A previous study from 2003–2004 entitled Life Chances 1 (LC1) investigated 364 patients with various types of repaired or palliated CHD [12]. These patients were treated by surgical or interventional procedures at the University Hospital of Goettingen, Germany, and then followed up in our ACHD clinic. The LC1 study included patients aged 13–50 years (median age, 26.4 years). Patients had undergone a medical history review, physical examination, electrocardiography (ECG), 2D-echocardiogram, blood sampling, metabolic exercise testing, and assessment of their psychological and socioeconomic status.

The present study is entitled Life Chances 2 (LC2). In 2017, all patients who had participated in LC1 were invited by phone, mail, or their family physician to attend the ACHD clinic for a follow-up examination [13].

The severity of CHD was assessed according to the 2020 European Society of Cardiology (ESC) Guidelines for the management of ACHD [14]. Patients were categorized into three groups: mild, moderate, or complex CHD. In patients with multiple cardiac lesions, the lesion with the highest complexity was used to assign the patient.

HF is not a single pathological diagnosis, but rather a clinical syndrome consisting of cardinal symptoms such as breathlessness, ankle swelling, and fatigue. For the purpose of this study, patients were classified as having developed HF if at least one of the following criteria was fulfilled:

(1) Patient had not been taking any HF medication (e.g., diuretic, beta blocker, ACE inhibitor, etc.) during LC1, but then started taking HF medication between LC1 and LC2. Although in a small number of patients these medications might have been used to treat hypertension, this condition could be considered a precursor for HF.

(2) Patient who required a surgical/interventional procedure for their underlying CHD, or had been admitted to a hospital for HF between LC1 and LC2.

(3) Patient who had died of HF between LC1 and LC2.

Only patients who had participated in the LC1 study were eligible for inclusion in LC2. Patients who were pregnant during the enrollment period for LC2 were excluded.

All patients provided written informed consent. The first part of this study was reviewed and approved by the ethics committee of Hannover Medical School under no. 3710 (date: 04-10-2004) and by the University Medical Center Goettingen under no. 10/2/01 (date: 01-03-2001). The second part was reviewed and approved by the ethics committee of the University Medical Center Goettingen under no. 15/8/14.

All patients underwent physical examination and measurement of heart rate, blood pressure, body weight, height, and standard 12-lead ECG.

Of the initial 364 patients in LC1, a total of 249 patients (68%, 134 male and 115 female) were recruited to participate in LC2. The remaining 115/364 (32%) patients did not participate in LC2 for the following reasons: patient could not be reached or was lost to follow-up (48/364, 13%), patient declined to participate (45/364, 12%), or patient had died (22/364, 6%). Two other patients died shortly after inclusion in LC2.

The distribution of CHD severity between the LC1 and LC2 participants did not differ significantly. CHD severity in the 364 patients (58% male, 42% female) from LC1 was mild (81, 22%), moderate (199, 55%), and severe (84, 23%), while in the 249 patients (58% male, 42% female) from LC2 it was mild (52, 21%), moderate (150, 60%), and severe (47, 19%).

Patients were further classified according to their diagnosis and lesion complexity. Table 1 shows patient classification based on diagnosis, as well as the patient demographics and prevalence of HF at initial assessment (LC1-HF), and in patients who developed HF by the start of LC2 (LC2-HF). A total of 57 patients (23%) had already developed HF at LC1, while another 67 of the remaining 192 patients (35%) developed HF during the follow-up period. Of note, patients with ventricular septal defect (VSD) closure had the lowest risk of developing HF (10%), whereas all patients with single ventricle physiology (Fontan) had developed HF by LC2.

Table 2 (Ref. [18]) shows the classification of patients according to lesion severity, as outlined by the 2020 ESC Guidelines for the management of ACHD [14]. Fifty-two (21%) had mild CHD, 150 (60%) had moderate CHD, and 47 (19%) had complex CHD. Patient demographics at the time of clinical work-up for LC1 and LC2 are also shown in Table 2. There was no significant difference in mean age between the three ACHD groups (p = 0.160).

As stated above, 67 of the 192 patients (35%) who did not have HF at LC1 later developed HF during the follow-up in LC2.

Table 2 shows the distribution of new HF cases according to lesion complexity. Of note, the percentage of patients with new HF increased as the lesion complexity increased (mild CHD: 17%, moderate CHD: 37%, complex CHD: 57%). Significant differences (p 0.001) were found between the mild vs. moderate, mild vs. complex, and moderate vs. complex groups.

Table 2 and Fig. 1A,B show the levels of NT-proBNP in the three ACHD groups at LC1 and LC2, respectively. Significant increases in NT-proBNP were observed in all three groups at LC2 compared to LC1. Differences between the three groups at LC2 remained significant (mean complex ACHD = 560 pg/mL, mean moderate CHD = 182 pg/mL, and mean mild CHD = 142 pg/mL; p 0.001).

Fig. 1.

NT-proBNP levels (pg/mL) for mild, moderate, and severe CHD at LC1 (A) and LC2 (B). Mean $\pm$ SD values are shown.

Table 3 shows the NT-proBNP levels at LC1 in patients who later developed HF in LC2. These were already significantly higher compared to patients who did not develop HF.

Table 2 and Fig. 2A,B show the percentage of predicted VO${}_{\text{2max}}$ in all three groups at LC1 and LC2, respectively. No significant difference in the mean VO${}_{\text{2max\%}}$ between patients with mild and moderate ACHD (85% and 86%, respectively; p = 0.880) was found at LC2. However, the mean VO${}_{\text{2max\%}}$ was significantly lower in complex CHD patients (72%, p 0.001) compared to the other two groups at LC2. Table 3 shows that patients who developed HF by LC2 already had significantly lower VO${}_{\text{2max\%}}$ at LC1 compared to patients who did not develop HF by LC2. The mean VO${}_{\text{2max\%}}$ for all patients from LC1 was found to be significantly lower than that of all patients from LC2 (74% vs. 81% respectively; p = 0.048).

Fig. 2.

VO${}_{\text{𝟐𝐦𝐚𝐱}}$ (% predicted) for mild, moderate, and severe CHD (mean $\mathrm{\pm }$ SD) at LC1 (A) and LC2 (B).

Table 2 and Fig. 3A,B show the QRS duration in all three groups at LC1 and LC2, respectively. In LC1 there was no significant difference in mean QRS duration between patients with moderate and complex ACHD (121 vs. 115 ms, respectively; p = 0.4). However, the QRS complexes of these patients was significantly longer compared to that of mild ACHD patients (103 ms; p = 0.002). In LC2, the mean QRS duration in all three groups was significantly longer compared to the equivalent group from LC1: 112 ms for mild (p = 0.03), 130 ms for moderate (p = 0.0001), and 132 ms for complex CHD (p = 0.001). The QRS duration of LC2 patients with mild CHD was shorter than that of the other two groups (p = 0.001), but there was no significant difference between the moderate and complex CHD patients (p = 0.99).

Fig. 3.

QRS duration (ms) for mild, moderate, and severe CHD (mean $\mathrm{\pm }$ SD) at LC1 (A) and LC2 (B).

The following parameters were found to be the best predictors for the development of HF between LC1 and LC2: NT-proBNP $>$73 pg/mL, VO${}_{\text{2max\%}}$ 73% of predicted, and QRS $>$120 ms. It is also important to note that NT-proBNP $>$73 pg/mL at LC1 was 1.7-fold higher than the 99% confidence interval (43 pg/mL) for healthy individuals. Table 4 shows the AUC, sensitivity, and specificity for the development of HF in all three ACHD groups, and separately for patient groups with mild and moderate CHD. To find the strongest predictor of HF, the AUC of each parameter was calculated individually and with all possible combinations. Although each parameter alone showed good prediction for developing HF, the combination of all three parameters resulted in the highest AUC (0.75), sensitivity (0.75), and specificity (0.63) for all lesion complexity. Although the results obtained with the combined parameters and with each individual parameter were not statistically different, the best AUC was obtained by using the combination of all three parameters, as seen in Fig. 4.

Fig. 4.

Results of AUC, sensitivity, and specificity analyses for the development of HF in all three ACHD groups for individual parameters, and in combination. No statistically significant differences were found between results for the combined parameters and the individual parameters of NT-proBNP (p = 0.630), VO${}_{\text{2max\%}}$ (p = 0.238) and QRS (p = 0.404).

Biomarkers have major significance for the diagnosis of HF. Giannokoulas et al. [8] reported that elevated BNP levels ($>$78 pg/mL) were a predictor of death in ACHD patients, with a sensitivity of 100% and a specificity of 76.3%. None of the patients in their study with a BNP level 78 pg/mL died during an 8-year follow-up period. In a study of 595 clinically stable ACHD patients, Baggen et al. [21] found that NT-proBNP $>$14 pmol/L was strongly associated with severe cardiovascular events such as HF and death. The present data confirms that increased NT-proBNP level is a good surrogate biomarker for cardiovascular risk stratification, including for HF in ACHD patients. We found significant differences in NT-proBNP levels between the three CHD groups in LC1, with the highest levels in the complex group, followed by the moderate and mild groups. During the 15-year follow-up period, a significant number of patients with moderate and complex CHD developed HF. A two-fold increase in NT-proBNP levels between LC1 and LC2 was observed in all three CHD groups. The increased NT-proBNP levels are likely attributable to HF rather than aging, since approximately 35% of patients developed HF between LC1 and LC2 [22, 23].

VO${}_{\text{2max}}$ is considered to be a reliable parameter for the measurement of cardiorespiratory capacity [24]. In a study of 1375 ACHD patients over a period of 10 years, Inuzuka et al. [9] demonstrated that a combination of peak oxygen uptake and heart rate reserve was related to midterm survival. Based on a random survival forest analysis, the authors found that 64% of predicted VO${}_{\text{2max}}$ was the optimal cut-off value for the prediction of 5-year survival. Furthermore, Diller et al. [25] identified that VO${}_{\text{2max}}$ was a predictor for the hospitalization and death of ACHD patients. Our study identified a threshold of 73% of predicted VO${}_{\text{2max}}$ as the optimal cut-off for predicting the development of HF.

Patients in LC1 with complex ACHD already had significantly reduced VO${}_{\text{2max\%}}$ compared to other patients, with this difference remaining unchanged after 15 years of follow-up. Interestingly, the mean VO${}_{\text{2max\%}}$ of all three patient groups was higher in LC2 than in LC1, probably because of the different exercise protocols used. In support of this, Michalik et al. [26] reported that the RAMP protocol resulted in higher VO${}_{\text{2max}}$ values during shorter duration of testing compared to the STEP protocol.

Widening of the QRS complex has been identified as an independent predictor of adverse outcomes in ACHD [10, 27]. Müller et al. [28] conducted a multicenter retrospective investigation on 875 patients with tetralogy of Fallot. These authors reported that patients with a QRS duration of $\ge$170 ms and a predicted VO${}_{\text{2max}}$ $\le$65% had an 11.4-fold increased risk of death or sustained ventricular tachycardia. In the present study, patients with moderate or complex CHD at LC2 had a significantly prolonged QRS duration (132 ms and 130 ms, respectively), whereas the QRS complex remained within normal limits in mild ACHD patients (112 ms). In most cases, QRS prolongation in patients from our cohort was due to complete right bundle branch block (RBBB), which has previously been associated with myocardial dysfunction and the development of HF [29, 30].

In adults without CHD, QRS prolongation $\ge$120 ms is present in 14% to 47% of patients with HF. Left bundle branch block is far more common than RBBB in these patients, in contrast to our CHD population. It is well established that left-sided intraventricular conduction delay is associated with more advanced myocardial disease, worse left ventricular (LV) function, poorer prognosis, and higher all-cause mortality compared with narrow QRS complex [31].

The progression of QRS complex duration in our patients (mainly in the moderate and complex CHD groups) from LC1 to LC2 is assumed to reflect decreasing cardiac function as a pattern of electro-mechanical interaction. In patients without CHD, progressive increases in QRS duration were shown to predispose HF patients to an increased risk of ventricular tachyarrhythmias [32, 33].

In order to identify surrogate parameters for predicting the development of HF, we applied logistic regression models to combinations of the top-performing variables of interest. Regardless of the severity of the underlying HF, the best result was found to be a combination of three parameters: NT-proBNP, VO${}_{\text{2max\%}}$, and QRS complex duration (AUC = 0.75, sensitivity = 75%, and specificity = 63%). For example, patients with NT-proBNP $>$1.7 times the upper limit of normal (ULN) and otherwise normal values for VO${}_{\text{2max}}$ and QRS complex duration had a 71% probability of developing HF between LC1 and LC2. However, the risk of developing HF increased to 75% if the patients fulfilled all three criteria of NT-proBNP $>$1.7 ULN, VO${}_{\text{2max\%}}$ 73% of predicted, and QRS $>$120 ms.

Our findings indicate that assessment of these parameters in ACHD patients could provide predictive information on patients who are at high risk of developing HF. These parameters have the advantage of being investigator-independent and of not requiring a deep knowledge of CHD. Hence, they might be a useful screening tool to indicate the need for referral to large ACHD centers.