– betes mellitus. Lancet i:1397–1399, 1983 betes 40 (Suppl. 2):131–135, 1991 146, 1961 87. Karjalainen J, Knip M, Hyoty H, Linikki 101. Metzger BE, Cho NH, Roston SM, Rad- 73. Jadresic A, Banks LM, Child DF, Dia- P, Ilonen J, Kaar M-L, Akerblom HK: vany R: Prepregnancy weight and ante- mant L, Doyle FH, Fraser TR, Joplin GF: Relationship between serum insulin an- partum insulin secretion predict glucose The acromegaly syndrome. QJ Med 202: tibodies, islet cell antibodies and Cox- tolerance five years after gestational dia- 189–204, 1982 sackie-B4 and mumps virus-specific betes mellitus. Diabetes Care 16:1598– 74. Stenstrom G, Ernest I, Tisell L: Long- antibodies at the clinical manifestation 1605, 1993 term results in 64 patients operated up- of type I (insulin-dependent) diabetes. 102. Coustan DR, Carpenter MW, O’Sullivan on for pheochromocytoma. Acta Med Diabetologia 31:146–152, 1988 PS, Carr SR: Gestational diabetes: pre- Scan 223:345–352, 1988 88. Pak CY, Eun H, McArthur RG, Yoon J: dictors of subsequent disordered glu- 75. Berelowitz M, Eugene HG: Non-insulin Association of cytomegalovirus-infec- cose metabolism. Am J Obstet Gynecol dependent diabetes mellitus secondary tion with autoimmune type 1 diabetes. 168: 1139–1145, 1993 to other endocrine disorders. In Diabetes Lancet ii:1–4, 1988 103. Kjos SL, Buchanan TA, Greenspoon JS, Mellitus. LeRoith D, Taylor SI, Olefsky 89. Solimena M, Folli, Aparisi R, Pozza G, Montoro M, Bernstein GS, Mestman JH: JM, Eds. New York, Lippincott-Raven, De Camilli P: Autoantibodies to GABA- Gestational diabetes: the prevalence of 1996, p. 496–502 nergic neurons and pancreatic beta cells glucose intolerance and diabetes melli- 76. Conn JW: Hypertension, the potassium in stiffman syndrome. N Engl J Med 41: tus in the first two months postpartum. ion and impaired carbohydrate tolerance. 347–353, 1992 Am J Obstet Gynecol 163:93–98, 1990 N Engl J Med 273:1135–1143, 1965 90. Rimoin DL: Genetic syndromes associ- 104. Diabetes and pregnancy. In ACOG Tech- 77. Pandit MK, Burke J, Gustafson AB, Mi- ated with glucose intolerance. In The Ge- nical Bulletin 200, 1994 nocha A, Peiris AN: Drug-induced dis- netics of Diabetes Mellitus. Berlin, 105. Carpenter MW, Coustan DR: Criteria for orders of glucose tolerance. Ann Int Med Springer-Verlag, 1976 screening tests for gestational diabetes. 118:529–540, 1993 91. Barrett TG, Bundey SE, Macleod AF: Neu- Am J Obstet Gynecol 144:768–773, 1982 78. O’Byrne S, Feely J: Effects of drugs on rodegeneration and diabetes: UK nation- 106. Sacks DA, Abu-Fadil S, Greenspoon JS: glucose tolerance in non-insulin-depen- wide study of Wolfram (DIDMOAD) Do the current standards for glucose tol- dent diabetes (parts I and II). Drugs 40: syndrome. Lancet 346:1458–1463, 1995 erance testing in pregnancy represent a 203–219, 1990 92. Metzger BE, Organizing Committee: valid conversion of O’Sullivan’s original 79. Bouchard P, Sai P, Reach G, Caubarrere Summary and recommendations of the criteria? Am J Obstet Gynecol 161:638– I, Ganeval D, Assan R: Diabetes mellitus Third International Workshop-Confer- 641, 1989 following pentamidine-induced hypo- ence on Gestational Diabetes Mellitus. 107. Neiger R, Coustan DR: Are the current glycemia in humans. Diabetes 31:40–45, Diabetes 40:197–201, 1991 ACOG glucose tolerance test criteria 1982 93. Engelgau MM, Herman WH, Smith sensitive enough? Obstet Gynecol 78: 80. Assan R, Perronne C, Assan D, Chotard PJ, German RR, Aubert RE: The epide- 1117–1120, 1991 L, Mayaud C, Matheron S, Zucman D: miology of diabetes and pregnancy in 108. Naylor CD, Sermer M, Chen E, Sykora Pentamidine-induced derangements of the U.S., 1988. Diabetes Care 18:1029– K: Caesarean delivery in relation to birth glucose homeostasis. Diabetes Care 18: 1033, 1995 weight and gestational glucose toler- 47–55, 1995 94. Coustan DR: Gestational diabetes. In Di- ance. JAMA 275:1265–1270, 1996 81. Gallanosa AG, Spyker DA, Curnow RT: abetes in America. 2nd ed. Washington, 109. Pettitt DJ, Bennett PH, Hanson RL, Diabetes mellitus associated with auto- DC, U.S. Govt. Printing Office, 1995 Narayan KMV, Knowler WC: Compari- nomic and peripheral neuropathy after (NIH publ. no. 95–1468), p. 703–717 son of World Health Organization and Vacor poisoning: a review. Clin Toxicol 95. Langer O, Rodriguez DA, Xenakis EMJ, National Diabetes Data Group proce- 18:441–449, 1981 McFarland MB, Berkus MD, Arrendon- dures to detect abnormalities of glucose 82. Esposti MD, Ngo A, Myers MA: Inhibi- do F: Intensified versus convention- tolerance during pregnancy. Diabetes tion of mitochondrial complex I may al management of gestational diabetes. Care 17:1264–1268, 1994 account for IDDM induced by intoxica- Am J Obstet Gynecol 170:1036–1047, 110. Sacks DA, Greenspoon JS, Abu-Fadil S, tion with rodenticide Vacor. Diabetes 45: 1994 Henry HM, Wolde-Tsadik G, Yao JF: 1531–1534, 1996 96. Magee MS, Walden CE, Benedetti TJ: Toward universal criteria for gestational 83. Fabris P, Betterle C, Floreani A, Greggio Influence of diagnostic criteria on the in- diabetes: the 75-gram glucose tolerance NA, de Lazzari F, Naccarato R, Chiar- cidence of gestational diabetes and peri- test in pregnancy. Am J Obstet Gynecol amonte M: Development of type 1 dia- natal morbidity. JAMA 269:609–615, 172:607–614, 1995 betes mellitus during interferon alpha 1993 111. Deerochanawong C, Putiyanum C, therapy for chronic HCV hepatitis. Lan- 97. Cousins L: Obstetric complications. In Wongsuryrat M, Serirat S, Jinayon P: cet 340:548, 1992 Diabetes Mellitus and Pregnancy: Princi- Comparison of National Diabetes Data 84. Shiba T, Morino Y, Tagawa K, Fujino H, ples and Practice. 2nd ed. New York, Group and World Health Organization Unuma T: Onset of diabetes with high Churchill Livingstone, 1995, p. 455– criteria for detecting gestational diabetes titer anti-GAD antibody after IFN ther- 468 mellitus. Diabetologia 39:1070–1073, apy for chronic hepatitis. Diabetes Res 98. O’Sullivan JB, Mahan CM: Criteria for 1996 Clin Pract 30:237–241, 1996 the oral glucose tolerance test in preg- 111a.Metzger BE, Coustan DR: Summary and 85. Forrest, JA, Menser MA, Burgess JA: nancy. Diabetes 13:278, 1964 recommendations of the Fourth Interna-

DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002 S19 Committee Report

tional Workshop-Conference on Gesta- sylated hemoglobin to oral glucose betologia 39:298–305, 1996 tional Diabetes Mellitus. Diabetes Care tolerance: implications for diabetes 134. Finch CF, Zimmet PA, Alberti KGMM: 21 (Suppl. 2):B161–B167, 1998 screening. Diabetes 37:60–64, 1988 Determining diabetes prevalence: a ra- 112. Marquette GP, Klein VR, Niebyl JR: Effi- 124. Reaven GM: Role of insulin resistance in tional basis for the use of fasting plasma cacy of screening for gestational diabe- human disease. Diabetes 37:1595–607, glucose concentrations? Diabet Med tes. Am J Perinatology 2:7–14, 1985 1988 7:603–610, 1990 113. Dietrich ML, Dolnicek TF, Rayburn WR: 125. McCance, DR, Hanson RL, Pettitt DJ, 135. Peters AL, Davidson MB, Schriger DL, Gestational diabetes screening in a pri- Bennett PH, Hadden DR, Knowler WC: Hasselblad V, the Meta-analysis Re- vate, midwestern American population. Diagnosing diabetes mellitus: do we search Group on the Diagnosis of Diabe- Am J Obstet Gynecol 156:1403–1408, need new criteria? Diabetologia 40:247– tes Using Glycated Hemoglobin Levels 1987 255, 1997 1996: A clinical approach for the diag- 114. Lucas MJ, Lowe TW, Bowe L, McIntire 126. Knowler, WC: Screening for NIDDM: nosis of diabetes mellitus: an analysis us- DD: Class A1 gestational diabetes: a opportunities for detection, treatment ing glycated hemoglobin levels. JAMA meaningful diagnosis? Obstet Gynecol and prevention. Diabetes Care 17:445– 276:1246–1252, 1996 82:260–265, 1993 450, 1994 136. American Diabetes Association: Tests of 115. Charles MA, Fontboune A, Thibult N, 127. Harris MI, Hadden WC, Knowler WC, glycemia in diabetes (Position State- Warnet JM, Rosselin GE, Eschwege E: Bennett PH: Prevalence of diabetes and Risk factors for NIDDM in white popu- impaired glucose tolerance and plasma ment). Diabetes Care 25 (Suppl. 1):S97– lation: Paris Prospective Study. Diabetes glucose levels in the U.S. population S99, 2002 40:796–799, 1991 aged 20–74 yr. Diabetes 36:523–534, 137. American Diabetes Association: Stan- 116. Brunzell JD, Robertson RP, Lerner RL, 1987 dards of medical care for patients with Hazzard WR, Ensink JW, Bierman EL, 128. Stolk RP, Orchard TJ, Grobbee DE: Why diabetes mellitus (Position Statement). Porte D Jr: Relationships between fasting use the oral glucose tolerance test? Dia- Diabetes Care 25 (Suppl. 1):S33–S49, plasma glucose levels and insulin secre- betes Care 18:1045–1049; 1995 2002 tion during intravenous glucose toler- 129. McCance DR, Hanson RL, Charles MA, 138. Mayrhofer M, Rabin DU, Messenger L, ance tests. J Clin Endocrinol Metab Jacobsson LTH, Pettitt DJ, Bennett PH, Standl E, Ziegler AG: Value of ICA512 42:222–229, 1976 Knowler WC: Comparison of tests for antibodies for prediction and diagnosis 117. Fuller JH, Shipley MJ, Rose G, Jarrett RJ, glycated haemoglobin and fasting and of type 1 diabetes. Exp Clin Endocrinol Keen H: Coronary-heart disease risk and two hour plasma glucose concentrations Diab 104:228–234, 1996 impaired glucose tolerance: the White- as diagnostic methods for diabetes. BMJ 139. Bingley PJ, Christie MR, Bonifacio E, hall Study. Lancet i:1373–1376, 1980 308:1323–1328, 1994 Bonfanti R, Shattock M, Fonte M-T, 118. Charles MA, Balkau B, Vauzelle-Kervoë- 130. Engelgau MM, Thompson TJ, Herman Bottazzo G-F, Gale EAM: Combined den F, Thibult N, Eschwe`ge E: Revision WH, Boyle JP, Aubert RE, Kenny SJ, analyses of autoantibodies improves of diagnostic criteria for diabetes (Let- Badran A, Sous ES, Ali MA: Comparison prediction of IDDM in islet cell an- ter). Lancet 348:1657–1658, 1996 of fasting and 2-hour glucose and HbA1c tibody-positive relatives. Diabetes 43: 119. Jarrett RJ, Kahn H: Hyperglycemia and levels for diagnosing diabetes: diagnos- 1304–1310, 1994 diabetes mellitus. Lancet ii:1009–1012, tic criteria and performance revisited. 140. Verge CF, Gianani R, Kawasaki E, Chase 1976 Diabetes Care 20:785–791, 1997 H, Eisenbarth GS: Prediction of type 1 120. Klein R, Comor EB, Blount BA, Wingard 131. Jackson CA, Yudkin JS, Forrest RD: A diabetes in first degree relatives using DL: Visual impairment and retinopathy comparison of the relationships of the a combination of insulin, GAD and in people with normal glucose tolerance, glucose tolerance test and the glycated ICA512bc/ IA-2 autoantibodies. Diabe- impaired glucose tolerance and newly haemoglobin assay with diabetic vascu- tes 45:926–933, 1996 diagnosed NIDDM. Diabetes Care lar disease in the community: the Isling- 141. Hagopian WA, Sanjeevi CB, Kockum 14:914–918, 1991 ton Diabetes Survey. Diabetes Res Clin I, Landin-Olsson M, Karlsen AE, Sund- 121. McCartney P, Keen H, Jarrett RJ: The Pract 17:111–123, 1992 Bedford Survey: observations on retina 132. Beks PJ, Mackay AJC, de Neeling JND, kvist G, Dahlquist G, Palmer J, Lern- and lens of subjects with impaired glu- de Vries H, Bouter LM, Heine RJ: Pe- mark A: Glutamate decarboxylase-, cose tolerance and in controls with nor- ripheral arterial disease in relation to insulin-, and islet cell-antibodies and mal glucose tolerance. Diabete Metab glycaemic level in elderly Caucasian HLA typing to detect diabetes in a gen- 9:303–305, 1983 population: the Hoorn Study. Diabetolo- eral population-based study of Swedish 122. Reaven GM, Olefsky J, Farquhar JW: gia 38:86–96, 1995 children. J Clin Invest 95:1505–1511, Does hyperglycemia or hyperinsulin- 133. Mooy JM, Gootenhuis PA, de Vries H, 1995 aemia characterize the patient with Kostense PJ, Popp-Snijders C, Bouter 142. Harris MI: Undiagnosed NIDDM: clini- chemical diabetes? Lancet i:1247–1249, LM, Heine RJ: Intra-individual variation cal and public health issues. Diabetes 1972 of glucose, specific insulin and proinsu- Care 16:642–652, 1993 123. Little RR, England JD, Wiedmeyer H-M, lin concentrations measured by two oral 143. Klein R: Hyperglycemia and microvas- McKenzie EM, Pettit DJ, Knowler WC, glucose tolerance tests in general Cauca- cular and macrovascular disease in dia- Goldstein DE: Relationship of glyco- sian population: the Hoorn Study. Dia- betes. Diabetes Care 18:258–268, 1995

S20 DIABETES CARE, VOLUME 25, SUPPLEMENT 1, JANUARY 2002