R. G. Macfarlane

A Thrombin Generation Test: The Application in Haemophilia and Thrombocytopenia

Much recent research on blood clotting has dealt with prothrombin and factors which influence its activation. In the development of such investigations experimental procedures have become increasingly complex and artificial. They usually involve the use of anticoagulants, often multistage fractionation by precipitation, adsorption or filtration, and, almost invariably, the addition of tissue extracts to promote thrombin generation. These techniques have yielded information of great importance, but they must have a limited application, since they are far removed from the natural process of clotting. Normal blood taken by a clean venepuncture into a glass vessel Clots firmly in a few minutes without any stimulus other than surface contact. It must therefore have some intrinsic source of thromboplastic activity, in the sense that something is formed which converts prothrombin to thrombin. It is clear that a study of this intrinsic thromboplastin is essential to further understanding of normal clotting and its various defects, because the ability of blood to clot without artificial assistance probably determines its haemostatic efficiency. Fractionation techniques can give no information on the earliest stages of natural clotting, which must, initially, be studied in blood which has been altered as little as possible. Estimations of clotting time and of prothrombin consumption can be made on unaltered blood, but because both give only a single figure which is influenced by several different factors, they are of little help in analysing the causes of delayed or deficient coagulation. More information would be provided if a series of determinations of the thrombin content of whole blood were made at intervals during clotting. In 1901 Arthus carried out an experiment of this sort. He took serial samples of dogs blood which was defibrinated by beating, added sodium fluoride, and after centrifuging determined their ability to clot fluoride dog plasma. A simplified technique described in the following paper uses unaltered whole blood from which serial samples are simply transferred to fibrinogen, the clotting time of which indicates the thrombin concentration of the blood at the time of sampling. By plotting thrombin concentration against time, curves are obtained which show the speed of thrombin generation and destruction and thus may indicate the rate of production and the activity of blood thromboplastin. The generation of thrombin in whole blood can only reflect thromboplastic activity if irrelevant variable factors are controlled as much as possible. Variations in prothrombin, antithrombin, and calcium concentration are likely to have a considerable effect on thrombin generation, but they were avoided by the use of normal blood, or blood from cases of haemophilia or thrombocytopenia in which these factors are almost certainly normal.

Haemophilia and related conditions: a survey of 187 cases.

THE object of this paper is to present the results of rccent laboratory investigations on a series of cases derived from I 87 families with constitutional disorders of blood coagulation. By relating these results to the clinical findings it is possible to assess the value of laboratory tests in diagnosis and prognosis. The majority ofthe cases investigated are classified as ‘hacmophilia’ and there has been already a number of comprehensive surveys of haemophilia, including the monumental work of Bullock and Fildes (1911) and those of Carol Birch (1937) and Andreasscn (1943). Our own survey is not intended to be comprehensive, since it deals mainly with investigations of clotting function, and it is therefore a supplement rather than a coniplcte work. But supplementary surveys arc needed because new methods of investigations have changed diagnostic criteria. The term ‘haemophilia’ has for this reason suffered a series of changes of meaning since the appearance of the classical monographs, and what was then regarded as a homogeneous condition has now apparently disintegrated into a number of distinct entities, with a similar symptomatolog y. For the sake of later clarity, it is necessary to review briefly this disintegration, though it has bccn discussed elswhere (Macfarlane, 1954a, b; Pitney and Dacie, 1955). It is, of course, part of a process which is at work throughout medicinc, and indced throughout biological science. Until a few years ago, the diagnosis of haemophilia was mainly a clinical one, resting on the well-known haemorrhagic tendency existing from infancy in a male patient, and on the distinctive sex-linked inheritance. A prolongation of the clotting time was the only positive laboratory finding contributing to the diagnosis. A proportion of cases could not be diagnosed with confidence as classical haemophilia, because of some deviation from the accepted criteria such as an absence of a family history or an abnormal inheritance; or unusual findings, such as a prolongation of the bleeding time or increased capillary fragility. Such cases might be dismissed as ‘atypical’ or re-classified under onc of thc variety of names which abound in the literature. As a result of research in blood coagulation diagnostic criteria in the haemorrhagic disorders became more precise. The application of the prothrombin time test of Quick (1935) allowed a simple separation of clotting defects into two groups, thosc in which the prothrombin time was normal, and those in which it was prolonged. In the majority of the constitutional cases, including most cases then classified as haemophilia, the prothrombin time was normal. In sonic paticnts, however, haeniophilia-like symptoms werc associated with a long prothrombin time, and investigation of these revealed cases of constitutional deficiency of Factor V (Owren, 1947; Frank, Bilhan and Ekren, 1950), Factor VII (Koller, Locliger and Duckert, 1951; Alexander, Goldstein, Landwehr and Cook, 1951), and of prothrombin (Biggs and Douglas 1953a). Subsequent literature on these conditions has been reviewed by Ackroyd (1956) and Biggs and Macfarlane (1957). Patients whose blood gives a long one-stage prothrombin time are not considered here. The prothrombin consumption test (Quick, 1947) also proved valuable, because it revealed defects of coagulation not detectable by siniplc clotting time determinations. Using a test similar to that of Quick, Merskey (1950) investigated patients with the clinical manifestations of haemophilia but with normal or almost

The coagulant action of Russell's viper venom; the use of antivenom in defining its reaction with a serum factor.

IN 1903, Lamb and Hanna showed that the intravenous injection of Russell’s viper (Daboia) venom in animals caused massive thrombosis, but when the venom was tested on oxalated or citrated blood in vitro, Lamb (1903) and Martin (1905) found that it had little coagulant activity. Though Arthus (1919) pointed out that, like tissue thromboplastin, the venom required ionized calcium for its action, its coagulant power continued to be under-estimated, and Iswariah and David (1932) even maintained that there was no evidence that it increased the coagulability of the blood in vivo or in vitro. Barnett and Macfarlane (1934) investigated the coagulant action of venom from 22 varieties of snakes, using as an indicator haemophilic blood which required no anticoagulant. They found that Russell’s viper venom (RVV) was far more potent as a coagulant than any of the others tested and it was therefore used, in high dilution, as a local haemostatic for the control of haemophilic bleeding (Macfarlane and Bariiett, 1934). The venom then became commerically available under the name ‘Stypven’. The mode of action of this venom, which produces a demonstrable effect on clotting in concentrations of less than 0.01 ug. per ml. of blood, has since been much studied. Because it has no thrombin-like action commensurate with its ability to clot blood or plasma in the presence of ionized calcium it was assumed that the venom behaves like tissue thromboplastin and it was therefore used for a time as a substitute for brain extract in routine one-stage prothrombin time estimations (Fullerton, 1940; Witts and Hobson, 1940). Wide discrepancies between the results of this modification and Quick’s original method when applied in the control of drcoumarin therapy (Witts, 1942; Wilson, 1947) led to a realization that the venom and tissue thromboplastin must act in different ways. One difference had already been found, this being that venom appears to require phospholipid. The addition of (crude) lecithin or cephalin greatly increases its clotting power (Trevan and Macfarlane, 1938 ; Edsall, 1941) and the removal of platelets and lipids from plasma by centrifuging or extraction greatly reduces its coagulability by venom, this being restored by replacing the extracted fat or adding various phospholipid preparations (Macfarlane, Trevan and Attwood, 1941). Another difference is that Factor VII is not required for the action of the venom (Jenhns, 1954; Rapaport, Aas and Owren, 1954a) and thus the Factor-VII reduction caused by dicoumarin therapy may be undetected if venom is used instead of a tissue thromboplastin. The venom resembles tissue thromboplastin in requiring Factor V (Rapaport, Aas and Owren, 1954b), though Hall (1956) raised some doubts on this point. Factor X is almost by definition a co-factor for Russell’s viper venom, the reduction of this factor in the Stuart-Prower clotting defect being recognized and measured by delayed clotting with the venom (Hougie, 1956; Denson, 1958).

The Interaction of Factors VIII and IX

It has been shown by Macfarlane, Biggs, Ash and Denson (1) and Biggs, Macfarlane, Denson and Ash (2) that factors viii and ix, react in the presence of calcium chloride and traces of thrombin to form factor x activator (factor viiza) (figure 1). The curves shown in figure 1 were obtained using a one-stage method for following the development of factor x activator activity, and a two-stage assay method to follow the destruction of factor viii when mixtures of factor viii, ix a, thrombin, and calcium chloride were pre-incubated. It can be seen that coincident with the disappearance or consumption of factor viii in figure 1, there is development of an activator of factor x. It appeared that factor viii was a substrate in this reaction, and that formation of the factor x activator was catalysed by both activated factor ix (factor ixa) and thrombin. The effect of thrombin was unmistakebly clearcut. There appeared to be little doubt that even traces of thrombin had a powerful effect on factor viii alone, at first activating it and subsequently destroying it. Acceleration in the rate of formation of the factor x activator by different concentrations of thrombin is shown in figure 2. The overall reaction was depicted by the following equations In these experiments, the observed activity, which was attributed to the formation of viii a, could equally well have been due to contaminating amounts of factor x in the various preparations of factor ix used.

Estimation of Prothrombin in Dicoumarin Therapy

The concept of prothrombin as the specific precursor of thrombin dates from the so-called classical theory of blood coagulation put forward by Schmidt (1895) and Morawitz (1905). This theory, which considers that four factors, thromboplastin, prothrombin, calcium, and fibrinogen, are concerned in the formation of fibrin, has been the basis of most modern work. The theory was of more academic than practical importance until the discovery of vitamin K and the use of dicoumarin in treatment of thrombosis, both of which required the quantitative estimation of prothrombin. The available methods are all based on the assumption that the classical theory is fundamentally sound. Of the methods of prothrombin estimation that have been developed, none can measure prothrombin directly and therefore infer its concentration from observations on thrombin. The one-stage technique of Quick (1942) has been the most commonly used because it is simple, easily performed, requires little in the way of special reagents, and gives results which are in general agreement with clinical experience. The method consists of adding an optimum amount of brain thromboplastin and calcium to oxalated plasma, under which conditions the clotting time of the mixture is assumed to be proportional to its prothrombin concentration. The basis of this supposed relationship is the literal acceptance of the classical theory. If three of the four variables are controlled, then variations in the fourth must determine any alteration in the rate of fibrin production, which is in turn indicated by a change in the clotting time of the mixture. In practice thromboplastin and calcium are controlled, if possible at or near their optimum concentrations, and fibrinogen variations found in the ordinary clinical material are usually not of sufficient magnitude to influence the result. The remaining variable is therefore prothrombin. The method is essentially dynamic in principle, depending entirely on the rate of thrombin generation and not on the amount of thrombin that may finally be produced. It depends also on the assumption that there are no uncontrolled factors other than prothrombin that might modify this rate of thrombin generation. It is now known, of course, that this is not the case. Not only may the reaction time of fibrinogen vary, but accelerators and depressors of thrombin generation are present in normal plasma, and variation in these accelerators and depressors in pathological conditions may alter the results of this test. Quick (1942) himself has been the first to recognize this. The practical implications of the fallacious nature of this method are difficult to assess at present, but no alternative to it is available for routine use. Though theoretically the two-stage method appears to be on much firmer ground, since it seeks to estimate the total amount of thrombin generated, it is too complex for ordinary routine use. There are, moreover, many indications that its results may be equally fallacious. In dicoumarin therapy one component of prothrombin (prothrombin B of Quick, 1947) is deliberately lowered, and the responsibility for a reduction to haemorrhagic levels rests largely with the pathologist. A reliable method of prothrombin determination is therefore essential. There has recently been much discussion as to which of the modifications of the one-stage technique give the best results (Marsh, 1947, 1948; James, 1948; Cleland, 1947; Pivawar, 1947; Lempert, 1948; Canti and Robertson, 1948).