Hypertrophic cardiomyopathy (HCM)

HCM is the most common heritable cardiac condition and is the most common cause of sudden cardiac arrest in the young, affecting about 1 in 500 people. Familial hypertrophic cardiomyopathy is a heart condition characterized by thickening (hypertrophy) of the heart muscle, more specifically the ventricle. The thickened heart muscle, can make it challenging to keep up with the oxygenation demands of the body and the heart muscle itself. Many people with HCM have few, if any, symptoms and can lead normal lives without significant symptoms. However, this condition can also have serious consequences. Life threatening arrhythmias resulting in cardiac arrest can sometimes be the first symptom.

Symptoms include: Shortness of breath, Chest pain, Fainting, Palpitations

Implantable Cardioverter-Defibrillator (ICD) ⚡

Previous work

O’Mahony et al. (2014)

Problem

• Hypertrophic cardiomyopathy (HCM) is a leading cause of sudden cardiac death (SCD) in young adults. Current risk algorithms provide only a crude estimate of risk and fail to account for the different effect size of individual risk factors.

• The aim of this study was to develop and validate a new SCD risk prediction model that provides individualized risk estimates.

Data

Results

• The prognostic model was derived from a retrospective, multi-centre longitudinal cohort study. The model was developed from the entire data set using the Cox proportional hazards model and internally validated using bootstrapping.

• The cohort consisted of 3675 patients from six centres. During a follow-up period of 24,313 patient-years (median 5.7 years), 198 patients (5%) died suddenly or had an appropriate ICD shock.

• Of eight pre-specified predictors, associated with SCD/appropriate ICD shock at the 15% significance level age, maximal left ventricular wall thickness, left atrial diameter, left ventricular outflow tract gradient, family history of SCD, non-sustained ventricular tachycardia, and unexplained syncope

• final model to estimate individual probabilities of SCD at 5 years.

  • For every 16 ICDs implanted in patients with ≥4% 5-year SCD risk, potentially 1 patient will be saved from SCD at 5 years.

Standard Failure Probability

Define the Prognostic Index (PI) using the coefficients from the standard Cox model

\[ \begin{align*} \text{PI}_{\text{Naive}} &= 0.159 \times \text{Maximal wall thickness} \\ &\quad - 0.003 \times \text{Maximal wall thickness}^2 \\ &\quad + 0.026 \times \text{Left atrial diameter} \\ &\quad + 0.004 \times \text{Maximal LVOT gradient} \\ &\quad + 0.458 \times \text{Family history SCD} \\ &\quad + 0.826 \times \text{NSVT} \\ &\quad + 0.717 \times \text{Unexplained syncope} \\ &\quad - 0.018 \times \text{Age} \end{align*} \]

Impact on Decision-Making

Green et al. (2024)

Aims

• To conduct a contemporary cost-effectiveness analysis examining the use of implantable cardioverter defibrillators (ICDs) for primary prevention in patients with hypertrophic cardiomyopathy (HCM)

Methods

• A discrete-time Markov model was used to determine the cost-effectiveness of different ICD decision-making rules for implantation.

• Several scenarios were investigated, including the reference scenario of implantation rates according to observed real-world practice.

• A 12-year time horizon with an annual cycle length was used.

• Transition probabilities used in the model were obtained using Bayesian analysis.

Markov Model

Input Data

Transition Probability Matrix

\[ \begin{pmatrix} p_{11}^s & p_{12}^s & 0 & 0 & p_{15}^s \\ 1 - p_{15}^s & 0 & 0 & 0 & p_{15}^s \\ 0 & 0 & p_{33} & p_{34}^s & p_{35}^s \\ 0 & 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 0 & 1 \end{pmatrix} \]

Model Equations

Denote \(x\) as the observed number of transitions, \(p\) the probability of a transition and \(n\) as the total number of transitions from a given state. The hyperparameters \(\alpha\) characterise the prior knowledge on \(p\). Superscripts indicate the decision rule used.

DAG

State Transition Probability Posterior Distributions

Posterior Predictive Checks

Trace Plots

Cost-Effectiveness Analysis Plots

Deterministic Sensitivity Analysis (DSA)

Global First-Order Variance-Based Probability Sensitivity Analysis

Cost-Effectiveness Acceptability Curves for ICD Costs

Subcutaneous ICD (S-ICD)

• Structural Model Extension: Adapt the existing Markov framework to differentiate between Transvenous (TV-ICD) and Subcutaneous (S-ICD) devices.

• Differential Cost-Effectiveness: Evaluate the economic trade-off between the higher upfront capital costs of S-ICDs against the potential long-term savings.

• Subgroup Stratification: Younger cohorts?

Markov Models with Device Replacement

Expected Value of Partial Perfect Information (EVPPI)

From Model Uncertainty to Decision Uncertainty

• Transition to EVPPI: Shift the focus from global variance sensitivity to the Expected Value of Partial Perfect Information (EVPPI).
• Isolate Decision Drivers: Identify which specific clinical parameters (e.g., LV wall thickness coefficients, baseline hazards) actively cause patients to cross the 6% 5-year risk threshold and alter the ICD implantation decision.
• Quantify Research Value: Calculate the maximum expected Net Monetary Benefit (NMB) of resolving uncertainty for these high-impact parameters to prioritize future clinical data collection.

Expected Value of Partial Perfect Information (EVPPI)

For a specific parameter of interest (\(X_i\)) is calculated as:

• \(X_i\): The specific parameter of interest (e.g., LV wall thickness coefficient).
• \(\boldsymbol{X}_{\sim i}\): All other uncertain parameters in the model.
• \(d \in \{0, 1\}\): The binary clinical decision (Implant ICD).
• \(\text{NMB}\): Net Monetary Benefit.

Cumulative Incidence Functions (CIFs)

Conclusions

• Individualized Risk Stratification: The novel risk prediction model provides individualized 5-year probabilities for Sudden Cardiac Death, offering a more precise tool for clinical decision-making than previous crude guidelines.

• Robust Economic Evaluation: By embedding this prediction model within a Bayesian discrete-time Markov framework, we can formally quantify the cost-effectiveness of different ICD implantation thresholds under real-world uncertainty.

• Future Work:

  • Shifting from model uncertainty to decision uncertainty (EVPPI) will provide a more rigorous foundation for prioritizing future data collection
  • State-of-the-art device evaluation (e.g., S-ICDs).