Carney, Caroline P. MD, MSc; Jones, Laura BS, BA; Woolson, Robert F. PhD; Noyes, Russell Jr., MD, and; Doebbeling, Bradley N. MD, MSc
BC/BS = Blue Cross/Blue Shield;, DSM-IV = Diagnostic and Statistical Manual of Mental Disorders;, ICD-9 = International Classification of Diseases-9;, SEER = Surveillance Epidemiology and End Results Cancer Registry;, OR = odds ratio;, CI = confidence interval.
Descriptions of an association between depression and pancreatic cancer date to the early 1930s when Yaskin described a triad of symptoms including anxiety, sense of impending doom, and depression in patients diagnosed with pancreatic cancer (1). Although guidelines for assessing and treating psychiatric conditions that develop after persons have been diagnosed and/or are under treatment for pancreatic cancer have been published (2, 3). Most research, however, describes an association between the conditions cited below and refers to psychiatric symptoms present before or at the time of diagnosis of pancreatic cancer (2, 3). This research is briefly reviewed below.
Early case reports and retrospective studies may have misattributed symptoms of pancreatic malignancies (ie, weight loss, pain) to depression (4–8). The authors of a retrospective chart review comparing patients treated for pancreatic and gastric cancers found depression in 14% of pancreatic cancer patients but in only 4% of stomach cancer patients (9).
Later research compared the presenting symptoms of patients with pancreatic tumors and other abdominal malignancies. Patients with pancreatic tumors had psychological symptoms occurring as early as 43 months before somatic symptoms. Compared with those with other abdominal malignancies, pancreatic cancer patients more commonly reported symptoms of anxiety, loss of ambition, and depression (76% vs. 20% of respondents) (10). In an elegantly designed study, controlling for demographic and medical characteristics, Holland et al. found that patients with pancreatic cancer had higher self-ratings of depression, tension-anxiety, fatigue, confusion-bewilderment, and total mood disturbance compared with patients with other advanced abdominal neoplasms (11).
Studies describing the association between pancreatic cancer and psychiatric symptoms have all been limited to clinical samples of persons with newly or recently diagnosed malignancies. Thus, a patient’s recollection of past psychiatric symptoms may be colored by the cancer diagnosis, resulting in recall bias.
Whether an association between these conditions exists in the general population has not been explored. The issue is important for several reasons. It is possible that the previously described association is spurious, the result of biased sampling methods. A valid association would lead one to pursue etiological links between the two conditions. The question that needs answered is whether depression or its treatments cause pancreatic cancer by altering protective immune functioning or whether pancreatic cancer itself causes depression. For instance, is there a subtype of depression that is the result of the hormonal changes brought on by the malignancy? Could alterations in gastrointestinal tract hormones caused by this or other conditions be manifest as depression? Alternatively, should new onset depression in certain populations be considered a clinical heralding sign of pancreatic cancer?
As part of a larger population-based study evaluating the association between mental disorders and incident malignancies, we examined whether an association between pancreatic cancer and mental disorders occurs in a general population of insured persons. Although our study is not able to answer whether a causal relationship exists between the two conditions, we sought to determine whether an association is present using data that were not influenced by recall bias. We also investigated whether depression occurred more frequently in persons later diagnosed with pancreatic cancer compared with other malignancies, including those of the gastrointestinal tract.
This study is based on all (100%) Wellmark Blue Cross/Blue Shield of Iowa and South Dakota (Wellmark) inpatient, outpatient, and provider claims data from January 1, 1989 to December 31, 1993. Enrollment files were not available. The study is considered population-based because during these years Wellmark insured nearly one half of the population of Iowa and the eastern South Dakota counties where it has penetration.
Subjects for this investigation were all 18- to 64-year-old men and women who filed claims for mental disorders and/or pancreatic cancer during the years of interest. The mental health cohort was composed of subjects with mental disorders occurring before pancreatic disorders. The control cohort was composed of subjects with pancreatic cancer who never filed claims for mental disorders at any time during the 5 years. Proportions were calculated using patients with mental disorders before pancreatic cancer over all patients with mental disorders before all cancers compared with all patients with pancreatic cancer and no mental disorders over all patients with all cancers and no mental disorders.
Nearly 93% of the sample was composed of residents of Iowa (12). Subjects 65 years of age and older were excluded given low representation in the data set due to Medicare reimbursement. Subjects younger than 18 years were excluded given low rates of psychiatric disorders and cancer.
The University of Iowa’s Institutional Review Board approved the study. Data were coded and individuals could not be identified.
Classification of Mental Disorders
Classification of subjects into the mental disorder cohort was based on criteria proposed by Lurie et al. for work with administrative databases (13). In order to assure the best specificity of a mental disorder into a major diagnostic class, such as mood disorders as defined by the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV), patients must have filed:
1. at least one inpatient psychiatric claim as the primary or secondary diagnoses filed during 1989 to 1993, or
2. two outpatient mental health claims made by any provider in any position filed during 1989 to 1993, or
3. any single mental health claim filed by a psychiatrist during 1989 to 1993 (13, 14).
Comparisons were limited to the subjects whose psychiatric claims were filed before the cancer diagnosis. Patients who never filed a claim for a mental disorder or who had only one outpatient psychiatric claim were classified as nonpsychiatric patients and considered in the comparison group. Subjects with psychiatric claims first filed after the cancer diagnosis were excluded from the analyses in order to avoid bias resulting from conditions that were more possibly related to the malignancy.
ICD-9 codes were mapped to broad diagnostic categories (ie, psychotic disorders) according to classification by the DSM-IV. This diagnostic manual has organized mental disorders into 16 major diagnostic classes. The official coding system used in the DSM-IV is the ICD-9 (11). ICD-9 codes selected from the claims data (290.00–319.00, 607.84, 608.89, 625.00, 625.80, 780.09, 780.52, 780.54, 780.59, 787.60) were sorted into the major diagnostic classes according to the organizational system provided by the DSM-IV. Specific diagnoses (ie, schizophrenia) were also considered. In cases in which an ICD-9 code was not listed in the DSM-IV, the condition was designated in an “other” category. Patients were classified into a single category based on their initial mental health claim (14, 15).
Classification of Cancer
Malignancies were categorized based on primary site. Codes for metastatic cancer sites were excluded as we were interested in determining incidence of primary malignancies. ICD-9 codes were grouped into major diagnostic categories (ie, gastrointestinal) designated by the National Cancer Institute’s SEER program (16). Pancreatic cancer was identified by codes 157.00 to 157.90. Gastrointestinal cancers included esophageal, stomach, small intestine, colon, rectum, liver, gallbladder, and other biliary malignancies (150.0–159.0; 159.8–159.9). The validity of cancer diagnoses in the Wellmark Blue Cross/Blue Shield data has been documented in a linkage study with the SEER State Health Registry of Iowa (17).
Statistical analyses included comparison of means and proportions using chi-square analyses for categorical variables and t tests for continuous variables. Fisher’s exact tests were used in comparisons with cell sizes of less than five. ORs were calculated to determine differences in incident cancers between cases and controls. The Mantel-Haenszel estimator and its 95% CI were calculated to adjust for age. The type I error rate was set at 0.05 and p values were two-tailed. Analyses were conducted with SAS, version 8.2.
Overall, 748,466 people filed claims during the 5 years of observation. The annual dropout rate was approximately 7% in both psychiatric and nonpsychiatric samples, suggesting that there was no influence of the mental illness on dropout rates; 74,959 (10%) individuals met criteria for a mental disorder. Of this group, 1,399 (1.9%) developed malignancies following the diagnosis of the psychiatric condition. The control cohort was composed of 673,507 individuals who filed claims during the 5-year study period but did not meet the defined psychiatric criteria. Of this group, 18,083 (2.7%) developed some type of malignancy.
Nine men and five women developed pancreatic cancer following the mental illness designation, representing 1% of all patients who developed cancer after mental health claims were filed. In comparison, 60 men and 41 women without insurance claims for psychiatric conditions were diagnosed with pancreatic cancer during the same period, representing 0.6% of the sample without mental illness, but who developed malignancies (OR 1.99, CI 1.13–3.50, p = 0.0144). Men with mental disorders were more likely to develop pancreatic cancer than those without mental disorders (OR 2.4, CI 1.15–4.78, p = 0.015). In contrast, no differences were noted for women (OR 1.6, CI 0.64–4.11).
The categories of those persons who were diagnosed with a mental illness before receiving the pancreatic cancer diagnosis were further analyzed. Ten of the 14 cases (71%) had been diagnosed with major depression. Other mental disorder diagnoses assigned before pancreatic cancer included anxiety disorder not otherwise specified (N = 1), psychotic disorder not otherwise specified (N = 2), and alcohol-related disorder (N = 1).
Among those subjects with depression diagnosed before the development of cancer (N = 485), we calculated the relative odds of developing pancreatic cancer (N = 10) compared any cancer of the gastrointestinal system (N = 39) or any other kind of cancer (N = 475). Pancreatic cancer also was diagnosed more commonly than other gastrointestinal malignancies in patients with preexisting depression (OR 4.56, CI 1.07–19.38, p = 0.03). Subjects with depression were also more likely to be diagnosed with pancreatic cancer compared with developing any other kind of cancer following claims for major depression (OR 4.10, CI 1.05–16.0, p = 0.028).
Given that depressive symptoms such as weight loss and fatigue might be manifestations of pancreatic cancer, we evaluated each case to determine the length of time between the mental health claim and the first claim for pancreatic cancer. The median number of days between psychiatric diagnosis and cancer was 553 days (range 44–1531 days). Only 3 of 14 cases had the psychiatric condition diagnosed within 6 months of the cancer diagnosis (44, 63, and 116 days each).
In this population-based assessment, pancreatic cancer occurred more commonly in insured persons with psychiatric conditions than in persons without such conditions. In those patients with psychiatric conditions who later develop pancreatic cancer, the majority had been diagnosed with depression. Our methods obviate the possibility of potential recall that may have influenced previous research on patients who complete interviews at the time of the cancer diagnoses. We also found that depression more commonly antedates the development of pancreatic cancer than it does gastrointestinal cancers and all other nonpancreatic cancers, supporting a mechanism uniquely related to pancreatic malignancies. Further, our data suggest that the interval between the mental health diagnosis and the first claim for pancreatic cancer was nearly 1.5 years. Given that the time from onset of symptoms to diagnosis of pancreatic cancer on average is 10 weeks, it is unlikely that the major depression diagnosed in all but three of these subjects was really misattributed physical symptoms of pancreatic cancer (18, 19).
Whether or not a unique biological relationship exists between pancreatic cancer and depression remains unknown. Several theories have been advanced that support a connection between the two conditions. Two theories involve serotonin. The first is immunological and postulates that pancreatic tumor cells secrete antibodies that either block central nervous system serotonin receptors or reduce the synaptic availability of serotonin (20, 21). Pancreatic tumors also have been shown to have increased secretion of serotonin. It has been postulated that this exuberant metabolism depletes central nervous system stores of serotonin, thereby resulting in depression (21). Another proposed mechanism involves a paraneoplastic syndrome that leads to nonsuppression of the dexamethasone suppression test (22). However, nonsuppression is not unique to depressed patients with pancreatic tumors, and the evidence linking these two is limited to case reports (23, 24). Other cancers, especially squamous cell lung cancer, are associated with paraneoplastic syndromes that may be first manifest with psychiatric symptoms. Finally, malignancies of the neuropeptide-rich gastrointestinal tract may secrete biogenic amines that alter mood (11, 25, 26).
When considering the association between mental disorders such as depression as causal factors for the development of malignancies, one must recognize that confounding factors may influence the results. Perhaps the most important of these factors for patients with mental illness is cigarette smoking. At least 56 studies report higher rates of cigarette smoking among patients with psychiatric diagnoses than among controls, including rates of at least 50% in persons with depressive, anxiety, alcohol, and schizophrenic disorders. The prevalence of pancreatic cancer is 1.4 to 6 times higher in smokers than in nonsmokers (27–30). Whether or not tobacco use among the depressed subjects accounted for the difference in incidence between cases and controls in this study is not known. Accurate smoking data are not available from this data source, so we are unable to measure the possible effect of smoking in our subjects. Other risk factors implicated in the etiology of pancreatic cancer include high fat diet and obesity (27, 31–33). Because patients with psychiatric symptoms may have a higher prevalence of obesity compared with the general population, this factor must also be considered when interpreting these findings (34–38). As with tobacco use, we are unable to discern whether cases with depression and pancreatic cancer had higher rates of obesity compared with controls.
Whether depression itself predisposes the development of malignancies remains an area of debate (39). Depression has been implicated in immune dysfunction through various mechanisms, including diminished cellular activity, lowered number of white blood cells, increased cortisol levels that impair healing, and defects in DNA repair of cells. Depression and stress may alter immune functioning, but it is important that other factors are considered before assuming depression itself is the cause. For instance, recent work suggests that depression-related mechanisms such as increased smoking, alcohol use, physical inactivity, or impaired nutritional intake may play an important role in immune functioning (40–46). Fox reviewed early epidemiological studies, finding that seven showed a positive relationship between depression or stressful conditions and later incidence of cancer or mortality from cancer, and seven concluded an absence of any relationship (47). Fox later reviewed five cohort studies published since 1980 and found no association between mental stress and cancer (48). Importantly, in a 13-year follow-up sample of community-dwelling depressed adults living in the Baltimore Epidemiological Catchment Area, no overall association between depression and cancer incidence was found. However, major depression was associated with the onset of breast cancer (49). In contrast, neuroticism and low life satisfaction were not associated with breast cancer risk in Finnish women (50). At best, our findings with respect to depression and cancer onset are inconsistent, probably secondary to disparate methodologies. Studies involving populations may also be unable to control for potentially important variables such as tobacco use, family history, or other carcinogenic exposures.
A limitation of research involving claims data merits mention. In any population, patients may be suffering from mental disorders that have gone undiagnosed by providers. Consequently, the prevalence of a mental disorder such as major depression is underestimated by claims data. Using the strict diagnostic criteria described in “Methods,” we found a 6.3% prevalence of mood disorders in this population, which likely represents an underestimate of the true prevalence of major and minor depression in the population (12, 51, 52). This, however, makes it likely that our comparison group was enriched by persons with undiagnosed psychiatric conditions, including mood disorders, before the development of cancer. However, given that only 5% met the criteria for being a psychiatric case after the cancer diagnosis, the control group seems unlikely to have been enriched by psychiatric disorders. Further, if subjects who truly had had depression but had not filed claims were moved into the mental health cohort, this likely would have only strengthened the association between the conditions. Finally, because persons who are treated for depression have a greater symptom burden than those were not treated, the association we observed between depression and cancer may have been influenced by a dose-effect relationship (53).
The sample size of subjects with mental disorders coded before the development of pancreatic cancer is small (N = 14 in mental health cohort; 101 cases in comparison cohort). The wide confidence intervals reflect these small samples. Importantly, these low numbers are consistent with the incidence of pancreatic cancer in the U.S. general population. According to the National Cancer Institute, the age-adjusted annual rates per 100,000 individuals were 12.1 cases in males and 9.5 cases in females (16). Our overall population who filed claims during 1989 to 1993 was composed of 748,466 adults. Given that 115 individuals filed claims for primary pancreatic cancer over the 5 years, the annual rate of pancreatic cancer in this sample 3.1/100,000. The low rate is likely secondary to the exclusion of subjects 65 years and older. Given the small sample size, caution should be given when interpreting these findings.
Claims data allow researchers to evaluate subjects only during the window of time that is available for study. In this case, we had 5 consecutive years of data. Although this was enough time to show that an association between the two conditions was present, it likely was not long enough to determine whether depression was a causal factor in the development of pancreatic cancer. For this to be studied, a long-term cohort study would need to be performed in order to make a determination about whether the burden of depression, duration of treatment, and treatment modalities were related to the development of malignancies. Unfortunately, these types of studies are extremely expensive and may take years of follow-up before associations can be made, especially for rare conditions such as pancreatic cancer.
Finally, these findings may not generalize to uninsured persons who lack access to medical and psychiatric services. However, previous studies have shown the Wellmark data to be representative of the state’s population for those younger than 65 years and the cancer diagnostic codes to be valid (17). These data include subjects 18 to 64 years of age. Given that rates of cancer increase with age, it is possible that the findings would not hold true for an older population. Pancreatic cancer is associated with advancing age, such that the majority of cases are diagnosed in persons older 64 years (16). One important finding of this study is that the results were age-adjusted. Although this cannot compensate for the small sample under study, it does remove the effect of age from the findings.
These data highlight the importance of population-based epidemiological studies in evaluating associations between conditions, especially rare conditions such as pancreatic cancer. Physicians may be likely to believe that associations exist between conditions when only clinical samples have been evaluated. Claims data provide a rich source of information with which to study relationships between conditions in naturalistic settings.
Importantly, these data suggest that pancreatic cancer occurs more often in patients who have had clinically recognized depression. This finding may not be secondary to only the physical symptoms of cancer itself, as we would have expected to see a similar finding when comparing pancreatic cancer to other gastrointestinal malignancies, and all other malignancies. The association between depression and pancreatic cancer remains unique and invites further investigation in order to determine which direction a causal relationship may exist. TABLE
This research was supported by NIMH K08 MH01932-01A1. The authors thank Dr. Sheila McGuire Riggs, Vice-President and Senior Epidemiologist of Wellmark Blue Cross Blue Shield of Iowa and South Dakota for her support.
1. Yaskin J. Nervous symptoms at earliest manifestations of cancer of the pancreas. JAMA 1931; 96: 1664–8.
2. Alter CL. Palliative and supportive care of patients with pancreatic cancer. Semin Oncol 1996; 23: 229–40.
3. Passik SD, Breitbart WS. Depression in patients with pancreatic carcinoma: diagnostic and treatment issues. Cancer 1996; 78: 615–26.
4. Savage CB, Noble D. Psychiatric manifestations in pancreatic disease. J Clin Psychopathol 1952; 13: 9–16.
5. Latter K. Psychic and neurological manifestations of carcinoma of the pancreas. Mayo Clin Proc 1937; 12: 457–62.
6. Perlas AF. Psychiatric manifestations of carcinoma of the pancreas. Am J Psychiatry 1964; 121: 182.
7. Karlinger W. Psychiatric manifestations of cancer of the pancreas. NY State J Med 1967; 56: 2251–2.
8. Wallen GC, Gittleson NL. A case of carcinoma of the pancreas with a psychiatric presentation. Br J Clin Pract 1972; 26: 132–3.
9. Jacobsson L, Ottosson JO. Initial mental disorders in carcinoma of pancreas and stomach. Acta Psychiatr Scand Suppl 1971; 221: 120–7.
10. Fras IL, Pearson JS. Comparison of psychiatric symptoms in carcinoma of the pancreas with those in some other intra-abdominal neoplasms. Am J Psychiatry 1967; 123: 1553–62.
11. Holland JC, Korzun AH, Tross S, Silberfarb P, Perry M, Comis R, Oster M. Comparative psychological disturbance in patients with pancreatic and gastric cancer. Am J Psychiatry 1986; 143: 982–6.
12. Carney CP, Woolson RF, Doebbeling BN. Access to mental health care in a rural state. Selected abstracts of the Society of General Internal Medicine Annual Meeting, J Gen Intern Med 2002; 17 (Suppl. 1): 186.
13. Lurie N, Popkin M, Dysken M, Moscovice I, Finch M. Accuracy of diagnoses of schizophrenia in Medicaid claims. Hosp Commun Psychiatry 1992; 43: 69–71.
14. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. Washington DC, American Psychiatric Association; 1994.
15. Carney CP, Woolson RF, Doebbeling BN. Mental health diagnostic patterns in rural state. Selected Abstracts of the Society of General Internal Medicine Annual Meeting, Poster 64, Atlanta, GA. J Gen Intern Med 2002; 17 (Suppl. 1): 110.
16. National Cancer Institute. Surveillance, epidemiology, and end results (SEER) Program; 1999.
17. Doebbeling B, Wyant D, McCoy K, Riggs S, Woolson R, Wagner D, Wilson R, Lynch C. Linked insurance-tumor registry database for health services research. Med Care 1999; 37: 1105–13.
18. Allen-Mersh TER. Pancreatic cancer in England and Wales: Surgeons look at epidemiology. Assoc R Coll Surg Engl 1986; 68: 154–8.
19. Tilleman EBT, Gourma DJ. Differences in diagnostic work-up for pancreatic carcinoma between centers and general hospitals. Presentation at Annual Meeting of the Society for Surgery of the Alimentary Tract, San Deigo, CA, 2000.
20. Jacobsson LO. Initial mental disorders in carcinoma of pancreas and stomach. Acta Psychiatr Scand 1971; 220: 120–7.
21. Brown JH, Paraskevas F. Cancer and depression. Cancer presenting with depressive illness: an autoimmune disease? Br J Psychiatry 1982; 141: 227–32.
22. Joffe RT, Rubinow DR, Denicoff KD, Maher M, Sindelar WF. Depression and carcinoma of the pancreas. Gen Hosp Psychiatry 1986; 8: 241–5.
23. Joffe RT, Adsett CA. Depression and carcinoma of the pancreas. Can J Psychiatry 1985; 30: 117–8.
24. Pomara N, Gershon S. Treatment-resistant depression in an elderly patient with pancreatic carcinoma: case report. J Clin Psychiatry 1984; 45: 439–40.
25. Asa SL, Lee YC, Drucker DJ. Development of colonic and pancreatic endocrine tumours in mice expressing a glucagons-SV40 T antigen transgene. Virchows Arch 1996; 427: 595–606.
26. El-Salhy M, Mahdavi J, Norrgard O. Colonic endocrine cells in patients with carcinoma of the colon. Eur J Gastroenterol Hepatol 1998; 10: 517–22.
27. Isaksson B, Jonsson F, Pedersen NL, Larsson J, Feychting M, Permert J. Lifestyle factors and pancreatic cancer risk: a cohort study from the Swedish Twin Registry. Int J Cancer 2002; 98: 480–2.
28. Maruchi WB, Ludwig J, Elveback LR, Kurland LT. Cancer of the pancreas in Olmstead County Minnesota 1935–1974. Mayo Clinic Proc 1979; 54: 245.
29. Gold EB. Epidemiology of and risk factors for pancreatic cancer. Surg Clin North Am 1995; 75: 819–43.
30. Fraumini J. Cancers of the pancreas and biliary tract: epidemiological consideration. Cancer Res 1975; 35: 34–37.
31. Farrow DC, Davis S. Diet and the risk of pancreatic cancer in men. Am J Epidemiol 1990; 132: 423–31.
32. Mills PK, Beeson WL, Abbey DE, Fraser GE, Phillips RL. Dietary habits and past medical history as related to fatal pancreas cancer risk among Adventists. Cancer 1988; 61: 2578–85.
33. Norell SE, Ahlbom A, Erwald R, Jacobson G, Lindberg-Navier I, Olin R, Tornberg B, Wiechel KL. Diet and pancreatic cancer: a case-control study. Am J Epidemiol 1986; 124: 894–902.
34. Wadden TA, Sarwer DB, Womble LG, Foster GD, McGuckin BG, Schimmel A. Psychosocial aspects of obesity and obesity surgery. Surg Clin North Am 2001; 81: 1001–24.
35. Lamberg L. Psychiatric help may shrink some waistlines. JAMA 2000; 284: 291–3.
36. Britz B, Siegfried W, Ziegler A, Lamertz C, Herpertz-Dahlmann BM, Remschmidt H, Wittchen HU, Hebebrand J. Rates of psychiatric disorders in a clinical study group of adolescents with extreme obesity and in obese adolescents ascertained via a population based study. Int J Obes Relat Metab Disord 2000; 24: 1707–14.
37. Clark MM, Niaura R, King TK, Pera V. Depression, smoking, activity level, and health status: pretreatment predictors of attrition in obesity treatment. Addict Behav 1996; 21: 509–13.
38. Ford DE, Mead LA, Chang PP, Cooper-Patrick L, Wang NY, Klag MJ. Depression is a risk factor for coronary artery disease in men: the precursors study. Arch Intern Med 1998; 158: 1422–6.
39. Mathe G. Depression, stressful events and the risk of cancer. Biomed Pharmacother 1996; 50: 1–2.
40. Anderson JL. The immune system and major depression. Adv Neuroimmunol 1996; 6: 119–29.
41. Irwin M. Immune correlates of depression. Adv Exp Med Biol 1999; 461: 1–24.
42. Maes M, Thompson P. Analysis of partial variance to control for day-to-day variability in functional immune tests in depression. Neuropsychobiology 1997; 36: 107–11.
43. Maes M. Major depression and activation of the inflammatory response system. Adv Exp Med Biol 1999; 461: 25–46.
44. Miller GE, Cohen S, Herbert TB. Pathways linking major depression and immunity in ambulatory female patients. Psychosom Med 1999; 61: 850–60.
45. Natelson BH, Denny T, Zhou XD, LaManca JJ, Ottenweller JE, Tiersky L, DeLuca J, Gause WC. Is depression associated with immune activation? J Affect Disord 1999; 53: 179–84.
46. Zaharia MD, Ravindran AV, Griffiths J, Merali Z, Anisman H. Lymphocyte proliferation among major depressive and dysthymic patients with typical or atypical features. J Affect Disord 2000; 58: 1–10.
47. Fox BH. Epidemiologic aspects of stress, aging, cancer and the immune system. Ann NY Acad Sci 1988; 521: 16–28.
48. Fox BH. Depressive symptoms and risk of cancer. JAMA 1989; 262: 1231.
49. Gallo JJ, Armenian HK, Ford DE, Eaton WW, Khachaturian AS. Major depression and cancer: the 13-year follow-up of the Baltimore epidemiologic catchment area sample (United States). Cancer Causes Control 2000; 11: 751–8.
50. Lillberg K, Verkasalo PK, Kaprio J, Helenius H, Koskenvuo M. Personality characteristics and the risk of breast cancer: a prospective cohort study. Int J Cancer 2002; 100: 361–6.
51. Lepine JP, Gastpar M, Mendlewicz J, Tylee A. Depression in the community: the first pan-European study DEPRES (Depression Research in European Society). Int Clin Psychopharmacol 1997; 12: 19–29.
52. Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Eshleman S, Wittchen HU, Kendler KS. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: results from the National Comorbidity Survey. Arch Gen Psychiatry 1994; 51: 8–19.
53. Coryell W, Endicott J, Winokur G, Akiskal H, Solomon D, Leon A, Mueller T, Shea T. Characteristics and significance of untreated major depressive disorder. Am J Psychiatry 1995; 152: 1124–9.