Impact of ROI placement on H/CL ratio in imaging cardiac amyloidosis with PYP.

Objectives: Amyloid is a condition in which misfolded proteins form deposits that can accumulate in the heart and other organs. It is important to clinically identify and differentiate AL from ATTR cardiac amyloidosis because they require different management strategies. Until recently the standard method to identify and differentiate the type of cardiac amyloidosis required an endomyocardial biopsy (EMB). ^99mTc-PYP imaging has emerged as a non-invasive test capable of detecting ATTR and differentiating it from AL-related and non-amyloid heart failure; currently based on a calculated heart-to-contralateral (H/CL) ratio of 1.3 in our practice. The purpose of this study was to evaluate the impact of region of interest (ROI) size and shape on the calculated H/CL ratio used for diagnosing cardiac amyloidosis on planar PYP scans.

Methods: This IRB, HIPPA compliant, retrospective study included patients who underwent ^99mTc-PYP imaging for cardiac amyloidosis from 1/1/2013 to 9/1/2016. Patients were included if they had an EMB confirming ATTR amyloid presence or absence, or biopsy elsewhere confirming amyloid and echocardiographic findings consistent with cardiac amyloid, including increased left ventricular wall thickness in the absence of other causes, diastolic dysfunction, and characteristic abnormalities of strain imaging. Patients were excluded if they had spinal surgery within the past 4 weeks, scoliosis, or incidental micro-amyloid deposits detected on biopsy performed for other reasons. Out of 178 consecutive patients, 107 patients met our inclusion criteria. 2D planar images, acquired for 5 minutes, were performed 15 minutes (blood pool) and 3 hours (delayed) after injection of Tc-99m PYP. These images were evaluated utilizing five ROI techniques. ROI₁ was a small 50 mm circle contained within the myocardium. ROI₂ was drawn free-hand following the outline of the myocardium. ROI₃ encompassed the entire heart in an oval shape including some background. ROI₄ used the same oval ROI but moved it lateral to the right atria of the heart. ROI₅ used the same oval, but excluded adjacent osseous structures by free-hand modifications. ROIs were applied to the blood pool image and were pasted onto the delayed image for H/CL ratio determination.

Results: 107 patients were evaluated (mean age 75.5 years, 9 F, 98 M), 69% ATTR (N=74); 12% AL (N=13); 19% heart failure with preserved ejection fraction (HFpEF) without amyloidosis (N=20). All patients had a tissue biopsy, and 29% had EMB (N=31). All patients had echocardiography imaging consistent with an infiltrative cardiomyopathy such as ventricular wall thickening and increased heterogeneous echogenicity. The best 2D H/CL method for differentiating ATTR from non-ATTR amyloidosis was to use ROI₅, the ROI that excluded adjacent osseous structures near rib ends and fractures (area under the curve (AUC) of 0.889, sensitivity of 93% and specificity of 82% with a cut-off of H/CL = 1.3). The next best method was to use ROI₁, a standard small ROI (AUC of 0.886, sensitivity of 95% and specificity of 76% with a cut-off of H/CL = 1.3). When limited to patients with endomyocardial biopsy confirmed diagnoses (N=31), both methods showed 100% specificity, while the small ROI (ROI₁) method had sensitivity of 96% and the method avoiding bone trauma near the rib ends (ROI₅) had a sensitivity of 91%. The other ROI methods were slightly inferior.

Conclusion: As anticipated, avoiding areas of high PYP activity secondary to bone trauma improved the accuracy of the H/CL ratio for ATTR in our study, resulting in high test accuracy. This supports a practice change from what has previously been published on methods of drawing ROIs. We were somewhat surprised that the small standard ROI performed nearly as well. Perhaps, a future step could be to combine the methods of small ROI and avoiding bone trauma near the rib ends and fractures.