Genetically Defined Forms of Diabetes in Children

ALTHOUGH type 1 diabetes has its highest incidence in the pediatric age range, there are also other forms of diabetes that affect children. From a clinical point of view, it is important to recognize these other forms of diabetes because of the important implications with respect to both prognosis and therapy. In addition, these relatively uncommon forms of diabetes provide an instructive model for the pathophysiological mechanisms that cause diabetes. Some syndromes are caused by insulin deficiency[ e.g. maturity-onset type diabetes of the young (MODY)]. Unlike patients with type 1 diabetes who exhibit near total absence of insulin due to autoimmune destruction of the β cell, MODY is characterized by a milder form of insulin deficiency caused by mutations in genes required for normal β cell function (1–6). However, not all cases of diabetes are caused by insulin deficiency. Indeed, some patients develop diabetes even though their β cells secrete sufficient insulin to achieve significant hyperinsulinemia (e.g. lipoatrophic diabetes and syndromes due to mutations in the insulin receptor gene) (7–10). MODY was originally defined by three diagnostic criteria (1): 1) onset at less than 25 yr of age; 2) autosomal dominant inheritance; and 3) noninsulin-dependent physiology. However, recent progress has led to the identification of at least five genes that are responsible for causing MODY (2–6). Thus, based on the locus of the mutation causing disease, the heterogeneous group of MODY patients can be subdivided into relatively homogeneous groups of patients with subtypes of MODY (see below). Assignment to a specific subtype of MODY has important prognostic implications. For example, MODY2 is characterized by mild hyperglycemia that has onset early in childhood and is nonprogessive (11). Indeed, only half the patients have sufficient hyperglycemia to meet the criteria for diagnosis of diabetes. As predicted by the mild degree of hyperglycemia, MODY2 does not carry a high risk of developing chronic complications of diabetes. In contrast, the degree of hyperglycemia and the risk of diabetic complications is much greater in MODY1, MODY3, and MODY5 (1). MODY4 seems to be intermediate in severity (5). Different authorities have published various estimates of the prevalence of the subtypes of MODY, but MODY2 and MODY3 are the most common, together accounting for up to 80% of cases. Like MODY, the syndrome due to mutations in the insulin gene has early onset and is transmitted with autosomal dominant inheritance (12). Although heterozygosity for a mutation in the insulin gene does not necessarily cause diabetes, it seems to be a genetic risk factor that predisposes to the development of insulin deficiency and diabetes in some patients. However, unlike MODY, patients with mutations in the insulin gene have elevated levels of circulating immunoreactive insulin because of the accumulation of abnormal molecules with markedly reduced biological activity. Diabetes due to mutations in mitochondrial DNA is transmitted with matrilineal inheritance and incomplete penetrance. In pedigrees where the mutation is transmitted to daughters (who can transmit the mutation to their own children), the pattern can resemble autosomal inheritance. Furthermore, because this syndrome may lead to early-onset diabetes, it may occasionally be confused with MODY. However, unlike MODY, diabetes due to mutations in mitochondrial DNA is often associated with other clinical features, such as deafness (13–15). Wolfram’s syndrome is another genetic form of diabetes mellitus with onset in childhood, but this syndrome has multiple features that distinguish it from MODY: autosomal recessive inheritance and association with optic atrophy, diabetes insipidus, and deafness. The gene for Wolfram’s has been mapped to chromosome 4p and has recently been cloned (16). Glucokinase catalyzes the first step in intracellular glucose metabolism (i.e. phosphorylation of glucose) (17, 18). Whereas hexokinases are capable of phosphorylating all hexoses, glucokinase is specific for glucose. In addition, whereas hexokinases have such low Km that they are saturated at physiological concentrations of glucose, glucokinase has a Km of ≈10 mm so that the flux through the enzyme is sensitive to changes in the glucose concentration within the the physiological range. Furthermore, glucokinase is specifically expressed in hepatocytes and pancreatic β cells, where it plays a crucial role in glucose metabolism. Abundant evidence has demonstrated that glucokinase is a necessary component in the β cell glucose-sensor. Indeed, the fact that inactivating mutations in the glucokinase gene lead to impaired insulin secretion and diabetes (MODY2) provides one of the most convincing pieces of evidence proving the role of glucokinase in regulating insulin secretion (3). This conclusion is further reinforced by the observation that an activating mutation in glucokinase led to hyperinsulinism and hypoglycemia (19). Consistent with the dominant inheritance in this syndrome, patients with MODY2 have inherited one mutant allele of the glucokinase gene. Furthermore, studies in an animal model have demonstrated that selective knock-out of glucokinase expression in the β cell can cause diabetes, suggesting that impaired insulin secretion is the major mechanism whereby these mutations cause diabetes (20, 21). Nevertheless, there is some evidence that the defect in hepatic glucokinase may also contribute to the pathophysiology of the disease. The remaining four subtypes of MODY are caused by mutations in the genes encoding transcription factors that regulate the expression of genes in the β cell (22): hepatocyte nuclear factor-4α (HNF-4α) in MODY1 (2), HNF-1α in MODY 3 (4), insulin promoter factor-1 (IPF-1) in MODY4 (5), and HNF-1β in MODY5 (6). IPF-1 contributes to the regulation of the insulin gene as well as additional genes that are important for β cell function. Indeed, homozygosity for inactivating mutations of IPF-1 leads to pancreatic agenesis (23). [In contrast, heterozygosity for a mutation in IPF-1 causes MODY (5).] As suggested by their names, the family of HNF also plays a role in regulation of genes expressed in the liver. Although it is likely that mutations in HNF-1α, -1β and -4α cause diabetes because they impair insulin secretion, it is possible that metabolic defects within the hepatocyte also contribute to disease pathogenesis. Two clinical features are commonly observed in patients with extreme insulin resistance, regardless of the pathogenetic mechanism causing the disease: acanthosis nigricans and hyperandrogenism (7–9, 24). Acanthosis nigricans is a hyperkeratotic, hyperpigmented skin lesion located primarily in skin folds such as the axillae and antecubital fossae (Fig. 1). Hyperandrogenism results from hypersecretion of androgens by the ovaries (7, 24). Clinically, the elevated levels of testosterone are manifested as a syndrome of polycystic ovaries, oligomenorrhea, hirsutism, and, sometimes, virilization. Several lines of evidence suggest that both acanthosis nigricans and hyperandrogenism are caused by hyperinsulinemia—presumably exerting “toxic” effects on the skin and ovaries. For example, when patients with type B insulin resistance enter remission because of a decrease in the titer of autoantibodies directed against the insulin receptor, this results in amelioriation (even disappearance) of acanthosis nigricans and hyperandrogenism. Similarly, in insulin-resistant patients with polycystic ovarian syndrome, administration of drugs such as metformin and thiazolidinediones leads to a decrease in plasma insulin levels and a concomitant amelioriation in hyperandrogenism (25, 26). Although the mechanism whereby hyperinsulinemia elicits these toxic effects has not been elucidated, it is likely not to be mediated by conventional insulin receptors inasmuch as these are functionally impaired as the result of inactivating mutations. Photograph of patient with Rabson-Mendenhall syndrome. Photograph of patient with Rabson-Mendenhall syndrome. In addition to the clinical signs that the various syndromes share in common, distinct syndromes are characterized by the presence or absence of specific features. For example, there are at least three clinical syndromes caused by mutations in the insulin receptor gene. Type A insulin resistance is defined by the triad of insulin resistance, acanthosis nigricans, and hyperandrogenism in the absence of obesity or lipoatrophy (27). Patients with leprechaunism have multiple abnormalities, including intrauterine growth retardation, fasting hypoglycemia, and death within the first 1–2 yr of life (Fig. 2) (28–30). Interestingly, patients with leprechaunism are reported to exhibit striking degrees of β cell hyperplasia. Thus, near-total absence of insulin receptors does not impair the ability of human β cells to become hyperplastic, a reaction that can be rationalized as an attempt to compensate for severe insulin resistance. The Rabson-Mendenhall syndrome is associated with short stature, protruberant abdomen, and abnormalities of teeth and nails (Fig. 3); pineal hyperplasia was a characteristic in the original description of this syndrome (31). In addition to genetic defects in insulin receptor function, there is an autoimmune form of insulin resistance due to autoantibodies to the insulin receptor (type B insulin resistance) (27). Although this syndrome has been reported in the pediatric age range, most of the patients have onset during adulthood. Photograph of patient with leprechaunism. Photograph of patient with leprechaunism. Photograph of patient with congenital generalized lipoatrophy (Seip-Berardinelli syndrome). Photograph of patient with congenital generalized lipoatrophy (Seip-Berardinelli syndrome). Lipoatrophic diabetes, another clinical syndrome associated with insulin resistance, is characterized by a paucity of subcutaneous fat, hypertriglyceridemia, and hepatic steatosis. Although the mutations causing genetic forms of lipoatrophic diabetes have not yet been identified, current evidence suggests that insulin resistance is secondary to the lipoatrophy and there may not be a primary genetic defect in the insulin action pathway (32, 33). The genes causing two forms of the lipoatrophy have been mapped: an autosomal recessive form of congenital, generalized lipoatrophy (Seip-Berardinelli syndrome; Figs. 3 and 44) to chromosome 9q34 (34), and an autosomal dominant form of partial (face-sparing) lipodystrophy with onset during adolescence (Dunnigan’s syndrome; Figs. 4 and 55) to chromosome 1q21–22 (35). In addition to the genetic forms of lipoatrophy, there are also syndromes that are believed to have autoimmune pathogenesis inasmuch as they are associated with well-defined autoimmune diseases, such as juvenile dermatomyositis. Magnetic resonance imaging of the mid-thigh region of patients with lipoatrophy. A, A normal 42-yr-old woman with a body mass index of 24 kg/m2. The normal individual demonstrates both sc adipose tissue (a circumferential band encircling the thigh) and deep adipose tissue located between muscle bundles. B, A 37-yr-old woman with Dunnigan’s syndrome. This shows a marked diminution in the sc fat, but relative preservation of deeper fat. C, a 29-yr-old woman with Seip-Berardinelli syndrome. This patient shows near total absence of both compartments of adipose tissue, but the brightness of the muscle density is indicative of increased triglyceride content in muscle tissue. Magnetic resonance imaging of the mid-thigh region of patients with lipoatrophy. A, A normal 42-yr-old woman with a body mass index of 24 kg/m2. The normal individual demonstrates both sc adipose tissue (a circumferential band encircling the thigh) and deep adipose tissue located between muscle bundles. B, A 37-yr-old woman with Dunnigan’s syndrome. This shows a marked diminution in the sc fat, but relative preservation of deeper fat. C, a 29-yr-old woman with Seip-Berardinelli syndrome. This patient shows near total absence of both compartments of adipose tissue, but the brightness of the muscle density is indicative of increased triglyceride content in muscle tissue. Photograph of patient with face-sparing lipoatrophy (Dunnigan’s syndrome). There is marked decrease in sc adipose tissue in the legs. Please note that this patient has undergone plastic surgery to achieve this normal facial appearance; before surgery, she had a marked increase in the fat in her face and neck. Photograph of patient with face-sparing lipoatrophy (Dunnigan’s syndrome). There is marked decrease in sc adipose tissue in the legs. Please note that this patient has undergone plastic surgery to achieve this normal facial appearance; before surgery, she had a marked increase in the fat in her face and neck. Numerous (>70) mutations have been identified in the insulin receptor gene in various insulin-resistant patients. The mutations can be classified based on the mechanisms whereby they impair receptor function (8–10). In some cases, the mutations decrease the number of receptors expressed on the cell surface [e.g. by decreasing the rate of receptor biosynthesis (class 1), accelerating the rate of receptor degradation (class 5), or inhibiting the transport of receptors to the plasma membrane (class 2)]. In addition, some mutations impair the intrinsic function of the receptor[ e.g. by decreasing the affinity of insulin binding (class 3) or inactivating receptor tyrosine kinase activity (class 4)]. Leprechaunism and the Rabson-Mendenhall syndrome are both transmitted with autosomal recessive patterns of inheritance; affected individuals have inherited two inactivating mutations in the insulin receptor gene (36–38). There are two forms of type A insulin resistance: a severe form with autosomal recessive inheritance (39) and a milder form with autosomal dominant inheritance, most commonly caused by dominant negative mutations in the tyrosine kinase domain of the insulin receptor (40, 41). Among patients who have inherited two mutant alleles of the insulin receptor gene, it is not entirely clear why different patients develop different clinical syndromes. At least two different mechanisms have been proposed to correlate phenotype with genotype (8, 9, 42). First, the severity of the defect in receptor function seems to correlate with the severity of the clinical syndrome, with the most severe mutations causing leprechaunism and less severe mutations causing type A insulin resistance. The Rabson-Mendenhall syndrome is associated with defects that are intermediate in severity. Second, differences in genetic background (i.e. genotypes at other loci) may modulate the severity of the clinical syndrome. Class 1, decreased rate of receptor biosynthesis.Several genetic defects can interfere with receptor biosynthesis. In one patient with leprechaunism, there was a total deletion of the insulin receptor gene (43). In several patients, there appear to be regulatory mutations that decrease the level of insulin receptor messenger RNA without altering the protein coding sequence of the gene (37). Finally, various premature chain termination mutations can lead to the synthesis of truncated receptors that are essentially inactive (36–38). Class 2, impaired transport of receptors to the plasma membrane.Some mutations appear to prevent folding of receptors into their normal conformations. Various chaperone molecules within the endoplasmic reticulum serve a “quality control” function, and appear to target the receptors for intracellular degradation, resulting in inefficient transport to the plasma membrane (39, 44, 45). Most class 2 mutations have been identified in the N-terminal half of theα -subunit of the receptor (Fig. 6), but some mutations in the tyrosine kinase domain have also been reported to promote intracellular degradation (46). Mapping of mutations in the insulin receptor gene within the protein coding sequence of the complementary DNA. This figure depicts the regions of the insulin receptor that are the most common sites of missense mutations giving rise to the various classes of defects in receptor function. Mapping of mutations in the insulin receptor gene within the protein coding sequence of the complementary DNA. This figure depicts the regions of the insulin receptor that are the most common sites of missense mutations giving rise to the various classes of defects in receptor function. Class 3, decreased affinity of insulin binding.The N-terminal half of the α-subunit has been implicated in the ligand binding domain of the insulin receptor and the homologous type 1 insulin-like growth factor receptor. Thus, it is not surprising that most of the mutations that inhibit insulin binding map to that domain of the receptor (Fig. 6) (8–10). In addition, at least two class 3 mutations have been mapped near the C-terminal domain of the α-subunit. Class 4, impaired receptor tyrosine kinase activity.Most class 4 mutations are located in the tyrosine kinase domain in the intracellular portion of the β-subunit (Fig. 6) (8–10). However, mutations in the extracellular domain of the receptor have also been reported to uncouple insulin binding from the ability to activate the receptor tyrosine kinase (47). Unlike most other classes of mutation, mutations in the kinase domain of the receptor exert a dominant negative effect to cause insulin resistance (48, 49). The resulting dominant inheritance pattern may be explained as follows: formation of hybrid receptors (i.e. association of a mutantαβ -dimer to a mutant αβmutant-dimer) leads to the formation of a catalytically inactiveα 2ββmutant-tetramer. Thus, in addition to being inactive itself, the mutant receptor inhibit the activity of wild type receptors. Class 5, accelerated degradation of the receptor.This is the least common class of mutations, with only two mutant alleles having been reported to accelerate degradation of receptors after having been inserted in the plasma membrane (50, 51). Both of these mutations are located near the middle of the β-subunit: Lys460Glu (36, 50) and Asn462Ser (38, 51). In other words, although there seems not to be an intrinsic defect in the ability of these mutant receptors to mediate insulin action. Accelerated degradation of mutant receptors leads to a marked decrease in the number of receptors expressed on the plasma membrane. The impairment in the ability of the acid pH within endosomes to dissociate insulin from the receptor is associated with accelerated receptor degradation (36, 50, 52). Although the mechanism has not been elucidated, this change in pH sensitivity is associated with preferential sorting of internalized receptors away from the pathway for recycling to the plasma membrane and toward the pathway for lysosomal degradation. In the past, the term “juvenile diabetes” was used to refer to diabetes caused by autoimmune destruction of pancreatic β cells (now called type 1 diabetes). This term was discarded for a variety of reasons; among them, the fact that there are multiple types of diabetes with onset in childhood. As emphasized in this brief review, clinicians had previously described numerous forms of early-onset diabetes, some of which seemed to follow simple Mendelian forms of inheritance. As a result of the incredible progress in molecular genetics in recent years, the genes responsible for ≈10 forms of diabetes have been identified. These developments provide a paradigm for future progress toward the goal of using molecular medicine to diagnose specific forms of diabetes in the hope that this will lead to develop therapeutic approaches that are tailored to specific molecular defects. In some cases, this may turn out to be gene replacement therapy, but it is also likely that the efficacies of specific drugs will be shown to correlate with the patient’s genotype. Finally, molecular diagnosis holds the promise of diagnosing diabetes early in life, before the onset of overt diabetes. This will facilitate early interventions designed to prevent diabetes. This avenue of research has the potential to lead to significant advances that will benefit patients suffering from the disease.