Introduction
Gray's reinforcement sensitivity theory (RST; Gray, Reference Gray1970) proposes that there are two brain systems regulating respectively the avoidance of aversive stimuli and the approach to appetitive stimuli and that these systems are reflected in personality traits. He called these systems the behavioural inhibition system (BIS) and the behavioural approach system (BAS), the first regulating response to conditioned cues of punishment, that is the individual's sensitivity to punishment (SP), and the second regulating response to rewarding stimuli, that is the individual's sensitivity to reward (SR). The BIS is sensitive to signals of punishment, non-reward and novelty, and its activation inhibits behaviours leading to negative or painful outcomes, causing inhibition of movement toward goals. The BAS system is sensitive to cues of reward, so its activation allows the person to begin (or to increase) movement toward goals. In terms of personality characteristics, individuals with high BIS sensitivity should exhibit a greater proneness to anxiety whereas those with high BAS sensitivity should exhibit a greater proneness to engage in goal-directed efforts and to experience positive feelings when exposed to cues of impending punishment or reward respectively.
Literature data indicate that differences in SR and SP may be of relevance in modulating individual reactions to food and it has been suggested that SR and SP could have a role in the genesis of eating problems and eating disorders (EDs). In particular, in non-clinical samples, positive correlations between motivational systems and dysfunctional eating patterns, such as bingeing, purging and preoccupation with weight and dieting, have been detected (Loxton & Dawe, Reference Loxton and Dawe2001, Reference Loxton and Dawe2006, Reference Loxton and Dawe2007; Hasking, Reference Hasking2007; Silva et al. Reference Silva, Ortiz, Quiñones, Vera-Villarroel and Slachevsky2011; Matton et al. Reference Matton, Goossens, Braet and Vervaet2013). Moreover, increased SR has been found to be associated with vulnerability to develop binge eating behaviour whereas high SP has been postulated to predispose to restriction of food intake (Dawe & Loxton, Reference Dawe and Loxton2004).
It is known that personality modulates the individual's sensitivity to stressors (Bolger & Schilling, Reference Bolger and Schilling1991; van Os & Jones, Reference van Os and Jones1999; Rijsdijk et al. Reference Rijsdijk, Sham, Sterne, Purcell, McGuffin, Farmer, Goldberg, Mann, Cherny, Webster, Ball, Eley and Plomin2001; Kendler et al. Reference Kendler, Kuhn and Prescott2004), which affects the hypothalamus–pituitary–adrenal (HPA) axis activity. Although several studies have examined HPA axis activity in relation to temperament or personality in adults, measured mainly by Cloninger's Tridimensional Personality Questionnaire (TPQ; Cloninger, Reference Cloninger1987) or Temperament and Character Inventory (TCI; Cloninger, Reference Cloninger1999), the results have been mixed. In fact, some studies reported no significant associations between personality and different measures of HPA axis activity (Kirschbaum et al. Reference Kirschbaum, Bartussek and Strasburger1992; Schommer et al. Reference Schommer, Kudielka, Hellhammer and Kirschbaum1999) whereas others showed significant positive or negative associations (McCleery & Goodwin, Reference McCleery and Goodwin2001; Zobel et al. Reference Zobel, Barkow, Schulze-Rauschenbach, Von Widdern, Metten, Pfeiffer, Schnell, Wagner and Maier2004). However, although the Cloninger model of personality stems partly from the same background as Gray's RST, the TPQ and TCI were not thought to provide an ideal operationalization of SP and SR, and more specific instruments, the BIS/BAS scales, were developed (Carver & White, Reference Carver and White1994).
To our knowledge, no previous study has assessed HPA axis activity in relation to SP and SR using the BIS/BAS scales in either healthy subjects or clinical populations. This topic is of particular relevance in EDs because an association between exposure to stressful life events and the onset or course of anorexia nervosa (AN) and bulimia nervosa (BN) has been demonstrated (Hagan et al. Reference Hagan, Shuman, Oswald, Corcoran, Profitt, Blackburn, Schwiebert, Chandler and Birbaum2002; Jacobi et al. Reference Jacobi, Hayward, de Zwaan, Kraemer and Agras2004; Fairburn & Bohn, Reference Fairburn and Bohn2005; Pike et al. Reference Pike, Wilfley, Hilbert, Fairburn, Dohm and Striegel-Moore2006; Rojo et al. Reference Rojo, Conesa, Bermudez and Livianos2006; Corstorphine et al. Reference Corstorphine, Mountford, Tomlinson, Waller and Meyer2007) and, as stated earlier, an individual's SP or SR could affect both HPA axis activity and eating behaviour. Therefore, a study of HPA axis activity in relation to a subject's personality characteristics of SP and SR could help to shed more light on the pathophysiology of AN and BN.
The cortisol awakening response (CAR) has received increasing attention as a useful method to assess the integrity of the HPA axis (Schmidt-Reinwald et al. Reference Schmidt-Reinwald, Pruessner, Hellhammer, Federenko, Rohleder, Schürmeyer and Kirschbaum1999; Wüst et al. Reference Wüst, Federenko, Hellhammer and Kirschbaum2000; Bartels et al. Reference Bartels, Van den Berg, Sluyter, Boomsma and de Geus2003). Therefore, we conducted an explorative study to assess the CAR in patients with symptomatic AN or BN compared to matched healthy controls, examining correlations that might emerge between this measure of HPA axis activity and subjects’ SP or SR measured by the BIS/BAS scales.
Method
Subjects
A total of 67 subjects volunteered for the study: 48 female patients consecutively attending the out-patient facilities of the Eating Disorders Centre of the Department of Mental Health of the University of Naples SUN and 19 healthy women. According to DSM-5 criteria, 13 patients met the diagnosis of AN binge/purge subtype (ANB/P), 12 the diagnosis of AN restrictive subtype (ANR) and 23 the diagnosis of BN. Diagnostic assessment was made by a trained interviewer using the SCID-I/P (First et al. Reference First, Spitzer, Gibbon and Williams1995). At the time of the study, the patients had never taken psychotropic medications and none of them had a current or past history of alcohol or drug abuse or a current co-morbid Axis I disorder; five patients with BN had a lifetime major depression disorder.
Control women were drug free and physically and mentally healthy, as assessed by physical examination, routine medical interview and the SCID-I/NP (First et al. Reference First, Gibbon, Spitzer and Williams1996). Information on their eating habits was carefully collected by an ad-hoc structured clinical interview to exclude EDs, which could have occurred at subclinical severity and thereby confounded the results. All controls were menstruating regularly.
Procedure
The experimental protocol was approved by the local ethics committee and all subjects gave their written consent after being fully informed of the nature and procedures of the study. Menstruating patients and healthy women were tested in the follicular phase of their menstrual cycle (day 5–10 from menses).
All subjects underwent a structured clinical interview aiming to collect historical and clinical data concerning their illness. Sociodemographic data and eating habits were collected by means of a semistructured clinical interview that we usually adopt in our clinical assessment of patients with EDs; this is a validated instrument based on the ED module of the SCID-I/P. All subjects filled in the BIS/BAS scales (Carver & White, Reference Carver and White1994), which comprise a 20-item questionnaire with answers rated on a four-point Likert scale, ranging from 1 (I strongly agree) to 4 (I strongly disagree). The questionnaire contains a subscale for BIS, measuring reactions to anticipated punishment, and three subscales for BAS, measuring drive (BAS-D), fun seeking (BAS-FS) and reward responsiveness (BAS-RR). BAS-D measures the persistent pursuit of desired goals; BAS-FS measures both a desire for new rewards and a willingness to approach a potentially rewarding event on the spur of the moment; and BAS-RR measures positive responses to the occurrence or anticipation of reward. For this study we used the Italian version of the BIS/BAS scales, validated by Leone et al. (Reference Leone, Pierro and Mannetti2002), which differs from the original version only for the integration of an intermediate scoring on the Likert scale going from 1 to 5.
The functional status of the HPA axis was evaluated by assessing salivary CAR. To this purpose, participants were instructed to collect saliva samples at home on a work day immediately on awakening and 15, 30 and 60 min post-awakening. During this time they had to refrain from eating, drinking (except water), smoking and brushing teeth. In addition, participants were asked to record information on the duration of their sleep the night before the sampling procedure and their time of awakening on the morning of sampling. All these procedures were checked by a relative. Cortisol concentrations in saliva were determined by using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Biochem Immunosystems, Italy). Intra- and interassay coefficients of variation were less than 8% and 8.7% respectively.
For each subject, the CAR was evaluated as the cortisol area under the curve (AUC) by applying the trapezoidal rule, the cortisol Δ-max, that is the maximum cortisol concentration in saliva after awakening minus the baseline value, or the slope from baseline to peak (SBP), that is the slope of the line connecting the cortisol value at awakening to the peak value after awakening (Fekedulegn et al. Reference Fekedulegn, Andrew, Burchfiel, Violanti, Hartley, Charles and Miller2007).
Statistics
The BMDP statistical software package (Dixon, Reference Dixon1985) was used for data analysis. Data were assessed for normal distribution by the Wilk–Shapiro test. Salivary cortisol values and anthropometric and clinical variables were normally distributed whereas salivary cortisol AUC, Δ-max and SBP values were not. Therefore, a two-way ANOVA, with group as the between-subjects factor, the repeated measure of time (0, 15, 30 and 60 min after awakening) as the within-subjects factor and cortisol levels as the dependent variable, was used to test differences in salivary cortisol responses to awakening among the groups. A one-way ANOVA was used to assess differences in demographic and anthropometric variables among groups. Correlation analyses between salivary cortisol AUC, Δ-max, SBP values and anthropometric and clinical variables were performed by using the non-parametric Spearman rank correlation test. As recommended for exploratory studies, we did not perform multiple test adjustments (Bender & Lange, Reference Bender and Lange2001).
Results
Clinical and demographic data
AN patients were of similar age to the healthy controls but had significantly lower values for body mass index (BMI). No significant differences were found in either age or BMI between BN patients and healthy women (Table 1). Patients with AN and BN showed statistically significantly higher scores in the BIS measure but did not differ in BAS subitem scores compared to healthy controls (Table 1). No significant differences emerged between patients and controls in the mean time of awakening or in the mean duration of sleep the night before sampling. Times of awakening ranged from 06:45 to 07:50 h in healthy subjects, from 06:55 to 08:00 h in AN patients and from 06:45 to 07:45 h in BN patients.
Table 1. Demographic data and personality characteristics of sensitivity to punishment and reward of the study sample
ANB/P, Anorexia nervosa binge/purge subtype; ANR, AN restrictive subtype; BN, bulimia nervosa; BMI, body mass index; BIS, behavioural inhibition system; BAS, behavioural approach system; BAS-RR, BAS-reward responsiveness; BAS-D, BAS-drive; BAS-FS, BAS-fun seeking; df, degrees of freedom.
Values given as mean ± standard deviation.
a F 1,30 = 6.02, p = 0.02; b F 1,29 = 25.4, p < 0.0001; c F 1,30 = 7.37, p = 0.01; d F 1,29 = 19.47, p = 0.0001; e F 1,40 = 6.36, p = 0.01 compared to healthy women.
CAR test
The group × time repeated-measures ANOVA yielded a significant main effect for group (F 2,64 = 10.37, p = 0.0001) and time (F 3,192 = 23.94, p < 0.00001) but no significant group × time interaction (F 6,192 = 1.36, p = 0.2), indicating that salivary cortisol levels changed significantly after awakening, with significant differences among the groups in the amount but not in the timing of the hormone response to awakening. An ANCOVA showed no significant effect of current BMI on the magnitude of salivary CAR (F 1,63 = 0.13, p = 0.7). In a two-group comparison, when AN patients were compared to healthy controls, a two-way ANOVA with repeated measures showed a significant main effect for group (F 1,42 = 7.32, p = 0.009) and time (F 3,126 = 13.72, p < 0.00001) but no significant group × time interaction (F 3,126 = 1.08, p = 0.2); the ANCOVA showed no significant effect of current BMI on the magnitude of salivary CAR (F 1,41 = 1.11, p = 0.2). When BN patients were compared to healthy controls, a two-way ANOVA with repeated measures showed a significant effect for time (F 3,120 = 17.38, p < 0.00001) but no significant effect for group (F 1,40 = 1.57, p = 0.2) and no significant group × time interaction (F 3,120 = 0.89, p = 0.4) (Fig. 1 a). No significant differences emerged in the CAR between patients with ANR and those with ANB/P. Compared to the healthy controls, the CAR, evaluated as cortisol AUC, Δ-max or SBP, was significantly higher in AN patients (F 1,42 = 7.84, p = 0.007 for AUC; F 1,42 = 6.35, p = 0.01 for Δ-max; F 1,42 = 6.15, p = 0.01 for SBP) but not in BN patients (F 1,40 = 0.99, p = 0.3 for AUC; F 1,40 = 0.01, p = 0.9 for Δ-max; F 1,40 = 0.22, p = 0.6 for SBP) (Fig. 1 b–d). ANCOVAs showed no significant effect of current BMI on the observed differences between AN and healthy subjects in salivary cortisol AUC (F 1,41 = 0.93, p = 0.3), Δ-max (F 1,41 = 1.01, p = 0.3) or SBP (F 1,41 = 0.37, p = 0.5).
Fig. 1. (a) Salivary cortisol awakening response (CAR), (b) cortisol area under the curve (AUC), (c) cortisol Δ-max and (d) cortisol slope from baseline to peak (SBP) in drug-free patients with anorexia nervosa (AN), bulimia nervosa (BN) and healthy controls. Data are expressed as mean ± standard deviation. * p < 0.01, ** p < 0.007 versus healthy women (one-way ANOVA).
Correlations
In healthy women, correlation analyses showed significant negative associations between BAS-FS scores and CAR AUC (r = –0.67, p = 0.001), Δ-max (r = –0.73, p = 0.0004) or SBP (r = –0.60, p = 0.006) and between BAS-RR scores and CAR AUC (r = –0.43, p = 0.05) or Δ-max (r = –0.57, p = 0.01); the correlation between BAS-RR scores and SBP was slightly above the threshold of statistical significance (r = –0.43, p = 0.06).
In the AN group, significant positive correlations emerged between BAS-FS scores and CAR AUC (r = 0.43, p = 0.03), Δ-max (r = 0.46, p = 0.01) or SBP (r = 0.55, p = 0.004); moreover, BAS-D scores were positively correlated with SBP (r = 0.62, p = 0.009). When correlation analyses were performed separately in the two subtypes of AN patients, significant positive correlations between BAS-FS scores and CAR AUC (r = 0.63, p = 0.01), Δ-max (r = 0.85, p = 0.0005) or SBP (r = 0.72, p = 0.001), and also between BAS-D scores and CAR Δ-max (r = 0.64, p = 0.01) or SBP (r = 0.75, p = 0.002), were detected only in the ANB/P patient group. In BN patients, significant positive correlations emerged between BAS-FS scores and CAR Δ-max (r = 0.62, p = 0.001) or SBP (r = 0.60, p = 0.002) and between BAS-D-scores and CAR Δ-max (r = 0.47, p = 0.02) or SBP (r = 0.49, p = 0.01).
No significant correlations were detected between cortisol measures and subjects’ age, BMI, age at onset of the illness or duration of illness.
Discussion
To our knowledge this is the first study exploring the activity of the HPA axis in patients with EDs and in healthy controls in relation to personality characteristics of SP and SR, assessed according to Gray's RST. The results can be summarized as follows: (1) compared to healthy controls, patients with AN exhibited an increased CAR whereas patients with BN did not show any abnormality in post-awakening cortisol secretion; (2) both AN and BN patients exhibited higher SP but no different SR than healthy controls, as measured by the BIS/BAS scales; and (3) CAR was significantly associated with BAS measures, negatively in healthy women but positively in patients with ANB/P or BN.
According to our previous findings (Monteleone et al. Reference Monteleone, Scognamiglio, Monteleone, Mastromo, Steardo, Serino and Maj2011b ), symptomatic AN patients showed a CAR that did not differ from normal controls in the timing of secretion but occurred at higher hormone levels, with an enhanced total secretion of cortisol and a higher secretory peak. Moreover, for the first time, the CAR was found to be normal in symptomatic patients with BN. These results were not explained by severity of body weight changes or duration of the illness because no significant correlations emerged between these variables and cortisol measures. The CAR is a response to awakening in the morning characterized by an immediate rise in cortisol secretion that is distinct from the circadian rise in HPA axis activity in the morning hours and reflects phasic psychophysiological processes specific to the sleep–wake transition (Kudielka et al. Reference Kudielka, Hellhammer and Wüst2009). The CAR is generally considered to be an indicator of individual differences in HPA axis activity, and changes in the CAR can yield important information on individual responsiveness to stressors (Clow et al. Reference Clow, Thorn, Evans and Hucklebridge2004). The enhanced salivary CAR detected in our symptomatic AN women but not in our BN women is consistent with a large body of literature showing hyperactivity of the HPA axis in AN but not in BN, although inconsistencies have also been reported (Lo Sauro et al. Reference Lo Sauro, Ravaldi, Cabras, Faravelli and Ricca2008; Monteleone et al. Reference Monteleone, Scognamiglio, Canestrelli, Serino, Monteleone and Maj2011a ). Therefore, the present findings provide further support for the evidence of dysregulation of HPA axis activity in AN but not in BN.
The second finding of our study was that AN and BN patients scored higher than healthy women on the BIS scale but did not differ significantly in BAS scale scores. In accordance with our results, three studies that used the BIS/BAS scales found that BIS scores were increased in both AN and BN subjects (Kane et al. Reference Kane, Loxton, Staiger and Dawe2004; Claes et al. Reference Claes, Nederkoorn, Vandereycken, Guerrieri and Vertommen2006; Harrison et al. Reference Harrison, Treasure and Smillie2011), whereas data on BAS subitem scores were inconsistent. Indeed, Claes et al. (Reference Claes, Nederkoorn, Vandereycken, Guerrieri and Vertommen2006) and Harrison et al. (Reference Harrison, Treasure and Smillie2011) found that, in people affected by an ED, BAS scores were lower than healthy controls, whereas Kane et al. (Reference Kane, Loxton, Staiger and Dawe2004) reported that this was not the case in BN patients with or without alcohol abuse, who showed the opposite. Therefore, the literature data and our present findings demonstrate that ED patients have an enhanced SP, as expressed by the BIS score, but their SR, as measured by the BAS scale, requires further study.
Finally, we found that the CAR was significantly associated with BAS measures, negatively in healthy women but positively in ANB/P and BN patients. To our knowledge, there have been no previous studies investigating HPA axis activity in relation to the RST model of personality. Studies exploring the activity of the HPA axis in relation to personality characteristics have generally used measures of personality based on Cloninger's model (Cloninger, Reference Cloninger1999). In this model, Cloninger's novelty seeking (NS) personality dimension reflects the tendency to respond with excitement to cues of reward or relief from punishment leading to active approach behaviours in pursuit of rewards, whereas Cloninger's harm avoidance (HA) dimension is a tendency to respond to cues of punishment, resulting in passive avoidance of situations that may result in punishment. Although differences have been pointed out between the RST and Cloninger's model of personality, it can be assumed that the BIS measure can be related to HA and the BAS measure to NS. Studies performed mainly in adult populations have shown either no association or significant relationships between Cloninger's personality dimensions and HPA axis activity. In positive studies, however, the directions of those associations were mixed. For instance, Tyrka et al. (Reference Tyrka, Wier, Price, Rikhye, Ross, Anderson, Wilkinson and Carpenter2008) showed that, in normal adults, the cortisol response to the dexamethasone/corticotrophin-releasing hormone (Dex/CRH) test was negatively correlated with NS and positively correlated with HA. Instead, the CAR has been reported to be both positively and negatively associated with neuroticism, a personality trait associated with HA (Portella et al. Reference Portella, Harmer, Flint, Cowen and Goodwin2005; Vedhara et al. Reference Vedhara, Tuinstra, Stra, Miles, Sanderman and Ranchor2006; Mangold et al. Reference Mangold, Mintz, Javors and Marino2012). Although differences may exist between the different constructs, our data, taken together with those of Tyrka et al. (Reference Tyrka, Wier, Price, Rikhye, Ross, Anderson, Wilkinson and Carpenter2008), suggest that, in healthy subjects, higher levels of BAS scores or NS are associated with less HPA axis activation whereas the BIS score, unlike HA or neuroticism, does not seem to be related to the activity of the corticosteroid system. Differences in the methods used to assess HPA axis function and in the instruments used to measure SP may explain such discrepancies. Moreover, in women with BN or ANB/P, our findings show a significant correlation between CAR and BAS scores, which is opposite to that detected in healthy controls. The reasons for such an inverse association are not clear. One hypothesis could be that healthy women and binge-purging patients differ in variables other than personality characteristics that could be relevant to HPA axis regulation. For example, it has been ascertained that a history of childhood trauma or abuse may have long-lasting functional effects on the HPA axis and that the positive or negative direction of these effects may vary in relation to the type of trauma, its duration and the subject's age at exposure (Pryce et al. Reference Pryce, Rüedi-Bettschen, Dettling, Weston, Russig, Ferger and Feldon2005; Carpenter et al. Reference Carpenter, Carvalho, Tyrka, Wier, Mello, Mello, Anderson, Wilkinson and Price2007). About 30% of adults with bulimia spectrum disorders report a history of childhood sexual abuse and at least 30% a history of physical abuse, rates that are significantly higher than in healthy controls (Fullerton et al. Reference Fullerton, Wonderlich and Gosnell1995; Wonderlich et al. Reference Wonderlich, Brewerton, Jocic, Dansky and Abbott1997; Léonard et al. Reference Léonard, Steiger and Kao2003). In this connection, patients with BN or ANB/P who reported a positive history of childhood trauma or abuse were found to exhibit a hypersensitive HPA axis response to dexamethasone (Diaz-Marsa et al. Reference Díaz-Marsá, Carrasco, Basurte, Pastrana, Sáiz-Ruiz and López-Ibor2007) and a reduced secretion of cortisol after a serotonergic challenge test (Steiger et al. Reference Steiger, Gauvin, Israël, Koerner, Ng, Paris and Young2001). Therefore, a positive history of childhood trauma or abuse in our binge-purging patients, which we did not assess in our study, could have a role in the determinism of our findings.
The neural basis of the BAS probably involves brain reward circuits, including mainly mesolimbic dopaminergic pathways from the ventral tegmental area to the nucleus accumbens and the medial orbitofrontal cortex. It is known that stress modulates brain reward circuits (Koob, Reference Koob2008) and this effect seems to be mediated by activation of the HPA axis (McKee et al. Reference McKee, Sinha, Weinberger, Sofuoglu, Harrison, Lavery and Wanzer2011). Our findings of an opposite association between CAR and BAS measures in binge-purging ED patients and in healthy controls may suggest a deranged HPA axis modulation of brain reward pathways in patients, which could be relevant for the pathophysiology of EDs. In this regard it is noteworthy that EDs have been conceptualized as reward-related disorders (Kaye et al. Reference Kaye, Fudge and Paulus2009; Monteleone & Maj, Reference Monteleone and Maj2013). Finally, the present findings, if confirmed in future studies, might contribute to characterizing the biological background of the Positive Valence Systems domain, recently proposed by the National Institute of Mental Health (NIMH) Research Domain Criteria (www.nimh.nih.gov/research-priorities/rdoc/index.shtml).
Some limitations of our study need to be discussed. First, the participants collected morning saliva samples at home under the supervision of a relative and not under strict observation in our clinical research unit. Although the non-expert supervised sampling at home may have affected the results, this procedure is commonly followed in clinical studies assessing the CAR. Second, evidence suggests that CAR is very sensitive to state variations in day-to-day activities (Kudielka et al. Reference Kudielka, Hellhammer and Wüst2009), and therefore it would have been preferable to study each participant on more than one work day, which would have enabled a more robust estimate of the CAR from each group to be determined. Therefore, the present data should be confirmed in studies taking into account the variables affecting the awakening cortisol secretion. Third, a larger sample size would have increased the statistical power of our study. However, a post-hoc power analysis revealed that the present study was sufficiently powered because our sample had a power of 0.94 to detect a small to medium effect size (w = 0.2) at an α value of 0.05 to determine significant differences in the CAR among the groups.
In conclusion, the present findings confirm that the CAR is enhanced in symptomatic patients with AN and show, for the first time, that it is normal in symptomatic women with BN. Moreover, the CAR seems to be associated with the personality characteristic SR, as conceptualized by Gray's RST, in opposite directions in binge-purging ED patients and in healthy controls. The relevance of these findings in the pathophysiology of EDs and the modulation of HPA axis activity require further study for clarification.
Declaration of Interest
None.
