A number of clinics around the world perform heart transplantation. One of the most difficult problems in this field is rejection of the transplanted tissue. It is critically important to know the degree of this rejection. Today the most accurate method to measure it is biopsy (analysis of a small piece of tissue taken from the patient). However, making a biopsy of the heart tissue is a complex, dangerous, and undesirable action. In the present study, each time the biopsy was made, a blood sample was also taken. Each blood test provides a great number of immunological parameters that may reflect the degree of rejection. Taking blood samples is obviously much easier than performing a biopsy. Thus, the possibility of predicting the rejection from immunological data would be extremely beneficial for patients with a transplanted heart.
NRCS runs PolyAnalyst PRO on Windows NT 4.0.
The data consisted of 470 training pairs: “blood immunology data”, “degree of rejection”. Immunology data included 13 parameters (attributes). The rejection degree (measured through biopsy) was expressed as integer numbers ranging from 1 to 4.
The results obtained by PolyAnalyst have a surpassing scientific value and large clinical implications. They allow us to significantly reduce the number of performed heart biopsies by predicting the transplanted heart rejection degree. PolyAnalyst obtained a nonlinear single-parameter formula which provides correct prediction of the rejection degree with an accuracy of about 70%. This model has been clinically used in NRCS for more than a year, allowing to reduce the total number of heart biopsies by 44%.
Professor of Immunology
National Research Center for Surgery