| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
C.N.R. Institute of Clinical Physiology and Department of Internal Medicine, University of Pisa, 56126 Pisa, Italy
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. Preliminary IssuesIII. Insulin Resistance vs....IV. SummaryV. Insulin Resistance: The...VI. ProspectsReferences |
|---|
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. Preliminary IssuesIII. Insulin Resistance vs....IV. SummaryV. Insulin Resistance: The...VI. ProspectsReferences |
|---|
As often is the case, semantics provide some fuel to the controversy. Thus, the question of whether insulin deficiency or insulin resistance is the central element (or equivalent term) of NIDDM is intrinsically confusing. In fact,
1. Clinically, NIDDM is diagnosed on the basis of raised plasma glucose levels (no other substrate or hormone being considered) that can be controlled without exogenous insulin (although proneness to ketosis may eventually appear) (4). Such a broad definition pigeonholes a heterogeneous group of conditions, ranging from maturity-onset diabetes of the young (MODY) to the common form of NIDDM to latent autoimmune diabetes of the adult [or LADA, as evidenced by the finding of autoantibodies against a variety of islet cell and other antigens (5)]. Obesity, estimated to accompany NIDDM in 80% of the patients in the westernized world, is featured in some subclassifications of diabetes (6) but is not required for its diagnosis. As obesity is often associated with insulin resistance, the latter feature is passed on to NIDDM with high frequency. As a consequence, many NIDDM patients may be presumed to have some degree of insulin resistance and, by definition, every diabetic patient secretes less insulin than necessary for his/her level of insulin sensitivity. Moreover, the mix of insulin resistance and insulin deficiency is likely to be different in each patient and, in any patient, may vary during the course of the disease. Clinically, to regard insulin deficiency or insulin resistance as the central element of the individual patient with NIDDM is rather a matter of preference than an evidence-based judgment.
2. Pathophysiologically, both insulin resistance alone and insulin
deficiency alone can alter plasma glucose levels. Rare inherited forms
of extreme insulin resistance (7) and secondary diabetes (8) are clear
examples for each case. Common forms of hyperglycemia, however, are
mostly mixed cases. In fact, insulin secretion and action govern
glucose homeostasis in a dual regulatory cycle: a rise in glucose
concentration stimulates insulin secretion, which in turn lowers plasma
glucose in a time- and concentration-dependent manner (outer
circle in Fig. 1
). In addition,
sustained hyperinsulinemia inhibits both insulin secretion (9) and
action (10). In turn, chronic hyperglycemia can impair both the
insulin-secretory response to glucose (11) and cellular insulin
sensitivity [i.e., glucose toxicity (12)] (inner
circle in Fig. 1). Due to the operation of these loops, it is
a priori unlikely that chronic hyperglycemia is due
exclusively to a deficit in insulin secretion or to an isolated defect
in insulin action. Instead, the prevalent scenario must be one in which
chronic hyperglycemia is associated with a compound of insulin
deficiency and resistance, in proportions that are also influenced by
the relative gains and time constants of the feedback effects.
|
Whether insulin deficiency or insulin resistance is the primary
determinant in the etiology of NIDDM is altogether a
different question. In general, the etiology of NIDDM could follow
three models: genetic (e.g., Down’s syndrome),
environmental (e.g., trauma), or mixed. Once again, the
unanimous opinion is that both genetic and environmental influences
confer risk for NIDDM. Given that both insulin secretion and insulin
action are under genetic control, mutations in either set of genes
(those encoding insulin secretion and those responsible for insulin
action) could theoretically be the primary event in NIDDM. When either
mutation is functionally expressed in the phenotype as ß-cell failure
or insulin resistance, however, the companion abnormality is likely to
follow suit, according to the interactions in Fig. 1
. Equally
admissible is a paradigm in which an environmental factor initially
strikes either insulin secretion or insulin sensitivity, thereby
starting the cycle. The general question, then, is which
event—mutation or environmental insult, affecting insulin secretion or
insulin resistance—is the cause of typical NIDDM? The solution
theoretically admits numerous combinations, especially if there is
linkage dysequilibrium and/or interactions between genes as well as
between environment and genes. Thus, solving the etiology of NIDDM is a
highly complex undertaking. It may be useful to review the approaches
and identify the critical issues as they apply to in vivo
investigation.
|
II. Preliminary Issues |
|---|
|
TopAbstractI. IntroductionII. Preliminary IssuesIII. Insulin Resistance vs....IV. SummaryV. Insulin Resistance: The...VI. ProspectsReferences |
|---|
The lean, insulin-sensitive NIDDM phenotype is an interesting model. It attests to the possibility that an unknown insult to the ß-cell can be such as to produce a slowly evolving diabetic syndrome without marked insulin resistance. In these patients, it is possible that insulin sensitivity was supernormal in the prediabetic phase but decreased to ’normal’ levels once hyperglycemia ensued. The nature of their ß-cell dysfunction, genetic defect or environmental (nonautoimmune) damage, remains undetermined. To the best of available evidence, the prevalence of this type of NIDDM should be low, because filtering out MODY, LADA, and overweight patients leaves only a thin segment of the entire NIDDM population.
B. Methodology
The definitions of insulin deficiency and insulin resistance are
dependent on methodology. Insulin action in vivo can be
measured with the use of the euglycemic hyperinsulinemic clamp
technique, which has become the gold standard against which other
methods [insulin suppression test, minimal model analysis of the
intravenous glucose tolerance test (ivGTT), Constant Infusion of
Glucose with Model Assessment (CIGMA), Homeostatic Model Assessment
(HOMA), etc. (19)] are validated. Merits of the insulin clamp
are that its estimates are relatively free of assumptions, are derived
under conditions approximating a steady state, and are well
reproducible [
10% intraindividual coefficient of variation (19)].
One limitation is that the clamp, in its simplest version, offers a
point estimate of insulin action: one substrate (glucose), one stimulus
(insulin concentration x time), both at one level only. Full
dose-response curves for in vivo insulin-mediated glucose
uptake have been constructed by performing multiple euglycemic (20) or
hyperglycemic (21) clamps sequentially or on different days, but this
laborious approach is feasible only in small numbers of subjects. An
additional difficulty is that insulin sensitivity estimated during
insulin administration may not bear a close relation to the insulin
sensitivity of the fasting state, when hepatic glucose output and
non-insulin-dependent glucose utilization dominate glucose homeostasis.
Despite its limitations, the simple version of the euglycemic insulin
clamp has become prevalent enough to generate databases in Pima Indians
(22) and, more recently, in healthy white Europeans [the European
Group for the Study of Insulin Resistance (EGIR) study] (23).
The minimal model analysis of insulin sensitivity—probably the closest
relative of the insulin clamp—has also been applied to a
population-based sample of young healthy subjects (24) and to the
tri-ethnic cohort in the IRAS study (25). Thus, insulin resistance can
now be defined by statistical criteria, and modulation of insulin
action by such common factors as age, obesity, fat distribution, and
life-style has been given a quantitative description (22, 23, 24, 25).
There is no equivalent standard for insulin secretion. Acute insulin release (AIR), measured as the sum of the peripheral hormone concentrations during the initial 10 min of an ivGTT, has been interpreted as the unloading of preformed insulin in response to maximal impulsive stimulation (26). The plasma insulin concentrations measured at later times during the ivGTT represent augmented hormone synthesis in response to sustained glucose stimulation (26). Indeed, when in vivo insulin secretion is accurately reconstructed from peripheral insulin concentrations by means of deconvolution analysis, the ß-cell response to intravenous glucose is frequently multiphasic in healthy subjects, with progressively damped peaks of release in phase with plasma glucose oscillations (27). Continuous glucose infusion evokes ultradian insulin secretory oscillations (28), which are superimposed on the more rapid, spontaneous pulsatility of fasting ß-cell activity (29). A mixed meal also elicits a roughly biphasic pattern of insulin release, with a first peak of peripheral insulinemia at 30–45 min followed by a sustained response occurring in the face of steadily declining plasma glucose levels (30, 31). In all of these tests, in which the glucose stimulus is not standardized, the insulin secretory response is commonly assessed as the plasma insulin increment divided by the plasma glucose increment. The hyperglycemic version of the glucose clamp, on the other hand, can maintain a square wave of hyperglycemia. The insulin response to this format of glucose challenge is again biphasic, a typical AIR being followed by a gradually increasing secretory activity. The hyperglycemic clamp is optimal to compare responses to a single equipotent stimulus and to construct dose-response functions by creating ramp-like hyperglycemic plateaus (19). Finally, intravenous arginine pulses can be superimposed on hyperglycemia to explore glucose potentiation (26).
While much has been learned from the application of these techniques in physiological studies, the distribution and physiological correlates of specific insulin-secretory responses have not been well characterized. Only recently (24), modulation of AIR by gender, body fatness, physical fitness, and other life-style factors has been described at the population level in a sample of young healthy volunteers (24). In contrast, many small case-control studies have examined the pathogenetic significance of impaired insulin secretion. Thus, short (60-min) glucose infusions at fixed rates have been used to characterize low and high insulin responders (32). Loss of AIR has long been known to be a good predictor of subsequent development of NIDDM (33). A blunted early insulin response to a mixed meal has been shown to be in part responsible for the excessive prandial hyperglycemia of NIDDM patients (34). Even the ability of exogenous glucose infusions to entrain cyclic insulin secretory pulses has been shown to have discriminating power for ß-cell function in humans (28). Nevertheless, a number of important questions remain unanswered. For example, we do not know whether AIR is related to the oscillatory patterns of insulin release, whether it is influenced by habitual diet or antecedent physical activity, or whether it aggregates in the family [as does insulin resistance (35)]. Likewise, the relationship between AIR and the early insulin response to a mixed meal is poorly understood. In everyday life there is no AIR; the early prandial insulin response is much slower than AIR and is strongly influenced by the gastrointestinal release of insulinotrophic hormones. Yet, an impairment of the ’initial’ response of the ß-cell to either intravenous glucose or a meal is so consistently found in hyperglycemic states to constitute a hallmark of diabetes. The cellular basis for the predictivity of AIR for the global failure of insulin release that ultimately leads to hyperglycemia is obscure. Less specifically, we do not know whether and which other parameters of the complex insulin secretory dynamics relate or anticipate the ß-cell defect of NIDDM.
In summary, insulin action can be assessed in vivo with good accuracy, specificity, and reproducibility, and its measurement at the population level continues to produce valuable data. For insulin secretion, no single test appears to convey information as densely as the insulin clamp does for insulin sensitivity, and none has gained sufficient diffusion.
C. Source data
Information relevant to the etiology of NIDDM can be derived from
epidemiological surveys and case-control studies. Both sources have
inherent limitations. In epidemiological studies, in which only simple
measurements can usually be obtained, insulin sensitivity and insulin
secretion have both been inferred from plasma insulin concentrations,
either fasting or postglucose (36, 37, 38). The only published studies in
which insulin resistance has been measured by a direct method (the
insulin clamp) at the population level are those performed in the Pima
Indians of the Southwestern United States (22) and the EGIR study (23).
Although peripheral insulin levels do reflect insulin sensitivity, the
extent of this relationship can be appreciated from the EGIR data (Fig. 2): lower insulin sensitivity is
associated with higher fasting insulin concentrations, but the scatter
around the relationship is wide. Similar data relating insulin
secretion [independently measured from C-peptide kinetics (31)] and
plasma insulin concentrations are simply lacking. Thus, epidemiological
findings based on plasma insulin measurements, even when they are
consistent, are intrinsically insufficient to discriminate insulin
resistance from insulin deficiency.
|
The validity of case-control studies critically depends on the selection of subjects as well as on the matching of cases and controls on all factors known to affect the test variable. Here the disparity between current information on insulin resistance and insulin secretion becomes critical. Many factors that affect insulin sensitivity have been identified: age and gender (23), body fat (39) and its distribution (40), physical fitness (41), arterial blood pressure (42), family history of diabetes (43) or hypertension (44), smoking (45), and presence of ischemic heart disease (46, 47), to name only those that would be recorded in a clinical chart. In contrast, less is known about the physiological correlates of the insulin response to glucose (gender, ethnicity, antecedent diet, time of day, and duration of fast, etc.). Thus, even when cases and controls are matched on the main confounders of insulin sensitivity, comparison of ß-cell function [as done by Pimenta et al. (48) in first-degree relatives of NIDDM patients] may still be biased by some unmeasured confounder of insulin secretion that has no detectable effect on insulin action.
In summary, sources such as epidemiological and case-control studies are useful tools to identify patterns and disease-related changes but generally have limited power to establish the etiological primacy of events.
|
III. Insulin Resistance vs. Insulin Deficiency |
|---|
|
TopAbstractI. IntroductionII. Preliminary IssuesIII. Insulin Resistance vs....IV. SummaryV. Insulin Resistance: The...VI. ProspectsReferences |
|---|
Another model of antecedence is provided by subjects who carry a genetic risk of developing the disease later in life. In 12 pairs of identical twins discordant for NIDDM, Vaag et al. (54) detected both insulin resistance and delayed or reduced insulin response to oral or intravenous glucose in the nondiabetic twins whether they had normal or impaired glucose tolerance. A number of case-control studies have reported the presence of insulin resistance in nondiabetic first-degree relatives of diabetic patients, at a time when their glucose tolerance is still within normal lipits (e.g., Refs. 55, 56). Detailed metabolic investigations have traced the defect in insulin action to the skeletal muscle tissue (56), where stimulation of both the tyrosine kinase activity of the insulin receptor and glycogen synthase was found to be reduced (57, 58). In other studies in this high-risk group, however, insulin resistance was not a prominent feature (59), and impaired early-phase insulin release on a hyperglycemic clamp (48) or loss of the normal oscillatory pattern of insulin release (60) have been described. The available literature is interpreted by Gerich (61) as offering preponderant evidence in favor of normal insulin sensitivity in NIDDM relatives. More recently, however, an analysis of the largest published series of clamp studies (the EGIR study) showed that the subjects with a positive family history for NIDDM (n = 235) were significantly more insulin resistant (by 13% on average) than subjects without a family history of NIDDM (n = 564) after adjusting for gender, age, and body mass (62). Incidentally, given the high between-subject variability of insulin sensitivity in nondiabetic individuals [coefficient of variation of 30% in the lean and 40% in the obese (23)], a difference such as that found in the EGIR study could only be detected in a very large sample.
These discrepancies cannot be decided by simply weighing the negative against the positive findings (61). On the one hand, the disparity of results confirms that case-control studies can be misleading because of the potential for multiple errors due to small sample size, selection bias, and variable accuracy and precision of measurements. On the other hand, if NIDDM is multifactorial, there is no good reason to expect that the phenotype of individuals at risk of NIDDM should be uniform. With regard to this, a recent study (63) reported that, in lean, nondiabetic offspring of NIDDM parents, insulin sensitivity showed a trimodal distribution, but a clear deficit in AIR (on the ivGTT) was only detectable in the very insulin-resistant subgroup. Thus, the population of NIDDM offspring sampled in that study apparently was a mixture of subjects who had normal sensitivity and secretion, insulin resistance but adequate ß-cell function, or insulin resistance and insulin deficiency. Even more interestingly, another study (64) has shown that the distribution of defects, insulin deficiency or insulin resistance, in 38 nondiabetic offspring of one diabetic proband fell into line with that of the proband cohort: the offspring of diabetics with a low fasting C peptide level were insulin deficient but had normal insulin sensitivity, whereas the offspring of diabetics with a high fasting C peptide showed the reverse pattern. Significantly, the prevalence of IGT among this randomly selected sample was the same irrespective of the prevalent defect, suggesting that insulin resistance and insulin deficiency can be inherited separately and can produce equivalent amounts of glucose intolerance in the progeny.
Follow-up of cohorts at high risk would generate more compelling data. Thus, if a large number of at-risk subjects [with familial NIDDM or gestational diabetes (65), or homozygous twins discordant for NIDDM (54)] were accurately phenotyped for insulin action (by a clamp technique) as well as secretion (by a range of tests), and then followed up for a long enough time to accumulate incident NIDDM, comparison of insulin secretion and sensitivity between confirmed prediabetics and matched controls would suggest which abnormality is the earlier factor precipitating the disease in individuals with a high a priori risk. This approach was followed by Warram et al. (66) in a large group of nondiabetic offspring of two diabetic parents. On the ivGTT, both slower glucose removal rates and higher acute and late insulin responses characterized the offspring of diabetics, who then went on to develop NIDDM at a rate that was 8 times that of the general population.
Longitudinal observations of random samples of the general population
are commonly used to identify predictors of NIDDM. In prospective
studies, the presence of hyperinsulinemia in nondiabetic individuals
has predicted NIDDM consistently (37, 67). So has, however, the plasma
glucose concentration itself, independently of insulin (67). In other
words, NIDDM has been more frequent among individuals who at baseline
were hyperinsulinemic and had relatively higher plasma glucose levels
than among subjects with the opposite characteristics. What exactly do
such findings tell about etiology? An example may help clarify this
point (Fig. 3). Fasting plasma glucose is
a tracking variable: in an individual in whom its value falls into the
upper half of the distribution of the population, fasting glucose tends
to remain above the population mean over time despite swinging in
response to environmental stimuli. Plasma insulin concentrations will
also tend to be on the high side of the distribution, as tracking has
also been described for insulin levels in both children of school age
(68) and adults (69). Assuming the subject represents a subgroup of
nondiabetic individuals who are found to be hyperglycemic at follow-up,
bivariate analysis of development of diabetes in the population will
establish that diabetes was independently predicted by both higher
glucose and higher insulin levels. This statistical association does
not tell how higher baseline insulin levels led to incident
diabetes. In fact, insulin levels may have been elevated because of a
high gain for glucose-induced insulin secretion (primary
hypersecretion) or because of underlying insulin resistance
(compensatory hypersecretion). Indeed, insulin hypersecretion might
have been the initial abnormality, rapidly followed by down-regulation
of insulin action (69) and exhaustion of a predisposed ß-cell. On the
other hand, fasting glucose may have been set at a higher level by a
mechanism separate from insulin (e.g., glucagon excess).
These (and other) interpretations are all compatible with the primary
data. Thus, even in prospective, population-based studies,
hyperinsulinemia does suggest potential mechanisms for the development
of NIDDM but falls short of proving primacy in the etiology of the
condition.
|
In synthesis, available evidence from high-risk groups and prospective population studies demonstrates that both insulin resistance and its surrogate, fasting hyperinsulinemia, are associated with the development of NIDDM. The corresponding data on insulin secretory dysfunction are both less consistent and of inferior quality than those on insulin resistance. Nevertheless, the evidence is not fully sufficient to defend the primacy of either insulin resistance or insulin deficiency in the sense of 1) coming first in time, 2) having stronger genetic determination, and 3) being quantitatively more important.
B. The problem by mechanism
Ultimately, it may prove impossible to dissociate insulin
resistance from insulin deficiency precisely because changes in insulin
secretion and action are coordinated by strong physiological
interactions (71). How likely is this possibility?
Information about the relationship between insulin secretion and
insulin action is surprisingly meager. Insulin output in humans
increases as a sigmoidal function of circulating glucose
concentrations. This relationship has been interpreted as a dependence
on the actual level of plasma glucose as well as its rate of change
(27). Thus, a quick rise in glycemia stimulates, and a sudden drop
inhibits, insulin secretion independently of the initial and final
levels of glucose. Furthermore, glucose potentiation and inhibition
(72), neural influences (73), gastrointestinal potentiation (74),
spontaneous oscillatory cycles (29), and glucose toxicity to the
ß-cell (75) are all phenomena that have been characterized in
isolation. However, they have not been integrated into a comprehensive
model of ß-cell response in vivo. On the other side of the
feedback, the in vivo dose-response relationship between
insulin concentrations and glucose fluxes also is curvilinear (76),
compatible with saturation kinetics. It is therefore expected that the
relationship between insulin secretion and insulin action on glucose
levels be a highly nonlinear one. In fact, in a large group of lean and
obese subjects with normal glucose tolerance (77), declining insulin
sensitivity (as assessed by the clamp technique) was associated with a
hyperbolic increase in basal posthepatic insulin delivery rate (Fig. 4). Assuming that fractional hepatic
insulin extraction is 50%, and that 24-h insulin release is twice the
fasting rate of insulin output (31), it can be calculated (from the
extremes of the distribution shown in Fig. 4) that normal glucose
tolerance can be achieved with as little as 10, but may require as much
as 400, U of insulin per day. This huge range reflects the variability
of individual setpoints (i.e., the product of insulin
secretion and insulin sensitivity) as well as the impact of obesity.
Indeed, when body mass (BMI) is factored in, the relationship between
insulin sensitivity and secretion can be plotted as a simultaneous
function of degree of obesity. It can be seen (Fig. 5) that at each level of insulin
sensitivity, obesity strongly potentiates insulin release, and that the
same decline in insulin sensitivity is associated with a larger
increase in insulin secretion in obese than in lean subjects. Thus,
insulin resistance and obesity each augment the functional demand on
the pancreas in a nonlinear manner.
|
|
27.9 kg·m-2).
These findings are typical of studies investigating the impact of
obesity on the development of NIDDM (e.g., Ref. 79).
However, in a very obese population like the Pima Indians, obesity was
not an independent precursor of NIDDM when plasma glucose and plasma
insulin levels (= insulin resistance) were used to predict the disease
(70). Thus, it can be argued that only the insulin-resistant segment of
an obese population is at increased risk for NIDDM. In the data used
for Fig. 5, only 52% of the obese subjects (defined as those with a
BMI 28 kg·m-2) were insulin resistant
(i.e., they had an insulin sensitivity value below the 10th
percentile of the lean group). Since the incidence of NIDDM in the
British cohort was 0.47% per year among the obese (BMI 28
kg·m-2) as compared with 0.14% per year in the lean, it
can be calculated that 40 yr of obesity would convert 32% of the
insulin-resistant obese phenotype into NIDDM. In line with
this prediction, in a cross-sectional, hospital-based analysis of the
impact of obesity duration on NIDDM, the majority of the subjects who
had been obese for longer than 40 yr were found to be diabetic (80).
Thus, exhaustion of the ß-cell secretory capacity under the joint
pressure of obesity and insulin resistance can be presumed from the
quantitative relationships in Fig. 5 and is compatible with the
epidemiology of NIDDM and obesity. Stress failure of insulin secretion
must be a more frequent event in the natural history of glucose
intolerance than previously thought. A further way in which insulin resistance can impair ß-cell function is glucose toxicity (11, 12). Insulin resistance is compensated for by hyperinsulinemia through small increases in plasma glucose, which signal to the ß-cell the increased cellular need for the hormone. The rise in plasma glucose may be small (e.g., in the case of weight gain), but in the long run it may nevertheless exert a negative effect on insulin-secretory function. Thus, although we ignore the exact relationship between exposure (hyperglycemia x time) and ß-cell damage in humans, it is conceivable that obesity strains the endocrine pancreas at the same time as it slowly intoxicates it over many years.
There is another mode of cross-talk between insulin secretion and its action, for which the term lipotoxicity has been coined. By inhibiting triglyceride hydrolysis, insulin restrains delivery of oxidizable fatty substrates (FFA) to target tissues, thereby favoring carbohydrate oxidation in a substrate competition cycle (Randle’s cycle). Resistance of lipolysis to insulin inhibition impedes glucose metabolism by curtailing both glucose oxidation and glucose storage in glycogen as well as glucose transport and phosphorylation in skeletal muscle (81, 82). FFA levels have long been known to be elevated in both obesity and NIDDM (83, 84), and raised fasting and postglucose FFA concentrations have recently been associated specifically with the presence of insulin resistance irrespective of body adiposity (85). In addition to the circulating FFA pool, triglyceride content in skeletal muscle (86, 87) and the fatty acid composition of skeletal muscle phospholipids (88) have been found to correlate with insulin sensitivity in vivo. Intramuscular triglyceride stores are increased in muscle tissue of patients with IGT (89). Circulating FFAs sustain hepatic glucose production through stimulation of gluconeogenesis (90); under certain circumstances, high FFA levels can exacerbate glucose overproduction in NIDDM (91). The epidemiological counterpart of these physiological connections is the evidence from prospective studies that FFAs predict NIDDM independently of insulin sensitivity (92, 93).
On the other side of the loop, FFA can inhibit insulin secretion. Thus, raising FFA for 48 h (through the infusion of a triglyceride emulsion) impairs the ß-cell response to glucose in rats (94), and long-term incubation of rat islets with fatty acids depresses their ability to release and synthesize insulin in response to glucose stimulation (95). Recent studies in healthy volunteers have shown that, while experimental increases in FFA enhance the acute insulin response to intravenous glucose in the short term (6 h), longer (24-h) exposure to raised FFA reduces AIR by almost 50% (96).
In sum, analysis of the etiology problem by mechanism does not solve the ambivalence. Rather, it suggests that insulin resistance and deficiency are so tightly connected through multiple mechanisms that the a priori likelihood that they may segregate in distinct phenotypes even in the early stages of NIDDM (i.e., IGT, high-risk groups) should not be high. Obesity, a prevalent and powerful factor, further reduces insulin action and strains the ß-cell, thereby confounding the relative contribution of insulin resistance and insulin deficiency to the genesis of hyperglycemia.
C. The genetic approach
Genetic analysis and genetic epidemiology are increasingly popular
strategies to establish etiology. Segregation analysis can link NIDDM
with anonymous polymorphic loci in the genome (97); positional cloning
can then be applied to map out relevant genes (98). With the
availability of a large number of polymorphic markers and an improved
sampling design [based on affected sibling pairs (99)], genome-wide
search of disease loci has become possible (e.g., Ref. 100).
Quantitative traits rather than clinical outcome can be used for
genetic analysis (101). For example, in more than 200 family members of
patients with NIDDM (102), fasting insulin levels segregated as an
autosomal recessive allele, with a frequency of 0.25. In
Mexican-American families, segregation of postglucose insulin levels
was best described by a single dominant locus, with residual polygenic
and/or environmental effects (103). Clearly, these estimates depend on
the number and quality of pedigrees suitable for analysis and
ultimately represent only the best statistical fit of the data from a
range of compatible models. It should also be noted that, when used as
a surrogate for insulin sensitivity, the fasting insulin concentration
behaves better (in terms of strength of association with other
variables) than stimulated insulin levels. However, in the RIA, fasting
insulin is measured with less precision than stimulated insulin.
Furthermore, proinsulin and split proinsulins make up a greater
percentage of fasting than stimulated insulin concentrations (104).
These propeptides may carry their own message as markers for NIDDM, as
recently suggested (105). Despite these limitations, it appears
indisputable that the fasting insulin concentration, although subject
to modulation by many external factors, is a moderately inheritable
trait. Insulin sensitivity, as measured by direct methods, shows a
tri-modal distribution in Pima Indians (106) and a skewed distribution
in nondiabetic Europeans (77); it clusters in the family in Pima
Indians (107) and Caucasians (35). Comparatively less is known on the
distribution, familial aggregation, and heritability of indices of
insulin secretion (e.g., Ref. 108).
Direct genetic analysis can identify mutations in genes encoding the
various functions involved in insulin secretion [insulin, glucagon,
glucokinase, glucose transporter 2 (GLUT2), sulfonylurea
receptor, etc.] (e.g., Ref. 109) or action [insulin
receptor, glucose transporter 4 (GLUT4), insulin receptor
substrate-1, hexokinase II, fatty acid-binding protein,
etc.] (e.g., Ref. 110). Overall, genetic analysis has been
successful in the monogenic forms of diabetes, including MODY (111),
isolated mutations in the insulin gene (112), the insulin receptor gene
(110), and in the mitochondrial genome (113). Typical late-onset NIDDM,
however, is heterogeneous both genetically as well as phenotypically.
As a result, genetic analysis has made limited progress: so far,
responsible genes have been identified in only a very small fraction
(5%) of adult diabetes (114).
Whether NIDDM will eventually be shown to be truly polygenic (numerous
genes with small effects) or whether a small number of major genes
(diabetogenes) will account for the majority of cases is unknown at
present. Several problems limit success in genetic analysis (97, 98, 99, 100).
Among them, diabetogenes and susceptibility genes may vary among
different ethnic groups just as much as does the prevalence of NIDDM
[e.g., 50% in Pima Indians vs. <2% in
mainland China (115)]. Thus, though ethnic differences may be useful
to map genes for their relevance to the disease, results obtained in
any one population may not be generalizable. Moreover, the search for
genes may be confounded by factors acting during early intrauterine
life, as implied by the work of Hales and Barker (116). Such
transmissible influences, marked by low birth weight, need to be
distinguished from true heredity. Third, environmental factors
(e.g., hyperglycemia, diet) may induce susceptibility genes
and do so differentially according to ethnicity and life-style.
Finally, even when a major NIDDM gene has been associated with the
disease, its pathogenic role must be proved by identifying the gene
product, measuring it, and quantitatively relating its malfunction to
the insulin resistance or insulin deficiency observed in
vivo.
In sum, the genetic approach has the best a priori chances of solving the etiology problem. The answer, however, may be a long way off and may eventually turn out to be complex and particular rather than simple and general.
|
IV. Summary |
|---|
|
TopAbstractI. IntroductionII. Preliminary IssuesIII. Insulin Resistance vs....IV. SummaryV. Insulin Resistance: The...VI. ProspectsReferences |
|---|
Despite these uncertainties, the weight of current evidence supports the view that insulin resistance is very important in the etiology of typical NIDDM for the following reasons: 1) it is found in the majority of patients with the manifest disease; 2) it is only partially reversible by any form of treatment (117); 3) it can be traced back through earlier stages of IGT and high-risk conditions; and 4) it predicts subsequent development of the disease with remarkable consistency in both prediabetic and normoglycemic states. Of conceptual importance is also the fact that the key cellular mechanisms of skeletal muscle insulin resistance (defective stimulation of glucose transport, phosphorylation, and storage into glycogen) have been confirmed in NIDDM subjects by a variety of in vivo techniques [ranging from catheter balance (118) to multiple tracer kinetics (119) to 13C nuclear magnetic resonance spectroscopy (120)], and have been detected also in normoglycemic NIDDM offspring (121).
If insulin resistance is a characteristic finding in many cases of NIDDM, insulin-sensitive NIDDM does exist. On the other hand, given the tight homeostatic control of plasma glucose levels in humans, ß-cell dysfunction, relative or absolute, is a sine qua non for the development of diabetes. If insulin deficiency must be present whereas insulin resistance may be present, is this proof that the former is etiologically primary to the latter? If so, do we have convincing evidence that the primacy of insulin deficiency is genetic in nature? The answer to both questions is negative on several accounts. The defect in insulin secretion in overt NIDDM is functionally severe but anatomically modest: ß-cell mass is reduced by 20–40% in patients with long-standing NIDDM (122). Moreover, the insulin secretory deficit is progressively worse with more severe hyperglycemia (123) and recovers considerably upon improving glycemic control (124). These observations indicate that part of the insulin deficiency is acquired (through glucose toxicity, lipotoxicity, or both). In addition, although insulin deficiency is necessary for diabetes, it may not always be sufficient to cause NIDDM. In fact, subtle defects in the ß-cell response to glucose may be widespread in the population (108, 125) and only cause frank hyperglycemia when obesity/insulin resistance stress the secretory machinery. Conceivably, there could be ß-cell dysfunction without NIDDM just as there is insulin resistance without diabetes. Incidentally, any defect in insulin secretion, whether in normoglycemic or hyperglycemic persons, could be due to other factors than primary ß-cell dysfunction: amyloid deposits in the pancreas (126), changes in insulin secretagogues (amylin, GLP-1, GIP, galanin) (127, 128, 129, 130), early intrauterine malnutrition (131). Finally, the predictive power of early changes in insulin secretion for the development of typical NIDDM is generally lower than that of insulin resistance. Thus, the balance of current evidence indicates that, while ß-cell dysfunction is the precipitating factor in the emergence of hyperglycemia, its molecular mechanisms (132) and genetic basis remain more elusive than those of insulin resistance.
It should be emphasized that, even if the NIDDM disease loci in the
genome should eventually prove to be numerous and, possibly, disparate,
NIDDM can always be reduced to a bifactorial problem, i.e.,
one of insulin resistance and insulin deficiency (Fig. 6). As long-term human studies indicate
that insulin resistance and insulin deficiency can be traced far back
in the natural history of the disease, NIDDM can be thought to evolve
under the joint drive of these two etiological factors. Because of the
multiple physiological connections, in general, insulin deficiency and
resistance will co-vary. However, variability of the two functions at
each stage and overlap between successive stages will be large on
account of differences in genetic pressure and environmental
influences. MODY and LADA, falling off the line to the left, may
exemplify monogenic (or oligogenic) hyperglycemia; obesity will curve
the relationship to the right.
|
|
V. Insulin Resistance: The Cluster Concept |
|---|
|
TopAbstractI. IntroductionII. Preliminary IssuesIII. Insulin Resistance vs....IV. SummaryV. Insulin Resistance: The...VI. ProspectsReferences |
|---|
|
VI. Prospects |
|---|
|
TopAbstractI. IntroductionII. Preliminary IssuesIII. Insulin Resistance vs....IV. SummaryV. Insulin Resistance: The...VI. ProspectsReferences |
|---|
It is increasingly clear that accurate phenotyping of patients and at-risk groups is key to the genetic analysis of heterogeneous disorders. Different forms of diabetes can cluster in the same family, thereby confounding the pattern of transmission of hyperglycemia. Extensive testing for a panel of autoantibodies and HLA typing will 1) elucidate the basis for the cosegregation of IDDM and NIDDM (145), 2) distinguish autoimmune (LADA or IDDM) from metabolic diabetes, and 3) improve the prediction of the clinical course of the disease (146).
In vivo techniques can be refined so as to obtain a reliable estimate of both insulin sensitivity and secretion from a single test. The minimal model analysis of the ivGTT and the hyperglycemic clamp (19) are attempts in this direction, but recent developments in non-steady-state modeling (147), coupled with the use of stable isotopes (148), are likely to lead to improved tests to measure insulin release and action simultaneously (i.e., the slope of the relationship between secretion and biological effects of insulin as an integrated index of insulin control of the glucose system). The application of refined measurements in field studies will increase information on the distribution and modulation of both traits, insulin sensitivity and secretion. The natural history of both functions in normal subjects progressing to hyperglycemia will be described in more detail.
Obesity and fat distribution are likely to be investigated in more
depth. Although the role of obesity in amplifying the risk of NIDDM has
been known for a long time, intriguing circumstances have recently
emerged. Thus, the close association of NIDDM with obesity may be due
not only to insulin resistance but also to linkage/interactions between
the respective genetic background of the two conditions (149). The
insulin hypersecretion of obesity (77), particularly in some ethnic
groups (150), may turn out to be the primary etiological factor in a
subset of NIDDM patients in whom insulin resistance and ß-cell
exhaustion both evolve as long-term consequences of
paroxysmal stimulation of insulin release (Fig. 1). In the central
nervous system, both feeding and stress signals originate: this is also
a candidate site for the link between insulin hypersecretion and fat
distribution (77, 151), eventually leading to excess cardiovascular
disease in subjects with android obesity (152). With regard to this, it
is interesting that in Japanese-American men, impaired insulin release
precedes visceral adiposity in the natural course of NIDDM (153), a
sequence that reverses the paradigm of body fat distribution as a
determinant of insulin action and secretion (154).
Finally, and perhaps most importantly, more information will be forthcoming on the relative role of insulin deficiency and insulin resistance in the appearance of diabetic complications. Here, the bifactorial version of the NIDDM etiology problem becomes drastically insufficient, as it is clear that additional genetic and environmental influences drive the natural history of complications. By way of example, ACVD mortality among diabetic patients varies greatly between countries (155); microvascular and macrovascular complication cross-predict one another (156); microalbuminuria segregates with insulin resistance (17) and predicts mortality in NIDDM (157); in Chinese-Americans both the insulin receptor and apolipoprotein gene contribute to the emergence of NIDDM (158); familial hypertension and glycemic control are major predictors of diabetic nephropathy (159). In this area, insulin resistance is likely to play a greater part than insulin deficiency due to its many links and its putative role as a direct antecedent of ACVD (3, 133). After all, insulin resistance is a larger problem than hyperglycemia in the population; for complexity and implications, it will therefore remain a threat to the patient, a concern to the clinician, and a challenge to the investigator.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractI. IntroductionII. Preliminary IssuesIII. Insulin Resistance vs....IV. SummaryV. Insulin Resistance: The...VI. ProspectsReferences |
|---|
