Victor J. Marder, Division of Hematology and Medical Oncology, Department of Medicine, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA. Tel.: +1310 825 4469; fax: +1310 825 0914. E-mail: firstname.lastname@example.org
Summary. Two issues have held the focus of thrombolysis research for over 50 years, namely, choosing between a plasminogen activator (PA) or plasmin as the best therapeutic agent and choosing between systemic or local administration. The original plasmin product of the 1950s was both ineffective and contaminated with PA, and catheter technology was not yet developed for routine clinical use. For decades, clinical practice has focused on PA and systemic administration, but today, PAs are often administered by catheter into thrombosed vessels, notably for peripheral arterial and graft occlusion and deep vein thrombosis, and increasingly for acute ischaemic stroke. Despite using catheter-delivered therapy, bleeding complications still occur, most severely expressed as symptomatic intracranial haemorrhage. New experimental data indicate that we should now reconsider plasmin as a viable, even preferable, thrombolytic agent. Plasmin requires catheter delivery to achieve thrombolysis, but this technical issue has been solved with modern technology and widespread presence of interventional suites. After local administration, plasmin will lyse thrombi; thereafter, any plasmin in the circulation will be rapidly neutralised. Pre-clinical studies confirm that plasmin has marked haemostatic safety advantage over t-PA. After more than 50 years, the field has come full circle, and plasmin as the thrombolytic agent and catheter use for local delivery of agent may represent a step forward in thrombolytic therapy.
The modern era of fibrinolysis research and clinical development began in 1933 with the serendipitous discovery by Tillett and Garner  of a fibrinolytic component contained within a broth culture of haemolytic Streptococci. Tillett moved from Johns Hopkins University to New York University, and in 1949, he and Sol Sherry administered this fibrinolytic component, called streptokinase (SK), to patients – not for vascular thrombosis, but rather into the pleural space for treatment of fibrinous adhesions, often with strikingly beneficial results . A very active interest also developed for the therapeutic potential of plasmin, at the time called ‘fibrinolysin’, and by 1957, Cliffton’s  group had administered this preparation to patients with all manner of arterial and venous thrombotic disease. Questions arose regarding the agent of choice, either a plasminogen activator (PA) or the active enzyme plasmin, and there was also considerable interest in the best route of administration for therapy, either by direct infusion into a thrombosed vessel or systemically by the intravenous (IV) route. In 1960, Boyles et al.  showed that plasmin recanalised the occluded canine coronary artery when administered locally, but not when administered intravenously. In retrospect, this observation explains the variable pre-clinical and clinical results of IV-administered plasmin, but the reasons for these conflicting data were not evident until later.
Milestones of fibrinolysis research
It is appropriate to recount the biochemical, pre-clinical and clinical milestones that have brought us to our current understanding of fibrinolysis, but finding consensus for what were the important events, and especially for what the future holds is problematic. As to the published historical narratives (Table 1), those of Koller  in 1960 and Macfarlane  in 1964 could be called ‘dispassionate overviews’, while the historical summaries of Cliffton in 1960 , Sherry in 1981  and 1989  and Braunwald in 2002  could be titled ‘personal narratives’. Other narratives were as much directed to a specific agent, often recombinant t-PA (rt-PA), as to the person’s involvement in events [11–15], or on a specific therapeutic target, for example, the coronary artery [16,17], or the perspective of a specific medical specialty, such as one devoted to vascular intervention [18,19].
Table 1. Types of published historical accounts of fibrinolysis research
Perhaps the field of fibrinolysis originated with Morgagni in 1761  noting that blood was uncoagulable after sudden death, with Denis in 1838  observing spontaneous clot dissolution, and Denys and De Marbaix in 1889  postulating a dormant blood fibrinolytic enzyme, but the phenomenon of post-mortem fibrinolysis was exploited by Skundina et al.  and by Yudin  in the mid-1930s to provide a source of unclotted blood for transfusion. The term ‘fibrinolysis’ was coined by Dastre in 1893 , but the modern era of fibrinolysis began with the observations of Tillett and Garner in 1933 and 1934 of the fibrinolytic potential of ‘streptococcal fibrinolysin’ [1,26,27]. Many investigators, including Milstone , Tagnon , Christensen [30,31], Astrup and Permin [32,33], Mullertz [34,35], Williams , Sobel et al. , Lack , Ratnoff et al. , Kline , Guest  and others elucidated and purified fibrinolytic system components (Table 2A). A balanced scheme of fibrinolytic activation and inhibition evolved, described eloquently by Macfarlane and Biggs in 1948 , Macfarlane being the same person who along with Davie and Ratnoff modernised our concept of coagulation as a cascade  or waterfall , depending on which brand of English one uses.
Preclinical studies assessed the potential of thrombolytic agents to recanalise occluded vessels in animal models. In 1952, Johnson and Tillett  showed that a rabbit ear vein clot could be dissolved by IV-administered SK, and groups led by Cliffton [46,47] and Ambrus  in the mid-1950s showed encouraging results with IV ‘fibrinolysin’. In an impressive experiment that compared IV vs. intra-arterial (IA) injection of ‘fibrinolysin’ in canine models of coronary and cerebral artery occlusions, Boyles et al. in 1960  showed that only local IA administration was successful. Their conclusion was in conflict with a concurrent study that reported successful lysis of canine coronary artery thrombi after IV administration . These glaringly discrepant results regarding the efficacy of IV-‘fibrinolysin’ are explained as follows: failure of ‘fibrinolysin’ would be due to neutralisation by antiplasmin , and successes would be due contamination of ‘fibrinolysin’ with PA (SK) .
In a prescient clinical experiment reported in 1960, Boucek and Murphy  devised a catheter method of delivering ‘fibrinolysin’ into the aortic root of eight patients with myocardial infarction (MI), effectively injecting agent into the coronary arteries. They noted that ‘segmental arterial infusion appears to offer distinct advantages’, but catheter technology in 1960 was not sufficiently developed for routine thrombolytic use, especially as concerned time-sensitive pathologies such as acute MI and acute ischaemic stroke. Studies of the same era tried IV ‘fibrinolysin’ for various thrombotic disorders [3,52], but in retrospect, it is unlikely that the supposed active agent (plasmin) ever reached or dissolved the target thrombus.
Two issues have emerged, regarding the choice of agent (plasminogen vs. plasmin) and route of administration (systemic vs. local) (Fig. 1). As of 1960–1970, plasmin (as exemplified by ‘fibrinolysin’) was discredited, knowing that it was rapidly inhibited upon IV administration  and that its beneficial thrombolytic activity was likely due to contaminating PA . Furthermore, a powerful foundation for the superiority of PA over plasmin was formulated by Sherry in 1954  and by Alkjaersig, Fletcher and Sherry in 1959 , based upon two principles. First, plasmin is neutralised by antiplasmin after IV administration, a reaction that was later shown to be extremely rapid and irreversible . Second, PA (in sufficient dosage) circulates after IV administration, activating thrombus plasminogen to plasmin, which degrades fibrin in an environment protected from antiplasmin.
This scheme provided a foundation for IV administration of PA, a situation that had been exploited in humans in 1959 for lysis of experimental venous thrombi , and which was especially suited for treating acute MI, for which the St. Louis group assessed SK therapy in 1958 . Although better lysis may have been possible with IA administration, there was no feasible catheter technology or support facility available at the time and trials of thrombolytic therapy focused on IV administration of PA.
Meanwhile, vascular surgeons initiated a new era of catheter-directed therapy (CDT), starting with the landmark study by Dotter et al. in 1974  that showed successful lysis of a peripheral arterial thrombosis by IA SK. This report noted a somewhat surprising and still largely overlooked occurrence, bleeding complications despite ‘local’ therapy. We now know that CDT improves thrombolysis, but this delivery route does not restrict PA presence to the site of thrombus. Bleeding at remote sites still occur, just as with systemic therapy.
Major clinical trials (1980–2000)
The era of multicentre study of thrombolytic therapy was initiated by studies of pulmonary embolism coordinated by Sherry and colleagues, reported in the early 1970s. Results showed better recanalisation with UK than with heparin  and equal results with UK and SK , most evident in patients with symptom onset of < 48 h .
From 1980 to the present time, there has been an explosion of work in thrombolytic therapy, and successful demonstrations of relevant clinical success, all with PA, delivered both systemically and CDT.
Thrombosis was definitively established as the central pathologic event of acute MI by DeWood et al. in 1980 . With this foundation, Boucek and Murphy’s  concept of ‘segmental perfusion of coronary arteries’, expounded in 1960, was exploited by intra-coronary infusion of PA by Chazov et al. in 1976  and by Rentrop et al. in 1979 . Lysis of coronary artery thrombi was documented in real time, and the potential of IV administration to decrease delay from event to treatment was shown by Schroder et al. in 1983 . The development of tissue-type PA (t-PA) coincided with these remarkable observations, with ultimate purification and synthesis in recombinant systems by the efforts of Rijken and Collen  and Pennica et al. , demonstration of efficacy in an animal model of pulmonary embolism by Matsuo et al.  (Table 2B), first use of t-PA in humans , and IV-administered rt-PA in patients with acute MI . An enormous number of patients with acute MI (> 100 000) were subjected to clinical trial of IV PA vs. placebo from 1986 to 1988, showing survival advantage for SK [71,72], rt-PA  and anistreplase . Head-to-head comparative trials showed equivalent mortality for SK and rt-PA in two studies (GISSI-2 AND ISIS-3) [75,76] and an advantage favouring rt-PA over SK (6.3% vs. 7.2%) in a third trial (GUSTO) . Although there was controversy regarding interpretation of these comparative trials , the studies clearly document the strong survival benefit of thrombolytic reperfusion. Mortality benefit is greatest if initiated in the first 2 h (40% reduction), less so at 6 h (25%), but still with measurable at 12 h . A similar effort to establish clinical benefit of PAs in acute ischaemic stroke was expended in the 1990s, with significant functional improvement established for rt-PA when administered IV in the 0–3 h time window . There was a suggestion of similar effect with SK by subgroup analysis of patients treated in the 0–3 h period , but study results did not meet statistical significance . Subsequent study of rt-PA in acute stroke has extended the time window to 4.5 h . In the early 1980s, CDT with SK or UK was used increasingly [84,85] for acute peripheral arterial (graft) occlusion. In 1994, Ouriel et al.  showed that IA CDT with UK (vs. early surgical repair) for occluded peripheral arterial graft occlusion showed a dramatic decrease in the need for invasive surgery and more importantly, a decreased 1-year mortality (16% vs. 42%). Subsequent comparison of UK with t-PA showed no difference in clinical outcome .
Persistent problems: limits of efficacy and bleeding complications
Significant problems with PA therapy persisted, specifically, practical limits of efficacy and bleeding complications, especially intracranial haemorrhage (ICH) (Fig. 2). Efforts to improve recanalisation and clinical results include CDT with or without other recanalisation techniques, ultrasound  and micro-vesicle tools, adjunctive pharmacologic agents (anticoagulant and anti-platelet) and novel PA such as recombinant derivatives of rt-PA , staphylokinase , and recombinant bat salivary derivative desmoteplase  (Table 3). An immense literature attests to these efforts, and a review of anticoagulants, anti-platelet agents and novel PA, whether recombinant varieties of t-PA or of bacterial or haematophagous origin are not the subject of this review. Rather, this narrative will focus on the confluence of two developments, namely, the progress made using catheter technology and a renewed consideration of plasmin as a potentially safer thrombolytic agent than PA.
Table 3. Approaches to improve efficacy of plasminogen activator therapy
Catheter-directed thrombolysis (CDT)
PA infusion alone
PA infusion plus other recanalisation approaches (thrombectomy, angioplasty, stenting)
Anticoagulants (glucose-aminoglycans, direct anti-thrombins)
Anti-platelet agents (aspirin, ADP and IIb/IIIa inhibitors)
Improved recanalisation with CDT and endovascular procedures
In the decades since the first documented advantages of local delivery of agent in 1960 , CDT has become the preferred approach for the management of arterial and venous thrombosis, with referral to the Interventional Radiology suite now routine for agent delivery and for combined mechanical recanalisation approaches.
Thrombolytic therapy for acute ischaemic stroke is currently limited to a small portion of patients [91,92] during a window of opportunity of up to 4.5 h [80,83]. To maximise cerebral artery recanalisation, a ‘multimodal reperfusion therapy’ approach may be applied  wherein all potential advantages provided by catheters can be brought to bear, including IA administration, thrombus disruption and extraction, angioplasty and stent deployment. To this end, IA administration of PA induces significant vascular recanalisation , clot retrieval devices have been deployed at up to 8 h after symptom onset , including one that also serves as an expandable stent , and micro-bubble infusion has been used as adjunct to IA-rt-PA as rescue for occluded cerebral arteries . Recanalisation is also accelerated by transcranial ultrasonography administered during systemic rt-PA therapy .
Controlled trials in patients with acute MI show that early coronary artery angioplasty provides superior clinical benefit than is achieved by IV rt-PA, provided that patient transfer to a specialised facility can be accomplished efficiently (< 2 h) [99,100]. Of special value is the decreased risk of symptomatic ICH (sICH), the report by Grines et al.  showing a 2% incidence in 200 t-PA-treated patients vs. 0% in 195 patients who underwent angioplasty (P = 0.05).
Significant (> 90%) thrombolysis can be achieved in about 50% of patients with deep vein thrombosis (DVT) treated with only IV PA [101,102]. However, current practice often combines CDT with ‘minimally-invasive endovascular strategies’ , and observational studies using CDT show high patency rates of up to 70% [104,105], with 75% of occluded vessels also subjected to angioplasty, stent placement, mechanical thrombectomy and/or surgery . Although CDT has not been compared with IV PA alone, CDT has been compared with heparin alone for ileofemoral DVT, reporting a superior 6-month patency rate of 64% (vs. 36% for heparin) . The endovascular therapies provided in addition to PA likely explain the very high patency rates with CDT [105,107]. More thrombolysis likely translates to fewer cases of post-thrombotic syndrome , and the additional endovascular strategies may be especially useful for older occlusions that are not amenable to thrombolysis, and for conditions with intractable extravascular compression, such as the May-Thurner syndrome .
Bleeding complications with PA therapy
Every thrombolytic agent currently approved by federal regulatory agencies for treatment of pathologic thrombi is a PA , and haemorrhagic complications have been and continue to be the most dangerous adverse events associated with PA use . CDT has improved the vascular recanalisation results achieved with PA but the agent is not restricted to the thrombus locale, especially as vessel recanalisation occurs. Thus, bleeding complications are anticipated, based upon lysis of susceptible haemostatic plugs at sites of vascular injury [110–113]. Despite the varied molecular structure and source of PA, they all have in common the properties of (i) capacity to circulate to sites of thrombosis or haemostatic plug presence despite plasma inhibitors, and (ii) conversion of plasminogen to plasmin to lyse a thrombus or haemostatic plug . Not surprisingly, adjunct anticoagulant and anti-platelet therapy exaggerate the bleeding risk . Among the PA, more potent agents such as t-PA and its recombinant variants carry a greater risk of inducing ICH after IV administration than does SK . This effect is clearly documented in comparative trials of IV PA in patients with acute MI, with rates of 0.7% for t-PA vs. 0.35% for SK in GISSI-2 and ISIS-3 [75,76].
The risk of inducing sICH by PA use is greatest in the treatment of ischaemic stroke, exemplified in the pivotal NINDS study that showed a 10-fold greater rate for rt-PA-treated patients (6.5% vs. 0.6% for placebo) . The pathologic causation of sICH is multifactorial, involving ischaemic vasculopathy and loss of microvascular integrity, leading to extravasation of blood and parenchymal injury , to a great degree initiated directly by toxic effects of t-PA on the blood-brain barrier and neural cells . The data suggest that sICH occurs at the same rate at 0–3 and 3–4.5 h [116,117], albeit still 10-fold more often than with placebo . The incidence of ‘any’ ICH is also higher for t-PA than for placebo (27% vs. 17.6%, P = 0.001) , a worrisome increase, as asymptomatic ICH (haemorrhagic transformation rather than parenchymal haemorrhage) may be a harbinger of poor clinical outcome . Advanced age (> 80 years) does not necessarily connote a higher risk of sICH . However, IV-rt-PA is associated with a lower rate of sICH (5.2%) than IA-rt-PA (12.5%) or combined IV + IA rt-PA (20%) . Recent reports show sICH for tenecteplase (0.4 mg kg−1) at 15.8% (vs. 3.2% for rt-PA)  and of desmoteplase (90 or 125 μg kg−1) at 3.5% and 4.5% (vs. 0% for placebo) , reinforcing the association of sICH with any PA used for stroke treatment.
Although CDT using PA is an effective approach for acute peripheral arterial and graft occlusion, serious haemorrhagic complications occur not-infrequently. In the TOPAS report, major haemorrhage was noted in 12.5% of 256 patients who received recombinant UK, including four episodes of ICH (1.6%)  and similar rates of major haemorrhage and ICH were reported for the STILE  and ‘Rochester’  reports. A retrospective compilation of experience in 125 cases of peripheral vascular disease (90% of which were arterial or graft occlusions) from 2005 to 2008 showed haemorrhagic complications in 22.4%, a 5.6%‘stroke rate’, and mortality because of haemorrhagic complications in 4 (3%) of patients .
Therapeutic success with CDT for DVT without a high rate of major haemorrhage has been reported as only two of 53 (3.8%) , three of 50 (6%) , and seven of 178 (3.9%) . However, serious bleeding is still a significant limitation to PA use, as shown by six of 30 (20%) patients with major haemorrhage or large haematomas requiring cessation of treatment , significant haematoma formation in up to 11% of published studies , and a persistent occurrence of fatal ICH in one of 68 (1.5%)  and one of 103 (1%) patients . About 10% of patients with pulmonary embolism who received PA therapy had in-hospital bleeding complications  and 21.4% of such patients receiving CDT had ‘treatment-related haemorrhagic complication’ . The overall risk of serious bleeding is less in patients with DVT than in patients with peripheral arterial thrombotic disease, but bleeding nevertheless represents a clinical complication of significant degree that warrants search for an improved (safer) thrombolytic agent.
The re-discovery of plasmin and its proposal for use as a safer thrombolytic
A proposal made in 2001 by Marder et al.  on the basis of pre-clinical data, supported by editorial commentary , stated that plasmin administered by catheter could have the desired pharmacologic characteristics of both thrombolytic efficacy and haemostatic safety (Fig. 3).
1For efficacy: Plasmin delivered by catheter binds to fibrin, and protected from inhibition by antiplasmins, induces thrombolysis.
(i)As catheters are now almost ubiquitous for thrombolytic administration, plasmin can be utilised routinely.
2For safety: Plasmin that enters the circulation is neutralised by antiplasmin, so bleeding is prevented.
(i)Bleeding complications would still occur with rt-PA, even when administered by catheter.
Thus, the biochemical property that nullifies the thrombolytic effect of IV-administered plasmin [49,53] would be turned to advantage for haemostatic safety of catheter-administered plasmin . Furthermore, the wide use of CDT by the 1980s eliminated the technical limits to local plasmin use that existed in the 1950s and 1960s. As to the production flaw of ‘fibrinolysin’ that allowed contamination with PA , the plasmin used in current study is free of PA [132,134].
Pre-clinical studies that compared plasmin with t-PA have supported these hypothetical advantages for plasmin. Whereas rt-PA in thrombolytic dosages caused bleeding in an animal model of fibrinolytic haemorrhage , and in a dose-related effect, even in sub-therapeutic dosages , plasmin was free of haemorrhagic complications even at dosages up to 4-fold needed for vascular recanalisation. Purposely-toxic dosages of plasmin, which totally depleted plasma fibrinogen, induced prolongation of primary bleeding after vascular trauma, but so long as fibrinogen levels were not totally depleted, reasonable haemostasis was maintained . As for venous or arterial thrombolysis, pre-clinical models showed plasmin to be as effective as t-PA, and even with possible superiority under conditions of limited plasminogen presence [132,134]. Plasmin has also shown ex vivo thrombolysis equivalent to that for rt-PA for thrombi retrieved from the middle cerebral artery of patients with acute ischaemic stroke . This combined result of at-least equal efficacy to go with greater haemostatic safety reflects a more favourable benefit-to-risk ratio for plasmin over rt-PA .
Plasmin is the prototypic example of a group of fibrinolytic agents, classified as direct-acting, all of which exert their effect independent of plasminogen as substrate, but with differing properties for fibrin-binding and inhibitor neutralisation. These agents are arbitrarily divided into plasmin and related species and fibrino(geno)lytic molecules derived from haematophagous animals, details of biochemical, preclinical and clinical testing summarised in a recent review . Several agents have been or are under current clinical study, including microplasmin, composed of the serine protease domain alone , alfimeprase, a recombinant derivative of the Copperhead snake venom fibrolase , and full-length plasmin, prepared in a low pH formulation that activates upon exposure to neutral pH in plasma (Talecris Biotherapeutics, Inc, Research Triangle Park, NC, USA). Plasmin is currently in Phase 2 trial for peripheral arterial and graft occlusion [NCT NCT 00418483) and in a Phase 1 safety study for acute ischaemic stroke [NCT 01014975]. A novel recombinant plasmin molecule has been synthesised, Δ (K2-K5) plasmin, consisting of the first kringle of plasmin adjoined to the serine protease domain, with virtually the same biochemical  and haemostatic safety characteristics  as full-length, plasma-derived plasmin.
Plasmin (as an impure preparation called ‘fibrinolysin’) was first studied more than 50 years ago and found to be either ineffective or inferior to plasminogen activators. With the current widespread use of catheters to deliver thrombolytic therapy, plasmin has been re-discovered as a potentially valuable agent, with a firm pharmacologic foundation for effective and haemostatically safe application in thrombotic disease.
Disclosure of Conflict of Interests
V.J. Marder has received research grants and is a paid consultant for Talecris Biotherapeutics, Inc, Research Triangle Park, NC, USA.