Between May 2006 and September 2007, a total of 146 potentially eligible patients were admitted to our department 2–3 days after an acute coronary syndrome (ACS). The present study enrolled patients 30–75 year old with an ACS, defined as previously reported [19]. Specific exclusion criteria with respect to depression, anti-inflammatory drugs and inflammatory diseases, were as follows:

Other exclusion criteria included unstable medical or neurologic condition and patients who were unable to communicate reliably (e.g., because of cognitive dysfunction or not speaking French).

The Ethics Committee review board of the hospital (CPP SOOM III, Bordeaux) approved the study protocol, and all participants provided informed consent prior to participation. All standard of care cardiovascular treatments were permitted during the trial. Concomitant medication, including steroids, COX-2 selective inhibitors and other non-steroidal anti-inflammatory treatments were recorded during the study.

ACS was defined and treated according to established guidelines [20]. A coronary angiogram was performed on each patient included in the study, allowing precise evaluation of coronary lesions, and optimized acute phase management. The most suitable treatment was delivered during the first week, in compliance with current guidelines [20].

Patients meeting the inclusion criteria were invited to participate as soon as they were medically stable and had been informed of their diagnosis, on average 2 days after their ACS (range, 2–4 days). After explaining the study and obtaining written informed consent, a research psychologist (S.T.) conducted all baseline psychiatric interviews, and gathered routine demographic data, including age, gender, education, living arrangements, current partner status, socioeconomic status and employment status (employed or unemployed). Information on history of familial and personal depression, or other psychiatric illness was collected during hospitalization.

During the assessment period, if the MINI was positive for a diagnosis of MDD, the patient was excluded from the study. If it was negative, the other psychiatric interviews were performed, as well as an evaluation of hsCRP and fibrinogen levels along with other classical risk factors.

All patients benefited from the CEPTA program with an optimization of secondary prevention measures and educational classes as previously described [19,23].

Patients were followed at 1, 3 and 9 months after ACS. At each time point, cardiovascular risk factors, treatment, levels of hsCRP and fibrinogen were recorded, and psychiatric interviews were performed. The cardiologists and the care team were blinded to the depression status over the 9 month follow-up period. Patients with depressive symptoms were not treated with antidepressant medication, due to the lack of evidence of a clear benefit of depression screening, and of antidepressant medication in this patient population, on cardiovascular disease (CVD) outcome. Instead these patients were sent to a mental health professional for psychotherapy alone.

To assess the incidence of MDD, the psychologist administered the MINI at 3 days and 1, 3 and 9 months, after an ACS. The psychologist also conducted the HDRS-17 and MADRS interviews at 3 days, 1, 3 and 9 months, to score the intensity and evolution of depression symptoms.

To assess the inflammatory status, hsCRP and fibrinogen levels were measured at 3 and 7 days, 1, 3 and 9 months. hsCRP was measured by immuno-turbidimetry, with a detection limit of 0.02 mg/L (Rock Diagnoses). The assays were performed blinded to patients’ depression status; plasma samples drawn at baseline were frozen at −70°C for subsequent measurement of inflammatory [19] markers.

The 9-month survival status was determined for all patients after their ACS. Cause of death was established from hospital and general practitioner records and death certificates. Deaths were classified as cardiac or non-cardiac by a consulting cardiologist blinded to baseline data.

Continuous variables were described by mean and standard deviation, and compared using Student’s t tests. Categorical variables were described by numbers and proportions, and compared using Chi-square tests. Comparisons were considered statistically significant when the p value was <0.05, and corrected according to the Bonferroni rule in case of multiple testing. The incidence rate of MDD was estimated by the proportion of patients presenting with at least one positive MINI at any time point divided by the total number of patients followed during the 9-month period after an ACS. Ninety-five percent confidence intervals (95% CI) of proportions and incidence rates were calculated by the exact binomial method. The association between patient characteristics and the diagnosis of MDD was assessed by logistic regression models, using standard recommended methods. The strength of the association between predictors and MDM was estimated by calculating the odds ratios (OR) and their 95% CIs. Linearity of the logit was systematically checked for every continuous variable. A limited number of possible predictive variables were chosen on the basis of previous reports (age, female gender, familial and personal history of depression, low socio-economic level, living alone, atherosclerosis burden, LDL cholesterol, metabolic syndrome, smoking and γGT). Their association with the occurrence of depression was first tested in univariate logistic regression analysis. Then, all variables associated with depression with a p value < 0.20 were introduced into the multivariable model. The variable of interest, hsCRP level at discharge, was forced into the model. The full model was kept without further selection, as the aim was to obtain an estimate of the association of hsCRP with depression, adjusted based on possible confounding factors.

In cases where there was missing data for either hsCRP or MINI, sensitivity analyses were carried out. Exploratory analysis of the association between the evolution of depressive symptoms, as assessed by MADRS, and the evolution of hsCRP was carried out using a mixed linear model.

All analyses were carried out using SAS® software v9.1.3 (SAS Institute Inc., Cary, North Carolina, USA).

The study population flow chart is shown in Figure 1. At the time of ACS, none of the 87 patients included in the study had a MDD, or a history of MDD less than 6 months before their initial hospitalization. New onset of depression was recorded in 8.0% of the patients at 1 month, 10.3% at 3 months and 10.3% at 9 months. A patient was categorized as part of the “depression group” if he or she had a MDD at any time point during follow up, as assessed by the MINI. The overall 9-month incidence of a MDD was 28.7% [95% CI: 19.5 – 39.4]. At 9 months, 19.5% of patients (prevalence) still met the criteria for diagnosis of depression.

There were no significant differences in general characteristics and the risk factor profile, between patients who remained free of MDD (non-depression group) versus those who developed MDD (depression group) at any of the follow up time points, with the exception of prior history of depression (Table 1 and Additional file 1: Table S1). There were significantly more patients in the depression group who reported having a prior history of depression, than in the non-depression group (88%, vs 37.1% p < 0.0001).

There was no association found between depressive symptom status and atherosclerosis burden (Table 1), or major cardiovascular events (such as reinfarction, stroke or death) during the 9 months follow-up period (3 ACS and 1 stroke occurred in the non-depression group vs. no event in the depression group).

Acute phase management and optimal medical treatment were similar in both groups of patients (MDD vs. no depression) at discharge (Table 1) and during the follow-up period (not shown).

According to HDRS-17 and MADRS scales, the majority of these MDD patients were classified as having mild to moderate depression (Table 2). No difference in hsCRP levels was detected between the depression and non-depression groups at any time-point (Table 2). The evolution of mean hsCRP levels was similar in the two groups and even individual values did not show any trend (Figure 2 and Additional file 1: Figures S1 and S2). In univariate, unadjusted analysis, hsCRP measured at day 7 was not associated with the occurrence of a MDD during the 9-month follow-up period after the ACS. The odds ratio for a MDD during the first 3 months was 0.96 (95% CI: 0.88 - 1.05; p = 0.37) and the odds ratio for a MDD during the first 9 months was 1.02 (95% CI: 0.97 - 1.07; p = 0.53) (Table 3). We also estimated the association between hsCRP measured at 1 and 3 months and the occurrence of a MDD during the first 3 or 9 months, but all odds ratios were close to 1 and not statistically significant (Table 3). In multivariable analysis, i.e. when taking into account other characteristics, hsCRP was not associated with the occurrence of MDD during the first nine months period after ACS. The adjusted OR was 1.07 (95% CI: 0.77 -1.48; p = 0.70) (Table 4).

Statistics were done on 87 of 98 patients originally included. Eleven patients were not analysed because at least one MINI was missing at a time point. Sensitivity analyses were carried out by replacing missing values on MINI and hsCRP according to different strategies. All 11 missing depression statuses were replaced by absence, then by presence of depression, without generating a significant association between hsCRP D7 and 9-month depression occurrence. When the 9 missing values of hsCRP D7 were replaced by the highest hsCRP value (46.3) for the patients of the depressed group and the lowest hsCRP value (0.3) for the patients of the non-depressed group, no significant association was found.

Moreover, we found no association between hsCRP at day 7 and either HDRS-17 or MADRS scores at months 1, 3 and 9 (not shown). We also used a mixed linear model for studying the association between the evolution of depressive symptoms, as assessed by MADRS, and the evolution of hsCRP, between inclusion and one-month follow-up. No association was found (p = 0.69). The evolution of hsCRP values was quite similar after the one-month follow-up (Figure 2).

Patients diagnosed as depressed at any time point, had higher fibrinogen levels at baseline and at their 9 months assessment, than their non-depressed counterparts, but this difference was not statistically significant at 1 and 3 month assessment (Table 2). Moreover, there was no significant association between fibrinogen values and the occurrence of depression during different periods of time, as analyzed by univariate logistic regression (Table 3).

We evaluated the association between quantitative and non-quantitative characteristics and CV risk factors of the population, as well as parameters of atherosclerosis burden, with MDD occurrence. A limited number of possible predictive variables were chosen on the basis of previous reports in the literature. In univariate (Additional file 1: Table S2) as well as in the above-mentioned multivariable analyses (Table 4), only the personal history of depression (more than 6 month before ACS) was significantly associated with the onset of a depressive episode during the 9-month follow-up period, with an OR of 11.02 (95% confidence interval [CI], 2.74 to 44.34; p = 0.0009) adjusted for gender, hsCRP, Gamma-GT at day 7 and HDRS-17 score at inclusion (Table 4).

A comprehensive literature review of hospitalized post-MI patients, revealed a prevalence rate for MDD of approximately 20%, when measured using clinical interviews [24,25]. Our results show that, in the 9 months following an ACS, 28.7% of patients developed MDD. We chose to exclude patients with a history of depression within the last six months and those who were depressed at the time of initial hospitalization for their ACS [26]. Therefore, this study highlights the high incidence of new cases of MDD after an ACS and the long delay in occurrence after the onset of ACS.

Depression can be a risk factor for the development of CAD, and often exacerbates the outcome, when present in patients with established CAD [7-9]. Atherosclerosis is fundamentally an inflammatory disease and inflammatory markers are powerful predictors of CAD events [10,11]. Depression and inflammation have each been independently associated with CAD [6,12,13,15,17,18,27-33]. The interplay between depression, inflammation, and CVD or mortality is complex, and may be analyzed in several different ways based on literature [6-13,15,17,18,27-32] : (i) depression may induce inflammation, and the latter mediate independently the relationship between depression and CVD; (ii) both depression and inflammation may cause CVD but through separate pathways, being synergistic or not; (iii) depression and inflammation may have a common precursor that is linked to CVD; or (iv) inflammation may cause depression and each may or may not have a causal role on CVD [6].

The primary focus of this study was to determine whether or not inflammation may cause depression, because of the importance of inflammation in atherosclerosis. In this study, we found no significant association between atherosclerosis burden, hsCRP or fibrinogen levels at any time point and depressive symptoms after an ACS. The point estimation of the OR was close to one (1.07), favoring the conclusion of an absence of association. However, due to the limited number of patients, the precision of the estimate was low (95% confidence interval: 0.77 - 1.48). Various sensitivity analyses were conducted by imputing extreme values for missing data but none resulted in an association between hsCRP and MDD. Exploratory analyses did not show an association between the evolution of hsCRP and the evolution of depression scales. These arguments reinforce the confidence in this null result. Therefore, this study supports the hypothesis that patients after an ACS may have a MDD independent of an elevated hsCRP or fibrinogen level.

These data are in contrast to epidemiological and clinical studies which have suggested that inflammation precedes the occurrence of depression as in patients with cancer treated with immunotherapy [17,18,27,29,33,34]. Previous studies on atherosclerosis, inflammation and depression [6] have been limited to selected clinical samples and/or have failed to control for important confounding factors [12]. It is not clear to what extent these associations are due to coexisting cardiovascular risk factors and comorbidity, or the greater severity of atherosclerotic disease among patients with depression [6]. There are several factors which may explain why our results are not consistent with these previous reports of the association between ACS, depression, and biomarkers of inflammation.

In our study, we used a structured diagnostic interview, the gold standard for defining clinical depression, while very few reports in the literature have used an interview-based study to examine the relationships between CVD, depression and inflammation. The majority of studies have used self-reported questionnaires such as the Beck Depression Inventory, to define depression, which are not specific for clinical depression. Thus, an elevated score may reflect transient sadness, anxiety, other psychiatric disorders, general distress secondary to a life-threatening experience, or even the symptoms of the medical illness [35].

In our study, patients who had a diagnosis of depression at study entry were excluded. If individuals with a known diagnosis of depression were not excluded, the cohorts would have had a greater prevalence of depression, than those where only individuals without clinically recognized depression were screened, and the relationship with inflammation could be skewed [26].

We also assessed depression at four time points, where as in the majority of studies examining the interaction of depressive symptoms with inflammation in patients with ACS have measured depressive symptoms only at the time of the hospitalization for ACS. Depressive symptoms at this stage may reflect transient “reactive” depression that may undergo spontaneous remission [5].

Finally, the association between depression and inflammation reported in the literature could be explained by inconsistencies in monitoring risk factors for CVD, co-morbidity, and CAD severity [6]. In our study we were able to control these factors, which could confound the association between depression and inflammation [6]. One difficulty in interpretation of reports in the literature, is the impact of medication such as statins [36]. In addition to their effects on cholesterol synthesis, statins also act as immunomodulators with pronounced anti-inflammatory properties [37]. Because of the optimal medical treatment provided to all patients during the 9 month follow-up period, we were able to minimize these confounding factors.

Due to the negative impact of depression on CVD outcome, in patients with CAD, identifying the onset of a MDD is particularly relevant for prevention and intervention [25,26,38-40].

Among all the markers tested in our model only personal history of depression was strongly associated with the occurrence of a MDD during the 9 months following an ACS. Personal history of depression identified patients with a greater vulnerability that were prone to depression. This marker is easy to assess, and may allow physicians to determine which patients could benefit from depression screening.

We acknowledge that this study has a limited number of patients that limit the power of some of statistical analyses and conclusions, namely in showing associations between clinical parameters and MDD occurrence (i.e. interaction between CRP and depression depending on the gender).

It is worth noting that levels of hsCRP and fibrinogen may not recapitulate all inflammatory processes, namely central nervous system inflammation. On the other hand, most often an increase in the levels of TNF-α, IL-1β or IL-6 will result in an increase in hsCRP or fibrinogen levels. However, it may be interesting to introduce other inflammatory markers in future studies.

This study is an exploratory study that may generate new knowledge and hypothesis, but it needs to be replicated.