Background and aim of the study: An in-vitro technique has been developed to assess the flow-induced thrombosis of artificial heart valves, using renneted milk as a blood analog. Previous studies have demonstrated similarities between the clotting of blood and milk on both microscopic and macroscopic scales. The study aim was to further validate the milk test by comparing the locations of milk clot to those of thrombus formation on a wide selection of mechanical heart valves.
Methods: Nine different valves were tested in the aortic position of a model heart chamber in the Edinburgh milk rig. These included caged-ball valves (Starr-Edwards silastic ball with bare struts and metal ball with cloth-covered struts), tilting-disc valves (Björk-Shiley Standard, Björk-Shiley Monostrut, Medtronic-Hall, and Ultracor), and bileaflet valves (St. Jude Medical, Edwards-Duromedics, and CarboMedics). Renneted

milk was pumped through the valves for 30 min at 2 l/min, 70 bpm pulsatile flow. After each run, valves were photographed for comparison with documented sites of thrombosis.
Results: All valves developed milk clot in specific, reproducible locations when run in the aortic position. Milk clot was found on the struts of caged-ball valves and tilting-disc valves, and around the hinge mechanism of the bileaflet valves. This compared favorably to documented cases of thrombosis in vivo.
Conclusion: Renneted milk may be used to model flow-induced thrombus formation and to predict the thrombogenic sites of mechanical heart valves. Whilst it is not suggested that milk mimics the entire blood coagulation cascade, these results indicate that such behavior may not be necessary for predicting fluid mechanically induced clotting.
The Journal of Heart Valve Disease 2004;13:509-520