Altered compositions of monocyte, T lymphocyte and NK cell subsets in heart failure of adult congenital heart disease

Abstract

Background

Adults with congenital heart disease (ACHD) display changes in adaptive immunity related to early open-heart surgery and subsequent incidental thymectomy. In acquired heart failure (HF) systemic inflammation, innate and adaptive immune cells play an important role. However, cellular immune alterations of monocyte, T lymphocyte and natural killer (NK) cell subsets have not been related to HF in ACHD.

Methods and results

This cross-sectional study included 209 ACHD outpatients (mean age: 35.3 ± 11.0 years; NYHA class I/II/III-IV 64.1%/21.5%/14.4%; 59.8% male) and 21 healthy controls (29.8 ± 12.6 years; 47.6% male). Patients with clinical signs of infection, inflammatory diseases or malignancies were excluded. Flow cytometry of fresh whole blood revealed significantly elevated levels of CD14⁺⁺CD16⁺ (Mon2) and CD14⁺CD16⁺⁺ (Mon3) monocytes that increased in advanced NYHA class. Immature CD14⁺HLA-DR^neg/low monocytes were significantly higher in ACHD, but did not relate to NT-proBNP or NYHA class. Frequencies of Mon2, CD14⁺HLA-DR^neg/low and neutrophils related to plasma S100A8/A9 levels. Furthermore, ACHD had decreased T helper and cytotoxic T cell counts, however NKT and CD3^bright T cells remained unchanged compared to control. The cytokine-producing CD56^bright NK subset related to NT-proBNP and expanded in advanced HF stages. Patients with elevated Mon2 and CD14⁺HLA-DR^neg/low monocytes exhibited a worse outcome regarding all-cause mortality/cardiac decompensations after a mean follow-up of 5.5 years. High Mon2, low total lymphocyte and low T helper cell counts were independent prognostic markers for the onset of adverse cardiac events.

Conclusions

Altered immune cell compositions related to heart failure and exhibited prognostic relevance. This suggests an important link between immunity, inflammation and HF in ACHD bearing potential for refined clinical risk stratification.

Graphical abstract

Graphical abstract: changes in lymphocyte and monocyte subsets are present in adults with congenital heart disease (ACHD) relating to heart failure. NLR, neutrophil-to-lymphocyte ratio.

1. Introduction

Leading causes of premature death in adult patients with congenital heart disease (ACHD) are heart failure (HF), infections such as pneumonia and endocarditis as well as sudden cardiac death [1]. Lifesaving early open-heart surgery substantially improved the survival of children born with complex congenital heart disease into adulthood and is often required depending on the heterogeneous malformations. However, to enable better access to the operation field, thymectomy is routinely performed in some patients, but known to cause changes in adaptive immune cell composition, such as reduced total T cell counts and increased immune senescence [2].

Inflammation and immune activation are well-described mediators of disease progression in acquired HF [3]. Innate and adaptive immune mediators, such as monocytes, T lymphocytes and natural killer (NK) cells may play a pivotal role for pathophysiological mechanisms and progression of HF [[4], [5], [6], [7], [8]]. Elevated circulating inflammatory markers, such as tumour necrosis factor (TNF)-α, interleukin (IL)-6 and C-reactive protein (CRP) already gained attention in ACHD [9,10]. However, knowledge about monocytes, T lymphocytes, NK cells and their subpopulations in HF is still lacking regarding the heterogeneous ACHD collective.

In disease contexts monocytes exhibit both, harmful and beneficial functions through cytokine production, cell recruitment, angiogenesis and extracellular matrix remodelling [4]. Three human mature blood monocyte subsets were described. The classical (Mon1) phenotype consists of CD14⁺⁺CD16⁻ monocytes accounting for 80–90% of peripheral blood monocytes. CD14⁺⁺CD16⁺ (Mon2, intermediate) and CD14⁺CD16⁺⁺ (Mon3, non-classical) monocytes were associated with inflammatory diseases [11,12]. Regarding heart disease, Mon2 counts correlated with HF characteristics in different types of acquired HF [[13], [14], [15]]. Within inflammatory responses Mon1 and Mon2 bind to monocyte chemoattractant protein-1 initiating tissue invasion to continue their local inflammatory actions. Mon3 cells are known to release IL-1β and TNF-α upon activation and are considered to have an endothelial patrolling function, as these cells express high levels of the adhesion-related fractalkine receptor (CX₃CR₁) [16].

Immature CD14⁺HLA-DR^neg/low monocytes gained much interest in pathomechanisms of severe COVID-19, cancer, inflammatory and autoimmune diseases and have been described to expand in the peripheral blood of patients with chronic HF and correlated positively with disease severity and high levels of cytokines [17]. In respective HF animal models a protective role of this subset expansion was suggested [17]. In acute myocardial infarction, CD14⁺HLA-DR^neg/low cells related to myocardial damage and dysfunction [18,19]. Neither mature nor immature monocyte subpopulations have been investigated in ACHD.

Classified as innate lymphoid cells, natural killer (NK) cells represent an interface between the adaptive and innate immune system. NK cells exhibit complex tissue-dependent functions. They can stimulate other effectors, such as dendritic cells, macrophages and T cells to mediate paracrine cytoprotective effects or contribute to chronic inflammation and progression of pathological cardiac remodelling [20]. Mainly two peripheral blood NK subpopulations can be classified in humans: accounting for 90% of circulating NK cells, CD56^dim NK cells have high cytotoxic properties and exhibit rather low cytokine production. Unlike CD56^bright NK cells, which are potent producers of cytokines, such as interferon-γ, TNF and granulocyte–macrophage colony-stimulating factor [21]. Although several studies investigated NK cells in ACHD none related this cell type to parameters of HF [2,22,23].

T cells are central regulators of immune processes in both, acute and chronic cardiac injury and subsequent remodelling [20]. Altered lymphocyte subset composition including reduced total CD4⁺ and CD8⁺ T cells and pronounced immune senescence were reported extensively in ACHD following infant thymectomy or in protein-losing enteropathy of Fontan patients [2,[24], [25], [26]]. However, associations of T cell frequencies and HF grade in ACHD remain unclear.

Using flow cytometry, we described monocyte, T lymphocyte and NK cell subsets in relation to HF functional capacity, NT-proBNP, CRP and S100A8/A9, a heterodimer known for its critical role in modulating inflammatory responses. Parameters of the immune phenotyping as well as routine blood biomarkers were related to clinical outcomes using combined primary and secondary endpoints. This observational investigation aimed to shed light into adaptive and innate immune cell compositions in HF of ACHD, which can give rise to further mechanistic research. In addition, respective insights could be of mayor clinical interest to estimate the risk for infections, auto-immunity, long-term consequences of thymectomy as well as for investigating chronic inflammation in the context of HF pathogenesis and the onset of associated comorbidities.

2. Methods

2.1. Cross-sectional prospective pilot study

A total of 209 consecutive patients with congenital heart disease were recruited from the ACHD outpatient clinic at Hannover Medical School between December 2013 and January 2016. In parallel, 21 healthy controls were recruited. Exclusion criteria were age <18 years, pregnancy, cancer, any clinical sign of ongoing infection as well as relevant chronic infectious or inflammatory disease to avoid potential confounders in terms of immune activation and function. All patients were examined by the same certified ACHD cardiologist who classified patients to heart failure symptoms (New York Heart Association [NYHA] class), anatomical subgroups and remained blinded to immunophenotyping analysis. Due to flow cytometry capacity reasons not more than three patients were included per day to ensure immediate analysis. Full patient's medical history and routine blood tests were available. Routine echocardiography data were obtained but not included to the final analysis because of the heterogeneity and inconsistently comparable anatomy. The study was performed in accordance with the declaration of Helsinki and was approved by the local ethic committee of Hannover Medical School. All patients provided written informed consent.

2.2. Patient population and classification

Five patients were diagnosed with DiGeorge syndrome, four had congenital heart disease associated with Down syndrome. NYHA class was chosen for graduation of HF symptoms. Approximately 64% of the enrolled patients presented with asymptomatic HF according to NYHA class I, whereas 22% were classified to NYHA II. As only two patients with NYHA class IV HF complied with study criteria, we decided to pool NYHA class III and IV patients together resulting in a NYHA III/IV collective accounting for 14% of all surveyed patients. At the time of blood sampling none of the patients were hospitalised and none of them were receiving i.v. inotropes. Median-sternotomy was performed at least once in 158 patients (75.6%). In order to estimate the impact of early median sternotomy with subsequent removal of thymus tissue, we analysed the patient records regarding timepoint and surgical access and obtained these data completely in 159 patients (76,1%). Information about the amount of removed thymus tissue were not available. The group that underwent median sternotomy at the age of one year or younger (≤1Tx, n = 46) was compared to those having median sternotomy later in life or never (no ≤ 1Tx, n = 113) and group characteristics were summarised in table S5. In case the first surgery was not associated with median sternotomy, but another at a later date was, the later timepoint was used for classification (≤1 or >1 year old). Surgical interventions following the initial/first median sternotomy were not further considered. To investigate possible differences between the heterogeneous cardiovascular malformations, we generated five anatomical subgroups including minor valvular lesions and corrected simple shunts, left heart disease including coarctation, right heart disease including pulmonary artery lesions, native single ventricles with Eisenmenger syndrome and Fontan circulation. We defined a group with mild cardiac impairment including repaired shunts without pulmonary arterial hypertension and mild congenital valvular lesions. Left heart disease summarises all subaortic ventricles irrespective of right or left ventricular morphology, together with moderate to severe left-sided valvular lesions and the presence of coarctation. Right heart disease includes entities with valvular lesions and ventricular dysfunction in the sub-pulmonary position and pulmonary artery anomalies. Native single ventricles, either unpalliated or palliated with aorto-pulmonary or modified Blalock Taussig shunts, and patients with Eisenmenger Syndrome were grouped together. As a common feature, both entities present with pulmonary arterial hypertension and cyanosis.

2.3. Outcome analysis

The primary outcome of our analysis was the combination of all-cause mortality and cardiac decompensation, which was defined as hospital admission for heart failure decompensation. The combined secondary outcomes of interest were: cardiac decompensation (defined by hospitalisation), death, arrhythmia (leading to either hospitalisation, cardioversion, implantation of a pacemaker/defibrillator, or initiation/change of anti-arrhythmic therapy), thromboembolism (myocardial infarction, ischaemic cerebrovascular events, or pulmonary embolism), onset of endocarditis (diagnosed by transoesophageal echocardiography or PET-CT with/or without pathogen detection in blood cultures).

2.4. Blood sampling and routine analysis

Peripheral venous blood samples were drawn by venepuncture after the patients and controls had spent 25 min at rest in a quiet room. Routine blood measurements were performed by the hospital's laboratory facilities. An automated Sysmex XT-2000i haematology analyser determined complete blood cell counts. EDTA-anticoagulated blood for flow cytometry analysis was left at room temperature and processed within 60min following collection.

2.5. Flow cytometry and blood cell counts

For multiparameter flow cytometry mouse anti-human monoclonal fluorochrome conjugated antibodies were mixed with 100 μL of fresh whole blood and incubated at room temperature in the dark for 20min. The following antibodies were used for monocyte subsets: anti-CD14 (Clone MφP9, 1:50, BD Biosciences), anti-CD16 (Clone CB16, 1:20, eBioscience), anti-HLA-DR (Clone L243, 1:66.6, BioLegend), anti-CCR2 (Clone K036C2, 1:50, BioLegend) and anti-CX3CR1 (Clone 2A9-1, 1:100, BioLegend). For verification of the CD14⁺HLA-DR^neg/low monocyte gating strategy anti-CD33 (Clone WM53, 1:50, BioLegend) and anti-CD15 (Clone HI98, 1:50, BioLegend) were used in a subgroup of patients. To stain lymphocyte subpopulations mouse anti-human monoclonal fluorochrome conjugated anti-CD56 (Clone MEM-188, 1:20, BioLegend), anti-CD8 (Clone SK1, 1:20, BioLegend), anti-CD4 (Clone RPA-T4, 1:20, BD Biosciences), anti-CD3 (Clone SK7, 1:20, BD Biosciences) and anti-CD16 (Clone CB16, 1:20, eBioscience) were used. Red blood cells were lysed by 2 mL Veralyse Lysing Solution® (Beckman Coulter) for 10min followed by washing twice with 4 mL Hanks' Balanced Salt Solution (HBSS) and immediate flow cytometry. After a pre-selection using side scatter (SSC) and forward scatter (FSC) to exclude cell debris, single cells were gated in a FSC vs. Time-of-Flight (TOF) dot plot to eliminate doublets. In order to distinguish positive from negative staining during gating analysis, fluorescence minus one controls were analysed. Flow cytometry was performed using a GalliosTM flow cytometer and corresponding software (Beckman Coulter). All experiments were performed and analysed blinded to patient's clinical data. Gating was performed as demonstrated below for monocyte subsets, CD14⁺HLA-DR^neg/low and for T and NK cells (Supp. Fig. S1). Percentages of gated subsets were multiplied by total monocyte or lymphocyte counts obtained from differential blood counts to quantify absolute numbers (cells/μL). Lymphocyte subsets were measured in a subgroup of 160 ACHD patients and in 20 healthy controls from 09/2014 to 01/2016.

2.6. S100A8/A9 plasma levels

To investigate S100A8/A9 (MRP8/14) plasma levels a sandwich ELISA was used (BioLegend, 439707) according to the manufacturer's instructions. Due to a limited number of plasma aliquots, S100A8/A9 was only measured in 97 ACHD patients and in all 21 controls.

2.7. Statistical analysis

Data are expressed as mean ± SD, SEM or as median [interquartile range] as indicated respectively. Equality of variances was assessed by Levene's test. Shapiro-Wilk test was used to assess normal distribution of the data. Normally distributed data were analysed by Student's 2-tailed t-test or the multiple comparison Tukey's HSD test as appropriate. Non-normally distributed variables were compared using Mann-Whitney U test for two and Kruskal-Wallis test for more than two groups. Spearman's correlation coefficient was calculated for comparisons of data including non-normally distributed variables. Subgroup analysis for different cardiac malformations was not performed for lymphocyte immunophenotyping results because of insufficient subgroup sizes. Survival analysis used the comparison between the top (Mon2, CD14⁺HLA-DR^neg/low monocyte) or bottom (lymphocyte, T helper cell) tertile and the respective other bottom/top two tertiles combined. Using the log-rank test, we performed univariate survival analysis. Time-to-event was defined from the date of blood sampling to the date of the first clinical event. Multivariable survival analysis was performed for every cell subset of interest (Mon2, CD14⁺HLA-DR^neg/low, total lymphocytes, T helper cells) individually using Cox regression, with backward selection including all immune phenotyping and routine blood test derived variables associated with log of p < 0.05 in univariate analysis and one respective leukocyte subset as the basis for the multivariable model. In case of a correlation of two biologically linked parameters with r > 0.4 (e.g. creatinine and urea), only the stronger one from the univariate analysis was included. Adjusting for age and gender had no influence on the multivariable analysis, except for T helper cells. Therefore, a second model was build including all clinical parameters such as age, gender, NYHA class and comorbidities as well as biomarkers. Data sets with single missing values were not entirely excluded and imputation was not performed. Values of p < 0.05 were considered statistically significant. Statistical analysis and graphical representation were performed using SPSS 23.0 and GraphPad Prism 9 software.

3. Results

Baseline characteristics of ACHD compared to controls are shown in Table 1. Gamma-glutamyltransferase and NT-proBNP plasma concentrations were significantly higher in patients. Composition of cardiovascular malformations and medical treatment details are summarised in the supplemental material (Table S1) such as the characterisation of HF groups (Table S2) demonstrating an elevated NT-proBNP and a worsening of renal function with rising NYHA class. Complete blood counts displayed significantly higher peripheral blood neutrophil and monocyte, but similar lymphocyte counts in patients compared to healthy controls. Between different stages of HF there were no significant distinctions regarding circulating counts of neutrophils, monocytes or lymphocytes. In addition, no differences were observed in eosinophil and basophil cell counts between all groups (Table 2). Platelet counts were significantly decreased in NYHA III/IV, whilst haemoglobin levels were increased (Table 2) as some of the patients had polycythaemia driven by hypoxia resulting in a worse functional capacity. Neither arterial hypertension, nor nicotine consumption represented statistically relevant inflammation-associated confounders of immune phenotypes in a subgroup analysis, except for higher Mon1 levels in smoking compared to abstinent ACHD patients (Fig. S6).

Table 1.

Group characteristics.

	Healthy controls (n = 21)	ACHD (n = 209)
Age, years	29.81 ± 12.6	35.34 ± 11.13 §
Male, (%)	10 (47.6)	125 (59.8)
BMI (kg/m2)	–	25.15 ± 0.30
NYHA class (I/II/III + IV)	–	134/45/30
Cardiovascular risk factors/comorbidities
Art. hypertension, (%)	2 (9.5)	47 (22.5)
Smoker, (%)	3 (14.3)	34 (16.3)
Diabetes, (%)	–	3 (1.4)
CRP (mg/L)	1.25 ± 1.78	2.99 ± 3.9
NT-proBNP (ng/L)	38.44 ± 36.39	371.30 ± 773.6 *
Renal/liver function
Creatinine (μmol/L)	80.62 ± 14.02	85.24 ± 43.37
Urea (mmol/L)	4.45 ± 1.37	5.15 ± 1.88
Albumin (g/L)	42.62 ± 5.59	42.17 ± 4.48
GGT (IU/L)	18.69 ± 13.56	41.60 ± 44.38 §
AST (IU/L)	28.92 ± 19.57	28.97 ± 9.25

Values are expressed as absolute numbers or mean ± SD; Statistical analysis performed using t-test or Mann-Whitney U test, §p < 0.05, *p < 0.001. ACHD, adult congenital heart disease; BMI, body mass index; CRP, C-reactive protein; GGT, gamma-glutamyltransferase; NYHA, New York Heart Association; TSH, thyroid stimulating hormone.

Table 2.

Blood counts according to heart failure functional capacity.

	Healthy controls (n = 21)	ACHD (n = 209)	NYHA I (n = 134)	NYHA II (n = 45)	NYHA III/IV (n = 30)
Haemoglobin (g/dL)	13.77 ± 1.87	15.19 ± 2.75	15.02 ± 2.43	14.85 ± 2.42	16.62 ± 4.66 § ‡
Platelets (103/μL)	218.85 ± 79.55	207 ± 60.43	215.75 ± 55.91	211.11 ± 65.2	162.70 ± 54.77 † ¶
Neutrophils (103/μL)	3.07 ± 1.05	4.23 ± 1.45 *	4.06 ± 1.39 §	4.63 ± 1.54 *	4.34 ± 1.53 §
Lymphocytes (103/μL)	1.73 ± 0.43	1.73 ± 0.58	1.78 ± 0.58	1.77 ± 0.53	1.49 ± 0.66
Monocytes (103/μL)	0.54 ± 0.14	0.63 ± 0.15 §	0.62 ± 0.23	0.64 ± 0.2	0.68 ± 0.16 §
Eosinophils (103/μL)	0.19 ± 0.23	0.18 ± 0.29	0.18 ± 0.23	0.19 ± 0.2	0.14 ± 0.11
Basophils (103/μL)	0.03 ± 0.018	0.04 ± 0.02	0.04 ± 0.02	0.04 ± 0.027	0.05 ± 0.033

Values are expressed as mean ± SD; Statistical analysis comparing controls vs. total ACHD was performed using Student's t-test, differences between the NYHA groups and controls were analysed using ANOVA with post hoc Tukey's test: §p < 0.05 vs. control, *p < 0.001 vs. control, ‡p < 0.05 vs. NYHA I, †p < 0.001 vs. NYHA I, ¶p < 0.05 vs. NYHA II; NYHA, New York Heart Association.

3.1. Markers of systemic inflammation in ACHD heart failure

The commonly used high sensitivity C-reactive protein (CRP), as a circulating marker of systemic inflammation, was not higher in the total ACHD cohort compared to controls (Fig. S2 A). However, CRP levels were significantly increased in severe stages of HF compared to asymptomatic ACHD patients (Fig. S2 B). The highest value of CRP, 25.3 mg/L, was measured in an individual with severe HF due to a single right ventricle anatomy. Patient's CRP values correlated robustly with NT-proBNP (spearman r = 0.32, p < 0.001). The neutrophil-to-lymphocyte ratio (NLR) represents an easy to determine marker of systemic inflammation and was significantly higher in ACHD patients. Furthermore, NLR increased significantly with worsening NYHA class (Fig. S2C-D). Unlike CRP, NLR was already significantly increased in asymptomatic NYHA I patients compared to controls.

3.2. Monocyte subsets

Composition of circulating blood monocyte subsets differed significantly between ACHD and controls (Fig. 1, Table S3). Absolute CD16 expressing Mon2 and Mon3 counts were both significantly elevated when compared to controls reflecting a shift towards a possibly further pro-inflammatory state in ACHD (Fig. 1 F–H, Table S3). Expanding CD16⁺ subsets lead to a diminished proportion of Mon1 when comparing ACHD vs control or NYHA III/IV vs. NYHA I patients, but absolute Mon1 counts remained similar (Table S3). Advanced stages of HF (NYHA III/IV) were associated with significantly elevated Mon2, Mon3 counts compared to asymptomatic NYHA class I patients or controls (Fig. 1J–L). Absolute counts of all three monocyte subsets related to CRP levels. Mon2 and Mon3 showed a weak, but statistically significant correlation with NT-proBNP (Fig. 1 N). Subgroup analysis of different haemodynamic defects showed no significant difference between Mon2 or other monocyte subsets among the ACHD subgroups (Fig. 1 M, Table 3). However, patients with native single ventricle physiology or severe pulmonary hypertension leading to Eisenmenger Syndrome exhibited the highest mean Mon2 levels (Fig. 1 M). This subgroup, next to left heart disease/coarctation and right heart disease/pulmonary artery lesions, showed significantly higher Mon2 counts compared to controls, but these results should be interpreted with caution as no adjustment for NYHA class was performed. The subgroup analysis is limited by small group sizes.

Fig. 1.

Increased circulating levels of CD16⁺monocytes in heart failure of ACHD (A–E) Gating strategy for monocyte subsets: Mon1 (classical, HLA-DR⁺CD14⁺⁺CD16⁻), Mon2 (intermediate, HLA-DR⁺CD14⁺⁺CD16⁺), Mon3 (non-classical, HLA-DR⁺CD14⁺CD16⁺⁺). Gating was verified by back-gating to CCR2 (D) and CX3CR1 (E) expression. (F–H) Representative flow cytometry density plots. (J–L) Frequency of monocyte subsets in controls vs. ACHD according to New York Heart Association (NYHA) class. (M) Circulating Mon2 in subgroups of ACHD malformations, coarctation (CoA), pulmonary artery lesions (PA), Eisenmenger syndrome and pulmonary hypertension (PH), n = 21/26/79/76/15/13. (J–M) n = 21/134/45/30. Values expressed as mean ± SEM. Statistical analysis performed using ANOVA with post hoc Tukey's test: §p < 0.05 vs. control, *p < 0.001 vs. control, †p < 0.001 vs. NYHA I. (N) Spearman-correlation matrix of Mon1–3 (cells/μL), CRP (mg/L) and NT-proBNP (ng/L), correlation coefficient is expressed by colour intensity and numeric R values are shown in the top right half, p value indicated as *p < 0.05, **p < 0.01, ***p < 0.001.

Table 3.

Morphologic subgroup differences.

		Healthy controls (n = 21)	Minor valvular lesion/ corrected shunts (n = 26)	Left heart disease, CoA (n = 79)	Right heart disease, PA (n = 76)	Native single ventricles, PHT (n = 15)	Fontan (n = 13)
Age, years		29.81 ± 12.6	35.9 ± 11.2	35.3 ± 10.2	35.0 ± 13.0	39.4 ± 11.4	29.2 ± 5.4
Male, (%)		10 (47.6)	14 (53.8)	57 (72.2)	39 (51.4)	6 (40%)	9 (69.2)
NYHA class I/II/III + IV		/	24/2/0	51/22/6	54/9/13	0/6/9	5/6/2
Mon1	(10/μL) (%)	45.6 ± 11.7	49.4 ± 19.5	54.0 ± 15.9	48.0 ± 15.3	51.3 ± 15.3	49.4 ± 15.6
		84.1 ± 5.3	80.7 ± 5.6	81.3 ± 5.2	80.0 ± 5.7 *	77.9 ± 5.7 *	80.9 ± 6.5
Mon2	(10/μL) (%)	2.4 ± 1.5	4.2 ± 1.9	4.4 ± 2.5 *	4.2 ± 2.4 *	6.0 ± 3.3°	3.8 ± 1.2
		4.6 ± 3.0	6.9 ± 2.9	6.5 ± 3.2	6.8 ± 2.9 *	8.6 ± 3.3°	6.8 ± 2.7
Mon3	(10/μL) (%)	3.1 ± 1.9	4.1 ± 2.1	4.3 ± 2.2	4.5 ± 3.0	5.1 ± 3.2	3.9 ± 2.0
		5.6 ± 3.1	6.6 ± 2.9	6.6 ± 3.2	7.3 ± 3.2	7.9 ± 3.7	7.1 ± 3.7
NLR		1.67 ± 0.37	2.47 ± 1.36	2.56 ± 1.14	2.74 ± 1.32	3.31 ± 1.92 *	3.58 ± 1.4°
Lymphocytes (103/μL)		1.73 ± 0.4	1.8 ± 0.54	1.9 ± 0.6	1.63 ± 0.6	1.68 ± 0.65	1.2 ± 0.43
CRP (mg/L)		1.25 ± 2.8	2.0 ± 1.4	3.0 ± 3.8	2.8 ± 3.7	6.0 ± 5.0 *	2.6 ± 2.1
NT-proBNP (ng/L)		38.4 ± 33	85 ± 87	284 ± 382	398 ± 918	1113 ± 1450°	577 ± 872
Creatinine (μmol/L)		80.6 ± 11.1	75.5 ± 12.3	81 ± 14.8	92.3 ± 70.6	92.6 ± 29.4	80.6 ± 11.0
Albumin (g/L)		42.6 ± 4.5	41.5 ± 3.2	43.1 ± 3.9	42.6 ± 3.9	37.3 ± 6.4 *	41.3 ± 6.9
GGT (IU/L)		18.9 ± 29.7	32.6 ± 37.8	32.8 ± 35.1	43.6 ± 36.2	32.3 ± 17.8	116.8 ± 107.5°

Values are expressed as mean ± SD; Statistical analysis performed using ANOVA with post hoc Tukey's test or Kruskal-Wallis (K–W) test as appropriate: *p < 0.05 vs. control; °p < 0.01 vs. control. CoA, coarctation; CRP, C-reactive protein; GGT, gamma glutamyltransferase; Mon1, CD14++CD16⁻; Mon2, CD14++CD16⁺; Mon3, CD14⁺CD16++; NLR, neutrophil-to-lymphocyte ratio; PA, pulmonary artery lesions; PHT, Eisenmenger syndrome and pulmonary hypertension.

3.2.1. Immature CD14⁺HLA-DR^neg/low monocytes

Circulating immature CD14⁺HLA-DR^neg/low absolute counts and percentage from total monocytes were significantly expanded in ACHD compared to controls (Fig. 2), which was consistent in all subgroups of HF severity compared to control. We observed a cluster of some individual patients with a substantially expanded cell population. The highest level among all groups was observed in a 42 years old, male NYHA class IV patient with single right ventricle causing severe hypoxia. Between the NYHA groups there was no significant difference in CD14⁺HLA-DR^neg/low frequencies (Fig. 2 H + L). Expansion of the immature CD14⁺HLA-DR^neg/low subset in ACHD was not associated with a significantly reduced expression of HLA-DR on total monocytes, as shown by the mean fluorescence intensity (Fig. S4 A). CD14⁺HLA-DR^neg/low absolute and relative frequencies correlated significantly with CRP, but not with NT-proBNP (Fig. 2 M, Fig. S4 B).

Fig. 2.

Expansion of a CD14⁺HLA-DR^neg/lowmonocyte subset in ACHD patients. (A–F) Gating strategy to quantify CD14⁺HLA-DR^neg/low monocytes. (E–F) Gating was verified in a subgroup by CD33 expression of CD14⁺ monocytes. (B–C) Representative flow cytometry density plots of a control and adult congenital heart disease (ACHD) patient classified to New York Heart Association (NYHA) class III. (J + M) Linear regression analysis between CD14⁺HLA-DR^neg/low counts and (J) S100A8/A9 plasma or (M) serum CRP levels in ACHD patients. (G-H, K-L) CD14⁺HLA-DR^neg/low in control vs. ACHD and according to patient's NYHA class. Percentages indicate HLA-DR^neg/low cells within CD14⁺ monocytes. Absolute counts were calculated by multiplying percentages from gated monocytes with monocyte frequencies from blood cell counts. n = 21/209 (G, K) or n = n = 21/134/45/30 (H,L). Values expressed as median (IQR). Statistical analysis performed using Mann-Whitney U or Kruskal-Wallis test: §p < 0.05 vs. control, *p < 0.001 vs. control.

3.3. Lymphocyte subsets

3.3.1. T cells

ACHD patients displayed significantly lower circulating CD8⁺ cytotoxic T cells as well as CD4⁺ T helper cells compared to control. NYHA III/IV patients had significantly less absolute cytotoxic and T helper cells compared to control, but no differences were observed between different stages of HF (Fig. 3). NKT cell and CD3⁺⁺CD56⁺ counts were similar between all groups. The proportions of all studied T cell subsets within lymphocytes did not reveal any relevant differences (Table S4).

Fig. 3.

ACHD patients exhibited decreased T cell subset counts next to increased frequencies of CD56^brightNK cells. (A–D) Frequency of T and NK subpopulation cell counts in controls vs. ACHD, n = 20/160. (E–H) Absolute CD8⁺ and CD4⁺ T cell and NK subset counts in controls vs. ACHD according to New York Heart Association (NYHA) heart failure class, n = 20/101/40/19. Values are expressed as median [interquartile range]. Statistical analysis performed using Mann-Whitney U or Kruskal-Wallis test: §p < 0.05 vs. control. (J) Linear regression analysis between CD8⁺ T cell counts and Mon3 frequency in ACHD patients.

3.3.2. Natural killer cells

Counts of total NK cells and CD56^dim NK subsets did not differ significantly between ACHD and controls and were not affected by HF functional capacity (Table S4). In contrast, the cytokine producing CD56^bright NK subset was significantly elevated in all ACHD and in NYHA II patients compared to control (Fig. 3 D,H, Table S4). The proportion of this subset was increased in advanced stages of HF (NYHA III/VI) compared to asymptomatic patients (Table S4). In addition, Spearman correlation analysis revealed a significant, but mild correlation between the CD56^bright NK proportion and NT-proBNP levels (correlation coefficient r = 0.143, p = 0.034), potentially indicating an indirect association with HF severity.

3.4. Relations between S100A8/A9, T lymphocyte, NK cell and monocyte subsets

Upon activation, neutrophils and, to a smaller extent, monocytes/macrophages are known to secrete high amounts of the heterodimer S100A8/A9. However, plasma levels did not differ significantly between the analysed groups, although some individuals exhibited markedly high concentrations (Fig. S3). In line with the biological function, a strong correlation was found between S100A8/A9 and neutrophils (r = 0.489, p = 3.85e⁻⁸), such as less robustly between S100A8/A9 and Mon2 (r = 0.201, p = 0.019) and CD14⁺HLA-DR^neg/low (Fig. 2 J). In addition, neutrophil and Mon2 counts were significantly related (r = 0.389, p = 4.38e⁻⁹). Mon3 absolute counts were significantly associated with the CD56^bright NK subset counts (r = 0.206, p = 0.004). No relevant correlation was found between the other monocyte or NK subpopulations. Low CD8⁺ cytotoxic T cell counts correlated negatively with an expanding Mon3 subset (Fig. 3 J).

3.5. Early/infant median sternotomy

Patients who underwent early/infant sternotomy (≤1 year old) including an assumed thymectomy were significantly younger at the time of study recruitment and had a lower reported LVEF compared to those receiving surgery later in life or never (Table S5). In addition, early sternotomised patients underwent significantly more surgical interventions between birth and date of study recruitment (Table S5). Absolute counts of monocyte, T cell and NK cell subsets seem not to be altered by early/infant median sternotomy including possible incidental thymectomy (Fig. S5, Table S5). In addition, neutrophil-to-lymphocyte ratio and CRP were unchanged (Table S5).

3.6. Outcome analysis

After a mean follow-up of 66.4 ± 24.68 months (5.5 years; max. 99 months/8.25 years), 23 (11.0%) patients experienced the primary combined endpoint, including 14 (6.7%) who died. The secondary composite endpoint was reached by 76 (36.4%) patients and detailed numbers about the different adverse cardiac events are described in table S6. There was no loss of follow-up. Kaplan-Meier plots show that patients with high (top tertile) Mon2 and CD14⁺HLA-DR^neg/low monocytes had a worse outcome, regarding all-cause mortality and hospitalisation for cardiac decompensation, compared to those with low (bottom two tertiles) counts (Fig. 4A–B). High Mon2 and patients with low (bottom tertile) total lymphocyte and low T helper cell counts experienced a worse outcome regarding the onset of adverse cardiac events and these changes represented statistically independent prognostic markers (Fig. 4 C-E, Table S8). However, this was not the case for the primary endpoint. Surprisingly, CRP levels were not related to clinical outcome in this small study (Table S7-S8). High Mon2 counts and low lymphocyte counts exhibited prognostic significance, however NT-proBNP seems to be a much better biomarker for ACHD outcome in our multivariate analysis (Table S7-S8).

Fig. 4.

Kaplan-Meier survival plots of the primary and secondary outcome in ACHD. (A–B) The top tertile of the Mon2 (intermediate, HLA-DR⁺CD14⁺⁺CD16⁺) and CD14⁺HLA-DR^neg/low monocyte subset war compared to the bottom two tertiles in all ACHD patients presented according to the primary outcome (all-cause mortality and hospitalisation for cardiac decompensation). (C–E) The top tertile of the Mon2 (C) and the bottom tertile of total circulating lymphocytes (D) and T helper cells (E) was compared to the respective two remaining tertiles. The cut-off ranges of the tertile groups are shown in the figure. Maximum follow-up period was 8.25 years. Log-rank test was performed, n = 209.

4. Discussion

This study is the first performing immunophenotyping of innate and adaptive immune cells with regard to parameters of heart failure (HF) in adult congenital heart disease (ACHD).

4.1. Systemic inflammation in ACHD

Accumulating evidence indicates an inflammatory state including elevated proinflammatory cytokines associated with HF in ACHD patients [9,10,27]. Increased low-grade CRP levels, neutrophil-to-lymphocyte ratio (NLR) and expansion of inflammation-associated monocyte and NK cell subsets underline the presence of systemic inflammation in patients with advanced HF related to ACHD. Although noticeable differences between individual patients were observed in our study, no significant differences between the morphological malformations were present. Though the small group sizes and variations of heart defects including NYHA class as potential confounder are limiting the interpretation. The multiple sources of local or systemic inflammation and their individual impact on immune effector cells are still not sufficiently explored in the context of ACHD. Possible causes and underlying mechanisms can be diverse: HF-induced innate immune activation by myocardial injury, congestion, hypoxia, comorbidities or immune deficiencies due to thymectomy and genetic syndromes aggravating pathogen elimination were discussed [2,9,[28], [29], [30]]. Thus, the induction of the inflammation-associated markers and leukocyte subsets such as Mon2, Mon3, CD14⁺HLA-DR^neg/low or CD56^bright NK cells may result from multifactorial sources. Regardless the causing mechanisms, CRP was already shown to be an independent predictor of mortality and non-elective cardiovascular hospitalisation in ACHD [10]. Neutrophil-to-lymphocyte ratio (NLR) has not been sufficiently investigated in ACHD, although it represents a promising and cheap biomarker to monitor systemic inflammation covering both, an innate immune activation as well as a decline in lymphocyte counts. This combination might be a reason for the increased NLR even in asymptomatic NYHA I patients, whereas CRP levels were still similar to healthy controls. Interestingly, NLR was significantly elevated in Fontan patients, but monocyte subsets were unchanged compared to control. An elevated NLR was already associated with protein-losing enteropathy in the Fontan physiology [24,31]. The small group size did not allow further statistical analysis. Targeting systemic immune activation by immunomodulatory therapy for well selected patients represents a promising future therapeutic strategy.

4.2. Monocyte subsets in heart failure of ACHD

Previous studies in ACHD revealed an association between HF symptoms and circulating pro-inflammatory cytokines such as TNF-α and IL-6 [9,10,27]. The Mon2 subset was found to produce high amounts of cytokines and reactive oxygen species and demonstrated an enhanced potential for the induction of angiogenesis, T cell proliferation and stimulation [12,32]. Even though we did not investigate monocytic cytokine production in ACHD, we assume that the expanding circulating Mon2 subset might contribute directly to the pronounced cytokine levels reported in these patients. In acquired heart disease the number of circulating Mon2 increased in several types of HF and related to NYHA class and impaired LV function [12,13,19,33]. In chronic HF of various aetiology Mon2 counts were independently associated with all-cause mortality [14]. Unlike Mon2, most of the referred studies did not observe changes in circulating Mon3 counts in the context of acute myocardial infarction or unstable angina [12,19]. However, as recently reported Mon3 migrate rapidly from the bloodstream into the coronary vasculature after acute myocardial injury [34]. The functional role of Mon3 in chronic HF remains less clear. Single reports are in line with our observations, describing expanding Mon3 in chronic systolic HF patients compared to control [35]. Whether increased circulating Mon3 might reflect enhanced or reduced transcoronary monocyte migration remains an open question in ACHD patients, such as the specific role of the recruited mature monocytes within the failing heart.

Also immature immune cells can be recruited to the injured myocardium and might promote proinflammatory mechanisms in cardiac healing and remodelling [19]. CD14⁺HLA-DR^neg/low monocytes have an immature phenotype and were already extensively studied in human cancer, sepsis, autoimmune disorders and severe COVID-19, where this population expanded, predicted outcome and exhibited significant immune suppressive functions [18,[36], [37], [38]]. Defined by these immunoregulatory properties CD14⁺HLA-DR^neg/low monocytes were also called monocytic myeloid derived suppressor cells (MDSC). In patients undergoing haemodialysis, circulating monocytic MDSC related to acute HF events and represented an independent predictor of mortality and stoke events [39]. Inside the failing human heart a population of HLA-DR^neg/low monocytes is present [40]. In acute myocardial infarction the increased CD14⁺HLA-DR^neg/low cells did not exhibit immunosuppressive properties, but frequencies correlated with myocardial injury and function [19]. We did not define the functional role of CD14⁺HLA-DR^neg/low in ACHD-HF, but detailed insights into immature monocyte functions will be crucial to understand inflammatory HF pathomechanisms in ACHD. The calcium-binding proteins S100A8/A9 are immunoregulatory alarmins produced by neutrophils, monocytes and MDSCs that are involved in granulopoiesis and mitochondrial dysfunction after myocardial injury [41,42]. Some individual ACHD patients presented markedly increased S100A8/A9 plasma concentrations, although no significant difference was found compared to controls. This might be explained by a sub-clustering of some patients with and others without S100A8/A9 excess and the rather small number of patients analysed. Identifying and further characterising individuals with a surge in plasma S100A8/A9 levels is necessary to fully understand our first observations in ACHD and to finally rate the role of the heterodimer in HF pathophysiology. In our study, circulating S100A8/A9 was associated with CD14⁺HLA-DR^neg/low and neutrophil counts, but not with any HF parameter or prognosis, though the protein is known to directly impair acute cardiac remodelling [43,44]. High Mon2 and CD14⁺HLA-DR^neg/low levels seem to be linked to adverse outcome in ACHD and might be added to other biomarkers, such as NT-proBNP or CRP for risk stratification and to identify patients with pronounced immune activation. Although, we did not perform comparisons of different predictions models and combinations of biomarkers.

4.3. T lymphocytes and natural killer cells in heart failure of ACHD

Although T cell subsets and postnatal thymic function have been extensively studied by immunologists using incidentally thymectomised ACHD patients, we are the first describing T cell counts in relation to HF functional capacity in ACHD. In parallel with our observations, reduced counts of circulating blood T helper and cytotoxic T cells have been found by several studies, predominantly driven by a decrease in naïve T cells contributing to pronounced immunosenescence [2,26]. Infant thymectomy during open-heart surgery seems to be a major cause for adaptive immune alterations, however HF itself could also be related to T cell deficiency. Indeed, in thymus competent acquired HF patients low lymphocyte counts were observed and related to adverse cardiac remodelling and worse prognosis [6,45]. We found NYHA III/IV patients exhibiting the lowest T cell numbers. In addition, low total lymphocyte and low T helper cell counts were related to a worse outcome in ACHD regarding the occurrence of adverse cardiac events. However, no significant difference in lymphocyte frequency was found between patients who received early/infant (≤1 year old) median sternotomy with suggested thymectomy compared to those having this kind of surgery later in life or never. This was surprisingly, as prior studies reported an impact of infant thymectomy on the immune system later in life [2,30]. However, our study was not designed to determine the effect of thymectomy, as actual residual thymus tissue could not be measured. In addition, the early sternotomised cohort comprised significantly younger patients having lower LVEF. ACHD patients who underwent sternotomy later or never included a large proportion of rather mild malformations, but also the most symptomatic patients (NYHA III/IV). Therefore, our data estimating the impact of thymectomy compared to HF should be interpreted with caution, due to patient selection and lacking records about thymus function. Generally, differentiation between both causing factors, thymectomy and HF will be challenging, as more severe congenital heart disease usually causes a greater need for early open-heart surgery. There is a major demand for basic science derived mechanistic insights using appropriate animal models and human specimens to elucidate the interplay of thymus removal, adaptive immunity, immunosenescence and HF in ACHD. Whether early thymectomy induces clinically relevant long-term consequences remains under discussion [30]. Increased incidences of cancer, auto-immune and infectious diseases have been reported in these patients [46]. Recently, auto-immunity was identified as important driver of HF progression after myocardial injury [47].

Being bone-marrow-derived, NK cells are not dependent on thymus function. Accordingly, a number of studies in ACHD did not find differences in NK subsets, neither shortly nor years after thymectomy [2,22]. However, Ramos et al. reported an increased number of the CD56^bright subset in thymectomised ACHD [23]. The authors suggested that the absence of the thymus promoted a differentiation of the common T and NK cell precursors into NK cells. The patient cohort in our study was also characterised by increased CD56^bright cells possibly relating to HF grade. Therefore, it can be suggested that HF-induced immune activation may expand the cytokine-producing CD56^bright NK subset, however NK cells did not relate to patient's outcome. NK cells expand at inflammatory sites and interact with monocytes by mutual cytokine-induced stimulation. This positive feedback loop between both cells is suggested to contribute to the pathogenesis of chronic inflammatory diseases [48]. Possible functional interactions between Mon3 and CD56^bright NK cells need to be further investigated. Limitations of this descriptive study are based on the small and unmatched control group. In addition, due to rather low correlation coefficients in some analyses, although being statistically significant, these data need to be interpreted with caution and potential confounders cannot be excluded. In addition, we are not able to provide mechanistic insights about interactions of the described immune cells.

4.4. Conclusions

In summary, systemic inflammation with elevated CRP and neutrophil-to-lymphocyte ratio is present in ACHD and increasing with heart failure severity. Immune activation by expansion of Mon2, Mon3, CD14⁺HLA-DR^neg/low monocytes and CD56^bright NK cells was observed in ACHD. In contrast, CD4⁺ T helper and CD8⁺ cytotoxic T cells were reduced in ACHD compared to control. Elevated Mon2 and low lymphocyte and T helper cell counts represent independent predictors of adverse cardiac events in ACHD. These insights open up a new perspective on inflammation and immunity in ACHD. However, further prospective studies are required to determine the role of mature and immature monocyte subsets, NK and T cells in ACHD heart failure pathomechanisms and onset of comorbidities.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: