Impact of cardiac magnetic resonance imaging on outcomes and downstream therapeutic management in heart failure: a propensity matched big data cohort study

With the prevalence of heart failure (HF) rising, it is unclear how cardiac magnetic resonance imaging (CMR) utilization impacts management and outcomes for HF populations.

To assess whether CMR referral correlates with improved outcomes (mortality, MACE, all-cause hospitalizations) in the US HF population.

A retrospective multicenter cohort study was performed using TriNetX, a global electronic medical records database, for adults age > 18 years and ICD-10 coding for HF (I50). HF was categorized as HFrEF (HF with reduced ejection fraction or EF), HFmrEF (HF with mildly reduced EF), and HFpEF (HF with preserved EF). Subjects were divided into CMR and non-CMR cohorts based on CMR referral 1 year before or after receiving the HF diagnosis, and propensity matched for characteristics prior to ICD-10 diagnosis: age, race, sex, hypertension, diabetes, chronic kidney disease, atrial fibrillation/flutter, coronary artery disease, hyperlipidemia, obesity, guideline-directed medical therapy (GDMT), valvular surgery, coronary intervention, rheumatoid arthritis, sarcoidosis, systemic lupus erythematosus, COVID-19, hemochromatosis, C reactive protein, brain natriuretic peptide (BNP), and troponin T. The primary outcomes included: all-cause mortality, major adverse cardiac events (MACE), and all-cause hospitalizations. Odds of GDMT initiation, pacemaker/implantable cardioverter-defibrillator placement, coronary intervention, and valvular surgery were compared in the CMR vs non-CMR cohorts. Outcome were assessed at 10 years after HF diagnosis.

Prior to propensity matching, 2,967,257 patients with HF and CMR were identified, and 81,239 with HF without CMR. Patient cohorts with HF classification based on EF included: HFrEF (n=9,097 with MRI; n=193,943 without), HFmrEF (n=3,609 with MRI; n=180,026 without) and HFpEF (n=6,532 with MRI, n=470,605 without). After propensity matching, the following cohorts emerged: n=80,458 for any HF, n=8,919 for HFrEF, n=3,607 for HFmrEF, and n=6,530 for HFpEF with and without MRI. 10-year mortality risk for HF overall was 13.34% vs 19.31% (p<0.0001), for HFrEF 12.86% vs 20.07% (p<0.0001), for HFmrEF 11.31% vs 16.95% (p<0.0001), and for HFpEF 14.09% vs 18.20% (p<0.0001) in the CMR vs no CMR cohorts, respectively. MACE was also lower in the CMR cohorts: for HF overall 25.05% vs 31.47% (p<0.0001), for HFrEF 32.18% vs 36.42% (p=0.0003), for HFmrEF 23.39% vs 27.83% (p=0.0002), and for HFpEF 25.15% vs 28.69% (p=0.0008). Odds of GDMT/procedural utilization was overall significantly higher in the CMR cohorts (figures 1 and 2).

CMR utilization is exceedingly low in patients diagnosed with heart failure (2.7%). In this propensity matched big data study, CMR is associated with significantly improved outcomes across the HF classification spectrum and has the potential to guide optimal therapy for HF patients. HF populations may benefit from increased access to care of CMR.Figure 1