- Email: [email protected]
How many components for the metabolic syndrome? Results of exploratory factor analysis in the FIBAR study
Nutrition, Metabolism & Cardiovascular Diseases (2007) 17, 719e726
www.elsevier.com/locate/nmcd
How many components for the metabolic syndrome? Results of exploratory factor analysis in the FIBAR study E. Mannucci a,b, M. Monami b, C.M. Rotella a,* a
Section of Endocrinology, Department of Clinical Pathophysiology, University of Florence, Viale Pieraccini 6, 50139 Florence, Italy b Geriatric Unit, Critical Care Department, University of Florence, Florence, Italy Received 23 June 2006; received in revised form 30 August 2006; accepted 11 September 2006
KEYWORDS Metabolic syndrome; Exploratory factor analysis; Hyperuricaemia; Glucose intolerance
Abstract Background and aims: Factor analysis can be used as a basis for the determination of diagnostic criteria for the metabolic syndrome (MS). This approach can be used as a basis for the determination of diagnostic criteria for MS. Methods and results: Exploratory factor analysis of Adult Treatment Panel (ATP)-III and International Diabetes Federation (IDF) criteria for MS, entered as dichotomic variables, was performed on 2945 patients enrolled in a screening programme for diabetes. The ability of calculated factors to identify patients with MS-related conditions (glucose intolerance, hyperuricaemia, and elevation of alanine aminotransferase; ALT) was assessed through Receiver Operator Characteristics (ROC) curve analysis. Alternative sets of criteria based on ATP-III and IDF definitions of MS were also assessed. A two-factor structure was found for both ATP-III and IDF criteria. Factor 1 (associated with fasting hyperglycaemia, hypertension, and elevated waist circumference) was capable of identifying subjects with MS-related conditions to a greater extent than factor 2 (low HDL-cholesterol and hypertriglyceridaemia). When a composite variable (low HDL-cholesterol and/or hypertriglyceridaemia) was used for dislipidaemia, a single factor structure was obtained both for ATP-III and IDF definitions; this factor structure was retained when hyperuricaemia was added as a fifth component of MS. Such a modified definition of MS was not inferior to original ATP-III and IDF criteria in the identification of subjects with glucose intolerance and elevated ALT.
* Corresponding author. Tel.: þ39 055 794930; fax: þ39 055 794 9660. E-mail address: [email protected] (C.M. Rotella). 0939-4753/$ - see front matter ª 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.numecd.2006.09.003
720
E. Mannucci et al. Conclusions: A modification of current ATP-III or IDF criteria is necessary in order to obtain a single-factor structure. Alternative definitions of MS, including additional features, such as hyperuricaemia, can maintain a monofactorial structure, and an association with related conditions not inferior to that of original criteria. ª 2006 Elsevier B.V. All rights reserved.
Introduction The choice of what parameters are needed for the diagnosis of metabolic syndrome (MS) has been criticized as arbitrary by several authors [1e3]. In fact, several conditions known to be associated with insulin resistance and increased cardiovascular risk, such as hyperuricaemia [4], non-alcoholic fatty liver disease (NAFLD) [5], and polycystic ovary syndrome (PCOS) [6], have not been included among components of MS by Adult Treatment Panel (ATP)-III [7] or International Diabetes Federation (IDF) [8]. In fact, ATP-III criteria allow the diagnosis of MS when three of the following conditions are satisfied: elevated waist circumference (>102 cm and >88 cm in men and women, respectively), elevated blood pressure (systolic blood pressure 130 mmHg and/or diastolic blood pressure 85 mmHg and/or treatment with antihypertensive drugs) low HDL-cholesterol (<40 mg/dL and <50 mg/dL in men and women, respectively), triglyceride 150 mg/dL, fasting glycaemia 110 mg/dL. Conversely, IDF recommends to perform diagnosis of MS when waist exceeds 93 cm and 79 cm in men and women, respectively, and two of the following are present: elevated blood pressure (see above), low HDL-cholesterol (see above), triglyceride 150 mg/dL, fasting glycaemia 100 mg/dL (or diabetes mellitus). Furthermore, current ATP-III criteria confer the same diagnostic weight to the different components of MS, despite the fact that they show a different strength of association with insulin resistance and cardiovascular risk. Factor analysis is a technique derived from social sciences and psychometric investigations, which is commonly used to detect heterogeneous areas explored by interviews and questionnaires and to identify related scales accordingly. This analysis identifies a small number of virtual variables (termed ‘‘factors’’) related to a larger number of measured parameters, and capable of explaining most of their variance. Factor analysis has been used to assess the homogeneity or heterogeneity of pathogenetic mechanisms underlying a series of clinically detectable alterations, which are associated in epidemiological studies. In this case, each identified
factor can be assumed to correspond to a different pathophysiological mechanism. In other words, a multifactorial structure of a syndrome strongly suggests a pathogenetic heterogeneity of components of the syndrome, despite their epidemiological association; conversely, a single factor structure supports the hypothesis of a common pathogenetic background underlying all clinical manifestations of the syndrome. In particular, factor analysis has been applied to components of the metabolic syndrome in several studies [9], identifying from one to seven distinct factors [10e12]. Most investigations performed so far have identified three or four factors, suggesting a possible heterogeneity of the metabolic syndrome [10e12]. Differences in results among different studies can be partly due to heterogeneity of the populations enrolled; however, a major reason for discrepancies could be the difference in the list of parameters considered. In fact, some studies include insulin levels or other indexes derived from insulinaemia [13], while others do not; the inclusion of hyperuricaemia [13] is also variable. It has also been observed that the use in factor analysis of two parameters closely related to each other, such as diastolic and systolic blood pressure, could interfere with factor analysis, as the two variables tend to identify a distinct factor [9]; this phenomenon can also be observed with fasting and post-load glycaemia, body mass index and waist circumference, and low HDL-cholesterol and hypertriglyceridaemia. Available studies on factor structure of the metabolic syndrome usually considered each parameter of the syndrome as a continuous variable. This procedure increases the sensitivity of the analysis, but it may cause some problems in interpretation of results. In fact, some of the parameters included in the metabolic syndrome, such as high blood pressure, hyperglycaemia, and hypertriglyceridaemia, are the target of specific pharmacological treatments; if values under treatment are considered, this produces a distortion in the analysis. Conversely, if patients receiving any pharmacological treatment for one or more of the components of the metabolic syndrome are excluded, this introduces a relevant selection bias in the sample. Notably, factor analysis can be easily
Metabolic syndrome and factor analysis applied to a set of dichotomic variables, which is often the case for psychometric questionnaires. In the present study, we performed a factor analysis using the five components defined by ATPIII for the diagnosis of metabolic syndrome [7]. These components were used as yes/no categorical variables, so as to include treated as well as untreated patients.
Methods These analyses were performed on the sample of subjects enrolled in the FIrenze-Bagno A Ripoli (FIBAR) study, which has been described elsewhere [14]. Briefly, inhabitants of a section of the city of Florence and of the nearby town of Bagno a Ripoli, aged 40e75 years and without known diabetes, were invited to participate in a screening programme for diabetes. All participants, who gave their informed consent, underwent a physical examination, with measurement of height, weight and waist circumference, according to World Health Organization (WHO) recommendations [15]; blood pressure was measured according to WHO guidelines [16]. A complete medical history, including any previous or current known illness and current medications, alcohol use and smoking habits, was collected. Alcohol use (average daily number of alcoholic drinks) and smoking habits (average daily number of cigarettes) were reported by the patients, in a non-structured interview. Plasma glucose, total and HDL-cholesterol, triglycerides, uric acid, and serum aspartate aminotransferase (ALT), were measured in a sample of venous blood drawn in the morning, after an overnight fast, using an automated method (Aeroset, Abbott Laboratories). All patients underwent a standard Oral Glucose Tolerance Test (OGTT), with measurement of plasma glucose 120 min after the oral ingestion of a 75 g glucose load in a 50% solution in water. Metabolic syndrome was classified according either to ATP-III [7] or IDF criteria [8]. The characteristics of the sample studied are summarized in Table 1. An exploratory factor analysis, using the method of principal components, was used to assess the factor structure of ATP-III and IDF-defined MS, entering each component as a 0/1 dichotomic variable. Varimax and direct oblimin rotations were used to assess the correlation of individual items with underlying factors after extraction. The ability of calculated factors to identify patients with other features related to MS was assessed
721 Table 1
Principal characteristics of the sample Entire sample
Number (% female) Age (years) BMI (kg/m2) Men Women Waist (cm) Men Women FPG (mmol/L) Total cholesterol (mg/dL) HDL-cholesterol (mg/dL) Men Women ALT (IU/L) g-GT(IU/L) Triglycerides (mg/dL) Uric acid (mmol/L) Alcohol consumption (%) Current smokers (%) 2 h PPG 7.8 mmol/L (%) 2 h PPG 11.1 mmol/L (%) Elevation of ALT (%) Hypertension (%) Low HDL-cholesterol (%) Hypertriglyceridaemia (%) High uric acid (%) FPG (ATP-III, %) FPG (IDF, %) Waist (ATP-III, %) Waist (IDF, %)
2945 (56.3%) 55.2 11.5 26.8 3.6 25.4 4.5 97.6 10.3 84.4 11.5 5.49 1.07 5.39 0.96 52.9 13.5 66.7 15.9 23.2 14.4 23 [16; 33] 154 [108; 215] 0.26 0.08 51.8 19.2 12.9 3.0 8.0 47.1 14.0 21.1 6.0 17.0 37.4 30.9 63.9
PPG, post-prandial glucose; FPG, fasting plasma glucose.
through Receiver-Operator Characteristics (ROC) curve analysis. The following conditions associated with MS were considered: (1) Glucose intolerance (120-min post-load glycaemia 7.8 mmol/L); (2) Hyperuricaemia (serum uric acid >0.39 mmol/L, i.e. 6.5 mg/L); (3) ALT >40 U/L, in patients reporting no alcohol consumption and no previous known HBV or HCV infection. This parameter was taken as an approximate measure of Non-Alcoholic Fatty Liver Disease (NAFLD), which is known to be associated with MS and insulin resistance [5,17]. Considering that the two components of MS related to lipid profile (hypertriglyceridaemia and low HDL-cholesterol) are highly correlated, and that this might interfere with the assessment of factor structure of the syndrome [9], we repeated all the analyses described above, for both ATP-III
722
E. Mannucci et al.
and IDF definitions of the syndrome, combining hypertriglyceridaemia and low HDL-cholesterol in a composite ‘‘dyslipidaemia’’ component. An alternative definition for ATP-III and IDF MS, in which low HDL-cholesterol and hypertriglyceridaemia were combined in a single composite dyslipidaemia variable, and hyperuricaemia (>6.5 mg/dL) was added as a fifth component, was studied through factor analysis; its association with glucose intolerance and elevated ALT in patients reporting no alcohol consumption or previous HBV/HCV infection was also assessed, as described above. Data with normal or non-normal distribution are expressed as mean SD and median [quartiles], respectively. Chi-square test was used for comparisons. All statistical analysis was performed using SPSS 12.0.
Results Exploratory factor analysis on the ATP-III-defined metabolic syndrome revealed a two-factor structure (Table 2), accounting for 55.7% of variance. Similar results were obtained when women and men were analysed separately (data not shown). The rotated solution of factor analysis (Table 3) showed that factor 1 was determined mainly by fasting hyperglycaemia, hypertension, and elevated waist circumference, while factor 2 was related to low HDL-cholesterol and hypertriglyceridaemia. The same result was obtained with direct oblimin rotation (data not shown). Similar results were obtained with IDF definition of MS (Tables 2 and 3). When those two factors were used for the identification of subjects with post-load glycaemia 7.8 mmol/L, factor 1, but not factor 2, was significantly associated with glucose intolerance. Similarly, higher scores on factor 1 were associated with hyperuricaemia and, in those who did not report alcohol consumption or HBV/HCV infection, with elevated ALT; however, those two conditions were also associated, although to a lesser extent, with factor 2 (Tables 4 and 5). Table 2 factors Factor 1 2 3 4
Table 3 Rotated solution of factor analysis of components of MS, ATP-III and IDF criteria MS parameters
ATP-III
IDF
Factor 1 Factor 2 Factor 1 Factor 2 High blood 0.53* pressure Waist 0.44* circumference Fasting plasma 0.45* glucose HDL-Cholesterol 0.16 Triglycerides 0.02
0.12
0.75*
0.01
0.01
0.69*
0.12
0.04
0.66*
0.10
0.69* 0.56*
0.02 0.23
0.85* 0.73*
*P < 0.001.
In subjects aged 65 years (N ¼ 760), ATP-IIIdefined MS showed a two-factor structure similar to that observed in younger individuals (data not shown). Factor 1 (i.e., elevated glucose, waist circumference and blood pressure) was associated with glucose intolerance both in younger and older subjects (area under the ROC curve: 0.74 [0.71e0.78] and 0.69 [0.64e0.75], respectively; both P < 0.01), while factor 2 (dyslipidaemia) was associated with post-load hyperglycaemia in older individuals only (area under the ROC curve 0.59 [0.54e0.64]; P < 0.05). Conversely, hyperuricaemia was associated with higher scores of both factor 1 and 2 in those aged less than 65 years, and only with factor 1 in elderly subjects (data not shown). When a single composed variable, including both low HDL-cholesterol and hypertriglyceridaemia, was used for dislipidaemia, a single factor structure was obtained both for ATP-III and IDF definitions (Table 2), which explained 40.3% and 41.0% of variance, respectively. Similar results were obtained when younger (<65 years) and older (65 years) subjects were analysed separately. The correlation of the underlying factor with hyperglycaemia, hypertension, elevated waist circumference, and dislipidaemia, was 0.64, 0.67, 0.68, and 0.54, respectively, for ATP-III definition, and 0.75, 0.68, 0.69, and 0.53, respectively, for
Principal component analysis of the metabolic syndrome. Eigenvalue (% variance explained) of individual ATP-III
IDF
Original
4-components
Modified
Original
4-components
Modified
1.74 (34.6) 1.05 (21.1) 0.79 0.75
1.61 (40.3) 0.87 0.79 0.73
1.72 (34.2) 0.98 0.96 0.85
1.76 (35.2) 1.06 (21.3) 0.79 0.73
1.65 (41.0) 0.87 0.78 0.70
1.73 (34.6) 0.94 0.85 0.78
Metabolic syndrome and factor analysis Table 4
723
Areas under ROC curves for factors underlying ATP-III-defined MS (with original and modified criteria) MODEL 1
2 h PG 11.1 mmol/L 2 h PG >7.8 mmol/L Hyperuricaemiaa AST >40 IUb
MODEL 2
MODEL 3
Factor 1
Factor 2
Factor 1
Factor 1
0.86 0.74 0.70 0.61
0.47 0.50 0.56 0.54
0.87 0.76 0.72 0.63
0.86 [0.83e0.90] 0.76 [0.73e0.79] e 0.64 [0.58e0.69]
[0.83e0.90] [0.72e0.77] [0.67e0.74] [0.55e0.67]
[0.39e0.55] [0.46e0.53] [0.51e0.61] [0.47e0.60]
[0.84e0.91] [0.74e0.79] [0.69e0.76] [0.57e0.69]
Model 1: 5 parameters, original criteria; Model 2: 4 parameters; Model 3: 5 parameters, modified criteria. PG, plasma glucose. a Uric acid >0.39 mmol/L. b In those who do not report alcohol consumption or HBV/HCV infection.
IDF definition. The single factor derived from this analysis on either ATP-III or IDF criteria was capable of predicting post-load hyperglycaemia, hyperuricaemia, and elevation of AST, at least to the same extent as Factor 1 of the previous analysis (Tables 4 and 5). If hyperuricaemia was entered as a component of MS, together with ATP-III-defined fasting hyperglycaemia and elevated waist circumference, high blood pressure, and dyslipidaemia (hypertriglycaeridaemia and/or low HDL-cholesterol), factor analysis revealed a single-factor structure, accounting for 34.2% of variance (Table 2). The correlation with the underlying factor of hypertension, elevated waist circumference, hyperglycaemia, dyslipidaemia, and hyperuricaemia was 0.65, 0.64, 0.62, 0.54, and 0.45, respectively. When a similar analysis was performed on IDF criteria for MS, a single factor structure was also found, accounting for 34.6% of variance; the correlation of this factor with hypertension, elevated waist circumference, hyperglycaemia, dyslipidaemia, and hyperuricaemia was 0.67, 0.66, 0.62, 0.53, and 0.42, respectively. When the calculated factor derived from ATP-III or IDF criteria was used in the prediction of glucose intolerance, diabetes, or elevated ALT, the area under the ROC curve was similar to that observed with factor 1 of the original model, or with the single factor of the 4-component model of MS (Tables 4 and 5).
Table 5
The prevalence of glucose intolerance and elevation of ALT increased with the number of components of MS, using both the standard ATP-III definition and the newly proposed definition, which includes hyperuricaemia (Fig. 1). Using this new definition of MS, the prevalence of the syndrome in the sample was not significantly different from that observed with ATP-III criteria (17.1% vs. 17.6%); of the 517 patients with ATP-III-defined NCEP, 453 (86.0%) also fulfilled the alternative set of diagnostic criteria; conversely, 52 (2.1%) of those who did not fulfil ATP-III criteria would have received a diagnosis of MS if the alternative set of criteria had been used. Similarly, the modification of IDF diagnostic criteria did not alter prevalence in a relevant manner (29.7% vs. 29.9%). Among the 874 patients with MS, as defined by original IDF criteria, 847 (96.9%) also fulfilled modified criteria, while 34 (1.6%) of the patients who did not meet original IDF criteria would have been classified as affected by MS should the modified criteria have been used. Patients fulfilling both original and modified ATPIII criteria for MS showed a significantly (P < 0.01) higher prevalence of glucose intolerance in comparison with the rest of the sample. Those who fulfilled original, but not modified, criteria, showed a prevalence of glucose intolerance that was significantly lower (P < 0.05) than those fulfilling both original and modified criteria. Conversely, the prevalence
Areas under ROC curves for factors underlying IDF-defined MS (with original and modified criteria) MODEL 1
2 h PG 11.1 mmol/L 2 h PG 7.8 mmol/L Hyperuricaemiaa AST >40 IUb
MODEL 2
MODEL 3
Factor 1
Factor 2
Factor 1
Factor 1
0.80 0.73 0.71 0.64
0.48 0.49 0.55 0.49
0.82 0.74 0.72 0.65
0.80 [0.76e0.85] 0.74 [0.71e0.77] e 0.66 [0.60e0.71]
[0.76e0.84] [0.70e0.75] [0.67e0.74] [0.59e0.70]
[0.40e0.55] [0.45e0.52] [0.50e0.60] [0.42e0.55]
[0.77e0.86] [0.72e0.77] [0.68e0.75] [0.60e0.71]
Model 1: 5 parameters, original criteria; Model 2: 4 parameters; Model 3: 5 parameters, modified criteria. PG, plasma glucose. a Uric acid >0.39 mmol/L. b In those who do not report alcohol consumption or HBV/HCV infection.
724
E. Mannucci et al. 40,0
Panel A 50,0
35,0
45,0 30,0
(%)
40,0
25,0
30,0
20,0
%
35,0
25,0
15,0
20,0 10,0
15,0
5,0
10,0 5,0
0,0
0,0 0
1
2
3
4
5
Components of MS
MS-original
-
-
+
+
MS-modified
-
+
-
+
Figure 2 Prevalence (%) of glucose intolerance (120 min post-load glycaemia 7.8 mmol/L) in patients fulfilling original and modified ATP-III criteria.
Panel B 50,0 45,0 40,0 35,0
(%)
30,0 25,0 20,0 15,0 10,0 5,0 0,0 Normal waist
0
1
2
3
4
Components of MS
Figure 1 Prevalence of glucose intolerance (120 min post-load glycaemia 7.8 mmol/L) in patients with a different number of components of MS, with original (black line) and modified (dotted black line) diagnostic criteria. Panel A: ATP-III; panel B: IDF.
of glucose intolerance in individuals identified as affected by MS through modified, but not original, ATP-III criteria was not significantly different from that observed in those positive for both sets of criteria (Fig. 2).
Discussion Factor analysis of the components of MS revealed two distinct factors, the first of which was related to adiposity, hyperglycaemia and hypertension, while the second was determined by lipid alterations. Of the two factors, the first appears to be more directly related to insulin resistance; in fact, it is predictive of other conditions which are known to be associated with reduced insulin sensitivity, such as hyperuricaemia [18], glucose intolerance
[12], and non-alcoholic fatty liver disease [19]. Conversely, the other factor appears to be only marginally associated with such conditions. It has been suggested that the inclusion in the model of closely associated variables can interfere with factor analysis; in fact, each couple of related variables tends to identify a distinct factor, which often disappears if only one of the associated parameters is considered [9]. Such a phenomenon was described for fasting and post-load glycaemia, systolic and diastolic blood pressure, body mass index and waist circumference, and low HDL-cholesterol and hypertriglyceridaemia [20]. In fact, when the two lipid alterations of MS were replaced with a single, composite variable, including both of them, a single factor was sufficient to explain most of the variance, confirming previous observations on the monofactorial structure of MS [9e12]. The factor derived from this second analysis, based on a four-component model of MS, was at least as effective in the prediction of related metabolic abnormalities as the usual fivecomponent model proposed by ATP-III. In fact, the consideration of two closely related conditions, such as low HDL-cholesterol and hypertriglyceridaemia, as distinct components of MS, could determine an overestimation of the role of dyslipidaemia associated with insulin resistance. A definition of MS in which the two alterations are considered as one component is plausible. Similarly, elevated systolic and diastolic blood pressure, which are similarly closely related are considered as one, and not two, components of MS. Among conditions associated with the metabolic syndrome, but not included in current ATP-III or
Metabolic syndrome and factor analysis IDF diagnostic criteria, hyperuricaemia deserves some special consideration. In fact, this condition is known to be associated with insulin resistance [18,21] and increased cardiovascular risk [22], and it was included in earlier descriptions of the metabolic syndrome. Our data show that the inclusion of hyperuricaemia among the diagnostic criteria of MS does not interfere with its monofactorial structure. This result suggests that a homogeneous pathophysiological mechanism could underlie hyperuricaemia and the other components of MS, which were included in ATP-III and IDF diagnostic criteria. Using both ATP-III and IDF criteria, combining low HDL-cholesterol and hypertriglyceridaemia, and adding hyperuricaemia as the fifth component, a single factor structure is preserved. Furthermore, the association of MS identified using these modified criteria does not seem to be inferior to that of the syndrome defined with the original sets of diagnostic criteria. On the other hand, on the basis of this cross-sectional analysis, the addition of hyperuricaemia to the definition of MS does not seem to improve the association of the syndrome with other conditions related to insulin resistance. Some limitations of the present study should be recognized. The sample enrolled cannot be considered representative of the general population. In fact, although all residents of the area were invited, only a minority of subjects participated in this screening programme for diabetes; those with risk factors for glucose intolerance were overrepresented, and this accounts for the relatively high prevalence of components of MS in the sample. Another limitation is represented by the fact that insulin measurements were not available, preventing an assessment of insulin sensitivity in this sample. Even though the adequacy of any definition of MS in the evaluation of insulin resistance has been questioned [1], such a measure would have provided additional information on the pathophysiological mechanisms related to factors identified through factor analysis. Furthermore, other parameters of possible interest, such as adiponectin or C-reactive protein, were not available in the present survey; in fact, low adiponectin or elevated C-reactive protein, which are known to be associated with insulin resistance [23] and cardiovascular morbidity [24], deserve to be evaluated for inclusion among criteria for MS. On the other hand, the FIBAR sample can provide a typical picture of subjects undergoing screening for diabetes and glucose intolerance. Furthermore, the over-representation of MS in this sample, in comparison with the general population, can facilitate the statistical analyses performed in the present study.
725 In conclusion, factor analysis of MS, based on the five components identified by ATP-III or IDF, reveals either a one- or two-factor structure, depending on the use of low HDL-cholesterol or hypertriglyceridaemia as two distinct or one composite variable. The combination of the two lipid alterations of MS allows obtaining a single-factor structure, which is maintained if hyperuricaemia is added among the diagnostic components. These alternative, modified definitions of MS do not seem inferior to the original ATP-III and IDF sets of criteria in their association with glucose intolerance and NAFLD, although longitudinal studies are needed to assess the predictive value of different definitions of MS for diabetes and cardiovascular disease. Other conditions, different from hyperuricaemia, could be taken into account for a possible inclusion among diagnostic criteria for MS; however, factor analysis should be considered a valuable aid in the definition of criteria for diagnosis of composite categories, such as the metabolic syndrome. An alternative definition of MS, which includes hyperuricaemia and unifies low HDL-cholesterol and hypertriglyceridaemia as a single composite component, deserves further investigation.
Acknowledgements This work was supported by grants from Menarini Diagnostics International, Florence, Italy, and from the Italian Ministry of University and Scientific Research (PRIN Projects).
References [1] Reaven GM. The metabolic syndrome: requiescat in pace. Clin Chem 2005;51:931e8. [2] Reynolds K, Muntner P, Fonseca V. Metabolic syndrome underrated or underdiagnosed? Diabetes Care 2005;28: 1831e2. [3] Kahn R, Buse J, Ferrannini E, Stern M. The metabolic syndrome: time for a critical appraisal. Joint statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetologia 2005;48: 1684e99. [4] Schachter M. Uric acid and hypertension. Curr Pharm Des 2005;11:4139e43. [5] Bugianesi E, Gastaldelli A, Vanni E, Gambino R, Cassader M, Baldi S, et al. Insulin resistance in non-diabetic patients with non-alcoholic fatty liver disease sites and mechanisms. Diabetologia 2005;48:634e42. [6] Ehrmann DA. Polycystic ovary syndrome. N Engl J Med 2005;352:1223e36. [7] Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood
726
[8] [9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
E. Mannucci et al. Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001;285:2486e97. Alberti KG, Zimmet P, Shaw J. The metabolic syndromeda new worldwide definition. Lancet 2005;366:1059e62. Pladevall M, Singal B, Williams LK, Brotons C, Guyer H, Sadurni J, et al. A single factor underlies the metabolic syndrome A confirmatory factor analysis. Diabetes Care 2006;29:113e22. Wang J, Qiao Q, Miettinen ME, Lappalainen J, Hu G, Tuomilehto J. The metabolic syndrome defined by factor analysis and incident type 2 diabetes in a Chinese population with high postprandial glucose. Diabetes Care 2004; 27:2429e37. Oh JY, Hong YS, Sung YA, Barrett-Connor E. Prevalence and factor analysis of metabolic syndrome in an urban Korean population. Diabetes Care 2004;27:2027e32. Hanson RL, Imperatore G, Bennett PH, Knowler WC. Components of the ‘‘metabolic syndrome’’ and incidence of type 2 diabetes. Diabetes 2002;51:3120e7. Boyko EJ, de Court, Zimmet PZ, Chitson P, Tuomilehto J, Alberti KG. Features of the metabolic syndrome predict higher risk of diabetes and impaired glucose tolerance a prospective study in Mauritius. Diabetes Care 2000;23:1242e8. Mannucci E, Ognibene A, Sposato I, Brogi M, Gallori G, Bardini G, et al. Fasting plasma glucose and glycated haemoglobin in the screening of diabetes and impaired glucose tolerance. Acta Diabetol 2003;40:181e6. Obesity preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser 2000;89:42e53. Chalmers J, MacMahon S, Mancia G, Whitworth J, Beilin L, Hansson L, et al. 1999 World Health Organization-International Society of Hypertension Guidelines for the management of hypertension. Guidelines sub-committee of the World Health Organization. Clin Exp Hypertens 1999;21: 1009e60.
[17] Marchesini G, Avagnina S, Barantani EG, Ciccarone AM, Corica F, Dall’Aglio E, et al. Aminotransferase and gamma-glutamyltranspeptidase levels in obesity are associated with insulin resistance and the metabolic syndrome. J Endocrinol Invest 2005;28:333e9. [18] Clausen JO, Borch-Johnsen K, Ibsen H, Pedersen O. Analysis of the relationship between fasting serum uric acid and the insulin sensitivity index in a population-based sample of 380 young healthy Caucasians. Eur J Endocrinol 1998; 13:863e9. [19] Choudhury J, Sanyal AJ. Insulin resistance in NASH. Front Biosci 2005;10:1520e33. [20] Meigs JB, D’Agostino Sr RB, Wilson PW, Cupples LA, Nathan DM, Singer DE. Risk variable clustering in the insulin resistance syndrome. The Framingham Offspring Study. Diabetes 1997;46:1594e600. [21] Tsunoda S, Kamide K, Minami J, Kawano Y. Decreases in serum uric acid by amelioration of insulin resistance in overweight hypertensive patients effect of a low-energy diet and an insulin-sensitizing agent. Am J Hypertens 2002;15: 697e701. [22] Conen D, Wietlisbach V, Bovet P, ShamLaye C, Riesen W, Paccaud F, et al. Prevalence of hyperuricemia and relation of serum uric acid with cardiovascular risk factors in a developing country. BMC Public Health 2004;49: 155e7. [23] Yuan G, Zhou L, Tang J, Yang Y, Gu W, Li F, et al. Serum CRP levels are equally elevated in newly diagnosed type 2 diabetes and impaired glucose tolerance and related to adiponectin levels and insulin sensitivity. Diabetes Res Clin Pract 2005;72:244e50. [24] Sattar N, Gaw A, Scherbakova O, Ford I, O’Reilly DS, Haffner SM, et al. Metabolic syndrome with and without C-reactive protein as a predictor of coronary heart disease and diabetes in the West of Scotland Coronary Prevention Study. Circulation 2003;108:414e9.
Available online at www.sciencedirect.com
www.elsevier.com/locate/nmcd
How many components for the metabolic syndrome? Results of exploratory factor analysis in the FIBAR study E. Mannucci a,b, M. Monami b, C.M. Rotella a,* a
Section of Endocrinology, Department of Clinical Pathophysiology, University of Florence, Viale Pieraccini 6, 50139 Florence, Italy b Geriatric Unit, Critical Care Department, University of Florence, Florence, Italy Received 23 June 2006; received in revised form 30 August 2006; accepted 11 September 2006
KEYWORDS Metabolic syndrome; Exploratory factor analysis; Hyperuricaemia; Glucose intolerance
Abstract Background and aims: Factor analysis can be used as a basis for the determination of diagnostic criteria for the metabolic syndrome (MS). This approach can be used as a basis for the determination of diagnostic criteria for MS. Methods and results: Exploratory factor analysis of Adult Treatment Panel (ATP)-III and International Diabetes Federation (IDF) criteria for MS, entered as dichotomic variables, was performed on 2945 patients enrolled in a screening programme for diabetes. The ability of calculated factors to identify patients with MS-related conditions (glucose intolerance, hyperuricaemia, and elevation of alanine aminotransferase; ALT) was assessed through Receiver Operator Characteristics (ROC) curve analysis. Alternative sets of criteria based on ATP-III and IDF definitions of MS were also assessed. A two-factor structure was found for both ATP-III and IDF criteria. Factor 1 (associated with fasting hyperglycaemia, hypertension, and elevated waist circumference) was capable of identifying subjects with MS-related conditions to a greater extent than factor 2 (low HDL-cholesterol and hypertriglyceridaemia). When a composite variable (low HDL-cholesterol and/or hypertriglyceridaemia) was used for dislipidaemia, a single factor structure was obtained both for ATP-III and IDF definitions; this factor structure was retained when hyperuricaemia was added as a fifth component of MS. Such a modified definition of MS was not inferior to original ATP-III and IDF criteria in the identification of subjects with glucose intolerance and elevated ALT.
* Corresponding author. Tel.: þ39 055 794930; fax: þ39 055 794 9660. E-mail address: [email protected] (C.M. Rotella). 0939-4753/$ - see front matter ª 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.numecd.2006.09.003
720
E. Mannucci et al. Conclusions: A modification of current ATP-III or IDF criteria is necessary in order to obtain a single-factor structure. Alternative definitions of MS, including additional features, such as hyperuricaemia, can maintain a monofactorial structure, and an association with related conditions not inferior to that of original criteria. ª 2006 Elsevier B.V. All rights reserved.
Introduction The choice of what parameters are needed for the diagnosis of metabolic syndrome (MS) has been criticized as arbitrary by several authors [1e3]. In fact, several conditions known to be associated with insulin resistance and increased cardiovascular risk, such as hyperuricaemia [4], non-alcoholic fatty liver disease (NAFLD) [5], and polycystic ovary syndrome (PCOS) [6], have not been included among components of MS by Adult Treatment Panel (ATP)-III [7] or International Diabetes Federation (IDF) [8]. In fact, ATP-III criteria allow the diagnosis of MS when three of the following conditions are satisfied: elevated waist circumference (>102 cm and >88 cm in men and women, respectively), elevated blood pressure (systolic blood pressure 130 mmHg and/or diastolic blood pressure 85 mmHg and/or treatment with antihypertensive drugs) low HDL-cholesterol (<40 mg/dL and <50 mg/dL in men and women, respectively), triglyceride 150 mg/dL, fasting glycaemia 110 mg/dL. Conversely, IDF recommends to perform diagnosis of MS when waist exceeds 93 cm and 79 cm in men and women, respectively, and two of the following are present: elevated blood pressure (see above), low HDL-cholesterol (see above), triglyceride 150 mg/dL, fasting glycaemia 100 mg/dL (or diabetes mellitus). Furthermore, current ATP-III criteria confer the same diagnostic weight to the different components of MS, despite the fact that they show a different strength of association with insulin resistance and cardiovascular risk. Factor analysis is a technique derived from social sciences and psychometric investigations, which is commonly used to detect heterogeneous areas explored by interviews and questionnaires and to identify related scales accordingly. This analysis identifies a small number of virtual variables (termed ‘‘factors’’) related to a larger number of measured parameters, and capable of explaining most of their variance. Factor analysis has been used to assess the homogeneity or heterogeneity of pathogenetic mechanisms underlying a series of clinically detectable alterations, which are associated in epidemiological studies. In this case, each identified
factor can be assumed to correspond to a different pathophysiological mechanism. In other words, a multifactorial structure of a syndrome strongly suggests a pathogenetic heterogeneity of components of the syndrome, despite their epidemiological association; conversely, a single factor structure supports the hypothesis of a common pathogenetic background underlying all clinical manifestations of the syndrome. In particular, factor analysis has been applied to components of the metabolic syndrome in several studies [9], identifying from one to seven distinct factors [10e12]. Most investigations performed so far have identified three or four factors, suggesting a possible heterogeneity of the metabolic syndrome [10e12]. Differences in results among different studies can be partly due to heterogeneity of the populations enrolled; however, a major reason for discrepancies could be the difference in the list of parameters considered. In fact, some studies include insulin levels or other indexes derived from insulinaemia [13], while others do not; the inclusion of hyperuricaemia [13] is also variable. It has also been observed that the use in factor analysis of two parameters closely related to each other, such as diastolic and systolic blood pressure, could interfere with factor analysis, as the two variables tend to identify a distinct factor [9]; this phenomenon can also be observed with fasting and post-load glycaemia, body mass index and waist circumference, and low HDL-cholesterol and hypertriglyceridaemia. Available studies on factor structure of the metabolic syndrome usually considered each parameter of the syndrome as a continuous variable. This procedure increases the sensitivity of the analysis, but it may cause some problems in interpretation of results. In fact, some of the parameters included in the metabolic syndrome, such as high blood pressure, hyperglycaemia, and hypertriglyceridaemia, are the target of specific pharmacological treatments; if values under treatment are considered, this produces a distortion in the analysis. Conversely, if patients receiving any pharmacological treatment for one or more of the components of the metabolic syndrome are excluded, this introduces a relevant selection bias in the sample. Notably, factor analysis can be easily
Metabolic syndrome and factor analysis applied to a set of dichotomic variables, which is often the case for psychometric questionnaires. In the present study, we performed a factor analysis using the five components defined by ATPIII for the diagnosis of metabolic syndrome [7]. These components were used as yes/no categorical variables, so as to include treated as well as untreated patients.
Methods These analyses were performed on the sample of subjects enrolled in the FIrenze-Bagno A Ripoli (FIBAR) study, which has been described elsewhere [14]. Briefly, inhabitants of a section of the city of Florence and of the nearby town of Bagno a Ripoli, aged 40e75 years and without known diabetes, were invited to participate in a screening programme for diabetes. All participants, who gave their informed consent, underwent a physical examination, with measurement of height, weight and waist circumference, according to World Health Organization (WHO) recommendations [15]; blood pressure was measured according to WHO guidelines [16]. A complete medical history, including any previous or current known illness and current medications, alcohol use and smoking habits, was collected. Alcohol use (average daily number of alcoholic drinks) and smoking habits (average daily number of cigarettes) were reported by the patients, in a non-structured interview. Plasma glucose, total and HDL-cholesterol, triglycerides, uric acid, and serum aspartate aminotransferase (ALT), were measured in a sample of venous blood drawn in the morning, after an overnight fast, using an automated method (Aeroset, Abbott Laboratories). All patients underwent a standard Oral Glucose Tolerance Test (OGTT), with measurement of plasma glucose 120 min after the oral ingestion of a 75 g glucose load in a 50% solution in water. Metabolic syndrome was classified according either to ATP-III [7] or IDF criteria [8]. The characteristics of the sample studied are summarized in Table 1. An exploratory factor analysis, using the method of principal components, was used to assess the factor structure of ATP-III and IDF-defined MS, entering each component as a 0/1 dichotomic variable. Varimax and direct oblimin rotations were used to assess the correlation of individual items with underlying factors after extraction. The ability of calculated factors to identify patients with other features related to MS was assessed
721 Table 1
Principal characteristics of the sample Entire sample
Number (% female) Age (years) BMI (kg/m2) Men Women Waist (cm) Men Women FPG (mmol/L) Total cholesterol (mg/dL) HDL-cholesterol (mg/dL) Men Women ALT (IU/L) g-GT(IU/L) Triglycerides (mg/dL) Uric acid (mmol/L) Alcohol consumption (%) Current smokers (%) 2 h PPG 7.8 mmol/L (%) 2 h PPG 11.1 mmol/L (%) Elevation of ALT (%) Hypertension (%) Low HDL-cholesterol (%) Hypertriglyceridaemia (%) High uric acid (%) FPG (ATP-III, %) FPG (IDF, %) Waist (ATP-III, %) Waist (IDF, %)
2945 (56.3%) 55.2 11.5 26.8 3.6 25.4 4.5 97.6 10.3 84.4 11.5 5.49 1.07 5.39 0.96 52.9 13.5 66.7 15.9 23.2 14.4 23 [16; 33] 154 [108; 215] 0.26 0.08 51.8 19.2 12.9 3.0 8.0 47.1 14.0 21.1 6.0 17.0 37.4 30.9 63.9
PPG, post-prandial glucose; FPG, fasting plasma glucose.
through Receiver-Operator Characteristics (ROC) curve analysis. The following conditions associated with MS were considered: (1) Glucose intolerance (120-min post-load glycaemia 7.8 mmol/L); (2) Hyperuricaemia (serum uric acid >0.39 mmol/L, i.e. 6.5 mg/L); (3) ALT >40 U/L, in patients reporting no alcohol consumption and no previous known HBV or HCV infection. This parameter was taken as an approximate measure of Non-Alcoholic Fatty Liver Disease (NAFLD), which is known to be associated with MS and insulin resistance [5,17]. Considering that the two components of MS related to lipid profile (hypertriglyceridaemia and low HDL-cholesterol) are highly correlated, and that this might interfere with the assessment of factor structure of the syndrome [9], we repeated all the analyses described above, for both ATP-III
722
E. Mannucci et al.
and IDF definitions of the syndrome, combining hypertriglyceridaemia and low HDL-cholesterol in a composite ‘‘dyslipidaemia’’ component. An alternative definition for ATP-III and IDF MS, in which low HDL-cholesterol and hypertriglyceridaemia were combined in a single composite dyslipidaemia variable, and hyperuricaemia (>6.5 mg/dL) was added as a fifth component, was studied through factor analysis; its association with glucose intolerance and elevated ALT in patients reporting no alcohol consumption or previous HBV/HCV infection was also assessed, as described above. Data with normal or non-normal distribution are expressed as mean SD and median [quartiles], respectively. Chi-square test was used for comparisons. All statistical analysis was performed using SPSS 12.0.
Results Exploratory factor analysis on the ATP-III-defined metabolic syndrome revealed a two-factor structure (Table 2), accounting for 55.7% of variance. Similar results were obtained when women and men were analysed separately (data not shown). The rotated solution of factor analysis (Table 3) showed that factor 1 was determined mainly by fasting hyperglycaemia, hypertension, and elevated waist circumference, while factor 2 was related to low HDL-cholesterol and hypertriglyceridaemia. The same result was obtained with direct oblimin rotation (data not shown). Similar results were obtained with IDF definition of MS (Tables 2 and 3). When those two factors were used for the identification of subjects with post-load glycaemia 7.8 mmol/L, factor 1, but not factor 2, was significantly associated with glucose intolerance. Similarly, higher scores on factor 1 were associated with hyperuricaemia and, in those who did not report alcohol consumption or HBV/HCV infection, with elevated ALT; however, those two conditions were also associated, although to a lesser extent, with factor 2 (Tables 4 and 5). Table 2 factors Factor 1 2 3 4
Table 3 Rotated solution of factor analysis of components of MS, ATP-III and IDF criteria MS parameters
ATP-III
IDF
Factor 1 Factor 2 Factor 1 Factor 2 High blood 0.53* pressure Waist 0.44* circumference Fasting plasma 0.45* glucose HDL-Cholesterol 0.16 Triglycerides 0.02
0.12
0.75*
0.01
0.01
0.69*
0.12
0.04
0.66*
0.10
0.69* 0.56*
0.02 0.23
0.85* 0.73*
*P < 0.001.
In subjects aged 65 years (N ¼ 760), ATP-IIIdefined MS showed a two-factor structure similar to that observed in younger individuals (data not shown). Factor 1 (i.e., elevated glucose, waist circumference and blood pressure) was associated with glucose intolerance both in younger and older subjects (area under the ROC curve: 0.74 [0.71e0.78] and 0.69 [0.64e0.75], respectively; both P < 0.01), while factor 2 (dyslipidaemia) was associated with post-load hyperglycaemia in older individuals only (area under the ROC curve 0.59 [0.54e0.64]; P < 0.05). Conversely, hyperuricaemia was associated with higher scores of both factor 1 and 2 in those aged less than 65 years, and only with factor 1 in elderly subjects (data not shown). When a single composed variable, including both low HDL-cholesterol and hypertriglyceridaemia, was used for dislipidaemia, a single factor structure was obtained both for ATP-III and IDF definitions (Table 2), which explained 40.3% and 41.0% of variance, respectively. Similar results were obtained when younger (<65 years) and older (65 years) subjects were analysed separately. The correlation of the underlying factor with hyperglycaemia, hypertension, elevated waist circumference, and dislipidaemia, was 0.64, 0.67, 0.68, and 0.54, respectively, for ATP-III definition, and 0.75, 0.68, 0.69, and 0.53, respectively, for
Principal component analysis of the metabolic syndrome. Eigenvalue (% variance explained) of individual ATP-III
IDF
Original
4-components
Modified
Original
4-components
Modified
1.74 (34.6) 1.05 (21.1) 0.79 0.75
1.61 (40.3) 0.87 0.79 0.73
1.72 (34.2) 0.98 0.96 0.85
1.76 (35.2) 1.06 (21.3) 0.79 0.73
1.65 (41.0) 0.87 0.78 0.70
1.73 (34.6) 0.94 0.85 0.78
Metabolic syndrome and factor analysis Table 4
723
Areas under ROC curves for factors underlying ATP-III-defined MS (with original and modified criteria) MODEL 1
2 h PG 11.1 mmol/L 2 h PG >7.8 mmol/L Hyperuricaemiaa AST >40 IUb
MODEL 2
MODEL 3
Factor 1
Factor 2
Factor 1
Factor 1
0.86 0.74 0.70 0.61
0.47 0.50 0.56 0.54
0.87 0.76 0.72 0.63
0.86 [0.83e0.90] 0.76 [0.73e0.79] e 0.64 [0.58e0.69]
[0.83e0.90] [0.72e0.77] [0.67e0.74] [0.55e0.67]
[0.39e0.55] [0.46e0.53] [0.51e0.61] [0.47e0.60]
[0.84e0.91] [0.74e0.79] [0.69e0.76] [0.57e0.69]
Model 1: 5 parameters, original criteria; Model 2: 4 parameters; Model 3: 5 parameters, modified criteria. PG, plasma glucose. a Uric acid >0.39 mmol/L. b In those who do not report alcohol consumption or HBV/HCV infection.
IDF definition. The single factor derived from this analysis on either ATP-III or IDF criteria was capable of predicting post-load hyperglycaemia, hyperuricaemia, and elevation of AST, at least to the same extent as Factor 1 of the previous analysis (Tables 4 and 5). If hyperuricaemia was entered as a component of MS, together with ATP-III-defined fasting hyperglycaemia and elevated waist circumference, high blood pressure, and dyslipidaemia (hypertriglycaeridaemia and/or low HDL-cholesterol), factor analysis revealed a single-factor structure, accounting for 34.2% of variance (Table 2). The correlation with the underlying factor of hypertension, elevated waist circumference, hyperglycaemia, dyslipidaemia, and hyperuricaemia was 0.65, 0.64, 0.62, 0.54, and 0.45, respectively. When a similar analysis was performed on IDF criteria for MS, a single factor structure was also found, accounting for 34.6% of variance; the correlation of this factor with hypertension, elevated waist circumference, hyperglycaemia, dyslipidaemia, and hyperuricaemia was 0.67, 0.66, 0.62, 0.53, and 0.42, respectively. When the calculated factor derived from ATP-III or IDF criteria was used in the prediction of glucose intolerance, diabetes, or elevated ALT, the area under the ROC curve was similar to that observed with factor 1 of the original model, or with the single factor of the 4-component model of MS (Tables 4 and 5).
Table 5
The prevalence of glucose intolerance and elevation of ALT increased with the number of components of MS, using both the standard ATP-III definition and the newly proposed definition, which includes hyperuricaemia (Fig. 1). Using this new definition of MS, the prevalence of the syndrome in the sample was not significantly different from that observed with ATP-III criteria (17.1% vs. 17.6%); of the 517 patients with ATP-III-defined NCEP, 453 (86.0%) also fulfilled the alternative set of diagnostic criteria; conversely, 52 (2.1%) of those who did not fulfil ATP-III criteria would have received a diagnosis of MS if the alternative set of criteria had been used. Similarly, the modification of IDF diagnostic criteria did not alter prevalence in a relevant manner (29.7% vs. 29.9%). Among the 874 patients with MS, as defined by original IDF criteria, 847 (96.9%) also fulfilled modified criteria, while 34 (1.6%) of the patients who did not meet original IDF criteria would have been classified as affected by MS should the modified criteria have been used. Patients fulfilling both original and modified ATPIII criteria for MS showed a significantly (P < 0.01) higher prevalence of glucose intolerance in comparison with the rest of the sample. Those who fulfilled original, but not modified, criteria, showed a prevalence of glucose intolerance that was significantly lower (P < 0.05) than those fulfilling both original and modified criteria. Conversely, the prevalence
Areas under ROC curves for factors underlying IDF-defined MS (with original and modified criteria) MODEL 1
2 h PG 11.1 mmol/L 2 h PG 7.8 mmol/L Hyperuricaemiaa AST >40 IUb
MODEL 2
MODEL 3
Factor 1
Factor 2
Factor 1
Factor 1
0.80 0.73 0.71 0.64
0.48 0.49 0.55 0.49
0.82 0.74 0.72 0.65
0.80 [0.76e0.85] 0.74 [0.71e0.77] e 0.66 [0.60e0.71]
[0.76e0.84] [0.70e0.75] [0.67e0.74] [0.59e0.70]
[0.40e0.55] [0.45e0.52] [0.50e0.60] [0.42e0.55]
[0.77e0.86] [0.72e0.77] [0.68e0.75] [0.60e0.71]
Model 1: 5 parameters, original criteria; Model 2: 4 parameters; Model 3: 5 parameters, modified criteria. PG, plasma glucose. a Uric acid >0.39 mmol/L. b In those who do not report alcohol consumption or HBV/HCV infection.
724
E. Mannucci et al. 40,0
Panel A 50,0
35,0
45,0 30,0
(%)
40,0
25,0
30,0
20,0
%
35,0
25,0
15,0
20,0 10,0
15,0
5,0
10,0 5,0
0,0
0,0 0
1
2
3
4
5
Components of MS
MS-original
-
-
+
+
MS-modified
-
+
-
+
Figure 2 Prevalence (%) of glucose intolerance (120 min post-load glycaemia 7.8 mmol/L) in patients fulfilling original and modified ATP-III criteria.
Panel B 50,0 45,0 40,0 35,0
(%)
30,0 25,0 20,0 15,0 10,0 5,0 0,0 Normal waist
0
1
2
3
4
Components of MS
Figure 1 Prevalence of glucose intolerance (120 min post-load glycaemia 7.8 mmol/L) in patients with a different number of components of MS, with original (black line) and modified (dotted black line) diagnostic criteria. Panel A: ATP-III; panel B: IDF.
of glucose intolerance in individuals identified as affected by MS through modified, but not original, ATP-III criteria was not significantly different from that observed in those positive for both sets of criteria (Fig. 2).
Discussion Factor analysis of the components of MS revealed two distinct factors, the first of which was related to adiposity, hyperglycaemia and hypertension, while the second was determined by lipid alterations. Of the two factors, the first appears to be more directly related to insulin resistance; in fact, it is predictive of other conditions which are known to be associated with reduced insulin sensitivity, such as hyperuricaemia [18], glucose intolerance
[12], and non-alcoholic fatty liver disease [19]. Conversely, the other factor appears to be only marginally associated with such conditions. It has been suggested that the inclusion in the model of closely associated variables can interfere with factor analysis; in fact, each couple of related variables tends to identify a distinct factor, which often disappears if only one of the associated parameters is considered [9]. Such a phenomenon was described for fasting and post-load glycaemia, systolic and diastolic blood pressure, body mass index and waist circumference, and low HDL-cholesterol and hypertriglyceridaemia [20]. In fact, when the two lipid alterations of MS were replaced with a single, composite variable, including both of them, a single factor was sufficient to explain most of the variance, confirming previous observations on the monofactorial structure of MS [9e12]. The factor derived from this second analysis, based on a four-component model of MS, was at least as effective in the prediction of related metabolic abnormalities as the usual fivecomponent model proposed by ATP-III. In fact, the consideration of two closely related conditions, such as low HDL-cholesterol and hypertriglyceridaemia, as distinct components of MS, could determine an overestimation of the role of dyslipidaemia associated with insulin resistance. A definition of MS in which the two alterations are considered as one component is plausible. Similarly, elevated systolic and diastolic blood pressure, which are similarly closely related are considered as one, and not two, components of MS. Among conditions associated with the metabolic syndrome, but not included in current ATP-III or
Metabolic syndrome and factor analysis IDF diagnostic criteria, hyperuricaemia deserves some special consideration. In fact, this condition is known to be associated with insulin resistance [18,21] and increased cardiovascular risk [22], and it was included in earlier descriptions of the metabolic syndrome. Our data show that the inclusion of hyperuricaemia among the diagnostic criteria of MS does not interfere with its monofactorial structure. This result suggests that a homogeneous pathophysiological mechanism could underlie hyperuricaemia and the other components of MS, which were included in ATP-III and IDF diagnostic criteria. Using both ATP-III and IDF criteria, combining low HDL-cholesterol and hypertriglyceridaemia, and adding hyperuricaemia as the fifth component, a single factor structure is preserved. Furthermore, the association of MS identified using these modified criteria does not seem to be inferior to that of the syndrome defined with the original sets of diagnostic criteria. On the other hand, on the basis of this cross-sectional analysis, the addition of hyperuricaemia to the definition of MS does not seem to improve the association of the syndrome with other conditions related to insulin resistance. Some limitations of the present study should be recognized. The sample enrolled cannot be considered representative of the general population. In fact, although all residents of the area were invited, only a minority of subjects participated in this screening programme for diabetes; those with risk factors for glucose intolerance were overrepresented, and this accounts for the relatively high prevalence of components of MS in the sample. Another limitation is represented by the fact that insulin measurements were not available, preventing an assessment of insulin sensitivity in this sample. Even though the adequacy of any definition of MS in the evaluation of insulin resistance has been questioned [1], such a measure would have provided additional information on the pathophysiological mechanisms related to factors identified through factor analysis. Furthermore, other parameters of possible interest, such as adiponectin or C-reactive protein, were not available in the present survey; in fact, low adiponectin or elevated C-reactive protein, which are known to be associated with insulin resistance [23] and cardiovascular morbidity [24], deserve to be evaluated for inclusion among criteria for MS. On the other hand, the FIBAR sample can provide a typical picture of subjects undergoing screening for diabetes and glucose intolerance. Furthermore, the over-representation of MS in this sample, in comparison with the general population, can facilitate the statistical analyses performed in the present study.
725 In conclusion, factor analysis of MS, based on the five components identified by ATP-III or IDF, reveals either a one- or two-factor structure, depending on the use of low HDL-cholesterol or hypertriglyceridaemia as two distinct or one composite variable. The combination of the two lipid alterations of MS allows obtaining a single-factor structure, which is maintained if hyperuricaemia is added among the diagnostic components. These alternative, modified definitions of MS do not seem inferior to the original ATP-III and IDF sets of criteria in their association with glucose intolerance and NAFLD, although longitudinal studies are needed to assess the predictive value of different definitions of MS for diabetes and cardiovascular disease. Other conditions, different from hyperuricaemia, could be taken into account for a possible inclusion among diagnostic criteria for MS; however, factor analysis should be considered a valuable aid in the definition of criteria for diagnosis of composite categories, such as the metabolic syndrome. An alternative definition of MS, which includes hyperuricaemia and unifies low HDL-cholesterol and hypertriglyceridaemia as a single composite component, deserves further investigation.
Acknowledgements This work was supported by grants from Menarini Diagnostics International, Florence, Italy, and from the Italian Ministry of University and Scientific Research (PRIN Projects).
References [1] Reaven GM. The metabolic syndrome: requiescat in pace. Clin Chem 2005;51:931e8. [2] Reynolds K, Muntner P, Fonseca V. Metabolic syndrome underrated or underdiagnosed? Diabetes Care 2005;28: 1831e2. [3] Kahn R, Buse J, Ferrannini E, Stern M. The metabolic syndrome: time for a critical appraisal. Joint statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetologia 2005;48: 1684e99. [4] Schachter M. Uric acid and hypertension. Curr Pharm Des 2005;11:4139e43. [5] Bugianesi E, Gastaldelli A, Vanni E, Gambino R, Cassader M, Baldi S, et al. Insulin resistance in non-diabetic patients with non-alcoholic fatty liver disease sites and mechanisms. Diabetologia 2005;48:634e42. [6] Ehrmann DA. Polycystic ovary syndrome. N Engl J Med 2005;352:1223e36. [7] Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood
726
[8] [9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
E. Mannucci et al. Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001;285:2486e97. Alberti KG, Zimmet P, Shaw J. The metabolic syndromeda new worldwide definition. Lancet 2005;366:1059e62. Pladevall M, Singal B, Williams LK, Brotons C, Guyer H, Sadurni J, et al. A single factor underlies the metabolic syndrome A confirmatory factor analysis. Diabetes Care 2006;29:113e22. Wang J, Qiao Q, Miettinen ME, Lappalainen J, Hu G, Tuomilehto J. The metabolic syndrome defined by factor analysis and incident type 2 diabetes in a Chinese population with high postprandial glucose. Diabetes Care 2004; 27:2429e37. Oh JY, Hong YS, Sung YA, Barrett-Connor E. Prevalence and factor analysis of metabolic syndrome in an urban Korean population. Diabetes Care 2004;27:2027e32. Hanson RL, Imperatore G, Bennett PH, Knowler WC. Components of the ‘‘metabolic syndrome’’ and incidence of type 2 diabetes. Diabetes 2002;51:3120e7. Boyko EJ, de Court, Zimmet PZ, Chitson P, Tuomilehto J, Alberti KG. Features of the metabolic syndrome predict higher risk of diabetes and impaired glucose tolerance a prospective study in Mauritius. Diabetes Care 2000;23:1242e8. Mannucci E, Ognibene A, Sposato I, Brogi M, Gallori G, Bardini G, et al. Fasting plasma glucose and glycated haemoglobin in the screening of diabetes and impaired glucose tolerance. Acta Diabetol 2003;40:181e6. Obesity preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser 2000;89:42e53. Chalmers J, MacMahon S, Mancia G, Whitworth J, Beilin L, Hansson L, et al. 1999 World Health Organization-International Society of Hypertension Guidelines for the management of hypertension. Guidelines sub-committee of the World Health Organization. Clin Exp Hypertens 1999;21: 1009e60.
[17] Marchesini G, Avagnina S, Barantani EG, Ciccarone AM, Corica F, Dall’Aglio E, et al. Aminotransferase and gamma-glutamyltranspeptidase levels in obesity are associated with insulin resistance and the metabolic syndrome. J Endocrinol Invest 2005;28:333e9. [18] Clausen JO, Borch-Johnsen K, Ibsen H, Pedersen O. Analysis of the relationship between fasting serum uric acid and the insulin sensitivity index in a population-based sample of 380 young healthy Caucasians. Eur J Endocrinol 1998; 13:863e9. [19] Choudhury J, Sanyal AJ. Insulin resistance in NASH. Front Biosci 2005;10:1520e33. [20] Meigs JB, D’Agostino Sr RB, Wilson PW, Cupples LA, Nathan DM, Singer DE. Risk variable clustering in the insulin resistance syndrome. The Framingham Offspring Study. Diabetes 1997;46:1594e600. [21] Tsunoda S, Kamide K, Minami J, Kawano Y. Decreases in serum uric acid by amelioration of insulin resistance in overweight hypertensive patients effect of a low-energy diet and an insulin-sensitizing agent. Am J Hypertens 2002;15: 697e701. [22] Conen D, Wietlisbach V, Bovet P, ShamLaye C, Riesen W, Paccaud F, et al. Prevalence of hyperuricemia and relation of serum uric acid with cardiovascular risk factors in a developing country. BMC Public Health 2004;49: 155e7. [23] Yuan G, Zhou L, Tang J, Yang Y, Gu W, Li F, et al. Serum CRP levels are equally elevated in newly diagnosed type 2 diabetes and impaired glucose tolerance and related to adiponectin levels and insulin sensitivity. Diabetes Res Clin Pract 2005;72:244e50. [24] Sattar N, Gaw A, Scherbakova O, Ford I, O’Reilly DS, Haffner SM, et al. Metabolic syndrome with and without C-reactive protein as a predictor of coronary heart disease and diabetes in the West of Scotland Coronary Prevention Study. Circulation 2003;108:414e9.
Available online at www.sciencedirect.com