Abstract 2574: Limitations of Molecular Expression Profiling (Allomap score) in Predicting Endomyocardial Biopsy Findings Following Cardiac Transplantation; a Phase IV Surveillance Study
Background: Non-invasive molecular profiling of immune pathways implicated in cardiac allograft rejection has been recently approved for clinical use, with Allomap scores (AS) = 30 (range: 0 – 40) consistent with cellular rejection, although the score has not been extensively validated in clinical practice.
Methods: We obtained blood samples for determination of AS at the time of endomyocardial biopsy using the commercially available quantitative PCR technique (n=46 samples from 29 cardiac allograft recipients, at least 6 months post-transplant).
Results: Biopsies interpreted blinded to AS were categorized as no rejection (ISHLT 0/4, n=13), mild rejection (ISHLT 1A/1B, n=23) and moderate or severe rejection (ISHLT 2/4 or higher, n=10). Patients were clinically regarded as “non-rejectors” (NR) with ISHLT 0, 1A/1B biopsies (n=36) and rejectors with ISHLT 2/3 (R, n=10). The mean Allomap score in the entire cohort was 29.9±1 and followed a non-Gaussian distribution with the majority of values in the upper quartile [AS: 0 –9: (2%), 10 –19: 3 (7%), 20 –29 (17%), 30 –39 (74%)]. There was no significant difference in AS between groups (NR: 30.3±1.2, R: 28.6±1.9, p~0.49). Patients with ISHLT 0/4 biopsies exhibited a paradoxical trend to have overall higher scores (32.4±0.9) than patients with either ISHLT 1A (29.1±1.6) or higher grade rejections. Using AS =30 as threshold, sensitivity was 60%, specificity 22%, positive predictive value (PPV) 18% and negative predictive value (NPV) 66%. To account for the skewed distribution, we re-analyzed the data using a cut-off AS of = 34, demonstrating sensitivity of 20%, specificity 69 %, PPV 15 % and NPV 75 %, still below those observed in previously published data. Thus, ROC characteristics of the test were suboptimal in our clinical setting.
Conclusions: Although molecular profiling of candidate genes to predict cardiac allograft rejection remains an attractive, non-invasive and physiologically sound technique, the clinical utility of the assay at its present form remains limited, especially for moderate (ISHLT 2/4) rejection.