- Email: [email protected]
Serum cholesterol and impulsivity in a large sample of healthy young men
Psychiatry Research 120 (2003) 239–245
Serum cholesterol and impulsivity in a large sample of healthy young men夞 Flavio Pozzia,*, Alfonso Troisib, Marco Cerillia, A.M. Autorec, C. Lo Castroc, D. Ribattid, Gaetano Frajesea a
Department of Internal Medicine, Unit of Endocrinology, University of Tor Vergata, Via di Tor Vergata 135, 00133, Rome, Italy b Department of Neurosciences, University of Rome Tor Vergata, Via di Tor Vergata 135, 00133, Rome, Italy c National Selection and Recruitment Center of the Arma dei Carabinieri, Rome, Italy d General Headquarters of the Arma dei Carabinieri, Rome, Italy
Abstract Studies that have investigated the association between cholesterol levels and impulsivity are relatively few in number and have yielded equivocal results. In this study, we investigated the relationship between impulsivity, depression and serum lipids wtotal cholesterol, high-density lipoprotein (HDL) cholesterol, and triglyceridesx in a large sample (Ns2051) of healthy young men who were remarkably homogeneous in terms of age, educational level, and socioeconomic conditions. Depression was assessed using the depression scale of the Minnesota Multiphasic Personality Inventory-2, and impulsivity was measured using the impulse control scale of the Big Five Questionnaire (BFQ). We found that subjects with a low serum cholesterol, defined as the lowest tenth of the total cholesterol distribution (F3.7 mmolyl), scored significantly lower on the impulse control scale of the BFQ. There was no significant association between depression and cholesterol concentrations. In addition, in a multiple regression model, both lower levels of total cholesterol and higher levels of HDL cholesterol emerged as significant predictors of impulsivity. However, since the regression model accounted for only 0.6% of the variance in the score on the impulse control scale of the BFQ, the biological significance of these correlations was negligible. Taken together, these findings suggest that, in healthy young men, a relationship between cholesterol and impulsivity emerges only when the statistical analysis focuses on subjects with very low levels of cholesterol. 䊚 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Cholesterol; Impulsivity; Depression; Personality; Healthy men
夞 This study is part of a collaborative program between the Department of Internal Medicine, Chair of Endocrinology, University ` of Rome Tor Vergata and ‘Comando Generale dell’Arma dei Carabinieri-Direzione di Sanita’. *Corresponding author. Tel.: q39-0-672-596-522; fax: q39-0-672-596-522. E-mail address: [email protected] (F. Pozzi). 0165-1781/03/$ - see front matter 䊚 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0165-1781Ž03.00192-6
240
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
1. Introduction Recent evidence has indicated a significant relation between low cholesterol levels and a variety of psychological and behavioral variables, including depression (Morgan et al., 1993; Brown et al., 1994; Steegmans et al., 2000; Troisi et al., 2001), anxiety (Suarez, 1999; Troisi et al., 2002), aggression (Golomb, 1998; Hillbrand and Spitz, 1999; Golomb et al., 2000), and suicidal behavior (Gallerini et al., 1995; Zureik et al., 1996; Papassotiropoulos et al., 1999). By comparison, studies that have examined the association between cholesterol levels and impulsivity are few in number and have yielded equivocal results. In a sample of 152 adolescent psychiatric inpatients, Apter et al. (1999) found no correlation between total serum cholesterol and impulsivity. Using the Eysenck and Eysenck Impulsivity Questionnaire, Steegmans et al. (2000) found no differences between the scores of men with chronically low cholesterol levels (F4.5 mmolyl) and the scores of a reference group of men with cholesterol levels between 6 and 7 mmolyl. In a sample of 168 cocaine-dependent patients, Roy et al. (2001) found no significant correlations between total serum cholesterol levels and scores on the Barratt Impulsivity Scale. In contrast with the negative findings of these reports, other studies have found an inverse association between low cholesterol levels and impulsivity. Buydens-Branchey et al. (2000) found a moderate negative correlation (rsy0.24) between high-density lipoprotein (HDL) cholesterol and scores on the Barratt Impulsivity Scale in 38 personality-disordered cocaine addicts. New et al. (1999) found that patients with borderline personality disorder had significantly lower serum cholesterol levels than patients with other non-impulsive personality disorders. Garland et al. (2000) found a significant negative correlation between total serum cholesterol and self-reported scores of impulsivity in 100 consecutive patients who had attempted suicide. Inconsistencies in the results of studies reported to date may reflect various limiting factors. For example, most of the evidence comes from studies of psychiatric patients with heterogeneous diagnoses. It is clear that in these clinical populations
the presence of other symptoms, such as depression and hostility, can confound the relationship between cholesterol and impulsivity. A second limiting factor is the relatively small samples of previous studies, which were likely to include few individuals with naturally occurring low levels of cholesterol. Lastly, the majority of studies have examined the relation of impulsivity to total cholesterol only. Inclusion of additional measures, such as triglycerides and HDL cholesterol, may lead to a better determination of the degree to which impulsivity is associated with low lipid and lipoprotein concentrations. In the present study, we investigated the relationship between impulsivity, depression and serum lipids (total cholesterol, HDL cholesterol, and triglycerides) in a large sample of healthy young men who were remarkably homogeneous in terms of age, educational level, and socioeconomic conditions. 2. Methods 2.1. Subjects The original data base included a consecutive series of 3016 male volunteers in the Italian national recruitment for permanent duty in the ‘Arma dei Carabinieri’, during the period between March and June 1999. Mandatory requirements for application were a high school degree and demonstration of good social and legal conduct. All subjects underwent a complete physical examination, an endocrinological examination, a thorax Xray, an electrocardiogram, and a complete blood and urine analysis, including a toxicological screening for drugs of abuse. During physical examination, subjects’ body mass index (BMI) was recorded. 2.2. Biochemical determinations Venous blood (15 ml) was taken from each subject at 07:30 h, after an overnight fast of at least 8 h, and divided into three samples for determination of hematology (Vacutainer with K3 EDTA), biochemistry and thyroid-stimulating hormone (TSH) (vacutainer with gel and clot activa-
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
tor). Total cholesterol, HDL cholesterol, triglycerides, plasma glucose, urea, creatinine, total and direct bilirubin, and transaminases were measured at 08:30 h, with an automatic Olympus AU 800 for quantitative analysis. Standard urine analysis was conducted with automatic Supertron Roche. Urine screening for cocaine, amphetamine, opiates, cannabis, barbiturates and alcohol was performed with MegayMerck by Dada-Behring. Laboratory standardization was carried out by means of Olympus commercial standards. TSH was analyzed with the hypersensitive human TSH immunoenzymatic chemoluminescent two-sided test, by Access Immunoassay System. The range between functional sensitivity and the highest concentration calibration point was 0.01–100 mUIy ml, with normal range values of 0.34–5.6 mUIy ml. 2.3. Psychometric assessment The subjects who had passed the medical and toxicological screening were administered a battery of psychological tests that included the 370-item version of the Minnesota Multiphasic Personality Inventory-2 (MMPI-2; Hathaway and McKinley, 1995) and the Big Five Questionnaire (BFQ; Caprara et al., 1993). Depression was measured using the depression scale of the MMPI-2, and impulsivity was measured using the impulse control scale of the BFQ (lower scores on this scale reflect greater impulsivity). The BFQ is a 132-item self-report inventory that measures five factors corresponding to the Big Five: Energy, Friendliness, Conscientiousness, Emotional Stability and Openness. Every dimension is organized into two facets, and every facet scale contains 12 items. The Emotional Stability scale includes emotion control and impulse control. The subjects have a five-choice answer scale to reply to each item, ranging from complete disagreement to complete agreement. The BFQ measures the same five constructs measured by the NEO Personality Inventory (Costa and McCrae, 1985). 2.4. Data analysis Of the original sample of 3016 subjects, 900 were excluded because of inconsistencies in the
241
compilation of tests, because they decided to withdraw their consent to participate, or because they did not pass the medical or toxicological screening. In particular, those subjects with TSH above normal range or hyperglycemia were excluded because these conditions are often associated with negative mood (Cleare et al., 1995) and hypercholesterolaemia (Oki, 1995; Michalopoulou et al., 1998). Of the remaining 2116 subjects who were administered the psychometric battery, 65 were excluded because of abnormal psychological profiles as assessed by the MMPI-2. Thus, the final data base included 2051 healthy subjects (mean"S.D. age: 23.28"2.36 years; range: 19– 31 years). This selected sample does not necessarily correspond to enrollment in the ‘Arma dei Carabinieri’, which is actually based on several other non-medical parameters. Statistical analysis included bivariate correlations, multiple regression analysis and analysis of covariance (ANCOVA). 3. Results Table 1 reports the physiological and psychometric data for the 2051 subjects included in the data analysis. To examine the relationship between lipid levels and psychological variables, separate multiple regression analyses were carried out with independent variables including serum levels of triglycerides, total cholesterol (TC), and HDL cholesterol. The serum levels of LDL cholesterol were strongly correlated with total cholesterol levels (rs0.93) and, therefore, were not included among the independent variables. The scores on the impulse control scale of the BFQ and the depression scale of the MMPI-2 were used as the dependent variables. The BMI and the TSH levels were entered into the regression models to control for the confounding effect of the subjects’ dietary habits and thyroid function, since TSH, even if in the upper normal ranges, may be a marker for subclinical hypothyroidism. In the model that included depression as the dependent variable, none of the serum lipids emerged as a significant predictor. The only significant predictor of the depression scale of the
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
242
Table 1 Physiological and psychometric data (mean"S.D.) for the total sample (Ns2051), the lowest tenth of the total cholesterol distribution (Ns206), and the highest tenth of the total cholesterol distribution (Ns211) Variable
Total 2
Body mass index (kgym ) TSH (mUIyml) Total cholesterol (mmolyl) HDL cholesterol (mmolyl) Triglycerides (mmolyl) BFQ-CI MMPI-D
24.70 1.50 4.75 1.25 1.00 34.18 17.94
Lowest (4.10) (0.68) (0.87) (0.29) (0.48) (7.92) (2.18)
24.24 1.49 3.39 1.30 1.38 32.83 17.85
(2.89) (0.71) (0.25) (0.30) (0.61) (7.24) (2.21)
Highest 25.69 1.49 6.42 1.13 0.76 34.76 18.28
(3.23) (0.66) (0.62) (0.26) (0.30) (8.69) (2.06)
Note: BFQ-CI, impulse control scale of the Big Five Questionnaire; MMPI-D, depression scale of the Minnesota Multiphasic Personality Inventory-2.
MMPI-2 was BMI: subjects with a lower BMI reported more depressive symptoms (bsy0.06, Ps0.006; R 2s0.006, Fs2.3, Ps0.04). In the model that included impulse control as the dependent variable, both HDL and total cholesterol levels emerged as significant predictors (R 2s0.006, Fs 2.6, Ps0.02). The subjects with lower scores on the impulse control scale of the BFQ had higher HDL cholesterol levels (bsy0.07, Ps0.006) and lower total cholesterol levels (bs0.06, Ps 0.01). When the analyses were repeated in the subsample (72.5% of the total sample) of subjects with total cholesterol levels within the normal range (F5.18 mmolyl or 200 mgydl), it did not change the results. Even though these results were statistically significant because of the large size of the sample, their biological and clinical significance was negligible, as suggested by the percentage of variance explained (0.6%) and the beta values (ranges0.06–0.07). Because previous studies have reported that the relation between cholesterol and psychological variables may be non-linear, TC was subsequently investigated as a dichotomized variable. The lowest tenth of the cholesterol distribution was selected as the low-cholesterol group, a cut-off point that has been used in previous studies (Rosengren et al., 1996; Horsten et al., 1997). In the present study population, the mean TC was 4.75 mmolyl and the low-cholesterol group (Ns206) had TC concentrations of 3.7 mmolyl or less. An ANCOVA model controlling for the BMI and TSH levels showed that subjects in the low-cholesterol group, compared with subjects with higher TC concentra-
tions, had significant lower scores on the impulse control scale of the BFQ (Fs6.90, d.f.s1, 2043, Ps0.009). The two groups did not differ on the depression scale of the MMPI-2 (Fs0.53, d.f.s 1, 2043, Ps0.47). 4. Discussion In this large sample of healthy young men, we found that the subjects with a low serum cholesterol, defined as the lowest tenth of the total cholesterol distribution (F3.7 mmolyl), scored significantly lower on the impulse control scale of the BFQ. In addition, in a multiple regression model, both lower levels of TC and higher levels of HDL cholesterol emerged as significant predictors of impulsivity. However, since the regression model accounted for only 0.6% of the variance in scores on the impulse control scale of the BFQ, the biological significance of these correlations was negligible. Taken together, these findings suggest that, in healthy young men, a relationship between cholesterol and impulsivity emerges only when the statistical analysis focuses on subjects with very low levels of cholesterol (Table 2). The clinical relevance of these findings should be evaluated in the light of two considerations. First, impulsivity is a complex construct, and research over the last 20 years has clearly shown that the causes of impulsivity are multiple in nature (Evenden, 1999). Therefore, any single predictor is likely to account for only a relatively small proportion of the variance in impulsivity. Second, highly selected healthy subjects, like those enrolled
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
243
Table 2 Univariate correlations between serum lipids and psychometric variables (impulsivity and depression) in the total sample Total cholesterol Total cholesterol HDL cholesterol LDL cholesterol Triglycerides BFQ-CI MMPI-D
rs0.16 rs0.93 rs0.38 rs0.04 rs0.03
HDL cholesterol
LDL cholesterol
Triglycerides
BFQ-CI
MMPI-D
rs0.16
rs0.93 rsy0.12
rs0.38 rsy0.30 rs0.25
rs0.04 rsy0.05 rs0.06 rs0.01
rs0.03 rs0.01 rs0.02 rs0.03 rs0.01
rsy0.12 rsy0.30 rsy0.05 rs0.01
rs0.25 rs0.06 rs0.02
rs0.01 rs0.03
rs0.01
Note: BFQ-CI, impulse control scale of the Big Five Questionnaire; MMPI-D, depression scale of the Minnesota Multiphasic Personality Inventory-2.
in this study, are likely not to be sufficiently extreme in impulsivity to allow the detection of strong linear correlations with cholesterol levels. Another finding of this study was that of the two psychological variables used as dependent variables, only impulsivity proved to be related to cholesterol levels. A possible explanation is that interindividual variation in depression is lower in a population of young healthy males than interindividual variation in impulse control. As a consequence, impulsivity, but not depression, emerged as a significant psychological correlate of cholesterol levels. Regardless of the factors that determined the different relations between cholesterol levels and impulsivity and depression, our results confirm the importance of measuring multiple constructs in studies exploring the relationship between cholesterol and psychological profile. We found that the lipid fractions related to impulse control were both TC and HDL cholesterol. This finding is relevant to the question of what lipid fractions are more strongly correlated with psychological and behavioral variables. The results of some studies suggest that the most important lipid fraction is HDL cholesterol. In healthy women, Horsten et al. (1997) found a negative linear association between depressive symptoms and HDL cholesterol but not between depressive symptoms and TC. In their study of depressed patients and control subjects, Maes et al. (1999) found that the difference between the two groups in HDL cholesterol was more significant than the difference in total cholesterol. In the study of BuydensBranchey et al. (2000), the lipid fraction correlated with impulsivity was HDL cholesterol, and the
patients who had a past history of aggression had significantly lower HDL cholesterol. In contrast, other studies found that the lipid parameter most strongly correlated with psychological variables was TC or the LDL fraction. Lindberg et al. (1994) found that TC and low-density lipid (LDL) cholesterol values were lower in those men who had experienced low mood during the past month compared with those who had not. Muldoon et al. (1997) found that better performance on a measure of fluid intelligence was correlated with higher total and LDL cholesterol concentrations. In a study of new mothers by Troisi et al. (2002), TC emerged as a stronger correlate of postpartum mood symptoms than HDL cholesterol. These conflicting results are difficult to explain and reflect our limited understanding of the neurobiological actions of lipids and lipoproteins. In a seminal article, Engelberg (1992) was the first to hypothesize that a lowered plasma cholesterol concentration might induce a decrease in brain cell membrane cholesterol that ultimately would result in a reduced serotonergic function. Diebold et al. (1998) proposed a model in which a decrease in plasma total cholesterol or LDL cholesterol would induce a relative increase in brain cell membrane fluidity, with increased presynaptic serotonin reuptake and decreased postsynaptic serotonin function. In contrast to this model, Buydens-Branchey et al. (2000) found that the lipid fraction associated with neuroendocrine indices of reduced serotonin function was low HDL cholesterol, not TC or LDL cholesterol. Finally, Hibbeln et al. (1998) observed a strong correlation between brain serotonin metabolites and essential fatty acids (EFAs), a finding
244
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
suggesting that cholesterol, in previous studies, was acting as a ‘surrogate marker’ for EFAs (Hibbeln et al., 2000). It is clear that a better understanding of the neurobiological actions of lipids and lipoproteins will be a precondition to answering the question of whether it is total cholesterol, one of its subfractions, or other lipids that play the most important role in psychological and behavioral problems. In conclusion, this study gives further support to existing evidence indicating an association between low cholesterol and impulsivity. However, it also indicates that, in healthy young men, such an association is much less evident than in a selected clinical sample of psychiatric patients with impulse control disorders. References Apter, A., Laufer, N., Bar-Sever, M., Har-Even, D., Ofek, H., Weizman, A., 1999. Serum cholesterol, suicidal tendencies, impulsivity, aggression, and depression in adolescent psychiatric inpatients. Biological Psychiatry 46, 532–541. Brown, S.L., Salive, M.E., Harris, T.B., Simonsick, E.M., Guralnik, J.M., Kohout, F.J., 1994. Low cholesterol concentrations and severe depressive symptoms in elderly people. British Medical Journal 308, 1328–1332. Buydens-Branchey, L., Branchey, M., Hudson, J., Fergeson, P., 2000. Low HDL cholesterol, aggression and altered central serotonergic activity. Psychiatry Research 93, 93–102. Caprara, G.V., Barbaranelli, C., Borgogni, L., Perugini, M., 1993. The ‘Big Five Questionnaire’: a new questionnaire to assess the five factor model. Personality and Individual Differences 15, 281–288. Cleare, A.J., McGregor, A., O’Keane, V., 1995. Neuroendocrine evidence for an association between hypothyroidism, reduced central 5-HT activity and depression. Clinical Endocrinology 43, 713–719. Costa, P.T., McCrae, R.R., 1985. The NEO Personality Inventory Manual. Psychological Assessment Resources, Odessa, FL. Diebold, K., Michel, G., Schweizer, J., Diebold-Dorsam, M., Fiehn, W., Kohl, B., 1998. Are psychoactive-drug-induced changes in plasma lipid and lipoprotein levels of significance for clinical remission in psychiatric disorders? Pharmacopsychiatry 31, 60–67. Engelberg, H., 1992. Low serum cholesterol and suicide. Lancet 339, 727–729. Evenden, J.L., 1999. Varieties of impulsivity. Psychopharmacology 146, 348–361. Gallerini, M., Manfredini, R., Caracciolo, S., Scapoli, C., Molinari, S., Fersini, C., 1995. Serum cholesterol concentrations in parasuicide. British Medical Journal 310, 1632–1636.
Garland, M., Hickey, D., Corvin, A., Golden, J., Fitzpatrick, P., Cunningham, S., Walsh, N., 2000. Total serum cholesterol in relation to psychological correlates in parasuicide. British Journal of Psychiatry 177, 77–83. Golomb, B.A., 1998. Cholesterol and violence: is there a connection? Annals of Internal Medicine 128, 478–487. Golomb, B.A., Stattin, H., Mednick, S., 2000. Low cholesterol and violent crime. Journal of Psychiatric Research 34, 301–309. Hathaway, S.R., McKinley, J.C., 1995. MMPI-2. Multiphasic Personality Inventory—2, Handbook. Organizzazioni Speciali, Firenze. Hibbeln, J.R., Umhau, J.C., George, D.T., Shoaf, S.E., Linnoila, M., George, D.T., Salem Jr, N., 2000. Plasma total cholesterol concentrations do not predict cerebrospinal fluid neurotransmitter metabolites: implications for the biophysical role of highly unsaturated fatty acids. American Journal of Clinical Nutrition 71, 331S–338S. Hibbeln, J.R., Umhau, J.C., Linnoila, M., George, D.T., Ragan, P.W., Shoaf, S.E., Vaughan, M.R., Rawlings, R., Salem Jr, N., 1998. A replication study of violent and non-violent subjects: cerebrospinal fluid metabolites of serotonin and dopamine are predicted by plasma essential fatty acids. Biological Psychiatry 44, 243–249. Hillbrand, M., Spitz, R.T., 1999. Cholesterol and aggression. Aggression and Violent Behavior 4, 359–370. Horsten, M., Wamala, S.P., Vingerhoets, A., Orth-Gomer, K., 1997. Depressive symptoms, social support, and lipid profile in healthy middle-aged women. Psychosomatic Medicine 59, 521–528. Lindberg, G., Larsson, G., Setterlind, S., Rastam, L., 1994. Serum lipids and mood in working men and women in Sweden. Journal of Epidemiology and Community Health 48, 360–363. Maes, M., Christophe, A., Delanghe, J., Altamura, C., Neels, H., Meltzer, H.Y., 1999. Lowered omega-3 polyunsaturated fatty acids in serum phospholipids and cholesteryl esters of depressed patients. Psychiatry Research 85, 275–291. Michalopoulou, G., Alevizaki, M., Piperingos, G., Mitsibounas, D., Mantzos, E., Adamopoulos, P., Koutras, D.A., 1998. High serum cholesterol levels in persons with ‘high normal’ TSH levels: should one extend the definition of subclinical hypothyroidism? European Journal of Endocrinology 138, 141–145. Morgan, R.E., Palinkas, L.A., Barrett Connor, E.L., Wingard, D.L., 1993. Plasma cholesterol and depressive symptoms in older men. Lancet 341, 75–79. Muldoon, M.F., Ryan, C.M., Matthews, K.A., Manuck, S.B., 1997. Serum cholesterol and intellectual performance. Psychosomatic Medicine 59, 382–387. New, A.S., Sevin, E.M., Mitropoulou, V., Reynolds, D., Novotny, S.L., Callahan, A., Trestman, R.L., Siever, L.J., 1999. Serum cholesterol and impulsivity in personality disorders. Psychiatry Research 85, 145–150. Oki, J.C., 1995. Dyslipidemias in patients with diabetes mellitus: classification and risks and benefits of therapy. Pharmacotherapy 15 (3), 317–337.
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245 Papassotiropoulos, A., Hawellek, B., Frahnert, C., Rao, G.S., Rao, M.L., 1999. The risk of acute suicidality in psychiatric inpatients increases with low plasma cholesterol. Pharmacopsychiatry 32, 1–4. Rosengren, A., Orth-Gomer, K., Wedel, H., Wilhelmsen, L., 1996. Low serum cholesterol, social support and predictors of mortality in middle aged men: a study of men born in 1933. Cardiovascular Risk Factors 6, 345–353. Roy, A., Gonzales, B., Marcus, A., Berman, J., 2001. Serum cholesterol, suicidal behavior and impulsivity in cocainedependent patients. Psychiatry Research 101, 243–247. Steegmans, P.H.A., Hoes, A.W., Bak, A.A.A., van der Does, E., Grobbee, D.E., 2000. Higher prevalence of depressive symptoms in middle-aged men with low serum cholesterol levels. Psychosomatic Medicine 62, 205–211.
245
Suarez, E.C., 1999. Relations of trait depression and anxiety to low lipid and lipoprotein concentrations in healthy young adult women. Psychosomatic Medicine 61, 273–279. Troisi, A., Moles, A., Panepuccia, L., Lo Russo, D., Palla, G., Scucchi, S., 2002. Serum cholesterol levels and mood symptoms in the postpartum period. Psychiatry Research 109, 213–219. Troisi, A., Scucchi, S., San Martino, L., Montera, P., d’Amore, A., Moles, A., 2001. Age specificity of the relationship between serum cholesterol and mood in obese women. Physiology and Behavior 72, 409–413. Zureik, M., Courbon, D., Ducimetiere, P., 1996. Serum cholesterol concentration and death from suicide in men: Paris prospective study I. British Medical Journal 313, 649–652.
Serum cholesterol and impulsivity in a large sample of healthy young men夞 Flavio Pozzia,*, Alfonso Troisib, Marco Cerillia, A.M. Autorec, C. Lo Castroc, D. Ribattid, Gaetano Frajesea a
Department of Internal Medicine, Unit of Endocrinology, University of Tor Vergata, Via di Tor Vergata 135, 00133, Rome, Italy b Department of Neurosciences, University of Rome Tor Vergata, Via di Tor Vergata 135, 00133, Rome, Italy c National Selection and Recruitment Center of the Arma dei Carabinieri, Rome, Italy d General Headquarters of the Arma dei Carabinieri, Rome, Italy
Abstract Studies that have investigated the association between cholesterol levels and impulsivity are relatively few in number and have yielded equivocal results. In this study, we investigated the relationship between impulsivity, depression and serum lipids wtotal cholesterol, high-density lipoprotein (HDL) cholesterol, and triglyceridesx in a large sample (Ns2051) of healthy young men who were remarkably homogeneous in terms of age, educational level, and socioeconomic conditions. Depression was assessed using the depression scale of the Minnesota Multiphasic Personality Inventory-2, and impulsivity was measured using the impulse control scale of the Big Five Questionnaire (BFQ). We found that subjects with a low serum cholesterol, defined as the lowest tenth of the total cholesterol distribution (F3.7 mmolyl), scored significantly lower on the impulse control scale of the BFQ. There was no significant association between depression and cholesterol concentrations. In addition, in a multiple regression model, both lower levels of total cholesterol and higher levels of HDL cholesterol emerged as significant predictors of impulsivity. However, since the regression model accounted for only 0.6% of the variance in the score on the impulse control scale of the BFQ, the biological significance of these correlations was negligible. Taken together, these findings suggest that, in healthy young men, a relationship between cholesterol and impulsivity emerges only when the statistical analysis focuses on subjects with very low levels of cholesterol. 䊚 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Cholesterol; Impulsivity; Depression; Personality; Healthy men
夞 This study is part of a collaborative program between the Department of Internal Medicine, Chair of Endocrinology, University ` of Rome Tor Vergata and ‘Comando Generale dell’Arma dei Carabinieri-Direzione di Sanita’. *Corresponding author. Tel.: q39-0-672-596-522; fax: q39-0-672-596-522. E-mail address: [email protected] (F. Pozzi). 0165-1781/03/$ - see front matter 䊚 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0165-1781Ž03.00192-6
240
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
1. Introduction Recent evidence has indicated a significant relation between low cholesterol levels and a variety of psychological and behavioral variables, including depression (Morgan et al., 1993; Brown et al., 1994; Steegmans et al., 2000; Troisi et al., 2001), anxiety (Suarez, 1999; Troisi et al., 2002), aggression (Golomb, 1998; Hillbrand and Spitz, 1999; Golomb et al., 2000), and suicidal behavior (Gallerini et al., 1995; Zureik et al., 1996; Papassotiropoulos et al., 1999). By comparison, studies that have examined the association between cholesterol levels and impulsivity are few in number and have yielded equivocal results. In a sample of 152 adolescent psychiatric inpatients, Apter et al. (1999) found no correlation between total serum cholesterol and impulsivity. Using the Eysenck and Eysenck Impulsivity Questionnaire, Steegmans et al. (2000) found no differences between the scores of men with chronically low cholesterol levels (F4.5 mmolyl) and the scores of a reference group of men with cholesterol levels between 6 and 7 mmolyl. In a sample of 168 cocaine-dependent patients, Roy et al. (2001) found no significant correlations between total serum cholesterol levels and scores on the Barratt Impulsivity Scale. In contrast with the negative findings of these reports, other studies have found an inverse association between low cholesterol levels and impulsivity. Buydens-Branchey et al. (2000) found a moderate negative correlation (rsy0.24) between high-density lipoprotein (HDL) cholesterol and scores on the Barratt Impulsivity Scale in 38 personality-disordered cocaine addicts. New et al. (1999) found that patients with borderline personality disorder had significantly lower serum cholesterol levels than patients with other non-impulsive personality disorders. Garland et al. (2000) found a significant negative correlation between total serum cholesterol and self-reported scores of impulsivity in 100 consecutive patients who had attempted suicide. Inconsistencies in the results of studies reported to date may reflect various limiting factors. For example, most of the evidence comes from studies of psychiatric patients with heterogeneous diagnoses. It is clear that in these clinical populations
the presence of other symptoms, such as depression and hostility, can confound the relationship between cholesterol and impulsivity. A second limiting factor is the relatively small samples of previous studies, which were likely to include few individuals with naturally occurring low levels of cholesterol. Lastly, the majority of studies have examined the relation of impulsivity to total cholesterol only. Inclusion of additional measures, such as triglycerides and HDL cholesterol, may lead to a better determination of the degree to which impulsivity is associated with low lipid and lipoprotein concentrations. In the present study, we investigated the relationship between impulsivity, depression and serum lipids (total cholesterol, HDL cholesterol, and triglycerides) in a large sample of healthy young men who were remarkably homogeneous in terms of age, educational level, and socioeconomic conditions. 2. Methods 2.1. Subjects The original data base included a consecutive series of 3016 male volunteers in the Italian national recruitment for permanent duty in the ‘Arma dei Carabinieri’, during the period between March and June 1999. Mandatory requirements for application were a high school degree and demonstration of good social and legal conduct. All subjects underwent a complete physical examination, an endocrinological examination, a thorax Xray, an electrocardiogram, and a complete blood and urine analysis, including a toxicological screening for drugs of abuse. During physical examination, subjects’ body mass index (BMI) was recorded. 2.2. Biochemical determinations Venous blood (15 ml) was taken from each subject at 07:30 h, after an overnight fast of at least 8 h, and divided into three samples for determination of hematology (Vacutainer with K3 EDTA), biochemistry and thyroid-stimulating hormone (TSH) (vacutainer with gel and clot activa-
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
tor). Total cholesterol, HDL cholesterol, triglycerides, plasma glucose, urea, creatinine, total and direct bilirubin, and transaminases were measured at 08:30 h, with an automatic Olympus AU 800 for quantitative analysis. Standard urine analysis was conducted with automatic Supertron Roche. Urine screening for cocaine, amphetamine, opiates, cannabis, barbiturates and alcohol was performed with MegayMerck by Dada-Behring. Laboratory standardization was carried out by means of Olympus commercial standards. TSH was analyzed with the hypersensitive human TSH immunoenzymatic chemoluminescent two-sided test, by Access Immunoassay System. The range between functional sensitivity and the highest concentration calibration point was 0.01–100 mUIy ml, with normal range values of 0.34–5.6 mUIy ml. 2.3. Psychometric assessment The subjects who had passed the medical and toxicological screening were administered a battery of psychological tests that included the 370-item version of the Minnesota Multiphasic Personality Inventory-2 (MMPI-2; Hathaway and McKinley, 1995) and the Big Five Questionnaire (BFQ; Caprara et al., 1993). Depression was measured using the depression scale of the MMPI-2, and impulsivity was measured using the impulse control scale of the BFQ (lower scores on this scale reflect greater impulsivity). The BFQ is a 132-item self-report inventory that measures five factors corresponding to the Big Five: Energy, Friendliness, Conscientiousness, Emotional Stability and Openness. Every dimension is organized into two facets, and every facet scale contains 12 items. The Emotional Stability scale includes emotion control and impulse control. The subjects have a five-choice answer scale to reply to each item, ranging from complete disagreement to complete agreement. The BFQ measures the same five constructs measured by the NEO Personality Inventory (Costa and McCrae, 1985). 2.4. Data analysis Of the original sample of 3016 subjects, 900 were excluded because of inconsistencies in the
241
compilation of tests, because they decided to withdraw their consent to participate, or because they did not pass the medical or toxicological screening. In particular, those subjects with TSH above normal range or hyperglycemia were excluded because these conditions are often associated with negative mood (Cleare et al., 1995) and hypercholesterolaemia (Oki, 1995; Michalopoulou et al., 1998). Of the remaining 2116 subjects who were administered the psychometric battery, 65 were excluded because of abnormal psychological profiles as assessed by the MMPI-2. Thus, the final data base included 2051 healthy subjects (mean"S.D. age: 23.28"2.36 years; range: 19– 31 years). This selected sample does not necessarily correspond to enrollment in the ‘Arma dei Carabinieri’, which is actually based on several other non-medical parameters. Statistical analysis included bivariate correlations, multiple regression analysis and analysis of covariance (ANCOVA). 3. Results Table 1 reports the physiological and psychometric data for the 2051 subjects included in the data analysis. To examine the relationship between lipid levels and psychological variables, separate multiple regression analyses were carried out with independent variables including serum levels of triglycerides, total cholesterol (TC), and HDL cholesterol. The serum levels of LDL cholesterol were strongly correlated with total cholesterol levels (rs0.93) and, therefore, were not included among the independent variables. The scores on the impulse control scale of the BFQ and the depression scale of the MMPI-2 were used as the dependent variables. The BMI and the TSH levels were entered into the regression models to control for the confounding effect of the subjects’ dietary habits and thyroid function, since TSH, even if in the upper normal ranges, may be a marker for subclinical hypothyroidism. In the model that included depression as the dependent variable, none of the serum lipids emerged as a significant predictor. The only significant predictor of the depression scale of the
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
242
Table 1 Physiological and psychometric data (mean"S.D.) for the total sample (Ns2051), the lowest tenth of the total cholesterol distribution (Ns206), and the highest tenth of the total cholesterol distribution (Ns211) Variable
Total 2
Body mass index (kgym ) TSH (mUIyml) Total cholesterol (mmolyl) HDL cholesterol (mmolyl) Triglycerides (mmolyl) BFQ-CI MMPI-D
24.70 1.50 4.75 1.25 1.00 34.18 17.94
Lowest (4.10) (0.68) (0.87) (0.29) (0.48) (7.92) (2.18)
24.24 1.49 3.39 1.30 1.38 32.83 17.85
(2.89) (0.71) (0.25) (0.30) (0.61) (7.24) (2.21)
Highest 25.69 1.49 6.42 1.13 0.76 34.76 18.28
(3.23) (0.66) (0.62) (0.26) (0.30) (8.69) (2.06)
Note: BFQ-CI, impulse control scale of the Big Five Questionnaire; MMPI-D, depression scale of the Minnesota Multiphasic Personality Inventory-2.
MMPI-2 was BMI: subjects with a lower BMI reported more depressive symptoms (bsy0.06, Ps0.006; R 2s0.006, Fs2.3, Ps0.04). In the model that included impulse control as the dependent variable, both HDL and total cholesterol levels emerged as significant predictors (R 2s0.006, Fs 2.6, Ps0.02). The subjects with lower scores on the impulse control scale of the BFQ had higher HDL cholesterol levels (bsy0.07, Ps0.006) and lower total cholesterol levels (bs0.06, Ps 0.01). When the analyses were repeated in the subsample (72.5% of the total sample) of subjects with total cholesterol levels within the normal range (F5.18 mmolyl or 200 mgydl), it did not change the results. Even though these results were statistically significant because of the large size of the sample, their biological and clinical significance was negligible, as suggested by the percentage of variance explained (0.6%) and the beta values (ranges0.06–0.07). Because previous studies have reported that the relation between cholesterol and psychological variables may be non-linear, TC was subsequently investigated as a dichotomized variable. The lowest tenth of the cholesterol distribution was selected as the low-cholesterol group, a cut-off point that has been used in previous studies (Rosengren et al., 1996; Horsten et al., 1997). In the present study population, the mean TC was 4.75 mmolyl and the low-cholesterol group (Ns206) had TC concentrations of 3.7 mmolyl or less. An ANCOVA model controlling for the BMI and TSH levels showed that subjects in the low-cholesterol group, compared with subjects with higher TC concentra-
tions, had significant lower scores on the impulse control scale of the BFQ (Fs6.90, d.f.s1, 2043, Ps0.009). The two groups did not differ on the depression scale of the MMPI-2 (Fs0.53, d.f.s 1, 2043, Ps0.47). 4. Discussion In this large sample of healthy young men, we found that the subjects with a low serum cholesterol, defined as the lowest tenth of the total cholesterol distribution (F3.7 mmolyl), scored significantly lower on the impulse control scale of the BFQ. In addition, in a multiple regression model, both lower levels of TC and higher levels of HDL cholesterol emerged as significant predictors of impulsivity. However, since the regression model accounted for only 0.6% of the variance in scores on the impulse control scale of the BFQ, the biological significance of these correlations was negligible. Taken together, these findings suggest that, in healthy young men, a relationship between cholesterol and impulsivity emerges only when the statistical analysis focuses on subjects with very low levels of cholesterol (Table 2). The clinical relevance of these findings should be evaluated in the light of two considerations. First, impulsivity is a complex construct, and research over the last 20 years has clearly shown that the causes of impulsivity are multiple in nature (Evenden, 1999). Therefore, any single predictor is likely to account for only a relatively small proportion of the variance in impulsivity. Second, highly selected healthy subjects, like those enrolled
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
243
Table 2 Univariate correlations between serum lipids and psychometric variables (impulsivity and depression) in the total sample Total cholesterol Total cholesterol HDL cholesterol LDL cholesterol Triglycerides BFQ-CI MMPI-D
rs0.16 rs0.93 rs0.38 rs0.04 rs0.03
HDL cholesterol
LDL cholesterol
Triglycerides
BFQ-CI
MMPI-D
rs0.16
rs0.93 rsy0.12
rs0.38 rsy0.30 rs0.25
rs0.04 rsy0.05 rs0.06 rs0.01
rs0.03 rs0.01 rs0.02 rs0.03 rs0.01
rsy0.12 rsy0.30 rsy0.05 rs0.01
rs0.25 rs0.06 rs0.02
rs0.01 rs0.03
rs0.01
Note: BFQ-CI, impulse control scale of the Big Five Questionnaire; MMPI-D, depression scale of the Minnesota Multiphasic Personality Inventory-2.
in this study, are likely not to be sufficiently extreme in impulsivity to allow the detection of strong linear correlations with cholesterol levels. Another finding of this study was that of the two psychological variables used as dependent variables, only impulsivity proved to be related to cholesterol levels. A possible explanation is that interindividual variation in depression is lower in a population of young healthy males than interindividual variation in impulse control. As a consequence, impulsivity, but not depression, emerged as a significant psychological correlate of cholesterol levels. Regardless of the factors that determined the different relations between cholesterol levels and impulsivity and depression, our results confirm the importance of measuring multiple constructs in studies exploring the relationship between cholesterol and psychological profile. We found that the lipid fractions related to impulse control were both TC and HDL cholesterol. This finding is relevant to the question of what lipid fractions are more strongly correlated with psychological and behavioral variables. The results of some studies suggest that the most important lipid fraction is HDL cholesterol. In healthy women, Horsten et al. (1997) found a negative linear association between depressive symptoms and HDL cholesterol but not between depressive symptoms and TC. In their study of depressed patients and control subjects, Maes et al. (1999) found that the difference between the two groups in HDL cholesterol was more significant than the difference in total cholesterol. In the study of BuydensBranchey et al. (2000), the lipid fraction correlated with impulsivity was HDL cholesterol, and the
patients who had a past history of aggression had significantly lower HDL cholesterol. In contrast, other studies found that the lipid parameter most strongly correlated with psychological variables was TC or the LDL fraction. Lindberg et al. (1994) found that TC and low-density lipid (LDL) cholesterol values were lower in those men who had experienced low mood during the past month compared with those who had not. Muldoon et al. (1997) found that better performance on a measure of fluid intelligence was correlated with higher total and LDL cholesterol concentrations. In a study of new mothers by Troisi et al. (2002), TC emerged as a stronger correlate of postpartum mood symptoms than HDL cholesterol. These conflicting results are difficult to explain and reflect our limited understanding of the neurobiological actions of lipids and lipoproteins. In a seminal article, Engelberg (1992) was the first to hypothesize that a lowered plasma cholesterol concentration might induce a decrease in brain cell membrane cholesterol that ultimately would result in a reduced serotonergic function. Diebold et al. (1998) proposed a model in which a decrease in plasma total cholesterol or LDL cholesterol would induce a relative increase in brain cell membrane fluidity, with increased presynaptic serotonin reuptake and decreased postsynaptic serotonin function. In contrast to this model, Buydens-Branchey et al. (2000) found that the lipid fraction associated with neuroendocrine indices of reduced serotonin function was low HDL cholesterol, not TC or LDL cholesterol. Finally, Hibbeln et al. (1998) observed a strong correlation between brain serotonin metabolites and essential fatty acids (EFAs), a finding
244
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245
suggesting that cholesterol, in previous studies, was acting as a ‘surrogate marker’ for EFAs (Hibbeln et al., 2000). It is clear that a better understanding of the neurobiological actions of lipids and lipoproteins will be a precondition to answering the question of whether it is total cholesterol, one of its subfractions, or other lipids that play the most important role in psychological and behavioral problems. In conclusion, this study gives further support to existing evidence indicating an association between low cholesterol and impulsivity. However, it also indicates that, in healthy young men, such an association is much less evident than in a selected clinical sample of psychiatric patients with impulse control disorders. References Apter, A., Laufer, N., Bar-Sever, M., Har-Even, D., Ofek, H., Weizman, A., 1999. Serum cholesterol, suicidal tendencies, impulsivity, aggression, and depression in adolescent psychiatric inpatients. Biological Psychiatry 46, 532–541. Brown, S.L., Salive, M.E., Harris, T.B., Simonsick, E.M., Guralnik, J.M., Kohout, F.J., 1994. Low cholesterol concentrations and severe depressive symptoms in elderly people. British Medical Journal 308, 1328–1332. Buydens-Branchey, L., Branchey, M., Hudson, J., Fergeson, P., 2000. Low HDL cholesterol, aggression and altered central serotonergic activity. Psychiatry Research 93, 93–102. Caprara, G.V., Barbaranelli, C., Borgogni, L., Perugini, M., 1993. The ‘Big Five Questionnaire’: a new questionnaire to assess the five factor model. Personality and Individual Differences 15, 281–288. Cleare, A.J., McGregor, A., O’Keane, V., 1995. Neuroendocrine evidence for an association between hypothyroidism, reduced central 5-HT activity and depression. Clinical Endocrinology 43, 713–719. Costa, P.T., McCrae, R.R., 1985. The NEO Personality Inventory Manual. Psychological Assessment Resources, Odessa, FL. Diebold, K., Michel, G., Schweizer, J., Diebold-Dorsam, M., Fiehn, W., Kohl, B., 1998. Are psychoactive-drug-induced changes in plasma lipid and lipoprotein levels of significance for clinical remission in psychiatric disorders? Pharmacopsychiatry 31, 60–67. Engelberg, H., 1992. Low serum cholesterol and suicide. Lancet 339, 727–729. Evenden, J.L., 1999. Varieties of impulsivity. Psychopharmacology 146, 348–361. Gallerini, M., Manfredini, R., Caracciolo, S., Scapoli, C., Molinari, S., Fersini, C., 1995. Serum cholesterol concentrations in parasuicide. British Medical Journal 310, 1632–1636.
Garland, M., Hickey, D., Corvin, A., Golden, J., Fitzpatrick, P., Cunningham, S., Walsh, N., 2000. Total serum cholesterol in relation to psychological correlates in parasuicide. British Journal of Psychiatry 177, 77–83. Golomb, B.A., 1998. Cholesterol and violence: is there a connection? Annals of Internal Medicine 128, 478–487. Golomb, B.A., Stattin, H., Mednick, S., 2000. Low cholesterol and violent crime. Journal of Psychiatric Research 34, 301–309. Hathaway, S.R., McKinley, J.C., 1995. MMPI-2. Multiphasic Personality Inventory—2, Handbook. Organizzazioni Speciali, Firenze. Hibbeln, J.R., Umhau, J.C., George, D.T., Shoaf, S.E., Linnoila, M., George, D.T., Salem Jr, N., 2000. Plasma total cholesterol concentrations do not predict cerebrospinal fluid neurotransmitter metabolites: implications for the biophysical role of highly unsaturated fatty acids. American Journal of Clinical Nutrition 71, 331S–338S. Hibbeln, J.R., Umhau, J.C., Linnoila, M., George, D.T., Ragan, P.W., Shoaf, S.E., Vaughan, M.R., Rawlings, R., Salem Jr, N., 1998. A replication study of violent and non-violent subjects: cerebrospinal fluid metabolites of serotonin and dopamine are predicted by plasma essential fatty acids. Biological Psychiatry 44, 243–249. Hillbrand, M., Spitz, R.T., 1999. Cholesterol and aggression. Aggression and Violent Behavior 4, 359–370. Horsten, M., Wamala, S.P., Vingerhoets, A., Orth-Gomer, K., 1997. Depressive symptoms, social support, and lipid profile in healthy middle-aged women. Psychosomatic Medicine 59, 521–528. Lindberg, G., Larsson, G., Setterlind, S., Rastam, L., 1994. Serum lipids and mood in working men and women in Sweden. Journal of Epidemiology and Community Health 48, 360–363. Maes, M., Christophe, A., Delanghe, J., Altamura, C., Neels, H., Meltzer, H.Y., 1999. Lowered omega-3 polyunsaturated fatty acids in serum phospholipids and cholesteryl esters of depressed patients. Psychiatry Research 85, 275–291. Michalopoulou, G., Alevizaki, M., Piperingos, G., Mitsibounas, D., Mantzos, E., Adamopoulos, P., Koutras, D.A., 1998. High serum cholesterol levels in persons with ‘high normal’ TSH levels: should one extend the definition of subclinical hypothyroidism? European Journal of Endocrinology 138, 141–145. Morgan, R.E., Palinkas, L.A., Barrett Connor, E.L., Wingard, D.L., 1993. Plasma cholesterol and depressive symptoms in older men. Lancet 341, 75–79. Muldoon, M.F., Ryan, C.M., Matthews, K.A., Manuck, S.B., 1997. Serum cholesterol and intellectual performance. Psychosomatic Medicine 59, 382–387. New, A.S., Sevin, E.M., Mitropoulou, V., Reynolds, D., Novotny, S.L., Callahan, A., Trestman, R.L., Siever, L.J., 1999. Serum cholesterol and impulsivity in personality disorders. Psychiatry Research 85, 145–150. Oki, J.C., 1995. Dyslipidemias in patients with diabetes mellitus: classification and risks and benefits of therapy. Pharmacotherapy 15 (3), 317–337.
F. Pozzi et al. / Psychiatry Research 120 (2003) 239–245 Papassotiropoulos, A., Hawellek, B., Frahnert, C., Rao, G.S., Rao, M.L., 1999. The risk of acute suicidality in psychiatric inpatients increases with low plasma cholesterol. Pharmacopsychiatry 32, 1–4. Rosengren, A., Orth-Gomer, K., Wedel, H., Wilhelmsen, L., 1996. Low serum cholesterol, social support and predictors of mortality in middle aged men: a study of men born in 1933. Cardiovascular Risk Factors 6, 345–353. Roy, A., Gonzales, B., Marcus, A., Berman, J., 2001. Serum cholesterol, suicidal behavior and impulsivity in cocainedependent patients. Psychiatry Research 101, 243–247. Steegmans, P.H.A., Hoes, A.W., Bak, A.A.A., van der Does, E., Grobbee, D.E., 2000. Higher prevalence of depressive symptoms in middle-aged men with low serum cholesterol levels. Psychosomatic Medicine 62, 205–211.
245
Suarez, E.C., 1999. Relations of trait depression and anxiety to low lipid and lipoprotein concentrations in healthy young adult women. Psychosomatic Medicine 61, 273–279. Troisi, A., Moles, A., Panepuccia, L., Lo Russo, D., Palla, G., Scucchi, S., 2002. Serum cholesterol levels and mood symptoms in the postpartum period. Psychiatry Research 109, 213–219. Troisi, A., Scucchi, S., San Martino, L., Montera, P., d’Amore, A., Moles, A., 2001. Age specificity of the relationship between serum cholesterol and mood in obese women. Physiology and Behavior 72, 409–413. Zureik, M., Courbon, D., Ducimetiere, P., 1996. Serum cholesterol concentration and death from suicide in men: Paris prospective study I. British Medical Journal 313, 649–652.