The critical question in immunosurveillance against cancer, which has so far remained unanswered, is whether differences between individuals with regard to immunologic host-defense mechanisms can predict the future development of cancer. Various in vivo studies in humans and mice have shown the significant role of different natural cytotoxic lymphocyte populations in preventing the development of cancer or its metastases.1–4 However, only a few studies have investigated NCA of peripheral blood lymphocytes among healthy individuals in a general population or those at increased risk of developing cancer.5–7 NCA has also been demonstrated to be an independent prognostic marker in patients with cancer.8, 9
Imai et al.10 showed that selected lifestyle factors, such as maintaining body weight, not smoking, increasing intake of green vegetables and moderate alcohol consumption, may have an effect on natural cytotoxicity and that NCA can be increased by altering several lifestyle factors. Other studies have shown that restricted lifestyle changes, such as increasing exercise, losing weight and reducing fat and alcohol intake, are associated with a significant reduction in breast cancer risk (reviewed by Chlebowski11).
Whether emotional distress is associated with the onset of cancer in general and of breast cancer in particular has long been debated. Some studies have suggested that the overall prevalence of breast cancer is increased in women who have experienced stressful life events.12, 13 In contrast, there is no good evidence pointing to any relation between stressful life events and the onset of breast cancer.14–16
The familial risk for breast cancer is widely acknowledged, and DBCP have deep feelings of anxiety about their health, in addition to a concern for their sick mothers.17, 18 As a result, DBCP live in a high degree of chronic stress.18 The association between chronic stress and impaired immune surveillance functions has long been established,19, 20 while the immunosuppressive effect of chronic stress is presumed to be mediated through excess secretion of the stress hormones cortisol and catecholamines.21, 22
Our study was aimed at clarifying the association between natural immunologic defense, stress hormone levels and psychologic status in DBCP. Our findings demonstrate that natural cytotoxicity, in particular against NK-resistant cells, is significantly impaired in DBCP and mostly affected by disturbed hormonal and immunologic mechanisms. Chronic emotional distress was identified as the main factor associated with both impaired immune surveillance functions and stress hormone levels.
MATERIAL AND METHODS
The study population was comprised of 80 DBCP, aged 20–45 years, divided into 2 groups: DBCP1 included 39 DBCP (mean age ± SD = 31.3 ± 7.9 years) whose mothers were found to have primary localized breast cancer 1 year prior to the beginning of the study; DBCP2 included 41 DBCP (mean age ± SD = 32.1 ± 7.5 years) whose mothers were found to have recurrent breast cancer (localized or metastatic) 1 year prior to the beginning of the study. Forty-seven women, matched for age (mean age ± SD = 30.9 ± 8.9 years), economic status and educational status, whose mothers were healthy and did not suffer from cancer in the past, were included in the control group.
DBCP and controls were healthy at the time of the study, were not pregnant or on any treatment and did not have a history of cancer or benign tumor. DBCP or controls who suffered from an acute viral infection or allergic reaction were studied 3 weeks or more after remission of symptoms.
Controls did not differ from DBCP in the number of negative life events during the 3 years prior to the study and in the health status of their father (data not shown). In addition, DBCP did not differ from controls in their smoking habits [F(2,123) = 0.12, p > 0.05], diet [F(2,123) = 0.95, p > 0.05], sleep habits [F(2,123) = 0.14, p > 0.05] or the degree of stress associated with blood drawing [F(2,123) = 2.13, p > 0.05].
Heparinized venous blood (30 ml) was drawn from all subjects between 8 and 9 A.M., and each participant answered a set of psychologic questionnaires. In addition, the first-morning urine sample was collected. Experiments were conducted in triplicate, and in each experiment subjects were counterbalanced with appropriate controls. The study was confirmed by the Helsinki committee of our hospital, and written informed consent was obtained from each participant.
Reagents and cell lines
RPMI-1640, FCS, penicillin-streptomycin, trypsin, EDTA, L-glutamine, NEAA and sodium pyruvate were purchased from Biological Industries (Beit-Ha'amek, Israel). PHA and LPS from Escherichia coli were purchased from Sigma (St. Louis, MO). 51Cr was from Amersham (Aylesbury, UK). rhIL-2 and rhIL-12 were purchased from R&D (Minneapolis, MN). FITC- or PE-conjugated mouse antihuman CD3 (Leu-4), CD4 (Leu-3a), CD56 (Leu-19), CD8 (Leu-2a), CD25 and CD14 MAbs were purchased from Becton Dickinson (San Jose, CA). ELISA kits for human IL-2, human IL-12 (highly sensitive) and human IFN-γ were from R&D. The human cell lines K562 (erythroleukemia), MCF-7 (breast cancer), COLO-205 (colon adenocarcinoma) and U937 (human premonocytic leukemia) were maintained in RPMI-1640 supplemented with 10% FCS at a concentration of 106 cells/ml in continuous-suspension cultures.
NCA against NK-sensitive K562 cells was tested, as previously described,23, 24 at E:T ratios of 50:1, 25:1 and 12:1. NCA is also presented as the number of LUs in 107 effector cells (1 LU is the number of effector cells that lyse 20% of target cells). LU was calculated as previously described25 with the aid of an appropriate computer program (gift from Dr. T. Whiteside, University of Pittsburgh, Pittsburgh, PA). In certain experiments, where NK-resistant cells (MCF-7, U937 or COLO-25) were used as targets, NCA was modified by preincubation with IL-2 or IL-12.26
IL-2, IL-12 and IFN-γ levels were measured in supernatants of stimulated PBMCs. For IL-2, 2 × 106 PBMCs were incubated in RPMI-FCS culture medium with PHA (5 μg/ml) for 32 hr at 37°C. For IL-12, 5 × 106 plastic-adherent PBMCs were incubated in 1 ml culture medium without FCS and stimulated with LPS (10 μg/ml) for 18 hr at 37°C. IFN-γ secretion from T cells was measured in supernatants of 1 × 106 PBMCs incubated with PHA (5 μg/ml) and IL-2 (120 U/ml) in enriched RPMI-FCS culture medium for 32 hr at 37°C,27 and IFN-γ secretion from NK cells was measured in supernatants of 1 × 106 PBMCs incubated with IL-12 (100 pM) in enriched RPMI-FCS culture medium for 32 hr at 37°C.28 Sensitivity threshold levels of ELISA kits (R&D) used for measuring cytokine concentration were as follows: 10 pg/ml for IL-2, 0.5 pg/ml for IL-12 and 3 pg/ml for IFN-γ. The assay for IL-12 recognizes the biologically active IL-12 heterodimer and not the monomeric form.29
Cortisol and catecholamine levels
The cortisol concentration in plasma was determined by the Gamma Coat Cortisol Radioimmunoassay (DiaSorin, Stillwater, MN). Plasma (10 μl) was incubated with 125I cortisol tracer in rabbit anticortisol serum-coated tubes. Tubes were washed and thereafter counted in a γ counter. Assay sensitivity was 5.8 nmol/l. Epinephrine and norepinephrine levels in urine were measured by HPLC with an electrochemical detector system (ESA Coulochem, Bedford, MA). Five milliliters of first-morning urine were mixed with internal standard (dihydroxybenzylamine and 1% EDTA in phosphate buffer) and loaded onto resin (Amberlite CG-50) columns. Catecholamines were eluted from the columns with 4% boric acid, diluted in 0.1 M HClO4 and injected into the HPLC.
The SCL-90R is composed of 90 items representing 9 categories of emotional symptoms and is suitable for assessing emotional response to physical illness.30 Internal consistency (α) of the categories ranged from 0.72 to 0.86
The BDI includes 21 items for identification of affective, behavioral, cognitive and somatic symptoms of depression. It is suitable for diagnosing depression associated with physical illness and stressful life events.31 Internal consistency (α) of the inventory was 0.83.
The State-Trait Anxiety Inventory consists of 2 subscales, 20 items each, which evaluate situational anxiety (the present experience of anxiety) and trait anxiety (the potential of the individual to experience anxiety symptoms when confronted with a threatening situation).32 Internal consistency (α) for each of the 2 subscales was 0.95.
The Ways of Coping Questionnaire is based on the psychologic approach to coping with stressful situations.33 The revised version, with 44 items, was used.34 Four categories of coping were identified by 3 professional referees: problem-focused, emotion-focused (avoidance), social support and positive focus. Internal consistency (α) of the categories ranged from 0.73 to 0.80.
A health-related questionnaire was composed for our study and included questions concerning negative life events, perceived level of stress, perceived control over breast cancer prevention, health status and health behavior patterns. Participants were also asked to grade their feeling of stress concerning the blood drawing on a scale from 0 (no stress) to 5 (extreme feeling of stress).
We used the SPSS (Chicago, IL) software package. One-way multivariate ANOVA was used to assess differences between groups of related variables (each NCA assay in all E:T ratios, each LU group, all hormonal measures, all SCL-90R subscales and other psychologic measures). When results were significant, variance analysis was conducted for each variable separately (univariate ANOVA). Significant main effects were followed up with post hoc Scheffe's test, to identify differences between pairs of groups. Interactions between psychologic, immunologic and hormonal variables were assessed on 2 levels: interactions between variables were examined using the 2-tailed Pearson correlation analysis and Fisher's Z analysis was conducted to compare correlation coefficients between the 3 groups, revealing no significant differences for correlations between psychologic distress, immune and hormonal variables, except for the variable of perceived control over breast cancer prevention. Hence, all correlation coefficients (except those with perceived control) are presented jointly. A Bonferroni procedure was applied to take into account the 10 correlations performed. Significance was set at p < 0.01.35 Thereafter, hierarchical regression analysis was conducted to determine the relative predictive values of the independent variables on the immune and stress hormone variables. It was conducted in 3 steps. In the first step, the “group” variable was entered as a dummy in 2 clusters (cluster 1 contained DBCP1 and DBCP2 together and controls as the second group, whereas in cluster 2 DBCP1 and controls were entered as a group and compared to DBCP2). In the second step, the independent variables were entered; and in the third step, the interaction between clusters and independent variables was entered.
Only a limited number of variables were used in the correlation and regression analysis, namely, NCA as represented by LU, coping strategies as represented by problem-focused and emotion-focused strategies and the global scores of the SCL-90 as represented by the general severity index score. BDI scores represent depression, whereas Spilberger's state anxiety questionnaire scores anxiety.
White blood cell counts
No significant differences were found between DBCP and controls in the absolute numbers of total white blood cells, lymphocytes and monocytes (Fig. 1a) in the proportion of total T lymphocytes (CD3+), T-helper lymphocytes (CD3+/CD4+), T-suppressor/cytotoxic lymphocytes (CD3+/CD8+), activated T lymphocytes (CD3+/CD25+) and monocytes (CD14+) (Fig. 1b) or in the proportion and absolute number of NK cells (CD56+/CD3–) (Fig. 1c).
NCA against NK-sensitive target cells (K562) was significantly lower in DBCP compared to controls (Fig. 2). No significant differences were found between DBCP1 and DBCP2 (Fig. 2). Preincubation of PBMCs with IL-2 or IL-12 increased NCA in both DBCP and controls (Fig. 3a,b). However, the increase in DBCP was significantly lower compared to controls (Fig. 3a,b). NCA against NK-resistant breast cancer target cells (MCF-7) was also lower in DBCP compared to controls (Fig. 4a,b). However, preincubation with IL-2 or IL-12 markedly augmented NCA in controls, whereas in DBCP it was much less increased (Fig. 4a,b). The same was observed for spontaneous (Fig. 5a) and cytokine-induced (Fig. 5b) NCA against 2 more NK-resistant target cell lines, COLO-205 and U937. When we calculated the differences between cytokine-induced and spontaneous NCA (by subtracting spontaneous NCA from cytokine-induced NCA), the degree of the increase of cytokine-induced NCA was significantly lower in DBCP compared to controls (Fig. 6).
In vitro secretion of Th1 cytokines
Secretion of IL-2, IL-12 and IFN-γ was lower in DBCP compared to controls (Fig. 7a,b).
Cortisol and catecholamine levels
Plasma cortisol level and urinary concentrations of epinephrine and norepinephrine were significantly higher in DBCP compared to controls (Fig. 8a,b).
DBCP, and in particular DBCP2, experienced significantly higher levels of distress (anxiety, depression, somatization, perceived level of stress) and lower perceived control over breast cancer prevention compared to controls (Fig. 9a–e). No differences were found between DBCP and controls regarding emotional symptoms (e.g., trait anxiety, paranoid ideation, Psychoticism), which are less reactive to stressful situations (Fig. 9a,c). Similarly, DBCP did not differ from controls in ways of coping with stressful life events (data not shown).
Effect of psychologic and hormonal factors on NCA and Th1 cytokine secretion
High levels of anxiety in DBCP were associated with reduced NCA (Fig. 10a), reduced IL-2 secretion (Fig. 10b) and elevated levels of norepinephrine (Fig. 10c). Using emotion-focused coping strategies was significantly and negatively associated with IL-2-induced NCA (Fig. 10d), IL-2 secretion and norepinephrine levels (data not shown). Problem-focused coping and perceived control were significantly and positively associated with IL-2-induced NCA (Fig. 10e) and negatively with norepinephrine levels (Fig. 10g) but not with IL-2 secretion (Fig. 10h). Cortisol levels (data not shown) and norepinephrine secretion could predict lower IL-2-induced NCA against MCF-7 target cells and lower IL-2 secretion (Fig. 10I,j). Norepinephrine and cortisol levels significantly predicted low IL-2 secretion (data not shown). The correlations between norepinephrine level and state anxiety, perceived control and coping strategies did not reach significance after implementing the Bonferroni procedure. No impact on immune functions was exerted by the acute stress of blood drawing or health behavior habits like sleeping or smoking (data not shown).
We examined the psychologic, immunologic and hormonal characteristics of DBCP and the associations between psychologic status, immunologic functions and stress hormone levels. We also examined the relative impact of the psychologic parameters on immunologic functions and stress hormone levels. Our main findings are that IL-2- and IL-12-induced cytotoxic activity against NK-resistant target cells is significantly reduced in DBCP, in particular in DBCP2, compared to controls (Figs. 3–6). In vitro secretion of IL-2, IFN-γ and, to a lesser extent, IL-12 is reduced in DBCP (Fig. 7). Impairment of immune functions could be predicted by increased emotional distress, lessened use of problem-focused coping strategy, increased use of emotion-focused coping and reduced sense of personal control over breast cancer prevention as well as by norepinephrine concentrations in urine (Fig. 10). Our findings confirm previous results that NCA is a sensitive immunologic measure of chronic psychologic stress.19, 20, 36 However, we show that cytokine-induced NCA against NK-resistant target cells is an even more reliable and sensitive measure of chronic stress, at least in DBCP. We also demonstrate direct associations between psychologic status, immune surveillance functions and stress hormone levels in DBCP.
Chronic stress induces inherent changes in lymphocyte functions, whereas immunologic impairment in acute stress is attributed to transient redistribution and, hence, to a change in the numbers of lymphocyte subpopulations in peripheral blood.37 NCA was tested in various chronic stress situations and found to be decreased compared to matched controls.19, 20 Reduced NK activity in relatives of breast cancer patients has been demonstrated in a few studies, which generally examined cytotoxic activity against NK-sensitive target cells.6, 38–40 In 1 study, NK activity against NK-resistant MCF-7 target cells in 11 female relatives of cancer patients was reduced; however, no association between NCA and psychologic status could be demonstrated.41
Th1 cytokine–stimulated NCA has been scarcely investigated in clinical studies in humans.19, 20 The impaired Th1 cytokine–induced NCA in DBCP found in our study could result from hyporesponsiveness of NK cells to in vitro stimulation by IL-2 or IL-12 due to previous in vivo exposure to high levels of IL-4.42 Elevated levels of IL-4 may be induced by increased levels of catecholamines or cortisol in vivo.22, 43 Although we did not test IL-4 levels, increased catecholamines and cortisol was a very prominent finding in our DBCP group (Fig. 8). Alternatively, as cytokine-induced NCA represents LAK activity,44 it may be hypothesized that the unique signal-transduction pathways involved in LAK activity are impaired in DBCP. This may result from sustained in vivo exposure to high levels of stress hormones.43
Th1 cytokines are vital to the initiation and augmentation of NCA. Decreased in vitro secretion of IFN-γ and IL-2 in chronic stress has already been reported,45–47 whereas IL-12 secretion has not as yet been tested. We found decreased in vitro secretion of IL-2, IFN-γ and IL-12 in DBCP, in particular in DBCP2 (Fig. 7), which was significantly associated with the degree of emotional distress (Fig. 10). As we did not find significant differences in the number of CD3+ T-cell subpopulations between DBCP and controls, the underlying mechanism of decreased in vitro Th1 cytokine secretion in DBCP remains unclear. Using intracellular cytokine staining would have pointed to a possible defect in the secretion and/or synthesis. Increased catecholamine levels may increase intracytoplasmic cAMP levels, which in turn may inhibit the production or secretion of certain Th1 cytokines, like IL-2.48, 49
Elevated levels of plasma cortisol and urinary catecholamines found in DBCP were related to the degree of emotional distress, though Bonferroni correction made the correlations less significant (Fig. 10c). Even a tiny increase in the level of stress hormones may have a significant impact on immune functions.21 This may explain the effect of stress hormones in DBCP1, even though significant differences in plasma cortisol and urinary catecholamine levels were observed only between DBCP2 and controls. However, some other neuroendocrine factors as well as Th1 and Th2 cytokines may affect immune functions22, 43 and in acute stress in humans β-blockade or benzodiazepines may modulate the immunologic response.50, 51
Our DBCP study group was strictly defined and matched with normal controls for age, sociodemographic characteristics and psychologic factors. Thus, the differences found between DBCP and controls are inherently attributed to DBCP. However, the strict definition of the study group imposed some methodologic problems, e.g., difficulties reaching a larger group of DBCP and the possibility that the DBCP who agreed to participate in the study represent a biased group.
Contradictory results have been reported concerning the intensity of emotional distress among DBCP17, 18, 52 or among relatives of breast cancer patients.53, 54 The consistency of increased emotional distress found in our DBCP may be attributed to the homogeneity of the study group. It is not clear whether the high distress level in DBCP is due to the mother's illness or to the awareness to personal risk. The combination of high anxiety and somatization (Fig. 9a,c) points to the possibility that concern for their own health is the main factor in the distress of DBCP.
The association between emotional distress and the risk of breast cancer currently rests on inferences from observational studies and preclinical data. To test this association, we studied immune functions like NCA and Th1 cytokines, which have a specific influence on the health state and defense against cancer.21, 37 Although the association between impaired NCA and the development of breast cancer has not been established, research on humans and other species points to a significant role of NCA in preventing the development of cancer, including breast cancer, and its metastases.1–4, 8, 9 Additional studies document severe emotional distress that precedes the onset of breast cancer.12, 13 We suggest that the combination of a genetic predisposition together with lifestyle10, 11 and enduring emotional distress may impair immune surveillance functions in DBCP and expose them to increased risk of breast cancer onset. Thus, an interventional trial to alter psychologic status and lifestyle may improve immune functions in DBCP and reduce the risk of breast cancer onset.