We conducted an international multicentric retrospective analysis of consecutive HCM patients undergoing cardiac magnetic resonance (CMR) for diagnostic confirmation and/or risk stratification at Instituto do Coração (InCor), University of São Paulo (São Paulo, Brazil) and Hospital de Santa Cruz (Lisbon, Portugal). The Brazilian site included patients between 2003 and 2017, while the Portuguese site included patients between 2009 and 2022.

The diagnosis of HCM was made by experienced cardiologists and based on available clinical data and LV wall thickness ≥15 mm in the absence of loading conditions or other disease that could explain the ventricular hypertrophy.

Exclusion criteria were: (1) age <16 years; (2) alternative diagnosis such as athlete's heart, Anderson-Fabry disease, cardiac amyloidosis or sarcoidosis that could explain ventricular hypertrophy; (3) CMR inconsistent with HCM or with previously stated alternative diagnosis; (4) LGE pattern consistent with prior myocardial infarction; (5) missing essential echocardiographic and/or 24 h Holter monitoring data: (6) moderate or severe aortic/mitral disease; (7) indication for ICD for secondary prevention; (8) previously implanted intracardiac electronic device.

The study protocol was reviewed and approved by each institution's review board which waived the need for specific informed consent.

Our primary endpoint was a composite of SCD, appropriate ICD discharge and sustained VT. SCD was defined as witnessed sudden death with or without documented ventricular arrhytmia (VA) or death within one hour of new onset symptoms or nocturnal death with no antecedent history of worsening symptoms. Events were ascertained by reviewing electronic medical records including ICD electrograms. When death occurred outside of the hospital, the circumstances of death were determined by phone call interview with a family member.

Echocardiography, 24h Holter monitoring and exercise stress test data within a six-month window to the CMR were collected from the patient files. If no left ventricular outflow obstruction (LVOTO) was identified, we assumed an LVOT gradient of 3 mmHg since this is the minimum accepted by the HCM Risk-SCD calculator. Left atrial diameter was determined by M-Mode or 2D in the parasternal long axis plane and maximal LV wall thickness (MWT) was defined as the greatest dimension at any site within the LV myocardium. Non-sustained ventricular tachycardia (NSVT) on 24 h Holter monitoring was defined as ≥3 consecutive ventricular beats at a rate of ≥120 beats per minute. Abnormal blood pressure response on stress test was defined as either a failure to increase ≥20 mmHg or a drop of ≥20 mmHg during exercise.

The five-year HCM Risk-SCD score was calculated for each patient. Eligibility criteria for ICD according to the ESC 2014 (HCM), ESC 2022 (ventricular arrhythmias – VAs), ESC 2023 (Cardiomyopathies), ACC 2011, ACC 2020 and ACC 2024 (all HCM guidelines) was determined for each patient and subdivided into three categories: Class IIa indication for ICD (ICD should be considered); Class IIb indication for ICD (ICD may be considered) and; Class III indication (ICD not recommended).

All CMR scans were performed using a 1.5 T system. Images were transferred to a core laboratory where an experienced cardiologist in CMR, blinded for clinical events, analyzed all CMR data using a dedicated software (Circle Cardiovascular Imaging®, Calgary, Canada). LV volume, mass and EF were measured by using a standard volumetric technique. LV endocardial and epicardial borders on cine images were manually adjusted to define the myocardium, taking care to exclude papillary muscles and the intertrabecular blood pool. LV maximal wall thickness (MWT) was defined as the greatest dimension at any site within the LV myocardium.

In both centers, LGE images were acquired 10 min after the administration of intravenous gadolinium chelate contrast agent in the dose of 0.2 mmol/kg, with breath-hold two-dimensional segmented inversion-recovery sequences. Imaging was performed in short-axis views of the LV with 8 mm slice thickness and 2 mm gaps. The typical in-plane spatial resolution was 1.5 mm×1.5 mm. Inversion time was optimized to null normal myocardial signal. The LV short-axis LGE images were first visually accessed for the presence of LGE, followed by quantification if LGE was present.

Late gadolinium enhancement was defined as areas of signal intensity ≥6 standard deviations from normal myocardium and was expressed as the percentage of myocardial mass (LGE%). Any areas that were identified as LGE by the software, but visually deemed artefactual on visual analysis were manually excluded.

Categorical variables are presented as frequencies and percentages, while continuous variables are presented as the mean±standard deviation for normally distributed data or as the median and interquartile range for non-normally distributed data. The Student's t-test, Mann–Whitney U test, and Fisher's exact test were used for comparisons, as appropriate.

Diagnostic accuracy statistics, including sensitivity, specificity, positive predictive value, and negative predictive value, were calculated for each individual set of guidelines and for various LGE cut-off points.

Weighted κ was used to assess concordance in risk assessment between different guidelines. The strength of agreement between each classification was considered poor (κ<0.2), fair (κ=0.21–0.4); moderate (κ=0.41–0.6), good (κ=0.61–0.8) or very good (κ=0.81–1.0).

Univariate cox regression analysis was performed to determine predictors of arrhythmic events. Clinically relevant variables and/or variables with a p-value <0.10 on individual analysis were included in multivariate models. Net reclassification index (categorical NRI) was used to ascertain improvements in the risk stratification strategies over time in the American and European guidelines.

Kaplan–Meier survival curves were plotted for each guideline recommendation class classified per LGE% strata. The log-rank test was used to assess significant differences in time to endpoint between the risk strata.

Statistical significance was set at p-value <0.05 (two-sided). All analyses were performed using SPSS® 27.0 and MedCalc® 9.3.8.0.

We included a total of 531 patients (median age was 49 (IQR 35–61), 57% male). Over a median follow-up of 49 (IQR 21–88) months, 28 events occurred (15 SCDs, 6 appropriate ICD discharges and 7 sustained VTs).

Clinical risk factors were evaluated for each patient. In this cohort, 13% of patients had a Family history of SCD (n=71), 12% had had a recent episode of unexplained syncope which was deemed to be of possible cardiac etiology, and 19% (n=101) had at least one NSVT episode on Holter monitoring at the index evaluation.

Regarding transthoracic echocardiography parameters, median MWT was 19 (IQR 16–22) mm, LV gradient was 5 (IQR 3–50) mm and left atrium (LA) diameter was 43 (IQR 38–47) mm. Of these parameters, only LA diameter was significantly different between groups (47 [IQR 43–51] mm vs. 42 [IQR 38–47]mm; p=0.004).

The median HCM risk score in the cohort was 2.4 (IQR 1.6–4.1)%. Only 12% (n=65) of the patients had a calculated HCM Risk Score greater or equal to 6%. Although there is a numerical difference in the percentage of patients with an HCM risk score greater or equal to 6% between groups (21% vs. 12%), this did not achieve statistical significance (p=0.128). Table 1 shows the baseline population characteristics for our cohort separated by those who met the primary endpoint and those who did not.

The diagnostic accuracy statistics of the European and the American society guidelines have evolved similarly over time: sensitivity and positive predictive value increased, negative predictive value remained high and discriminative ability also increased (see Table 2 and Figure 4 for details).

The concordance in risk assessment is only moderate between the ESC 2022 and ACC 2020 (κ=0.60 (95% CI: 0.54–0.67); p<0.001) or the ESC 2023 and ACC 2024 guidelines (κ=0.56 (95% CI: 0.49–0.62); p<0.001).

The most recent 2020 (and 2024) AHA guidelines show a net reclassification index of 0.15 (p=0.01, mainly due to the correct reclassification of patients with arrhythmic events to receive an ICD [event 0.11; non-event 0.048]). When comparing the ESC's 2014 guidelines to the more recent recommendations in the 2022 and 2023 guidelines, reclassification also moved towards correctly classifying patients with arrhythmic events for ICD implantation. The NRI from the 2014 to the 2022 guidelines was 0.30 (p=0.027) (event 0.43; non-event −0.13), while the NRI from the 2014 to the 2023 guidelines was 0.17 (p=0.063) (event 0.21; non-event −0.05). However, the same cannot be said of the NRI from the 2022 to the 2023 ESC guidelines (NRI=−0.20 [p=0.039]; event 0.25; non-event 0.06).

Late gadolinium enhancement was present in 80% of patients, with a median LGE% of 3.2% (IQR 0.5–8.4%) and predicted arrhythmic events in this population (see Figure 1). LGE has remained an independent predictor of arrhythmic events after adjustment to known confounders (family history of SCD, unexplained syncope, NSVT, MWT, LVOT gradient, LA diameter, LVEF, apical aneurysm) with an aHR of 1.09 per 1% increase in LGE% (95% CI 1.05–1.12; p<0.001), and receiver operating characteristic analysis showed an AUC of 0.80 (0.72–0.88) (p<0.001) (see Figure 3).

Figure 1.

Kaplan–Meier curve showing event-free survival stratified by late gadolinium enhancement burden categories.

The Youden test showed the best cut-off value for LGE burden in our cohort of 8.0% (Youden index 0.513), which represented a sensitivity of 75% and specificity of 76%, with negative and positive predictive values of 15% and 98%, respectively.

Alternatively, the cut-off used by the most recent ESC guidelines to define a high burden of LGE (>15%) showed to have high specificity (88%) at the expense of a much lower sensitivity (46%), with positive and negative predictive values of 16% and 97%, respectively.

Regardless of the guideline publisher, among patients with any recommendation for ICD (classes IIa and IIb), the absence of LGE identified those who experienced no arrhythmic events during follow-up (see Figure 2, top). Furthermore, among patients for whom ICD implantation is not recommended (class III), those who reached the endpoint (and therefore experienced arrhythmic events) all showed some degree of LGE on CMR (see Figure 2 – bottom).

Figure 2.

Kaplan–Meier curves showing event-free survival stratified by late gadolinium enhancement burden in patients with class II recommendation (IIa should be considered/IIb may be considered) (top) and class III recommendation (not recommended) (bottom) according to the different most recent iterations of the guidelines.

Figure 3.

Receiver operator curve showing area under curve of late gadolinium enhancement burden.

Figure 4.

ROC curve showing AUC of the different iterations of the ESC (top) and ACC/AHA (bottom) guidelines.