Optimization of therapeutic procedure during LDL-apheresis - verification of the computerized model in clinical practice.

UNLABELLED: LDL-apheresis is a very effective method in the treatment of resistant hypercholesterolemia when other therapies fail. To maximize the efficacy of the use of LDL-absorbers, we created a computerized model. The aim of this study is to verify it in clinical use.

PATIENTS AND

METHODS

A therapeutic technique of immunoadsorption was used, applying a pair of Lipopak columns (Pocard, Russia). Plasma was separated by a continuous-flow plasma separator (Cobe Spectra, USA); adsorption was controlled by adsorption-desorption equipment Adasorb (Medicap, Germany). 494 LDL-apheresis procedures had been performed in nine patients with primary hypercholesterolemia in the earlier, initial study; 47 other procedures (202 therapeutic cycles) were used in this verification study. The program for procedure planning uses Microsoft Excel for Windows. Complex metabolism of the LDL-cholesterol was neglected (owing to the short-time period of the procedure) and the procedure calculated as continuous filtration. The input enterer into the program includes basic patient data (mass, height, sex and initial plasma LDL level in mmol/L).

RESULTS

The results show a very good match between calculated levels and the real laboratory results in most procedures, but in some procedures we observed minor differences (0.05 mmol/L), which was caused by procedure adjustments due to technical reasons. However, some methodological and medical details must be carefully observed (initial cholesterol level, correct calculation of plasma volume, and the precise capacity of adsorbers that must not be overshot); as they influence the correct match between calculated and real results significantly.

CONCLUSIONS

Although our software uses a fairly simplified model of the LDL-cholesterol kinetics during the LDL-apheresis, it is providing a great aid in the procedure planning. It is also suitable for practical use because it only requires a few commonly used and readily available input values.