Fax: (011) 49 30 450513911
Biomarkers of depression in cancer patients
Article first published online: 11 OCT 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 11, pages 2723–2729, 1 December 2006
How to Cite
Jehn, C. F., Kuehnhardt, D., Bartholomae, A., Pfeiffer, S., Krebs, M., Regierer, A. C., Schmid, P., Possinger, K. and Flath, B. C. (2006), Biomarkers of depression in cancer patients. Cancer, 107: 2723–2729. doi: 10.1002/cncr.22294
- Issue published online: 17 NOV 2006
- Article first published online: 11 OCT 2006
- Manuscript Accepted: 7 SEP 2006
- Manuscript Revised: 14 AUG 2006
- Manuscript Received: 24 APR 2006
- hypothalamic-pituitary-adrenal (HPA) axis
Inflammation and perturbation of the hypothalamic-pituitary-adrenal (HPA) axis function appears to play a putative role in the etiology of depression. Patients with metastatic cancer demonstrate elevated prevalence rates for depression. The objective of the current study was to illustrate the efficacy of interleukin-6 (IL-6) and HPA axis function as adjuncts to support the diagnosis of depression in cancer patients.
Plasma concentrations of IL-6 and cortisol were measured in 114 cancer patients with and without depression. The relative diurnal variation of cortisol (cortisol VAR), expressed as a percentage, was calculated. Receiver operating characteristics analysis was performed.
Depression was associated with increased plasma concentrations of IL-6 (18.7 pg/mL vs. 2.7 pg/mL; P < .001) and higher cortisol concentrations at 8 AM and 8 PM. The relative cortisol VAR (11.7% vs. 60.6%, respectively; P < .001) was found to be decreased in cancer patients with depression, indicating a disturbed circadian function of the HPA axis. As a biomarker of depression, IL-6 yielded at a cutoff value of 10.6 pg/mL, a sensitivity of 79%, and a specificity of 87% (area under the curve [AUC] = 0.86; 95% confidence interval [95% CI], 0.78–0.94), whereas cortisol VAR demonstrated a sensitivity of 81% and a specificity of 88% (AUC = 0.85; 95% CI, 0.74–0.97) at a cutoff value of 33.5%.
Depression is associated with increased plasma IL-6 concentrations in patients with cancer. These patients demonstrate a dysfunction of the HPA-axis, characterized by a decreased diurnal variation of cortisol. The high sensitivity and specificity of these parameters biomarkers of depression make IL-6 and cortisol VAR helpful tools in the diagnosis of depression in patients with cancer. Cancer 2006. © 2006 American Cancer Society.
Most people value the quality of life as much as the quantity. This is an important issue in treating and caring for patients with advanced cancer. A significant portion of patients develop a depressive disorder, ranging from a mild depressive mood to major depression, as defined by the 4th edition of the Diagnostic and Statistical Manual of Mental Disorders(DSM-IV).1 It is important in this context to distinguish between understandable sadness and clinical depression, which requires sufficient recognition and treatment. Depression reduces compliance with treatment and clearly lowers quality of life in cancer patients. Therefore, depression may have a negative impact on outcome, considering that psychologic intervention to relieve distress in breast cancer patients has demonstrated a significant increase in overall survival time.2
Research on depression in the general population has identified several physiologic alterations that are strongly associated with depression, possibly even a causative role in the pathogenesis of depression. In particular, 2 findings have sparked interest in developing a new view of the etiology of depression. First, extensive research has shown marked alterations of the hypothalamus-pituitary-adrenal (HPA) axis function, resulting in an increased secretion and flattened circadian rhythm of cortisol in these patients.3, 4 The combined dexamethasone suppression (dex)/corticotropin-releasing hormone (CRH) stimulation test5, 6 suggests an impaired corticosteroid receptor function (leading to a malfunction of the glucocorticoid-mediated feedback inhibition) to be responsible for the observed hyperactivation of the HPA axis in depression. Paradoxically, the adrenocorticotropic hormone (ACTH) and cortisol response to the combined dex/CRH test is elevated during depression, but tends to normalize after successful treatment and has been proposed as an additional response marker for antidepressant treatment.7–9
The concept of sickness behavior, a syndrome that shares many characteristics with depression,10 has lead to the speculation that proinflammatory cytokines might also play a role in depression. Several studies have shown a significant association of cytokines, especially interleukin-6 (IL-6), with depression in patients with and without cancer.11, 12 The cytokine IL-6 is produced by a variety of immune cells, such as monocytes, lymphocytes, and glial cells, but also by nonimmune cells such as endothelial cells, endocrinal cells, and tumor cells.13, 14 IL-6 is also a hormonally regulated cytokine. Its production is stimulated by catecholamines and suppressed by glucocorticoids.15, 16 In healthy individuals, the blood levels of IL-6 increase with age.17 The level of IL-6 has been related to tumor progression, tumor size, recurrence, and shorter survival.18–20
An unfortunate symbiosis between the failing of medical staff to recognize depression in cancer patients and a reluctance of the patients to reveal emotional distress has led to the problem that too few cancer patients receive adequate antidepressant therapy. This is due in part to the fact that the diagnostic criteria for major depression include somatic symptoms that are difficult to distinguish clearly from the effects of chemotherapy or symptoms inherent to cancer (ie, fatigue, weight loss, and loss of libido). Therefore, several attempts to modify the diagnostic criteria have been employed. Omitting such somatic symptoms as anorexia and fatigue has been suggested, or placing more emphasis on feelings such as hopelessness.21, 22 In conjunction with various cutoff scores of different diagnostic and screening tools, depression has been identified in 2% to 57% of cancer patients, depending on which assessment procedure was used.23–25 Such differences in prevalence rates has led to a certain degree of uncertainty in the clinical setting as to who should receive an antidepressant therapy and who should not. In view of this, the use of biologic markers such as IL-6 and the HPA axis function might be helpful adjuncts to support the diagnosis of depression in cancer patients.
MATERIALS AND METHODS
In this cross-sectional study, 114 inpatients were assessed for depression in the Department of Oncology at the University Hospital Charite Campus Mitte (Berlin, Germany) from March 2003 to December 2004. The study was approved by our Institutional Review Board. All patients had advanced metastatic cancer (Stage IV) and provided written informed consent. Patient characteristics such as age, gender, cancer type, disease status (progressive disease, stable disease, or disease remission) and Karnofsky performance status26 were recorded.
All subjects presented with a Mini-Mental State Examination (MMSE) score of at least 24. Exclusion criteria were untreated endocrine diseases, renal insufficiency requiring dialysis, or any form of neurologic malfunction (ie, meningeosis carcinomatosa or brain metastasis). None of the patients were receiving anti depressant drugs, anticonvulsants, or other psychotropic medication for treatment of affective disorders. Other exclusion criteria were current evidence of substance/alcohol abuse, long-term steroid use (>8 weeks), and any use of steroids within the last 8 weeks, which could interfere with HPA-axis function.
The diagnosis of major depression was established using the depression section of the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID).27 To measure the level of depression, all participants completed the Hospital Anxiety and Depression Scale (HADS-D), a well-validated,28 self-rated 14-item scale that was developed especially for assessing depressive symptoms in medically ill patients.29 A threshold score of >11 on the depression scale (range, 0–21) is considered indicative of a probable diagnosis of clinical depression in patients with somatic disease.30
Patients were routinely woken at 7:00 AM by the nursing staff. Blood samples for the measurement of IL-6 and cortisol were collected at 8:00 AM At 8:00 PM, blood collection for the evening cortisol concentration was performed. Plasma IL-6 concentrations were measured with commercially available kits using the manufacturer's protocol (Immulite/Immulite 1000 IL-6; DPC Biermann, Germany). Samples were assayed in duplicate and IL-6 concentrations were derived from a standard curve comprised of serial dilutions (range, 2–625 pg/mL) of recombinant human IL-6. Assay sensitivity was <2 pg/mL. The mean interassay and intraassay coefficients of variation were 10.9% and 3.6%, respectively. Plasma concentrations of cortisol were measured by using a radioimmunoassay (GammaCoat; Incstar, Stillwater, MN). The mean interassay and intraassay coefficients of variation were 8.8% and 6.6%, respectively. All samples were assayed by personnel who were blind to the diagnostic identity of the study subjects.
All statistical analysis was performed using SPSS version 12.0 software (SPSS Inc., Chicago, IL). Results with normal distributions are presented as the mean ± the standard deviation (SD). The Student t-test was used to compare these variables. Due to the skewed distribution of IL-6, cortisol plasma concentrations, and cortisol variations (VAR), the medians and ranges were calculated. The relative diurnal variation of cortisol (cortisol VAR), expressed as a percentage, was calculated as the change in cortisol (Δ cortisol = cortisol 8:00 AM minus cortisol 8:00 PM) divided by cortisol 8:00 AM and multiplied by 100. The Mann-Whitney U-test was used to compare these variables between the groups. All results were considered significant at P < .05 (2-tailed). In addition, a multiple logistic regression analysis was performed to evaluate whether tumor status and IL-6 level were independently associated with depression.
To test the specificity and sensitivity of the biological markers IL-6 and cortisol VAR to detect depression, receiver operating characteristics (ROC) analysis was performed. The correlation between the true-positive rate (sensitivity) and the false-positive rate (1-specificty) was represented as a curve. The cutoff point was chosen to minimize the sum of false-positive and false-negative test results. Efficient screening instruments are indicated by ROC curves with a high area under the curve (AUC). In addition, the positive predictive value (PPV) was calculated.
Thirty-one patients diagnosed with clinical depression and 83 cancer patients with no past or present history of depression were recruited for comparison. The 114 patients evaluated in the analysis had a mean age of 60.7 years (SD of 9.7 years) and presented at the time of assessment with a mean Karnofsky index of 63.9% (SD of 13.3%). All patients were being treated with chemotherapy and 62% of the patients (71 patients) had progressive disease, whereas 38% (43 patients) were classified as exhibiting stable disease at the time of study inclusion. None of the patients were in disease remission. The chi-square test revealed that the presence of depression was dependent on tumor status (ie, progressive disease [chi-square: 16.1; degrees of freedom (df): 1; P < .001]). In addition, the multiple logistic regression analysis with depression as the dependent variable and tumor status and IL-6 as the independent variable showed that tumor status (odds ratio [OR] = 6.0; 95% confidence interval [95% CI], 1.2–29.7 [P = .029]) and IL-6 concentration (OR = 1.1; 95% CI, 1.0–1.2 [P = .001]) were independently associated with depression.
Table 1 shows a comparison of the characteristics of patients with and without depression. All patients were white. The mean age and Karnofsky index were not found to be significantly different between cancer patients with depression and cancer patients without depression (age of 62.7 years [SD of 10 years] vs. 59.4 years [SD of 9.2 years] [P = .073] and Karnofsky index of 62.1% [SD of 14.5%] vs. 65.2% [SD of 12.3%] [P = .385], respectively). There was a significant difference noted in median plasma concentration of IL-6 between the cancer patients with and without depression (Table 2). Patients with depression showed an elevated median IL-6 concentration of 18.7 pg/mL (range, 0–199 pg/mL), whereas the patients without depression exhibited a median concentration of only 2.7 pg/mL (range, 0–43 pg/mL) (Fig. 1). The significance between these 2 groups was P < .001. To assess the usefulness of this biomarker as an adjunct in the diagnosis of depression in cancer patients, an ROC analysis was performed, shown in Figure 2. The optimal cutoff point for using IL-6 as a biomarker for depression was 10.6 pg/mL (Table 3). This cutoff point was associated with a sensitivity of 79% and a specificity of 87% (AUC = 0.86; 95% CI, 0.78–0.94). The PPV for IL-6 was 74%.
|Characteristic||Depression present (n = 31)||Depression absent (n = 83)||P student t test|
|Head and neck||4||9|
|Age, y||62.7 ± 10||59.4 ± 9.2||P = .073 (NS)|
|Karnofsky index (%)||62.1 ± 14.5||65.2 ± 12.3||P = .385 (NS)|
|HADS-D score (0–21)||14.7 ± 2.9||4.27 ± 1.9||P < .001|
|Biomarkers||Depressionpresent (n = 31)||Depressionabsent (n = 83)||P Mann-Whitney U-test|
|IL-6 (pg/mL)||18.7 (0–199)||2.7 (0–43)||P < .001|
|8:00 AM||506.5 (16–894)||387.5 (29–663)||P = .003|
|8:00 PM||445.5 (105–922)||130.5 (12–618)||P < .001|
|8:00 AM to 8:00 PM||75.5 (−131–512)||211.5 (−145–308)||P < .001|
|Cortisol VAR (%)||11.7% (9–71.6)||60.6% (27.6–96.3)||P = .037|
|Cutoff||Sensitivity (%)||Specificity (%)|
Plasma cortisol concentrations were also significantly higher in depression at 8 AM and 8 PM (Table 2). To assess the presence of a disturbed circadian function of the HPA axis, the diurnal amplitude in cortisol concentration was evaluated by calculating the cortisol VAR. Cortisol VAR was found to be significantly decreased in the cancer patients with depression compared with the patients without depression (11.7% vs. 60.6%, respectively; P = .001), indicating a reduced diurnal amplitude in the cortisol concentration (Fig. 3). The ROC curve is shown in Figure 4. The optimal cutoff point for using cortisol VAR as a biomarker for depression was 33.5% (Table 4). This cutoff point was associated with a sensitivity of 81% and a specificity of 88% (AUC = 0.85; 95% CI, 0.74–0.97). The PPV for cortisol VAR yielded a value of 76%.
|Cutoff Point||Sensitivity (%)||Specificity (%)|
The fear of cancer alone frequently fosters the notion that depression is an understandable and manifest reaction to this life-threatening disease. The incidence of depression in cancer patients, however, is comparable to that observed in patients with other medical conditions associated with inflammation (ie, rheumatoid arthritis and coronary artery disease).31, 32 In this cross-sectional study, cancer patients with depression exhibited elevated IL-6 plasma concentrations. This is consistent with other studies of depressed patients with and without cancer.12, 33 However, there also have been inconclusive studies regarding this association of elevated IL-6 concentration and depression.34, 35 An explanation for this could be the small sample sizes and inhomogeneity of the study populations with regard to comorbidity (ie, cancer type and stage, age, and performance status). Age and performance status are important confounders of HPA activity.17 Our study consisted of a homogenous study population regarding these confounding factors. The groups compared were not found to differ with regard to age, cancer stage, or Karnofsky performance index.
Depression and disease status were not found to be independent variables. We found that the presence of depression is dependent on a progressive disease status. However, disease status and IL-6 concentration appear to be independently associated with depression. One explanation for this could be the increased IL-6 production induced by tumor growth and inflammation.13
IL-6 is a potent activator of the HPA-axis, characterized by an increased production of cortisol and depressive mood.36, 37 To our knowledge, this mechanism is still not fully understood, but proinflammatory cytokines demonstrate the ability to inhibit the glucocorticoid-receptor (GR) translocation and function in vitro, thereby inducing a state of GR-resistance, leading to malfunction of the glucocorticoid-mediated regulatory feedback loop of the HPA axis. The result is a chronic pathologic activation of the HPA axis.38 In our study, depression was found to be significantly associated with a disturbance of the circadian function of the HPA axis, as noted by a decrease in the relative diurnal variance of cortisol. The dysregulation of the HPA function has been reported extensively in depressive patients without cancer.3, 6 To our knowledge, however, this has not been shown conclusively in cancer patients with depression. Maes et al.39 demonstrated a significant correlation between IL-6 and HPA function (postdexamethasone cortisol values) in depressed patients without cancer. However, to our knowledge, such a correlation has not been established to date in a representative sample of cancer patients.
There are a few limitations to the results of the current study, with one being the omitting of pain evaluations. Although none of the patients was in great pain, a high percentage of cancer patients were receiving chronic pain medication. To our knowledge to date, the role of pain in the pathophysiology of depression remains unclear. Some authors suggest that depression may represent a consequence of pain, whereas others find depression contributes significantly to the intensity of pain.22, 40
The lack of recognition of depression in cancer patients is well documented.41 Diagnosing depression in cancer patients is often not clearcut42 because the symptoms inherent to cancer disease often overlap with symptoms used as diagnostic criteria for depression, as established by the DSM-IV. Therefore, some authors have suggested using an exclusive approach by removing such criteria as anorexia and fatigue from the list of diagnostic symptoms.43 However, we believe that excluding such symptoms increases the possibility of misdiagnosis because the origin of overlapping symptoms can rarely be identified with certainty, and therefore we suggest that for clinical purposes all DSM-IV criteria should be taken into account. The HADS-D has gotten mixed reviews for its ability to identify depression as a diagnostic tool among cancer patients. The sensitivity and specificity of the HADS-D are reported to range from 65% to 91% and 66% to 96%, respectively, in cancer patients, depending on the cutoff values used and the homogeneity of the sample.44 However, the HADS-D has been shown to perform well in assessing symptom severity in depressed somatic patients.45
The pharmacologic treatment of depression in patients with cancer has proven to be effective, and not only improves the quality of life but may also prolong survival and compliance with chemotherapy.46–48 In view of these difficulties, we evaluated the clinical usefulness of IL-6 and the HPA axis function as biologic markers in the diagnosis of depression in cancer patients.
At a cutoff point of 10.6 pg/mL, IL-6 provided sufficient sensitivity (79%) and specificity (87%) as a screening tool and as a helpful adjunct in the diagnosis of depression in cancer patients. As an indicator of a flattened circadian rhythm of cortisol, the relative diurnal variation (cortisol VAR) of cortisol demonstrated an optimal cutoff point at 33.5%. The dysfunction of the HPA axis exhibited an even better sensitivity (81%) and specificity (88%) as a biomarker for depression, emphasizing its putative core role in the pathophysiology of depression. These markers could be helpful tools in the diagnosis of depression in cancer patients in the often uncertain clinical setting with regard to which patients should be treated for depression.
- 1American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association 1994.
- 2Effect of psychosocial treatment on survival of patients with metastatic breast cancer. Lancet. 1989; II: 888–891., , , et al.
- 10Behavioral effects of cytokines: an insight into mechanisms of sickness behavior. Methods Neurosci. 1993; 17: 130–144., , , .
- 22Depression. In: HollandJC, RowlandJH, editors. Handbook of Psychooncology: Psychological Care of the Patient with Cancer. New York: Oxford University Press; 1989: 283–290..
- 26The clinical evaluation of chemotherapeutic agents in cancer. In: Evaluation of Chemotherapeutic Agents. New York: Columbia University Press; 1949: 191., .
- 27Structured clinical interview for DSM-IV axis I disorders. Research version, patient ed. New York: Biometrics Research, New York State Psychiatric Institute; 1999., , , .
- 43Diagnostic assessment of depression in the medically ill. In: StoudemireA, FogelB, editors. Psychiatric Care of the Medical Patient. New York: Oxford University Press; 1993: 53–70., , .