1. Goal of the Study:
The goal of this prospective observational study is to develop and validate a novel, non-invasive method for predicting the prognosis of patients with light-chain cardiac amyloidosis (AL-CA). This method integrates advanced multi-modal imaging techniques and artificial intelligence (radiomics) to provide early and accurate assessment of treatment response and survival outcomes.
2. Main Question:
Can a multi-modal radiomics model, based on the fusion of \[¹⁸F\]FAPI PET/CT (assessing fibroblast activation) and 3D Cardiac MRI (CMR) (assessing structural damage) imaging data, accurately predict 12-month all-cause mortality and dynamically track disease progression in patients with AL-CA receiving standard care?
3. Participants:
Population: Patients diagnosed with AL-CA (confirmed by endomyocardial biopsy or extracardiac biopsy plus specific cardiac criteria: NT-proBNP \>332 pg/mL, mean left ventricular wall thickness \>12 mm, excluding hypertension/other causes).
Setting: Single-center study at Beijing Anzhen Hospital, Capital Medical University.
Number: 49 patients (calculated sample size accounting for dropouts).
Key Criteria:
Inclusion: Confirmed AL-CA diagnosis, receiving standard AL-CA treatment (chemotherapy e.g., Daratumumab-based regimen + supportive cardiac care).
Exclusion: Active infection, advanced malignancy (life expectancy \<12 months), severe cognitive impairment/immobility affecting imaging compliance/follow-up.
4. Study Design \& Procedures:
Design: Single-center prospective cohort study.
Intervention: Participants receive standard-of-care treatment for AL-CA as per guidelines (chemotherapy regimen based on Daratumumab, Bortezomib, Cyclophosphamide, Dexamethasone; tailored cardiac support including diuretics, rate control, anticoagulation if needed).
Procedures:
Baseline: Upon enrollment, participants undergo comprehensive assessment: \[¹⁸F\]FAPI PET/CT scan, 3D CMR scan, blood tests (NT-proBNP, troponin, free light chains, etc.), clinical staging (Mayo 2012), functional assessment (NYHA class), quality of life questionnaire (KCCQ).
Imaging: Specialized software (Siemens True D) performs cross-platform fusion of PET/CT and 3D CMR images. Radiomics features are extracted from the fused images using dedicated software (Siemens FeAture Explorer).
Follow-up:
Clinical: Every 3 months (symptoms, medication adherence, adverse events, lab tests including NT-proBNP).
Imaging: Repeat \[¹⁸F\]FAPI PET/CT and 3D CMR scans at 6 months post-baseline. Radiomics features are extracted again.
Endpoints: Primary endpoint is 12-month all-cause mortality. Secondary endpoints include re-hospitalization rates and changes in NYHA class. Follow-up continues until the 12-month endpoint for all participants.
Data Analysis: Machine learning (LASSO-Cox regression) is used to select key radiomics features from baseline and 6-month scans and integrate them with quantitative imaging parameters (FAPI uptake volume, SUVmax, LGE burden, ECV) and clinical data to build prognostic models predicting 12-month survival.
5. Comparison:
Researchers will compare the predictive performance of the developed multi-modal radiomics model against:
* Traditional clinical biomarkers: NT-proBNP levels and Mayo Clinic staging.
* Standard quantitative imaging parameters alone: Such as myocardial FAPI uptake volume, SUVmax, or CMR-derived extracellular volume (ECV) measured at baseline and 6 months.
The goal is to demonstrate superior accuracy in predicting 12-month all-cause mortality using the integrated radiomics approach.