Mandeep R. Mehra, Jon A. Kobashigawa, Mario C. Deng, Kenneth C. Fang, Tod M. Klingler, Preeti G. Lal, Steven Rosenberg, Patricia A. Uber, Randall C. Starling, Srinivas Murali, Daniel F. Pauly, Russell Dedrick, Michael G. Walker, Adriana Zeevi, Howard J. Eisen, CARGO Investigators

We have previously demonstrated that a peripheral blood transcriptional profile using 11 distinct genes predicts onset of cardiac allograft rejection weeks to months prior to the actual event. In this analysis, we ascertained the performance of this transcriptional algorithm in a Bayesian representative population: 28 cardiac transplant recipients who progressed to moderate to severe rejection; 53 who progressed to mild rejection; and 46 who remained rejection-free. Furthermore, we characterized longitudinal alterations in the transcriptional gene expression profile before, during and after recovery from rejection. In this patient cohort, we found that a gene expression score (range 0 to 40) of <or =20 represents very low risk of rejection in the subsequent 12 weeks: 0 progressed to treatable (ISHLT Grade > or =3A) rejection; 16 of 53 (30%) from the intermediate group (those who progressed to ISHLT Grade 1B or 2) and 13 of 46 (28%) controls (who remained Grade 0 or 1A) had scores < or =20. A gene score of > or =30 was associated with progression to moderate to severe rejection in 58% of cases. These two extreme scores (< or =20 or > or =30) represented 44% of the cardiac transplant population within 6 months post-transplant. In addition, longitudinal gene expression analysis demonstrated that baseline scores were significantly higher for those who went on to reject, remained high during an episode of rejection, and dropped post-treatment for rejection (p < 0.01). The use of gene expression profiling early after transplantation allows for separation into low-, intermediate- or high-risk categories for future rejection, permitting development of discrete surveillance strategies.