A Biobehavioral Model of Cancer Stress and Disease Course

Barbara L. Andersen, Janice K. Kiecolt-Glaser, Ronald Glaser. Am Psychol. 1994 May; 49(5): 389–404.

Approximately 1.3 million Americans and over 120,000 Australians (2010) are diagnosed with cancer each year and must cope with the disease and treatments. Many studies have documented the deteriorations in quality of life that occur. These data suggest that the adjustment process is burdensome and lengthy. There is ample evidence showing that adults experiencing other long-term stressors experience not only high rates of adjustment difficulties (e.g., syndromal depression) but important biologic effects, such as persistent down-regulation of elements of the immune system, and adverse health outcomes, such as higher rates of respiratory tract infections. Thus, deteriorations in quality of life with cancer are underscored if they have implications for biological processes, such as the immune system, relating to disease progression and spread. Considering these and other data, a biobehavioural model of adjustment to the stresses of cancer is offered, and mechanisms by which psychological and behavioural responses may influence biological processes and, perhaps, health outcomes are proposed. Finally, strategies for testing the model via experiments testing psychological interventions are offered.

The Cancer Stressor and Psychological Pathways
A cancer diagnosis and cancer treatments are objective, negative events. Although negative events do not always produce stress and an altered quality of life, data from many studies document severe emotional distress accompanying these cancer-related events. Several years ago, we studied gynecologic cancer patients within days of their diagnosis and prior to treatment (Andersen, Anderson, & deProsse, 1989a). Using the Profile of Mood States (POMS; McNair, Lorr, & Droppleman, 1981), we found that the scores for women with cancer were significantly greater than scores from matched women who were awaiting treatment for benign gynecologic diagnoses, which in turn were higher than the stresses of everyday life reported by a matched group of healthy women. These data underscore the acute stress of a life-threatening diagnosis. However, it is also clear that lengthy cancer treatments and the disruptions in major life areas that subsequently occur can produce chronic stress. For example, in a study of 60 male survivors of Hodgkin’s disease, long after treatment had ended, men reported lowered motivation for interpersonal intimacy, increased avoidant thinking about the illness (which is a characteristic of posttraumatic stress), illness-related concerns, and difficulty in returning to predisease employment status (Cella & Tross, 1986). Also, the majority of the patients (57%) reported low levels of physical stamina (Yellen, Cella, & Bonomi, 1993). Other permanent sequelae from cancer treatments, such as sexual problems or sterility (e.g., Kaplan, 1992), which have ramifications for intimate relationships and social support, are well documented (see Andersen, 1986, for a review). Unemployment, under-employment, job discrimination, and difficulty in obtaining health insurance can be problems for a substantial minority (e.g., 40% of bone marrow transplant survivors reported difficulty in obtaining insurance; Winegard, Curbow, Baker, & Piantadosi, 1991). Thus, many chronic stressors (e.g., continued emotional distress, disrupted life tasks, social and interpersonal turmoil, and fatigue and low energy) can occur with cancer.

We have considered the qualities of stressors that not only cause distress and a lowered quality of life, but that are also powerful enough to produce biological changes. Some of the effects may covary with the extent of the disease (which usually necessitates more radical treatment).2 Considering psychological factors, Herbert and Cohen’s (1993b) recent meta-analysis of the stress and immunity literature provides clarification. Their comparison of objective stressful events, such as bereavement, divorce, or caregiver stress, with self-reports of stress, such as reports of hassles, life events, or perceived stress, revealed that greater immune alteration (e.g., lower natural killer [NK] cell activity) occurred with objective events. Furthermore, analyses of stressor duration showed that long-term naturalistic stressors (e.g., bereavement) may have a more substantial impact on NK cell function than do short-term stressors. Also, when events have interpersonal components, they too are related to greater immune alteration than are nonsocial events. The specific immune alterations that were sensitive to these stressor characteristics were NK cell activity, the helper:suppressor ratio, and the percentage of suppressor/cytotoxic T cells. Considering the objective, acutely stressful events of diagnosis and treatment and the chronic and interpersonally disruptive aspects of cancer recovery and survivorship described above, it would appear that cancer as a stressor includes the attributes that have documented linkages to immunity. Lowered NK cell activity may represent one of the more reliable markers of this process.

Summary
Data from healthy samples suggest that stress variables are predictive of immune downregulation, and accumulating data with cancer groups support the same general conclusion. Because the phenomena appear to be reliable, investigators are beginning to examine the clinical relevance of the effect. That is, are these immune changes related to health consequences? Here the data are sparse, but the most controlled analysis, the study by Cohen et al. (1991), shows a direct, replicable relationship between stress and infections and colds. Many researchers question whether the latter findings are relevant to persons with cancer or to those with other chronic illnesses, such as HIV infection or AIDS. The most frequent argument waged against psychological and behavioral effects (even if large) significantly affecting biologic processes is because of the presumed downregulating effects of the disease or the treatments. That is, both our data and those of Levy and colleagues come from correlational designs that tested the contribution of psychological factors, and the directional hypotheses for all variables—psychological–behavioral and disease–treatment effects—are in the same direction, downregulation.

In this context we offer one observation and one suggestion for continued study. First, the directional effects of cancer on the immune system remain to be documented; however, there are sufficient data to state that cancer effects are not unidirectional (i.e., downward), within or across sites. For example, T-cell levels may be slightly decreased, whereas T-cell function appears to be intact in early stage ovarian cancer (Mandell, Fisher, Bostick, & Young, 1979). In contrast, cervical cancer patients often have an increase in B cells and gamma interferon levels, with a decrease in T-cell function (Sharma, Gupta, & Kadian, 1979). Variability on other immune measures, such as delayed hypersensitivity to antigens and contact allergens, can occur across cervical, endometrial, and ovarian cancers (Khoo, MacKay, & Daunter, 1979; Nalick, DiSaia, Rea, & Morrow, 1974). The model we and other psychologists test, however, is in line with the immune theory that posits an influential role of NK cells in host resistance against metastasis (Gorelik & Herberman, 1986). The latter theory rests on the association between depressed immune activity and increased metastases in animal models.

We have proposed a biobehavioral model of variables and pathways believed to lead to immune and health effects. Regarding the health effects, we have discussed disease progression and related variables (e.g., DFI) as endpoints; however, the model may be relevant to other health outcomes, such as a higher incidence of infections and infectious (viral) illnesses (infections are a major obstacle in health care management of cancer patients and also are a major cause of death; Bodey, 1986). Further correlational studies of diagnosed cancer patients would need to be performed to document the reliability of the stress–immunity–cancer link. However, a stronger test would be, of course, to experimentally manipulate a variable in the model. Despite the numerous difficulties entailed, a randomized intervention trial offers a powerful test. A psychological–behavioral intervention is powerful because the prediction for the intervention would be for immune enhancement, more positive health outcomes, or both, in contrast to the prediction for a no-treatment control of downward regulation, poorer health outcomes, or both.

We have reviewed the effects of psychological interventions on moderating immunity in noncancer populations (Kiecolt-Glaser & Glaser, 1992). There we noted that researchers have used diverse strategies to modulate immune function, including relaxation, hypnosis, exercise, self-disclosure, and cognitive–behavioral interventions, and these interventions have generally produced positive changes. For example, our PNI laboratory tested the immune effects of a relaxation training and social contact intervention (Kiecolt-Glaser et al., 1985) with a cancer-relevant sample, elderly adults (mean age of 74 years) living in independent living facilities. Subjects were randomized to (a) relaxation training, (b) social contact, or (c) no contact. Subjects in the two intervention conditions were seen individually three times a week for a month (12 sessions). Blood samples and self-report data were obtained at pretreatment, posttreatment, and one-month follow-up. Analysis of the the psychological data indicated that the relaxation intervention had quality of life effects, as indicated by significant reductions on the Hopkins Symptom Checklist (Derogatis, Lipman, Rickels, Uhlenhuth, & Covi, 1974), a measure of affective distress. Also, all subjects (including controls) reported a significant increase in self-rated quality of sleep, a positive health behavior. Analyses of the immune data indicated that the relaxation intervention produced significant increases (approximately 30%) in NK cell activity (see Figure 1 in Kiecolt-Glaser et al., 1985), with the highest percentage lysis of target cells occurring immediately after treatment.