Ruling in or out the presence of cardiac amyloidosis based on cut-off values using T1-mapping – caution is required regarding the control group

Abstract

Abstract Background Cardiac amyloidosis (CA) is an infiltrative disease that is characterized by accumulation of amyloid deposits in the interstitium of the myocardium. In contrast, hypertrophic cardiomyopathy (HCM) is caused by a disorganized arrangement of myocyte hypertrophy as well as expanded extracellular matrix, composed of interstitial and replacement fibrosis. Purpose A diagnostic algorithm based on (native) T1-mapping using cardiovascular magnetic resonance (CMR) was suggested in a recent study for the diagnosis of CA: A native T1 <1,036ms was mentioned to have a 98% negative predictive value (NPV) for ruling out CA whereas a native T1 >1,164ms showed a 98% positive predictive value (PPV) for the presence of CA. In the present study, we critically addressed the calculation of such cut-off values considering possible differences in the composition of the control group. Methods N=30 patients with CA, N=20 patients with HCM and N=15 healthy controls without relevant cardiac disease underwent dedicated CMR studies on a 1.5-T MR scanner. The CMR protocol comprised standard sequences for cine-imaging, native and post-contrast MOLLI-based T1-mapping and late-gadolinium-enhancement (LGE). ECV measurements were based on pre- and post-contrast T1-mapping images. Results Native T1 and ECV were significantly increased in CA compared to HCM and receiver operating characteristic (ROC) analyses revealed an area-under-the-curve (AUC) = 0.984 for native T1 (p<0.001) and AUC = 0.985 for ECV (p<0.001) regarding the diagnosis of CA). When CA patients were compared to HCM patients (excluding healthy controls), a native T1 <1,036ms or an ECV <33% were associated with a 99% NPV for ruling out CA whereas a native T1 ≥1,082ms or an ECV ≥41% were associated with a 99% PPV for diagnosis of CA. However, when CA patients were compared to healthy controls (excluding HCM patients), a native T1 <1,025ms or an ECV <34% were associated with a 99% NPV for ruling out CA whereas a native T1 ≥1,025ms or an ECV ≥34% were associated with a 99% PPV for diagnosis of CA since there was no overlap in native T1 and ECV values between CA patients and healthy controls. Conclusion Cut-off values for native T1 or ECV derived from ROC analyses (in a specific group of study patients) for ruling in or out the presence of CA are – amongst others - determined by the native T1 and ECV values of the respective “control group”. A different composition of the control group (e.g. HCM patients vs. healthy volunteers) will result in different cut-off values. Hence, previously suggested cut-off values obtained in single center studies need to be considered carefully – with a special attention to the control group of the underlying study – and should not be transferred to other centers carelessly. Funding Acknowledgement Type of funding source: None