The impact of hypertension on the prognosis of patients with hypertrophic cardiomyopathy: a single-center retrospective study

Study patients

This retrospective, single-center cohort study was performed at West China Hospital of Sichuan University, which a tertiary center located in Chengdu, China. From December 2008 to November 2018, we included 546 hospitalized patients with a discharge diagnosis of HCM in total. Baseline characteristics were collected from medical records by experienced physicians. Data entry was performed using the twice-entry method. The data would be entered into the database when the values of the two entries were same; otherwise, the raw data would be examined. After reviewing the medical records, nine patients were excluded for other conditions that may cause cardiac hypertrophy, and 69 patients were excluded due to missing baseline data after the first evaluation or loss to follow-up (Fig. 1). Eventually, 468 adult patients in total were included for the present analysis. The Biomedical Research Ethics Committee of West China Hospital of Sichuan University approved the study (approval number: 2019-1147), and the study was conducted according to the principles of the Declaration of Helsinki. Because of the retrospective nature of the study, informed consent was waived. The study has been registered at the Chinese Clinical Trial Registry with a registration number of ChiCTR2000029352; the registration information can also be accessed via International Clinical Trials Registry Platform (https://trialsearch.who.int/default.aspx). Some other detailed information has been reported in the recently published study (Wang, Zhao & He, 2021).

Diagnostic criteria

Diagnosis of HCM was made in accordance with the 2014 European Society of Cardiology guideline criteria (Zamorano et al., 2014), and it was principally supported by the echocardiographic finding of increased left ventricular (LV) wall thickness (> = 15 mm)—that was not solely explained by aberrant loading conditions. Each patient had > =1 of the following disease features of HCM (Maron et al., 2013; Wang, 2019): LV wall thickness > =18 mm (n = 310); systolic anterior motion (SAM) of mitral valve (n = 146); family history of HCM (n = 40); family history of sudden cardiac death (SCD) (n = 15); intervention of obstruction (n = 52); LV apical aneurysm (n = 10); asymmetric septal hypertrophy (n = 333; IVS/LVPW > = 1.5); hyperdynamic LV function (n = 208; LV ejection fraction > = 70%); non-long standing and refractory hypertension (n = 468; prior to admission, blood pressure was controlled within the target range).

The diagnosis of hypertension was made when the resting systolic blood pressure was > = 140 mm Hg and/or the diastolic blood pressure was > = 90 mm Hg, or when there was a history of antihypertensive drug usage (Mancia et al., 2018).

Study outcome

Based on the previous studies (Songsirisuk et al., 2019), the study’s outcome were described as a composite of three common types of HCM-related death and additional types of particular HCM-related death, such as: (1) HF-related death, which was defined as death preceded by signs and/or symptoms of HF lasting more than one hour, in the context of progressive cardiac decompensation more than a year before death (Coats et al., 2013); (2) stroke-related death, which occurred as a result of probable or proven ischemic stroke (Haruki, Minami & Hagiwara, 2016). Here we did not distinguish cardioembolic stroke from other ischemic subtypes in the present study; (3) SCD, which was defined as nocturnal death or witnessed sudden cardiac death within one hour of new symptoms in patients who previously had a relatively stable or uneventful clinical course (Elliott et al., 2000); (4) other specific HCM-related death, which was defined as death due to HCM, but not belonged to the aforementioned three conditions.

Follow-up was carried out via medical records or telephone contact with the patients themselves and/or referring relatives. All patients were followed from the first evaluation up to the endpoint or the most recent evaluation.

Statistical analysis

Frequencies and descriptive statistics were used to summarize patient baseline characteristics for the research population.

Cox proportional-hazards regression models were used to estimate the association between hypertension and the study outcome. Initially, demographic factors, clinical factors, laboratory tests and therapies were included in a multivariable Cox regression model. propensity scores (PS) were developed to account for potential confounding by observed baseline characteristics given the observational nature of the current study (Austin, 2011). With PS methods, more potential confounders can be considered than with traditional regression techniques, which allows for the replacement of the whole set of baseline data with a single composite score. With the use of the multivariable logistic regression model, which took into account all the factors listed in Table 1, the propensities of each individual with hypertension were calculated. Associations between hypertension and the study outcome were then estimated by Cox regression models with the use of multiple propensity-score methods.

In the primary analysis, we used PS matching (PSM), in which 1:1 matching without replacement was carried out using a nearest neighbor matching algorithm with a fixed caliper width of 0.2. Additionally, in order to ensure the reproducibility of the analysis results by PSM, stabilized inverse probability treatment weighting (IPTW) (Austin, 2016) and overlap weighting (Thomas, Li & Pencina, 2020) were also carried out. The overall PS distributional curves and the absolute standardized differences (ASD) for each covariate (an ASD ≥ 0.10 indicates imbalance) were used to assess the covariate differences before and after PSM, as well as IPTW and overlap weighting (Austin, 2016). Then, we presented the Kaplan Meier curves and Cox models that used the above-mentioned PS, and we also showed the Cox model that included PS as an additional covariate. As the sensitivity analyses, we evaluated the relationship between hypertension and some specific HCM-related deaths, such as HF-related death, stroke-related death and SCD.

The statistical analyses were performed with the use of R software, version 4.1.0 (R Core Team, 2021). A two-sided p value of 0.050 was regarded as statistically significant for all statistical analyses.

Baseline Characteristics and study outcome

The study population consisted of 468 HCM patients, including 149 (31.8%) with hypertension and 319 (68.2%) without hypertension. HCM patients with hypertension were older, had lower prevalence of family history of HCM and SCD, had higher chest pain but lower pre-/syncope. More importantly, HCM patients with hypertension had a greater prevalence of prior thromboembolism, arial fibrillation and diabetes. Other drug applications except beta-blockers and warfarin were higher in hypertension group, while the intervention of obstruction and devices therapy were higher in non-hypertension group. In addition, HCM patients with hypertension tended to have larger LV end-diastolic diameter, higher prevalence of left ventricular outflow tract obstruction (LVOTO), lower prevalence of LV apical aneurysm and systolic anterior motion of the mitral valve. The relatively thinner interventricular septum (IVS) and thicker left ventricular posterior wall (LVPW) resulted in a lower IVS/LVPW ratio in hypertension group. Other detailed information is shown in Table 1.

During a follow-up time of 2170.7 person-years (PYs) (median, 4.6 years; IQR, 2.1–6.8 years), a total of 55 HCM-related death occurred with a mortality rate of 2.53 (95% confidence interval [CI]: 1.87–3.19) per 100 PYs. The specific causes of deaths were as follows: 28 HF-related deaths, 10 stroke-related deaths, 15 SCDs, two HCM-related postoperative deaths.

Bivariable analysis

In bivariable analysis (unweighted sample), the HCM-related death rates were 2.75 (95% CI [1.92–3.58]) and 2.06 (95% CI [0.99–3.13]) per 100 PYs in the non-hypertension and hypertension groups, respectively. Kaplan–Meier curves demonstrated that the cumulative HCM-related death rate did not significantly differ between the two groups (Fig. 2A). We used the function, namely cox.zph, from the package “survival” to test the proportional hazard assumption for hypertension, which was not violated (p = 0.550). Univariate Cox regression analysis indicated that there was no significant difference regarding HCM-related death between the two groups (HR = 0.74; 95% CI [0.40–1.36]; p value: 0.329). After adjusting potential confounding factors, HR for HCM-related death in hypertension group versus non-hypertension group is 0.77 (95% CI [0.35–1.71], p value: 0.521).

Propensity score matching analysis

Results of multivariable logistic regression analysis that predicted hypertension are listed in Table 2, and the C-statistic of the model was 0.918. The distribution of PS in the hypertension and non-hypertension groups before matching is shown in Fig. 3A. The lesser overlap of PS curves of the two groups indicated a greater risk of confounding. With the use of PSM, 71 HCM patients with hypertension were matched with 71 HCM patients without hypertension. After PSM, PS curves for hypertension and non-hypertension groups were superimposed, indicating well-balanced baseline covariates between the two groups (Fig. 3B). Few baseline variables remained unbalanced with ASD ≥ 0.1 were added as covariates in the multivariable Cox regression analysis (Fig. 3E). Kaplan–Meier analysis also demonstrated that the cumulative HCM-related death rate did not significantly differ between the two groups after matching with log rank p at 0.84 (Fig. 2B). Furthermore, in the primary analysis with PSM, both univariable and multivariable Cox regression analysis indicated that hypertension was not significantly associated with HCM-related death with HR at 0.90 (95% CI [0.34–2.41]; p value: 0.838) and 0.77 (95% CI [0.35–1.71]; p value: 0.521), respectively (Table 3).

ITPW and overlap weighting analyses

After IPTW and overlap weighting, as compared to the crude cohort, the baseline differences between the hypertension group and non-hypertension group were largely reduced (Figs. 3C and 3D). The remained unbalanced variables were further adjusted in the multivariable Cox regression analysis as well (Fig. 3E). Kaplan–Meier curves indicated that the cumulative HCM-related death was comparable between the two groups after IPTW and overlap weighting with log rank p at 0.887 and 0.842, respectively (Figs. 2C and 2D). No strong evidence for an association was observed between hypertension and HCM-related death in both univariate and multivariate Cox regression models in IPTW and overlap weighting cohort (Table 3).

Sensitivity analysis

As a sensitivity analysis, we explored the association between hypertension and some specific HCM-related death, which included HF-related death, stroke-related death and SCD. In crude analysis, no significant difference was found between the two groups for HF-related death (HR = 0.71; 95% CI [0.30–1.67]; p value: 0.434), stroke-related death (HR = 1.55; 95% CI [0.44–5.51]; p value: 0.498) and SCD (HR = 0.53; 95% CI [0.15–1.89]; p value: 0.330). In the primary analysis with PSM, unadjusted HRs of hypertension for HF-related death, stroke-related death and SCD were 0.87 (95% CI [0.22–3.49]; p value: 0.846), 0.96 (95% CI [0.13–6.89]; p value: 0.967) and 0.91 (95% CI [0.13–6.49]; p value: 0.926), respectively. The results remained unchanged with further adjustment of age and sex in the PSM cohort. Additionally, multivariable analysis and other PS analyses unanimously demonstrated that hypertension was not significantly associated with those specific HCM-related death (Table 4).