It is now well established that ingestion of aspirin (acetylsalicylic acid) inhibits platelet aggregation, thereby conferring some degree of protection against thrombosis-mediated vascular events in high-risk patients. The basis for this protective effect derives from multidisciplinary studies by many investigators. The contributions of the author are reported below. The route that led to the early discoveries began with studies on patients with unexplained bleeding disorders. ‘Midway upon the journey of our life, I found myself…’[1] doing thromboplastin generation tests in a small laboratory of the Walter Reed Army Institute of Research in Washington, DC. In 1959, I had just completed my residency training in internal medicine and hematology in New York and received from my draft board a somewhat Orwellian invitation (the term used was ‘obligated-voluntary’) to enlist in the United States Army Medical Corps. I was assigned to the Department of Hematology at Walter Reed, then under the direction of Colonel William H. Crosby. Bill Crosby is a man of extraordinary talents and interests: he spent the early years of his retirement translating and publishing the major poems of Baudelaire. He encouraged the members of his department to pursue their own research interests, in addition to carrying out the more mission-oriented studies for which the Institute had been established. Bill had made important contributions to the fields of hemolytic anemias, hemoglobin, and iron metabolism and, having learned that I had spent a year during my hematology fellowship studying the interaction of haptoglobin with globin, assigned me to the hemoglobin and iron lab. He assigned a fellow draftee, Munsey Wheby, to be chief of the coagulation lab, a position I secretly coveted, having been attracted to the field in medical school by the lectures of Benjamin Alexander and Robert Goldstein. Hearing that Munsey, who is now president of the American College of Physicians, had a research interest in hemoglobin and iron metabolism, I asked whether he would be willing to switch laboratory assignments. He agreed to do so, thereby determining the next 40 years of my life. For this, I am grateful. The tests in the coagulation laboratory were then being performed by a super technologist, Jim Eichelberger, 6 feet 9 in (206 cm) tall, and late of the Harlem Globetrotters. Jim had introduced into the lab almost every coagulation test then known and so, neck extended, I had only to observe and question him for the next several months to complete my novitiate period as a clotter. Besides conducting research, the lab also served as the clinical coagulation laboratory for Walter Reed, and we thus had occasion to study a large number of patients, both military and civilian, who were referred to the Hospital because of bleeding problems. One of the tests that we used in these studies was the Biggs–Douglas thromboplastin generation test (TGT) [2] that measured the rate at which ‘thromboplastin’ (in all likelihood activated factor [F] X) was generated in a mixture of serum, adsorbed plasma, and washed platelets. Using a slight modification of the method for preparing the platelets, we found an abnormality in the platelet component of the test in some of the patients referred to us. This deficiency in the procoagulant property of platelets, then known as platelet factor 3, could also be demonstrated in a test that assessed the formation of thrombin from purified prothrombin. The latter was provided to me during one of our weekly string quartet sessions by Sandy Shapiro, who was at the National Institutes of Health at the time. A report of these patients with platelet factor 3 deficiency, then called thrombopathia, was published in the American Journal of Medicine[3]. Having completed my tour at Walter Reed, the time had come to consider my professional future, and I began looking for a staff appointment at several New York hospitals in anticipation of establishing a practice in internal medicine and hematology. When Bill Crosby heard that the reason for my decision was mainly acute research anxiety, he commanded me to pursue a research career. As I was still in the army, I decided to obey. For this, too, I am grateful. Returning to New York in 1962, first in Lewis Thomas's Department of Medicine at New York University, and then in Louis R. Wasserman's Department of Hematology at the Mount Sinai Hospital, I continued to study patients in whom the platelet component of the TGT was abnormal. The latter test, however, was cumbersome and I began to look for a simpler method for detecting defects in platelet factor 3. Roger Hardisty [4] and Ted Spaet [5] had recently shown that platelet factor 3 activity could be ‘made available’ by incubating platelet-rich plasma with kaolin. Utilizing a modification of their test, I found platelet factor 3 availability to be abnormal in two of my Walter Reed thrombopathia patients whom I had occasion to restudy, as well as in four newly discovered New York patients. Roger had also shown that an absolute requirement for making factor 3 available was the prior release from the platelets of adenosine diphosphate (ADP) [4], the platelet activating substance that had been recently identified by Anne Gaarder and coworkers [6]. Using this information, I explored the possibility that the impaired kaolin-induced platelet factor 3 availability in my thrombopathic patients might be due to a defect in the release of platelet ADP, and found this to be the case. In all of the patients, the impairment in kaolin-induced platelet factor 3 availability was associated with impaired ADP release, and was corrected by the prior addition of ADP. The studies were published in the American Journal of Medicine in 1967 [7]. While these studies were in progress, I had the extreme good fortune of learning from Marion Barnhardt in Detroit that an outstanding technologist, John Rogers, was looking for a position in New York. I quickly hired John, who remained with me for the next 30 years, and without whom many of my subsequent studies on platelets, von Willebrand factor (VWF), and FVIII, would have been difficult, if not impossible. An impairment of kaolin-induced platelet factor 3 availability and ADP release was an unlikely explanation for the prolongation of the bleeding time that was frequently observed in patients with thrombopathia, since the bleeding time is normal in most patients with coagulation defects. I became aware of reports by Torstein Hovig [8], and Ted Spaet and Marjorie Zucker [9], that platelets also released ADP, and aggregated, when exposed to an active component of connective tissue, later shown to be collagen. This raised the possibility that a defect in connective tissue-induced ADP release, analogous to that observed with kaolin, might be detected as impaired connective tissue-induced platelet aggregation. This also turned out to be case. Utilizing the method of platelet aggregometry that had recently been introduced by Gustave Born, and independently by John O'Brien, I found that connective tissue-induced platelet aggregation and ADP release were impaired in all of the patients with thrombopathia, but were normal in von Willebrand disease (VWD). Primary aggregation by ADP was also normal, thereby distinguishing these patients from those with Glanzmann thrombasthenia. These results were also published in the 1967 American Journal of Medicine[7]. The impairment of connective tissue (collagen)-induced platelet aggregation in these patients with thrombopathia, as in those reported independently by Roger Hardisty [10], suggested that an impaired response of their platelets to components of the vessel wall and surrounding connective tissue during the early stages of hemostasis might provide a better explanation for their prolonged bleeding time than the defect in platelet factor 3 availability, since the latter type of coagulation defect, as in hemophilia, would not be anticipated to affect primary hemostasis. Subsequently, I found that the impaired release of ADP in some of my thrombopathic patients was due to a decreased platelet content of this adenine nucleotide. In later studies done in collaboration with Holm Holmsen, with whom I also played string quartets, we found that this was associated with a general reduction in substances stored in platelet-dense granules. But these studies on storage pool deficiency are another story. At about the time I was studying patients with thrombopathia, Armand Quick had been publishing studies on the ‘aspirin tolerance test’[11]. Confirming previous reports by French investigators [12, 13], he reported that aspirin, in doses far below those required to affect the prothrombin time, not only prolonged the bleeding time in normal subjects, but produced a disproportionate prolongation in patients with VWD [11]. To explain these findings, he proposed a theory that aspirin induced in normal subjects the same kind of defect that accounted for the prolonged bleeding time in VWD. The findings that we had obtained in thrombopathic patients suggested an alternative hypothesis. Might the prolongation of the bleeding time in normal subjects who ingest aspirin, as in thrombopathic patients, be due to an impairment of platelet aggregation, possibly associated with defective ADP release? To test the above hypothesis, I designed an experiment to see whether the anticipated prolongation of the bleeding time after aspirin ingestion might be associated with an impairment of platelet responses, particularly connective tissue-induced aggregation and ADP release. A description of the methodology from the paper that was later published in the Lancet[with current comments in parentheses] read as follows: ‘10 normal male subjects [six of whom were physicians], age 25–40 years, received capsules, of identical appearance each containing either 0.3 g of aspirin or the same amount of lactose. Pretreatment studies were done 2 h after a light, fat-free breakfast on day 1. Paired subjects then took 10 capsules, in 3 divided doses, on days 1 and 2, and four capsules with breakfast on day 3, and the studies were repeated.’…‘2 weeks later, the drugs received by the paired subjects were reversed and the studies repeated, so that the 10 subjects each received both aspirin [as we now know, we could have used a lower dosing regimen] and the placebo. The entire study was conducted by the double blind technique’. (Ralph Zalusky dispensed the coded pills and provided the information to break the code at the end of the experiment.) The tests done included most of those I had set up in my laboratory, and included the bleeding time, platelet aggregation by connective tissue and ADP, platelet factor 3 availability, ADP release induced by connective tissue and kaolin, and platelet retention in glass bead filters, thought (incorrectly) at the time to be a measurement solely of platelet adhesion. Considering the possibility that aspirin might affect platelet ATP or ATPase, I asked Lou Aledort, and he kindly agreed, to assay them, utilizing methods he had recently learned during a fellowship with Bob Weed and Stan Troup in Rochester, New York. The first entry in the index page of the laboratory notebook, with my query, ‘does aspirin affect platelets’, and the entry on April 21, 1967 by John Rogers that initiated the laboratory studies (preparation of platelet-poor plasma substrate for the platelet factor 3 assay) are shown in Fig. 1. The first of two aspirin studies begins. I ask ‘does aspirin affect platelets’ on the index page of the lab book, and on page 1, dated April 21, 1967, John Rogers records the preparation of platelet-poor plasma from a normal subject (name removed) to be used as substrate for the platelet factor 3 assays prior to initiating the formal studies 3 days later. The results of the study are shown in Fig. 2, taken from the publication in the Lancet on September 2, 1967 [14]. In the 10 normal subjects, ingestion of aspirin resulted in significantly less platelet aggregation by connective tissue than after the placebo, and this was associated with impaired release of ADP. The bleeding time was also prolonged, confirming previous reports. Although kaolin-induced ADP release was also decreased after aspirin ingestion, this was not associated with an impairment of platelet factor 3 availability (I later found that the impairment of ADP release induced by both kaolin and collagen was significantly less in aspirin-treated normal subjects than in patients with thrombopathia [15]). No significant changes in other measured values were observed, including primary aggregation by ADP, platelet adhesion/retention, and platelet ATP/ATPase. The studies thus demonstrated that aspirin ingestion in normal human subjects produced the same type of platelet aggregation defect, associated with impaired ADP release, that I had observed in patients with unexplained bleeding disorders. Aspirin inhibits platelet responses. Compared with results obtained with a placebo, aspirin ingestion by 10 normal subjects resulted in impaired connective tissue-induced aggregation and adenosine diphosphate (ADP) release, and prolonged the bleeding time. Primary aggregation by ADP was unaffected. P-values calculated by paired Student's t-test. (Reprinted from [14], with permission.) At about the same time, Fraser Mustard, Marian Packham, Geoff Evans, and colleagues in Hamilton and Toronto, proceeding along an entirely different pathway, reported that anti-inflammatory agents, including aspirin, inhibited the aggregation of platelets obtained from several species of animals [16-18]. Could aspirin, and similarly acting drugs, be clinically useful antithrombotic agents? I wrote in the Lancet paper ‘the results suggest that these agents may have antithrombotic properties’[14]. In the summer of 1967, I began to pursue questions that had not been answered by the first study. The studies initiated at that time were published the following year in the Journal of Clinical Investigation[19]. They extended those published in the Lancet in showing that aspirin, besides inhibiting connective tissue (collagen)-induced aggregation and the release of ADP (and ATP as well), also inhibited the secondary aggregation responses that are observed with agonists such as epinephrine and ADP. Similar results were published in 1968 by Marjorie Zucker, who showed that aspirin also inhibited the release of platelet serotonin [20]. We also found that ingestion of sodium salicylate, unlike aspirin, did not inhibit platelet aggregation or ADP release, consistent with Quick's observation (which we confirmed) that it had no effect on the bleeding time [11]. Similar findings were published the same year by John O'Brien [21], and studies published several years later from Aaron Marcus's lab demonstrated the uptake into platelet fractions of aspirin radiolabeled in the acetyl, but not the carboxyl group [22]. The findings from these several studies suggested that the acetyl group was somehow involved in the mechanism by which aspirin inhibited platelet aggregation. In order to determine how long the aspirin effect lasted, we followed connective tissue-induced platelet ADP release for 10 days after ingestion of a single dose of aspirin. A decrease in ADP release, compared with control values, was observed when first studied 2 h after aspirin ingestion. The abnormality persisted on the second and third day when salicylate was no longer detectable in the subject's blood, and did not completely disappear until 4–7 days after aspirin ingestion. Thus, we concluded, ‘the defect produced by aspirin occurred rapidly and persisted for a period roughly equal to the platelet life span, suggesting an irreversible effect on the platelet’. I presented the findings of this, my second aspirin study, at the International Society of Hematology Congress in New York in 1968. The presentation was scheduled to be the last paper on the last day of the Congress, but to my pleasant surprise, the seats were fully occupied. I reiterated my previous suggestion that aspirin might be a useful antithrombotic drug, and my next efforts were directed at exploring this possibility. Fraser Mustard's group had shown that aspirin reduced the amount of deposit formed in extracorporeal shunts in rabbits [18], and I decided to see whether it might protect against an experimental model of arterial thrombosis. In exploring this possibility, I was fortunate to enlist the collaboration of two excellent vascular surgeons, Callisto Danese and Choudary Voleti. The model we decided upon was to injure, either by endarterectomy or chemical means, segments of the common carotid and femoral arteries of dogs, and to measure subsequently the degree of thrombosis that was induced in these segments. The results, published in Thrombosis et Diathesis Haemorrhagica in 1971, were dramatic [23]. As seen in Fig. 3, aspirin ingestion reduced the incidence of total occlusions by either endarterectomy or chemical injury from 43% and 29% to 11% and 2%, respectively, although it did not completely prevent the deposition of some thrombotic material within the vessel segments. These studies provided experimental evidence that aspirin might be a useful antithrombotic agent, and we concluded ‘Although no clinical implications can necessarily be drawn from the findings in the present study, the results would appear to support the previously discussed possibility that aspirin has antithrombotic properties that may be useful in preventing arterial thrombosis in man, and clinical trials with this ubiquitous and relatively non-toxic drug appear warranted’. Aspirin protects against the development of arterial occlusions. Segments of the internal carotid or femoral artery of dogs were injured by chemical means or endarterectomy. Administration of aspirin significantly inhibited the occurrence of complete occlusions. (Adapted from [23], with permission.) In 1969, I moved from Mt Sinai to the Columbia-affiliated Roosevelt Hospital in New York, having been recruited by Nicholas P. Christy, the Chairman of Medicine, to direct the clinical and research program in hematology at the Hospital. Nick was extremely generous in providing me with the facilities and initial support that enabled me to continue my research efforts on platelets, VWF, and FVIII. Having shown that aspirin inhibits platelet aggregation and experimentally induced arterial thrombus formation, it remained to be seen whether it could inhibit the deposition of platelet thrombi, derived from human blood, on an injured vascular surface under arterial flow conditions. To perform these studies, I was fortunate to be engaged, at the time, in collaborative studies with Hans Baumgartner, who had developed in Basel a unique technique for studying the interaction of platelets with the subendothelial surface of rabbit aorta. In this technique, citrated or directly sampled human whole blood was perfused through a flow chamber (containing vessel segments) under flow conditions that are similar to those in arteries. Tom Tshopp was then working in our laboratory and, having previously worked with Hans, returned briefly to Basel to learn the technique. After it was introduced into our laboratory, we were able, with Hans, to demonstrate the impairment in platelet adhesion that accounts for the hemostatic defect in VWD and the Bernard–Soulier syndrome. The technique thus seemed ideal for studying the effect of aspirin on platelet–subendothelium interactions under flow conditions that simulated those in the arterial circulation. We found that aspirin ingestion by human subjects did not affect platelet adhesion, but inhibited subsequent platelet–platelet cohesion; thrombus formation was thereby decreased [24] (Fig. 4). These studies provided further support for its potential clinical use as an antithrombotic agent. Aspirin inhibits platelet thrombus formation. Citrated blood from normal subjects was exposed to everted, de-endothelialized rabbit aorta at arterial shear rates. Shown is a representative finding of platelet interaction with the subendothelium (A) before aspirin, and (B) 2.5 h after ingesting 0.9 g. (Reprinted from [24], with permission.) The basic mechanism by which aspirin inhibits platelet aggregation was elucidated in a series of seminal studies in the 1970s that have been extensively reviewed, and will only be briefly summarized. Studies in 1971 by John Vane [25], and independently by Brian Smith and Jim Willis the same year [26], demonstrated that aspirin inhibits the transformation of arachidonic acid to prostaglandin E2 (PGE2) and PGF2α. In 1973, Jim Willis and Doug Kuhn demonstrated that aspirin inhibited the formation of a labile aggregation-stimulating substance (LASS) that is formed when arachidonic acid is incubated with prostaglandin synthetase preparations from seminal vesicles [27]. Bengt Samuelsson's group in Stockholm showed that the active substances in these incubates were endoperoxide intermediates (PGG2 and PGH2) that induced platelet aggregation, and whose formation was inhibited by aspirin [28]. They then went on to demonstrate that the final product of arachidonate transformation that could induce platelet aggregation, and whose overall synthesis was blocked by aspirin, was an unstable derivative of endoperoxides which they named thromboxane A2[29, 30]. Hence, aspirin inhibited platelet aggregation by inactivating the enzyme system (cyclooxygenase) that converts arachidonic acid to PGG2 (H2), the precursors of thromboxane A2. The demonstration by Gerry Roth and Phil Majerus that aspirin irreversibly acetylates the cyclooxygenase enzyme, thereby inactivating it, completed this part of the story [31]. Further details on the mechanism by which aspirin inhibits cyclooxygenase were elucidated some years later, and have been reviewed [32]. Finally, the inhibition by aspirin of thromboxane-dependent platelet aggregation (such as that induced by epinephrine and ‘low’ concentrations of thrombin or collagen) has the effect of also inhibiting the aggregation-dependent release of ADP. Thus, the overall effect of aspirin on platelets is to inhibit both thromboxane-dependent aggregation responses, and also the amplification of these responses by released ADP [33]. Some of the investigators to whom I have referred in this review are shown in Fig. 5. Some of the players. A fragment of the photograph taken at the Gordon Conference on Hemostasis in Proctor, New Hampshire, June 10–14, 1974 shows some of the investigators referred to in this review. (1) T. H. Spaet, (2) the author, (3) H. R. Baumgartner, (4) M. A. Packham, (5) A. J. Marcus, (6) P. W. Majerus, (7) J. B. Smith, (8) M. B. Zucker, (9) J. F. Mustard. The benefit of aspirin in offering protection against thrombosis-mediated clinical events has now been well established. However, I found some years ago that the uniqueness of my suggestion to this effect in 1967 had been somewhat blunted by an observation made over a decade before. I discovered this quite by accident. Sometime in 1971 or 1972, I gave a talk in Ann Arbor, Michigan on platelet disorders, ending with my usual pep talk about the possible use of aspirin as an antithrombotic agent. On the way back to the airport, I shared a taxi with a physician who had attended my talk. After telling me how much he had enjoyed it, he asked if I was aware of an article reporting the use of aspirin in preventing heart attacks, published about a decade before, by an author whose name he could not remember. He also could not remember where the article had been published but, in true Yankee spirit, thought it was in ‘some Southern journal’. Determined to trace its provenance, I mentioned this to our hospital librarian, one Winifred Lieber, who was both a librarian's librarian and, as it turned out, a contemporary of Marjorie Zucker at Vassar College some years before. The challenge proved to be irresistible, and an hour later I received a telephone call from Miss Lieber telling me that she had found a reference to the article, which was by Dr Lawrence L. Craven, entitled ‘Experiences with aspirin (acetylsalicylic acid) in the non-specific prophylaxis of coronary thrombosis’, and published in the Mississippi Valley Medical Journal in 1953 [34]. (The ‘Southern’ journal was actually published in Quincy, Illinois, about 100 miles due west of Abraham Lincoln's home.) As our library did not carry this journal, Miss Lieber requested a reprint of the article from the New York Academy of Medicine Library, and it arrived a few days later. Dr Craven, a general practitioner, reasoned that since aspirin ingestion was known to cause bleeding in some patients, it might have anticoagulant properties. He reported that he had prescribed daily aspirin to 1465 healthy male subjects, mainly between the ages of 45 and 65 years, ‘who were overweight and known to lead a sedentary life’. He added that they returned at regular intervals, ‘and not one of them developed coronary occlusion or coronary insufficiency’, although he was perplexed by the effectiveness of the relatively ‘small’ dose (325–650 mg), which was less than that usually associated with a prolongation of the prothrombin time. A footnote to the paper indicated that it had won third (my italics) prize in the 1952 Mississippi Valley Medical Society Essay Contest. His report was a model of clinical insight, proven to be correct by later studies, and I cited it in my next [35], and subsequent reviews on antiplatelet drug therapy. My aspirin journey began with platelet factor 3, and in some ways ended with it. I had begun looking for possible platelet aggregation defects in patients with thrombopathia (the background for the aspirin studies) because I conjectured that the defect in platelet procoagulant activity was an unlikely explanation for their prolonged bleeding time. In 1973, the evidence to support this hypothesis was provided when Mary Ann Scott was referred to me for work-up of an undiagnosed bleeding disorder. Mrs Scott, whose disorder now bears her name, proved to have a selective defect in platelet procoagulant activity (and completely normal platelet aggregation responses), the basis for which has been the subject of many reports [36]. Her bleeding time was repeatedly normal. Based on studies using the Baumgartner chamber, I have suggested that a drug which produces a ‘Scott-like’ platelet defect in phospholipid exposure might be a useful antithrombotic agent [36, 37]. Perhaps Dr Lawrence Craven knew of such a drug. Our current understanding of the inhibitory effects of aspirin on platelet function, and its role as an antithrombotic agent, derives from the studies of many investigators encompassing a wide variety of disciplines. My own contributions began with studies of patients with poorly understood bleeding disorders. This route, proceeding from patient studies to basic discoveries to clinical applications, has led to many advances in the field of hemostasis and thrombosis. The term ‘bench-to-bedside’ has become the mantra for describing the pathway by which the benefits of research are transferred to clinical practice. Perhaps in some instances, ‘bedside-to-bench-to-bedside’ might be an equally appropriate description.