Background: Depression and obesity, the two common ailments of modern society, are associated with increased risk of coronary artery disease and raised C-reactive protein (CRP) levels. Are the effects of depression and obesity related or do they influence CRP levels independently?
Objective: In 493 consecutive patients presenting for obesity surgery, we explored the relationship between symptoms of depression and raised CRP levels after controlling for confounding factors.
Methods and Procedures: Depression was measured using the Beck Depression Inventory (BDI). Confounding variables were age, gender, BMI, waist and hip measures, smoking and alcohol habits, medications, biochemical measures of the metabolic syndrome, and indirect measures of insulin resistance. General linear regression sought variables independently associated with CRP levels.
Results: These patients had a BMI range from 31 to 91 kg/m2, participants age ranged from 14 to 71 years, and 76% were women. The median CRP concentration was 7.7 mg/l (interquartile range: 3.9–14), 40% had an abnormally raised concentration (>10 mg/l). The mean BDI score was 17.0 ± 9.0, indicating symptoms of moderate depression. We found five independent factors associated with raised CRP levels. In order of strength of association, these were: higher BMI (β = 0.36, P < 0.001), female gender (β = −0.19, P < 0.001), estrogen therapy (β = 0.18, P < 0.001), higher BDI score (β = 0.11, P = 0.01), and insulin resistance index (β = 0.11, P = 0.01), and with a combined R2 = 0.24, (P < 0.001).
Discussion: In obese patients, symptoms of depression were associated with raised CRP levels after controlling for confounding variables. Obese women on estrogen therapy are at risk of high CRP levels.
Recent large-scale epidemiological, clinical and experimental studies have increased our understanding of the mechanisms generating cardiac risk and have provided evidence indicating that depression is an independent risk factor for the development of coronary heart disease (1,2,3,4). Although the mechanism of increased cardiac risk attributable to depression remains unknown, the observation of elevated plasma inflammatory cytokine concentrations in patients (5,6,7) raises the possibility that enhanced inflammatory-mediated atherogenesis may be a contributing factor in generating cardiac risk. Furthermore, a recent study has suggested that a low-grade inflammatory state with raised C-reactive protein (CRP) levels is associated with depression (8). It has been suggested that an association between CRP levels and depression may increase the risk of atherosclerotic disease in these patients (9,10). CRP is an acute-phase inflammatory marker produced mainly in the liver (11) as a result of stimulation by interleukin-6 (12). CRP has been shown to be an independent predictor of death from cardiovascular incidents and myocardial infarction in both men and women, in people of different ages and from different ethnic backgrounds (11).
Obesity is also associated with low-grade inflammation, possibly related to adipose tissue cytokine stimulation of liver CRP production (13). A German study of 3,204 men indicated a significant synergistic effect between depressed mood and obesity, and raised CRP concentrations in a subgroup obese men (n = 726, 23%) (14). The North Finland 1966 birth cohort study reported an association between depressive illness and highly sensitive CRP levels in men, but not in women (10). Another population-based study from Finland, the Cardiovascular Risk in Young Finns study, reported a similar association but found the effect was attenuated when controlled for obesity and triglyceride concentrations, concluding that much of the effect could be attributed to obesity (15). In patients with the metabolic syndrome, high-sensitivity plasma CRP concentrations are reduced after weight loss (16).
CRP levels, metabolic disturbance, and symptoms of depression rise with increasing BMI (17). The interaction between CRP levels, depression and obesity-related factors has been described as producing potentially important confounding effects when examining the relationship between CRP levels and depression. Patients presenting for bariatric surgery form a unique group where these interactions can be examined in high-risk individuals. They present with significant physical and psychological problems associated with their obesity. In particular, they have both high levels of metabolic and inflammatory disturbance and high levels of depressive symptoms (18,19,20).
The aim of this study was to investigate the following in obese patients presenting for bariatric surgery: (i) a possible relationship between CRP levels and depressive symptoms, (ii) determinants of CRP levels, and (iii) after controlling for identified determinants, a relationship between CRP concentrations and symptoms of depression, as measured using the Beck Depression Inventory I (BDI I) (21).
Methods and Procedures
Patients with a BMI >30 kg/m2, suffering significant medical, physical, and/or psychosocial disabilities, and who had failed at significant attempted weight reduction by other means for at least several years were considered for surgery. Preoperative assessment were medical assessment, documentation of comorbidity, medication usage, anthropometric measurements, biochemical tests, and completion of standardized questionnaires including the BDI. Informed written consent was obtained from all patients before surgery. The study was carried out in accordance with the Declaration of Helsinki.
For this study, we selected 493 (95%) of 517 consecutive patients presenting for primary laparoscopic adjustable gastric banding surgery. Twenty-four patients (4%) who had missing questionnaire, biochemical, or anthropometric data were excluded. These excluded patients did not differ to the included group and were excluded only because of missing data. Three patients (1%) with baseline outlier CRP concentrations >50 mg/l were also excluded from the study.
The BDI was used to assess symptoms of depression and therefore the risk of depression within a population. A clinical diagnosis of depression using the Diagnostic and Statistical Manual 4th Edition (DSM-IV) criteria in individual cases was not made and therefore the BDI has been used to measure the risk of depression within groups. The full 21-item version was used and total scores were calculated (22). There was no secondary or subanalysis of specific items or groups of items. Cutoff values for total BDI scores were used for descriptive presentation. We chose to use the following cutoff values: normal, 0–9; mild depressive symptoms, 10–15; moderate depressive symptoms, 16–22; and severe depressive symptoms, 23–63 (refs. 23,24,25). The BDI contains questions to assess all nine characteristic attitudes and symptoms listed in the DSM-IV criteria for a major depressive episode.
Detailed clinical evaluations were used to obtain demographic details, medical and psychiatric history, medication usage, and history of smoking and alcohol consumption. Smoking history was assessed with never, past, or current recorded, as well as pack years. Alcohol consumption was assessed using three categories: nil or no regular alcohol, 10–200 g per week, and >200 g/week. In addition, standard anthropometric measures of height, weight, waist, hip, and neck measurements were taken along with biochemical measures to assess metabolic risk and nutritional status. Fasting biochemical measures were CRP, insulin, glucose, total cholesterol, triglycerides, and high-density lipoprotein-cholesterol levels. An index of insulin resistance was calculated using the loge-transformed fasting plasma insulin and glucose product (26).
CRP scores (median ± interquartile range) were not normally distributed and required loge transformation before parametric and general linear regression analysis. BDI scores were normally distributed with mean ± s.d. reported. Raw correlations were assessed using Spearman correlation coefficients. ANOVA, using the Tukey method of post-hoc analysis, was used to assess differences between the four groups based on their BDI scores. Linear regression analysis was performed using forward and backward modeling to find independent predictors of preoperative loge-transformed CRP concentrations. Linear regression models were built progressively to find relevant and independent factors associated with CRP levels. The model built is described in the results. β-Coefficients and combined R2 have been reported. Binary regression analysis, using conditional stepwise forward and backward approach, was used to characterize those with favorable CRP levels (i.e., <3.0 mg/l). All analysis was performed using SPSS for Windows, Version 14.0 (SPSS, Chicago, IL).
A total of 493 eligible consecutive patients completed the BDI preoperatively and had CRP levels and all other baseline biochemical and anthropometric measures. Characteristics of the group are shown in Table 1. These patients had a BMI range from 31 to 91 kg/m2, were from 14 to 71 years of age, and 76% were women.
Table 1. The 494 consecutive obese patients grouped by their preoperative Beck Depression Inventory Score
The mean preoperative BDI score was 17.0 ± 9.0, indicating that the mean score for these patients is within the moderate depression range. The characteristics of patients grouped by BDI scores indicating normal, mild, moderate, and severe are shown in Table 1. The Spearman correlation coefficients for associations between BDI scores, and CRP levels, sex, age, and BMI are shown in Table 2. Thus, we show a significant crude positive relationship (r = 0.18, P < 0.001) between BDI scores and CRP levels. However, significant confounders of female gender (β = 0.21, P < 0.001) and higher BMI (β = 0.12, P = 0.007) were independently associated with higher BDI scores (R2 = 0.06, P < 0.001).
Table 2. Univariate Spearman correlation coefficients showing the associations between age, sex, BMI, BDI scores, and CRP concentrations in 494 consecutive obese patients
The median CRP concentration was 7.7 mg/l (interquartile range: 3.9–14). Forty percent of all patients had an abnormally raised CRP concentration (>10 mg/l). The mean BDI score for patients with a normal CRP level was 15.1 (s.e.m.: 0.47), and for an abnormally raised patient it was 17.5 (s.e.m.: 0.72) P = 0.005. The Spearman correlation coefficients for associations between CRP levels, and sex, age, BMI, and BDI scores are shown in Table 2.
With the whole group divided into quartiles based on their CRP levels, the mean ± 95% confidence interval for BDI score and BMI are shown in Figures 1 and 2, respectively. Progressively higher mean BDI scores, and mean BMI levels are seen with higher quartiles of CRP.
Associations with CRP concentration
Using linear regression analysis we looked for associations with raised log-transformed CRP levels in this obese population (Table 3). In an initial model, BDI scores were included with BMI, age, and gender. In order of strength of association with higher CRP concentrations were: higher BMI (β = 0.36, P < 0.001), female gender (β = −0.19, P < 0.001), lower age (β = −0.11, P < 0.016), and higher BDI score (β = +0.09, P < 0.026) with a combined R2 = 0.21 (P < 0.001). BDI score was thus independently associated with higher CRP concentration when controlled for age, gender, and BMI.
Table 3. Factors influencing the variance of CRP concentration modeled in women and men separately
Other specific confounders known to influence CRP levels or factors that may confound an association between CRP levels and depression were also examined in additional models. These factors were taking antidepressant medications (selective serotonin reuptake inhibitors medications were taken by 78 (16%) of patients), smoking, alcohol consumption, lipid-lowering medications (hydroxymethylglutaryl CoA reductase inhibitors, taken by 59 (10%) of patients) and estrogen therapy. In this obese population, two of these factors were associated with variance of the CRP level when modeled with age, gender, and BMI. Regular light to moderate alcohol consumption (>10 gm and <200 gm/week) was associated with lower CRP concentrations and estrogen therapy with the oral contraceptive pill or hormone replacement therapy was associated with higher CRP concentrations. In this population currently taking antidepressant or lipid-lowering medications, current smoking or the extent of smoking with pack years was not associated with significant variance in the CRP concentration.
BDI scores were then modeled with age, gender, BMI, estrogen therapy, and light to moderate alcohol consumption for association with CRP concentration. Four factors were associated with CRP levels. In order of strength of association these were: higher BMI (β =0.37, P < 0.001), female gender (β = −0.17, P < 0.001), estrogen therapy (β = 0.17, P < 0.001), and higher BDI score (β = +0.11, P < 0.01) with a combined R2 = 0.23 (P < 0.001). Thus, BDI score remained a significant association. In this model, age was no longer a significant contributor, but it was included in further models. There were six women with major depressive symptoms taking estrogen therapy. They had a mean serum CRP concentration of 28 mg/l and individual levels were 9, 15, 27, 27, 41, and 49 mg/l.
Transformed CRP scores were examined for associations with weight distribution (waist-to-hip ratio) and the metabolic syndrome (a diagnosis of hypertension, diabetes, systolic and diastolic blood pressure, fasting plasma insulin, glucose, triglycerides, and high-density lipoprotein-cholesterol). The only factors that were associated with a higher CRP were low waist-to-hip ratio, higher fasting plasma insulin levels, and higher insulin resistance index (ln-transformed insulin glucose product (26)). When modeled with age, gender, BMI, estrogen therapy, and BDI scores, there were now five independent variables explaining some variance of the transformed CRP level. In order of strength of association these were: higher BMI (β = +0.36, P < 0.001), female gender (β = −0.19, P < 0.001), estrogen therapy (β = +0.18, P < 0.001), higher BDI score (β = +0.11, P = 0.01), and insulin resistance index (β = +0.11, P = 0.01), and with a combined R2 = 0.24, (P < 0.001). Thus after controlling for a wide range of confounders, BDI scores were significantly associated with raised CRP levels.
The analysis was repeated with data from women and men examined separately (Table 3). For women (n = 377), the BMI, estrogen therapy, BDI scores, and insulin resistance index contributed to variance of CPR levels. Men (n = 117), who made up only 24% of the total group were less likely to have symptoms of depression, and BMI was the only factor associated with CRP concentration.
Only 19.8% of patients had CRP levels in the low to moderate coronary risk range of 0–3 mg/l. Binary regression analysis showed that these patients were more likely to be men (b = 5.6, 95% confidence interval 3.2–9.8), have lower BMI (b = 0.89, 95% confidence interval 0.85–0.93), and less insulin resistant (b = 0.61, 95% confidence interval 0.42–0.89). We did not measure highly sensitive CRP, but this did not appear to hamper the analysis unduly in these obese patients.
In agreement with our recent demonstration of elevated CRP plasma concentrations in patients with major depressive disorder (8), we have found a durable association between BDI scores and CRP concentrations in obese patients after adjusting for a range of potentially confounding variables. Obesity and depression are two of the most common conditions impairing health and quality of life in our communities. We confirm that these two conditions have independent low-grade inflammatory effects which may play an important role in predisposing these patients to atherosclerosis (14). We demonstrate that degree of obesity and the severity of depressive symptoms are independently and incrementally associated with higher CRP levels (Figures 1 and 2).
Obese women, particularly those on estrogen therapy, appear to be at the greatest risk of this inflammatory effect. This study of predominantly obese women shows the association between CRP and symptoms of depression. This contrasts with the National Health and Nutrition Examination Study III (NHANES III) data which suggested the relationship between depression and raised CRP levels were stronger in men (27), and the data from Finland where an association was only found in men (10). Our analysis in men may have been limited by the smaller number of male patients within the study, the limited BMI range, and men reporting fewer symptoms of depression. The World Health Organization (WHO) recommends caution regarding the use of estrogens containing oral contraceptive medications in obese women principally related to an increased risk of venous thromboembolism. Raised CRP levels may contribute to this thrombotic risk and to the risk of major cardiovascular events in obese women taking oral contraceptives. The results of this study clearly indicate that obese women, who often have a range of metabolic and cardiovascular risk factors, should be cautioned and a risk benefit assessment made before commencing oral contraceptives or hormone replacement therapy.
Cross-sectional studies such as this demonstrate relationships but attributing causality or counter causality is problematic. BDI measures symptoms of depression and many of these symptoms are common in severely obese patients especially those associated with poor body image, low self value, self loathing, and somatic symptoms such as lethargy, tiredness, daytime sleepiness, and nocturnal sleep disturbance. Of course an inflammatory state may also be symptomatic with symptoms of lethargy, tiredness, and impaired physical function. It would be interesting to look at a similar analysis in the weight loss state, however, causality can be difficult to interpret as substantial weight loss lowers CRP levels (16,28,29), BDI scores (18), and improved quality of life and metabolic factors associated with obesity (30). Interestingly, recent reports provide evidence of a link between vagal activity and inflammation. Dubbed the “cholinergic anti-inflammatory pathway,” efferent vagal activity results in acetylcholine release which, in turn, suppresses the inflammatory response (31). This is consistent with our previously published data in patients with major depressive disorder, reduced vagal activity, as indicated by parallel reductions in heart rate variability and the slope of the cardiac baroreflex, is accompanied by increased release of proinflammatory cytokines, including CRP (8). Weight loss in obese patients with the metabolic syndrome is associated with improved vagal function and, as mentioned previously, a concomitant reduction in plasma CRP concentrations (16).
Although the results of this study indicate a strong relationship between CRP and depression, there are several limitations of this study which need to be considered. All patients were presenting for bariatric surgery, they were usually severely obese and often had high levels of physical and mental impairment of quality of life, and high levels of medical comorbidity and metabolic disturbance. Although this significantly reduces the ability to generalize results to the broader community, it has enabled us to look at a high-risk group of patients where confounders in the relationship between CRP levels and depression are plentiful. There were relatively few men in this study reducing the ability to look at associations with CRP levels in obese men, but the association between depression and CRP level is more established in men (10,14).
We did not measure highly sensitive CRP levels enabling us to look at the nuances of “low” usual community values of <3 mg/l. However, only 20% of patients had values of ≤3 mg/l, and in this group of patients presenting for bariatric surgery we were not substantially restricted by this weakness. Standard CRP values provided sufficient variance in which to examine for clinically relevant associations without the need for high-level discrimination at lower levels.
In summary, in these obese patients we have confirmed a durable association between raised CRP levels and symptoms of depression after controlling for a wide range of confounding variables associated with obesity. Previous research has suggested that these confounders may account for the association between CRP levels and obesity. Raised CRP levels may play a part in the observed increased risk of cardiovascular disease in those with symptoms of depression.