Insulin resistance is a common biochemical abnormality preceding type 2 diabetes and other associated diseases. In 1988, G. Reaven proposed that this state should be studied as a disease and named it syndrome X, or unknown syndrome. Epidemiologic studies revealed the risk factors of this disease and ultimately the cause(s). The World Health Organization (WHO) gave a new name to this disease, metabolic syndrome, and established its diagnostic criteria in 1998. The metabolic syndrome is defined as the presence of at least 3 of the following symptoms or phenotypes in an individual; abdominal obesity (measured by waist circumference), high serum triglyceride level, low serum high density lipoprotein (HDL) level, hyperglycemia, and high blood pressure.

In 1998, my colleagues and I found that blood mitochondrial DNA (mtDNA) density in peripheral blood cells was lower in those subjects who would go on to develop diabetes in 2 years, in comparison to those who would not. Furthermore, people with lower blood mtDNA density had high blood pressure, high serum triglyceride levels and a high waist circumference ratio. These findings suggest that a decrease in mtDNA density in blood cells is causally related to the metabolic syndrome. In the following years, we accumulated evidence suggesting that mitochondrial dysfunction might be the central abnormality causing insulin resistance, and it might also be responsible for the insulin secretion defect found with type 2 diabetes. We presented evidence in support of this hypothesis and this evidence underwent several reviews [1,2].

In this lecture, I will argue again that the metabolic syndrome is a state of mitochondrial dysfunction, which is, in turn, identical to the state of insulin resistance. In this interpretation, cause(s) of type 2 diabetes are identical to cause(s) of the metabolic syndrome, and type 2 diabetes is later stage of this altered body state. The insulin secretion defect in type 2 diabetes might arise from the insulin resistance state per se, as it is a state of mitochondrial dysfunction and good mitochondrial function is needed for good insulin secretion. This view of the metabolic syndrome is thus narrower than the generally held view, which includes full-blown type 2 diabetes and atherosclerosis within the metabolic syndrome (Fig. 1).

Another important aspect of the mitochondria-based model is that it is a special case of the scaling law. We know that many body parameters correlate with body size; the bigger the body size, the larger the metabolism rate becomes. It is well known that the metabolic rate of mammals increases when body mass increases. There is a linear relationship between metabolic rate (Y), and body mass (M) characterized by the equation, Y= Y0×M(body mass)^3/4.

This law tells us that the larger the body size, the greater the decrease in the mass specific metabolic rate (Y/M). Most importantly, it was found that this relationship holds down to individual cells and their mitochondria as well as mitochondrial electron transfer chain activities [3]. In other words, if the body size of a mammal increases, its unit (cell or mitochondrial) metabolic power decreases (Fig. 2). We found that mitochondrial DNA copy numbers are decreased in the metabolic syndrome, and this is negatively correlated with serum insulin responses in health young male medical students. From these results, we reasoned that the metabolic syndrome is the result of the metabolic scaling law in action, in that body metabolism is changing in order to adapt to decreasing mitochondrial function. In other words, the state of insulin resistance is an adaptation of the body to the decrease in unit mitochondrial function. This reasoning is also discussed in detail in a previous paper [2].

One of the key principles of this process is the adaptation of enzymes to the body temperature of 37℃. At this temperature, most enzymes have the lowest Gibbs free energy. If mitochondrial dysfunction leads to a decrease in heat production, body temperature would be lower, and this has to be compensated for by an increase in fat mass (insulation), increased heat production, and thus more mitochondrial mass and nutrient supply, which in turn would require higher levels of serum insulin, glucose, and triglycerides.

Metabolic scaling is the only "law" generally accepted in biology. The term "law" is not debated in this usage. Scaling is applicable not only to the metabolic rate but to many aspects of body states, such as the relationship between body size and brain mass, aorta circumference, blood volume and almost every aspect of body metabolism. This phenomenon is also known as "allometry"; where the size of every part of body is quantitatively related to the whole (mass). It is not surprising that bone mass and bone strength should increase along with the body size of an animal; lung volume should increase with increasing metabolic demand, kidney size also increases in order to deal with the increasing metabolic waste of an increasing body mass and blood volume. This concept is illustrated in Fig. 3.

This kind of relationship predicts that if body mass increases (obesity), the circulatory system will enlarge in adaptation. However, there will be limit to this adaptation, in particular when unit cell (or mitochondrial function) is decreased, as in the insulin resistance state, a mismatch will develop. If heart metabolism (i.e., cardiac output) did not increase in order to adapt to increasing demand from the whole body, the result would be heart failure. If the coronary artery does not need to supply as much blood to the heart due to a decrease in energy demand, the coronary artery should decrease in size. I suspect that this may be a reason why coronary heart disease develops in type 2 diabetes.

These novel explanations might be unfamiliar to most clinicians, as they are biophysical and based mostly on bioenergetics. While many aspects of this theory need experimental support, I believe that a mitochondria-based model fits well with reality.

There are many "causes" in medicine; in fact "causes" are diverse in different disease states, and diseases are themselves diverse and variously defined. These features are illustrated in Fig. 1. Diabetologists may regard coronary heart disease as a complication of type 2 diabetes, while cardiologists might consider diabetes to be a cause of coronary heart disease. This difference arises because specialists treat a limited group of people with specific diseases, such as patients with type 2 diabetes or coronary heart disease. They examine why their patients get those illnesses (causes) and what might happen to them as a result (complications).

Causes are divided into genetic and environmental causes, which can in turn be divided again into their individual components. Medicine adopts the division of "causes" when there are different treatments or prevention methods available for these different causes. For example, diabetes is caused by insulin deficiency and it is treated with an insulin supply. If you were to ask whether we could restore insulin deficiency, the answer is no. If you were to further ask why we could not do this, there would be another answer(s); such as ageing, etc. In fact, we do not know.

We are, however, fairly certain of all of these chains of events from the genetic make-up (factors) and epigenetic changes which occur during development, particularly during the fetal stage, such as malnutrition and environmental factors. In the mitochondria-based model, there are 3 basic causes; genes and developmental and environmental toxins (and protectors). I will explain why the continuous accumulation of environmental toxins or so called persistent organic pollutants (POPs) in our body might be one of the most important factors.

A cause and effect relationship between a microorganism and disease is usually established when Koch's postulates are met; 1) the microorganism must be found in abundance in all organisms suffering from the disease, but should not be found in healthy organisms; 2) The microorganism must be isolated from a diseased organism and grown in pure culture; 3) The cultured microorganism should cause disease when introduced into a healthy organism; and 4) the microorganism must be re-isolated from the inoculated, diseased experimental host and identified as being identical to the original specific causative agent. The evidence provided above would be sufficient enough to meet Koch's postulates. As POPs are inorganic, very stable and persistent, the issues of isolation for POPs and pure culture as required in postulates 2 and 4 are irrelevant.

We are dealing with a novel problem here; the suspected causative agents, POPs, are not a single agent, but are diverse, very similar to one another, interact with each other (variously acting as antagonist or agonist), and exhibit a "low dose effect." How can we "know" if they cause a disease, when they accumulate over many years and exhibit such complicated behaviors? There is another problem as well, because the metabolic syndrome is not a well-defined disease entity. Lee DH found that serum levels of organochlorine pesticides are more closely associated with type 2 diabetes, and abdominal obesity and high serum triglyceride levels and furans and dioxins are more closely associated with hypertension. Should we divide the metabolic syndrome into two subgroups: type 1, organochlorine pesticide syndrome and type 2, furan/dioxin syndrome? Furthermore, there are 5 or more phenotypes characterizing the metabolic syndrome and only 3 of them are needed for a diagnosis. What would be the merit of diagnosing a disease, such as the metabolic syndrome, if serum POP levels (or burden of body POPs) is high? We need to reduce POP levels anyway, regardless of the subject's phenotype. It might be worthwhile to note that the NHANES data showed that POP levels interact with BMI in determining type 2 diabetes. When POP levels were lower, diabetes prevalence was not associated with BMI, while obese people with high serum POP levels had an increase in diabetes prevalence.

Measuring exposure to POPs in humans is particularly difficult as the current method of POP detection usually requires a large serum sample. There is no stored serum sample large enough to test this concept epidemiologically, only a few milliliters at best. Furthermore, measuring POPs chemically with the use of high resolution GC/MS, which is the gold standard, is very difficult, prohibitively expensive and requires a machine devoted to this use, facility and people.

We wondered if there would be an alternative method for measuring the body burden of POPs and found the CALUX assay promising. This assay has been successfully used by environmental scientists for the screening of dioxin-like substances for many years. It takes advantage of the fact that dioxin and dioxin-like substances act through aryl-hydrocarbon receptors almost exclusively. If one were to make a cell line transfected with the plasmid expressing aryl-hydrocarbon receptor (AhR) gene, it would respond to dioxin or dioxin-like chemical substances. If AhR is Chemical (dioxin-like substances) Activated and Luciferase Expression (CALUX) is measured, one could calculate the amount of dioxin or dioxin-like substances in samples.

Recently, KimPak Y at Kyung Hee University developed a modified CALUX assay and tested sera of pre-diabetes or type 2 diabetes for their AhR transactivation activity (AhRTA). To our surprise, the AhRT activities of sera were closely associated with the parameters of the metabolic syndrome. Results were consistent with the conclusion that serum dioxin-like activity is a "cause" of insulin resistance and type 2 diabetes. In short, AhRT activities and parameters of the metabolic syndrome were linear (Submitted for publication). This linear relationship is in stark contrast with the conclusion obtained from CARDIA cohorts, where a "low dose effect" predominates. I wonder if this study could be extended to a nested case control study like CARDIA.

Critics of the mitochondria-based model argue that excessive calorie intake and lack of exercise are the primary causes of obesity, and thus the metabolic syndrome. It is frequently argued that even though mitochondrial dysfunction is present in this state, it is regarded as secondary to obesity, induced by overfeeding or lack of exercise. Indeed, one can surely bring on obesity by consuming a high-fat diet. However, these critiques dismiss some important facts; 1) only certain high-fat diets cause obesity in susceptible animals, (this "high-fat diet" was found to be contaminated with POPs [unpublished observation]); 2) the propensity to return to previous body weight after fat removal or transplantation suggests the presence of a strong drive to maintain body weight; and 3) obese people do not eat more, once a certain increase in body weight is reached. I will not provide references supporting all of these facts, but I want to emphasize that the current gold standard of diet assessment, the doubly labeled water test, which is held to provide unequivocal evidence for excessive calorie intake as a cause of obesity, might be based on an erroneous assumption.

A basic assumption of the doubly labeled water (DLW) method to measure energy expenditure is that H of body water only leaves the body pool in the form of water. If deuterium (as a measure of H) is sequestered in anything other than body water during a DLW study, an overestimation of water turnover and therefore an underestimation of CO₂ production would result, thus indicating a high metabolic rate. Hydrogen is incorporated into fat, and deuterium incorporation into body fat has the greatest potential to be an error of the DLW method. This is significant and presents a potentially serious error in the determination of energy requirements, especially in those subjects with different intrinsic rates of fat synthesis. Haggarty et al. [17] reported de novo lipogenesis and cholesterol synthesis over the typical duration of a DLW study (2 weeks), as well as the likely effects of sequestration on the performance of the DLW method in adult human subjects.

Using whole-body fatty acid synthesis, they found little evidence of an error on DLW-derived CO₂ production for five of the six subjects studied; mean -0×5%, range 0 to -1×3%. However, there was a subject whose fatty acid synthesis rate was almost 10 times faster than others, was hypertrigliceridemic (five times higher) and had a slightly higher body fat mass. They argued that the very high value for circulating triacylglycerols in this subject was indicative of a high rate of de novo lipid synthesis and that the assumptions underlying the calculation of whole-body lipogenesis (assumed identical between subjects) may be invalid. In other words, in subjects with hypertriglyceridemia, the DLW test might result in rather high estimates of energy expenditure.

In the metabolic syndrome (with hyertriglyceridemia), a DLW test would result in a high estimate of energy expenditure compared with a diet intake measure (which is normal for the body mass). One would arrive at an erroneous conclusion that obese subjects under-report their dietary intake.

I presented evidence that POPs might be a cause of the metabolic syndrome, probably acting through AhR and a mitochondrial mechanism. There is a huge body of evidence supporting this hypothesis. However, establishing a cause and effect relation between POP exposure and the development of the metabolic syndrome will be difficult, because of the diversity of POPs and their bizarre low dose effect. This conundrum might be resolved by our finding that serum AhRT activity is linearly related with parameters of the metabolic syndrome in nested case control studies.

The mitochondria-based model could be considered as a special application of the metabolic scaling law as applied to a group of people with a specific disease. This model will be very useful in integrating various factors acting on mitochondrial function, such as genes, fetal malnutrition and environmental toxins and protectors. I believe that "genomic medicine" should be based on the mitochondrial genome, not only the nuclear genome, as our model suggests. Our model also suggests that if we could improve mitochondrial function, insulin resistance might decrease. This prediction is validated by recent studies with hydrogen gas, which quenches free radicals and stimulates mitochondrial function [18], and by studies with bile acid sequestrants, which could remove POPs and improve the metabolic syndrome [19].