A systematic search was performed in PubMed-Medline, EMBASE, the Cochrane Library and Web of Science databases using the terms “hypertrophic obstructive cardiomyopathy”, “idiopathic hypertrophic sub-aortic stenosis”, “asymmetric septal hypertrophy”, “septal reduction therapy”, “septal myectomy”, “Morrow septal myectomy”, “modified morrow septal myectomy”, “alcohol septal ablation” and “percutaneous transluminal myocardial ablation”. No time limit to the start date was applied, and the search was conducted up to October 2020. The detailed search strategies are presented in Supplementary Table 1. The inclusion criteria included (a) studies comparing the outcomes of ASA and SM; (b) enrolled patients $\ge$18 years; (c) published language restricted to English or a complete English translation version; and (d) follow-up the studies were more than 30 days.

Two examiners (WT, ML) independently screened the titles and abstracts (if available) of the entries identified in different databases. Next, the full text of all studies that met the eligibility criteria or those with insufficient information from the titles or abstracts to make a decision, were obtained for the next screening phase. All studies that did not meet the criteria were excluded, and the reasons for exclusion were noted. Case reports and series, review articles, editorials and duplicate reports were excluded. For those studies that reported the same study or utilized the repeated data at different follow-up intervals, we pooled all the relevant details together and used the most comprehensive data for further analysis. A third member (JL) would further check the data whenever there was disagreement until a consensus was reached. Data from the included studies were then extracted by two independent reviewers (WT, ML) based on a predesigned outline. The extracted information included the design of the study, study population (number of participants, age, and sex), length of follow-up, clinical characteristics and outcomes such as the pre- and post-procedure left ventricular outflow tract pressure gradient, sustained ventricular tachycardia or ventricular fibrillation (VT/VF) events (including appropriate implantable cardioverter defibrillator (ICD) intervention) during and post-procedure, SCD and resuscitated sudden cardiac arrest (SCA) events post-procedure, the ICD implantation rate and permanent pacemaker (PPM) implantation rate post procedure, the reintervention rate and all-cause or cardiac mortality post procedure. Additional details are presented in Table 1 (Ref. [10, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31]). VF/VT events were divided into an early-phase (which occurred during the procedure, during hospitalization or within 30 days after the procedure) and a late-phase (which occurred $>$30 days after the procedure and out of hospital) subgroup.

Since all of the included studies were observational, the assessment of the risk of bias was evaluated by a modified version of the Newcastle-Ottawa scale (NOS), which is a quality assessment tool for nonrandomized studies in three domains: the selection of participants, comparability of study groups, and the outcome of interest. The risk of bias in each study was evaluated by calculating the aggregate score on the 9 items. The detailed assessment of each study is shown in Supplementary Table 2.

The data analysis was conducted by Review Manager 5.4 (The Cochrane Collaboration, Oxford, England) and Stata (version 15.1, StataCorp, College Station, TX, USA). Continuous variables were reported as the means $\pm$ standard deviation (SD) if they were normally distributed; otherwise, they were reported as medians and interquartile ranges (IQRs). The pooled effects are presented as the relative ratio (RR) or weighted mean difference (WMD) with a 95% confidence interval (CI), and sensitivity analyses were performed when significant heterogeneity was observed. The heterogeneity between studies was considered significant with an I${}^2$ of more than 50% and a p value less than 0.05. The meta-analyses were performed with the fixed model when the heterogeneity between studies was not significant; otherwise, the randomized effect model was used, and sensitivity analyses were also needed. Publication bias was evaluated using funnel plots and Egger’s test. p 0.05 was considered statistically significant.

A total of 1185 articles were initially retrieved from PubMed, Embase, the Web of Science, and the Cochrane Library. After removing duplicates, 580 articles were left for title and abstract review. Thirty-five full-text articles were further assessed for eligibility, and 20 studies were ultimately included for data extraction and analysis [10, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31] (Fig. 1). All studies were observational studies. The risk of bias was evaluated by the Newcastle Ottawa Scale (Supplementary Table 2). Among them, 5 studies acquired 6/9 points, 12 studies acquired 7/9 points, and the remaining 3 studies acquired 8/9 points.

Fig. 1.

Flow chart of literature search.

The study characteristics are presented in Table 1. A total of 8025 patients were included in the 20 studies. Among them, 3860 patients received ASA treatment, and the remaining 4165 patients underwent SM. The mean follow-up periods varied from 3 months to 120 months. Of the 20 enrolled studies, twelve studies had a specific description of VT/VF events (including appropriate ICD intervention), eight studies contained descriptions about SCD and resuscitated SCA, and ten studies collected data about ICD implantation. The baseline LVOT pressure gradient was higher in the ASA groups (WMD = 5.89; 95% CI: 3.03–8.75; p 0.0001; I${}^2$ = 1%, Supplementary Fig. 1a). The baseline interventricular septal diameter (IVSd) in diastole was slightly thinner in the ASA groups (WMD = –0.68; 95% CI: –1.35, –0.02; p = 0.04; I${}^2$ = 43%, Supplementary Fig. 2a), while the number of patients in NYHA class III/IV and the left ventricular end-diastolic diameter (LVEDd) prior to the intervention were similar between the two groups (p = 0.36, p = 0.51, Supplementary Figs. 1b,2b).

Among the 20 enrolled studies, 12 studies contained descriptions of VT/VF events (including ICD intervention). The pooled analysis showed that the incidence of total VT/VF events was almost twice as high in the ASA group (345/3312, 10.42%) than in the SM group (61/3227 patients, 4.99%) (RR = 1.98; 95% CI: 1.65–2.37; p 0.0001; I${}^2$ = 0%, Fig. 2a). When VT/VF events were classified as early-phase and late-phase, the data showed that more than 90% of VT/VF events occurred in the early-phase in both groups (ASA: 94.20%; SM: 94.41%). Further subgroup analysis indicated that VT/VF was significantly higher in the ASA group than in the SM group in both the early phase (RR = 1.94; 95% CI: 1.61–2.33; p 0.0001; I${}^2$ = 0%, Fig. 2b) and the late phase (RR = 2.80; 95% CI: 1.00–7.89; p = 0.05; I${}^2$ = 33%, Fig. 2c). The sensitivity analysis of the enrolled studies demonstrated that the removal of each of them did not change the result of the pooled analysis (Supplementary Table 3). In addition, possible publication bias was not found in the funnel plot (Supplementary Fig. 3) or Egger’s test (p = 0.101). Further meta regression showed no significant interaction between the incidence of VT/VF with LVOT pressure gradient reduction (p = 0.904/0.220), with baseline ejection fraction (EF) (p = 0.552/0.685), with baseline IVSd (p = 0.799/0.054), and with baseline NYHA class III/IV proportion (p = 0.165/0.364) in both ASA and SM cohorts (Supplementary Figs. 5–8).

Fig. 2.

Comparisons of ventricular fibrillation (VT)/ventricular tachycardia (VF)/appropriate ICD intervention events between ASA groups and SM groups. (a) Total VF/VT events. (b) Early-phase VF/VT events. (c) Late-phase VF/VT events.

The SCD and resuscitated SCA were combined as SCA to estimate the risk for SCD together in this analysis. In the 8 studies that described SCD and/or resuscitated SCA events, a total of 49 events were reported in the ASA group (SCD: 26; resuscitated SCA: 23), and 17 events were reported in the SM group (SCD: 13; resuscitated SCA: 4). The pooled analysis showed that the ASA group had a higher incidence of SCA (RR = 2.30; 95% CI: 1.35–3.94; p = 0.002; I${}^2$ = 0%, Fig. 3a). However, when SCD was considered alone, the pooled analysis showed no significant difference between the two groups (ASA cohorts: 28/824, 3.40%; SM cohorts: 13/730, 1.8%; RR = 1.70; 95% CI: 0.90–3.18; p = 0.10; I${}^2$ = 0%, Fig. 3b). Possible publication bias was also not found in the funnel plot (Supplementary Fig. 4) or Egger’s test (p = 0.667).

Fig. 3.

Comparison of the incidence of sudden cardiac death (SCD)/resuscitated sudden cardiac arrest (SCA) after ASA and SM. (a) Combined SCD/resuscitated SCA events. (b) SCD events alone.

All-cause mortality within 30 days and above 30 days after receiving ASA or SM was regarded as early-phase and late-phase mortality, respectively. For the ASA cohorts, there were 38 early-phase deaths and 89 late-phase deaths among 3361 (1.13%) and 1055 patients (8.43%), respectively, while in the SM cohorts, there were 75 early-phase deaths and 88 late-phase deaths among 3520 (2.13%) and 1304 patients (6.75%), respectively. The pooled analysis showed no significant differences between ASA and SM in either early-phase mortality (RR = 0.72, 95% CI: 0.37–1.41; p = 0.34; I${}^2$ = 31%) or late-phase all-cause mortality (RR = 1.04, 95% CI: 0.63–1.69; p = 0.89; I${}^2$ = 39%) (Table 2).

After ASA, 404 patients were implanted with a permanent pacemaker in the 3474 pooled patients (11.63%), which was significantly higher compared with (240/3864, 6.21%) after SM (RR = 1.99; 95% CI: 1.39–2.83; p = 0.0002; I${}^2$ = 50%; Table 2). In addition, the pooled analysis indicated that patients receiving ICD implantation were not significantly different between the two cohorts (ASA group: 123/1208, 10.18%; SM group: 89/1589, 5.60%; RR = 1.67; 95% CI: 0.98–2.86; p = 0.06; I${}^2$ = 64%, Table 2). The reintervention rate was significantly higher in the ASA group than in the SM group (ASA vs SM: 11.28% vs 0.56%; RR = 10.50; 95% CI: 5.10–21.64; p 0.0001; I${}^2$ = 0%; Table 2).