This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Steel, J. L. Articles by Carr, B. I. Search for Related Content PubMed Articles by Steel, J. L. Articles by Carr, B. I. Related Articles Related Correspondence

In Reply

University of Pittsburgh School of Medicine, Pittsburgh, PA

Marion C. Olek

University of Pittsburgh Medical Center, Pittsburgh, PA

Brian I. Carr

University of Pittsburgh School of Medicine and Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, PA

First, we appreciate the opportunity to respond to Coyne and Palmer's comment on our article entitled, "Depression, Immunity, and Survival in Patients with Hepatobiliary Carcinoma,"¹ and we are pleased the article generated interest and a response from its readers.² Understanding the prevalence and psychologic and biologic factors contributing to the development of depression in the context of cancer, as well as the potential influence of depression on health outcomes, warrants further research and debate.

Coyne and Palmer raise several important issues, many of which we agree with, including (1) that the Center for Epidemiological Studies-Depression scale does not reflect Major Depressive Disorder (MDD) and is the motivation for using the terminology "depressive symptoms" rather than MDD throughout the article. As a result of this distinction, we compared our prevalence rates of "depressive symptoms" with general and medical populations samples using the Center for Epidemiological Studies-Depression not samples with a diagnosis of MDD (Tables 3 and 4 in Steel et al¹). (2) Depressive symptoms may serve as a "risk marker" rather than "causal factor" for survival, which is one possible explanation of our findings; (3) the treatment of depression is a realistic goal in a cancer center, and evidence-based treatments should be available to patients; (4) health care providers should not convey to patients that psychologic treatment or maintaining a positive attitude (eg, fighting spirit) will enhance patients' immune system functioning or survival. Treatment to reduce depressive symptoms or psychologic distress should be a goal to improve quality of life and is as important of an outcome as survival; and (5) as Coyne and Palmer suggest, that perceived or objective health status (eg, exogenous depression) may contribute to depressive symptoms; however, there are likely other subtypes of depression that may involve biologic underpinnings (eg, endogenous depression). Analogous to pancreatic and colorectal carcinoma, a percentage of patients diagnosed with hepatobiliary carcinoma may experience the onset of depressive symptoms long before their diagnosis of cancer.^3,4

Unfortunately, there were several points that were expressed where we disagree with Coyne and Palmer. First, Coyne and Palmer reported in their Correspondence that we controlled for only "sociodemographic variables" when, more importantly, we included disease-specific factors as covariates that may influence survival (eg, size of tumor at diagnosis, number of lesions, vascular invasion). Our study addressed this as well as several other limitations of previous research by employing a prospective design and a standardized instrument to assess depressive symptoms; excluding other psychiatric symptoms or measurements of global psychological distress; controlling for the somatic symptoms associated with depression; and investigating the association between depression and survival in a virally-related cancer that is more likely to be immune-mediated.^5-21

Second, the examination of our data by Coyne and Palmer, done by calculating the "unadjusted odds ratios" from Figure 1 in Steel et al¹ to test the association between depressive symptoms and survival, is methodologically and statistically erroneous. Unadjusted odds ratios used in this context are uninterpretable as the length of follow-up for each patient varies and is not accounted for when using odds ratios. For decades, there have been well-known and readily available statistical methods to analyze survival that addresses the problem of variable follow-up (eg, Cox regression, Kaplan-Meier;^22,23). In addition, these methods also control for the variance associated with confounding factors and allow the use of censored data and time-dependent covariates unlike odds ratios.^22-24

Furthermore, Coyne and Palmer stated that "the regression they perform is likely overfitted, given the number of deaths being explained." First, 72% of the patients in our sample had died at the time of the analyses. As cited by Coyne, Babyak suggested that 10 to 15 observations per predictor should be included in a Cox regression analysis.^24,25 Accordingly, we included 10 predictors for our sample of 101 patients, thereby resulting in a comprehensive model while avoiding overfitting the data. The second analysis, which was exploratory, did not include an adequate number of observations per predictor. However, due to the significant associations found between depression and survival with natural killer (NK) cell number, further research is warranted to test this mediational model with a larger sample size. Coyne and Palmer also stated "the choice of covariates capitalized on chance by prescreening for significance." We would like to clarify that we did not statistically prescreen predictor variables for significance, but rather included all sociodemographic and disease-specific variables we had available and that, theoretically, were assumed to be associated with survival.

Finally, it does not appear that Coyne and Palmer understood the exploratory results concerning the mediational model that was tested ("Steel et al also report an analysis purported to demonstrate that NK cell number mediates the relationship between depression and mortality among a subsample of 23 individuals"). Using Baron and Kenny's criteria,²⁶ we tested the model of depressive symptoms and survival with NK cell number as a potential mediator but reported that the model was not supported. This exploratory model was likely not supported secondary to the small sample size. Other possible explanations may include: (1) other underlying biologic mechanisms may explain the relationship between depression and survival, or (2) depressive symptoms and changes in immunity are simultaneously associated with survival.

The results of our research are consistent with a recent consensus statement from the National Institutes of Health, in which Coyne was a coauthor. The consensus statement entitled "Mood Disorders in the Medically Ill: Scientific Review and Recommendations"²⁷ concluded that "a growing body of evidence suggests that biologic mechanisms underlie a bidirectional link between mood disorders and many medical illnesses. In addition, there is evidence to suggest that mood disorders affect the course of medical illnesses. Further prospective studies are warranted."²⁷ Our ongoing research will provide further data to test the mediational model between depression, immunity, and survival in patients with hepatobiliary carcinoma with a larger sample. Our team plans to investigate not only NK cell number and cytotoxicity, but also other potential mediators such as cytokines and other biologic markers that may be associated with both psychological factors and survival in patients diagnosed with hepatobiliary carcinoma.

We also agree with the conclusions of the consensus statement that suggest that "biologic mechanisms underlie a bidirectional link between mood disorders and many medical illnesses." Our team is currently testing an evidence-based intervention designed to reduce cancer-related symptoms to better understand the bidirectional relationship between cancer-related symptoms and biologic markers. To conclude, we look forward to continued research and debate from our colleagues to further understand this complex relationship between cancer-related symptoms, immunity, and survival in the context of cancer as well as other medical illnesses.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The author(s) indicated no potential conflicts of interest.

REFERENCES

1. Steel JL, Geller DA, Gamblin TC, et al: Depression, immunity, and survival in patients with hepatobiliary carcinoma. J Clin Oncol 25:2397-2405, 2007[Abstract/Free Full Text]

2. Coyne JC, Palmer SC: More claims about depression, immune function, and survival that exceeds the evidence. J Clin Oncol doi:10.1200/JCO.2007.13.8594

3. Passik SD, Breitbart WS: Depression in patients with pancreatic carcinoma–diagnostic and treatment issues. Cancer Supplement 78:615-626, 1996

4. Allen-Mersh TG, Glover C, Fordy C, et al: Relation between depression and circulating immune products in patients with advanced colorectal cancer. J R Soc Med 91:408-413, 1998[Abstract]

5. Shirai M, Watanabe S, Nishioka M: Depressed lymphokine-activated killer activity and analyses of the precursor cells in peripheral blood of patient with hepatocellular carcinoma. Hepatogastroenterology 37:465-468, 1990[Medline]

6. Shirai M, Watanabe S, Nishioka M: Intramural injection of OK432 and lymphokine-activated killer activity in peripheral blood of patients with hepatocellular carcinoma. Eur J Cancer 26:965-996

7. Shirai M, Watanabe S, Nishioka M: Defective immunological functions associated with abnormal lymphokine-activated killer activity in patients with hepatocellular carcinoma. J Gastroenterol Hepatol 5:542-548, 1990[Medline]

8. Wang Y, Chen H, Wu M, et al: Postoperative immunotherapy for patients with hepatocarcinoma using tumor-infiltrating lymphocytes. Chin Med J 110:114-117, 1997[Medline]

9. Haruta I, Yamauchi K, Aruga A, et al: Analytical study of the clinical response to two distinct adoptive immunotherapies for advanced hepatocellular carcinoma: Comparison between LAK cell and CTL therapy. J Int Assoc Physicians 15:451-460, 1996

10. Taketomi A, Shimada M, Shirabe K, et al: Natural killer cell activity in patients with hepatocellular carcinoma. Cancer 83:58-63, 1998[CrossRef][Medline]

11. Actis GC, Ponzetto A, D'Urso N, et al: Chronic active hepatitis. Interferonactivated natural killer like cells against a hepatoma cell line transfected with the hepatitis B virus nucleic acid. Liver 11:106-113, 1991[Medline]

12. Reference deleted.

13. Kawarabayashi N, Seki S, Hatsuse K, et al: Decrease of CD56+ cell and natural killer cells in cirrhotic livers with hepatitis C may be involved in their susceptibility to hepatocellular carcinoma. Hepatol 32:962-969, 2000[CrossRef][Medline]

14. Nakajima T, Mizushima N, Kanai K: Relationship between natural killer activity and development of hepatocellular carcinoma in patient with cirrhosis of the liver. Japan J Clin Oncol 17:327-332, 1988

15. Okanoue T, Itoh Y, Minami M, Set al: Interferon therapy lowers the rate of progression to hepatocellular carcinoma in chronic hepatitis C but not significantly in an advanced stage: A retrospective study in 1148 patients. J Hepatol 30:653-659, 1999[CrossRef][Medline]

16. Fang Y, Wang L, Jin J, et al: Focal adhesion kinase affect the sensitivity of human hepatocellular carcinoma cell line SMMC-7721 to tumor necrosis factor-alpha/cycloheximide-induced apoptosis by regulating protein kinase B levels. Eur J Biochem 268:4513-4519, 2001[Medline]

17. Tsujimoto T, Kuriyama S, Yamazaki M, et al: Augmented hepatocellular carcinoma progression and depressed Kupffer cell activity in rat cirrhotic livers. Int J Oncol 18:41-47, 2001[Medline]

18. Wang D, Yang E, Cheng LY: Effects of IFN-gamma, TNF-alpha and EGF on expression of HLA class I antigen and the proliferation of human hepatocellular carcinoma HepG2 cells. Anticancer Res 17:181-188, 1997[Medline]

19. Feng X, Tang X, Zheng Z: Preliminary studies on the effects of tumor necrosis factor gene transfer on the growth of human hepatocellular carcinoma cells in nude mice. Chinese Journal of Oncology 17:167-169, 1995[Medline]

20. Osawa Y, Nagaki M, Banno Y, et al: Possible involvement of reactive oxygen species in D-galactosamine-induced sensitization against tumor necrosis factor-alpha induced hepatocyte apoptosis. Journal of Cellular Physiology 187:374-385, 2001[CrossRef][Medline]

21. Atarashi Y, Yasumura S, Nambu S, et al: A novel human tumor necrosis factor alpha mutein, F4614, inhibits in vitro and in vivo growth of murine and human hepatoma: Implications for immunotherapy of hepatocellular carcinoma. Hepatology 28:57-67, 1998[CrossRef][Medline]

22. Cox DR: Regression models and life tables. J R Stat Soc [Ser B] 187-202, 1972

23. Altman DG, Andersen PK: Bootstrapping investigation of the stability of a Cox regression model. Stat Med 8:771-783, 1989[Medline]

24. Peduzzi PN, Concato J, Holford TR, et al: The importance of events per independent variable in multivariable analysis. II: Accuracy and precision of regression estimates. J Clin Epidemiol 48:1503-1510, 1995[CrossRef][Medline]

25. Babyak MA: What you see may not be what you get: A brief, nontechnical introduction to overfitting in regression-type models. Psychosom Med 66:411-421, 2004[Abstract/Free Full Text]

26. Baron RM, Kenny DA: The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. J Pers Soc Psychol 51:1173-1182, 1986[CrossRef][Medline]

27. Evans DL, Charney DS, Lewis L, et al: Mood disorders in the medically ill: Scientific review and recommendations. Biological Psychiatry 58:175-189, 2005[CrossRef][Medline]

Related Correspondence

• More Claims About Depression, Immune Function, and Survival That Exceed the Evidence
  James C. Coyne and Steven C. Palmer
  JCO 2007 25: 5328-5329 [Full Text]

This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Steel, J. L. Articles by Carr, B. I. Search for Related Content PubMed Articles by Steel, J. L. Articles by Carr, B. I. Related Articles Related Correspondence