6th ESACP Congress, Heidelberg, April 7-11, 1999

A143
PREDICTIVE MEDICINE BY CYTOMETRY
Valet G 1, Tarnok A 2, Kellermann W 3, van Driel BEM 4, van Noorden CJ 4, Tschöpe D 5, Otto F 6, Kahle H 1

1) AG Zellbiochemie, Max-Planck-Institut für Biochemie, Martinsried, 2) Pediatric Cardiology, Heart Centre Leipzig GMBH, University Hospital, Leipzig, 3) Anästhesiologische Abteilung, Krankenhaus Schwabing, TU München, Germany 4) Lab.Cell Biology & Histology, Academic Medical Centre, Univ.Amsterdam, Netherlands 5) Arb.Gruppe Zelluläre Hämostase & Klinische Angiologie, Diabetes Forschungsinstitut, Heinrich-Heine-Universität, Düsseldorf 6) Fachklinik Hornheide, Münster, Germany

Background: Multiparametric flow and image cytometric measurements contain high amounts of information which are frequently only partially evaluated either by restriction to certain cell populations e.g. lymphocytes, cancer cells etc. or to certain parameters e.g. percent cell frequency or S-phase cells. These limitations are either due to intellectual concept e.g. evaluation of only lymphocytes in lymphoma but frequently also to the lack of suitable software which can treat hundreds or thousands of data columns during learning processes and provide standardized i.e. portable as well as intelligible classifications suitable e.g. for scientific hypothesis formation. Aim: Several studies in different areas of clinical medicine aimed to determine whether exhaustive extraction and evaluation of multiparametric cytometric data in conjunction with their automated triple matrix classification (http://www.biochem.mpg.de/valet/classif1.html) was capable of either disease course predictions or risk assessments for individual patients. Methods & Results: Data derived mostly from flow cytometric list mode file analysis or cytophotometry with the analyzed cellular systems being closely involved in the various disease processes. Clinical chemistry parameters were also included in some instances. The results show that flow cytometric cell function measurements or immunophenotyping is capable to predict diseases course in between 90 to 100% correctly for intensive care patients (risk of sepsis, shock, death, preoperative prediction of postoperative capillary leak syndrome) for time periods between 1 and 10 days in advance. Furthermore risk assessment for myocardial infarction risk patients from peripheral thrombocyte activation analysis by immunophenotyping as well as for prediction on 10 year survival of melanoma patients at surgery was possible with reasonable certainty (80-95%). Lastly, the predictive capacity for survival/death in colo-rectal cancer patients from cytophotometric G6PDH, Zn/Mn SOD and lipid peroxidation reaction screening is clearly improved over predictions from Dukes classification or patho-histological grading. Conclusion: Exhaustive data analysis on cytometrically obtained biochemical cell and tissue parameters permits to prognosticate disease course with high accuracy in individual patients for a variety of clinically important disease states. This will allow to further optimize the respective therapies and reduce unwanted therapeutic side effects.