Factors of the Metabolic Syndrome

Baseline Interrelationships in the First Follow-up Cohort of the HDDRISC Study (HDDRISC-1)

From the Wynn Department of Metabolic Medicine, Imperial College School of Medicine, London, UK.

Correspondence to Ian F. Godsland, PhD, Wynn Department of Metabolic Medicine, Imperial College School of Medicine, 21 Wellington Rd, London NW8 9SQ. E-mail i.godsland{at}ic.ac.uk

	Abstract

Top Abstract Introduction Methods Results Discussion References

 
Abstract—Syndromes of risk factor disturbance may contribute to the development of coronary heart disease and non–insulin-dependent diabetes mellitus, but their definition and quantification remain problematic. Using factor analysis, constellations of risk factor variables that could indicate distinct syndromes of metabolic disturbance were explored in the baseline data of the first follow-up cohort of 742 men from the Heart Disease and Diabetes Risk Indicators in a Screened Cohort (HDDRISC) study. The primary analysis considered 16 intercorrelated variables measured in more than 90% of cohort participants. A missing-values estimation routine was used to ensure inclusion of all participants in the analysis. Subanalyses were undertaken, including a repeat of the primary analysis on the 522 individuals who had received measurement of HDL cholesterol, an oblique rather than orthogonal factor rotation procedure performed on primary and HDL subset analyses, a repeat of these two primary and HDL subset analyses using only those participants with complete measurements, and a repeat of these six analyses including only the seven variables conventionally associated with the metabolic syndrome. The principal factor that emerged in all analyses undertaken comprised oral glucose tolerance test insulin and glucose response, serum uric acid, and body mass index. Fasting serum triglyceride concentration was included in this factor in 11 of the 12 analyses undertaken, fasting plasma insulin in 8, fasting plasma glucose in 5, and mean arterial pressure in 3. HDL cholesterol factored in isolation from insulin in all analyses undertaken. These findings provide strong support for a core metabolic cluster, which is unlikely to include blood pressure and does not include HDL. The factor scores relating to this cluster will provide a means of assessing its quantitative importance in prospective analysis of the development of CHD and diabetes in this cohort.

Key Words: factor analysis • metabolic syndrome • cohort study • insulin • high density lipoprotein

	Introduction

Top Abstract Introduction Methods Results Discussion References

 
Coronary heart disease and diabetes mellitus share many risk factors in common,^{1 2} and disturbances in these risk factors are themselves highly intercorrelated.^{3 4 5} It has been proposed that such intercorrelations indicate a distinct metabolic syndrome in which insulin resistance may have a central role.^{6 7 8} Such a syndrome could include impaired glucose tolerance, hyperinsulinemia, high blood pressure, hypertriglyceridemia and low HDL concentrations, increased levels of the antifibrinolytic factor plasminogen activator inhibitor-1, increased levels of the atherogenic small, dense LDL, hyperuricemia, postprandial hyperlipidemia, and central obesity.⁹ Mechanistic links can be proposed to explain these relationships and support the existence of a distinct metabolic syndrome.

As yet, there appears to have been little attempt to quantify the metabolic syndrome or its clinical significance or even to determine to what extent it can be distinguished statistically as an independent entity.¹⁰ Traditionally, epidemiological analysis of risk factors for CHD and diabetes has been designed to isolate single independent risk factors from a series of candidate predictors for the disease in question. However, in the light of the manifold interrelationships that exist between potential risk factors,¹¹ it may be of value to identify and quantify distinct clusters of metabolic disturbance and then determine whether the measure thus derived better relates to subsequent disease development. One technique that can help achieve this goal is factor analysis.^{12 13} In the present analysis, we have applied this technique to baseline information gathered in the course of a prospective study of risk factors for CHD and diabetes mellitus. Within a panel of potential risk factors, interrelationships are investigated that might indicate the existence of a distinct metabolic syndrome. Factor analysis is then used to determine whether a factor can be distinguished that can represent such a syndrome. A variety of different approaches may be taken in factor analysis, both with regard to the way in which the data set is selected and organized and in the type of factor analysis employed, but there has been little systematic exploration of how the approach taken might affect the results obtained. In the present study, a series of subanalyses has been undertaken to determine how consistently a given factor appears in different analyses.

	Methods

Top Abstract Introduction Methods Results Discussion References

 
Design
The HDDRISC study is a prospective study of metabolic risk factors for the development of CHD and diabetes mellitus. The study derives from a company health program begun in 1971, in which participants receive a range of metabolic, clinical, and laboratory measurements at their first visit and, for those continuing in the program, at subsequent visits spaced at 2- or 3-year intervals. As the program has developed, further measurements have been included, leading to the delineation of distinct follow-up cohorts, defined by the range of measurements made at each participant's first visit. The first follow-up cohort (HDDRISC-1) comprised men seen between June 1971 and July 1988, the majority of whom (>90%) received an OGTT with measurement of plasma glucose and insulin concentrations at their first visit. Subsequent cohorts are being defined by inclusion of insulin resistance, HDL subfraction, body composition, or hemostatic factor measurements. A proportion of those included in the HDDRISC-1 study (70.4%) had HDL cholesterol measured, and these comprise a subset of the HDDRISC-1 study. The study received local ethics committee approval, and each participant gave his written informed consent.

Subjects
The HDDRISC-1 study group comprised 742 senior staff of a company with extensive transport and shipping interests. All are white males. The majority are land-based business executives, but the cohort also includes some senior seagoing staff (captain or chief engineer, n=120). At their first visit, men were aged 26.1 to 70.5 years, with BMI ranging from 18.9 to 46.9 kg/(m)². Fourteen participants (1.9%) were diagnosed as having CHD, as defined by previous documentation of a myocardial infarction or angina, coronary artery bypass surgery, coronary angiography showing evidence of CHD, presence of electrocardiographic findings consistent with a myocardial infarction (WHO criteria), or treatment with antianginal medication. Twenty-eight participants (3.8%) were diagnosed as having diabetes mellitus, as defined by a previous diagnosis of diabetes mellitus, 120-minute OGTT plasma glucose concentration >11.1 mmol/l, or treatment with hypoglycemic medication.

Procedures
Studies were carried out in our metabolic day ward. Participants were asked to consume more than 200 g/d carbohydrate in their diet for 3 days before their visit, to have fasted for 12 hours, and to have refrained from smoking on the morning of the test. Height and weight were measured on arrival at 9 AM. A clinician obtained a general medical history, including details of exercise habits and alcohol and tobacco consumption. After the subject had rested for 15 minutes in a semirecumbent position, systolic and diastolic blood pressures were measured by a cuff method with a mercury sphygmomanometer. First- and fifth-phase Korotkoff sounds were recorded. A cannula was inserted into an antecubital vein in one arm for sampling, the arm having been previously rested on a heating pad to assist blood flow. Blood samples were taken for a full blood count, measurement of hepatic, renal, and thyroid function tests, fasting plasma glucose and insulin concentrations, and serum lipid and lipoprotein concentrations. A further sample was taken for a second measurement of fasting plasma glucose and insulin concentrations. Each participant then underwent an OGTT. Glucose was given as 50% dextrose solution as a chilled, flavored drink, at a dose of 1 g/kg body weight, or 40 g/m² for those over 120% of their ideal body weight. Sampling for plasma glucose and insulin was at 30, 60, 90, 120, 150, and 180 minutes after ingestion of the glucose solution, and the participant remained semirecumbent throughout the procedure. Before leaving the day ward, a 12-lead resting ECG was obtained.

Laboratory Determinations
Routine hematological and biochemical parameters were measured by standard laboratory methods. Glucose, insulin, lipid, and lipoprotein measurements were undertaken in our own laboratory. Plasma glucose was determined on the same day using glucose oxidase procedures; until 1977, o-tolidine was used as chromogen,¹⁴ and thereafter 4-aminophenazone was used.¹⁵ Plasma insulin concentrations were measured on samples stored at -20°C by using the radioimmunoassay procedure of Albano et al.¹⁶ Until 1976, serum total cholesterol was measured by a Lieberman Burchard technique¹⁷ and triglycerides were measured by the procedure of Cramp and Robertson.¹⁸ Subsequently, fully enzymatic assays were employed.^{19 20} Concentrations of HDL were measured after separation by preparative ultracentrifugation until 1984 and by sequential precipitation with heparin/manganese ions thereafter.²¹ Whenever there was a change in methodology, extensive comparisons between methods were undertaken, and annual cohort means were inspected for any discontinuity. The only significant change found was in glucose values on adoption of the method of Trinder.¹⁸ This discrepancy necessitated addition of 0.42 mmol/L to all values before this change to ensure comparability of glucose measures among all cohort members. Within- and between-batch precision was monitored throughout the study by using frozen plasma and serum pools and commercially available lyophilized sera and by participation in national quality assurance schemes.

Data Analyses
Members of the HDDRISC-1 study with preexisting CHD or diabetes mellitus were excluded from the present analysis because our concern was primarily with predictors of disease that would develop in individuals who were healthy at baseline. Risk factor interrelationships between 17 potential risk factor variables in the first-visit data are considered here (Table 1). These were selected from the full range of measurements made, on the basis of previous evidence for their involvement in the development of CHD or diabetes mellitus or because they differed significantly between those who remained healthy and those who on follow-up developed CHD or diabetes mellitus (results of the 11.3-year clinical follow-up of the HDDRISC-1 study are to be the subject of separate reports, as are the influences of lifestyle characteristics, such as cigarette smoking, alcohol consumption, and physical activity).

Statistical analyses were carried out using the BMDP statistical package (BMDP Statistical Software Inc). Fasting plasma concentrations of glucose and insulin were taken as the mean of the two pretest samples. Incremental areas under the OGTT concentration profiles were calculated according to the formula: Incremental Area=[(MF/2)+C30+C60+C90+C120+C150 +(C180/2)]-[6xMF]

where MF is the mean fasting glucose or insulin concentration and C is the plasma concentration of glucose or insulin at the specified minutes during the OGTT. In the derivation of mean values, insulin measures and triglyceride concentrations were logarithmically transformed. For correlation and factor analysis, data were adjusted to the mean age of the cohort by using the univariate regression coefficient with age for each variable. Univariate Pearson correlation coefficients between variables were derived. In light of the number of correlations explored, univariate correlation significance values between .01 and .05 were considered borderline. Significant clustering of correlated metabolic disturbances was identified by factor analysis.

Factor Analysis
Factor analysis assumes that intercorrelations between observed variables are influenced by a smaller number of hypothetical underlying variables, termed factors, and allows derivation of a score that not only reflects the magnitude of the values of the individual measures that comprise the factor but also reflects the strength of their interrelationships.¹² Factors are characterized by the variables that, according to their so-called factor loadings, most strongly correlate with the factor concerned. If a factor explained all the variance in the data, it would be the only factor extracted, and each of the variables considered in the analysis would load on it with a value of 1. In practice, each factor identified accounts for only a fraction of the total variance (the percentage of the variance explained). Factor analysis consists of two main steps: factor extraction and factor rotation. The former step comprises the identification of linear combinations of variables that account for the maximum proportion of the total variance in the full set of variables. The latter step improves the discrimination between factors and provides values for the factor scores in each case. Various methods are available for performing each step. In the present analysis, methods were applied that would further the achievement of the simplest and most discriminatory identification of factors. Principal-components analysis was used to extract the initial factors. Principal components analysis identifies uncorrelated components that parsimoniously summarize the total variation in the data space, each component being a linear combination of the observed variables. In factor analysis, this relationship is reformulated such that each observed variable is made a linear combination of the components. Only components with eigenvalues (the sum of the squared factor loadings, representing the variance attributable to each principal component) >1 are considered. In the primary analysis that we undertook, factor rotation was achieved by using the varimax method of rotation. The correlation coefficient between the factors identified by the varimax method are predetermined to be zero, which further simplifies factor interpretation. Factor interpretation is usually made with reference to factors that load above a certain level. In the present analysis, we have adopted the relatively inclusive level of equal to or greater than 0.25, since it was found that certain variables (particularly serum triglycerides) would load with the principal factor at a relatively low level but highly consistently in a variety of different analyses. A single measure of mean arterial pressure ([(2xdiastolic)+systolic]/3) was employed in place of separate measures of diastolic and systolic blood pressure in the factor analysis. This procedure avoided the emergence of a separate factor consisting solely of the two highly correlated blood pressure measures (diastolic and systolic blood pressure and mean arterial pressure behaved, in any case, very similarly in univariate analysis and when entered as alternatives in the factor analysis).

In the present analysis, a series of 14 subanalyses was undertaken. In the primary investigation, factors were derived from the 17 potential risk factors that had been selected on the basis of their known involvement in CHD or diabetes or their showing baseline differences in the HDDRISC-1 study group between those who were identified in prospective study (data not shown) as subsequently developing CHD or diabetes and those who remained free of these diseases. HDL cholesterol was excluded from this primary investigation so that each variable included in the analysis would be nonmissing in more than 90% of cases. To ensure that the complete data set was employed in this primary analysis, missing values were estimated (BMDP program AM was employed, which provides an estimate of missing values based on coefficients from multiple linear regression prediction of the variable concerned from all other concurrently measured variables). Varimax factor rotation was used, and participants with CHD or diabetes were excluded. This analysis was then repeated in the subset of 522 participants who had received measurement of HDL cholesterol. In analysis, HDL cholesterol was included in the variable list. Both analyses were then repeated using quartimin rather than varimax factor rotation. This oblique factor rotation procedure allows for the identification of factors that may be intercorrelated. Both analyses were again repeated with varimax factor rotation, but without estimation of missing values (596 participants in the full data set analysis and 444 in the subset with HDL measured had complete data recorded). These six analyses were repeated, but with only the seven variables conventionally associated with the metabolic syndrome being considered. These variables included BMI; blood pressure; uric acid, triglyceride, and mean fasting insulin and glucose concentrations; OGTT incremental insulin and glucose areas; and, in the subset in which it was measured, HDL cholesterol. In a final analysis, the primary analysis, with estimation of missing values and varimax rotation, was repeated, with all cases and the subset with HDL cholesterol measured, but with inclusion rather than exclusion of those who entered the cohort with diagnosed diabetes or vascular disease.

	Results

Top Abstract Introduction Methods Results Discussion References

 
Of the 701 participants free of CHD or diabetes mellitus at first visit, 187 (26.7%) exceeded a BMI of 27 kg/(m)², 27 (3.8%) had a systolic blood pressure >160 mm Hg, and 73 (10.4%) a diastolic blood pressure >90 mm Hg. Five (0.7%) were taking antihypertensive medication (three beta-blockers and two other antihypertensive agents), 2 (0.2%) were taking systemic steroids, 9 (1.2%) were taking antihyperuricemic medication (allopurinol), and 1 (0.1%) was taking fibrate hypolipidemic medication. Analysis with and without participants taking these drugs gave essentially the same results, and data for the entire group of 701 are reported below. Of these 701, 522 (74.5%) had HDL cholesterol measurements recorded.

Table 1 summarizes means and ranges for all variables in the whole group. Means and ranges in the HDL subset were indistinguishable from those in the entire group. The mean (range) HDL cholesterol concentration was 1.30 (0.36 to 2.34) mmol/L. Significant correlations with age were found for BMI (r=.20, P<.001), systolic (.31, P<.001) and diastolic (.29, P<.001) blood pressures, erythrocyte sedimentation rate (.22, P<.001), total cholesterol (.25, P<.001), triglycerides (.16, P<.001), mean fasting glucose (.12, P<.001), incremental glucose area (.26, P<.001), and incremental insulin area (.10, P<.01).

Univariate correlation coefficients between the 17 potential age-standardized risk factors are shown in Table 2. Extensive intercorrelations were apparent, the majority significant at the P<.001 level. For example, mean arterial pressure, serum triglyceride concentration, and OGTT incremental glucose area were each significantly, or borderline significantly, correlated with 13 other variables. The variable that showed the least number of significant correlations was plasma sodium concentration, which was significantly correlated with four other variables. Similar intercorrelations to those in the group as a whole were found for the subset on whom HDL cholesterol concentration had been measured (data not shown). HDL cholesterol showed significant, or borderline significant, correlations with all variables except sodium, potassium, uric acid, globulin, total cholesterol, and fasting plasma glucose concentrations (Table 2).

Factor analysis of all potential predictors of CHD and diabetes (excluding HDL cholesterol), using all cases, with substitution of missing values and varimax factor rotation, identified a principal factor (ie, the factor explaining the greatest proportion of the variance) comprising OGTT glucose and insulin response, mean fasting insulin, uric acid, triglycerides, and BMI (Table 3). This factor will subsequently be referred to as the core metabolic cluster. Other factors represented correlated variation in blood proteins (factor 2), lipemia (factor 3), blood pressure and sodium (factor 4), and erythrocyte sedimentation rate and hemoglobin (factor 5). An almost identical pattern of factors was apparent when the subset of participants with HDL measurements was analyzed (Table 3). The principal factor again comprised the core metabolic cluster, with the additional inclusion of mean fasting glucose. HDL cholesterol comprised a separate factor (factor 6) When quartimin factor rotation was employed on the full data set, the principal factor comprised the core metabolic cluster, but without triglycerides, and when the HDL subset was analyzed, the principal factor again comprised the core metabolic cluster, but with mean fasting glucose included (Table 4). When the analysis was undertaken with only those cases with complete data, the principal factor was again the core metabolic cluster (results not shown). This factor was obtained when both the full data set and the HDL subset were analyzed, although mean fasting glucose was again an additional component of the principal factor when the HDL subset was analyzed.

Factor analysis of only those variables proposed to be components of the metabolic syndrome identified the core metabolic cluster as the principal factor in every analysis undertaken (Table 5). Analysis of the full data set (excluding HDL cholesterol) resulted in exclusion of mean fasting insulin from the core metabolic cluster, and this was also the case when quartimin rotation was used and when only those cases with complete data were considered. An additional component of the principal factor when cases with complete data were considered was mean arterial pressure (results not shown). Analysis of the subset with HDL cholesterol measured resulted in inclusion of mean arterial pressure and mean fasting glucose as components of the principal factor when either varimax or quartimin rotation was employed (Table 5) and exclusion of mean fasting insulin when only cases with complete data were considered (results not shown). In the analyses of putative components of the metabolic syndrome, a second factor (factor 2) was identified in each analysis of the full data set and in the analysis of only those participants with complete data when the subset with HDL cholesterol measured was analyzed. This second factor comprised mean fasting glucose and insulin and BMI, with variable inclusion of mean arterial pressure and OGTT glucose and insulin response. In the analysis of the HDL subset, HDL consistently comprised a separate factor, which also included triglycerides. Inclusion of those with diabetes or vascular disease made no difference in the factors obtained and very little difference in the factor loadings (results not shown).

	Discussion

Top Abstract Introduction Methods Results Discussion References

 
In this study, we have used factor analysis to evaluate the grouping of metabolic risk factors for cardiovascular disease in 701 men free of diabetes mellitus and heart disease. In a series of such analyses, the principal cluster that emerged consistently included four variables, each of which has previously been included among the proposed components of the metabolic syndrome: OGTT insulin and glucose response, serum uric acid, and BMI. Fasting serum triglycerides were included in this cluster in 11 of the 12 analyses undertaken, fasting plasma insulin in 8, fasting plasma glucose in 5, and mean arterial pressure in 3. HDL cholesterol factored in isolation from insulin in all analyses undertaken. This method justifies the proposed existence of a distinct metabolic syndrome, comprising the core cluster that was identified in the primary analysis described previously: OGTT insulin and glucose response, fasting plasma insulin, serum uric acid, and triglyceride concentrations, and BMI. Mean fasting glucose and blood pressure did not cluster with these variables sufficiently consistently to conclude that they too can be included in the syndrome. Various mechanisms may contribute to these associations, but one common link that could be responsible might reside in variation in the activity of the glycolytic pathway, specifically at the level of GAPDH. A reduction in the activity of this enzyme would diminish flux through the glycolytic pathway, thus causing insulin resistance and hyperinsulinemia. Back-up of metabolic intermediates before GAPDH would then be expected to result in accumulation of glycerol-3-phosphate, which would promote triglyceride synthesis,²² and also ribose-5-phosphate, phosphoribosyl pyrophosphate, and further intermediates that contribute to uric acid synthesis.

OGTT insulin response was a prominent feature of the core metabolic cluster. In subjects with normal glucose tolerance (89% of participants in the present study) this can act as a surrogate measure of insulin resistance,²³ supporting the possibility that insulin resistance is an important factor in the metabolic clustering we observed. However, the core metabolic cluster identified in the present analysis differs from the panel of putative components of the classic metabolic or insulin resistance syndrome²⁴ in that two of the major components, blood pressure and HDL cholesterol, were not included, despite showing significant univariate correlations with other proposed components of the syndrome. This finding emphasizes the potential importance of the common link described above in the metabolic clustering we observed. In the analyses employing all potential predictors, blood pressure was the highest loading component of a factor that was entirely separate from the metabolic syndrome cluster. When included in the analyses involving only putative components of the metabolic syndrome, blood pressure showed no consistent pattern of loading. There has been some uncertainty over the extent to which blood pressure should be included in the metabolic syndrome, in accord with its equivocal behavior in the present analysis.²⁵ The isolated factor in which HDL cholesterol was the major component generally included triglycerides, in inverse relationship. This observation probably reflects the activity of the various lipase enzymes that dictate the transfer of components from triglyceride-rich lipoproteins to HDL²⁶ and appear to dominate metabolic intercorrelations with HDL. A relatively weak univariate correlation between insulin-related variables and HDL cholesterol was also seen in an analysis of data acquired more recently during the course of the HDDRISC study.³ However, this data also included measurement of HDL₂ subfraction cholesterol, which showed a stronger correlation with insulin-related variables than did total HDL cholesterol. It is therefore possible that factor associations between HDL- and insulin-related measures might have been apparent in the HDDRISC-1 participants had HDL₂ subfraction measurements been available.

An important feature of the present analysis was the robustness of the principal factor in the variety of analyses that were undertaken. This was not the case for subsidiary factors, which could vary according to the variables included in the analysis, the completeness of the data set, and the type of factor analysis undertaken. Factors explaining lesser proportions of the variance are clearly less well determined, which should be borne in mind in conclusions based on factor analyses of the metabolic syndrome. Three such analyses have been published previously^{13 27 28} (two in abstract form^{27 28} ), and although they have provided general support for metabolic clustering, they have differed appreciably between themselves and with the present analysis. The cohort we examined was all male and contained a high proportion of participants who were moderately overweight. It would be of interest to compare factors of the metabolic syndrome between different study populations to establish whether and how clustering may vary. However, such comparisons would have to be undertaken with standardized panels of variables and factor analysis procedures.

This is the first analysis we are aware of in which a metabolic syndrome cluster has been distinguished from a background of intercorrelated risk factors, not all of which are proposed components of the syndrome. Of particular interest is the provision of a continuous variable score with which the intensity of the syndrome can be quantified. Such a score could provide the basis for a quantitative definition of the syndrome and, as Edwards and colleagues have pointed out,¹³ such a score may also prove useful in the prediction of clinical outcomes in prospective studies. Preliminary analysis suggests that this may indeed be the case with regard to diabetes in the HDDRISC-1 study group.²⁹ Thus, factor analysis might allow for a more discriminatory evaluation of metabolic antecedents of vascular disease.

	Selected Abbreviations and Acronyms

 

BMI = body mass index CHD = coronary heart disease OGTT = oral glucose tolerance test HDDRISC = heart disease and diabetes risk indicators in a screened cohort

	Acknowledgments

 
Financial support for this study was provided by the Heart Disease and Diabetes Research Trust, the Cecil Rosen Foundation, and the Ederman Trust. The study was initiated and established by Prof Victor Wynn. We also gratefully acknowledge the contribution of the many clinical, nursing, and laboratory staff who have contributed to this project since it began in 1971.

Received September 17, 1997; accepted September 26, 1997.

	References

Top Abstract Introduction Methods Results Discussion References

 
1. Jarrett RJ, Shipley MJ. Type 2 (non–insulin-dependent) diabetes and cardiovascular disease: putative associations via common antecedents; further evidence from the Whitehall study. Diabetologia. 1988;31:737–740.[Medline] [Order article via Infotrieve]

2. Stern MP. Diabetes and cardiovascular disease: the "common soil" hypothesis. Diabetes. 1995;44:369–374.[Abstract]

3. Godsland IF, Crook D, Walton C, Wynn V, Oliver MF. Influence of insulin resistance, secretion, and clearance on serum cholesterol, triglycerides, lipoprotein cholesterol, and blood pressure in healthy men. Arterioscler Thromb. 1992;12:1030–1035.[Abstract/Free Full Text]

4. Walton C, Lees B, Crook D, Worthington M, Godsland IF, Stevenson JC. Body fat distribution rather than overall adiposity influences serum lipids and lipoproteins in healthy men independently of age. Am J Med. 1995;99:459–464.[Medline] [Order article via Infotrieve]

5. Walton C, Lees B, Crook D, Worthington M, Godsland IF, Stevenson JC. Relationships between insulin metabolism, serum lipid profile, body fat distribution and blood pressure in healthy men. Atherosclerosis. 1995;118:35–43.[Medline] [Order article via Infotrieve]

6. Modan M, Halkin H, Almog S, Lusky A, Eshkol A, Shefi M, Shitrit A, Fuchs S. Hyperinsulinaemia: a link between hypertension, obesity and glucose intolerance. J Clin Invest. 1985;75:809–817.

7. Reaven G. Banting lecture 1988: role of insulin resistance in human disease. Diabetes. 1988;37:1595–1607.[Abstract]

8. Kaplan NM. The deadly quartet: upper-body obesity, glucose intolerance, hypertriglyceridemia, and hypertension. Arch Intern Med. 1989;149:1514–1520.[Abstract/Free Full Text]

9. Després J-P, Marette A. Relation of components of insulin resistance syndrome to coronary disease risk. Curr Opin Lipidol. 1994;5:274–289.[Medline] [Order article via Infotrieve]

10. Godsland IF, Stevenson JC. Insulin resistance: syndrome or tendency? Lancet. 1995;346:100–103.[Medline] [Order article via Infotrieve]

11. Davey Smith G, Phillips A. Declaring independence: why we should be cautious? J Epidemiol Community Health. 1990;44:257–258.[Free Full Text]

12. Stevens J. Applied Multivariate Statistics for the Social Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates; 1986:337–350.

13. Edwards KL, Austin MA, Newman B, Mayer E, Krauss RM, Selby JV. Multivariate analysis of the insulin resistance syndrome in women. Arterioscler Thromb. 1994;14:1940–1945.[Abstract/Free Full Text]

14. Cramp DG. New automated method for measuring glucose by glucose oxidase. J Clin Pathol. 1967;20:910–912.[Abstract/Free Full Text]

15. Trinder P. Determination of blood glucose using an oxidase-peroxidase system with non-carcinogenic chromogen. J Clin Pathol. 1969;22:158–161.[Abstract/Free Full Text]

16. Albano JDM, Ekins RP, Maritz G, Turner RC. A sensitive, precise radioimmunoassay of serum insulin relying on charcoal separation of bound and free hormone moieties. Acta Endocrinol. 1972;70:487–509.

17. Robertson G, Cramp DG. An evaluation of cholesterol determinations in serum and serum lipoprotein fractions by a semiautomated fluorimetric method. J Clin Pathol. 1970;23:243–245.[Abstract/Free Full Text]

18. Cramp DG, Robertson G. The fluorimetric assay of triglyceride by a semiautomated method. Anal Biochem. 1968;25:246–251.[Medline] [Order article via Infotrieve]

19. Bucolo G, David H. Quantitative determination of serum triglycerides by the use of enzymes. Clin Chem. 1973;19:476–482.[Abstract]

20. Seidel J, Hagele E, Ziagenhorn J, Wahlefield A. Reagent for the enzymatic determination of serum total cholesterol with improved lipolytic efficiency. Clin Chem. 1983;29:1075–1080.[Abstract/Free Full Text]

21. Warnick GR, Albers JJ. A comprehensive evaluation of the heparin-manganese precipitation procedure for estimating high density lipoprotein cholesterol. J Lipid Res. 1978;19:65–76.[Abstract]

22. Newsholme EA, Start C. Regulation in Metabolism. New York: John Wiley & Sons, Inc; 1980:218.

23. Laakso M. How good a marker is insulin level for insulin resistance? Am J Epidemiol.. 1993;137:959–965.[Abstract/Free Full Text]

24. Reaven GM. Role of insulin resistance in human disease (Syndrome X): an expanded definition. Annu Rev Med. 1993;44:121–131.[Medline] [Order article via Infotrieve]

25. Haffner SM. Insulin and blood pressure: fact or fantasy? J Clin Endocrinol Mtab.. 1993;76:541–542.[Medline] [Order article via Infotrieve]

26. Patsch JR, Gotto AMJ, Olivecrona T, Eisenberg S. Formation of high density lipoprotein 2–like particles during lipolysis of very low density lipoproteins in vitro. Proc Natl Acad Sci U S A. 1978;75:4519–4523.[Abstract/Free Full Text]

27. Wingard D, von Muhlen D, Barrett-Connor E, Kritz-Silverstein D. Factor analysis of proposed components of the insulin resistance syndrome. Diabetes. 1996;45(suppl 2):137A. Abstract.

28. Meigs JB, D'Agostino JB, Cupples LA, Wilson PWF, Nathan DM, Singer DE. Evidence that insulin resistance does not unify metabolic cardiovascular disease risk variable clustering: the Framingham Offspring Study. Diabetes. 1996;45(suppl 2):137A. Abstract.

29. Godsland IF, Leyva F, Bruce R, Walton C, Worthington M, Stevenson JC. The metabolic syndrome predicts development of diabetes in the first follow-up cohort of the HDDRISC study (HDDRISC-1): an application of factor analysis. Diabet Med. 1997;14(suppl 1):518. Abstract.

This article has been cited by other articles:

				L. G. Sanchez-Lozada, E. Tapia, P. Bautista-Garcia, V. Soto, C. Avila-Casado, I. P. Vega-Campos, T. Nakagawa, L. Zhao, M. Franco, and R. J. Johnson Effects of febuxostat on metabolic and renal alterations in rats with fructose-induced metabolic syndrome Am J Physiol Renal Physiol, April 1, 2008; 294(4): F710 - F718. [Abstract] [Full Text] [PDF]

				R. O. Halperin, H. D. Sesso, J. Ma, J. E. Buring, M. J. Stampfer, and J. Michael Gaziano Dyslipidemia and the Risk of Incident Hypertension in Men Hypertension, January 1, 2006; 47(1): 45 - 50. [Abstract] [Full Text] [PDF]

				E. Goodman, L. M. Dolan, J. A. Morrison, and S. R. Daniels Factor Analysis of Clustered Cardiovascular Risks in Adolescence: Obesity Is the Predominant Correlate of Risk Among Youth Circulation, April 19, 2005; 111(15): 1970 - 1977. [Abstract] [Full Text] [PDF]

				I. F. Godsland, D. Crook, A. J. Proudler, and J. C. Stevenson Hemostatic Risk Factors and Insulin Sensitivity, Regional Body Fat Distribution, and the Metabolic Syndrome J. Clin. Endocrinol. Metab., January 1, 2005; 90(1): 190 - 197. [Abstract] [Full Text] [PDF]

				U. Salmenniemi, E. Ruotsalainen, J. Pihlajamaki, I. Vauhkonen, S. Kainulainen, K. Punnonen, E. Vanninen, and M. Laakso Multiple Abnormalities in Glucose and Energy Metabolism and Coordinated Changes in Levels of Adiponectin, Cytokines, and Adhesion Molecules in Subjects With Metabolic Syndrome Circulation, December 21, 2004; 110(25): 3842 - 3848. [Abstract] [Full Text] [PDF]

				D. A. Lawlor, S. Ebrahim, M. May, and G. Davey Smith (Mis)use of Factor Analysis in the Study of Insulin Resistance Syndrome Am. J. Epidemiol., June 1, 2004; 159(11): 1013 - 1018. [Abstract] [Full Text] [PDF]

				J. M. McCaffery, R. Niaura, J. F. Todaro, G. E. Swan, and D. Carmelli Depressive Symptoms and Metabolic Risk in Adult Male Twins Enrolled in the National Heart, Lung, and Blood Institute Twin Study Psychosom Med, May 1, 2003; 65(3): 490 - 497. [Abstract] [Full Text] [PDF]

				B.-J. Shen, J. F. Todaro, R. Niaura, J. M. McCaffery, J. Zhang, A. Spiro III, and K. D. Ward Are Metabolic Risk Factors One Unified Syndrome? Modeling the Structure of the Metabolic Syndrome X Am. J. Epidemiol., April 15, 2003; 157(8): 701 - 711. [Abstract] [Full Text] [PDF]

				R. L. Hanson, G. Imperatore, P. H. Bennett, and W. C. Knowler Components of the "Metabolic Syndrome" and Incidence of Type 2 Diabetes Diabetes, October 1, 2002; 51(10): 3120 - 3127. [Abstract] [Full Text] [PDF]

				V.-P. Valkonen, M. Kolehmainen, H.-M. Lakka, and J. T Salonen Insulin resistance syndrome revisited: application of self-organizing maps Int. J. Epidemiol., August 1, 2002; 31(4): 864 - 871. [Abstract] [Full Text] [PDF]

				I. F. Godsland, J. R. Rosankiewicz, A. J. Proudler, and D. G. Johnston Plasma Total Homocysteine Concentrations Are Unrelated to Insulin Sensitivity and Components of the Metabolic Syndrome in Healthy Men J. Clin. Endocrinol. Metab., February 1, 2001; 86(2): 719 - 723. [Abstract] [Full Text]

				P. A. Sakkinen, P. Wahl, M. Cushman, M. R. Lewis, and R. P. Tracy Clustering of Procoagulation, Inflammation, and Fibrinolysis Variables with Metabolic Factors in Insulin Resistance Syndrome Am. J. Epidemiol., November 15, 2000; 152(10): 897 - 907. [Abstract] [Full Text] [PDF]

				J. B. Meigs Invited Commentary: Insulin Resistance Syndrome? Syndrome X? Multiple Metabolic Syndrome? A Syndrome At All? Factor Analysis Reveals Patterns in the Fabric of Correlated Metabolic Risk Factors Am. J. Epidemiol., November 15, 2000; 152(10): 908 - 911. [Abstract] [Full Text] [PDF]

				M. Pyorala, H. Miettinen, P. Halonen, M. Laakso, and K. Pyorala Insulin Resistance Syndrome Predicts the Risk of Coronary Heart Disease and Stroke in Healthy Middle-Aged Men : The 22-Year Follow-Up Results of the Helsinki Policemen Study Arterioscler Thromb Vasc Biol, February 1, 2000; 20(2): 538 - 544. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Leyva, F. Articles by Stevenson, J. C. Search for Related Content PubMed PubMed Citation Articles by Leyva, F. Articles by Stevenson, J. C.