This was an observational, multi-centre cohort study performed at Shaanxi Provincial People’s Hospital, China. The HCM cohort was evaluated to identify consecutive patients from February 2019 to November 2020 and formed the training dataset. An independent population study with an external HCM cohort was performed in Shaanxi, China at Pucheng County Hospital, Zhouzhi County Hospital, and Huazhou District People’s Hospital from March 2019 to March 2021.

Inclusion criteria for the study were as follows: (1) a diagnosis of HCM based on the presence of left ventricular hypertrophy on echocardiography (15 mm), which is not solely explained by abnormal loading conditions, according to the guidelines of the European Society of Cardiology (3); (2) age ≥18 and ≤70 years; (3) left ventricular ejection fraction ≥ 50%; and (4) New York Heart Association (NYHA) function class I or II.

The exclusion criteria were as follows: (1) other cardiac or systemic diseases that may produce left ventricular hypertrophy; (2) a history of hypertension; (3) a history of diabetic mellitus; (4) a history of coronary artery disease; (5) burden of premature beat ≥ 10%, second- or third-degree atrioventricular block, left bundle branch block (LBBB), interventricular conduction delay (IVCD), or acute pulmonary oedema; (6) previous episode of atrial fibrillation; (7) previous septal alcohol ablation or morrow surgery; (8) onset of atrial fibrillation during follow-up; and (9) onset of SCD during follow-up.

Our research focused on the feedback and input of patients through consultations. Patients were involved in the study design and conduction of this research. During the preliminary feasibility stage, the priority of the research question, and methods of recruitment were informed via panel discussions with patients and one structured on-line interviews. Furthermore, a patient joined the independent steering committee during the execution phases of the research. We intended to share the vital results of this study with our patients and planned to explore patient and public involvement in the development of an appropriate approach of dissemination.

Demographic characteristics including age, sex, body mass index (BMI), family history of sudden cardiac death (SCD), unexplained syncope ≤ 2 years, risk-SCD score (3), heart rate (HR), resting blood pressure, 6-min walk test (6-MWT), brain natriuretic peptide (BNP), and medication history were collected at enrolment.

Peripheral blood was sampled from patients in a fasting state in the morning following the enrolment day. Venous plasma concentrations of serum kalium, calcium, natrium, glycated haemoglobin, total cholesterol, low-density lipoprotein, high-density lipoprotein, triglyceride, lipoproteins, serum creatinine, N-terminal pro-B-type natriuretic peptide, red blood cells, haemoglobin, white blood cells, creatine, phosphokinase-myohaemoglobin (MB), myohaemoglobin (MB), troponin I, thyroid stimulating hormone, free triiodothyronine 3, and free triiodothyronine 4 were determined in the clinical laboratory department using standard biochemical techniques.

An ECG was performed as previously described (14). Twenty-four-hour ECG recordings were obtained using a digital Holter system. The transthoracic echocardiography protocol was based on standardised acquisition from the European Association of Echocardiography/American Society of Echocardiography (ASE) guidelines in all patients (15). Recordings in standardised views, including two-dimensional, M-mode, and colour flow Doppler and pulsed tissue Doppler imaging (TDI), were acquired with an EPIQ 7C system (S5-1 probe, Philips, Netherlands) for subsequent analysis. All measurements were calculated as the average of three consecutive cardiac cycles. Interventricular septum (IVS) thickness and left ventricular posterior wall (LVPW) thickness were defined as the greatest thickness in any single segment. Right ventricular diameters (RVDs) were evaluated as the greatest diameter at the end-diastolic stage in the apical four-chamber view and left ventricular long axis view, respectively. Left ventricular mass (LVM) and LVM index (LVMI) were calculated according to the Devereux formula: LVM (g) = 0.8 × 1.04 × [(IVS + LVPW + LVEDD)³-LVEDD³] + 0.6, LVMI (g/m²) = LVM/body surface area (BSA), respectively, and the body surface area (BSA) was calculated as follows: 0.0061 × height (cm) + 0.0128 × weight (kg)–0.1529. Left atrial volume (LAV) was calculated using the biplane area length method at end systole (16), LAVI (ml/m²) = LAV/BSA. Left ventricular ejection fraction (LVEF) was calculated using biplane Simpson’s rule (15). Left outflow tract gradients (LVOTG) were measured and automatically calculated from the flow velocities using the modified Bernoulli equation (17). The tricuspid regurgitation peak velocity was recorded using continuous wave Doppler.

Electromechanical coupling time courses (Qsb, Qst) in the interventricular septum (IVS) and lateral wall (Lat) of the left ventricle were measured using a combination of TDI echo and ECG. The Qsb time course is from the onset of the Q wave on the ECG to the beginning of the S wave on the TDI. The Qst time course is from the onset of the Q wave on the ECG to the top of the S wave on the TDI (Figure 1). HR is related to the electrical and mechanical activities of the heart (18). We determined all the data of electromechanical coupling time courses with HR correction, referenced as the formula in a previous study: electromechanical coupling time courses with HR correction = electromechanical coupling time courses without HR correction/HR × 60 (12, 18).

The primary end-point was HF admission, which was defined as follows: (1) exertional dyspnoea, (2) NYHA function class III-IV, and (3) diuretic is essential for improving symptoms. All patients were followed-up via face-to-face interviews in our hospital. The duration of follow-up was determined using the initial visit to the date of the first HF occurrence or the last visit (December 2021).

Data were collected using IBM SPSS statistical software (version 20.0) for Windows (SPSS Inc., Chicago, IL, United States) and R 4.0.1 (New York¹). Continuous variables were expressed as the mean and standard deviation. Categorical variables were expressed as frequencies and percentages. The Kolmogorov–Smirnov test was used to assess the normal distribution of the quantitative variables. The independent samples t-test was performed to compare parametric values between the HF and non-HF groups, whereas categorical variables were compared using the Chi-square test. Univariate and multivariate survival analyses were performed using the Cox regression analysis. Interclass correlation coefficient (ICC) was used to evaluate intra- and inter-observer variability in electromechanical coupling time parameters. The clinically significant predictors of the final regression model were formulated as a nomogram. Kaplan–Meier survival curves were used in the two groups according to the model score to further assess the prognostic value. The function “surv_cutpoint” in the R package “survminer” was applied to determine the optimal cut-off value of these chosen variables in our study. The accuracy of the predictive model was assessed using receiver operating characteristic curve (ROC) analysis and Harrell’s concordance index (C index). A calibration curve was derived to explore the consistency between predicted probabilities by the nomogram and observed HF rates. External validation was performed in an independent population. The nomogram was used to assess each patient in the validation dataset. Thereafter, the area under the curve (AUC), C-index and calibration curve were calculated to assess the discriminative performance and predictive accuracy of the nomogram model. Decision curve analysis (DCA) was used to calculate the net benefit by using the model in both training and validation datasets (19). The total scores of each patient were calculated based on the nomogram, and we developed an application for clinical use of the model.

All probability values were two-tailed. A p-value of less than 0.05 was considered statistically significant. The R code is submitted as a Supplementary Material.

Our study complied with the Declaration of Helsinki and was approved by the ethics committee of the Shaanxi Provincial People’s Hospital. Written informed consent was obtained from all patients.

Eighty-two patients fulfilled the study inclusion criteria out of a cohort of 142 patients with HCM. The reasons for exclusion were: 16 for having no second visit during the study period (lost to follow-up), 8 for coronary heart diseases, 4 for previous septal alcohol ablation, 2 for previous morrow surgery, 4 for previous acute HF, 2 for SCD, 10 for previous episode of atrial fibrillation, 11 for onset of atrial fibrillation during follow-up, and 2 for onset of SCD during follow-up (Figure 2). Clinical, demographic, echocardiographic characteristics, and electromechanical coupling time parameters of the population at the initial evaluation are presented in Table 1. The mean age was 42.20 ± 10.83 years, and 46 (56.09%) patients were men. The left ventricular wall thickness was 22.60 ± 5.04 mm (range, 15–35), and the mean ejection fraction was 59.50 ± 5.72%. All patients received standardised medical care during the clinical course.

Patients were followed-up for a median duration of 23 months (interquartile range, 11–34 months). Nineteen patients (23.17%) experienced HF. The baseline characteristics of patients with and without HF are outlined in Table 1. There were no significant differences in BMI, HR, IVS thickness, LVEF, LVOTG at rest, and provoked LVOTG. Compared with patients without HF, those who experienced HF had lower age, higher levels of creatine and triglyceride, obviously longer results for the 6-MWT, largest anteroposterior diameter of the right ventricle (RVD-1), larger right ventricular transverse diameter (RVD-2), and greater left atrial volume index (LAVI). Moreover, all four cardiac electromechanical coupling time courses (IVS-QSb, IVS-QSt, Lat-QSb, and Lat-QSt) were longer in patients who experienced HF (p < 0.001 for all comparisons).

Univariate Cox analyses to predict the endpoint are shown in Table 2. Significant predictors of HF events were IVS-QSb, IVS-QSt, Lat-QSb, Lat-QSt, SCD family history, 6-MWT, CREA, ApoA1, MB, RVD-1, RVD-2, LVEF, LAVI, P, QRS, and PR. Table 3 presents the results of multivariate Cox analysis, which demonstrated that Lat-QSb [hazard ratio (HR): 1.330; 95% confidence interval (95% CI): 1.083–1.633; p = 0.007], Lat-QSt (HR: 1.230; 95% CI: 1.026–1.463; p = 0.025), MB (HR: 1.004; 95% CI: 1.000–1.008; p = 0.042), PR (HR, 0.925; 95% CI, 0.864–0.989; p = 0.023), and LAVI (HR: 1.210; 95% CI: 1.006–1.454; p = 0.043) were significant and independent predictors of HF occurrence.

We plotted ROC curves to assess the accuracy of nomogram in predicting the risk of HF occurrence. The AUC for combining MB, PR, and LAVI was 0.751 (95% CI: 0.616–0.886), while that for our predictive nomogram was 0.948 (95% CI: 0.885–1.000, p < 0.001) (Figure 5A). Additionally, the nomogram displayed strong discrimination with a C-index of 0.918 (95% CI: 0.867–0.969) and strong calibration. The calibration curve of the accuracy was highly consistent with the diagonal, indicating that the predicted probability of HF was in accordance with the actual probability (Figure 5B).