Surgical outcomes of constrictive pericarditis and exploratory associations with inflammatory features: a 17-year single-center experience

Abstract

Background

Constrictive pericarditis (CP) remains a challenging disease with evolving etiological patterns and increasing emphasis on phenotype-based management. Pericardiectomy is the definitive treatment for CP. However, contemporary data on the etiological spectrum and the clinical relevance of inflammatory phenotypes in patients undergoing pericardiectomy are limited. We report the etiological spectrum, clinical phenotypes, and short- and long-term outcomes of pericardiectomy at a single center and explore whether a preoperative inflammatory phenotype is associated with adverse perioperative outcomes.

Methods

We retrospectively analyzed 36 consecutive patients who underwent pericardiectomy for CP between 2007 and 2024. A study-defined inflammatory phenotype—preoperative pericardial effusion plus C-reactive protein > 10 mg/L (n = 16)—was compared with all remaining patients (n = 20). Patients were evaluated according to etiology, radiologic morphology, inflammatory status, perioperative outcomes, and long-term survival. Survival was estimated by the Kaplan–Meier method.

Results

The mean age was 46.1 ± 17.6 years and 69.4% were male. Idiopathic disease was the most common etiology (44.4%), followed by tuberculosis (22.2%), while several uncommon causes, including familial Mediterranean fever, post-transplantation, sarcoidosis, Erdheim–Chester disease and post-COVID pericarditis, were identified. Hypoalbuminemia was more frequent in the inflammatory phenotype group (62.5% vs. 20.0%; p = 0.016). The inflammatory phenotype was also associated with a higher incidence of postoperative LCOS (p = 0.038). In-hospital mortality was observed in two patients (5.6%). After a median follow-up of 86.8 months, overall survival was 94.4%, 87.4% and 82.0% at 1, 5, and 10 years, respectively.

Conclusion

Pericardiectomy for constrictive pericarditis provided favorable short- and long-term outcomes in this single-center cohort. A study-defined preoperative inflammatory phenotype was associated with higher perioperative risk in this exploratory cohort. These preliminary findings highlight the potential role of perioperative inflammatory assessment in surgical patients, though validation in larger prospective registries is required.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12872-026-06535-9.

Introduction

Constrictive pericarditis (CP) is marked by chronic fibrous thickening of the pericardial sac, which leads to impaired diastolic function and increased filling pressures due to the loss of pericardial compliance [1]. The etiology of CP varies substantially by geography and healthcare setting. Tuberculosis remains an important cause in many low- and middle-income regions, whereas postsurgical, radiation-associated, and idiopathic cases are more commonly reported in Western series [2]. Furthermore, emerging etiological categories have recently been recognized, including IgG4-related disease, Tumor Necrosis Factor Receptor-Associated Periodic Syndrome (TRAPS), familial Mediterranean fever (FMF), post-COVID-19 pericarditis, and immune check pointinhibitor-induced cases. Within this changing etiological spectrum, Turkey may represent an epidemiological transition setting, where both Western-type idiopathic cases and traditional tuberculosis-related disease may coexist [3].

Patients typically present with symptoms resembling right heart failure, including dyspnea, peripheral edema, pulmonary congestion, hepatomegaly and ascites, despite preserved ventricular function [4]. The diagnosis of constrictive pericarditis relies on multimodality imaging, including echocardiography, computed tomography, cardiac magnetic resonance imaging, and cardiac catheterization, which together provide comprehensive structural and hemodynamic assessment [5, 6]. Surgical pericardiectomy remains the definitive therapy for CP and can be performed with relatively low perioperative mortality in experienced centers, making timely diagnosis and intervention crucial for patient outcomes [7].

In recent years, the diagnostic and therapeutic approach to constrictive pericarditis has undergone substantial evolution. The 2024 International Position Statement emphasized a multimodality imaging–basedphenotypic classification integrating inflammatory activity, hemodynamic status, and morphological findings into clinical decision-making [8]. Subsequently, the 2025 European Society of Cardiology (ESC) and American College of Cardiology (ACC) guidelines introduced an integrated management strategy for CP, combining inflammatory biomarkers, imaging findings, and surgical timing [4, 9]. Notably, the 2025 ACC pathway defined a “transient phenotype” within the subacute spectrum, characterized by elevated C-reactive protein (CRP) levels and cardiac magnetic resonance imaging of active inflammation, highlighting a potential therapeutic window for anti-inflammatory treatment prior to irreversible fibrotic remodeling [9].

While recent 2024/2025 ESC and ACC guidelines have introduced the concept of inflammatory phenotyping in pericardial disease, its application has largely been restricted to medical management and non-invasive imaging cohorts. The novelty of our study lies in translating this study-defined inflammatory framework specifically into a surgically treated constrictive pericarditis cohort undergoing pericardiectomy.

Pericardial effusion was selected as a clinical imaging feature frequently accompanying the inflammatory phenotype, but it should not be regarded as equivalent to CMR evidence of active inflammation. The exploratory contribution of the present study was to apply a pragmatic combination of routinely available preoperative findings—pericardial effusion and CRP > 10 mg/L—to a surgically treated cohort and to examine its association with early postoperative outcomes.

The study aimed to (1) analyse the current etiological landscape of CP in a tertiary center in Turkey, (2) explore the clinical associations of specific phenotypic classifications and a study-defined inflammatory phenotype, and (3) describe short and long-term survival outcomes following pericardiectomy.

Methods

Study design and ethical approval

This retrospective, single-center study aimed to compare clinical characteristics and surgical outcomes among patients with constrictive pericarditis based on the presence of pericardial effusion and calcification. We reviewed cases between 2007 and 2024, where patients received complete/pericardiectomy due to presence of constrictive pericarditis. Prior to the initiation of the study, ethical approval (approval number: 679379) was obtained from the Clinical Research Ethics Committee of Istanbul University-Cerrahpasa, Cerrahpasa Faculty of Medicine on 8 May 2023, ensuring compliance with the ethical guidelines outlined in the Declaration of Helsinki.

Patients and data collection

All consecutive patients undergoing pericardiectomy for clinically and/or radiologically confirmed constrictive pericarditis were included. Patients undergoing pericardial surgery for isolated effusion, malignancy without constrictive physiology, or non-constrictive indications were excluded. Between 2007 and 2024, a total of 36 patients underwent pericardiectomy procedures for constrictive pericarditis. Data was collected from the electronic medical records system of our institution, supplemented by information from surgical records and the national electronic health database. Cardiac output and cardiac index were not systematically recorded by thermodilution in our retrospective dataset and could therefore not be reported. Missing data, long-term survival, and clinical status during follow-up were verified through telephone interviews.

Definitions

The diagnosis of constrictive pericarditis was established through integration of clinical, imaging, and invasive hemodynamic findings, in line with the 2015 ESC and contemporary 2025 ESC/ACC pericardial guidelines [4, 9, 10]. CP was defined by clinical features of chronic right heart failure together with characteristic imaging findings — pericardial thickening or calcification on computed tomography, and echocardiographic evidence of impaired ventricular filling (septal bounce, respiratory variation in mitral inflow, or biatrial dilatation). When available, invasive hemodynamic assessment was used to confirm constrictive physiology which is defined as a difference between left ventricular and right ventricular end-diastolic pressures (LVEDP − RVEDP) ≤ 5 mmHg, a right ventricular end-diastolic pressure to right ventricular systolic pressure ratio (RVEDP/RVSP) > 1/3, and a pulmonary artery systolic pressure (PASP) < 55 mmHg. All patients received intraoperative confirmation at pericardiectomy. Preoperative CVP was measured on the operating table immediately prior to incision, whereas postoperative CVP was recorded upon admission to the intensive care unit (ICU).

Preoperative morphological phenotypes were defined using Computed Tomography (CT) and echocardiography as follows: effusion-dominant disease, mixed effusion-calcification disease, and calcification-dominant disease. The presence of effusion and calcification was determined from the original radiology and echocardiography reports. Low cardiac output syndrome (LCOS) was defined as the postoperative requirement for inotropic or vasopressor support beyond standard anesthesia weaning, the necessity for mechanical circulatory support (including intra-aortic balloon pump (IABP) or extracorporeal membrane oxygenation (ECMO), or persistent systemic hypotension despite achieving adequate intravascular volume status. Acute kidney injury was classified according to the KDIGO criteria. In-hospital mortality was defined as death occurring during the index admission or within 30 days postoperatively. The inflammatory phenotype was operationally defined as the presence of preoperative pericardial effusion together with a CRP level > 10 mg/L. This threshold was adopted from established clinical trial eligibility criteria for active pericardial inflammation [11–13]. To evaluate the stability of our endpoints across different diagnostic criteria, a sensitivity analysis was conducted by varying the preoperative CRP cutoffs from 5 to 25 mg/L.

Surgical procedure

Depending on individual patient characteristics, disease distribution, and anatomical considerations, surgical access was achieved through either median sternotomy or left antero-lateral thoracotomy. The relative use of the two approaches nevertheless changed over the study period, with seven of the eight left antero-lateral thoracotomies performed before 2016 and median sternotomy used almost exclusively thereafter. In the earlier years a left anterolateral thoracotomy was the customary route when the pericardium required direct inspection, whereas in later years median sternotomy was preferred for its wider exposure and the ability to institute cardiopulmonary bypass if required. All patients underwent complete pericardiectomy. The intended extent of resection was a broad anterior pericardiectomy between the phrenic nerves, with additional removal of the diaphragmatic pericardium and, when safely accessible, pericardial tissue extending toward the great vessels and posterior to the left phrenic nerve. The extent of resection was tailored according to operative exposure, the severity of adhesions, calcification, and hemodynamic stability.

During decortication, pericardial tissue was meticulously dissected from the ventricular surfaces down to the atrioventricular grooves, with careful preservation of the phrenic nerves and adjacent structures. Calcific deposits were removed when technically feasible, particularly from the diaphragmatic surface of the left ventricle and from the right atrial and caval regions when adequately exposed. Decortication was generally initiated over the left ventricular surface to facilitate early ventricular decompression, followed by progressive release of the right-sided chambers. Cardiopulmonary bypass was used selectively to allow safe completion of resection in patients with dense fibrotic adhesions, severe calcification, or intraoperative hemodynamic instability.

Statistical analysis

Continuous variables were summarized as mean ± standard deviation or median with interquartile range (IQR), as appropriate, and compared between groups using the Mann–Whitney U test, given the small sample size and non-normal distribution of several variables; the Kruskal–Wallis test was used for the three–group morphological phenotype analysis. Paired preoperative and postoperative measurements were compared with the Wilcoxon signed-rank test. Categorical variables were expressed as counts and percentages and compared using Fisher’s exact test. For the principal binary outcome comparisons, effect estimates are reported as risk ratios with 95% confidence intervals, together with absolute risk differences where informative. For continuous outcomes, median differences with percentile bootstrap 95% confidence intervals were calculated where feasible.

The inflammatory phenotype was evaluated as a study-defined binary variable (preoperative pericardial effusion plus CRP > 10 mg/L versus all remaining patients), with an exploratory sensitivity analysis across alternative preoperative CRP thresholds to assess the consistency of the direction of association rather than to derive an optimal cutoff. Given the small number of events, multivariable modelling was exploratory only. Binary logistic regression was performed for postoperative low cardiac output syndrome (13 events), entering no more than one covariate alongside the inflammatory phenotype in any model, corresponding to approximately 6.5 events per variable. Candidate covariates were age, hypoalbuminaemia and NYHA class III/IV; C-reactive protein was ineligible because it forms one of the two components of the phenotype definition. Follow-up duration was estimated by the reverse Kaplan–Meier method. Overall survival, defined as the time from pericardiectomy to death from any cause with patients censored at last contact, was estimated using the Kaplan–Meier method—reporting 1-, 5-, and 10-year survival with 95% confidence intervals—and compared with the log-rank test. All p-values were two-sided, with p < 0.05 considered statistically significant. Analyses were performed in Python 3 using pandas, SciPy, and lifelines.

Results

Demographics and etiology

Table 1 provides a comprehensive overview of the demographic characteristics and medical backgrounds of the 36 patients. Patients ranged in age from 19 to 83 years old, with a mean age of 46.1. Of those, 25 (69.4%) were male.

Table 1.

Patients demographics and medical background

Patients (n=36)	n (%)	Mean ±SD	Median (IQR)
Age (years)		46.1 ± 17.6
Gender (male)	25 (69.4%)
Presentations
Dyspnea	21 (58.3%)
Ascites	16 (44.4%)
Hepatomegaly	15 (41.7%)
NYHA
Class I&II	11 (30.6%)
Class III	11 (30.6%)
Class IV	14 (38.9%)
Etiology of Constrictive Pericarditis
Idiopathic	16 (44.4%)
Tuberculosis	8 (22.2%)
Familial Mediterranean Fever	2 (5.6%)
Post-Cardiac Surgery	2 (5.6%)
Erdheim–Chester disease	1 (2.8%)
Sarcoidosis	1 (2.8%)
Post-radiation	1 (2.8%)
Covid-19	1 (2.8%)
Cardiac Trauma	1 (2.8%)
Combined Heart-Kidney Transplantation	1 (2.8%)
Isolated Renal Transplantation	1 (2.8%)
Neoplastic (malignant mesothelioma)	1 (2.8%)
Preoperative Laboratory Findings
Albumin, g/dL		3.77 ± 0.80
Hypoalbuminemia	(< 3.5 g/dL)		14 (45.2%)
Urea (BUN), mg/dL			28.5 (22.8–51.2)
Creatinine, mg/dL			0.80 (0.68–1.02)
Elevated creatinine	(> 1.3 mg/dL)	3 (8.3%)

Values are expressed as numbers and percentages. The patient classified under "COVID-19" had a background diagnosis of POEMS syndrome, but there was no clinical, laboratory or imaging evidence of active or relapsing POEMS at the time the constrictive physiology developed; the respective contributions of the two conditions cannot be separated in a single case

The most common presenting symptom was dyspnea (58.3%). According to the New York Heart Association (NYHA) classification, 38.9% of patients presented with Class IV symptoms, whereas 30.6% were in Class III and 30.6% in Class I-II.

The most frequent etiology of constrictive pericarditis was idiopathic origin (44.4%, n = 16), followed by tuberculosis (22.2%, n = 8). One patient had neoplastic constrictive pericarditis due to malignant pleural mesothelioma. Specific systemic and autoimmune diseases included familial Mediterranean fever (5.6%, n = 2), Erdheim–Chester disease (2.8%, n = 1), and sarcoidosis (2.8%, n = 1). Regarding procedural and environmental etiologies, two patients had a history of cardiac surgery, one had post-radiation constrictive disease, and two patients developed constrictive pericarditis after transplantation — one following combined heart–kidney transplantation and the other following isolated renal transplantation. Additionally, isolated cases were linked to prior cardiac trauma and post-COVID-19 pericarditis with POEMS syndrome. In the patient with POEMS syndrome, the hematology department confirmed that there were no clinical, laboratory, or imaging findings of active or recurrent POEMS disease at the time of pericarditis development; therefore, the narrowing was attributed to a previous COVID-19 infection, not to POEMS.

Radiologic and hemodynamic findings

Echocardiographic findings revealed a median left ventricular ejection fraction (LVEF) of 55.0% (IQR 50.0–55.0). Tricuspid valve regurgitation and mitral valve regurgitation were detected in 52.8% of patients. Pericardial effusion was confirmed in 63.9%. Pericardial calcification was detected in 58.3% (n = 21/36) of the patient cohort during preoperative radiologic evaluations.

Hemodynamic evaluation demonstrated a statistically significant reduction in central venous pressure (CVP) following pericardiectomy. CVP decreased from a preoperative median of 24.0 mmHg (IQR 17.8–30.0) to a postoperative median of 11.5 mmHg (IQR 9.0–12.2) (Wilcoxon signed-rank test, p = 0.001).

Hemodynamic assessment demonstrated elevated filling pressures consistent with constrictive physiology, right ventricular end-diastolic pressure of 21.0 mmHg (IQR 18.5–24.8), pulmonary capillary wedge pressure of 23.0 mmHg (IQR 21.0–28.0), and left ventricular end-diastolic pressure of 25.0 mmHg (IQR 20.0–28.5) (Table 2).

Table 2.

Radiologic, echocardiographic, and hemodynamic findings

	n	Median (IQR)	n (%)
Echocardiography
LVEF (%)	36	55.0 (50.0–55.0)
Mitral Valve Regurgitation (≥ moderate)	36		19 (52.8)
Tricuspid Valve Regurgitation (≥ moderate)	36		19 (52.8)
Pericardial Effusion	36		23 (63.9)
Computed Tomography
Pleural Effusion	36		18 (50.0)
Pericardial Calcification	36		21 (58.3)
Catheterization
RV systolic pressure, mmHg	22	41.5 (38.0–45.0)
RV end-diastolic pressure, mmHg	22	21.0 (18.5–24.8)
PASP, mmHg	29	32.0 (28.0–41.0)
PCWP, mmHg	21	23.0 (21.0–28.0)
LV systolic pressure, mmHg	20	110.0 (104.8–126.5)
LV end-diastolic pressure, mmHg	20	25.0 (20.0–28.5)

Continuous variables are reported as median (interquartile range), categorical variables as n (%)

LV Left ventricular, LVEF Left ventricular ejection fraction, PASP Pulmonary arterial systolic pressure, PCWP Pulmonary capillary wedge pressure, RV Right ventricular

Surgical approach and postoperative outcomes

Median sternotomy was the predominant surgical approach, performed in 77.8% of patients, while left antero-lateral thoracotomy was utilized in selected cases. Cardiopulmonary bypass was required in four patients (11.1%). Indications for CPB included concomitant single-vessel CABG in one patient (n = 1), whereas the remaining three cases (n = 3) required CPB due to operational factors. The median intensive care unit stay was 2.0 days (IQR: 2.0–3.2), whereas the median length of hospital stay was 8.5 days (IQR: 7.0–11.2) (Table 3).

Table 3.

Operative characteristics and postoperative outcomes

Patients (n=36)	n (%)	Median (IQR)	95% CI
Surgical Approach
Median sternotomy	28 (77.8%)		61.9–88.3
Left antero-lateral thoracotom	8 (22.2%)		11.7–38.1
Cardiopulmonary bypass	4 (11.1%)		4.4–25.3
Postoperative Recovery
ICU stay, days		2.0 (2.0-3.2)
Hospital stay, days		8.5 (7.0–11.2)
Postoperative Complications
RBC Transfusion >3 Units	19 (52.8%)		37.0–68.0
AF	14 (38.9%)		24.8–55.1
LCOS	13 (36.1%)		22.5–52.4
Reoperation	3 (8.3%)		2.9–21.8
Acute Kidney Injury	3 (8.3%)		2.9–21.8
Need for ECMO	2 (5.6%)		1.5–18.1
Mortality
In-hospital mortality	2 (5.6%)		1.5–18.1
Overall all-cause mortality	5 (13.9%)		6.1–28.7

Data were complete for all variables in this table (n = 36); no denominators differ from the full cohort. Confidence intervals for proportions are Wilson score intervals. Confidence intervals are not reported for medians. Overall all-cause mortality includes the two in-hospital deaths

ICU Intensive care unit, RBC Red blood cell, AF Atrial fibrillation, ECMO Extracorporeal membrane oxygenation, ICU Intensive care unit, LCOS Low cardiac output syndrome

New-onset atrial fibrillation occurred in 38.9% of patients, followed by low cardiac output syndrome in 36.1% of cases. Reoperation due to bleeding was required in three patients (8.3%). Acute kidney injury developed in 8.3% of cases, and ECMO support was required in two patients (5.6%). No patient required IABP support. In-hospital mortality was observed in two patients (5.6%), while all-cause mortality during follow-up was 13.9%.

Morphological and inflammatory phenotype comparisons

The cohort was classified into three mutually exclusive imaging-defined morphological phenotypes based on the combined assessment of preoperative pericardial effusion and calcification: effusion-dominant disease without calcification (n = 15, 41.7%), mixed effusion-calcification disease (n = 8, 22.2%), and calcification-dominant disease without effusion (n = 13, 36.1%). No patient lacked both features. Demographic characteristics, including age, sex, and NYHA functional class, did not differ significantly across these phenotypes. Preoperative CRP demonstrated a graded decrease from the effusion-dominant group to the mixed and calcification-dominant groups, with median values of 34.3, 10.2, and 5.0 mg/L, respectively (p = 0.018). Other perioperative outcomes, including acute kidney injury, in-hospital mortality, ICU stay, hospital stay, and postoperative CRP, did not differ significantly across the three groups (Table 4).

Table 4.

Preoperative characteristics and perioperative outcomes by morphological phenotype (n = 36)

Variable	Group A Effusion Dominant	Group B Mixed	Group C Calcification Dominant	p (overall, 3-group)	p (A vs C)
	(n=15)	(n=13)	(n=13)
Preoperative characteristics
Age, years	49.0 (33.0–58.0)	54.5 (44.2–64.2)	35.0 (24.0–61.0)	0.222	0.645
Male, n (%)	11 (73.3)	6 (75.0)	8 (61.5)	0.805	0.689
NYHA III–IV	10 (66.7)	7 (87.5)	8 (61.5)	0.524	1.000
CRP, mg/L	34.3 (17.4–120.1)	10.2 (4.2–37.7)	5.0 (3.6–11.6)	0.018	0.006
Albumin, g/dL	3.3 (3.2–3.7)	3.8 (3.5–4.1)	4.3 (3.2–4.7)	0.668	0.518
	n=12a	n=8a	n=11a
Perioperative outcomes
LCOS, n (%)	6 (40.0)	5 (62.5)	2 (15.4)	0.093	0.221
AKI, n (%)	2 (13.3)	1 (12.5)	0 (0.0)	0.422	0.484
In-hospital	1 (6.7)	1 (12.5)	0 (0.0)	0.694	1.000
mortality, n (%)
ICU stay, days	2.0 (2.0–4.0)	3.0 (2.8–3.2)	2.0 (1.0–2.0)	0.235	0.184
Hospital stay, days	7.0 (5.0–10.5)	10.0 (6.8–15.2)	9.0 (8.0–10.0)	0.315	0.219
CRP	28.2 (21.4–60.2)	25.9 (10.1–83.5)	19.9 (14.8–26.9)	0.471	0.227
Postoperative, mg/L	n=12a	n=8a	n=7a

Continuous variables are reported as median (interquartile range). Categorical variables are reported as n (%). Group A = effusion-dominant; Group C = calcification-dominant. Three-group comparisons used the Kruskal–Wallis test for continuous and Fisher's exact test for categorical variables; A versus C comparisons used the Mann–Whitney U test or Fisher's exact test

CRP C-reactive protein, ICU Intensive care unit, NYHA New York Heart Association

Statistically significant values (p < 0.05) are shown in bold

^aDenominators are shown where data were incomplete: preoperative albumin was missing in 5 patients overall and postoperative CRP in 9. All other variables were complete for all 36 patients

Preoperative CRP was available for all 36 patients, with no missing values. Twenty-three patients had a pericardial effusion, of whom 16 also had CRP > 10 mg/L; these 16 constitute the inflammatory phenotype. The remaining seven had an effusion with CRP < = 10 mg/L. Of the 13 patients without an effusion, four had CRP > 10 mg/L and nine had neither feature. Because the phenotype requires both features, the non-inflammatory comparator (n = 20) is therefore composed of seven patients with effusion but low CRP, four with raised CRP but no effusion, and nine with neither.

Sensitivity analyses across CRP thresholds of > 5, >10, > 15, >20, and > 25 mg/L showed a consistent direction of association, although effect estimates and statistical precision varied across thresholds. No single threshold was uniformly optimal (Supplementary Table S1).

Patients with the inflammatory phenotype (n = 16) demonstrated a higher incidence of LCOS than the non-inflammatory group (n = 20) (56.2% vs. 20.0%; risk ratio 2.81, 95% CI 1.06–7.47; p = 0.038) and a longer ICU stay (median 3.0 vs. 2.0 days; rank-biserial r = 0.41; p = 0.034). Preoperative albumin, available in 31 patients, was lower in the inflammatory phenotype group (median 3.26 vs. 4.40 g/dL; median difference − 1.14, 95% CI − 1.45 to − 0.40; p = 0.013), and hypoalbuminemia (< 3.5 g/dL) was present in 10 of 13 versus 4 of 18 patients (76.9% vs. 22.2%; risk ratio 3.46, 95% CI 1.39–8.64; p = 0.004). Postoperative CRP, available in 27 patients, was higher in the inflammatory phenotype group (median 39.2 vs. 19.9 mg/L; median difference + 19.3, 95% CI 1.4 to 80.2; p = 0.027) (Table 5).

Table 5.

Outcomes by inflammatory phenotype

Variable	Inflammatory phenotype	Non-inflammatory	Effect size	p-value
	(n=16)	(n=20)	(95% CI)
LCOS	9 (56.2%)	4 (20.0%)	RR 2.81 (1.06–7.47)	0.038
Acute kidney injury	3 (18.8%)	0 (0.0%)		0.078
Reoperation	2 (12.5%)	1 (5.0%)		0.574
Atrial fibrillation	7 (43.8%)	7 (35.0%)		0.734
In-hospital mortality	2 (12.5%)	0 (0.0%)		0.190
ICU stay, days	3.0 (2.0–4.0)	2.0 (1.0–3.0)	r = 0.41	0.034
Hospital stay, days	9.5 (5.8–13.0)	8.0 (7.0–10.0)		0.713
Postoperative	39.2 (26.4–93.2)	19.9 (10.5–26.2)		0.027
CRP, mg/L	n =14	n =13
Preoperative	3.26 (3.20–3.47)	4.40 (3.73–4.70)		0.013
albumin, g/dL	n =13	n =18
Hypoalbuminemia (<3.5 g/dL), n (%)	10/13 (62.5 76.9%)	4/18 (20 22.2%)	RR 3.46 (1.39–8.64)	0.016 0.004

Continuous variables are reported as median (interquartile range). Categorical variables are reported as n (%). The inflammatory phenotype was defined as preoperative pericardial effusion together with C-reactive protein >10 mg/L (16 patients; the remaining 20 formed the comparator). Comparisons used the Mann–Whitney U test for continuous and Fisher's exact test for categorical variables. Preoperative albumin was not available in 5 patients (3 inflammatory, 2 non-inflammatory); all albumin analyses were therefore performed on available cases without imputation, and denominators are shown accordingly. Postoperative CRP was available in 27 patients. Preoperative C-reactive protein and all other variables were complete for all 36 patients. r denotes rank-biserial correlation

CI Confidence interval, CRP C-reactive protein, ICU Intensive care unit, OR Odds ratio

Statistically significant values (p < 0.05) are shown in bold

Exploratory logistic regression analyses were performed to assess the stability of the association between the study-defined inflammatory phenotype and postoperative LCOS. Because only 13 LCOS events occurred, each model included the inflammatory phenotype and only one additional covariate. Models involving albumin were restricted to the 31 patients with an available preoperative value; all 13 events occurred within this subset, so no outcome events were lost. In univariable analysis, the inflammatory phenotype was associated with LCOS (OR 5.14, 95% CI 1.18–22.48; p = 0.030). The association was attenuated and no longer significant after adjustment for age (OR 3.25, 95% CI 0.61–17.18; p = 0.166), whereas it persisted after adjustment for hypoalbuminemia (OR 7.16, 95% CI 1.08–47.56; p = 0.042), for albumin as a continuous variable (OR 7.41, 95% CI 1.27–43.23; p = 0.026) and for NYHA class III/IV (OR 5.05, 95% CI 1.11–23.04; p = 0.037). Age was itself associated with LCOS (OR 1.08 per year, 95% CI 1.01–1.15; p = 0.016), whereas neither hypoalbuminemia (OR 1.20, 95% CI 0.18–7.93; p = 0.853) nor albumin as a continuous variable (OR 0.91, 95% CI 0.29–2.87; p = 0.868) was independently associated with it. Confidence intervals were wide throughout, and these models were considered exploratory rather than evidence of independent prediction (Supplementary Table S2).

Long-term survival

The cohort contributed a total of 263 patient-years of follow-up. The median follow-up duration, estimated using the reverse Kaplan–Meier method, was 86.8 months (IQR 29.9–179.2; range: 0.5–227.2 months), during which a total of five deaths occurred. The overall estimated Kaplan–Meier survival was 94.4% (95% CI 79.6–98.6) at 1 year, 87.4% (95% CI 69.7–95.1) at 5 years, and 82.0% (95% CI 60.6–92.4) at 10 years (Fig. 1A). To achieve adequate subgroup sizes for survival comparison, patients were stratified into three major etiological groups: tuberculous (n = 8), idiopathic/post-viral (including idiopathic and post-COVID-19, n = 17), and inflammatory/other (comprising familial Mediterranean fever, sarcoidosis, Erdheim–Chester disease, post-cardiac surgery, post-transplantation, post-radiation, and cardiac trauma, n = 10). One patient with neoplastic (mesothelioma-associated) constrictive pericarditis was excluded from the stratified comparison as a single-patient stratum but remained in the overall analysis. When stratified (Fig. 1B), no deaths occurred in the tuberculous group (5- and 10-year survival: 100.0%), whereas the idiopathic/post-viral group demonstrated 5- and 10-year survival rates of 87.4% (2 deaths). The heterogeneous inflammatory/other group showed a 5-year survival of 85.7% and a 10-year survival of 57.1% (2 deaths). This between-group difference did not reach statistical significance (log-rank p = 0.392), likely reflecting limited statistical power due to small subgroup sizes and the small number of events.

Fig. 1.

Kaplan–Meier survival after pericardiectomy. A Overall survival for the entire cohort (n = 36); survival was 94.4%, 87.4%, and 82.0% at 1, 5, and 10 years, respectively. B Survival stratified by etiological subgroup (tuberculous [n = 8], idiopathic/post-viral [n = 17], and inflammatory/other [n = 10]; one patient with neoplastic/mesothelioma-associated constrictive pericarditis was excluded as a single-patient stratum); between-group difference by log-rank test, p = 0.392. Tick marks denote censored observations; the number of patients at risk is shown below each panel. The post-COVID-19 case with underlying POEMS syndrome was grouped with the idiopathic/post-viral category for survival analysis because the temporal clinical trigger was post-viral pericarditis; however, a contribution of POEMS-related systemic inflammation cannot be excluded

Discussion

This 17-year single-center cohort of 36 patients undergoing pericardiectomy for constrictive pericarditis yielded four principal observations. First, the etiological landscape of CP in our cohort differed considerably from earlier series, with tuberculosis declining and idiopathic causes predominating; notable contributions from rare etiologies including familial Mediterranean fever, post-transplantation, and post-COVID disease were also observed. Second, the morphological spectrum spanning from effusion-dominant to calcification-dominant disease was reflected biochemically by contrasting preoperative CRP levels, supporting the concept that these phenotypes may correspond to different stages or patterns of pericardial disease. Third, an active inflammatory state characterized by preoperative effusion and elevated CRP levels was significantly associated with adverse early postoperative outcomes, suggesting that active inflammation may adversely affect short-term surgical results. Finally, despite the heterogeneity of underlying etiologies, long-term survival following pericardiectomy was favorable, suggesting that satisfactory surgical outcomes can be achieved with appropriate perioperative management.

The etiological distribution observed in our cohort reflects an ongoing epidemiological transition in CP in Turkey. Tuberculosis accounted for 22.2% of cases, representing a substantial decline compared with earlier Turkish surgical series, in which tuberculous constrictive pericarditis had historically been the predominant etiology, with reported rates approaching 50% [14]. In parallel with this decline, idiopathic etiologies have become increasingly dominant in recent years, consistent with the gradual epidemiological shift observed in middle-income countries transitioning toward Western disease patterns. The diversity of contributing conditions in our series, including familial Mediterranean fever, post-cardiac and post-renal transplantation, sarcoidosis, Erdheim–Chester disease, and post-COVID pericarditis, mirrors the expanding etiological spectrum highlighted in recent reviews and underscores the importance of a broad differential diagnostic approach.

Particularly noteworthy within this diverse spectrum was a patient in our cohort who developed CP 14 months after combined heart and renal transplantation. The clinical course was characterized by progressive right-sided heart failure symptoms initially raising suspicion for graft rejection, while serial echocardiographic examinations demonstrated preserved or mildly reduced ventricular function accompanied by septal motion abnormalities and minimal pericardial effusion. Definitive diagnosis was ultimately established through invasive hemodynamic assessment and CT imaging demonstrating pericardial thickening without calcification. (Fig. 2-A) The patient successfully underwent pericardiectomy (Fig. 2-B), was postoperatively managed with venoarterial extracorporeal membrane oxygenation (VA-ECMO) support in the intensive care unit, and was subsequently discharged. This case highlights the diagnostic complexity of post-transplant constrictive pericarditis and emphasizes the importance of considering pericardial constriction in transplant recipients presenting with unexplained heart failure symptoms despite inconclusive routine imaging findings.

Fig. 2.

A Preoperative chest CT scan demonstrating post-transplant pericardial constriction. Axial and sagittal chest CT scans showing significant, non-calcified pericardial thickening in a patient presenting with progressive right-sided heart failure 14 months after combined heart and renal transplantation. B Intraoperative view of pericardiectomy in the post-transplant patient. The thickened and densely adherent pericardium is dissected from the underlying cardiac allograft during surgical release of constrictive physiology. C Invasive right ventricular pressure tracing from the same patient demonstrating the classic “dip-and-plateau” morphology (square root sign). Blue arrows indicate the early diastolic dip — the rapid pressure nadir occurring immediately after the brisk Y descent, while orange arrows indicate the sustained mid-to-late diastolic plateau, confirming severe constrictive physiology. Paper speed 25 mm/s

Our findings support the presence of distinct inflammatory and morphological phenotypes in CP. Our cohort showed three morphologically and biochemically distinct patterns: a predominantly effusive group with higher inflammatory marker levels, a predominantly calcific group with lower CRP levels, and an intermediate mixed group combining features of both. These findings are broadly consistent with the spectrum of inflammatory and chronic morphological patterns described in contemporary guidance, although the absence of systematic CMR assessment precludes direct classification according to these imaging-based phenotypes. The morphological intermediate (mixed) group showed numerically higher perioperative morbidity in our series, plausibly reflecting the combined burden of active pericardial inflammation and the technical surgical challenges imposed by pericardial calcification. Although our sample is too small to support definitive risk-stratification claims, the gradient of preoperative C-reactive protein across the three phenotypes (p = 0.018) provides convergent biochemical support for the proposed morphological spectrum and aligns conceptually with the imaging-based phenotype classifications increasingly used in pericardial disease management. The absence of cardiac magnetic resonance imaging in our cohort precludes direct mapping of these morphological phenotypes onto the CMR-defined transient/chronic categories described in the guideline framework, and represents a methodological limitation discussed further below.

Importantly, when an inflammatory phenotype characterized by pericardial effusion and elevated preoperative CRP levels was applied to our cohort in accordance with contemporary concepts of inflammatory pericardial disease, this subgroup demonstrated a significantly higher incidence of postoperative low cardiac output syndrome and prolonged intensive care unit stay. The lower albumin levels in the inflammatory phenotype group may reflect a greater systemic inflammatory burden, more advanced right-sided congestion, impaired nutritional status, or a combination of these mechanisms. However, the attenuation of the association after adjustment for age suggests that the CRP–effusion combination may partly capture underlying differences in age and overall disease severity rather than representing an independent inflammatory effect. The association persisted after adjustment for hypoalbuminemia and for NYHA class, but the wide confidence intervals and limited number of events prevent any conclusion regarding independent prediction.

The association between preoperative patient status and postoperative LCOS has also been demonstrated in larger surgical series. Yan et al. identified NYHA class III/IV, reduced left ventricular mass index and hypoalbuminemia as independent predictors of LCOS after pericardiectomy for tuberculous constrictive pericarditis, while Wang et al., in 212 patients undergoing isolated pericardiectomy, described markers of more advanced haemodynamic compromise, including elevated preoperative filling pressures, in this setting. Hypoalbuminemia was markedly more frequent in our inflammatory phenotype group (76.9% versus 22.2%; risk ratio 3.46, 95% CI 1.39–8.64), yet it was not independently associated with LCOS in our models, most likely reflecting our smaller sample rather than a genuine discrepancy. These observations are consistent with the possibility that the association observed in our cohort reflects not only inflammatory activity itself but also the broader burden of systemic congestion, nutritional impairment and advanced constrictive physiology. Accordingly, the study-defined inflammatory phenotype should not be interpreted as an independent predictor of LCOS.

The C-reactive protein threshold of > 10 mg/L was selected a priori based on its established use as the eligibility cutoff for active pericardial inflammation in the pivotal anti–interleukin-1 trials in recurrent pericarditis, and is the inflammatory threshold to which the 2025 American College of Cardiology Concise Clinical Guidance explicitly refers when discussing biologic therapy in pericarditis [9, 11–13]. Applying this established framework to a surgical cohort, we observed that the phenotype tracked with early postoperative outcomes. Our findings provide initial real-world evidence that inflammatory phenotyping carries direct clinical relevance for surgical risk stratification and perioperative optimization. In selected cases, a time-limited empiric anti-inflammatory trial (nonsteroidal anti-inflammatory drugs and colchicine, with corticosteroids in selected cases) may be considered prior to surgical referral, which may unmask transient constriction and avert pericardiectomy in a subset of patients, in line with current guideline-based management of inflammatory pericardial disease [4, 9, 15, 16]. Our findings suggest that recognition of an inflammatory phenotype may be clinically relevant when considering preoperative management, although our study was not designed to evaluate the efficacy of anti-inflammatory therapy. In contrast, calcific constrictive pericarditis may represent a surgically definitive but technically more demanding phenotype due to dense adhesions and rigid pericardial remodeling.

Postoperative mortality and long-term survival outcomes in our cohort were consistent with contemporary pericardiectomy series. In-hospital mortality was 5.6%, which falls within the range reported in recent surgical studies, where perioperative mortality generally varies between 0% and 8% depending on patient selection, etiology, and operative complexity [17–19]. The two in-hospital deaths occurred in distinct clinical settings: one followed cardiac tamponade on postoperative day 15 that progressed to refractory low cardiac output syndrome despite reintervention, whereas the other occurred in a patient with malignant mesothelioma and severe right heart failure who could not be weaned from postoperative ECMO support. These cases highlight the heterogeneous mechanisms underlying early mortality after pericardiectomy, including both procedure-related complications and advanced underlying disease.

The 36% incidence of low cardiac output syndrome observed in our cohort, although higher than in some published series, is consistent with reports identifying preoperative elevated central venous pressure as a predictor of LCOS and likely reflects the advanced functional status (NYHA III–IV in 69%) of patients referred for surgery in our institution. Kaplan–Meier analysis demonstrated favorable long-term survival, with outcomes remarkably consistent with previously published Turkish and international surgical series [14, 17, 20]. The reverse Kaplan–Meier–derived median follow-up was 86.8 months (7.2 years), allowing meaningful assessment of late postoperative outcomes. Notably, no mortality occurred among patients with tuberculous constrictive pericarditis during follow-up. This finding supports previous evidence suggesting that, when combined with appropriate antituberculous therapy, surgical pericardiectomy can provide favorable long-term results in carefully managed tuberculosis-related disease [20]. In contrast, patients within the inflammatory/other etiological subgroup demonstrated relatively lower long-term survival, although statistical significance was not reached, likely because of the limited number of events and small subgroup sizes. Nevertheless, this trend may indicate the prognostic heterogeneity associated with non-classical etiologies of constrictive pericarditis.

The role of cardiopulmonary bypass in pericardiectomy for constrictive pericarditis remains a subject of debate. In our cohort, CPB was utilized in 11.1% (n= 4) of cases, positioning our surgical approach on the selective side of the current international debate [17, 18]. This contrasting strategy differs from centers advocating for routine CPB adoption [21]. The rationale behind our selective CPB strategy relies on minimizing cross-clamping time and perioperative bleeding risks, provided that a complete resection can be safely achieved. Recent evidence confirms comparable outcomes between off-pump and on-pump pericardiectomy, supporting a case-by-case decision-making model [22]. Our institutional practice favors a tailored approach, reserving CPB for cases of severe hemodynamic compromise or dense fibrotic adhesions limiting safe off-pump dissection.

In one notable case, an inadvertent injury occurred at the descending aorta during the mobilization of the left pulmonary veins. The preoperative chest CT scan of this patient demonstrated extensive, dense calcific remodeling encasing the cardiac chambers (Fig. 3). Because the resulting hemorrhage could not be safely controlled under beating-heart conditions, systemic heparinization was performed, arterial and venous cannulation were rapidly established via the aorta and right atrium, and the injury was successfully repaired with 4 − 0pledgeted polypropylene sutures under CPB support. This complication underscores the critical importance of maintaining readiness for CPB even during intended off-pump procedures. In addition to vascular risks, structural decompression after removing the rigid pericardial shell can alter ventricular geometry and annular alignment. Illustrating this challenge, one patient with baseline mild-to-moderate mitral regurgitation (MR) progressed to severe MR within the first postoperative week, requiring mechanical mitral valve replacement. Median sternotomy was preferred in 77.8% of our patients, which is consistent with the literature [17, 18]. Our routine phrenic-to-phrenic decortication is consistent with contemporary recommendations favoring radical or complete over partial pericardiectomy to minimize residual constriction and recurrence, while our selective—rather than routine—use of cardiopulmonary bypass reflects an individualized balance between resection completeness and bleeding risk [7, 21].

Fig. 3.

Preoperative multiplanar CT imaging of the calcification-dominant phenotype. Computed tomography demonstrating extensive circumferential pericardial calcification surrounding the cardiac chambers in a patient who subsequently required cardiopulmonary bypass during pericardiectomy because of an intraoperative descending aortic injury

Our study has several limitations that should be acknowledged. First, the single-center, retrospective design inherently limits the generalizability of our results. Second, given the sample size of 36 patients, the statistical power of this study is inherently limited. Rather than offering definitive prognostic assertions, our observations are intended to serve as exploratory, hypothesis-generating data regarding the potential influence of baseline inflammatory status on post-pericardiectomy recovery. Several continuous variables showed non-normal or skewed distributions and were therefore analysed using rank-based methods, which do not require normally distributed observations. The greater limitation is the small sample size, which reduces precision and power to detect modest differences. In addition, cardiac magnetic resonance imaging was not routinely available during the earlier years of the study period, precluding systematic assessment of inflammatory activity and limiting integration of advanced imaging-based phenotyping recommended in recent international position statements. Etiological subgroup analyses were also underpowered because of the small number of patients within individual categories, which may explain the absence of statistically significant survival differences despite apparent trends. The cohort spans aetiologies that differ substantially in pathophysiology and prognosis, from idiopathic and tuberculous disease to rare causes such as Erdheim-Chester disease, radiation injury and post-transplant constriction; pooling them limits what can be inferred about biology, and the individual subgroups are far too small to analyse separately.

Cardiac magnetic resonance (CMR) imaging with late gadolinium enhancement or T2-weighted edema sequences was not systematically available across the 17-year study period, precluding direct validation of tissue-level inflammation against CMR imaging gold standards. Thus, our CRP- and effusion-based classification represents a pragmatic, study-defined surrogate rather than a CMR-validated disease stage.

Although we added exploratory two-variable logistic regression models, the occurrence of only 13 LCOS events precluded comprehensive multivariable adjustment. Each model included only one covariate in addition to the study-defined inflammatory phenotype, and confidence intervals remained wide. Residual confounding therefore remains likely, and the observed association should not be interpreted as evidence that the inflammatory phenotype independently predicts LCOS.

The association between the phenotype and low cardiac output syndrome could not be separated from age, which was independently associated with the outcome and which we could not adjust for alongside other covariates given only 13 events.

Finally, the single-centre retrospective design limits generalisability, the phenotype framework was applied retrospectively, and no systematic trial of anti-inflammatory therapy preceded surgery in this cohort, so we cannot exclude that some patients in the inflammatory-phenotype group had potentially reversible constriction and were operated within a window in which medical therapy might have succeeded.

Despite these limitations, our findings support further investigation of a phenotype-oriented approach to CP that integrates inflammatory status, imaging morphology, and individualized surgical planning.

Conclusion

In this contemporary 17-year single-center series of 36 patients undergoing pericardiectomy for constrictive pericarditis, idiopathic disease was the most frequent etiology, while tuberculosis accounted for a lower proportion of cases than reported in earlier Turkish surgical series. Effusion-positive and calcification-positive disease showed contrasting preoperative C-reactive protein profiles, suggesting that these imaging-defined groups may represent different points along an inflammatory-to-fibrotic disease spectrum. A study-defined preoperative inflammatory phenotype, defined by pericardial effusion and CRP > 10 mg/L, was associated with hypoalbuminemia and a higher burden of adverse perioperative outcomes, suggesting it may serve as a simple, readily available marker for preoperative risk stratification. Long-term survival after pericardiectomy was generally favorable in the overall cohort, although etiology-specific comparisons were limited by small subgroup sizes. Given the retrospective design, limited sample size, and low number of events, these findings should be considered hypothesis-generating and warrant confirmation in larger multicenter cohorts.

Abbreviations

ACC

American College of Cardiology

AKI

Acute kidney injury

CABG

Coronary artery bypass grafting

CMR

Cardiac magnetic resonance

CP

Constrictive pericarditis

CPB

Cardiopulmonary bypass

CRP

C-reactive protein

CT

Computed tomography

CVP

Central venous pressure

ECD

Erdheim–Chester disease

ECMO

Extracorporeal membrane oxygenation

ESC

European Society of Cardiology

FMF

Familial Mediterranean fever

IABP

Intra-aortic balloon pump

ICU

Intensive care unit

IQR

Interquartile range

KDIGO

Kidney Disease: Improving Global Outcomes

LCOS

Low cardiac output syndrome

LVEF

Left ventricular ejection fraction

LVEDP

Left ventricular end-diastolic pressure

MR

Mitral regurgitation

MVR

Mitral valve replacement

NSAIDs

Nonsteroidal anti-inflammatory drugs

NYHA

New York Heart Association

PASP

Pulmonary artery systolic pressure

POEMS

Polyneuropathy, organomegaly, endocrinopathy, monoclonal protein, skin changes

RVEDP

Right ventricular end-diastolic pressure

RVSP

Right ventricular systolic pressure

SD

Standard deviation

TRAPS

Tumor Necrosis Factor Receptor-Associated Periodic Syndrome

Authors’ contributions

CTU and LY conceived and designed the study. CTU, BA, LY, AOY, OFS, EA and AT collected the clinical data. LY and CTU performed the statistical analysis. CTU and LY drafted the manuscript and prepared the tables and figures. BA, OOB, DG, and SNO critically revised the manuscript. SNO supervised the study. All authors read and approved the final manuscript.

Funding

None.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Prior to the initiation of the study, ethical approval (approval number: 679379) was obtained from Clinical Research Ethics Committee of Istanbul University-Cerrahpasa, Cerrahpasa Faculty of Medicine on 8 May 2023, ensuring compliance with the principles outlined in the Declaration of Helsinki. Informed consent was obtained from all individual participants included in the study.

Consent for publication

Informed consent was obtained from all individual participants included in the study.

Competing interests

The authors declare no competing interests.