Multidomain Arterial Afterload and Left Atrial–Left Ventricular Coupling in Hypertensive Phenotypes

ABSTRACT

Brachial blood pressure may not fully reflect the multidimensional arterial load associated with left atrial–left ventricular (LA–LV) coupling. In this cross‐sectional study, we examined local carotid pulse wave velocity (PWV), cuff‐derived arterial pressure–volume index (API), and arterial velocity–pulse index (AVI), separately and as a composite afterload index (ACI), in 336 inpatients (71 controls, 183 with hypertension, and 82 with hypertension and type 2 diabetes). Covariate‐adjusted linear models related standardized exposures to the left atrial stiffness index (LASI) and LA–LV coupling index (LACI). Greater ACI was associated with higher LACI (β = 0.0142 per 1‐SD, 95% CI 0.0011–0.0273; p = 0.033). The LASI estimate was directionally similar but less precise (β = 0.0232, 95% CI −0.0024–0.0488; p = 0.076), and all single‐domain CIs included zero. These findings support further evaluation of accessible multidomain vascular assessment as a complement to brachial blood pressure. External calibration and longitudinal validation are needed to establish the clinical value of ACI.

1. Introduction

Brachial blood pressure alone may not fully capture the multidimensional arterial load imposed on the heart. Arterial stiffness and wave transmission determine when and how systolic pressure reaches the left ventricle (LV), meaning that comparable cuff pressures can accompany different myocardial loading patterns. In hypertension, prominent late‐systolic load has been associated with impaired left atrial (LA) function [1]. Ventricular—arterial stiffening and LA myopathy are also important features of heart failure with preserved ejection fraction, in which abnormal LA mechanics carry clinical and prognostic information [2, 3, 4]. Early vascular–cardiac dysfunction may therefore be present before it is apparent from brachial pressure alone.

Noninvasive vascular measures provide different views of this underlying load. Carotid–femoral pulse wave velocity (PWV) is the reference measure of aortic stiffness [5], whereas local carotid PWV reflects segmental carotid stiffness and wave propagation [6, 7, 8]. The cuff‐derived arterial pressure–volume index (API) and arterial velocity–pulse index (AVI) describe peripheral pressure–volume behavior and waveform augmentation [9, 10]. Central arterial stiffness has been associated with impaired LA mechanics independently of systolic blood pressure and LV structure [11]. Because these measures are related but not interchangeable, no single measure may fully capture the combined afterload burden.

In this study, LA–LV coupling was operationalized using complementary pressure–strain and volumetric indices. The LA stiffness index (LASI) combines estimated LV filling pressure with LA reservoir strain and has been proposed as a marker of early hypertension‐mediated cardiac injury [12]. Pulsatile arterial hemodynamics provide a mechanistic link between vascular load and cardiac dysfunction [13], while the LA–LV coupling index (LACI) provides a volumetric measure associated with diastolic dysfunction [14]. We evaluated local carotid PWV, API, and AVI separately and as a composite afterload index (ACI). We hypothesized that higher ACI would be associated with higher LASI and LACI. Mutually adjusted models examined whether individual domains retained residual associations, while additional analyses assessed the explanatory contribution beyond brachial blood pressure and possible modification by type 2 diabetes.

2. Methods

2.1. Study Design and Setting

This single‐center cross‐sectional study included consecutive adult inpatients admitted to the cardiology service between October 2024 and June 2025. Vascular assessments (PWV, API, and AVI) and transthoracic echocardiography, including speckle‐tracking strain and 3D volumetry, were performed during the same hospitalization and, whenever feasible, on the same day. Of the 403 individuals assessed, 67 were excluded for objective data‐quality or completeness reasons. Fifty‐three had vascular or echocardiographic recordings that were inadequate, nonstandardized, or unsuitable for reliable quantification. Fourteen did not have key variables required for the primary analysis. The remaining 336 participants formed the complete‐case analytic cohort. The sample size was determined by the number of eligible participants enrolled during the study period. No a priori sample‐size calculation was performed. The study complied with the Declaration of Helsinki and was approved by the Ethics Committee of the Affiliated Hospital of Guizhou Medical University (approval no. 2023 Ethics Review No. 279). All participants provided written informed consent.

2.2. Participants

Eligible participants were adults aged 18–85 years. HTN and T2D were identified from a documented diagnosis, condition‐specific treatment, or a diagnosis established during the index hospitalization. Participants were categorized as controls without HTN or T2D, HTN without T2D, or HTN with T2D (HTN+T2D). A T2D‐only phenotype was not included in the study design.

Cardiac exclusion criteria were persistent atrial fibrillation or another rhythm precluding reliable strain analysis; prior myocardial infarction; LV ejection fraction <50%; moderate‐to‐severe valvular disease; dilated or hypertrophic cardiomyopathy; or congenital heart disease. Patients with known pulmonary arterial hypertension or severe pulmonary disease were also excluded. Technical exclusions related to data quality and completeness were applied as described above. Participant flow is shown in Figure 1.

FIGURE 1.

Participant selection and formation of the analytic cohort. Abbreviations: ACI, composite afterload index; HTN, hypertension; LACI, left atrial–left ventricular coupling index; LASI, left atrial stiffness index; T2D, type 2 diabetes.

2.3. Vascular Measurements

Local carotid PWV was measured in the left common carotid artery, 1–2 cm proximal to the bifurcation, using a Mindray Resona R9 ultrasound system with the Holo‐PWV module (Mindray, China). Participants were examined at rest in the supine position. Five consecutive 1–4‐second recordings were obtained from the same straight arterial segment with a stable probe position. The device automatically reported the mean of the five recordings.

API and AVI were measured using a PASESA AVE‐2000 PRO system (Shisei Datum, Japan) as cuff‐derived indices of peripheral pressure–volume behavior and waveform morphology. Brachial blood pressure and heart rate were recorded simultaneously from the left upper arm in a quiet resting state. Three measurements were obtained and averaged.

2.4. Echocardiography

Transthoracic echocardiography was performed using a GE Vivid E95 system and analyzed offline with EchoPAC software (GE Healthcare). A stable electrocardiographic signal was used for cardiac phase identification. The R‐wave peak defined end‐diastole, the end of the T wave defined end‐systole, and the onset of the P wave defined the start of atrial contraction. Three consecutive sinus cycles were acquired. When occasional premature beats occurred, only sinus cycles with adequate image quality were analyzed. At least five cycles were recorded for spectral Doppler and M‐mode measurements. Two‐dimensional images were acquired at frame rates of 40–80 frames/s. For calculation of E/e′, e′ was defined as the mean of the septal and lateral mitral annular early‐diastolic velocities. LA reservoir strain (LASr) was measured in the apical four‐ and two‐chamber views. The final value was averaged across both views and three cardiac cycles. Three consecutive single‐beat, high‐volume‐rate full‐volume 4D datasets were acquired, with each dataset covering one complete cardiac cycle. The dataset with the best image quality was selected for volume analysis. 3D‐derived LAVmin and 3D LVEDV were measured using dedicated 4D LA and 4D LV analysis software on the GE platform. Echocardiographic analyses were performed without knowledge of the vascular measurements or clinical group.

2.5. Exposure and Outcomes

The primary exposures were local carotid PWV, API, and AVI. For participant i, each exposure X was standardized as:

$$z_{X,i}=\frac{X_i-\overline{X}}{s_X}$$

The composite afterload index was then calculated as:

$$AC{I}_i=\frac{z_{\mathrm{PWV},i}+z_{\mathrm{API},i}+z_{\mathrm{AVI},i}}{3}$$

Here, $\overline{X}$ and $(s_X)$ denote the cohort‐specific mean and SD, respectively. ACI represents relative multidomain afterload burden within this cohort rather than a fixed clinical scale. External use would require re‐standardization or calibration in an independent population.

The prespecified outcomes were:

and

$$LACI=\frac{3DLA{V}_{min}}{3DLVEDV}$$

Higher values indicated less favorable LA–LV coupling.

2.6. Statistical Analysis

Baseline characteristics were summarized overall and by group. Continuous variables were compared using one‐way ANOVA or Kruskal–Wallis tests, and categorical variables using χ² or Fisher's exact tests, as appropriate. Primary associations were examined in separate linear models for PWV, API, AVI, and ACI. Single‐domain models estimated the overall association of each measure with the outcomes, whereas the ACI model evaluated combined multidomain burden. All exposures, including ACI, were standardized before regression; coefficients are reported per 1‐SD increase. Models used HC3 robust standard errors and adjusted for age using a natural cubic spline (df = 4), sex, body surface area, heart rate, LV mass index, and clinical group.

Mutually adjusted models included PWV, API, and AVI simultaneously and estimated the residual association of each domain conditional on the other two. Collinearity was assessed using variance inflation factors and a condition index. Restricted cubic splines examined nonlinearity, with knots at the 10th, 50th, and 90th percentiles; overall and nonlinearity p values were obtained from nested‐model F‐tests. Because pulse pressure may lie downstream of arterial stiffness, PP‐anchored models were treated as contextual sensitivity analyses rather than estimates of direct effects.

Sensitivity analyses included standardized PCA‐derived PC1, an age‐overlap subset (40–70 years), and models further adjusted for hypertension duration, diabetes duration, and four antihypertensive medication classes: renin–angiotensin system inhibitors, calcium‐channel blockers, diuretics, and beta‐blockers. Among participants with hypertension, an ACI×T2D interaction term assessed modification by T2D. Pairwise interaction and subgroup analyses were exploratory and were evaluated using Benjamini–Hochberg false‐discovery‐rate correction within the relevant analysis families. Additional ACI‐tertile and bootstrap procedures are described in the Supplementary Methods.

Analyses used complete cases without imputation. Missingness and exclusion counts are reported in Tables S5–S6, and available characteristics of included and excluded participants are compared in Table S7. Analyses were conducted in Python 3.11.2; package versions, scripts, and machine‐readable outputs are provided in the submitted code archive. All tests were two‐sided, with P<0.05 indicating nominal statistical significance. Interpretation also considered effect size, 95% CI, multiplicity, and analysis hierarchy.

2.7. Intra‐Observer Repeatability

To assess intra‐observer repeatability, the same operator reanalyzed stored image datasets from 20 participants at least 2 weeks apart and was blinded to the first‐session results. Endocardial borders were retraced, and speckle tracking was repeated de novo. Agreement was assessed using ICC(A,1) with 95% CIs, Bland–Altman analysis, and the within‐subject coefficient of variation.

3. Results

Of 403 individuals assessed, 336 were included: 71 controls, 183 participants with HTN, and 82 with HTN+T2D. Participant flow and baseline characteristics are shown in Figure 1 and Table 1, respectively. Exclusion and missingness details are provided in Tables S5–S7.

TABLE 1.

Baseline characteristics of the complete‐case cohort.

Characteristic	Control (n = 71)	HTN (n = 183)	HTN + T2D (n = 82)	p‐value
Demographic and clinical characteristics
Age, years	50.0 ± 13.5	61.4 ± 9.8	61.7 ± 9.2	<0.001
Male sex, n (%)	22 (31.0)	75 (41.0)	41 (50.0)	0.058
Body surface area, m2	1.60 ± 0.15	1.66 ± 0.16	1.68 ± 0.15	0.004
Systolic blood pressure, mmHg	112.0 ± 10.4	136.9 ± 19.9	137.1 ± 19.0	<0.001
Diastolic blood pressure, mmHg	71.0 ± 8.8	78.2 ± 14.1	78.7 ± 13.0	<0.001
Pulse pressure, mmHg	41.2 ± 11.2	58.7 ± 18.1	58.2 ± 14.4	<0.001
Heart rate, beats/min	78.5 ± 12.2	75.6 ± 11.8	80.3 ± 11.5	0.008
Vascular measures
Local carotid PWV, m/s	5.85 ± 1.32	7.78 ± 1.76	7.49 ± 1.53	<0.001
API	24.5 ± 4.3	31.8 ± 8.5	30.4 ± 6.2	<0.001
AVI	15.8 ± 4.6	21.8 ± 7.4	20.9 ± 5.6	<0.001
ACI, mean z score	−0.72 ± 0.45	0.24 ± 0.73	0.08 ± 0.57	<0.001
Cardiac measures
LV mass index, g/m2	57.6 ± 10.7	72.9 ± 16.1	74.0 ± 17.0	<0.001
Relative wall thickness	0.26 ± 0.05	0.33 ± 0.08	0.34 ± 0.08	<0.001
LASI	0.225 (0.178–0.289)	0.377 (0.299–0.499)	0.358 (0.296–0.484)	<0.001
LACI	0.186 (0.152–0.202)	0.244 (0.189–0.312)	0.238 (0.179–0.304)	<0.001
Disease duration and medication use
Hypertension duration, years	—	3.00 (0.03–10.00)	6.00 (1.25–10.75)	0.002
Diabetes duration, years	—	—	7.00 (1.00–12.50)	—
RAAS inhibitor use, n (%)	0 (0.0)	44 (24.0)	21 (25.6)	<0.001
Calcium‐channel blocker use, n (%)	0 (0.0)	81 (44.3)	41 (50.0)	<0.001
Beta‐blocker use, n (%)	0 (0.0)	22 (12.0)	8 (9.8)	0.010
Diuretic use, n (%)	0 (0.0)	9 (4.9)	4 (4.9)	0.146
Insulin use, n (%)	0 (0.0)	0 (0.0)	19 (23.2)	<0.001
Metformin use, n (%)	0 (0.0)	0 (0.0)	39 (47.6)	<0.001
SGLT2 inhibitor or GLP‐1 receptor agonist use, n (%)	0 (0.0)	0 (0.0)	15 (18.3)	<0.001

Notes: Data are mean ± SD, median (IQR), or n (%).

Disease durations are shown only for applicable phenotypes.

Abbreviations: ACI, composite afterload index; API, arterial pressure–volume index; AVI, arterial velocity–pulse index; HTN, hypertension; LACI, left atrial–left ventricular coupling index; LASI, left atrial stiffness index; PWV, pulse wave velocity; T2D, type 2 diabetes.

Intra‐observer repeatability. Agreement was excellent for LACI (ICC[A,1] = 0.940, 95% CI 0.822–0.980) and LASI (ICC[A,1] = 0.957, 95% CI 0.831–0.991). Bland–Altman plots and component‐measure results are shown in Figure S3 and Table S3.

Primary associations. Higher ACI was associated with higher LACI (β = 0.0142 per 1‐SD, 95% CI 0.0011–0.0273; p = 0.033). The estimate for LASI was positive but less precise (β = 0.0232, 95% CI −0.0024–0.0488; p = 0.076). All single‐domain 95% CIs included zero (Table 2 and Figure 2). Adding ACI to a clinical model that included brachial systolic and diastolic blood pressure increased adjusted R² by 0.011 for LACI and 0.008 for LASI, indicating modest incremental explanatory value (Table S14).

TABLE 2.

Primary linear associations of afterload domains and composite burden with LASI and LACI.

Outcome	Exposure	Model	N	β per 1‐SD (95% CI)	p‐value
LASI	PWV	Single‐domain	336	0.0185 (−0.0025 to 0.0394)	0.084
LASI	API	Single‐domain	336	0.0131 (−0.0106 to 0.0369)	0.279
LASI	AVI	Single‐domain	336	0.0092 (−0.0108 to 0.0292)	0.367
LASI	ACI	Composite	336	0.0232 (−0.0024 to 0.0488)	0.076
LACI	PWV	Single‐domain	336	0.0094 (−0.0036 to 0.0223)	0.157
LACI	API	Single‐domain	336	0.0048 (−0.0069 to 0.0165)	0.424
LACI	AVI	Single‐domain	336	0.0103 (−0.0028 to 0.0235)	0.123
LACI	ACI	Composite	336	0.0142 (0.0011 to 0.0273)	0.033

Notes: Models used HC3 robust standard errors and adjusted for age using a natural cubic spline (df = 4), sex, body surface area, heart rate, LV mass index, and clinical group.

Coefficients are reported per 1‐SD increase.

Abbreviations: ACI, composite afterload index; API, arterial pressure–volume index; AVI, arterial velocity–pulse index; CI, confidence interval; LACI, left atrial–left ventricular coupling index; LASI, left atrial stiffness index; PWV, pulse wave velocity.

FIGURE 2.

Primary associations of afterload domains and composite burden with LASI and LACI.

Secondary and sensitivity analyses. PC1 yielded estimates quantitatively similar to those of ACI (Table 3 and Table S12). In mutually adjusted models, all domain‐specific 95% CIs included zero. VIFs were 1.15–1.23, and the condition index was 1.63, indicating limited residual domain‐specific information without evidence of severe numerical multicollinearity (Tables S1 and S13). Restricted cubic splines showed no evidence of nonlinearity over the observed range for LASI (p = 0.857) or LACI (p = 0.557). PP‐anchored models were interpreted only as contextual sensitivity analyses. Estimates remained directionally consistent after additional adjustment for disease duration and antihypertensive medication classes (Table 3 and Tables S10 and S12). Among participants with HTN (n = 265), there was no clear evidence that T2D modified the ACI association with LASI (P for interaction = 0.214) or LACI (P for interaction = 0.205) (Table 3 and Table S10).

TABLE 3.

Summary of key secondary and sensitivity analyses.

Analysis	LASI	LACI	Interpretation
PC1 alternative composite	β = 0.0233 (95% CI −0.0023 to 0.0490); p = 0.075	β = 0.0141 (95% CI 0.0010 to 0.0272); p = 0.034	Estimates closely matched ACI
RCS nonlinearity	P for nonlinearity = 0.857	P for nonlinearity = 0.557	No evidence of nonlinearity
Mutually adjusted domains	All domain‐specific 95% CIs included zero	All domain‐specific 95% CIs included zero	Limited residual domain‐specific information
ACI + pulse pressure	β = 0.0296; p = 0.046	β = 0.0185; p = 0.033	Contextual sensitivity analysis
ACI + disease duration	β = 0.0245; p = 0.061	β = 0.0150; p = 0.026	Similar to primary estimates
ACI + duration and medication classes	β = 0.0271; p = 0.035	β = 0.0157; p = 0.022	Similar to primary estimates
ACI × T2D among participants with HTN	Interaction β = 0.0379; p = 0.214	Interaction β = −0.0190; p = 0.205	No clear evidence of effect modification

Note: Values are β per 1‐SD unless stated. ACI, composite afterload index; BH FDR, Benjamini–Hochberg false‐discovery rate; HTN, hypertension; PC1, first principal component; PP, pulse pressure; RCS, restricted cubic spline; T2D, type 2 diabetes. The API×AVI interaction for LASI did not remain significant after BH FDR correction (nominal p = 0.039; q = 0.235). Sensitivity analyses were not used to upgrade the primary evidence.

Exploratory analyses. The API×AVI interaction for LASI did not remain significant after Benjamini–Hochberg false‐discovery‐rate (BH FDR) correction (nominal p = 0.039; q = 0.235; Figure S1A and Tables S2 and S4). In the age‐overlap subset (n = 256), the ACI–LACI estimate remained positive (Table S12). The AVI–LACI association did not remain significant after BH FDR correction (nominal p = 0.012; q = 0.073; Figure S1B and Tables S4 and S12). ACI‐tertile results are shown in Figures S4–S5 and Table S9. Bootstrap estimates were centered near the primary estimates and are reported in Figures S7–S8 and Table S8.

4. Discussion

This study examined whether accessible afterload surrogates provide separate or combined information about LA–LV coupling. The principal finding was that greater ACI was associated with higher LACI, whereas the LASI estimate was directionally similar but less precise. Single‐domain associations were weaker, and each domain contributed limited residual information when the three measures were modeled together. PC1 yielded results similar to those of ACI. Together, these findings show a modest association between integrated afterload burden and LACI, whereas the single‐domain estimates were less precise.

Hypertension is an important upstream contributor to HFpEF [15]. Current frameworks emphasize that diastolic dysfunction reflects the interaction of ventricular mechanics, vascular aging, and time‐varying arterial load [16, 17, 18, 19, 20, 21]. Previous studies have linked late‐systolic loading to LA dysfunction in hypertension [1], related arterial stiffness to subsequent cardiac remodeling in young adults with T2D [22], and identified distinct arterial waveform clusters [23]. Atrial dysfunction is also an important HFpEF phenotype, and atrial strain may assist in detecting hypertensive HFpEF [24, 25]. Our study extends this work by integrating local carotid stiffness and cuff‐derived waveform indices with LA strain and 3D volumetry. This design links accessible vascular phenotyping with complementary pressure—strain and volumetric measures of LA–LV coupling. Emerging reference data for LASI and LACI may further support interpretation of these coupling phenotypes [26].

ACI represents relative multidomain burden within the study population. PC1 explained 55.1% of the variance in the three measures and was almost perfectly correlated with ACI. The similar PC1 and ACI estimates support the stability of the integrated signal across two composite approaches. Mutually adjusted models addressed whether any domain retained a distinct association after accounting for the other two. Low VIFs indicated that severe numerical multicollinearity was unlikely. Each domain nevertheless contributed limited residual explanatory information once the other two were included.

Secondary analyses reinforced the overall pattern. Because PP may partly mediate stiffness‐related load, PP‐anchored models were interpreted as contextual analyses. Their larger ACI coefficients and minimal change in model fit did not alter the primary interpretation. The association was adequately represented by a linear model over the observed range. Its direction was maintained after adjustment for disease duration and antihypertensive medication classes, and no clear modification by T2D was detected. The API×AVI interaction did not remain significant after BH FDR correction. These findings extend broader evidence linking arterial stiffness to heart‐failure risk and shared vascular–metabolic pathways [27, 28, 29] while supporting continued validation of device‐specific vascular measurements [30].

Clinically, local carotid PWV can be obtained during routine vascular ultrasound and characterizes segmental stiffness in a proximal elastic conduit artery, whereas API and AVI can be acquired rapidly using an upper‐arm cuff assessment [6, 7, 8, 9, 10]. Their complementary accessibility provides a practical basis for multidomain vascular phenotyping. Greater ACI was associated with higher LACI and provided modest incremental explanatory value beyond clinical variables and brachial blood pressure. This supports ACI as an integrative research phenotype and motivates external calibration against established vascular‐aging and HFpEF frameworks [31, 32, 33]. Previous studies support the prognostic relevance of LACI and LASI [14, 34, 35], while echocardiographic LACI has also shown diagnostic value in HFpEF [36]. More broadly, ventricular—arterial coupling remains relevant to functional HF phenotyping [37, 38]. With longitudinal and external validation, automated or AI‐assisted models could integrate accessible vascular measures with LA–LV coupling indices to identify patients who may benefit from further cardiac assessment.

4.1. Limitations

Several limitations should be considered. The cross‐sectional design precludes temporal or causal inference. This was a single‐center inpatient study based on complete cases; individuals with greater clinical or imaging complexity may therefore be underrepresented. Residual confounding by disease duration, treatment intensity, metabolic or renal status, and unmeasured disease severity remains possible. Because no T2D‐only group was included, we could not distinguish the association of T2D alone with LA–LV coupling from that observed when T2D coexisted with hypertension. Local carotid PWV is a device‐dependent segmental measure and is not interchangeable with carotid–femoral PWV, while API and AVI are cuff‐derived surrogates of vascular load. ACI is standardized within the study cohort and has no fixed external reference or clinical cutoff. The sample size also limited the precision of interaction and nonlinear analyses, and the findings may not generalize to community‐based or untreated populations. Multicenter prospective studies are required for external calibration and clinical validation.

5. Conclusion

We found that greater multidomain arterial afterload burden, summarized by ACI, was associated with higher LACI. The LASI estimate was similar in direction but less precise. No single afterload measure retained a clear residual association when all three were modeled together. These findings support further evaluation of integrated vascular assessment as a complement to brachial blood pressure. External calibration should test whether ACI adds information beyond brachial blood pressure in community cohorts. Longitudinal studies should assess its ability to predict incident HFpEF and detect treatment‐related changes before clinical application.

Author Contributions

Yiran Luo and Lina Zhao contributed equally to this work. Sha Li: investigation, resources, and writing – review and editing. Fangfang Lu: investigation, resources, and writing – review and editing. Yiran Luo: conceptualization, methodology, data curation, formal analysis, visualization, writing – original draft, and writing – review and editing. Bei Zhang: supervision, conceptualization, methodology, project administration, funding acquisition, and writing – review and editing. Lina Zhao: methodology, investigation, data curation, validation, visualization, and writing – review and editing. Yueyao Zhao: investigation, data curation, and writing – review and editing.

Consent

The authors have nothing to report.

Funding

This work was supported by the National Natural Science Foundation of China (grant nos. 81960315 and 82360354); the Guizhou Provincial Science and Technology Department (grant nos. qiankehejichu‐ZK[2022]yiban359 and qiankehejichu‐ZK[2024]zhongdian045); and the Doctoral Research Startup Fund Project of the Affiliated Hospital of Guizhou Medical University (grant no. gyfybsky‐2026‐01). The funders had no role in study design; data collection, analysis, or interpretation; decision to publish; or preparation of the manuscript.

Conflicts of Interest

The authors report no relationships that could be construed as a conflicts of interest.

Supporting information

Supporting Information: jch70368‐sup‐0001‐SuppMat.docx

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.