Surrogate markers of insulin resistance in subjects with metabolic syndrome – data of the Berlin Aging Study II

Introduction

The prevalence of overweight and obesity in Germany is approx. 67% among men and 53% among women aged 18–79 years [1]. Visceral obesity, in particular, and the metabolic syndrome (MetS) associated with it represent a risk for cardiovascular disease (CVD) and type 2 diabetes (T2D) [2, 3]. Older patients are especially at risk of developing MetS due to the increased number of comorbidities, as well physiological changes that develop in old age – such as a percentage increase in body fat and decrease of muscle mass, the main site of action for insulin [4–7].

Insulin resistance (IR) refers to the diminished effect of the hormone insulin in the body, causing an increased demand for insulin to maintain glucose homeostasis [8]. Regardless of the underlying cause, IR does not only disrupt glucose metabolism; increased insulin levels, through increased synthesis of VLDL lipoproteins in the liver, can also favor changes in lipometabolism (increased triglyceride levels, reduction in HDL levels) and thus cause MetS [9].

The homeostasis model assessment index and other surrogate markers for IR, the homeostasis model of insulin resistance (HOMA-IR) and insulin sensitivity are well-established and frequently altered in advanced age, as well as pathological in MetS patients [10, 11]. Cutoff values for these surrogate markers to identify subjects with MetS have not been established. The objective of the current data analysis as part of the Berlin Aging Study II (BASE-II) was to determine the prevalence of the MetS in a group of (i) 446 younger (aged 22–37 years) and (ii) 1271 older subjects (>60 years) without diabetes. Taking into account potential influencing factors, the cutoff values for HOMA-IR were calculated above which MetS could be identified in our study population with the best possible sensitivity and specificity.

Materials and methods

Results

Complete cross-sectional data to determine MetS and calculate surrogate markers of insulin resistance and insulin sensitivity (HOMA-IR, QUICKI, Homa-b and G/I ratio) were available for 446 younger (53% women, 28±3 years old) and 1271 older subjects (52% women, 68±4 years old). The clinical characteristics of the younger and older group of test subjects have been summarized in Table 1.

According to the 2009 definition of IDF/AHA/NHLBI, the prevalence of MetS was 6.7% among the younger subjects and 28.3% among the older test subjects (4.7% of the younger subjects and 27.6% of the older subjects when the IDF definition was applied; Table 1). “Elevated blood pressure” (83.2% of the older and 37.1% of the younger subjects) and “increased waist size” (78.4% of the older and 25.1% of the younger subjects (Figure 1A) were the most frequent conspicuous MetS criteria. Overall, the older subjects exhibited statistically significant higher levels for fasting glucose and glucose following an oral glucose tolerance test (OGTT), as well as elevated insulin levels and HbA_1c that was higher on average. There were also significant changes in the surrogate markers for insulin resistance/sensitivity (HOMA-IR, HOMA-b, QUICKI and G/I ratio) in older test subjects. The other results refer to the 2009 MetS definition of IDF/AHA/NHLBI. Among older subjects with MetS, particularly “increased waist size”, “elevated blood pressure” and “insulin resistance” dominated, while “elevated blood pressure”, “elevated triglycerides” and “reduced HDL” dominated among the younger test subjects (Figure 1B).

Figure 1:

Distribution of MetS criteria among study subjects.

Distribution of MetS criteria (A) within the total population, (B) among subjects with MetS, (C) among subjects with MetS above the HOMA-IR cutoffs, and (D) among subjects with MetS below the HOMA-IR cutoffs.

Receiver operator characteristic (ROC) curves were analyzed for HOMA-IR and MetS (MetS vs. no MetS). The Youden index was calculated as sensitivity – (1-specificity), according to the coordinates of the ROC curve (Figure 2). The highest values were used for the Youden index to identify cutoff values for HOMA-IR. The group of younger subjects showed slightly lower cutoffs (HOMA-IR >1.88) than the older subjects (HOMA-IR >1.98). The sensitivity among the younger subjects – subjects above these HOMA-IR cutoffs also exhibited MetS – was 80%, specificity at 85.3%. The sensitivity among the older subjects was 73.6%, specificity at 72.9%.

Figure 2:

ROC curves for the identification of HOMA-IR cutoffs in (A) older and (B) younger test subjects.

The positive predictive value for the younger subjects was 80%, and 74% for the older subjects (negative predictive value of 15% for younger and 27% for older test subjects). As for test subjects above the computed HOMA-IR cutoffs, “increased waist size” and “insulin resistance” dominated among a subsample of younger subjects; the MetS criteria were evenly distributed among the older subjects (Figure 1C). In older test subjects below the HOMA-IR cutoffs, we also observed evenly distributed MetS criteria, while “elevated blood pressure”, “reduced HDL” and “elevated triglycerides” dominated among the younger subjects.

Finally, binary logistic regression models were calculated to determine odds ratios for MetS among test subjects above the computed cutoffs for HOMA-IR. Table 2 shows the results of different regression models after adjustment for an increasing number of Covariates. Under the most adjusted model (model 3, adjusted for age, gender, physical activity and computed separately for the two age groups), the MetS odds for the older test subjects were 5.7 (95% CI: 4.1–7.9) and 22.2 for the younger subjects (95% CI: 7.0–70.5) (if and when they were above the computed HOMA-IR cutoffs).

Discussion

The current analysis as part of BASE-II yielded a high prevalence for the MetS both in older and younger test subjects (6.7% among younger and 28.3% among older subjects) according to the 2009 definition of IDF/AHA/NHLBI. Overall, the parameters of glucose and lipid metabolism were elevated in the group of older test subjects (with a significant reduction in HDL cholesterol).

In the test subjects without diabetes, the HOMA-IR levels >1.88 (for younger subjects; 80% sensitivity, 85.3% specificity) and HOMA-IR >1.98 (for older subjects; 73.6% sensitivity, 72.9% specificity) proved to be good markers for the existence of MetS. The positive predictive values were at 80% for younger and at 74% for older test subjects (negative predictive value for younger subjects at 15%, and at 27% for older subjects).

Even today, the MetS constitutes a major challenge for the health system, and is associated with an up to five-fold risk of diabetes mellitus and twice the risk of heart attacks or strokes [2, 17]. Measurement of waist size and blood pressure is part of clinical routine, however determining all parameters included in the MetS cluster is not common. Moreover, this represents a time-consuming task, which is why MetS diagnoses are rare. This fact is complicated further as a result of different possible MetS definitions. Even though our analysis revealed only a minor difference in the prevalence of MetS according to IDF vs. IDF/AHA/NHLBI, other MetS definitions are also used that focus on insulin resistance (WHO 1999) or other cutoffs (NCEP 2001), thus creating significant divergence with respect to prevalence [18–20].

Visceral obesity and insulin resistance are central factors of the MetS [21, 22]. Free fatty acids (FFA), which are released in greater number in the case of abdominal obesity, may favor insulin resistance by inhibiting the insulin-dependent uptake of glucose in skeletal muscles. In addition, permanently elevated FFA impair the beta-cell function of the pancreas [23]. Obesity and insulin resistance are closely associated with elevated RR levels and dyslipidemia in this context [24, 25].

Although (visceral) obesity increases the risk of such changes, being overweight does not necessarily lead to IR, dyslipidemia or hypertension; these conditions also occur in people with normal weight [26]. Thus, as a cluster syndrome MetS may represent different phenotypes. For example, dyslipidemia, when combined with elevated blood pressure (as seen particularly in our younger test subjects), but also increased waist size, in combination with insulin resistance and elevated blood pressure, may translate to a diagnosis of MetS. Therefore identification of subjects with MetS on the basis of HOMA-IR, in particular in phenotypes that are not dominated by insulin resistance or increased waist size seems to be less appropriate. Still, our current data analysis did not only determine cutoffs for HOMA-IR as surrogate markers for insulin resistance, above which test subjects with MetS can be identified with a good degree of sensitivity and specificity, but, when adjusted for KO variables like BMI, age, gender and physical activity, test subjects above these cutoffs also exhibited significantly elevated odds for MetS. In this context, a diagnosis of MetS on the basis of HOMA-IR does not seem possible due to the positive predictive value of 80% for younger and 74% for older test subjects. The negative predictive value – 15% for younger and 27% for older test subjects – must also be taken into account.

The determination of insulin resistance is not a routine clinical test. The hyperinsulinemic-euglycemic clamp technique represents the gold standard for identifying IR [27]. With a constant insulin infusion, the patient is administered variable doses of glucose for the purpose of reaching the normal fasting blood glucose level. This test is time- and labor-intensive, which means that there is not yet a clinically routine alternative to IR testing on the basis of surrogate parameters.

These surrogate parameters include, in particular, the HOMA-IR index, a suitable parameter [11], which worked well in identifying MetS in our data analysis both among younger and older test subjects. Accordingly, it might also be used in a clinical context to identify patients at risk of MetS. The cutoffs obtained in this group of subjects without diabetes showed only minor differences between younger and older patients. Even though MetS cannot be diagnosed with sufficient reliability due to the positive and negative predictive values, these HOMA-IR cutoffs could still be used to estimate the probability of MetS. Patients above these cutoffs should be subjected to a full diagnostic range of the MetS parameters, which would probably help to increase the frequency of MetS diagnoses in clinical practice. A single surrogate parameter for IR could be used to determine whether the time-consuming analysis of all parameters involved in a MetS diagnosis made sense. Given that the negative predictive value is relatively low particularly in older patients, the MetS diagnosis cannot be excluded solely on the basis of HOMA-IR. These patients should continue to be examined for blood pressure or BMI, as is customary in clinical practice, and if MetS is suspected, even independently of HOMA-IR levels, further diagnostics should be ordered to identify those patients that exhibit an increased risk of developing diabetes – they would potentially benefit from early intervention and closer checks.