Psychiatry Research 215 (2014) 438–443
Contents lists available at ScienceDirect
Psychiatry Research journal homepage: www.elsevier.com/locate/psychres
Alexithymia as a risk factor for type 2 diabetes mellitus in the metabolic syndrome: a cross-sectional study Alexandra V. Lemche a, Oleg S. Chaban b, Erwin Lemche c,n a
Department of Psychosomatic Medicine, Charité University Medicine Berlin, Germany Section of Neuroses and Somatoform Disorders, Bogolomets National Medical University, Kiev, Ukraine c Section of Cognitive Neuropsychiatry, Institute of Psychiatry, King's College London, UK b
art ic l e i nf o
a b s t r a c t
Article history: Received 16 June 2012 Received in revised form 13 August 2013 Accepted 4 December 2013 Available online 14 December 2013
Alexithymia is a clinical trait consisting of diminished introspective and interoceptive capacities that has been shown to implicate elevated autonomic outflow and to bias for hypertension. To estimate relative risk associated with alexithymia in the metabolic syndrome (MetS), we conducted a cross-sectional analysis of patients with manifest type 2 diabetes mellitus (T2DM) or familial diabetes risk (N¼ 101; 67 females; age 45.6 713.96) in a nationwide sampled treatment cohort for MetS in the Ukrainian governmental health care system. Laboratory data of single components of the MetS according to International Diabetes Federation Consensus were dependent measures in multivariable regression models with self-reported alexithymia severity (TAS-20) and socio-demographic data. TAS-20 as the sole surviving psychometric predictor for T2DM in the simplest regression equation provided the best model fit: OR 1.073, Z ¼19.04, (95%CIs 1.065–1.081). For microalbuminuria, the best fitting model was OR 1.030, Z¼ 3.49 (95%CIs 1.013–1.048). TAS-20 predicted also triglyceride level at Wald-χ2 ¼1299.27, Z¼36.05 (95%CIs 0.052–0.058) and blood pressure maximum at Wald-χ2 ¼2309.05, Z¼48.05 (95%CIs 2.402– 2.606). Our results show that alexithymia severity contributes to MetS by covarying with several of its single components, and that it may be a substantial concurrent indicator of T2DM and cardiovascular risks in MetS. & 2013 Elsevier Ireland Ltd. All rights reserved.
Keywords: Metabolic syndrome Type 2 diabetes mellitus Cardiovascular disease Toronto Alexithymia Scale Relative risk Cross-sectional studies National cohort
1. Introduction Metabolic syndrome (MetS) is a constellation of conditions considered comprising crucial risks conducive to T2DM and cardiovascular disease (CVD). Over the last 12 years, MetS criteria have seen five major revisions in details. However, all operationalized definitions address specific metabolic abnormalities, hypertension and obesity (Eckel et al., 2010). MetS is present in 12.5%–19.1% of patients with normal glucose tolerance (Riediger and Clara, 2011), in 55% of those with impaired fasting glucose tolerance, and in 81% of those with T2DM (Ginsberg and Stalenhoef, 2003). The presence of MetS increased the risk of T2DM manifestation 24-fold during a 5-year period (Sattar et al., 2003). Furthermore, CVD risk climbed towards 20% once T2DM had developed in MetS patients (Girman et al., 2004). MetS risk increases with aging, being 44% in the seventh decade of life having the MetS compared to 7% prevalence in the third decade (Ford et al., 2002). Elevated MetS risks have been reported for depressive disorders and schizophrenia, and also psychological
n
Corresponding author. Tel.: þ 44 20 7848 5110; fax: þ44 20 7848 0379. E-mail address: [email protected] (E. Lemche).
0165-1781/$ - see front matter & 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.psychres.2013.12.004
distress increases the risk for MetS by a factor of more than two (Puustinen et al., 2011). Alexithymia is a trait that comprises impairments in the perception of bodily states, their cognitive representation, and verbal communication. Increasing evidence suggests that alexithymia plays a role in a range of psychiatric, neurological, and internal medical problems, such as schizophrenia (Maggini and Raballo, 2004), traumatic brain injury (Wood et al., 2009) or CVD (Kauhanen et al., 1994). Alexithymia has been suggested to consist of a neurodevelopmental cognitive deficit originating in parent–offspring transmission (Lemche et al., 2004). Several neuroimaging studies have revealed that alexithymia traits are based on brain regions subserving interoception and physiological monitoring (e.g. Lemche et al., 2013). General health behaviors, such as poor dieting habits and sedentary lifestyles known to contribute to MetS, were found influenced by alexithymia (Helmers and Mente, 1999). A corpus of findings in T1DM supports the assumption that mental factors such as alexithymia in parents and in diabetic offspring, and the ability to monitor bodily homeostasis, might have an impact upon glycemic control and hypoglycemic incidence rates (Chatzi et al., 2009; Housiaux et al., 2010; Luminet et al., 2006; Meunier et al., 2008; Topsever et al., 2006). It is therefore possible that interoceptive impairments could also be involved in the emergence of T2DM.
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
Given the evidence that the development of T2DM could be partially dependent on autonomic dysbalance, and perhaps also on the ability to monitor homeostasis of satiety levels, we tested the hypothesis, that alexithymia factors may contribute to diabetes risk in MetS. To test the possibility of such an influence, we measured severity of alexithymia and its subfactors in an at-risk sample for T2DM that fulfilled at least three criteria for MetS, and estimated relative risks from psychometric, sociodemographic and endocrinological variables.
2. Methods
439
2.3. Disease classification All patients had been consensus-diagnosed with one or more of the aforementioned primary diagnoses according to ICD-10 criteria by at least two cardiological and/or endocrinological specialists not involved in the study. The investigator ascertained MetS criteria conforming to International Diabetes Federation (IDF; www.idf.org/publications) consensus from patient files. The diagnosis of MetS was established as the presence of three or more of the following features: waist circumference 480 cm in women and 494 cm in men; fasting serum triglycerides Z 1.7 mmol/L, serum high-density lipoprotein (HDL) o 1.29 mmol/L in women and o1.03 mmol/L in men; systolic/diastolic blood pressure Z130/ 85 mm Hg or treated; fasting plasma glucose level Z5.6 mmol/L or T2DM. Laboratory analyses of all biological specimens were performed inhouse with enzymatic methods using commercially available reagents. We report descriptives of all measurements (Table 1), but statistical results pertaining to the IDF criteria only.
2.1. Eligibility criteria and recruitment 2.4. Self-report instruments The sample was drawn from Caucasian patients treated within the health care system of the Ukrainian Ministry of Transport – with a catchment area encompassing all Ukraine. The targeted participants constituted a national treatment cohort (N¼ 867) for metabolic disorders, who were seen once a year at specialized in- and outpatient departments of Railway Hospital No. 2 in Kiev, Ukraine, for preventive screenings. The patients of the total treatment cohort had been elected for annual re-examination based on documented familial risk factors for T2DM. Our internal inclusion criteria were the presence of valid MetS ranges, and exclusion criteria were T1DM as well as endocrinological problems other than MetS. From these 220 MetS patients, approximately 50% were interested in medication treatment only. N ¼107 of the identified MetS patients were willing to undergo further psychiatric examinations and self-report as part of their regular treatment. The patients fulfilling these motivational inclusion criteria were eligible for the present investigation. Of these were excluded patients with any history of psychiatric-neurological illness or substance abuse on the basis of screening interviews (N ¼6).
Alexithymia was measured using the official Russian language version (www. consilium-medicum.com/article/8766) of the Toronto Alexythymia Scale, 20-item version (TAS-20) (Bagby et al., 1994a; Bagby et al., 1994b). Each item is rated on a 5point Likert-type scale, ranging from ‘strongly disagree’ (1) to ‘strongly agree’ (5). According to the current clinical consensus, alexithymia severity is divided into three groups: subjects with a TAS-20 total score Z 61 are considered highalexithymic, and those with a score r 51 are considered low-alexithymic. Factor analyses have suggested that the TAS-20 has three factors quantifying different dimensions of alexithymia: Factor 1 assesses the capacity to identify emotional feelings and to distinguish them from bodily sensations (F1: Difficulty in Identifying Feelings); Factor 2 reflects the inability to communicate feelings and sensations to other people (F2: Difficulty in Describing Feelings); and Factor 3 indicates Externally Oriented Thinking (F3) (Table 1). These TAS factors are replicable across cultures, with well-established psychometric properties. Additional self-report instruments were administered for control purposes, but are not focus of this report.
2.2. Patient sampling
2.5. Biological laboratory data and statistical analysis
The Institutional Review Board of the National Medical University of the Ukraine had endorsed all procedures. The investigation was conducted in compliance with the Helsinki Declaration (www.wma.net/e/ethicsunit/helsinki.htm). All subjects gave written informed consent to the scientific use of their data, and were reimbursed for their participation. The final sample included N¼ 101 individuals (mean age 45.6 7 0.14 SEM years; education levels 4.73 70.11, 4¼ junior college level, 67 females) with a primary diagnosis of T2DM, obesity, and/or hypertension. It was in size and composition comparable to research samples typically used in biomarker detection studies in MetS, and in approximately counterbalancing diabetic and non-diabetic patients by design (e.g. Huang et al., 2013).
The laboratory of the Railway Hospital No. 2 was subject to quality control standards imposed by the Ukrainian Ministry of Public Health, and was certified accordingly. We conducted multivariable logistic regression and hierarchical linear regression analyses, following correlation analyses for variables pertaining to the MetS. STATAs/IC 11.2 for Mac (64-bit Intel version) (StataCorp LP, College Station, TX, U.S.A.) was used for the statistical analyses. For comparison and generalizability purposes, we utilized the online data of the well-known Nhanes2 study (N ¼ 10,531) (Center for Disease Control, Bethesda, MD: US Health and Nutrition Survey II; protocols at archive.nlm.nih.gov/proj/dxpnet/nhanes/docs/doc/nhanes2/labproc. pdf), accessed through STATAs journal (http://www.stata-press.com/data/r11/ nhanes2), to serve as control group.
Table 1 Biological and self-report questionnaire descriptives: comparison of gender differences. Variables
Men
Women
Student's t
P Value
Age group Education level Family size Number of children TAS-20 mean score TAS-20 factor 1 TAS-20 factor 2 TAS-20 factor 3 Waist cm Body-mass index Triglycerides mmol/L Cholesterol mmol/L HDL mmol/L LDL g/L Microalbuminuria % Systolic blood Pressure mm/Hg Diastolic blood Pressure mm/Hg Type 2 diabetes mellitus Fasting plasma glucose mmol/L
4.26 7 1.50 4.82 7 0.99 2.85 7 0.86 0.79 7 0.41 56.53 7 13.69 20.477 7.09 14.59 7 4.42 21.747 4.76 100.50 7 13.61 31.59 7 7.52 2.93 7 0.92 5.86 7 1.11 0.93 7 0.06 5.167 1.65 357 0.48 143.977 14.02 90.59 7 6.72 0.53 7 0.51 6.047 1.57
4.727 1.32 4.69 7 1.18 3.377 1.20 0.69 7 0.47 55.54 7 11.96 20.96 7 6.52 12.99 7 4.08 21.707 4.37 90.25 7 7.64 32.717 5.85 3.34 7 0.79 6.32 7 1.19 1.117 0.09 6.277 2.89 337 0.47 147.317 12.65 89.03 7 5.31 0.60 7 0.49 6.58 7 2.14
1.484 0.611 2.501 1.187 0.359 0.333 1.767 0.035 4.074 0.0757 2.197 1.923 10.874 1.460 0.243 1.169 1.179 0.639 1.091
0.143 0.543 0.014 0.239 0.721 0.740 0.082 0.972 0.000 0.452 0.032 0.059 0.000 0.154 0.809 0.247 0.244 0.525 0.281
Note: N ¼ 101. Plus-minus values are mean 7 SD. TAS-20 20-item Toronto Alexithymia Scale, HDL high-density lipoprotein, LDL low-density lipoprotein.
440
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
3. Results 3.1. Scale reliabilities and alexithymia distribution The internal consistencies (Cronbach's α coefficients) for the TAS and its subscales (TAS-20 α ¼ 0.847, F1 α ¼0.886, F2 α ¼ 0.705, F3 α ¼ 0.691) were satisfactory for subsequent analyses. Of the socio-demographic variables, only education level showed negative association with alexithymia (r ¼ 0.443, p o0.002). On the basis of the TAS-20 cut-off scores, patients were subdivided into three groups: a low (N ¼31), a moderate (N ¼33), and high alexithymia group (N ¼ 37). Fig. 1 shows the distribution of alexithymia severity for patients with and without previously diagnosed T2DM. The result of cross-tabulation shows that highalexithymics are significantly over-represented amongst T2DM patients (Goodman and Kruskal's coefficient Γ ¼0.481, p o0.001): Of T2DM patients (57.4% of the sample) are 46.6% high- and 34.5% mid-alexithymic, whereas 64.5% of non-diabetes patients are low-alexithymic. 3.2. Correlation analyses Waist circumference showed negative association with HDL (r ¼ 0.514, p o0.0001) and correlated positively with blood pressure minimum (r ¼0.304, p o0.002). There were significant positive correlations of triglycerides with microalbuminuria (r ¼0.308, p o0.002), blood pressure maximum (r ¼ 0.334, p o0.001), T2DM status (r ¼0.341, p o0.0001), and fasting plasma glucose levels (r ¼0.327, p o0.011). HDL exhibited inverse correlation with blood pressure minimum (r ¼ 0.203, p o0.042) and fasting plasma glucose levels (r ¼ 0.248, p o0.056). Microalbuminuria was positively associated with T2DM status (Cramer's V¼ 0.359, p o0.0001) and fasting plasma glucose levels (r ¼0.407, p o0.001). Blood pressure minima and maxima were correlated (r ¼0.564, p o0.001), and also T2DM and fasting plasma glucose levels (r ¼0.716, p o0.0001). Next to alexithymia severity (TAS-20), patients also filled selfreport forms with the Russian language versions of the Spielberger State-Trait Anxiety Inventory, the Buss–Durkee Hostility Scale, and the Zung Depression Scale for control purposes. These self-report scales are not the in-depth focus of this report; results are only reported here for control purposes. All self-report measures were significantly intercorrelated in zero-order correlations (rs 0.627– 0.274, ps 0.0001–0.005). Significant associations with T2DM status were observed positively for alexithymia severity (r ¼0.461, p o0.0001), and negatively for aggressivity (r ¼ 0.231, p o0.02) and trait anxiety level (r ¼ 0.179, p o0.08). All other measures, including depression, state anxiety, and hostility were unrelated to the diagnosis of T2DM (all rs o0.15).
Of the subscales of the TAS-20 (factors F1–F3), associations with laboratory variables remained on a low-to-moderate range (all rs o0.3, all ps 40.05), except for F2 and LDL (r¼ 0.509, po 0.001), F3 and microalbuminuria (r ¼0.5797, p o0.001), and F3 and T2DM status (r ¼0.545, p o0.001). 3.3. Comparison with US National Health and Nutrition survey For control purposes, the (standardized) variables systolic and diastolic blood pressures, triglycerides, total cholesterol, HDL and BMI were compared in their association patterns, separately for the diabetic and non-diabetic white groups. We computed correlations for those laboratory variable pairs, which were measured both in the Nhanes2 and the present sample. In the Nhanes2 diabetic group, 9 out of 15, and in the Nhanes2 non-diabetic group, 10 out of 15 pairwise associations had similar magnitudes of correlation coefficients as in the respective Kiev subsamples (results not shown). Association pairs similar in both comparisons were the following: total cholesterol–triglycerides, total cholesterol–HDL, systolic blood pressure–HDL, diastolic blood pressure– total cholesterol, diastolic–systolic blood pressure, BMI–triglycerides. These analyses suggested that the pathological ranges of the Kiev patient data are in majority well comparable to the basic pathophysiological associations found in a representative control sample. 3.4. Logistic regression analyses Following correlation analyses, we adjusted bootstrapped linear regression models for continuous variables and bootstrapped logistic regression for binary criterion variables. The TAS-20 total score (as well as other psychometric measures, data not reported here) was used as regressor, and all additionally measured biological and social confounder variables were entered into forward stepping procedures until the best fitting models (in terms of χ2 and significance levels) were determined. It remained that total TAS-20 proved as the main predictor for the clinical diagnosis of T2DM in the simplest logit model: Wald-χ2(1,101)¼ 362.70, OR 1.073, log likelihood¼ 61.091, Z¼ 19.04, po0.00001 (95%CIs 1.065–1.081; 100 iterations). The second best model was the inclusion of the age variable with TAS-20: Wald-χ2(2,101)¼ 20.13, log likelihood¼ 58.823, with TAS-20 OR 1.071, Z¼ 3.40, po0.001 (95%CIs 1.029–1.114) and age OR 1.405, Z¼ 2.07, po0.038 (95%CIs 1.018–1.938). For microalbuminuria, also multivariable logit models were tested, and again, alexithymia remained the sole significant psychometric predictor variable. However, with adjustment for different clustering between the two sexes, the best fitting model was Wald-χ2(2,101)¼12.21, OR 1.030, log likelihood¼ 62.990, Z¼3.49, po0.0005 (95%CIs 1.013–1.048; 100 iterations). Regarding contributions of the biological variables for the manifestation of T2DM, significant regression equations (in a multivariable bootstrapped logit model, Wald-χ2(1,101) ¼40.82, log likelihood ¼ 30.71, p o0.001 (95%CIs 1.032–1.115)) could be obtained only for fasting glucose levels (Wald-χ2 ¼7.74, p o0.005, 95%CIs 1.695–21.019) and microalbuminuria (Helmert contrast, Wald-χ2 ¼3.051, p o0.06, 95%CIs 0.021–1.248; 100 iterations) predicting T2DM, consistent with general clinical expectation. 3.5. Hierarchical linear regression
Fig. 1. Distribution of percentage type 2 diabetes mellitus across alexithymia severity groups.
In the statistical prediction of triglycerides with TAS-20 as the regressor, again the simplest regression equation provided the best model fit. With 100 bootstrap iterations, the final model converged at Wald-χ2 ¼1299.27, Z¼ 36.05, po 0.00001 (95%CIs 0.052–0.058). Despite significant sex differences in triglyceride levels (Table 1), the inclusion of the gender variable did not improve model fit.
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
441
Fig. 2. ROC curves of alexithymia scores predicting clinical diagnoses of type 2 diabetes. A (left panel) TAS-20 score alone; B (right panel) TAS-20 score combined with age.
For the prediction of blood pressure maximum with alexithymia, bootstrap replications were employed with 100 iterations. The model converged at Wald-χ2 ¼2309.05, Z¼48.05, p o0.00001 (95%CIs 2.402–2.606). Alexithymia also statistically predicted waist girth at Wald-χ2 ¼1418.64, Z¼37.66, p o0.00001 (95%CIs 1.151–1.681), when adjusted for sex difference. Alexithymia severity also predicted BMI: Wald-χ2 ¼1332.67, Z¼36.51, p o0.00001 (95%CIs 0.521–0.581; 100 iterations). For dependent variables with significant sex differences (Table 1), separate regression models were obtained for each sex (i.e. waist, triglycerides, HDL, and total cholesterol). These separate models did not differ from the combined-sex models in their significance levels. 3.6. Receiver operating characteristics Signal detection methods are often combined with regression models to ascertain clinical cut-off levels for clinical diagnoses. We further investigated in a confirmatory approach classification decisions for binary clinical outcomes at varying levels of alexithymia severity. Following logistic regression, receiver-operating characteristics (ROCs) were computed for binary variables. ROC models enable testing classification specificity of continuous measures for binary variables such as clinical diagnoses. The area under the ROC curve (AUC) is a graphical representation of the sensitivity for a binary classifier system (the diagnosis) as its discrimination threshold (the continuous variable) is varied. The graphical representation of the area reflects the percentage of correctly classified observations. With TAS-20 as discrimination variable and T2DM status as classifier, the AUC ¼0.711. Including the age variable, the AUC¼ 0.742 (Fig. 2). With TAS-20 and subject sex as discrimination variable and microalbuminuria as classifier, the AUC¼ 0.594 (Fig. 3). An AUC 40.7 is considered to indicate substantial risk (Szmukler et al., 2012). All ROC analyses, however, successfully refuted the null hypothesis of 0.5 in areal classification, and thus demonstrated sufficient specificity for the diagnoses.
4. Discussion We aimed to examine the potential risk of alexithymia, a clinical trait implicating interoceptive impairments, deficits in physiological–emotional introspection, and dysbalanced autonomic reactivity, for the development of T2DM amongst MetS patients. To this end, we investigated the relative risks of alexithymia for T2DM, and for biomarkers of MetS and CVD, in a cross-sectional sample of MetS patients with high treatment adherence and dieting compliance. Biological data relevant for the IDF diagnoses of MetS and possible socio-demographic confounders were subjected to correlation analyses, and to multivariable logistic and hierarchical linear regression analyses, followed by ROC statistics, where appropriate.
This cross-sectional study using regression statistics provides evidence that (a) alexithymia severity is a substantial psychometric predictor of T2DM in MetS patients, and (b) that alexithymia also statistically predicts indicators of obesity (waist girth, BMI), dyslipidemia (triglyceride level), hypertension (maximum blood pressure), and a subclinical biomarker of CVD risk (microalbuminuria) in the context of established MetS. Previous evidence had suggested that alexithymia might be related to obesity (Elfhag and Lundh, 2007) in relation with other personality traits, as well as with hypertension and CVD risks (Gathercole and Stewart, 2010). The novelty of the present results lie in the fact that alexithymia statistically predicts an ensemble of biomarkers of MetS and CVD risks in MetS patients, in addition to T2DM diagnosis. Age risk significantly contributed to T2DM prediction, and differential sex effects correlated with waist circumference. All other social–familial variables did not account for significant variance proportions in any of our regression models. The receiveroperating curves indicated that alexithymia severity reaches high sensitivity for the clinical T2DM diagnosis, and moderate sensitivity for microalbuminuria. Our results therefore support the notion that alexithymia severity may contribute to MetS by covarying with several of its single components, and is a substantial concurrent indicator of T2DM and CVD risks in MetS. Our main assumption that alexithymia severity, as an indicator of impaired interoception and introspection capacities, could contribute to T2DM, was supported. The magnitude of the estimated relative risks here is at a level comparable to each single of the hitherto known genetic T2DM risks (i.e. ORs 1–2). To our knowledge, attempts to predict T2DM with alexithymia have been unsuccessful in other clinical groups so far including eating disorders (e.g. Eriksson et al., 2012). However, a population study estimating the relative risk of alexithymia for atherosclerosis and hypertension reported similar odds ratios as in our study (Grabe et al., 2010). Paralleling our current findings, confoundation by socio-demographic measures was not observed in that study. Limited introspective and/or interoceptive capacities in MetS could imply that satiety states may not be well sensed in this group. However, there is neuroimaging evidence that satiety states can be interocepted in normal individuals by bottom-up cognitive hierarchies, in which the supreme levels consist of the caudolateral orbital cortex, the inferior parietal lobule and the hypothalamus (Batterham et al., 2007). The current results also imply that autonomic functioning may be altered. Systolic blood pressure is a physiological indicator of hypertension, and autonomic reactivity was found biased in alexithymia, according to previous findings (Lemche et al., 2004). Specifically, the interplay of sympathetic and vagal outflow in emotional states may be disturbed in alexithymic individuals (Neumann et al., 2004). It is therefore likely that an increase of alexithymia severity could also reflect abnormalities in the hypothalamic pituitary adrenocortical axis (HPA). Hypercortisolism, the most common HPA dysregulation, has been confirmed in alexithymics (de Timary et al., 2008). There is also a robust set of data implicating hypercortisolism in the pathogenesis of
442
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
Fig. 3. ROC curves of alexithymia scores predicting microalbuminuria. A (left panel) TAS-20 scores of patients without and with microalbuminuria; B (right panel) classification specificity of microalbuminuria according to TAS-20 score.
obesity, T2DM and the MetS. In addition, experimental evidence demonstrated improvements in metabolic profiles by inactivating free cortisol (Gathercole and Stewart, 2010). Overall, the findings of the current study offer initial evidence, but no final conclusion, to support the suggestion that cerebral and mental factors, such as limited perception of hunger–satiety states, nonperception of homeostatic processes, and non-recognition of bodily sensations and dysregulated emotional states, could contribute to T2DM. The current study is limited due to its sample size and crosssectional design, albeit provides robust statistical effects. A further limitation in this cross sectional design is the inability to include measures of treatment adherence or dieting compliance, two factors that previously explained the relationship between psychological factors such as diabetes distress and diabetes disequilibrium (Fisher et al., 2010a). While depression or negative affect has also been found to attenuate metabolic control in T2DM (Gois et al., 2011), recent findings suggest that diabetes distress and trait depression should be differentiated (Fisher et al., 2010b). In our study, depression was unrelated to T2DM, but associated with alexithymia. Our directions for future research recommend that the present findings should be re-investigated in a longitudinal and/or prospective cohort, with a population based sampling strategy. The inclusion of cerebral biomarkers (such as melanocortins, leptin, ghrelin), the concurrent measurement of a dysbalance of the HPA, and systematic testing of emotional dysregulation would be advisable. Controlling for major genetic risk loci, while conducting functional neuroimaging of the hypothalamus would enable attributing cerebral connectivity patterns to specific risks.
Funding A.V.L. acknowledges intramural research funding from the National Medical University, Kiev, Ukraine. O.S.C. was supported by a Ukrainian Ministry of Public Health center grant on “Quality of Life in Diabetes”.
Role of the funding sources The funding sources were not involved in the study design; collection, analysis, or interpretation of data; or in preparation or submission of the manuscript for publication.
References Bagby, R.M., Parker, J.D., Taylor, G.J., 1994a. The twenty-item Toronto Alexithymia Scale-I. Item selection and cross-validation of the factor structure. Journal of Psychosomatic Research 38, 23–32. Bagby, R.M., Taylor, G.J., Parker, J.D., 1994b. The Twenty-item Toronto Alexithymia Scale-II. Convergent, discriminant, and concurrent validity. Journal of Psychosomatic Research 38, 33–40. Batterham, R.L., ffytche, D.H., Rosenthal, J.M., Zelaya, F.O., Barker, G.J., Withers, D.J., Williams, S.C., 2007. PYY modulation of cortical and hypothalamic brain areas predicts feeding behaviour in humans. Nature 450, 106–109. Chatzi, L., Bitsios, P., Solidaki, E., Christou, I., Kyrlaki, E., Sfakianaki, M., Kogevinas, M., Kefalogiannis, N., Pappas, A., 2009. Type 1 diabetes is associated with alexithymia in nondepressed, non-mentally ill diabetic patients: a case-control study. Journal of Psychosomatic Research 67 (4), 307–313. de Timary, P., Roy, E., Luminet, O., Fillee, C., Mikolajczak, M., 2008. Relationship between alexithymia, alexithymia factors and salivary cortisol in men exposed to a social stress test. Psychoneuroendocrinology 33, 1160–1164. Eckel, R.H., Alberti, K.G., Grundy, S.M., Zimmet, P.Z., 2010. The metabolic syndrome. Lancet 375, 181–183. Elfhag, K., Lundh, L.G., 2007. TAS-20 alexithymia in obesity, and its links to personality. Scandinavian Journal of Psychology 48, 391–398. Eriksson, A.K., Gustavsson, J.P., Hilding, A., Granath, F., Ekbom, A., Ostenson, C.G., 2012. Personality traits and abnormal glucose regulation in middle-aged Swedish men and women. Diabetes Research and Clinical Practice 95, 145–152. Fisher, L., Glasgow, R.E., Strycker, L.A., 2010a. The relationship between diabetes distress and clinical depression with glycemic control among patients with type 2 diabetes. Diabetes Care 33, 1034–1036. Fisher, L., Mullan, J.T., Arean, P., Glasgow, R.E., Hessler, D., Masharani, U., 2010b. Diabetes distress but not clinical depression or depressive symptoms is associated with glycemic control in both cross-sectional and longitudinal analyses. Diabetes Care 33, 23–28. Ford, E.S., Giles, W.H., Dietz, W.H., 2002. Prevalence of the metabolic syndrome among US adults: findings from the third National Health and Nutrition Examination Survey. Journal of the American Medical Association 287, 356–359. Gathercole, L.L., Stewart, P.M., 2010. Targeting the pre-receptor metabolism of cortisol as a novel therapy in obesity and diabetes. J. Steroid. Biochem. Mol. Biol. 122 (1-3), 21–27. Ginsberg, H.N., Stalenhoef, A.F., 2003. The metabolic syndrome: targeting dyslipidaemia to reduce coronary risk. Journal of Cardiovascular Risk 10, 121–128. Girman, C.J., Rhodes, T., Mercuri, M., Pyorala, K., Kjekshus, J., Pedersen, T.R., Beere, P. A., Gotto, A.M., Clearfield, M., 2004. The metabolic syndrome and risk of major coronary events in the Scandinavian Simvastatin Survival Study (4S) and the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS). American Journal of Cardiology 93, 136–141. Gois, C., Barbosa, A., Ferro, A., Santos, A.L., Sousa, F., Akiskal, H., Akiskal, K., Figueira, M.L., 2011. The role of affective temperaments in metabolic control in patients with type 2 diabetes. Journal of Affective Disorders 134, 52–58. Grabe, H.J., Schwahn, C., Barnow, S., Spitzer, C., John, U., Freyberger, H.J., Schminke, U., Felix, S., Volzke, H., 2010. Alexithymia, hypertension, and subclinical atherosclerosis in the general population. Journal of Psychosomatic Research 68, 139–147. Helmers, K.F., Mente, A., 1999. Alexithymia and health behaviors in healthy male volunteers. Journal of Psychosomatic Research 47, 635–645.
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
Housiaux, M., Luminet, O., Van Broeck, N., Dorchy, H., 2010. Alexithymia is associated with glycaemic control of children with type 1 diabetes. Diabetes & Metabolism 36, 455–462. Huang, C.F., Cheng, M.L., Fan, C.M., Hong, C.Y., Shiao, M.S., 2013. Nicotinuric acid: a potential marker of metabolic syndrome through a metabolomics-based approach. Diabetes Care 36, 1729–1731. Kauhanen, J., Kaplan, G.A., Cohen, R.D., Salonen, R., Salonen, J.T., 1994. Alexithymia may influence the diagnosis of coronary heart disease. Psychosomatic Medicine 56, 237–244. Lemche, E., Klann-Delius, G., Koch, R., Joraschky, P., 2004. Mentalizing language development in a longitudinal attachment sample: implications for alexithymia. Psychotherapy and Psychosomatics 73, 366–374. Lemche, E., Surguladze, S.A., Giampietro, V.P., Brammer, M.J., Williams, S.C.R., Sierra, M., David, A.S., Phillips, M.L., 2013. Interoceptive–reflective regions differentiate alexithymia traits in depersonalization disorder. Psychiatry Research Neuroimaging 214 (1), 66–72. Luminet, O., de Timary, P., Buysschaert, M., Luts, A., 2006. The role of alexithymia factors in glucose control of persons with type 1 diabetes: a pilot study. Diabetes & Metabolism 32, 417–424. Maggini, C., Raballo, A., 2004. Alexithymia and schizophrenic psychopathology. Acta Bio Medica 75, 40–49. Meunier, J., Dorchy, H., Luminet, O., 2008. Does family cohesiveness and parental alexithymia predict glycaemic control in children and adolescents with diabetes? Diabetes & Metabolism 34, 473–481.
443
Neumann, S.A., Sollers 3rd, J.J., Thayer, J.F., Waldstein, S.R., 2004. Alexithymia predicts attenuated autonomic reactivity, but prolonged recovery to anger recall in young women. International Journal of Psychophysiology 53, 183–195. Puustinen, P.J., Koponen, H., Kautiainen, H., Mantyselka, P., Vanhala, M., 2011. Psychological distress predicts the development of the metabolic syndrome: a prospective population-based study. Psychosomatic Medicine 73, 158–165. Riediger, N.D., Clara, I., 2011. Prevalence of metabolic syndrome in the Canadian adult population. Canadian Medical Association Journal 183, E1127–1134. Sattar, N., Gaw, A., Scherbakova, O., Ford, I., O’Reilly, D.S., Haffner, S.M., Isles, C., Macfarlane, P.W., Packard, C.J., Cobbe, S.M., Shepherd, J., 2003. 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 108, 414–419. Szmukler, G., Everitt, B., Leese, M., 2012. Risk assessment and receiver operating characteristic curves. Psychological Medicine 42, 895–898. Topsever, P., Filiz, T.M., Salman, S., Sengul, A., Sarac, E., Topalli, R., Gorpelioglu, S., Yilmaz, T., 2006. Alexithymia in diabetes mellitus. Scottish Medical Journal 51, 15–20. Wood, R.L., Williams, C., Kalyani, T., 2009. The impact of alexithymia on somatization after traumatic brain injury. Brain Injury 23, 649–654.
Contents lists available at ScienceDirect
Psychiatry Research journal homepage: www.elsevier.com/locate/psychres
Alexithymia as a risk factor for type 2 diabetes mellitus in the metabolic syndrome: a cross-sectional study Alexandra V. Lemche a, Oleg S. Chaban b, Erwin Lemche c,n a
Department of Psychosomatic Medicine, Charité University Medicine Berlin, Germany Section of Neuroses and Somatoform Disorders, Bogolomets National Medical University, Kiev, Ukraine c Section of Cognitive Neuropsychiatry, Institute of Psychiatry, King's College London, UK b
art ic l e i nf o
a b s t r a c t
Article history: Received 16 June 2012 Received in revised form 13 August 2013 Accepted 4 December 2013 Available online 14 December 2013
Alexithymia is a clinical trait consisting of diminished introspective and interoceptive capacities that has been shown to implicate elevated autonomic outflow and to bias for hypertension. To estimate relative risk associated with alexithymia in the metabolic syndrome (MetS), we conducted a cross-sectional analysis of patients with manifest type 2 diabetes mellitus (T2DM) or familial diabetes risk (N¼ 101; 67 females; age 45.6 713.96) in a nationwide sampled treatment cohort for MetS in the Ukrainian governmental health care system. Laboratory data of single components of the MetS according to International Diabetes Federation Consensus were dependent measures in multivariable regression models with self-reported alexithymia severity (TAS-20) and socio-demographic data. TAS-20 as the sole surviving psychometric predictor for T2DM in the simplest regression equation provided the best model fit: OR 1.073, Z ¼19.04, (95%CIs 1.065–1.081). For microalbuminuria, the best fitting model was OR 1.030, Z¼ 3.49 (95%CIs 1.013–1.048). TAS-20 predicted also triglyceride level at Wald-χ2 ¼1299.27, Z¼36.05 (95%CIs 0.052–0.058) and blood pressure maximum at Wald-χ2 ¼2309.05, Z¼48.05 (95%CIs 2.402– 2.606). Our results show that alexithymia severity contributes to MetS by covarying with several of its single components, and that it may be a substantial concurrent indicator of T2DM and cardiovascular risks in MetS. & 2013 Elsevier Ireland Ltd. All rights reserved.
Keywords: Metabolic syndrome Type 2 diabetes mellitus Cardiovascular disease Toronto Alexithymia Scale Relative risk Cross-sectional studies National cohort
1. Introduction Metabolic syndrome (MetS) is a constellation of conditions considered comprising crucial risks conducive to T2DM and cardiovascular disease (CVD). Over the last 12 years, MetS criteria have seen five major revisions in details. However, all operationalized definitions address specific metabolic abnormalities, hypertension and obesity (Eckel et al., 2010). MetS is present in 12.5%–19.1% of patients with normal glucose tolerance (Riediger and Clara, 2011), in 55% of those with impaired fasting glucose tolerance, and in 81% of those with T2DM (Ginsberg and Stalenhoef, 2003). The presence of MetS increased the risk of T2DM manifestation 24-fold during a 5-year period (Sattar et al., 2003). Furthermore, CVD risk climbed towards 20% once T2DM had developed in MetS patients (Girman et al., 2004). MetS risk increases with aging, being 44% in the seventh decade of life having the MetS compared to 7% prevalence in the third decade (Ford et al., 2002). Elevated MetS risks have been reported for depressive disorders and schizophrenia, and also psychological
n
Corresponding author. Tel.: þ 44 20 7848 5110; fax: þ44 20 7848 0379. E-mail address: [email protected] (E. Lemche).
0165-1781/$ - see front matter & 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.psychres.2013.12.004
distress increases the risk for MetS by a factor of more than two (Puustinen et al., 2011). Alexithymia is a trait that comprises impairments in the perception of bodily states, their cognitive representation, and verbal communication. Increasing evidence suggests that alexithymia plays a role in a range of psychiatric, neurological, and internal medical problems, such as schizophrenia (Maggini and Raballo, 2004), traumatic brain injury (Wood et al., 2009) or CVD (Kauhanen et al., 1994). Alexithymia has been suggested to consist of a neurodevelopmental cognitive deficit originating in parent–offspring transmission (Lemche et al., 2004). Several neuroimaging studies have revealed that alexithymia traits are based on brain regions subserving interoception and physiological monitoring (e.g. Lemche et al., 2013). General health behaviors, such as poor dieting habits and sedentary lifestyles known to contribute to MetS, were found influenced by alexithymia (Helmers and Mente, 1999). A corpus of findings in T1DM supports the assumption that mental factors such as alexithymia in parents and in diabetic offspring, and the ability to monitor bodily homeostasis, might have an impact upon glycemic control and hypoglycemic incidence rates (Chatzi et al., 2009; Housiaux et al., 2010; Luminet et al., 2006; Meunier et al., 2008; Topsever et al., 2006). It is therefore possible that interoceptive impairments could also be involved in the emergence of T2DM.
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
Given the evidence that the development of T2DM could be partially dependent on autonomic dysbalance, and perhaps also on the ability to monitor homeostasis of satiety levels, we tested the hypothesis, that alexithymia factors may contribute to diabetes risk in MetS. To test the possibility of such an influence, we measured severity of alexithymia and its subfactors in an at-risk sample for T2DM that fulfilled at least three criteria for MetS, and estimated relative risks from psychometric, sociodemographic and endocrinological variables.
2. Methods
439
2.3. Disease classification All patients had been consensus-diagnosed with one or more of the aforementioned primary diagnoses according to ICD-10 criteria by at least two cardiological and/or endocrinological specialists not involved in the study. The investigator ascertained MetS criteria conforming to International Diabetes Federation (IDF; www.idf.org/publications) consensus from patient files. The diagnosis of MetS was established as the presence of three or more of the following features: waist circumference 480 cm in women and 494 cm in men; fasting serum triglycerides Z 1.7 mmol/L, serum high-density lipoprotein (HDL) o 1.29 mmol/L in women and o1.03 mmol/L in men; systolic/diastolic blood pressure Z130/ 85 mm Hg or treated; fasting plasma glucose level Z5.6 mmol/L or T2DM. Laboratory analyses of all biological specimens were performed inhouse with enzymatic methods using commercially available reagents. We report descriptives of all measurements (Table 1), but statistical results pertaining to the IDF criteria only.
2.1. Eligibility criteria and recruitment 2.4. Self-report instruments The sample was drawn from Caucasian patients treated within the health care system of the Ukrainian Ministry of Transport – with a catchment area encompassing all Ukraine. The targeted participants constituted a national treatment cohort (N¼ 867) for metabolic disorders, who were seen once a year at specialized in- and outpatient departments of Railway Hospital No. 2 in Kiev, Ukraine, for preventive screenings. The patients of the total treatment cohort had been elected for annual re-examination based on documented familial risk factors for T2DM. Our internal inclusion criteria were the presence of valid MetS ranges, and exclusion criteria were T1DM as well as endocrinological problems other than MetS. From these 220 MetS patients, approximately 50% were interested in medication treatment only. N ¼107 of the identified MetS patients were willing to undergo further psychiatric examinations and self-report as part of their regular treatment. The patients fulfilling these motivational inclusion criteria were eligible for the present investigation. Of these were excluded patients with any history of psychiatric-neurological illness or substance abuse on the basis of screening interviews (N ¼6).
Alexithymia was measured using the official Russian language version (www. consilium-medicum.com/article/8766) of the Toronto Alexythymia Scale, 20-item version (TAS-20) (Bagby et al., 1994a; Bagby et al., 1994b). Each item is rated on a 5point Likert-type scale, ranging from ‘strongly disagree’ (1) to ‘strongly agree’ (5). According to the current clinical consensus, alexithymia severity is divided into three groups: subjects with a TAS-20 total score Z 61 are considered highalexithymic, and those with a score r 51 are considered low-alexithymic. Factor analyses have suggested that the TAS-20 has three factors quantifying different dimensions of alexithymia: Factor 1 assesses the capacity to identify emotional feelings and to distinguish them from bodily sensations (F1: Difficulty in Identifying Feelings); Factor 2 reflects the inability to communicate feelings and sensations to other people (F2: Difficulty in Describing Feelings); and Factor 3 indicates Externally Oriented Thinking (F3) (Table 1). These TAS factors are replicable across cultures, with well-established psychometric properties. Additional self-report instruments were administered for control purposes, but are not focus of this report.
2.2. Patient sampling
2.5. Biological laboratory data and statistical analysis
The Institutional Review Board of the National Medical University of the Ukraine had endorsed all procedures. The investigation was conducted in compliance with the Helsinki Declaration (www.wma.net/e/ethicsunit/helsinki.htm). All subjects gave written informed consent to the scientific use of their data, and were reimbursed for their participation. The final sample included N¼ 101 individuals (mean age 45.6 7 0.14 SEM years; education levels 4.73 70.11, 4¼ junior college level, 67 females) with a primary diagnosis of T2DM, obesity, and/or hypertension. It was in size and composition comparable to research samples typically used in biomarker detection studies in MetS, and in approximately counterbalancing diabetic and non-diabetic patients by design (e.g. Huang et al., 2013).
The laboratory of the Railway Hospital No. 2 was subject to quality control standards imposed by the Ukrainian Ministry of Public Health, and was certified accordingly. We conducted multivariable logistic regression and hierarchical linear regression analyses, following correlation analyses for variables pertaining to the MetS. STATAs/IC 11.2 for Mac (64-bit Intel version) (StataCorp LP, College Station, TX, U.S.A.) was used for the statistical analyses. For comparison and generalizability purposes, we utilized the online data of the well-known Nhanes2 study (N ¼ 10,531) (Center for Disease Control, Bethesda, MD: US Health and Nutrition Survey II; protocols at archive.nlm.nih.gov/proj/dxpnet/nhanes/docs/doc/nhanes2/labproc. pdf), accessed through STATAs journal (http://www.stata-press.com/data/r11/ nhanes2), to serve as control group.
Table 1 Biological and self-report questionnaire descriptives: comparison of gender differences. Variables
Men
Women
Student's t
P Value
Age group Education level Family size Number of children TAS-20 mean score TAS-20 factor 1 TAS-20 factor 2 TAS-20 factor 3 Waist cm Body-mass index Triglycerides mmol/L Cholesterol mmol/L HDL mmol/L LDL g/L Microalbuminuria % Systolic blood Pressure mm/Hg Diastolic blood Pressure mm/Hg Type 2 diabetes mellitus Fasting plasma glucose mmol/L
4.26 7 1.50 4.82 7 0.99 2.85 7 0.86 0.79 7 0.41 56.53 7 13.69 20.477 7.09 14.59 7 4.42 21.747 4.76 100.50 7 13.61 31.59 7 7.52 2.93 7 0.92 5.86 7 1.11 0.93 7 0.06 5.167 1.65 357 0.48 143.977 14.02 90.59 7 6.72 0.53 7 0.51 6.047 1.57
4.727 1.32 4.69 7 1.18 3.377 1.20 0.69 7 0.47 55.54 7 11.96 20.96 7 6.52 12.99 7 4.08 21.707 4.37 90.25 7 7.64 32.717 5.85 3.34 7 0.79 6.32 7 1.19 1.117 0.09 6.277 2.89 337 0.47 147.317 12.65 89.03 7 5.31 0.60 7 0.49 6.58 7 2.14
1.484 0.611 2.501 1.187 0.359 0.333 1.767 0.035 4.074 0.0757 2.197 1.923 10.874 1.460 0.243 1.169 1.179 0.639 1.091
0.143 0.543 0.014 0.239 0.721 0.740 0.082 0.972 0.000 0.452 0.032 0.059 0.000 0.154 0.809 0.247 0.244 0.525 0.281
Note: N ¼ 101. Plus-minus values are mean 7 SD. TAS-20 20-item Toronto Alexithymia Scale, HDL high-density lipoprotein, LDL low-density lipoprotein.
440
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
3. Results 3.1. Scale reliabilities and alexithymia distribution The internal consistencies (Cronbach's α coefficients) for the TAS and its subscales (TAS-20 α ¼ 0.847, F1 α ¼0.886, F2 α ¼ 0.705, F3 α ¼ 0.691) were satisfactory for subsequent analyses. Of the socio-demographic variables, only education level showed negative association with alexithymia (r ¼ 0.443, p o0.002). On the basis of the TAS-20 cut-off scores, patients were subdivided into three groups: a low (N ¼31), a moderate (N ¼33), and high alexithymia group (N ¼ 37). Fig. 1 shows the distribution of alexithymia severity for patients with and without previously diagnosed T2DM. The result of cross-tabulation shows that highalexithymics are significantly over-represented amongst T2DM patients (Goodman and Kruskal's coefficient Γ ¼0.481, p o0.001): Of T2DM patients (57.4% of the sample) are 46.6% high- and 34.5% mid-alexithymic, whereas 64.5% of non-diabetes patients are low-alexithymic. 3.2. Correlation analyses Waist circumference showed negative association with HDL (r ¼ 0.514, p o0.0001) and correlated positively with blood pressure minimum (r ¼0.304, p o0.002). There were significant positive correlations of triglycerides with microalbuminuria (r ¼0.308, p o0.002), blood pressure maximum (r ¼ 0.334, p o0.001), T2DM status (r ¼0.341, p o0.0001), and fasting plasma glucose levels (r ¼0.327, p o0.011). HDL exhibited inverse correlation with blood pressure minimum (r ¼ 0.203, p o0.042) and fasting plasma glucose levels (r ¼ 0.248, p o0.056). Microalbuminuria was positively associated with T2DM status (Cramer's V¼ 0.359, p o0.0001) and fasting plasma glucose levels (r ¼0.407, p o0.001). Blood pressure minima and maxima were correlated (r ¼0.564, p o0.001), and also T2DM and fasting plasma glucose levels (r ¼0.716, p o0.0001). Next to alexithymia severity (TAS-20), patients also filled selfreport forms with the Russian language versions of the Spielberger State-Trait Anxiety Inventory, the Buss–Durkee Hostility Scale, and the Zung Depression Scale for control purposes. These self-report scales are not the in-depth focus of this report; results are only reported here for control purposes. All self-report measures were significantly intercorrelated in zero-order correlations (rs 0.627– 0.274, ps 0.0001–0.005). Significant associations with T2DM status were observed positively for alexithymia severity (r ¼0.461, p o0.0001), and negatively for aggressivity (r ¼ 0.231, p o0.02) and trait anxiety level (r ¼ 0.179, p o0.08). All other measures, including depression, state anxiety, and hostility were unrelated to the diagnosis of T2DM (all rs o0.15).
Of the subscales of the TAS-20 (factors F1–F3), associations with laboratory variables remained on a low-to-moderate range (all rs o0.3, all ps 40.05), except for F2 and LDL (r¼ 0.509, po 0.001), F3 and microalbuminuria (r ¼0.5797, p o0.001), and F3 and T2DM status (r ¼0.545, p o0.001). 3.3. Comparison with US National Health and Nutrition survey For control purposes, the (standardized) variables systolic and diastolic blood pressures, triglycerides, total cholesterol, HDL and BMI were compared in their association patterns, separately for the diabetic and non-diabetic white groups. We computed correlations for those laboratory variable pairs, which were measured both in the Nhanes2 and the present sample. In the Nhanes2 diabetic group, 9 out of 15, and in the Nhanes2 non-diabetic group, 10 out of 15 pairwise associations had similar magnitudes of correlation coefficients as in the respective Kiev subsamples (results not shown). Association pairs similar in both comparisons were the following: total cholesterol–triglycerides, total cholesterol–HDL, systolic blood pressure–HDL, diastolic blood pressure– total cholesterol, diastolic–systolic blood pressure, BMI–triglycerides. These analyses suggested that the pathological ranges of the Kiev patient data are in majority well comparable to the basic pathophysiological associations found in a representative control sample. 3.4. Logistic regression analyses Following correlation analyses, we adjusted bootstrapped linear regression models for continuous variables and bootstrapped logistic regression for binary criterion variables. The TAS-20 total score (as well as other psychometric measures, data not reported here) was used as regressor, and all additionally measured biological and social confounder variables were entered into forward stepping procedures until the best fitting models (in terms of χ2 and significance levels) were determined. It remained that total TAS-20 proved as the main predictor for the clinical diagnosis of T2DM in the simplest logit model: Wald-χ2(1,101)¼ 362.70, OR 1.073, log likelihood¼ 61.091, Z¼ 19.04, po0.00001 (95%CIs 1.065–1.081; 100 iterations). The second best model was the inclusion of the age variable with TAS-20: Wald-χ2(2,101)¼ 20.13, log likelihood¼ 58.823, with TAS-20 OR 1.071, Z¼ 3.40, po0.001 (95%CIs 1.029–1.114) and age OR 1.405, Z¼ 2.07, po0.038 (95%CIs 1.018–1.938). For microalbuminuria, also multivariable logit models were tested, and again, alexithymia remained the sole significant psychometric predictor variable. However, with adjustment for different clustering between the two sexes, the best fitting model was Wald-χ2(2,101)¼12.21, OR 1.030, log likelihood¼ 62.990, Z¼3.49, po0.0005 (95%CIs 1.013–1.048; 100 iterations). Regarding contributions of the biological variables for the manifestation of T2DM, significant regression equations (in a multivariable bootstrapped logit model, Wald-χ2(1,101) ¼40.82, log likelihood ¼ 30.71, p o0.001 (95%CIs 1.032–1.115)) could be obtained only for fasting glucose levels (Wald-χ2 ¼7.74, p o0.005, 95%CIs 1.695–21.019) and microalbuminuria (Helmert contrast, Wald-χ2 ¼3.051, p o0.06, 95%CIs 0.021–1.248; 100 iterations) predicting T2DM, consistent with general clinical expectation. 3.5. Hierarchical linear regression
Fig. 1. Distribution of percentage type 2 diabetes mellitus across alexithymia severity groups.
In the statistical prediction of triglycerides with TAS-20 as the regressor, again the simplest regression equation provided the best model fit. With 100 bootstrap iterations, the final model converged at Wald-χ2 ¼1299.27, Z¼ 36.05, po 0.00001 (95%CIs 0.052–0.058). Despite significant sex differences in triglyceride levels (Table 1), the inclusion of the gender variable did not improve model fit.
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
441
Fig. 2. ROC curves of alexithymia scores predicting clinical diagnoses of type 2 diabetes. A (left panel) TAS-20 score alone; B (right panel) TAS-20 score combined with age.
For the prediction of blood pressure maximum with alexithymia, bootstrap replications were employed with 100 iterations. The model converged at Wald-χ2 ¼2309.05, Z¼48.05, p o0.00001 (95%CIs 2.402–2.606). Alexithymia also statistically predicted waist girth at Wald-χ2 ¼1418.64, Z¼37.66, p o0.00001 (95%CIs 1.151–1.681), when adjusted for sex difference. Alexithymia severity also predicted BMI: Wald-χ2 ¼1332.67, Z¼36.51, p o0.00001 (95%CIs 0.521–0.581; 100 iterations). For dependent variables with significant sex differences (Table 1), separate regression models were obtained for each sex (i.e. waist, triglycerides, HDL, and total cholesterol). These separate models did not differ from the combined-sex models in their significance levels. 3.6. Receiver operating characteristics Signal detection methods are often combined with regression models to ascertain clinical cut-off levels for clinical diagnoses. We further investigated in a confirmatory approach classification decisions for binary clinical outcomes at varying levels of alexithymia severity. Following logistic regression, receiver-operating characteristics (ROCs) were computed for binary variables. ROC models enable testing classification specificity of continuous measures for binary variables such as clinical diagnoses. The area under the ROC curve (AUC) is a graphical representation of the sensitivity for a binary classifier system (the diagnosis) as its discrimination threshold (the continuous variable) is varied. The graphical representation of the area reflects the percentage of correctly classified observations. With TAS-20 as discrimination variable and T2DM status as classifier, the AUC ¼0.711. Including the age variable, the AUC¼ 0.742 (Fig. 2). With TAS-20 and subject sex as discrimination variable and microalbuminuria as classifier, the AUC¼ 0.594 (Fig. 3). An AUC 40.7 is considered to indicate substantial risk (Szmukler et al., 2012). All ROC analyses, however, successfully refuted the null hypothesis of 0.5 in areal classification, and thus demonstrated sufficient specificity for the diagnoses.
4. Discussion We aimed to examine the potential risk of alexithymia, a clinical trait implicating interoceptive impairments, deficits in physiological–emotional introspection, and dysbalanced autonomic reactivity, for the development of T2DM amongst MetS patients. To this end, we investigated the relative risks of alexithymia for T2DM, and for biomarkers of MetS and CVD, in a cross-sectional sample of MetS patients with high treatment adherence and dieting compliance. Biological data relevant for the IDF diagnoses of MetS and possible socio-demographic confounders were subjected to correlation analyses, and to multivariable logistic and hierarchical linear regression analyses, followed by ROC statistics, where appropriate.
This cross-sectional study using regression statistics provides evidence that (a) alexithymia severity is a substantial psychometric predictor of T2DM in MetS patients, and (b) that alexithymia also statistically predicts indicators of obesity (waist girth, BMI), dyslipidemia (triglyceride level), hypertension (maximum blood pressure), and a subclinical biomarker of CVD risk (microalbuminuria) in the context of established MetS. Previous evidence had suggested that alexithymia might be related to obesity (Elfhag and Lundh, 2007) in relation with other personality traits, as well as with hypertension and CVD risks (Gathercole and Stewart, 2010). The novelty of the present results lie in the fact that alexithymia statistically predicts an ensemble of biomarkers of MetS and CVD risks in MetS patients, in addition to T2DM diagnosis. Age risk significantly contributed to T2DM prediction, and differential sex effects correlated with waist circumference. All other social–familial variables did not account for significant variance proportions in any of our regression models. The receiveroperating curves indicated that alexithymia severity reaches high sensitivity for the clinical T2DM diagnosis, and moderate sensitivity for microalbuminuria. Our results therefore support the notion that alexithymia severity may contribute to MetS by covarying with several of its single components, and is a substantial concurrent indicator of T2DM and CVD risks in MetS. Our main assumption that alexithymia severity, as an indicator of impaired interoception and introspection capacities, could contribute to T2DM, was supported. The magnitude of the estimated relative risks here is at a level comparable to each single of the hitherto known genetic T2DM risks (i.e. ORs 1–2). To our knowledge, attempts to predict T2DM with alexithymia have been unsuccessful in other clinical groups so far including eating disorders (e.g. Eriksson et al., 2012). However, a population study estimating the relative risk of alexithymia for atherosclerosis and hypertension reported similar odds ratios as in our study (Grabe et al., 2010). Paralleling our current findings, confoundation by socio-demographic measures was not observed in that study. Limited introspective and/or interoceptive capacities in MetS could imply that satiety states may not be well sensed in this group. However, there is neuroimaging evidence that satiety states can be interocepted in normal individuals by bottom-up cognitive hierarchies, in which the supreme levels consist of the caudolateral orbital cortex, the inferior parietal lobule and the hypothalamus (Batterham et al., 2007). The current results also imply that autonomic functioning may be altered. Systolic blood pressure is a physiological indicator of hypertension, and autonomic reactivity was found biased in alexithymia, according to previous findings (Lemche et al., 2004). Specifically, the interplay of sympathetic and vagal outflow in emotional states may be disturbed in alexithymic individuals (Neumann et al., 2004). It is therefore likely that an increase of alexithymia severity could also reflect abnormalities in the hypothalamic pituitary adrenocortical axis (HPA). Hypercortisolism, the most common HPA dysregulation, has been confirmed in alexithymics (de Timary et al., 2008). There is also a robust set of data implicating hypercortisolism in the pathogenesis of
442
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
Fig. 3. ROC curves of alexithymia scores predicting microalbuminuria. A (left panel) TAS-20 scores of patients without and with microalbuminuria; B (right panel) classification specificity of microalbuminuria according to TAS-20 score.
obesity, T2DM and the MetS. In addition, experimental evidence demonstrated improvements in metabolic profiles by inactivating free cortisol (Gathercole and Stewart, 2010). Overall, the findings of the current study offer initial evidence, but no final conclusion, to support the suggestion that cerebral and mental factors, such as limited perception of hunger–satiety states, nonperception of homeostatic processes, and non-recognition of bodily sensations and dysregulated emotional states, could contribute to T2DM. The current study is limited due to its sample size and crosssectional design, albeit provides robust statistical effects. A further limitation in this cross sectional design is the inability to include measures of treatment adherence or dieting compliance, two factors that previously explained the relationship between psychological factors such as diabetes distress and diabetes disequilibrium (Fisher et al., 2010a). While depression or negative affect has also been found to attenuate metabolic control in T2DM (Gois et al., 2011), recent findings suggest that diabetes distress and trait depression should be differentiated (Fisher et al., 2010b). In our study, depression was unrelated to T2DM, but associated with alexithymia. Our directions for future research recommend that the present findings should be re-investigated in a longitudinal and/or prospective cohort, with a population based sampling strategy. The inclusion of cerebral biomarkers (such as melanocortins, leptin, ghrelin), the concurrent measurement of a dysbalance of the HPA, and systematic testing of emotional dysregulation would be advisable. Controlling for major genetic risk loci, while conducting functional neuroimaging of the hypothalamus would enable attributing cerebral connectivity patterns to specific risks.
Funding A.V.L. acknowledges intramural research funding from the National Medical University, Kiev, Ukraine. O.S.C. was supported by a Ukrainian Ministry of Public Health center grant on “Quality of Life in Diabetes”.
Role of the funding sources The funding sources were not involved in the study design; collection, analysis, or interpretation of data; or in preparation or submission of the manuscript for publication.
References Bagby, R.M., Parker, J.D., Taylor, G.J., 1994a. The twenty-item Toronto Alexithymia Scale-I. Item selection and cross-validation of the factor structure. Journal of Psychosomatic Research 38, 23–32. Bagby, R.M., Taylor, G.J., Parker, J.D., 1994b. The Twenty-item Toronto Alexithymia Scale-II. Convergent, discriminant, and concurrent validity. Journal of Psychosomatic Research 38, 33–40. Batterham, R.L., ffytche, D.H., Rosenthal, J.M., Zelaya, F.O., Barker, G.J., Withers, D.J., Williams, S.C., 2007. PYY modulation of cortical and hypothalamic brain areas predicts feeding behaviour in humans. Nature 450, 106–109. Chatzi, L., Bitsios, P., Solidaki, E., Christou, I., Kyrlaki, E., Sfakianaki, M., Kogevinas, M., Kefalogiannis, N., Pappas, A., 2009. Type 1 diabetes is associated with alexithymia in nondepressed, non-mentally ill diabetic patients: a case-control study. Journal of Psychosomatic Research 67 (4), 307–313. de Timary, P., Roy, E., Luminet, O., Fillee, C., Mikolajczak, M., 2008. Relationship between alexithymia, alexithymia factors and salivary cortisol in men exposed to a social stress test. Psychoneuroendocrinology 33, 1160–1164. Eckel, R.H., Alberti, K.G., Grundy, S.M., Zimmet, P.Z., 2010. The metabolic syndrome. Lancet 375, 181–183. Elfhag, K., Lundh, L.G., 2007. TAS-20 alexithymia in obesity, and its links to personality. Scandinavian Journal of Psychology 48, 391–398. Eriksson, A.K., Gustavsson, J.P., Hilding, A., Granath, F., Ekbom, A., Ostenson, C.G., 2012. Personality traits and abnormal glucose regulation in middle-aged Swedish men and women. Diabetes Research and Clinical Practice 95, 145–152. Fisher, L., Glasgow, R.E., Strycker, L.A., 2010a. The relationship between diabetes distress and clinical depression with glycemic control among patients with type 2 diabetes. Diabetes Care 33, 1034–1036. Fisher, L., Mullan, J.T., Arean, P., Glasgow, R.E., Hessler, D., Masharani, U., 2010b. Diabetes distress but not clinical depression or depressive symptoms is associated with glycemic control in both cross-sectional and longitudinal analyses. Diabetes Care 33, 23–28. Ford, E.S., Giles, W.H., Dietz, W.H., 2002. Prevalence of the metabolic syndrome among US adults: findings from the third National Health and Nutrition Examination Survey. Journal of the American Medical Association 287, 356–359. Gathercole, L.L., Stewart, P.M., 2010. Targeting the pre-receptor metabolism of cortisol as a novel therapy in obesity and diabetes. J. Steroid. Biochem. Mol. Biol. 122 (1-3), 21–27. Ginsberg, H.N., Stalenhoef, A.F., 2003. The metabolic syndrome: targeting dyslipidaemia to reduce coronary risk. Journal of Cardiovascular Risk 10, 121–128. Girman, C.J., Rhodes, T., Mercuri, M., Pyorala, K., Kjekshus, J., Pedersen, T.R., Beere, P. A., Gotto, A.M., Clearfield, M., 2004. The metabolic syndrome and risk of major coronary events in the Scandinavian Simvastatin Survival Study (4S) and the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS). American Journal of Cardiology 93, 136–141. Gois, C., Barbosa, A., Ferro, A., Santos, A.L., Sousa, F., Akiskal, H., Akiskal, K., Figueira, M.L., 2011. The role of affective temperaments in metabolic control in patients with type 2 diabetes. Journal of Affective Disorders 134, 52–58. Grabe, H.J., Schwahn, C., Barnow, S., Spitzer, C., John, U., Freyberger, H.J., Schminke, U., Felix, S., Volzke, H., 2010. Alexithymia, hypertension, and subclinical atherosclerosis in the general population. Journal of Psychosomatic Research 68, 139–147. Helmers, K.F., Mente, A., 1999. Alexithymia and health behaviors in healthy male volunteers. Journal of Psychosomatic Research 47, 635–645.
A.V. Lemche et al. / Psychiatry Research 215 (2014) 438–443
Housiaux, M., Luminet, O., Van Broeck, N., Dorchy, H., 2010. Alexithymia is associated with glycaemic control of children with type 1 diabetes. Diabetes & Metabolism 36, 455–462. Huang, C.F., Cheng, M.L., Fan, C.M., Hong, C.Y., Shiao, M.S., 2013. Nicotinuric acid: a potential marker of metabolic syndrome through a metabolomics-based approach. Diabetes Care 36, 1729–1731. Kauhanen, J., Kaplan, G.A., Cohen, R.D., Salonen, R., Salonen, J.T., 1994. Alexithymia may influence the diagnosis of coronary heart disease. Psychosomatic Medicine 56, 237–244. Lemche, E., Klann-Delius, G., Koch, R., Joraschky, P., 2004. Mentalizing language development in a longitudinal attachment sample: implications for alexithymia. Psychotherapy and Psychosomatics 73, 366–374. Lemche, E., Surguladze, S.A., Giampietro, V.P., Brammer, M.J., Williams, S.C.R., Sierra, M., David, A.S., Phillips, M.L., 2013. Interoceptive–reflective regions differentiate alexithymia traits in depersonalization disorder. Psychiatry Research Neuroimaging 214 (1), 66–72. Luminet, O., de Timary, P., Buysschaert, M., Luts, A., 2006. The role of alexithymia factors in glucose control of persons with type 1 diabetes: a pilot study. Diabetes & Metabolism 32, 417–424. Maggini, C., Raballo, A., 2004. Alexithymia and schizophrenic psychopathology. Acta Bio Medica 75, 40–49. Meunier, J., Dorchy, H., Luminet, O., 2008. Does family cohesiveness and parental alexithymia predict glycaemic control in children and adolescents with diabetes? Diabetes & Metabolism 34, 473–481.
443
Neumann, S.A., Sollers 3rd, J.J., Thayer, J.F., Waldstein, S.R., 2004. Alexithymia predicts attenuated autonomic reactivity, but prolonged recovery to anger recall in young women. International Journal of Psychophysiology 53, 183–195. Puustinen, P.J., Koponen, H., Kautiainen, H., Mantyselka, P., Vanhala, M., 2011. Psychological distress predicts the development of the metabolic syndrome: a prospective population-based study. Psychosomatic Medicine 73, 158–165. Riediger, N.D., Clara, I., 2011. Prevalence of metabolic syndrome in the Canadian adult population. Canadian Medical Association Journal 183, E1127–1134. Sattar, N., Gaw, A., Scherbakova, O., Ford, I., O’Reilly, D.S., Haffner, S.M., Isles, C., Macfarlane, P.W., Packard, C.J., Cobbe, S.M., Shepherd, J., 2003. 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 108, 414–419. Szmukler, G., Everitt, B., Leese, M., 2012. Risk assessment and receiver operating characteristic curves. Psychological Medicine 42, 895–898. Topsever, P., Filiz, T.M., Salman, S., Sengul, A., Sarac, E., Topalli, R., Gorpelioglu, S., Yilmaz, T., 2006. Alexithymia in diabetes mellitus. Scottish Medical Journal 51, 15–20. Wood, R.L., Williams, C., Kalyani, T., 2009. The impact of alexithymia on somatization after traumatic brain injury. Brain Injury 23, 649–654.
