Three-dimensional echocardiography-derived myocardial mechanistic insights into obstructive hypertrophic cardiomyopathy with moderate septal hypertrophy

Abstract

Background

In hypertrophic cardiomyopathy (HCM) patients with moderate septal hypertrophy (≤ 18 mm), left ventricular outflow tract obstruction (LVOTO) cannot be fully explained by septal thickness alone, thereby complicating therapeutic decision-making. This study aimed to investigate the association between three-dimensional speckle-tracking echocardiography (3D-STE)-derived myocardial mechanics and LVOTO in patients with moderate hypertrophic HCM.

Methods

We retrospectively analyzed 216 HCM patients with moderate septal hypertrophy, stratified into nonobstructive (n = 38), provokable LVOTO (n = 63), and resting LVOTO (n = 115) groups. LV geometry, LVOT diameter, and anterior mitral leaflet length were assessed using transthoracic echocardiography, and mitral subvalvular papillary muscle abnormalities were evaluated and composited into a SubMV score. 3D-STE quantified strain parameters, including global longitudinal, circumferential, radial, and area (GAS) strain, twist, and torsion. Measurement differences were analyzed among the LVOTO subgroups, and associations with LVOTO were evaluated using multivariate regression, restricted cubic spline (RCS) modeling, and receiver operating characteristic (ROC) curve analysis.

Results

Both LVOTO subgroups showed significantly augmented strain mechanics versus nonobstructive patients, most pronounced in resting LVOTO (p < 0.05). Twist and torsion were the strongest discriminators (p < 0.001). Multivariate regression showed that, beyond LVOT diameter and SubMV score, torsion (B = 9.47), twist (B = 1.92), and GAS (B = 1.29) were independently associated with provoked LVOT gradients (all p < 0.05). RCS revealed nonlinear twist/torsion–LVOTO relationships, with the obstruction risk rising above twist = 15° and torsion = 3°/cm, especially under provocation. Torsion achieved 81% specificity for identifying provokable LVOTO (cutoff 2.4°/cm, AUC = 0.72), outperforming all single structural parameters. Integrating 3D-STE mechanics with structural metrics improved discrimination (AUC 0.84 vs. 0.76, DeLong p = 0.003).

Conclusions

In moderately hypertrophic HCM, 3D-STE–derived strain mechanics, particularly LV torsional mechanics, are independently associated with LVOTO beyond established structural parameters including anatomical narrowing and cumulative papillary muscle abnormalities. These findings provide mechanics insights into LVOTO and support refined risk stratification and longitudinal monitoring in this heterogeneous population.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12880-026-02257-8.

Background

Hypertrophic cardiomyopathy (HCM) is the most common inherited cardiomyopathy, with an estimated prevalence of 1:200–1:500 in the general population [1]. A pivotal feature of HCM is left ventricular outflow tract obstruction (LVOTO), which occurs in up to 70% of patients at rest or under provocation and is strongly associated with heart failure symptoms, arrhythmias, and adverse outcomes [1–3]. HCM is defined by LV wall thickness (WT) ≥ 15 mm or ≥ 13 mm in genetically or familial affected individuals [1, 4]. Septal hypertrophy is traditionally considered the primary anatomical determinant of LVOTO. However, a substantial proportion of patients with moderate septal thickening (≤ 18 mm) exhibit significant LVOT gradients [5–7], suggesting that other factors beyond myocardial thickness contribute to obstruction in this subgroup. In this context, some investigators have proposed that mitral valve–associated abnormalities, including leaflet elongation and subvalvular papillary muscle changes, influence LVOTO and have advocated corresponding therapeutic strategies in patients with moderate hypertrophy; however, these strategies remain debated because of procedural risks and variable outcomes [6, 8–10].

Emerging evidence indicates that positive inotropic agents enhance myocardial contractility and may trigger or exacerbate LVOTO [11]. Accordingly, myosin inhibitors such as mavacamten, have been validated to relieve LVOTO by modulating myocardial contractility, independent of structural remodeling [12, 13]. In parallel, imaging studies using strain analysis have shown that enhanced myocardial deformation, including increased radial, circumferential, and rotational mechanics, is associated with LVOTO irrespective of anatomy structure [14–16].

Despite these findings, the association between myocardial deformation patterns in patients with moderate septal hypertrophy and LVOTO remains incompletely characterized. Three-dimensional echocardiography (3DE) with speckle-tracking enables the noninvasive assessment of myocardial strain and rotational mechanics with high temporal resolution and minimal geometric assumptions [17, 18]. Accordingly, we hypothesized that specific 3D-STE–derived myocardial mechanical features are associated with the presence and severity of LVOTO in patients with moderate hypertrophy. This study aimed to characterize mechanical patterns across different obstructive states and quantify their incremental association with LVOTO beyond established structural parameters, thereby adding the hemodynamic-mechanics insights into this heterogeneous subgroup.

Methods

Study design and population subgroups

This retrospective study screened 1,236 consecutive patients with HCM at Tongji Hospital between July 2023 and January 2025, with HCM defined according to the 2024 ESC criteria [1]. All patients underwent standard two-dimensional echocardiography (2DE), 3DE, and contrast echocardiography. Contrast echocardiography and/or CMR were used to identify patients with moderate septal hypertrophy (maximal WT ≤ 18 mm) [8], yielding 298 eligible subjects. The exclusion criteria were as follows: (1) prior septal reduction, valvular surgery, or major cardiovascular intervention; (2) significant intrinsic valvular disease except for systolic anterior motion (SAM)-related mitral valve (MV) regurgitation (MR); (3) congenital heart disease; (4) inadequate 3DE image quality; and (5) apical- or mid-ventricular obstruction. The final cohort was stratified into three subgroups based on LVOTO status: (1) Nonobstructive LVOTO, both resting and provoked LVOT gradients < 30 mmHg; (2) Provokable LVOTO, resting LVOT gradient < 30 mmHg but provoked gradient ≥ 30 mmHg; and (3) Resting LVOTO, both resting and provokable LVOT gradients ≥ 30 mmHg. Additionally, 32 healthy controls were included for comparison in the 3DE strain analysis.

The study was approved by the Ethics Committee of Tongji Medical College (approval number: 2022-S013-(1–4)). The procedures used in this study adhered to the tenets of the Declaration of Helsinki.

Clinical data collection and imaging acquisition

Clinical data was extracted from electronic medical records. Echocardiographic studies were conducted by experienced sonographers (W. Z. and J. T.) using the same standardized protocol and a commercial ultrasound system (Vivid E95, GE Healthcare, Horten, Norway).

For 2DE, standard parasternal long-axis, apical four-chamber, and apical two-chamber views were acquired with individualized optimization to maximize the frame rate and ensure a clear LV endocardial definition. Patients with resting LVOT gradients < 50 mmHg underwent Valsalva maneuvers as a provocation; if these failed to induce LVOTO, exercise echocardiography was performed using the modified Bruce treadmill protocol [19]. Frame rates of 25–35 frames/s and volume rates of ≥ 40 volumes/s were maintained to ensure adequate temporal resolution for 3D STE analysis. All datasets were reviewed for image quality, and only those with clear delineation of the endocardial and epicardial borders across the entire left ventricle were included in the strain analysis. A 12-slice display was used to verify the complete LV coverage before further processing.

All CMR examinations were performed using a 3-T system (MAGNETOM Skyra; Siemens Healthineers, Erlangen, Germany). Cine images were acquired in standard short-axis and long-axis views using ECG-gated, breath-hold balanced steady-state free precession (bSSFP) sequences (slice thickness, 8 mm; temporal resolution, 32 ms). Late gadolinium enhancement (LGE) imaging was performed 10 min after intravenous administration of gadobutrol (0.2 mmol/kg) using a breath-hold phase-sensitive inversion recovery (PSIR) sequence. Detailed imaging protocols are provided in Supplementary Material.

Image processing and analysis

All datasets were exported to a dedicated workstation (EchoPAC, Vision 204; GE Medical System) for offline analysis.

Structural measurements were obtained using 2DE and mainly included maximal wall thickness (Maximal WT), left ventricular outflow tract diameter (LVOTD), early diastolic transmitral flow velocity (E), early diastolic mitral annular velocity (e′), and anterior mitral leaflet (AML) length. LVOTD was measured at mid-systole 0.5–1 cm below the aortic annulus using the inner edge-to–inner edge method, while AML length at end-diastole, both obtained in the parasternal long-axis view [4]. Septal morphology was determined from the long-axis view [20]. SAM and MR grades were evaluated according to the current guidelines for the Management of HCM [1, 19, 21]. The left atrium volume index (LAVI) was derived using the biplane method at end-systole and indexed to body surface area (BSA).

A semi-quantitative scoring system, termed the SubMV score, was applied to evaluate the accumulation of mitral subvalvular papillary muscle abnormalities. Each component was weighed equally to simplify scoring and ensure feasibility in clinical practice. One point was assigned for one of the following anomalies, with a maximum total score of 4: (1) accessory papillary muscle (additional or abnormally positioned papillary muscle structures) or anomalous muscular bundles (aberrant myocardial structures extending from the LV apex to the basal septum or anterior wall) [4, 22, 23], (2) anterior displacement or bifurcation of the papillary muscles, (3) papillary muscle hypertrophy (an end-diameter ＞11 mm on imaging), and (4) insertion of papillary muscle into the MV leaflet. These characteristics were defined based on prior literature descriptions and expert consensus [22–24]. The representative features and scoring criteria are illustrated in Fig S1. The scoring was performed independently by two experienced sonographers with 6 and 13 years of experience. Discrepancies were resolved by consensus agreement. When available, the CMR images were reviewed for reference. The SubMV score was reassessed in a randomly selected subset of 20 patients, with repeated measurements performed independently at a later time to assess reproducibility.

3DE analysis was performed using the 4D AutoLVQ software (GE Healthcare) following standardized analysis protocols [25, 26]. Offline analysis began with semi-automated endocardial delineation using manually identified anatomical landmarks (LV apex and mitral annular hinge points) on the end-diastolic frame. A dynamic 3D LV model was generated to derive volume–time curves, from which the LV end-diastolic volume (EDV), end-systolic volume (ESV), and ejection fraction (EF) were calculated automatically. Volumetric indices were normalized to the body surface area to obtain EDV index (EDVI) and ESV index (ESVI). For myocardial deformation assessment, the endocardial contour was tracked throughout the cardiac cycle using a 17-segment LV model. The epicardial surface was automatically generated based on a predefined myocardial thickness model, and both endocardial and epicardial borders were visually inspected and manually adjusted as needed. 3D speckle-tracking was then performed across the cardiac cycle to assess myocardial motion in all directions, enabling the calculation of global longitudinal (GLS), circumferential (GCS), radial (GRS), and area (GAS) strains as the average of 17 segments. Rotational mechanics were quantified as LV twist (apical minus basal rotation at peak systole, °) and torsion (twist normalized to LV long-axis length (°/cm). All strain and rotational parameters were automatically derived, with quality control ensured via visual inspection and manual correction when necessary. A schematic of the analysis workflow is shown in Fig S2. All 3D-STE analyses were performed by a 6-year-experienced echocardiographer (Y.W.B.). Inter-study reproducibility was tested in 20 randomly selected patients who were reanalyzed one week apart.

Additionally, the LGE percentage (LGE%) was calculated as the LGE volume relative to the total myocardial volume, with regions exhibiting a signal intensity ≥ 5 SD above that of the normal myocardium defined as fibrotic. LGE burden was categorized as < 5%, 5–15%, and ≥ 15% for comparative analysis.

Statistical analysis

Continuous variables are presented as mean ± SD, and categorical variables as counts (%). Between-group comparisons were performed using one-way analysis of variance (ANOVA) or the Kruskal–Wallis test for continuous variables, and the χ² test or Fisher’s exact test for categorical variables as appropriate. Variables with p < 0.05 in the univariate analysis or with established clinical relevance were included in multivariable linear regression models, adjusted for potential confounders (e.g., age, BSA), to identify independent predictors of LVOT gradients, with regression coefficients (B) and 95% confidence intervals (CIs) reported. Restricted cubic spline (RCS) analyses based on logistic regression were used to assess potential nonlinear associations between tortional mechanics and the probability of LVOTO. Receiver operating characteristic (ROC) curves were generated to evaluate the discriminative performance of significant 3DE-derived strain parameters for identifying LVOTO phenotypes, and AUCs were compared using the DeLong test to determine the incremental diagnostic value of myocardial mechanics beyond conventional structural parameters. The optimal cutoffs were selected using Youden index. The reproducibility of the SubMV scores and 3DE measurements was assessed using intraclass correlation coefficients (ICC) and/or Bland–Altman analysis. Statistical analyses were performed using SPSS version 23.0 (IBM, Chicago, IL, USA) and MedCalc version 20.0.22 (MedCalc, Ostend, Belgium). A two-sided p value < 0.05 was considered statistically significant.

Results

Of the 298 patients with moderate HCM, 82 were excluded (Fig. 1), leaving 216 patients for analysis. Among the excluded patients, 21 were removed due to inadequate 3DE image quality, primarily caused by arrhythmia (n = 8), high BMI (n = 7), or chronic obstructive pulmonary disease (n = 6). The final cohort included 216 patients: 38 with nonobstructive, 115 with resting LVOTO, and 63 with provocable LVOTO. Among the 46 patients who underwent exercise echocardiography, 38 did not reach an LVOT gradient ≥ 30 mmHg at rest or during provocation and were classified in nonobstructive group. Three patients reached the LVOTO threshold only during exercise and were assigned to the provokable LVOTO group, while five patients with resting LVOT gradients ≥ 30 mmHg were classified as having resting LVOTO.

Fig. 1.

Workflow of the study enrollment. HCM: hypertrophic cardiomyopathy; LVOTO: left ventricular outflow tract obstruction, LV: left ventricular, 3DE: three-dimensional echocardiography

Baseline characteristics and echocardiographic parameters

The baseline clinical characteristics of the 216 patients with HCM, stratified by LVOTO status (nonobstructive, resting, or provokable), are summarized in Table 1. Systolic blood pressure was highest in the nonobstructive group (p < 0.05). Paroxysmal supraventricular tachycardia and dyspnea were most frequent in the resting LVOTO group (p < 0.05). NT-proBNP levels were significantly higher in both the nonobstructive and resting LVOTO groups than in the provokable group.

Table 1.

Population clinical characteristics

	All(n=216)	HCM with moderate hypertrophy			P
		No LVOTO (n=38)	Provokable LVOTO(n=63)	Resting LVOTO (n=115)	ANOVA
Resting LVOT gradient, mmHg	38 (11,98)	6 (4,9)	14 (9,19)	86 (59,125)	-
Provoked LVOT gradient, mmHg	91 (57,131)	18 (8,24)	70 (57,99)	120 (92,146)	-
Male, n (%)	130(60.2%)	29 (76.32%)#	40 (63.49%)	61 (53.04%)	0.032
Age, year	56(45,64)	52 (38,61)	57 (46,63)	56 (45.5,65)	0.241
BSA, m2	1.75±0.2	1.73±0.24	1.75±0.2	1.75±0.18	0.847
SBP, mmHg	128 (115,142)	137 (111,148)	132 (120,146)#	125 (113,137)*	0.027
DBP, mmHg	77±11	78±13	79±12#	74±10*	0.034
Heart rate, bpm	61 (56,69)	66 (61,73)	60 (54,68)	61 (58,69)	0.419
NYHA class	3 (2,3)	3 (1,3)	2 (2,3)	3 (2,3)	0.302
NT-proBNP, pg/ml	260 (115,583)	444 (149.5,875)*	206 (96.3,299.2)#	455 (194.3,893.7)*	0.020
cTnI, ng/ml	11(7.4,18)	23.5 (4.4,41.8)	6.8 (4.2,15)	7 (4,15.8)	0.087
Comorbidity
Hypertension	75 (34.72%)	11 (28.95%)	22 (34.92%)	42 (36.52%)	0.696
Diabetes	27 (12.5%)	3 (7.89%)	8 (12.7%)	16 (13.91%)	0.622
CAD	67 (31.02%)	7 (18.42%)	20 (31.75%)	40 (34.78%)	0.166
PSVT	29 (13.43%)	1 (2.63%)#	6 (9.52%)	22 (19.13%)	0.02
Chest pain	71 (32.87%)	11 (28.95%)	19 (30.16%)	41 (35.65%)	0.645
Dyspnea	102 (47.22%)	11 (28.95%)#	26 (41.27%)	65 (56.52%)	0.007
Syncope	28 (12.96%)	5 (13.16%)	7 (11.11%)	16 (13.91%)	0.867
Palpitation	66 (30.56%)	9 (23.68%)	17 (26.98%)	40 (34.78%)	0.334
Medication
β-blocker	85 (39.35%)	7 (18.42%)#*	26 (41.27%)	52 (45.22%)	0.013
Calcium channel antagonists	23 (10.65%)	0 (0%)*	12 (19.05%)	11 (9.57%)	0.009

Footnote: #indicates vs. resting LVOTO; *indicates vs. provokable LVOTO. Data are presented as mean ± standard deviation or median (interquartile range) for continuous variables and number (percentage) for categorical variables. Comparisons among groups were performed using one-way analysis of variance (ANOVA) or the Kruskal-Wallis test for continuous variables, and the chi-square test for categorical variables. Adjusted p-values were calculated using the Bonferroni correction for multiple comparisons. BSA: body surface area, LVOT: left ventricular outflow tract, SBP: systolic blood pressure, DBP: diastolic blood pressure, NYHA: New York Heart Association, CAD: coronary artery disease, PSVT: paroxysmal supraventricular tachycardia

Among the conventional parameters, maximal WT was slightly greater in the resting LVOTO group than in the provokable group (16 vs. 15 mm, p < 0.05). Overall LVEF was preserved (mean 66%), with the highest value observed in the resting LVOTO subgroup (71.2%), paralleling the trend in 2DE-derived GLS. Diastolic dysfunction was most pronounced in the resting LVOTO group, as reflected by the highest E/e′ ratio and largest LAVI (both p < 0.05). The severity of SAM and MR followed the patterns of obstruction, being greatest in the resting LVOTO group, followed by the provokable group, and lowest in the nonobstructive cohort. Morphologically, a sigmoid septum predominated in the resting LVOTO group (63.4%), whereas a reverse curve type was more common in the nonobstructive group (52.6%). A detailed comparison is presented in Table 2.

Table 2.

Conventional LV structural measurements of HCM subgroups

	All (n=216)	HCM with Moderate Hypertrophy			p-value
		Nonobstructive (n=38)	Provokable LVOTO(n=63)	Resting LVOTO (n=115)	ANOVA
Maximal WT, mm	16 (15,17)	16 (15,17)	15 (14,16)	16 (15,17)*	0.009
LVEDVI, ml/m2	49.7 (42.8,56.1)	51.9 (41.7,59.4)	54.2 (45.8,59.9)	58.5 (48.94,64.0)	0.030
LVESVI, ml/m2	18.72 (16.1,20.5)	18.2 (14.7,22.9)#*	15.91 (12.35,20.05)	15.12 (12.8,18.82)	<0.001
LVEF, %	66 (60,64)	63.0 (57.8,62.5)#	66.4 (63.1,73.9)#	71.2(65.01,77.3)*	<0.001
LAVI, ml/m2	38 (30,45.92)	35 (26,41.75)#	34.7 (28,39)#	42 (34.2,50.95)*	<0.001
E/e’	13 (10,16)	10.8 (8,12.75)#	12 (8.22,15)#	15 (12,18)*	<0.001
SAM	2 (0,3)	0 (0,1)#	1 (0,2)#	3 (2,4)*	<0.001
2D GLS, %	-15.99±3.12	-13.95±3.54#*	-16.16±2.75#	-16.56±2.91	<0.001
MR grade	2 (1,3)	1 (1,1)#	1 (1,2)#	3 (2,4)*	<0.001
Septal morphology
Sigmoid	137(63.4%)	8(21.1%)#*	46(73.0%)#	83(62.6%)*	<0.001
Reverse curve	48(22.2%)	20(52.6%)#*	9(14.3%)#	19(14.8%)*	<0.001
Others	31(14.4%)	10(26.3%)#	8(12.7%)	13(4.3%)	<0.001
CMR-derived LGE	150(69.4%)	19(50%)	38(60.3%)	93(80.9%)
<5%	126(84%)	9/19(47.4%)#*	31/38(81.6%)	86/93(92.5%)	<0.001
5-15%	19(12.7%)	8/19(42.1%)#*	4/38(10.5%)#	7/93(0.8%)*	<0.001
≥15%	5(3.3%)	2/19(10.5%)#*	3/38(7.9%)	0	<0.001
SubMV Score	1(0,1)	1 (0,1)#	1 (0,1)#	1 (1,2)*	<0.001
Accessory papillary muscle	40 (18.52%)	6 (15.79%)	10 (15.87%)	24 (20.87%)	0.234
Anterior displacement of papillary muscles	77 (35.65%)	11 (28.95%)	14 (22.22%)#	52 (45.22%)*	0.006
Papillary muscle hypertrophy	12 (5.56%)	2 (5.26%)	3 (4.76%)	7 (6.09%)	0.931
Papillary muscle multifurcation	25 (11.63%)	3 (7.89%)	1 (1.61%)#	21 (18.26%)*	0.003
Insertion of papillary muscle into the mitral leaflet	6(2.8%)	0(0)	1(1.6%)	5(4.3%)	0.291

Footnote: Data are presented as mean ± standard deviation or median (interquartile range) for continuous variables and number (percentage) for categorical variables. Comparisons among groups were performed using one-way analysis of variance (ANOVA) or the Kruskal-Wallis test for continuous variables, and the chi-square test for categorical variables. Adjusted p-values were calculated using the Bonferroni correction for multiple comparisons. #indicates vs. resting LVOTO, * indicates vs. provokable LVOTO. WT: wall thickness, LVEDVI: left ventricular end diastolic volume index, LVESVI: left ventricular end systolic volume index, LVEF: left ventricular ejection fraction, LAVI: left atrial volume index, SAM: systolic anterior motion, GLS: global longitudinal strain, MR: mitral regurgitation, CMR: cardiac magnetic resonance, LGE: late gadolinium enhancement

Of the 216 patients, 150 (69.4%) underwent CMR with LGE evaluation: 19 (12.7%) in nonobstructive group, 38 (25.3%) in provokable LVOTO, and 93 (62.0%) in resting LVOTO. Interestingly, the extent of LGE was significantly lower in both the resting (p = 0.001) and provokable (p = 0.018) LVOTO groups than in the nonobstructive group. SubMV scores were highest in the resting LVOTO group (p < 0.001), primarily driven by anterior papillary muscle displacement and bifurcation (p < 0.001).

3DE metrics comparison

As summarized in Table 3, patients with resting LVOTO showed the greatest augmentation in multidirectional mechanics, including GCS, GAS, GRS, twist, and torsion, followed by the provokable LVOTO and nonobstructive groups. The differences between each LVOTO subgroup and the nonobstructive group were statistically significant (all p < 0.05), whereas no significant differences were observed between the two LVOTO subgroups (all p > 0.05). Representative 3DE strain maps for each subgroup are shown in Fig. 2, visually illustrating the intergroup differences. Notably, twist and torsion showed the largest differences across the subgroups (both p < 0.001). Additionally, strain parameters in the nonobstructive group were most similar to those of the 32 healthy controls, with comparison data provided in Table S1.

Table 3.

Comparison of 3DE LV strain components of HCM subgroups

	HCM with Moderate Hypertrophy				P values of LVOTO vs. nonobstructive		P values of two LVOTO groups
	All (n=216)	Nonobstructive (n=38)	Provokable LVOTO(n=63)	Resting LVOTO(n=115)	Provokable LVOTO	Resting LVOTO
GLS, -%	14.4(12.0, 16.7)	12.5 (10.0, 13.3)	15.0 (13.0, 17.0)	15.0 (12.0, 17.0)	0.003*	0.002*	>0.999
GCS, -%	18.6(16.1, 21.1)	17.0 (14.0, 18.3)	18.0 (15.3, 21.0)	19.0 (17.0, 21.0)	0.023*	0.002*	0.213
GAS, -%	28.8(24.6, 32.3)	25.0 (22.0, 29.0)	29.0 (25.0, 32.8)	30.0 (26.0, 32.0)	0.020*	0.001*	>0.999
GRS, %	46.6(37.2, 53.5)	36.5 (32.0, 46.5)	48.0 (38.0, 54.0)	48.0 (40.0, 54.5)	0.012*	0.001*	>0.999
Peak twist, °	12.4(6.6, 17.3)	6.3 (3.1, 10.4)	13.2 (7.05, 17.4)	13.7 (8.4, 19.4)	0.004*	<0.001*	0.397
Peak torsion, °/cm	2.5±1.2	1.6 (0.9, 2.2)	2.5 (1.5, 3.1)	2.7 (2.1, 3.6)	0.031*	<0.001*	0.092

Footnote: Data are presented as mean ± standard deviation or median (interquartile range) for continuous variables and number (percentage) for categorical variables. Comparisons among groups were performed using one-way analysis of variance (ANOVA) or the Kruskal-Wallis test for continuous variables, and the chi-square test for categorical variables. *indicates statistic difference is significant (p<0.05). LVOTO: left ventricular outflow tract obstruction, GLS: global longitudinal strain, GCS: global circumferential strain, GAS: global area strain, GRS: global radial strain

Fig. 2.

Representative 3DE-derived strain maps in different LVOTO states. Representative 3DE strain maps are shown for individuals with nonobstructive (top row), provokable LVOTO (middle row), and resting LVOTO (bottom row). In each case, the first column presents the three-chamber view, followed by color-coded polar maps illustrating longitudinal strain, circumferential strain, area strain, radial strain, twist, and torsion. Segmental values are displayed numerically, with corresponding global values provided in the lower left corner of each map. Resting LVOTO exhibits the most pronounced multidirectional strain augmentation compared with other subgroups. 3DE: three-dimensional echocardiographic, LVOTO: left ventricular outflow tract obstruction

Linear associations and classification performance

Multivariate linear regression models adjusted for age, sex, and BSA identified LVOTD and SubMV score as consistent structural predictors of both resting and provoked LVOT gradients. To account for collinearity, ESVI was used to represent LV size, and twist and torsion were analyzed using separate models. Additionally, maximal WT independently predicted resting but not provoked gradients. Importantly, twist, torsion, and GAS remained independent mechanical predictors of both resting and provoked LVOT gradients (all p < 0.05 in models including torsion). The detailed results are presented in Table 4.

Table 4.

Univariate and multivariate linear regression analysis of the resting and provoked gradients of left ventricular outflow tract

	Provokable LVOT Gradient						Resting LVOT Gradient
	Univariate		Multivariable (Model 1)		Multivariable (Model 2)		Univariate		Multivariable (Model 1)		Multivariable (Model 2)
	B	P	B (95% CI)	P	B (95% CI)	P	B	P	B (95% CI)	P	B (95% CI)	P
Sex	-14.94	0.05					-21.11	0.003
Age	0.66	0.01					0.64	0.009
Maximal WT, mm	4.36	0.048	3.33(-0.90,7.55)	0.125	3.44(-0.96,7.83)	0.124	4.323	0.039	4.10(0.12,8.09)	0.044*	3.93(0.15,7.70)	0.043*
LVOTD, mm	-5.55	0.000	-4.95(-7.20,-2.70)	0.000*	-3.70(-6.3,-1.07)	0.006*	-6.561	<0.001	-4.89(-7.28,-2.49)	0.000*	-5.02(-7.05,-2.99)	0.000*
3DE EDVI, ml/m2	0.816	0.007	-		-		0.628	0.029	-		-
3DE ESVI, ml/m2	-1.54	0.011					-1.37	0.017
AML length, mm	1.563	0.062					1.235	0.122
SubMV score	17.762	0.000	11.56(3.0,20.11)	0.009*	13.04(4.3,21.75)	0.004*	17.148	<0.001	10.79(2.81,18.78)	0.008*	11.08(3.45,18.71)	0.005*
GLS, -%	2.334	0.037					1.512	0.158
GCS, -%	2.27	0.013					2.87	0.001
GAS, -%	1.898	0.004	1.29(0.35,2.39)	0.035*	2.86(-0.38,6.11)	0.084	2.0	0.001	1.21(-0.83,3.25)	0.241	1.24(0.15,2.33)	0.027*
GRS, %	0.426	0.127					0.385	0.149
Twist, °	1.997	0.000	1.92(1.02,2.82)	0.000*	-		1.644	0.001	1.35(0.48,2.23)	0.003*	-
Torsion, °/cm	13.386	0.000	-		9.47(3.63,15.31)	0.002*	14.416	<0.001	-		10.64(5.55,15.73)	0.000*

Footnote: Univariate variables with p < 0.05 were entered into the multivariate model using a stepwise approach. * Indicates statistical significance (p < 0.05). Model 1 included LV twist, whereas Model 2 incorporated LV torsion for model construction. LVOTO: left ventricular outflow tract obstruction; LVEDVI: left ventricular end diastolic volume index, LVESVI: left ventricular end systolic volume index, LVEF: left ventricular ejection fraction, 3DE: three-dimensional echocardiography, AML: anterior mitral leaflet; GLS: global longitudinal strain, GCS: global circumferential strain, GAS: global area strain, GRS: global radial strain

Further RCS analyses demonstrated load-dependent and nonlinear associations between LV twist/torsion and the risk of LVOTO, with evident threshold effects, as indicated in Fig. 3. For twist, in the resting state, the association was modest and near-linear, with odds ratios (ORs) increasing from 1.0 to approximately 2.0 as twist rose to 15° and plateauing above 15°. Under provocation, the association became more pronounced and nonlinear, with ORs rising sharply from 1.0 to approximately 4.0 across 5–15°, followed by a plateau above 15°. For torsion, the nonlinear relationship was more pronounced. At rest, ORs increased steeply from 1.0 to approximately 3.0 within the 1–3°/cm range, plateauing thereafter. Under provocation, the association was markedly nonlinear, with ORs rising rapidly and peaking near 6.0, indicating markedly increased LVOTO risk under hemodynamic stress.

Fig. 3.

RCS curves illustrating the associations of LV twist and torsion with the likelihood of LVOTO under resting and provoked conditions. Red lines denote ORs estimated from the RCS models, and shaded areas represent 95% confidence intervals. Overall association significance is indicated by the P value for the overall, while deviation from linearity is assessed by the P value for nonlinear. The upper panels show LV twist in relation to resting LVOTO (left) and provocable LVOTO (right); the lower panels depict LV torsion in relation to resting LVOTO (left) and provocable LVOTO (right). Both twist and torsion exhibited significant nonlinear associations with LVOTO, with obstruction risk increasing progressively at higher values and reaching a plateau or peak at approximately 15° for twist and 3°/cm for torsion. Notably, these associations were more pronounced under provoked conditions. RCS: Restricted cubic spline, OR: odds ratios LVOTO: left ventricular outflow tract obstruction

ROC analysis (Table 5) demonstrated LV torsion as the strongest determinant of provokable LVOTO, achieving 81% specificity at a threshold of 2.4°/cm (AUC = 0.72, p < 0.001), outperforming all structural metrics, with twist as the next best performer. For resting LVOTO, LVOTD yielded the highest AUC, whereas twist of 8.0° provided the highest sensitivity of 78% (p = 0.005). Integrating 3D STE mechanics with structural metrics further improved classification performance. For provokable LVOTO, the AUC increased from 0.75 to 0.84 (DeLong’s test p = 0.003), with specificity rising from 63% to 78%. Similarly, the AUC improved from 0.76 to 0.82 (DeLong’s test p = 0.013) for resting LVOTO, with sensitivity increasing from 68% to 86% (Fig. 4).

Table 5.

ROC results of variables for identifying resting and provokable LVOTO

Variates	Provokable LVOTO			Resting LVOTO
	AUC (95% CI)	P	Thresholds (Sen/ Spe)	AUC (95% CI)	P	Thresholds (Sen/ Spe)
LVOTD, mm	0.71(0.62,0.79)	<0.001	16.5(55%,72%)	0.68(0.60,0.75)	<0.001	16.5(65%,64%)
SubMV Score	0.60(0.50,0.70)	0.059	0.5(70%,44%)	0.67(0.60,0.72)	<0.001	1.5(36%,90%)
GAS, -%	0.68(0.58,0.78)	<0.001	25.5(75%,58%)	0.60(0.52,0.68)	0.011	27.5(67%,54%)
Twist, °	0.72(0.62,0.81)	<0.001	10.9(61%,78%)	0.62(0.54,0.69)	0.005	8.0(78%,44%)
Torsion, °/cm	0.72(0.62,0.82)	<0.001	2.4(60%,81%)	0.66(0.59,0.74)	<0.001	2.7(57%,71%)
Combined metrics
Structural metrics	0.75(0.68,0.83)	<0.001	0.80(77%,63%)	0.76(0.70,0.82)	<0.001	0.56(68%,73%)
3DE mechanics	0.75(0.66,0.85)	<0.001	0.83(68%,75%)	0.67(0.57,0.76)	<0.001	0.43(78%,50%)
Structure + 3DE	0.84(0.76,0.91)	<0.001	0.83(78%,78%)	0.82(0.76,0.87)	<0.001	0.41(86%,63%)

Footnote: LVOTO: left ventricular outflow tract obstruction, AUC: area under the curve, Sen: sensitivity, Spe: specificity, CI: confidence intervals, LVOTD: left ventricular outflow tract diameter, GAS: global area strain

Fig. 4.

The increased value of 3DE mechanics for differentiating LVOTO indicated by ROC curves. The left panel shows the ROC curves for provokable LVOTO, while the right panel shows those for resting LVOTO. The blue curves represent the ROC when 3DE mechanics are incorporated with structural metrics, and the yellow curves represent the ROC with established structural parameters. The area under the curve (AUC: [95CI]) values are provided in the legends for each condition. A higher AUC indicates better classified performance. LVOTO: left ventricular outflow tract obstruction, 3DE: three-dimensional echocardiography, ROC: receiver-operating characteristic, AUC: area under the curve

The SubMV score demonstrated excellent reproducibility, with an ICC of 0.929 (95% CI: 0.831–0.971). Minor discrepancies were observed in three cases, including two with subtle papillary muscle apical displacement and one with mild hypertrophy of papillary muscle. Bland-Altman plots (Fig S3) and ICCs confirmed good to excellent reproducibility of 3DE strain measurements, with ICCs ranging from 0.887 to 0.984 (Table S2).

Discussion

To our knowledge, this study is among the first to comprehensively characterize 3D STE myocardial mechanics in relation to LVOTO, specifically in HCM patients with moderate septal hypertrophy (WT ≤ 18 mm). Three key insights emerged: (1) both resting and provokable LVOTO exhibited significantly augmented multidirectional myocardial mechanics compared with the nonobstructive group, with LV torsional mechanics demonstrating the strongest discriminatory capacity. (2) Selective mechanics were independently associated with LVOT gradients beyond the established structural parameters, with nonlinear, threshold-dependent torsion–LVOTO relationships, especially under provocation. (3) LV torsional mechanics outperformed structural metrics in detecting provocable LVOTO and combining 3D STE mechanics with anatomy further improved discrimination.

This study primarily extends the current understanding of the multifactorial determinants of LVOTO in patients with moderately hypertrophic HCM. First, among the structural parameters, LVOTD demonstrated the strongest independent predictive value, with superior diagnostic performance compared to that of WT alone. This observation indicates that, in patients with moderate hypertrophy, systolic narrowing of the LVOT may be more relevant to obstruction severity than absolute myocardial thickness. Second, we developed a composite SubMV score that integrates multiple mitral subvalvular structural features. This quantitative approach demonstrated a strong discriminatory capacity across LVOTO phenotypes and remained independently associated with both resting and provoked LVOT gradients. Prior studies have consistently reported associations between subvalvular papillary muscle anomalies, such as hypertrophy, bifurcation, apical displacement, and anomalous insertion etc. and LVOTO, primarily through their effects on mitral–LVOT spatial relationships and systolic leaflet motion and trajectory [22, 23, 27, 28]. However, most previous assessments relied on the binary classification of individual abnormalities, which may insufficiently capture the cumulative burden of subvalvular remodeling in contributing to SAM and dynamic obstruction. Although individual components may exert heterogeneous hemodynamic effects, they are equally weighted to ensure feasibility and reproducibility. Among the individual components, apical displacement and bifid morphology of papillary muscle were the most frequently observed. However, their relatively lower prevalence compared with prior reports likely reflects the exclusion of apical and midventricular obstructive phenotypes, in which such abnormalities are more common [29, 30]. These findings support the relevance of integrated quantified subvalvular assessment in explaining the marked heterogeneity of LVOTO manifestations in patients with moderate hypertrophic HCM.

Most importantly, this study demonstrated a robust and independent association between augmented myocardial mechanics assessed by 3D STE and both the presence and severity of LVOTO in patients with moderate hypertrophic HCM. In contrast to prior studies that primarily linked impaired GLS to myocardial injury without stratification by hypertrophy severity [18, 31], we specifically examined a subgroup in which LVOTO could not be fully explained by hypertrophy alone, thereby often underling therapeutic uncertainty. In this cohort, despite preserved or mildly elevated EF, LV circumferential and rotational mechanics significantly increased while mildly despaired longitudinal strain, with these features being most pronounced in patients with LVOTO. Notably, exaggerated myocardial mechanics remained significantly associated with LVOT gradients beyond the established structural parameters, particularly under provocation. These findings are consistent with prior CMR feature-tracking studies demonstrating selectively elevated strain parameters in LVOTO independent of myocardial hypertrophy or fibrosis [14, 15], as well as 2D-STE evidence highlighting the contributory role of increased LV twist in provocable obstruction [16]. Extending these observations, our nonlinear analyses revealed threshold-dependent relationships between torsional mechanics and the risk of LVOTO. Mechanistically, hypercontractility in HCM represents both an early, mutation-driven phenotype related to specific genes [32] and also a compensatory role enhancing ejection efficiency, as observed in endurance training–induced LV twist augmentation [33]. Increased LV torsion reflects amplified apical–basal shear deformation that facilitates systolic ejection [17]; however, excessive rotational deformation over a specific critical thresholds, may pathologically distort LVOT geometry and promote flow convergence, thereby amplifying dynamic obstruction and potentially reinforcing hypercontractile behavior. In parallel, elevated GAS, an integrative metric of longitudinal and circumferential deformation [26], likely reflects enhanced centripetal myocardial shortening accompanying these rotational changes.

Morphologically, patients with obstruction in our cohort more frequently exhibited a sigmoid septal configuration with basal bulging and relatively limited myocardial fibrosis. In contrast, patients with nonobstructive HCM exhibit a reverse-curved phenotype with greater focal fibrosis and, in some cases, elevated BNP levels, suggesting more advanced myocardial remodeling [15, 18, 34]. Although derived from consecutive screenings, selection bias cannot be fully excluded in this tertiary referral setting, where asymptomatic, nonobstructive patients with early-stage moderate hypertrophy are less likely to undergo advanced imaging and may be underrepresented. Accordingly, our findings reflect the phenotypic spectrum of clinically referred populations rather than that of all individuals with moderate HCM. These observations support the dynamic concept of LVOTO, indicating that its absence does not necessarily reflect a latent or early disease stage [35, 36] and underscore the relevance of myocardial mechanical alterations, particularly in nonobstructive patients.

By capturing true volumetric motion, 3DE avoids geometric assumptions and reduces the out-of-plane tracking loss inherent to 2DE [18], allowing a more robust assessment of multidirectional deformation and rotational mechanics, an essential advantage in the marked heterogeneous myocardial morphology of HCM. Prior studies have shown strong agreement between 3DE-derived strain, 2DE strain, and CMR-based deformation metrics, with 3DE offering a shorter analysis time [37] and good reproducibility, as demonstrated in our cohort. Given its wide clinical availability, 3DE is well-suited for large real-world HCM populations in which access to CMR may be limited. In the context of emerging therapies such as cardiac myosin inhibitors, targeting myocardial hypercontractility while avoiding excessive hypocontractility [38, 39] Routine integration of 3D STE mechanical indices may provide incremental value for refined patient phenotyping, including identification of appropriate candidates for targeted intervention by cutoffs and longitudinal surveillance of patients at risk of functional “run-out,” even avoiding unnecessary mitral interventions when combining with SubMV score system, thereby refining individualized stratification management in HCM patients with moderately hypertrophy.

Limitations

This study has several limitations. First, it was conducted at a single tertiary center, and although 3DE datasets were acquired prospectively, all analyses were retrospective in nature, which may limit the generalizability of our findings. Second, the study population was restricted to patients with moderate septal hypertrophy, while those with severe hypertrophy or apical/mid-ventricular obstruction were excluded. In addition, genetic heterogeneity across HCM subtypes was not considered, and only 150 of 216 patients underwent CMR and diffuse fibrosis indices were not included, which limited comprehensive characterization of myocardial fibrosis. These factors may restrict the applicability and interpretability of our results across the broader HCM spectrum. Third, the SubMV score, a novel semi-quantitative index designed to capture mitral subvalvular anomalies, did not incorporate other potentially relevant structural features such as chordal elongation, posterior mitral leaflet length, or annular displacement. Moreover, the equal weighting of individual components may not fully reflect their differential hemodynamic contributions to LVOTO. Fourth, LVOTO is a highly dynamic phenomenon influenced by loading conditions, including afterload and heterogeneous physiological stressors. In addition, 3DE-derived strain analysis remains dependent on image quality, and strain measurements may vary across vendors, software versions and even sonographers. These physiological and technical factors inevitably limit the generalizability of specific mechanical thresholds for broader clinical application in HCM. Therefore, prospective, multicenter, and longitudinal studies are warranted to validate these findings and to clarify the temporal relationships between exaggerated myocardial mechanics and the development or progression of LVOTO.

Conclusions

In patients with moderate hypertrophic cardiomyopathy, 3D STE -derived hypercontractile mechanics, particularly LV torsional mechanics, are associated with both the presence and severity of LVOTO, independent of established structural parameters, including anatomical narrowing and cumulative papillary muscle abnormalities. These findings help explain the heterogeneity of LVOTO phenotypes in this population. The observed nonlinear, threshold-dependent relationships between myocardial mechanics and LVOTO provide mechanistic insight into the dynamic interaction between myocardial contractility and LVOT hemodynamics. Integration of 3D STE analysis into assessment and longitudinal monitoring procedure potentially refines risk stratification in this heterogeneous HCM population.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

2/3DE

2/3-dimensional echocardiography

AUC

Area under the curve

AML

Anterior mitral leaflet

BSA

Body surface area

CI

Confidence interval

CMR

Cardiac magnetic resonance

EDV/EDVI

End-diastolic volume/ end-diastolic volume index

EF

Ejection fraction

ESC

European Society of Cardiology

ESV/ESVI

End-systolic volume/ end-systolic volume index

GAS

Global area strain

GCS

Global circumferential strain

GLS

Global longitudinal strain

GRS

Global radial strain

HCM

Hypertrophic cardiomyopathy

ICC

Intraclass correlation coefficients

LAVI

Left atrium volume index

LGE

Late gadolinium enhancement

LV

Left ventricular

LVOTO

Left ventricular outflow tract obstruction

MR

Mitral valve regurgitation

MV

Mitral valve

OR

Odds ratio

ROC

Receiver operating characteristic

RCS

Restricted cubic splines

SAM

Systolic anterior motion

WT

Wall thickness

Author contributions

All authors made substantial contributions to this work. Yuwei Bao, Wei Zhou, and Jie Tian were primarily responsible for study conception, study design, and the acquisition and analysis of three-dimensional echocardiographic data. Jeffrey B. Geske, Si Fang, Shiliang Liu, Liming Xia, and Youbin Deng contributed to the interpretation of clinical findings and provided critical intellectual input during data analysis and manuscript revision. The first draft of the manuscript was prepared by Yuwei Bao, and all authors contributed to subsequent critical revisions for important intellectual content. Yani Liu, as the corresponding author, supervised the overall project, provided methodological oversight, and ensured the integrity of the analyses and the manuscript. All authors read and approved the final version of the manuscript and agree to be accountable for all aspects of the work.

Funding

This work was supported by National Natural Science Foundation of China (82472010 and 82272109), Cardiovascular Ultrasound Innovation Team of Yunnan Province (202305AS350021).

Data availability

Datasets and study materials (e.g., echocardiographic protocols, analysis templates) from this study are available from the corresponding author upon reasonable request. Data are securely stored at Tongji hospital in line with relevant regulations.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Tongji Medical College (Approval No.: 2022-S013-(1–4)) and complied with the latest Declaration of Helsinki (2013, Fortaleza, Brazil). Informed consent was obtained from all individual participants included in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.