ابراز هیجان و شیب کورتیزول روزانه در زنان مبتلا به سرطان پستان متاستاتیک در گروه درمانی بیانگر حمایتی: یک مطالعه مقدماتی

Abstract We examined coded emotional expression during an initial therapy session and its association with a known physiological risk factor for early death, aberrant diurnal cortisol slope, in women with metastatic breast cancer. Out of 64 women with metastatic breast cancer randomized to a multi-site clinical intervention trial of supportive-expressive group therapy (SET), a subsample of 29 met eligibility criteria for this study. We tested whether longer mean durations of primary negative affect (fear, sadness, and anger) expression were associated with steeper diurnal cortisol slopes after adjusting for speaking time, repressive-defensiveness, anxiety, and the interaction between repressive-defensiveness and anxiety. We found that steeper cortisol slopes were related to lower repressive-defensiveness and greater primary negative affect expression in line with a priori hypotheses. Additionally we explored whether coded positive affect, defensive/hostile affect, constrained anger, and the interaction between primary negative affect and repressive-defensiveness explained additional variance in diurnal cortisol patterns.

. Introduction Disruptions in the circadian rhythmicity of cortisol (Sephton et al., 2000), other hormones (Davis et al., 2001, Mormont and Levi, 1997 and Schernhammer et al., 2001), and rest/activity cycles (Mormont et al., 2000) have emerged as processes that may connect psychological states with increases in vulnerability to cancer incidence or faster cancer progression (Sephton and Spiegel, 2003). We previously found that aberrations in diurnal cortisol slope were associated with early death and also with greater self-report of repressive-defensiveness and anxiety in a sample of metastatic breast cancer patients (Giese-Davis et al., 2004 and Sephton et al., 2000). Lack of emotional expression, (suppression or repression) has also been linked with greater cancer incidence and faster progression (Giese-Davis and Spiegel, 2003 and Gross, 1989). Emotional expression may help to regulate physiological distress (Gross and Levenson, 1993) and is an important element of support groups for cancer patients (Spiegel and Classen, 2000). In the current study we sought to extend this research to examine associations between baseline diurnal log cortisol slope and observed emotional expression during a first supportive-expressive group psychotherapy session (SET) in women with metastatic breast cancer. Although a few studies have examined the behavioral emotional expression of repressors and high-anxious participants (Milmoe et al., 1967, Rosenthal et al., 1984, Scherer et al., 1991 and Tolkmitt and Scherer, 1986), nothing is known about how they express emotion in a cancer support group or whether expression of primary negative affect by repressors would be associated with better physiological function. Finding that behavioral expression helps to explain additional variance in the diurnal cortisol slope beyond self-report of repressive-defensiveness and anxiety would be important information for future studies examining links between cancer support groups, physiological change, and survival. Emotion-focused psychotherapies, such as SET, posit that direct expression of primary negative affect (fear, anger, and sadness) in a supportive setting during therapy sessions facilitates resolution of ongoing difficulties, releases pent-up emotion, and allows participants to adopt healthier thought processes (Giese-Davis et al., 2002, Giese-Davis et al., 2005 and Greenberg and Safran, 1984). Longer moments of such expression may allow a woman with metastatic breast cancer to be aware of her feelings and cognitions, have time to process her feelings, become aware of choices she can make that might alleviate some of her distress, and to receive support from group members (Giese-Davis et al., 2002; Giese-Davis et al., 2005). This process may lead to better response and recovery from life crises (Classen et al., 2001, Goodwin et al., 2001, Greenberg and Safran, 1989, Spiegel and Bloom, 1983 and Spiegel et al., 1981) and to better physiological function (van der Pompe et al., 1997). Emotion-regulation abilities may develop during psychotherapy for clients who have difficulty expressing or regulating affect (Giese-Davis et al., 2002 and Greenberg, 1993), or may exist as trait-like abilities (Gross, 2002, Lopes et al., 2005 and Salovey et al., 1995). Cortisol is a hormone with known immunosuppressive effects under chronic stress conditions (Dhabhar and McEwen, 1997). Aberrant cortisol slopes (flatter profiles or peaks and troughs at unexpected times) have been associated with chronic stress and conditions including depression, (Deuschle et al., 1997 and Ellenbogen et al., 2002), post-traumatic stress (Yehuda et al., 1996), repression and anxiety (Giese-Davis et al., 2004), chronic cancer fatigue (Bower et al., 2005a and Bower et al., 2005b), higher work load (Caplan et al., 1979), insecure attachment (Adam, 1999), the chronic stress of caring for a child with cancer (Miller et al., 2002), and cancer progression (Touitou et al., 1996 and Touitou et al., 1995). Relationships between observed emotional expression and aberrant cortisol slopes however have not been studied. Chronic experiences of negative life events, negative affectivity, or ineffective emotion-regulation may habitually tax the hypothalamic-pituitary-adrenal (HPA) system leading to higher allostatic load and debilitating health effects (McEwen, 1998). Ability to express and process negative affect in a supportive setting, however, may be associated with better, rather than worse, physiological function (Cruess and Antoni, 2000, McGregor et al., 2004 and van der Pompe et al., 1997). Studies examining cortisol rhythmicity over the day are most relevant to our hypotheses, but most research has instead examined mean cortisol levels assuming that lower mean levels are healthier. However, mean cortisol levels are not comparable across studies because each has utilized different time periods and sampling strategies. In addition, cortisol levels decline throughout the day making mean cortisol levels collected overnight, in the morning, and in the afternoon or evening very different. It is thus impossible to discern how the mean cortisol level is related to the diurnal slope. Examination of the expression of primary negative affect and mean cortisol levels is sparse. Most in line with our hypotheses, Mattsson et al. found that when physicians and nurses rated hemophiliacs as expressing negative emotion, lower rather than higher mean cortisol levels were found (Mattsson et al., 1971). However, contrary to our hypotheses, Vickers found that when participants undergoing stressful life events were rated by clinicians’ as visibly distressed, they had higher urinary cortisol levels (Vickers, 1988), and Davis et al. found that when parents rated children high on negative affect expression, the children had higher mean cortisol levels during the first week of school (Davis et al., 1999). Our primary interest is in the association between primary negative affect and cortisol because SET specifically encourages this expression; however we also explored positive affect, constrained anger, and defensive/hostile expression. Many people assume that expression of positive affect is important to cancer survival even though no research documents this association (Holland and Lewis, 2000). However, greater positive affect experience has been associated with lower mean cortisol levels (Berk et al., 1989, Buchanan et al., 1999, Hubert and deJong-Meyer, 1989, Hubert and deJong-Meyer, 1992 and Smyth et al., 1998). Some studies indicate that constrained anger may be damaging to cancer patients (Giese-Davis and Spiegel, 2003), making examination of its expression important. Lastly, greater experience of hostility has also been associated with higher mean cortisol levels (Pope and Smith, 1991 and Suarez et al., 1998). We found that participation in SET increased self-report of restraint of hostility (impulse control, suppression of aggression, responsibility, and consideration for others) (Giese-Davis et al., 2002), therefore examination of hostile expression and physiology may help to generate hypotheses mechanisms of regulation. We predicted that longer mean durations of expression of primary negative affect during a therapy session would be associated with a steeper (healthier) cortisol slope throughout the day, lower 8:00 h level, and lower mean cortisol level (between 8:00 and 21:00 h), after adjusting for speaking time in the session, and repression and anxiety in the linear model. In a secondary exploratory analysis, we examined whether repressors who expressed primary negative affect would have less aberrant diurnal cortisol slopes. We also explored whether longer mean durations of expression of positive affect, and shorter mean durations of both constrained anger and defensive/hostile affect would account for additional variance.

. Results We examined zero-order Pearson correlations among our study variables (Table 2). We noted that repression was negatively correlated with expression of primary negative affect (r = −0.37, p < 0.05). Table 2. Pearson correlations among variables for metastatic breast cancer patients (N = 29) Psychosocial variables and cortisol 1 2 3 4 5 6 7 8 9 10 1. Diurnal slope of log cortisol level – 2. 8:00 h log cortisol level 0.01 – 3. Mean log cortisol level 0.30 0.89*** – 4. Primary negative affect −0.51** −0.02 −0.20 – 5. Positive affect −0.01 −0.33 −0.38* −0.22 – 6. Defensive affect 0.07 −0.02 −0.05 −0.33 0.30 – 7. Constrained anger −0.32 0.05 −0.12 0.06 0.34 0.23 – 8. Speaking-time −0.16 0.21 −0.14 0.08 0.48** 0.00 0.41* – 9. Repressive-def/restraint composite 0.56** −0.03 0.13 −0.37* 0.05 0.02 −0.36 −0.05 – 10. Anxiety −0.35 0.09 0.11 0.02 0.01 0.16 0.16 0.03 −0.50** – Note. Significance tests were two-tailed. Study variables average duration (in seconds) of an instance of primary negative affect, positive affect, defensive/hostile affect, and constrained anger were from specific affect (SPAFF) coding of participants’ first session of supportive-expressive group therapy. Average duration of an instance of speaking time was from speaker coding of this first session. All coding was from videotapes. Repressive-defensive/restraint composite and anxiety were self-report from the Weinberger Adjustment Inventory Long Form (WAI). * p < 0.05. ** p < 0.01. *** p < 0.001. Table options Our a priori hypothesis was that longer durations of primary negative affect would significantly add to the variance accounted for by average speaking time, repression, anxiety, and the interaction between repression and anxiety and would be associated with less aberrant diurnal cortisol slopes (or steeper log cortisol slopes), lower 8:00 h log cortisol level, and lower mean log cortisol level. We examined these hypothesis in a hierarchical linear regression model entering mean duration speaking time (in seconds), repressive–defensive/restraint composite, anxiety, and the interaction of repressive–defensive/restraint composite and anxiety in the first step and with mean duration of primary negative affect (in seconds) in the second step. This model was significant for log cortisol slope (Table 3). The repressive–defensive/restraint composite was significantly related to the log diurnal slope of cortisol on step 1 (β = 0.50, p = 0.03), and primary negative affect was significantly related to the diurnal slope on step 2 (β = −0.47, p = 0.02). The model was not significant for 8:00 h log cortisol level, or for the mean log cortisol level (and controlling for average physical activity in the first step did not change the significance of these results). Table 3. Summary of linear regression analysis for emotion coding of first group therapy session predicting metastatic breast cancer patients’ diurnal slope of log cortisol level (N = 29) Diurnal Cortisol Slope Variable B S.E. β t p Step 1 Average seconds speaking time −0.001 0.001 −0.13 −0.79 0.44 Repressive–defensive/restraint composite 0.002 0.001 0.50 2.40 0.03 Anxiety −0.001 0.001 −0.09 −0.47 0.64 (Repressive–defensive/restraint composite) × anxiety 1.000 0.000 0.02 0.09 0.93 Step 2 Average seconds primary negative affect −0.006 0.002 −0.47 −2.54 0.02 Note: Step 1: adjusted R2 = 0.22 (F (4,24) = 2.99), p = 0.038. Step 2: adjusted R2 = 0.37, R2 change = 0 0.15 (F change (1,23) = 6.45), p = 0.018. Total model F (5,23) = 4.23, p = 0.007. Table options In Fig. 1 we graphed these results as the projection lines of cortisol slope for subjects at each level of primary negative affect and at the 25th, 50th, and 75th percentiles of the repressive–defensive/restraint composite from the Weinberger Adjustment Inventory. Image for figure Fig. 1 Fig. 1. Figure options 3.1. Secondary analyses We explored whether the other coded affect expressions (positive affect, defensive/hostile affect, and constrained anger), and the interaction between the repressive–defensive restraint composite and mean duration primary negative affect added predictive power in three hierarchical linear regressions. Mean duration speaking time (in seconds), repressive–defensive/restraint composite, anxiety, the interaction of repressive–defensive/restraint composite and anxiety and mean duration primary negative affect were in the first step using an “all-in” model; mean duration positive affect; defensive/hostile affect, constrained anger, and the interaction of repressive–defensive/restraint composite and primary negative affect were entered in a second step using stepwise forward entry, p < 0.05 to enter or remove. We could not demonstrate that the additional coded affects and interaction between repressive-defensiveness and primary negative affect added additional variance in predicting the diurnal log cortisol slope or 8:00 h log cortisol levels (all were excluded from the analysis on the second step). The only additional insight provided was for mean log cortisol level: positive affect added significant additional predictive power in the second step making the resulting total model significant F (6,22) = 2.55, p = 0.050 ( Table 4). Table 4. Summary of linear regression analysis for emotion coding of first group therapy session predicting metastatic breast cancer patients’ mean log cortisol level (N = 29) Mean log cortisol level Variable B S.E. β t p Step 1 Average seconds speaking time −0.02 0.02 −0.14 −0.79 0.44 Repressive–defensive/restraint composite −0.01 0.02 −0.12 −0.45 0.65 Anxiety 0.009 0.03 0.06 0.27 0.79 (Repressive–defensive/restraint composite) × anxiety 0.006 0.003 0.48 2.22 0.037 Average seconds primary negative affect −0.09 0.06 −0.34 −1.52 0.14 Step 2 Average seconds positive affect −0.30 0.12 −0.48 −2.46 0.022 Note: Step 1: adjusted R2 = 0.09 (F (5,23) = 1.52), p = 0.22. Step 2: adjusted R2 = 0.25, R2 change = 0.16 (F change (1,22) = 6.03), p = 0.022. Total model F (6,22) = 2.55, p = 0.050.