ArticleInsulin resistance and cardiometabolic indexes: a comparison of gender-related differences in working-age subjects with overweight and obesity

Luisella Vigna Ospedale Maggiore di Milano Policlinico: Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico Amedea Silvia Tirelli Ospedale Maggiore di Milano Policlinico: Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico Melania Gaggini Istituto di Fisiologia Clinica Consiglio Nazionale delle Ricerche Salvina Di Piazza Ospedale Maggiore di Milano Policlinico: Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico Laura Tomaino Polytechnic University of Marche: Universita Politecnica delle Marche Stefano Turolo La Fondazione IRCCS Ospedale Maggiore Policlinico: Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico Gianluca Moroncini Azienda Ospedaliera Ospedali Riuniti Marche Nord Kyriazoula Chatzianagnostou CNR di Pisa: Consiglio Nazionale delle Ricerche Area della Ricerca di Pisa Fabrizia Bamonti Università degli Studi di Milano: Universita degli Studi di Milano Cristina Vassalle (  cristina.vassalle@ftgm.it ) CNR di Pisa: Consiglio Nazionale delle Ricerche Area della Ricerca di Pisa https://orcid.org/0000-0003-3438-6450


Introduction
Insulin resistance (IR) and cardiometabolic abnormalities affect a most of the middle-aged population of the working-age, being particularly associated with overweight and obesity [1]. Monitoring IR indexes and cardiometabolic risk is crucial in occupational medicine, because overweight/obesity is a risk factor for the development of many chronic conditions and well-being reduction, and, moreover, may have a signi cant long-term impact on health care costs of working-age adults [2]. Thus, it is particularly important to identify subjects at the pre-disease stage with the possibility of life-style and therapeutic intervention.
IR, due to a diminution of tissue response to insulin activity, is associated with metabolic syndrome, type 2 diabetes mellitus (T2D), obesity, and cardiovascular disease. There are several methods to assess IR state and selecting one or another depends on the purpose of their use and, often, on the available economic resources. Direct methods such as the "Hyperinsulinemic Euglycemic Glucose Clamp" or the "Insulin Suppression Test" have a low margin of error and considered the gold standard methods [3;4]. However, their application in the clinical practice or in epidemiological studies is limited as they are more complex and expensive. For these reasons, indirect methods have been developed (i.e. Insulin Sensitivity Index-ISI derived from the Oral Glucose Tolerance Test-OGTT, or Homeostasis Model Assessment-HOMAIR and Quantitative Insulin-sensitivity Check Index-QUICKI) which, even less precise, are simpler and cheaper than direct ones5-7. Nonetheless, comparison studies suggest that there could be differences in the prevalence of insulin resistance detected among different indirect indexes [8]. Recently two other indirect routine indexes based on the lipid pro le and fasting glucose levels were employed: Triglycerides/HDL-cholesterol ratio (Ty/HDLC) and Triglycerides-Glucose concentrations (TyG) [9][10][11]. Among indirect methods, ISI, even the most expensive, better correlates with direct methods [6]. In addition, the insuTAG (fasting insulin × fasting triglycerides) has been proposed as predictor of insulin resistance (IR) and metabolic syndrome (MetS) [12].
Moreover, other indexes are currently used as indicators of cardiometabolic risk, some considering only circulating lipids (e.g. Castelli risk indexes 1 and 2, non-HDLC, tryglicerides rich lipoporotein cholesterol-TRL-C and Atherogenic Index of Plasma-AIP), whereas others combine anthropometric measurements (such as waist circumference-WC, and body mass index-BMI) and lipid pro le, as in the visceral adiposity index-VAI13-16. Accordingly, we aimed to evaluate these indexes of insulin resistance and cardiometabolic risk in a large population of workers with overweight or obesity, evidencing possible gender-related differences, in order to identify a possible e cient, cheap and simple strategy to apply in clinical practice of workers surveillance and public economic management.

Patients
Blood samples were drawn after an overnight fast, to be analyzed for glucose (Gl), insulin (Ins), triglycerides (Ty), total cholesterol (T-Chol), high and low density lipoprotein-cholesterol (HDLC, LDLC), gammaglutaryltransferase, creatinine, uric acid and C reactive protein (serum), and a complete blood count (CBC), brinogen and homocysteine levels (EDTA) by automated Modular D and P (Roche Diagnostics International Ltd, Basel, Switzerland).. Samples were frozen and stored at -20°C until analysed.
As Ty concentrations increase during chronic hyperinsulinemia (because adipose tissue releases free fatty acids in the blood circulation) and increase in liver and heart, this index can also provide information about risk of developing cardiovascular diseases and to discriminate a prediabetes condition [20][21][22][23].
Triglycerides/HDL cholesterol (Ty/HDL Chol): Ty/HDLC ratio is closely associated with fasting glucose levels. IR is present when this ratio is ≥3.5 mg/dL. Ty/HDLC is a predictor for insulin resistance and metabolic syndrome, also providing information about the risk of developing the cardiovascular disease CASTELLI index I: This index, obtained by the total cholesterol/HDLC ratio, has close relationship with cardiovascular risk in both sexes [13].
Subjects at TC/HDL ratio greater than or equal to 4.5 were found to be at higher cardiovascular risk in the Framingham study. Optimal values are <5 and <4.5 for men and women, respectively [24] non-HDLC: simply calculated as the difference between TC and HDL cholesterol, represents bad cholesterol including cholesterol carried on VLDL, LDL as well as chylomicron remnants and lipoprotein(a). Several important guidelines recommended the use of non-HDLC for cardiometabolic prevention strategies and indicate this biomarker as therapeutic target [14]. A cut-off for non-HDLC values 130 mg/dL is commonly used in clinical setting [25] Atherogenic index-AI CASTELLI index II: This index, obtained by the LDLC/HDLC ratio, is an indicator of cardiovascular risk [13]. Ideal values are <3.5 and 3 in men and women [24]. . Remnant-C ≥30 mg/dL identi es subjects at a higher cardiovascular risk [27] Atherogenic index of plasma-AIP: calculated as logarithmically transformed ratio of molar concentrations of triglycerides to HDL-cholesterol, resulting in a reliable predictor of cardiometabolic risk [14].

Results
Clinical and demographic parameters of the studied population are summarized in Table 1. There were no signi cant differences among males and females regarding mean age and BMI. However, levels of total cholesterol, HDL, CRP and brinogen were higher in women. Instead, smoking history (current-ex smokers) was more frequent in men, who presented higher blood pressure, triglycerides, fasting glucose and insulin, creatinine, homocysteine, GGT, WBC, and uric acid levels. In Table 2 were reported the calculated IR and cardiometabolic risk indices, all higher in men, except for the QUICKI. Moreover Table 3 reported the correlation between the different indirect cardiometabolic indices and CV risk factors according to gender. CV risk factors more frequently correlated with cardiometabolic indices in women. Interestingly, homocysteine was more commonly associated with cardiometabolic indices in men. In Fig. 1 is reported the categorization of the IR/cardiometabolic risk status through the test utilized according to gender. It is noteworthy the variability of the percent of the workers identi ed as insulin resistant, IR+, or at higher cardiometabolic risk greatly vary according to the different index used.

Discussion
The present study mostly focused on a large sample of working-age subjects with overweight and obesity, where alterations in cardiometabolic indexes could be indicative of their further worsening toward overt T2DM and cardiovascular risk.
The main ndings of the present study are: 1) The prevalence of IR and cardiometabolic risk was higher among males for all indices; 2) CV risk factors were correlated more frequently with IR/cardiometabolic indexes in women, especially when considering aging, waist circumference, BMI, blood pressure, glucose, CRP, brinogen and uric acid. Instead, homocysteine was more commonly associated with IR/cardiometabolic indexes in men; 3) The worker percentage of IR+ and subjects at higher cardiometabolic risk greatly varies according to the different index used to measure risk.
Isolated impaired fasting glucose (IFG) is generally found more prevalent in men, whereas impaired glucose tolerance, likely more related to cardiovascular risk, is more common in women [30;31]. Accordingly, we found a higher percentage of women with altered OGTT (19% versus 13%, data not shown), which was performed in a subgroup of subjects (316 women and 106 men). However, as insulin resistance is linked closely to IFG, it is not surprising to nd IR indices more elevated in men. Women remain more insulin sensitive, also despite weight gain, thanks to their capacity to expand subcutaneous fat, as we also observed in our population (higher QUICKI levels in women) [32]. It is known that at a given BMI the risk to develop T2D is higher in men than in women, presumably because of male higher insulin resistance [33;34]. This means that women have to increase more weight, and more adipose tissue, to develop insulin resistance and subsequently T2D. Moreover, fat distribution and quality differ by sex and, in general, men have greater visceral and hepatic fat compared with women [35]. In men ectopic fat accumulates in different organs, including the liver where it determines an increase of hepatic enzymes and triglycerides, rendering organs insulin-resistant [35]. Mechanisms determining increase of toxic lipid derivatives (ceramides or diglycerides) or other molecules are involved in this process, which, in turn, alters insulin pathway [35]. Moreover, men generally together with higher fasting glucose levels, had lower HDLC levels than women, as con rmed in our population [35].
Obesity, insulin resistance and insulin hypersecretion are likely to be key mediators of development of T2D in men. However, excess weight in women may be associated with a greater deterioration in the cardiometabolic risk. In fact, waist circumference and BMI resulted more frequently correlated with IR/cardiometabolic indexes in women. Moreover, the correlation more frequent between aging and IR/cardiometabolic indexes in women evidences the importance of menopause (sex hormone-e.g., oestrogen-suddenly falling with a relative growth in male hormones-e.g., testosterone). Testosterone increase in women is related to insulin resistance, hyperglycemia, increased prevalence of central obesity and hypertension. Previous evidence suggested that at the stage of prediabetes especially women have a higher risk of cardiovascular disease [36]. Biological causes related to this risk pattern involve a loss of the protective female sex hormones and their imbalance in hyperglycemic conditions leading to higher oxidative stress and in ammation, endothelial dysfunction, hypertension [37]. Accordingly, we found a more frequent correlation of CRP and uric acid (biomarkers related to in ammation and oxidative stress) with cardiometabolic indexes in women. Also the more frequent relationship between brinogen and cardiometabolic indexes in women may evidence an higher CV risk, as brinogen contributes to increased blood viscosity, brin formation and platelet aggregation [38]. Interestingly, Hcy was more commonly associated with IR/cardiometabolic indexes in men. According to previous data, Hcy was found higher in male than in females, and increases with aging and creatinine levels [39;40]. Different determinants affect Hcy levels such as diet, smoking habit, genetic and environmental factors, and hyperhomocysteinemia might be an additional risk factor in abdominal obesity [40]. Accordingly, we observed a signi cant relationship between VAI and homocysteine only in men (r=0.13, p<0.05, data not shown).
For the discrepancy observed in the identi cation of IR+ subjects between different IR indexes, it is important to remember that these indexes have different derivation. IR has genetic and environmental components and different pathogenetic origins (e.g. mitochondrial, in ammatory, metabolic, and adipocytic)s41-45. Moreover, the indices included in their determination different biomarkers that re ect different pathophysiological pathways. For example, the Ty/HDLC has been conceived as a marker of cardiovascular risk and, as such, suitable to better identi cation of subjects with altered lipid pro le within those affected by IR, whereas TyG, based on glucose and triglycerides values, provides information about subjects that have high risk to develop diabetes and dyslipidemia. However, the main advantage of the TyG and Ty/HDLC is that they derive from the fasting state measurements of easily available laboratory parameters, without requiring the more expensive quanti cation of serum insulin.
From a statistical point of view, the use of the IR indexes is, often, not accompanied by adequate information about their calculation, which may account for the discrepancy in the de nition of IR-/IR+ subjects. According to literature, no explanation was reported about the selected coe cients giving an idea of the limitations of this comparison. In fact, it is di cult to establish if the coe cient is a constant, derived by others measurement, or is a parameter. In fact, a parameter, derived for example by a mean or median, should be always reported with its standard deviation, but in many cases it is di cult to nd out the original formula and the consequent errors might invalidate the result of the following analyses. For example, the parameter 22.5 of HOMAIR formula represents the fasting plasma glucose/C peptide levels ratio; but it was calculated only on six healthy subjects and six diabetic patients [5] without the possibility to compute its standard deviation; therefore, it is probably not so correct to extend to a whole population a parameter calculated in a very small sample. The same problems were found for ISI 0,120 and TyG. There is no information regarding multiplying body weight by 0.19 or dividing the triglycerides and glucose product by 2 before converting it into logarithmic form. QUICKI has no parameters or constants, but its validation can be disputable, because the population sample used by Katz was small (out of 56 subjects, only 13 subjects were affected by obesity) and, moreover, all the subjects were mostly Caucasian[6].
Importantly, it can also be observed that reference intervals, decision values and cut-offs for IR and cardiometabolic risk indexes are determined in various ways. For example, majority of published studies report information about indirect insulin resistance indexes values, given as mean or median [40]. When used,

Conclusions
Men appear more IR than females, also presenting higher levels in cardiometabolic risk indexes. In women the IR/cardiometabolic indexes were more frequently correlated with in ammatory/oxidative stress-related biomarkers (CRP and uric acid), fasting glucose, brinogen, hypertension and WC and BMI. Instead, homocysteine was correlated more recurrently with IR/cardiometabolic indexes. These gender-related differences in the relationship between CV risk factors and IR/cardiometabolic indexes could be a potential pathophysiological determinant for the sex-related epidemiological differences in cardiometabolic diseases.
Thus, with a small group of biomarkers and anthropometric measures it is possible to calculate a number of IR/cardiometabolic indexes, which interpretation may assist for a personalized evaluation of IR and cardiometabolic risk (Figure 1). It is true that different indexes may measure different physiopathological aspects of dysglycemia and cardiometabolic risk, giving different information, which however may also represent an advantage, being more revelatory than one single parameter re ecting only a level of the whole phenomenon.
The major advantage for the use of a panel of indices is that it requires only few biomarkers easily available and normally tested in the general patient evaluation (anthropometric measurements or simple laboratory parameters such as glucose and lipids; only insulin a little more expensive and less diffuse). In any case, it avoids the high costs and di culties for example associated with the euglycemic-hyperglycemic clamp technique for IR+ identi cation, di cult to apply in the clinical practice.