Predictive value of estimated plasma volume for postoperative hypotension in percutaneous intramyocardial septal radiofrequency ablation treating for hypertrophic obstructive cardiomyopathy

We performed this retrospective analysis (NCT06003478 [22/08/2023]) of patients prospectively enrolled in the registry trials of PIMSRA (NCT02888132 & NCT04355260) and the primary results were published earlier [3, 4]. Data on consecutive patients who were diagnosed with HOCM followed by the 2023 ESC Guidelines for the management of cardiomyopathies and had undergone PIMSRA at the Xijing Hospital between October 2016 and December 2022 were retrospectively collected. We excluded 3 patients with preoperative hypotension (mean systolic blood pressure (SBP) < 90 mmHg preoperatively measured by ambulatory blood pressure monitoring). Informed consent for the procedure and clinical research was obtained from all individual participants before procedure. Informed consent for their data to be used for research was obtained from all patients before they received PIMSRA. This retrospective analysis was based on anonymous data and the exception to the requirement of informed consent was approved by the Institutional Ethics Committee of the Xijing Hospital. Research procedures followed the amended Helsinki Declaration of 1975. The study was approved by the Institutional Ethics Committee of the Xijing Hospital.

Information of medical history was collected from the admission note and the family history was collected from the latest follow-up records. Laboratory blood tests, transthoracic echocardiogram examinations, 12-lead electrocardiograms, computed tomography angiography, ambulatory blood pressure monitoring, and Holter electrocardiographic monitoring were performed perioperatively. The blood test was performed within one week before PIMSRA by our research protocol. Baseline blood pressures and heart rates were recorded using daytime mean blood pressures preoperatively obtained from ambulatory blood pressure monitoring, and missing data were filled by results of routine blood pressure measurement at admission. Mitral regurgitation and systolic anterior motion of the mitral valve were evaluated by echocardiogram following the established classification methods [15, 16]. The results were collected from the Hospital Information System of Xijing Hospital. ePVS was calculated using hematocrit and hemoglobin from the Strauss-derived Duarte formula as follows [9]:

Information of procedure duration time was acquired from the operation note and calculated as total working time of ablation.

A consensus statement on postoperative blood pressure was provided in the Perioperative Quality Improvement (POQI) 3 document, which pointed out that a systolic pressure of less than 90 mmHg is likely to put the majority of patients at risk of end-organ damage [17]. This criterion is also consistent with the definitions used by the majority of researchers in earlier investigations of perioperative hypotension [18]. Based on these, our study defined patients with postoperative hypotension (POH) as follows: regardless of the dosage of vasopressor agents utilized, recording at least once the SBP less than 90 mmHg obtained by invasive or noninvasive blood pressure measurements, with parallel hypotension-related symptoms such as dizziness, nausea, fatigue, palpitations, or syncope, which necessitate adjustment to the dose of vasopressor agents. The Hospital Information System and postoperative care records were analyzed to obtain data of the patient's blood pressure, symptoms, and medicine usage.

Categorical variables were expressed as frequencies (percentages). Several continuous variables cannot be considered to obey the normal distribution in this study so continuous variables were expressed as median (interquartile range, IQR). Logarithmic transformation was applied to N-terminal pro-brain natriuretic peptide (NT-proBNP) concentrations. Missing values were interpolated by multiple interpolation methods. In detail, we used the “mice” package to perform multiple interpolations with the method of predictive-mean matching to produce five imputed data sets. The subsequent analysis is based on a pooled analysis of the results of the imputed data sets. Comparisons of demographic, clinical, biological, and echocardiographic parameters among patients with or without POH were analyzed using Chi-square tests for categorical variables and a two-sided t-test or Kruskal–Wallis test for continuous variables. The above analyses were also performed in the groups with lower or higher ePVS. Logistic regression models were used to obtain unadjusted and covariate-adjusted odds ratios (ORs) between ePVS and postoperative hypotension. Covariates used for adjustment included age, sex, NYHA class, baseline log-BNP, baseline medication, creatinine which were used to evaluate the renal function and time duration of the procedure. In addition, variables with significant differences between patients with and without POH were used as candidates for another multivariable logistic regression which was performed using a global optimization search. To determine the cut-off value of ePVS for prediction of the POH risk, receiver operating characteristic (ROC) curve analysis was performed. Effects of ePVS on POH were assessed using both continuous and categorical variables which were transformed by the quartiles and best threshold determined by ROC curve analysis. The forecast capacities of each threshold were calculated. As a sensitive analysis, we also calculated the ePVS derived from the Hakim-Kaplan formula using the measured weight instead of dry weight, which was calculated as follows:

The 405 patients had a median (IQR) age of 49 (39~58) years and a total of 150 (37.0%) were women. 243 (60.0%) patients were with severe HF symptoms and diagnosed as NYHA class III or IV. 30 (7.5%) of all patients were taking diuretics and 17 (4.2%) had previously undergone SRT. The median (IQR) of SBP was 116 (109~126) mmHg (Table 1). There were 53 (13.1%) patients suffering from POH. Between patients with or without POH, there were statistical differences in the distribution of sex, BMI, hypertension, and history of syncope or presyncope.

All patients who underwent PIMSRA were measured with baseline median (IQR) LVOT pressure gradient of 78 (50~107) mmHg at rest and 123 (92~156) mmHg at provoked peak. Several characteristics measured by echocardiogram and results of lab examinations were significantly different between patients with or without postoperative hypotension. After adjusted with body surface area, the left ventricular volume indexes, both in the end-diastolic and systolic stages, were smaller in patients with POH. There was no significant difference in the degree of myocardial hypertrophy (Table 2).

The distribution of ePVS levels in the 405 patients who underwent PIMSRA is shown in Fig. 1. The median (IQR) of ePVS in overall patients was 3.77 (3.27 ~ 4.40) mL/g. The median (IQR) of ePVS in patients with POH was 4.40 (3.85 ~ 5.01) mL/g, which was significantly higher than those without POH of 3.70 (3.23 ~ 4.26) mL/g (Table 2).

Among the 405 patients included in this study, 53 (13.1%) patients were observed with symptomatic postoperative hypotension. By univariable Logistic regression analysis, higher ePVS was found to have a significant association with the risk of POH, with the odds ratio (OR) of 1.669 (95% confidence interval [CI] 1.299 ~ 2.144) per mL/g (Supplement Table). After being adjusted by potential confounders, ePVS remained independently associated with POH. In other words, patients with high ePVS exhibited an increased risk for POH (Fig. 2).

Using ePVS to predict POH, the area under the ROC curve (Fig. 3) achieved 0.698 (95% CI, 0.622 to 0.774). It was worth emphasizing that using the best cutoff value, the ePVS achieved an appreciable predictive performance with a sensitivity of 0.811 and a specificity of 0.528.

Using several thresholds including the best cut-off value determined by ROC curve analysis and quartiles, ePVS was transformed into categorical variables which showed different classification performances as shown in Table 3. Using the lower quartile, the sensitivity achieved 92.5%, and instead, using the upper quartile, the specialty achieved 78.7%.

Using the ePVS derived from the Hakim-Kaplan formula instead of the Duarte formula, the ePVS showed concordant estimations in univariable and multivariable logistic regression analysis (Table 4). The ePVS using two formulas demonstrated an approximate ROC curve and the two AUROCs were not statistically significant (Fig. 4).