Introduction

Effusive-constrictive pericarditis (ECP) is a clinical syndrome in which compressive pericardial fluid and a constricting visceral pericardium occur simultaneously [1]. The definitive diagnosis is established by demonstrating hemodynamics consistent with constrictive pericarditis, following complete drainage of pericardial effusion with normalization of intra-pericardial pressure [2]. The importance of recognizing ECP is that drainage of pericardial fluid alone when visceral constriction is present is ineffective in relieving pericardial compression; in the absence of visceral pericardiectomy, clinical evidence of hemodynamic impairment may persist and there may be progression to constrictive pericarditis over a short period of time [3]. Current prospective studies suggest that ECP is an uncommon pericardial syndrome [4]. The largest prospective study to date enrolled 15 patients after screening 1184 patients with pericarditis over 15 years; the prevalence of ECP amongst participants with pericardial disease was 1.3% and 6.8% amongst those with tamponade [1].

ECP is recognized as an important manifestation of pericardial tuberculosis [5], a disease that is associated with high morbidity and mortality [6,7]. There are, however, no studies of tuberculous ECP that have used combined pericardiocentesis and cardiac catheterization, the ‘gold standard’ for the diagnosis of effusive-constrictive disease [8]. Furthermore, it is unknown whether ECP has a biomarker profile that would shed light on the pathogenesis and diagnostic approach of this unique pericardial syndrome that is associated with inflammation and fibrosis. Cytokines such as interleukin-1β (IL-1β), tumor necrosis factor- α (TNF-α) and transforming growth factor β (TGF-β) have been implicated in the development of fibrosis in the context of a chronic inflammatory disease such as tuberculosis [9]. We postulated that the unique combination of fluid exudation and visceral constriction in effusive-constrictive disease may have a distinctive cytokine profile. We present the results of the first study of the prevalence and cytokine profile of proven effusive-constrictive disease in patients with tuberculous pericardial effusion.

Methods

This is a single center sub-study of the ongoing Investigation of Management of Pericarditis in Africa (IMPI Africa) registry of patients with suspected tuberculous pericarditis which was approved by the University of Cape Town Human Research Ethics Committee (HREC REF: 402/2005) [6]. This sub-study was conducted at Groote Schuur Hospital in Cape Town, South Africa. All participants gave written informed consent. The study complies with the Declaration of Helsinki.

Statistical Methods

The sample size required to provide adequate power to estimate the prevalence rate of ECP with a precision of 5% was 54, based on the anticipated 15% prevalence rate of tuberculous ECP from published literature [8]. The Shapiro-Wilk test was used to assess the distribution of continuous data. Baseline summary data are presented as medians with inter-quartile ranges (IQR) for variables with skewed distribution or means with standard deviation for variables with normal distribution. Differences in continuous measures between participants with or without ECP were tested using the Student t-test or Mann-Whitney test. Categorical data were compared using Pearson χ² test or, where appropriate, the Fisher exact test.

Logistic regression models were fitted to determine factors associated with ECP. The following variables that were tested: age, gender, history of tuberculosis, duration of symptoms, New York Heart Association Functional Class, HIV status, clinical signs of heart failure (i.e., jugular venous pressure, pulse rate, pulsus paradoxus, peripheral edema), electrocardiographic changes (i.e., heart rate, rhythm, and chest lead voltage), chest radiograph abnormalities (i.e., presence of pleural effusion, active pulmonary tuberculosis or evidence of previous tuberculosis), serum analysis (creatinine phosphokinase (CK), CK-MB fraction, NT-pro brain natriuretic peptide level, CD4 T cell count (in HIV positive individuals), lactate dehydrogenase (LDH), and total protein), pericardial fluid analysis (adenosine deaminase level, LDH, total protein, microscopy for acid fast bacilli, M. tuberculosis culture, and PCR), and invasive cardiac hemodynamic findings (pericardial fluid volume, intra-pericardial pressure before and after pericardiocentesis, right atrial pressure before and after pericardiocentesis, and presence of hemodynamic changes of cardiac tamponade). Variables found to be non-normally distributed were modeled as binary categorical variables using the first or the third quartile as a cut-off. Factors found significantly associated with ECP in univariable models were included in the final multivariable model. All tests were two-sided, and a p-value <0.05 was considered significant. Data analysis was performed with SPSS (version 17.0).

Results

Cytokine Profile

Table 2 shows the pericardial and serum levels of cytokines in the two groups. Patients with ECP had higher median levels of IL-10 (38.5 versus 0.2 pg/ml, P<0.001) and TGF-β (121.5 versus 29.1 pg/ml, P=0.02) in the serum compared to those with effusive non-constrictive pericarditis. In the pericardial fluid, participants with ECP had higher median levels of IL-10 (84.7 versus 20.4 pg/ml, P=0.006) and IFN-γ (2568.0 versus 906.6 pg/ml, P=0.03) than those without. No differences were noted between serum and pericardial fluid concentrations of IL-1β, IL-6, IL-17, IL-22, and TNF-α. These findings were independent of age, HIV status and whether or not the pericardial fluid was culture positive for M. tuberculosis. IFN-γ was not detected in the serum probably because ex-vivo levels tend to be lower than the detection threshold of the assay of 50 pg/ml[13]. Likewise, TGF-β was not detectable in pericardial fluid. Bioactive TGF-β, which was measured in this study, is often very low or undetectable in tuberculous pericarditis which is known to be a delayed type hypersensitivity response driven predominantly by Th-1 cytokines (e.g., IL-10) and characterized by the absence of TGF-β [14-16].

Compartment	Cytokine		Effusive-constrictive pericarditis		Effusive non-constrictive pericarditis	P-value
		N	Median (IQR)	N	Median (IQR)
Serum	IL-1β	33	7.1 (0.0-9.9)	29	7.8 (0.0-9.5)	0.90
	IL-6	31	7.9 (0.0-50.4)	28	29.1 (0.0-86.8)	0.65
	IL-10	31	38.5 (0.4-563.2)	29	0.2 (0.0-2.6)	<0.001
	IL-17A	31	0.0 (0.0-13.0)	29	5.0 (0.0-14.9)	0.50
	IL-22	31	74.5 (0.0-597.4)	28	60.1 (0.0-846.0)	0.96
	IFN-γ	31	0.0 (0.0-0.0)	29	0.0 (0.0-37.6)	NA
	TGF-β	31	121.5 (26.5-279.6)	29	29.1 (0.0-136.1)	0.02
	TNF-α	33	14.7 (0.0-109.8)	28	13.0 (0.0-50.3)	0.33
Pericardium	IL-1β	33	8.7 (0.8-25.5)	28	9.8 (0.4-11.7)	0.79
	IL-6	31	1608.2 (1218.4-4248.1)	28	1885.4 (1310.5-4993.4)	0.27
	IL-10	31	84.7 (0.9-392.7)	29	20.4 (0.0-62.2)	0.006
	IL-17A	31	0.0 (0.0-12.2)	29	3.6 (0.0-12.2)	0.49
	IL-22	31	349.5 (128.3-901.2)	28	210.8 (21.0-1165.1)	0.41
	IFN-γ	31	2568.0 (1318.0-6260.6)	29	906.6 (3.3-3447.4)	0.03
	TGF-β	31	0.0 (0.0-0.0)	30	0.0 (0.0-0.0)	0.97
	TNF-α	31	36.9 (8.6-113.0)	28	14.5 (7.7-50.0)	0.19

Table 2. Serum and pericardial cell-free fluid levels of cytokines at baseline in patients with tuberculous effusive constrictive and effusive non-constrictive pericardial disease (all values in pg/ml).

IQR, inter-quartile range IQR: inter-quartile ranges

CSV

Download CSV

Figure 1 is the visual representation of significant cytokine differences when comparing effusive with effusive-constrictive tuberculous pericarditis in the blood and pericardial fluid compartments (see Table 2 for results of all cytokines). Each colored disc represents the median value for the particular cytokine; the areas are proportional to the concentration of the cytokine. It is evident that the IFN-γ response in pericardial fluid is significantly greater in effusive-constrictive disease compared to that in effusive non-constrictive pericarditis. While the IFNγ response is thought to be protective, this is countered by an increased regulatory IL-10 response in pericardial fluid in the effusive-constrictive form of the disease. The most striking finding, however, is the extent to which IL-10 is elevated in serum in effusive-constrictive tuberculous pericarditis, at a ratio of 158 compared to effusive non-constrictive disease. Similarly, the fibrotic and regulatory cytokine, TGF-β is significantly elevated in the serum of effusive-constrictive tuberculous pericarditis, leading to the hypothesis that the immunopathology seen in ECP is largely driven by a pathological cross-regulatory and fibrotic response which abrogates the protective IFN-γ response, and leads to structural changes in the visceral pericardium which are manifested by constrictive physiology.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Visual representation of significant cytokine differences when comparing effusive with effusive-constrictive tuberculous pericarditis in the blood and pericardial fluid compartments (see Table 2 for results of all cytokines).

Each colored disc represents the median value for the particular cytokine; the areas are proportional to the concentration of the cytokine. IL-10, interleukin 10; TGF- β, transforming growth factor beta; IFN- γ, interferon gamma.

https://doi.org/10.1371/journal.pone.0077532.g001

Discussion

We have described the first prospective study of the prevalence and cytokine profile of hemodynamically proven effusive-constrictive disease in tuberculous pericardial effusion. We show that, contrary to non-tuberculous forms of pericardial disease where effusive-constrictive disease is uncommon, ECP occurs in half of cases of tuberculous pericardial effusion. In addition, we demonstrate that tuberculous ECP has a unique clinical, hemodynamic, and cytokine profile that is characterized by a younger age of onset (about a decade earlier than effusive non-constrictive disease), a high pre-pericardiocentesis right atrial pressure on cardiac catheterization, elevated serum and pericardial levels of IL-10, serum TGF-β and pericardial IFN-γ compared to effusive non-constrictive disease. Finally, we demonstrate for the first time that pre-pericardiocentesis right atrial pressure and IL-10 level were independently associated with effusive-constrictive disease in patients with tuberculous pericardial effusion.

Previous studies of ECP have suggested that it is an uncommon pericardial syndrome with a prevalence ranging from 2.4% to 14.8%, possibly because they were conducted in countries with a low prevalence of tuberculosis [8]. The current state of knowledge on the frequency of tuberculous ECP is based on studies that used clinical methods of diagnosis which are less accurate than invasive hemodynamic tests. The first prospective case series on the frequency of tuberculous ECP suggested a low prevalence of 2.6% based on non-invasive diagnostic criteria [17]. Subsequently, The IMPI Africa Registry reported that tuberculous ECP accounted for 15% of cases of suspected tuberculous pericarditis in the sub-Saharan African region, based on clinical and echocardiographic evaluation of survivors of tuberculous pericarditis [6]. The present study provides a more accurate estimate of the prevalence of effusive-constrictive disease in patients with tuberculous pericardial effusion based on an appropriate sample size [8], an acceptable case definition of tuberculous pericarditis [10], and a proven diagnosis of effusive-constrictive disease using the combination of pericardiocentesis and cardiac catheterization [1].

The finding that there was a relationship between the height of the pre-pericardiocentesis right atrial pressures and the diagnosis ECP was consistent with a similar observation by Hancock in the original case series on ECP [2]. The explanation provided then, to which we subscribe, was that the impact of the compression from the surrounding fluid and constricting visceral pericardium was additive. Interestingly the group of patients with ECP in that study as in ours, was approximately a decade younger than those with effusive non-constrictive pericarditis. It is possible that age related changes in the activity of cell mediated responses to infections like tuberculosis may offer an explanation for this finding [18]. We were not able to demonstrate an association between ECP and participant HIV status, or the degree of immunosuppression as measured by CD4 count. This is important because of suggestions that HIV might protect against the development constrictive pericarditis [19].

There have been no studies hitherto of the pathogenesis of ECP. We have studied the levels of cytokines that play a role in inflammation and fibrosis to shed light on the immunological basis of tuberculous ECP. We have found that tuberculous ECP differs from effusive non-constrictive disease, having a distinctive cytokine profile characterized by elevated serum IL-10 and TGF-β and high pericardial fluid IL-10 and IFN-γ. IL-10 plays a role in the regulation of the immune response to M. tuberculosis [18], while IFN-γ is a major inflammatory cytokine responsible for the control and resistance to mycobacterial infection (Figure 1) [20,21]. Furthermore, IL-10 and TGF-β have important roles in fibrosis following chronic inflammation [22-25]. While the combination of elevated IL-10 and IFN-γ within the pericardium may reflect a chronic unresolving inflammatory response to tuberculosis, the elevated serum levels of IL-10 and TGF-β may be a signal of early activation of pathways involved in long-term fibrosis offering a potential explanation for the constrictive physiology that was observed.

This pattern of cytokine expression may not only establish effusive-constrictive syndrome as biologically distinct in addition to the hemodynamic profile, but may have pathophysiological significance for pericardial remodeling following injury. The distensible visceral pericardium makes a vital contribution to cardiac compliance during diastole [26]. Disruption of the architecture of the visceral monolayer of mesothelial cells by inflammatory injury may impair epicardial elasticity and compliance, resulting in constriction and diastolic dysfunction. It has been suggested that in response to tuberculous antigens, TNF-α may lead to greater tissue injury in the presence of a mixture of regulatory and inflammatory cytokines than it does in the presence of either regulatory or inflammatory cytokines alone [27].

This study has a number of limitations. First, we enrolled patients with a moderate to a large pericardial effusion that was amenable to safe pericardiocentesis. This strategy may have introduced a degree of selection bias, and therefore the findings of this study apply to patients with moderate to large pericardial effusions with a probable and a definite diagnosis of tuberculous pericarditis. Second, a proven diagnosis of tuberculosis was made in 60% of cases. It is possible therefore that the 40% of participants with a presumed diagnosis may have had other inflammatory causes of pericarditis such as viral infection and autoimmune disease. The proportion of proven cases of tuberculosis is however typical of experience in tuberculosis endemic regions of the world [28]. Finally, while the sample size of 36 cases of ECP may appear to be small, this study is the largest series of proven cases of ECP to date. Nevertheless, a larger sample size will be required to provide reliable information on the outcome of ECP compared to effusive non-constrictive tuberculous pericarditis[8].

In conclusion, this report provides new information on the prevalence, hemodynamics, clinical and cytokine profile of cases of proven and presumptive tuberculous ECP. We show that effusive-constrictive disease is a common form of tuberculous pericardial effusion that affects younger patients with an extremely high pre-pericardiocentesis right atrial pressure and markedly elevated pericardial and serum IL-10 levels. The cytokine profile is consistent with chronic unresolving inflammation and fibrosis. These observations lay the basis for studies of outcome and management of effusive-constrictive tuberculous pericarditis, a clinical entity that has hitherto not been addressed in guidelines for the management of pericardial disease [29].

Acknowledgments

We are grateful to Sisters Naomi Hare, Joanne Hartnick, Veronica Francis, Unita September, Melanie Stahl, and Maitele Tshifularo, Mrs. Carolina Lemmer, Ms. Lerato Motete, Ms. Sharon Mosie, Ms. Margaret van den Berg, Ms. Connie Talliard, Ms. Lowena van Wyk, Mr. Simphiwe Nkepu and Mr. Jimmy Williams for assistance with this research.