Comparison of left ventricle ejection fraction by echocardiography and cardiac magnetic resonance imaging in day to day clinical practice

Introduction

Left ventricle ejection fraction (LVEF) is the most commonly used marker of LV function. It generally does not change over a short period of time in the absence of a major cardiovascular event or medication change. Cardiac magnetic resonance imaging (CMRI) is considered gold standard for LVEF assessment, however echocardiography is more widely available and most commonly used tool to assess LVEF.

Purpose

To evaluate the differences in clinically relevant LVEF ranges as measured by echocardiography and CMRI and to assess the reproducibility of CMRI LVEF calculations of experienced CMRI physicians compared to minimally-trained medical residents.

Methods

One hundred consecutive patients who underwent both echocardiography and CMRI from 9/01/05 to 6/18/07 at a community hospital were retrospectively reviewed. Echocardiography LVEF was derived by standard techniques. CMRI LVEF measurements were derived by standard techniques, utilizing short axis stack of TRUFISP cine images over the entire LV volume with volumetric analysis on the ARGUS work station. LVEF cutoffs for major clinical decision making were defined as <30% (prophylactic ICD implantation) and <50% (diagnosis of cardiomyopathy and discontinuation of cardiotoxic chemotherapy agents), and were the basis for comparisons. Mean time frame between the echocardiogram and CMRI was 15.8 days (range 14–49 days). CMRI LVEF calculated by two internal medicine residents blinded to patient information and original CMRI was compared with experienced CMRI physician calculated LVEF. Residents received limited training in selecting appropriate CMRI images and working with ARGUS workstation. We compared physician read echocardiography and CMRI LVEF, as well as physician read and resident read CMRI derived LVEF.

Results

Physician read echocardiography and CMRI LVEF comparisons divided subjects into 4 groups. Group 1 had LVEF <30% on echocardiography (n = 16). Of these, 8 (50%) had corresponding LVEF by CMRI as >30% with differences of 5–10% in 1 subject (6%) and >10% in 7 subjects (44%). Group 2 had LVEF <30% on CMRI (n = 9). Of these, 1 subject (5%) had corresponding LVEF by echocardiography >30%, with a difference of >10%. Group 3 had LVEF ≥ 30 to <50% on echocardiography (n = 22). Of these, 10 (45%) had corresponding LVEF by CMRI as >50% with differences of 5–10% in 4 subjects (18%) and >10% in 6 subjects (27%). Group 4 had LVEF ≥ 30 to <50% on CMRI (n = 27). Of these, 5 subjects (18%) had corresponding LVEF by echocardiography as > 50% with differences of 0–10% in 2 subjects (7%) and >10% in 3 subjects (11%). When LVEF was assessed as <30% by either tool, the discordance was 36%; when assessed as ≥ 30 to <50% by either tool, the discordance was 35%.

Physician read and resident read CMRI LVEF comparisons divided subjects into 2 groups. Group A had physician calculated CMRI LVEF <30% (n = 9). Of these, 8 (89%) had corresponding LVEF by resident calculated CMRI as <30%. The one subject with a divergent LVEF was over read by the resident by 10.4%. Group B had LVEF ≥ 30 to <50% on CMRI (n = 27). Of these, 17 (63%) had corresponding LVEF by resident read CMRI as ≥ 30% to < 50% with differences of < 5% in 6 subjects (22%) and > 5% in 4 subjects (15%). Overall discordance between physician and resident calculated CMRI LVEF for <30% was 11%. When LVEF was assessed as ≥ 30 to <50%, the discordance was 37% (22% with a difference of <5%; 15% with a difference of >5%).

Conclusion

Prior to making major management decisions based on echocardiography derived LVEF in clinically relevant ranges, CMRI confirmation is strongly suggested, particularly in patients with suboptimal images. These results also suggest that the analysis of CMRI data to derive accurate LVEF can be performed without extensive training and is reproducible.