Answers are grounded in this paper and cite their sources.
The burden of heart disease continues to increase (Sans et al, 1997), and coronary artery bypass grafting (CABG) increases with it: to 400 000/year in the U.S.A. (Motwani & Topol, 1998) and >30 000/year in the U.K. Despite developments in 'minimally-invasive' surgery (Bryan & Angelini, 1996), nearly all CABG — together with surgery for valvular and congenital heart disease, transplantation and major aortic procedures — is performed on unbeating hearts supported by cardiopulmonary bypass (CPB). CPB pumps blood through an extracorporeal circuit with an oxygenator which exchanges O2 and CO2. This process consequently disturbs haemostasis, often leading to excess perioperative bleeding. Consequent allogeneic transfusion of red cells (RBC) comprises 20% of all RBC transfused in the U.S.A. (Johnson et al, 1992). U.K. figures are unavailable, but one Regional Cardiothoracic Centre, doing 4% of U.K. heart surgery, calls for 4000 RBC units/year: 17.6% of transfused RBC, which is more than all other surgery, including trauma, combined. This has produced a rapidly growing cost pressure on cardiac services; in the U.K. universal leucodepletion of blood products, due to concern about possible transmission of the agent of new-variant Creutzfeld-Jakob disease (nvCJD) by lymphocytes (Aguzzi, 1997), together with nucleic acid testing for blood-borne viruses, may double the cost in the near future. Understanding disordered haemostasis in CPB would benefit health and economics, but has proved difficult. Proponents of complexity and chaos theories in biomedicine can find rich material in the haemostasis of CPB. The evidence base is dispersed: surgeons, anaesthetists and coagulationists tend to publish in different journals and to have different preconceptions, aims and methodologies. Determined investigators have crossed these tribal boundaries to produce substantial data and therapeutic advances. This article attempts to ask if anything important has changed since a magisterial review by Woodman & Harker (1990); readers seeking earlier references can find them there. After 60 years of evolution, blood anticoagulated with the same dose of heparin received by Gibbon's cats in 1937 (300 u/kg) is pumped over 1.5–2 m2 of non-biological surface, mostly in the oxygenator where flow is engineered to be non-laminar to maximize O2 transport (Stammers, 1997); the recent generation of hollow-fibre oxygenators have reduced these factors to a minimum. Surface composition of current CPB devices includes polyurethane, polyester, plasticized PVC, polypropylene and polycarbonate (Hsu, 1997). During surgery the sternum, mediastinum, pericardium, great vessels and heart are incised and cannulated, the aorta cross-clamped, and the pulmonary circulation converted to a low-flow venous conduit fed by bronchial vessels. Exposed myocardium and pericardium are bathed in a mixture of blood and cardioplegia solution, which is drained into the cardiotomy reservoir and often re-infused directly via the CPB circuit. Resulting cross-talk between artificial surfaces, cells and proteins in the circulating blood, traumatized vascular endothelium and extravascular tissues, creates enormous complexity which is impossible to model in vitro. Seeking a single key to this complexity, whether platelet function defect or fibrinolysis, is a fallacy which can lead to conceptual problems in research, selective readings of the literature, and polarized advocacy of one mechanism or drug effect above all others. Given differences in bleeding rates and definitions of what is normal, excess bleeding can probably be defined as >1 litre per procedure. There is twice the risk of bleeding after valve surgery than after CABG (Hardy et al, 1991), and repeat operations redouble this risk; significantly, 19% of CABG patients need re-grafting within 10 years (Motwani & Topol, 1998). Bleeding is usually manifest postoperatively, after protamine reversal of heparin, and shed from the operative field into mediastinal and pleural drains rather than seen at remote sites. If aspirin (ASA) or other antiplatelet therapy has been given, the operation may be 'wet' from the start. Critical rates of blood loss, formulated by Kirklin & Barrett-Boyes (1986), are >500 ml in the first postoperative hour, >400 ml/h in the first 2 h, >300 ml/h in the first 3 h, >1litre in 4 h, or >1.2 litres in 5 h. If bleeding attains these rates, is acute and massive, or begins again after ceasing, resternotomy becomes unavoidable. Volume per se can mislead if there is progression from blood to serosanguineous drainage; the haematocrit of the fluid in the drain tubing (not the reservoir) may indicate that blood loss is being overestimated if the patient is otherwise stable. The need for re-sternotomy entails a 30% increase in perioperative mortality, and hence is a crucial endpoint in CPB studies. In 67% of cases bleeding vessels are found, often small mediastinal arteries or the aortotomy incision; such vessels might not bleed if haemostasis were normal, but a routine check of repeat offenders before closure is worthwhile. In the remainder, general microvascular oozing is seen (Unsworth-White et al, 1995). Allogeneic RBC transfusion post-CPB shows wide, apparently irrational, variance between centres (Goodnough et al, 1991). Female gender, by determining preoperative haematocrit and blood volume, is a consistent risk-factor for transfusion (reviewed by Stehling, 1998). Some centres transfuse RBC to fewer than a third of standard-risk CPB patients, others to more than two-thirds (Stover et al, 1998), despite access to evidence-based guidelines (Goodnough et al, 1990), with similar inconsistency in platelet and fresh frozen plasma use. The U.K. situation may be similar, although CPB is more centralized. These differences confound studies using volume as an end-point, since a transfusion-prone team out-ranks other risk-factors for RBC transfusion (Stover et al, 1998). As pointed out by Laupacis et al (1997), transfusion risk reflects the number of allogeneic exposures; studies reporting only transfused volumes (particularly as means, when marked skewing is likely) do not adequately measure this risk. Witholding RBC unless the systemic haematocrit fell to <25% (Hb <8 g/dl) post-CPB had no adverse clinical or physiological impact in standard-risk patients (Johnson et al, 1992). Observing this threshold is likely to reduce allogeneic transfusion as much as any foreseeable method of improving haemostasis. The role of cardiac surgeons and anaesthetists in U.K. hospital transfusion committees will be vital as a means toward audit and effective guidelines in this area; without such participation, increasing scrutiny of cardiac surgery outcomes may lead to a defensive retreat from rational transfusion practice. Space disallows consideration of reducing allogeneic exposure by pre-operative autodonation or acute normovolaemic haemodilution; a useful annotated bibliography is provided by Stowell et al (1998). This account covers CPB events in sequence; to avoid repetition, relevant interventions are discussed alongside, e.g. aprotinin vis-à-vis fibrinolysis. Bleeding increases if aspirin (ASA) continues up to surgery (Kallis et al, 1994b). This effect is eliminated if ASA is stopped 7 d before, and re-started 1–6 h after, surgery. If ASA cannot be stopped, haemostasis should be enhanced by antifibrinolytic therapy (see below). The calcium antagonist nimodipine was associated with excess post-CPB bleeding in one report (Wagenknecht et al, 1995), but this has not been confirmed, nor found with other drugs of this class (Hynynen et al, 1996). Coumarin anticoagulation (e.g. in transplantation when a donor heart arrives too suddenly to omit warfarin) requires replacement therapy with prothrombin complex concentrate (PCC) containing factor VII (e.g. Beriplex®) if the INR is >1.7. Coronary angioplasty/stenting with the hybrid anti-GpIIb/IIIa monoclonal agent abciximab (c7E3, ReoProTM) may need urgent conversion to CABG (Boehrer et al, 1994). Intra- and post-operative bleeding occurs, particularly if the interval between abciximab and CPB is <12 h (Gammie et al, 1998) or if standard doses of heparin are used for CPB. Reducing heparin (to ACT 400 s) with postoperative (±pre-operative) platelets is one approach (Ferguson et al, 1998). Others suggest aprotinin with platelet transfusion (6 units) given at the end of bypass (Alvarez, 1998). True and pseudo-thrombocytopenia can occur after abciximab therapy and must be distinguished from heparin-induced thrombocytopenia; an algorithm for this was provided by Berkowitz et al (1997). Patients with congenital heart disease may acquire a deficiency of high-molecular weight von Willebrand factor (VWF; Gill et al, 1986) which rarely poses a problem since it corrects immediately post surgery (Turner-Gomes et al, 1992). Right-to-left shunts can lead to increased platelet size with misleading automated counts (check the blood film). Children with Noonan syndrome may have coagulation factor deficiencies (Sharland et al, 1992) and can bleed during surgery for heart defects; they need a haemostatic work-up before cardiac surgery. Patients with haemophilia of any degree of severity may require CPB; the safest operative cover in all cases is high-purity plasma-derived or recombinant VIII:C or IX:C. In von Willebrand's disease a product with a reliably high content of high-molecular-weight (HMW-VWF) multimers should be used. All such patients should be managed at centres with comprehensive expertise in haemophilia. Pre-operative thrombocytopenia compounds the bypass-induced platelet function defect (see below). The minimum acceptable pre-operative platelet count is 100 × 109/l, requiring steroid or intravenous immunoglobulin (IVIG) therapy in immune thrombocytopenia. Platelet counts fall by 25–60% within 15 min of first passage of blood through the primed CPB circuit (Salzman, 1963). Simultaneously, a platelet function defect marked by the bleeding time develops (Harker et al, 1980). The speed of this effect is due to the exposure of circulating platelets to early conditions of bypass: the high concentration of unfractionated heparin (UFH), haemodilution by oxygenator prime, and the non-biological surface of the extracorporeal circuit. A direct platelet effect of UFH (3–8 U/ml) was suggested by studies that showed the fall in platelet count to begin between heparinization and the first passage of CPB (Mazer et al, 1995; Wahba et al, 1996). Blood–surface interactions begin with adsorption of fibrinogen on the surface, which recruits and activates phagocytes (Tang & Eaton, 1993) initiating the post-CPB inflammatory state (Hall et al, 1997); it also attracts and activates platelets (Sly et al, 1995). In flow chamber experiments, platelet adhesion to an artificial surface in the presence of UFH is no different from that in non-anticoagulated blood (Badimon et al, 1987). In-vitro, adsorbed fibrinogen is replaced within 15 min of continued exposure to plasma (Vroman & Evans, 1967). The dominant protein becomes kinin-free high-molecular weight kininogen [HKa], bradykinin having been released by surface-activated XIIa and kallikrein (Schmaier et al, 1984; Brash et al, 1988). HKa has anti-adhesive effects on platelets (Asakura et al, 1992) and blocks their activation by thrombin (Meloni & Schmaier, 1991), so 'passivation' of the artificial surface may be mediated by HKa, although this has not been proved in CPB. During this period a shower of microparticles can be detected in the circulating blood, many of which are aggregates of platelets and white cells, predominantly monocytes (Rinder et al, 1992), and the peripheral monocyte count decreases (Parrat & Hunt, 1998). There is incomplete consensus on the pathophysiology of the CPB platelet lesion, but it is plausible that contact with surface-adsorbed fibrinogen and thrombin early in CPB activates platelets, causing transient adhesion, releasing alpha-granule contents and down-modulating surface receptors for VWF and fibrinogen. The resulting circulating platelets are functionally compromised (reviewed by Woodman & Harker, 1990; Addonizio, 1990). Platelet binding to surface-bound fibrinogen is a statistical phenomenon, rates of adhesion and detachment at different shear forces determining the density of the platelet layer (Jen et al, 1996). At high shear rates platelets adopt fully spread GpIIb/IIIa-dependent forms; further shear stress tears central holes in them (Wu et al, 1997). By analogy, in CPB, GpIIb/IIIa 'footprints' are found on the oxygenator surface (Wenger et al, 1989) and platelet microparticles in the blood (Abrams et al, 1990). In CPB, platelet alpha-granule constituents are detectable in the plasma and both ultrastructure and aggregometry reveal alpha-granule depletion (reviewed by Woodman & Harker, 1990). Flow cytometry, particularly using platelet-rich plasma (Shigeta et al, 1997; Wahba et al, 1996) but also in whole blood (Mazer et al, 1995), may reveal alpha-granule P-selectin on the platelet surface; several studies using whole blood dispute this (Kestin et al, 1993; Ray & Martin, 1997; Maquelin et al, 1998). Agonist-induced alpha-granule depletion is a subtle platelet function defect; disaggregated thrombin-aggregated platelets are refractory to thrombin and vulnerable to plasmin dissociation of new aggregates (Kinlough-Rathbone et al, 1992), both defects relevant to the CPB environment. Many flow cytometric studies in CPB demonstrate reduction of surface GpIb (van Oeveren et al, 1990), whole-platelet GpIb by radioassay being preserved (Orchard et al, 1993). Ristocetin-induced platelet aggregation (RIPA; GpIb-dependent) decreases during CPB (Lu et al, 1991); both RIPA and surface GpIb have been found to correlate with blood loss (Unsworth-White et al, 1996). Aprotinin, which reduces post-bypass bleeding (Laupacis et al, 1997; see below), seems to preserve surface GpIb when used during CPB (Van Oeveren et al, 1990), and permit the return of GpIb when used at the end (Kallis et al, 1994a). GpIb reduction is consistent with the nature of CPB bleeding; the oozing seen in 23% of patients is from small vessels with high shear rates, predictable if the GpIb–VWF binding critical to haemostasis in such vessels (Goto et al, 1998) were disturbed. These GpIb effects would fit protease-driven receptor trafficking between the platelet surface and the canalicular system (Nurden, 1997), but evidence for this phenomenon in CPB is circumstantial, trafficking being inferred rather than directly demonstrated. Glycoprotein trafficking, both in and out of CPB, is the subject of radical disagreement: whether it occurs at all (White et al, 1995); in CPB, whether platelet-surface GpIb and GpIIa/IIIb changes are artefactual (Kestin et al, 1993); and whether aprotinin has any effect on platelet membrane GpIb in CPB (Ray et al, 1997). Kestin et al (1993) came close to falsifying the CPB platelet hypothesis by finding no loss of GpIb, alteration in surface markers, or alpha-granule secretion in CPB using whole-blood flow cytometry. They suggested that findings by others (e.g. van Oeveren et al, 1990) were due to using centrifuged and gel-filtered platelets, compared to minimally disturbed whole blood, for flow cytometry. However, ~35% loss of surface GpIb has been found in subsequent studies using whole blood and FITC-conjugated monoclonal antibodies (McAbs) (Kallis et al, 1994a; Mazer et al, 1995; Unsworth-White et al, 1996; Maquelin et al, 1998). Differences in sampling, case-mix, CPB techniques, or the use of biotinylated versus FITC-conjugated McAbs may account for this; in CPB the scope for such variation between studies is vast (Royston, 1992). Kestin et al (1993) analysed platelets shed from skin incisions during CPB, and proposed an extrinsic platelet defect due to heparin-mediated thrombin deficiency, but it is now known that thrombin generation is supra-basal throughout CPB (Boisclair et al, 1993a). These unexplained inconsistencies call the GpIb hypothesis of the CPB platelet lesion into question. Positive studies find surface GpIb reduced by ~35%; not enough to cause bleeding in Bernard-Soulier heterozygotes, nor to reduce platelet adhesion to collagen under flow conditions (Van Zanten et al, 1998). Others have demonstrated that the tethered-ligand thrombin receptor (PAR1) is down-regulated during CPB, also a potential protease-induced traffic flow, while the high-affinity binding site for thrombin located on GpIb is preserved (Ferraris et al, 1996). No doubt studies of PAR3 'et al' in CPB will follow. The CPB platelet defect is evidently composed of multiple lesions, individually weak, but in concert disabling parallel platelet systems to such an extent that a temporary but dangerous bleeding disorder occurs. Clinical correlates permit the model of platelet malfunction at the of this more If haemostasis it that circulating VWF is a of post-CPB bleeding. plasma VWF et al, and platelet VWF (Kallis et al, to be found in CPB; others have detected post-CPB depletion of if only in patients bleed the (Unsworth-White et al, 1996). However, no effect in the of of (Laupacis et al, 1997). of et al, 1998) any benefit by increasing post-bypass (see below). VWF is associated with blood et al, of to see if haemostasis was to of blood would be of If platelet in CPB the with surface-adsorbed it might preserve platelet function during CPB. Platelet adhesion to can be reduced by which also binding et al, of the not changes in platelet of GpIb or P-selectin et al, 1997), but of to the reduced platelet adhesion to the oxygenator membrane by at the cost of 4% (Sly et al, 1996). This direct of the function of the surface both and further studies are receptor with monoclonal despite being associated with post-CPB platelet counts after urgent conversion to CABG (Ferguson et al, 1998), is too and resulting in increased blood However, a antagonist of the fibrinogen has been used with in a et al, and further platelets can bypass the bypass via pre-operative platelet the platelet autodonation to the patient post-CPB et al, of this of the platelet hypothesis of post-CPB bleeding are in The platelets are to shear stress and artificial surfaces, and time and are is to transfuse 2 × of platelets post-CPB when excess blood becomes the platelet for 23% of platelet in one and supported by the evidence there is no data this or to indicate the dose or However, platelet transfusion is associated with reduction in the of blood loss in fresh whole blood was given a in a which haemostatic et al, but it is and would blood in product it is CPB is a for the new central of the coagulation in which factor with factor VII thrombin by both factor and factor factor kallikrein and are no to to thrombin generation in & 1995). This the that contact with the CPB circuit via a which seems to be in for such an activation has been rather than in CPB; factor XIIa and kallikrein were detected in in CPB et al, but in a a in the of to in CPB, was the only detectable et al, 1997). In the new model this contact activation bradykinin and activates rather than by of (Lu et al, 1991), complex and prothrombin activation (Boisclair et al, & Hunt, 1998), throughout CPB despite high UFH and et al, 1996), and is in the platelet lesion and of CPB et al, 1992). The of this thrombin is although on the CPB circuit et al, 1998), it is before factor activation (Boisclair et al, and despite of factor et al, to an in contact activation and during CPB. of factor during CPB also factor that through factor activation is the rather than factor XIIa et al, 1995). potential site of is the surface of circulating monocytes aggregates detected during CPB (Rinder et al, 1992) are et al, 1997), and likely to be by platelet P-selectin et al, 1989) which can et al, 1994). by flow was found to be increased in CPB blood by et al but for h et al, or & Hunt, 1996) by others. by after heparin is during CPB (Unsworth-White et al, 1996; et al, 1996), with a of et al, 1996), so if is for thrombin generation in CPB it must be to plasma and from must also the which blocks et al, 1995). Proponents of the of thrombin in CPB need an for from the heparin-mediated double on is by to the oxygenator et al, 1996) and in blood shed into the et al, 1996). The surface receptor activates factor directly & 1996), and increased generation by without evidence of has been detected in clinical CPB by & also found of on the of oxygenator at different by different are the plausible of thrombin in CPB. This to the blood that has drained the surface to with in the cardiotomy reservoir and be re-infused into the CPB circuit. as material the need for allogeneic RBC, the haematocrit of cardiotomy is usually and it access to the As as et al, 1996), cardiotomy microparticles from platelets and at 10 the concentration found in systemic CPB blood et al, 1997). A to find if thrombin generation during CPB on this would be The cardiotomy reservoir is also a site of et al, 1993; see below). A cause of thrombin generation in CPB, despite heparin, would be in CPB is by devices which measure a whole blood the The ACT is by factors other than heparin concentration and can mislead as a measure of circulating heparin, particularly in et al, 1997). of the (e.g. or which heparin protamine have been in an to heparin and reversal by by these more heparin and protamine to be given et al, 1996) and there is evidence of thrombin but in bleeding and blood product use have not been et al, 1998). during CPB et al, to a degree that heparin in et al, 1997). by plasma-derived concentrate reduces of thrombin generation during CPB et al, 1994). of in CPB are The has been by heparin to the surface of the CPB circuit and the a of heparin the Despite a fall in of the inflammatory to CPB, and reduction of systemic heparin dose to such have not reduction of bleeding or exposure to allogeneic blood (reviewed by & 1996). protamine from heparin if the UFH dose is by (see there concern over including and which occurs in up to of et al, 1998). immune via the particularly in with deficiency of et al, 1997). (Unsworth-White et al, 1996) after protamine but protamine has been for and in any heparin reversal can be to protamine & and recombinant platelet factor 4 they heparin rapidly and but these do not the increase in The drug heparin-induced thrombocytopenia may of and in CPB. to of CPB patients antibodies et al, 1996), but only are likely to thrombocytopenia et al, 1995), usually at the postoperative and although thrombocytopenia and venous have been d post-CPB et al, 1994). cases should be in repeat CPB is of are rarely and the rates of aggregometry and that the to out heparin for CPB may have to be on (e.g. a platelet count a mixture of is the for UFH in CPB 1993). A dose of to the prime, if et al, excess bleeding. is by by and has a not with in has been in CPB with in the presence of et al, 1997). Pre-operative of the CPB in CABG patients when plasma fibrinogen is et al, and has been used for CPB in et al, effect can be by fibrinogen In one of CPB with in despite a plasma fibrinogen of this that patients, have are not to fit in their to and may need e.g. of and et al, 1998). into the require between and CPB is no detectable is not in of the of but and and have been used with (reviewed by & 1997). of these is The of the heart and coronary circulation are rich in plasmin
No takes yet. Share an insight, caveat, or question.
David Bevan (1999) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: