Psychological symptoms and disease-free and overall survival in women with stage II breast cancer. Cancer and Leukemia Group B.

BACKGROUND: The possible link between psychological factors and length of cancer survival has generated a literature of contradictory findings. Associations usually have not been found when general psychological symptoms are assessed. Associations usually have been found for predictors related to expressive versus repressive emotional coping (e.g., depression, "fighting spirit," hostility, and type C personality); however, even these associations have been relatively small, when compared with those for medical factors. Yet few studies have adequately controlled for medical and treatment-related factors. PURPOSE: Within a Cancer and Leukemia Group B (CALGB) national clinical trial of four adjuvant therapy regimens for stage II breast cancer (CALGB 8082), this study prospectively examined the contribution of potential psychological predictors to length of disease-free and overall survival over a 15-year period. METHODS: Subjects were 280 women with stage II breast cancer, out of a total of 899, who were randomly assigned to receive CMFVP (cyclophosphamide-methotrexate-fluorouracil-vincristine-prednisone) for two 6-week cycles or six 4-week cycles, then subsequently randomly assigned to receive or not to receive VATH (vinblastine-doxorubicin-thiotepa-fluoxymesterone). Subjects were recruited during the period between October 1980 and August 1984, inclusive, and followed until January 1996. Prior to chemotherapy, psychological symptoms were assessed using the Symptom Check List-90-Revised (SCL-90-R). SCL-90-R scores were trichotomized into categories representing high, medium, and low distress. Basic base-line sociodemographic data (including age, ethnicity, education, and marital status) and medical data (including lymph node status, estrogen receptor status, menopausal status, and performance status) were collected. Subjects with psychosocial data differed from those without psychosocial data solely in their higher percentage of classification in the mild limitation category of the Zubrod (Eastern Cooperative Oncology Group) performance status rating (subjects with psychosocial data: 14%; subjects without psychosocial data: 8%). RESULTS: In stepwise Cox regression analyses that controlled for sociodemographic and medical variables, there was no significant predictive effect of the level of distress (as measured by the SCL-90-R trichotomized scores) on length of disease-free and overall survival of the study subjects. Risk ratios for low versus high distress were 1.01 (95% confidence interval [CI] = 0.62-1.66) for disease-free survival and 1.03 (95% CI = 0.58-1.82) for overall survival. CONCLUSIONS: This study failed to provide evidence that psychological factors contributed to length of disease-free or overall survival of women who received adjuvant chemotherapy (either CMFVP alone or CMFVP followed by VATH) for treatment of stage II breast cancer. IMPLICATIONS: In the context of far more potent medical factors, the contribution of psychological factors to disease-free and overall survival is likely to be relatively small. Future research should focus on specific theory-driven predictors rather than on general psychological symptoms. Moreover, it should be based on clinical studies using a controlled, prospective design, in which the effects of medical factors may be distinguished and psychological predictors are clear antecedents of survival outcomes.

Duke Scholars

Author James Emmett Herndon II Biostatistics & Bioinformatics, Division of Biostatistics