- Email: [email protected]
Safety of stringent prophylactic platelet transfusion policy for patients with acute leukaemia
Saturday
16 November 1991
No 8777
ORIGINAL ARTICLES
Safety of stringent prophylactic platelet transfusion policy for patients with acute leukaemia
Early studies suggested that the risk of haemorrhagic complications become unacceptable when platelet counts drop below 20 × 109/l. Because there are insufficient data to define 20 x 109/l as the threshold for prophylactic platelet transfusions, the practicability of a more restrictive transfusion policy has been assessed prospectively in 102 consecutive patients being treated for acute leukaemia. Besides platelet count, the transfusion protocol took into consideration factors such as presence of bleeding, fever, coagulation disorders, and intention to do therapeutic procedures. 31 major bleeding episodes occurred on 1·9% of the study days when platelet counts were 10 × 109/l or less and on 0·07% of study days when counts were 10-20 × 109/l. The findings indicate that the threshold for prophylactic transfusions can safely be set at 5 × 109/l in patients without fever or bleeding manifestations and at 10 × 109/l in patients with such For patients with coagulation disorders or anatomical lesions, or for those on heparin, the threshold should be at least 20 × 109/l. Such a restrictive platelet transfusion policy, which is applicable not only to thrombocytopenia associated with acute leukaemia but also to other forms of reduces hypoproliferative thrombocytopenia, of such to blood donors and exposure patients results in substantial health-care savings.
signs.
Introduction Data on the incidence of haemorrhagic manifestations of in relation to blood platelet counts in patients with acute leukaemia were first published in the early 1960s.1 That study suggested a substantial increase in
thrombocytopenia
haemorrhagic episodes at platelet counts below 20 x 109/1. With the development of aggressive chemotherapeutic
regimens for neoplasias the common practice of transfusing platelets at this level has increased the demand for platelets enormously and exposes patients to a large number of donors.2 However, recent studies do not indicate that this platelet level is critica1.3-5 Accordingly, in 1986 the US National Institutes of Health Consensus Development Conference on Platelet Transfusion Therapy recommended that the critical platelet level "might safely be below 20 x 109/1 for some patients".’’ Several uncertainties-due to scarcity of data on the safety of lower platelet thresholds in leukaemic patients, especially when they are receiving drugs that may interfere with the survival and/or function of transfused platelets, and on reproducibility of platelet counts at very low levels-may have contributed to the fact that in many institutions the platelet transfusion policy has not changed. Our adoption of a more restrictive platelet transfusion strategy as a basis for platelet alloimmunisation studies in leukaemia patients7,8 gave us the opportunity to examine prospectively the frequency of bleeding complications at various platelet levels among a representative population of patients being treated for acute leukaemia according to a stringent platelet transfusion
protocol. Methods
Study design In 1983 we started a prospective study on platelet transfusions in with newly diagnosed acute leukaemia.8 A standardised transfusion platelet protocol was used (table i). The analysis includes data on consecutive patients followed-up to Dec 31,1990. The patients were examined daily for evidence of haemorrhage as long as the platelet count was below 50 x 109/1; fundoscopy was a routine part of the examination. In addition all patients were scrutinised at least twice a week by one of us (J. G., J. B., U. S., or A. S.). Special attention was paid to the avoidance of aspirin or non-steroidal anti-inflammatory drugs affecting platelet function.
patients
ADDRESS: Department of Internal Medicine, University Hospital of Zürich, Switzerland (Prof J. Gmur, MD, J. Burger, MD, U. Schanz, MD, J Fehr, MD, A. Schaffner, MD). Correspondence to Prof J. Gmur, MD, Department of Internal Medicine, University Hospital, Ramistrasse 100, CH-8091 Zürich, Switzerland.
1224
TABLE I-PLATELET TRANSFUSION PROTOCOL
Platelet counting The platelet counting procedure was strictly adhered to. Platelets counted daily from entry until discharge, death, or attainment of stable values of > 50 x 109/1, whichever occurred first. The counts were made on edetic acid (EDTA) anticoagulated blood in a ’H*I’ flow cytometer (Technicon Instruments Corporation, Tarrytown, New York, USA). If the automated platelet count was below 50 x 10*’/I a manual platelet count was done on EDTA-blood diluted 1/10 in TIaxan’ (Sorinter SA, 1410 Waterloo, Belgium) and mixed for 3-5 min on a Coulter-roller (Coulter Electronics, Luton, UK). Platelets were counted manually at a magnification of 400 x in an improved Neubauer haematocytometer (5 x 16 fields) after all had been allowed to sediment for 20 min in a humidified petri dish. At counter values below 50 x 109/1 the manual count was always applied. If the first manual count and the automated count deviated by more than 20%, the manual value was verified by a second manual count. were
Paracetamol was used as an antipyretic. Standard antimicrobial therapy consisted of netilmicin and piperacillin; patients who did not respond were given vancomycin with ceftazidime or ciprofloxacin, and if there was still no response, amphotericin B was added. Haemoglobin concentrations were kept at 60-80 g/l by leucocyte-poor packed-red-cell transfusions. Induction chemotherapy consisted of 2-4 marrow-ablative courses according to national therapy protocols. Patients were followed up until hospital discharge, recovery of platelet counts to more than 100 x 109/1, or death, whichever occurred first. All 15 patients who died underwent necropsy.
Platelet transfusion protocol For platelet replacement, single-donor platelets collected on a blood-cell separator (Model V50, Haemonetics, Braintree, Massachusetts, or CS 3000, Baxter, Deerfield, Illinois, USA) were given. Random (non-HLA-typed) donors were used unless alloimmunisation necessitated HLA-matching (11/102 patients, 45/972 transfusions). Most of the 972 platelet preparations were given within 6 h of collection; after extended storage platelet packs became available. Storage for at most 72 h at room temperature on a continuous motion shaker was allowed (about 220/972 transfusions). The criteria for a platelet transfusion on the same day were based on the morning platelet count plus findings on careful clinical examination of the patient. 1 h posttransfusion platelet counts of less than 10 x 109/1 (35/972 transfusions 3-7%) were not routinely taken as an indication for further platelet transfusion on the same day (except for 13 instances in which such transfusions were given). Additional platelet counts were not done, nor were minor bleeding episodes taken into account until the next
morning. Definition of bleeding manifestations Minor
bleeding episodes (WHO grade 1) included any mucocutaneous haemorrhages or haematomas not requiring redcell transfusions, as well as retinal haemorrhages without impairment of vision. Major bleeding episodes included (1) melaena, haematemesis, haematuria, and haemoptysis whether or not blood transfusions were required (WHO grade >_ 2); (2) epistaxis and vaginal or soft tissue bleeding requiring red-cell transfusions (WHO grade 3-4); (3) fresh retinal haemorrhages with impairment of vision. TABLE II-FREQUENCY OF PLATELET TRANSFUSIONS ON SAMEE DAY AS PLATELET COUNT
Tx
=
platelet transfusion
Statistics Mean values were compared by Student’st test, and frequencies by X2 test or, for numbers below 75, by Fisher’s exact test. Linear regression was computed with a computer-based program (’Graph PAD’, San Diego, California, USA). Results are presented as mean values (standard deviation).
Results The study included 102 patients (mean age 41-8, range 15-71 years) with acute non-lymphocytic (n = 87) or acute lymphocytic (n = 15) leukaemia. All 7 patients with acute promyelocytic leukaemia (FAB M3) and 4 of the 80 with
other AML FAB subtypes presented with disseminated intravascular coagulation (DIC). 11 patients had haemoglobin concentrations of less than 6 g/dl and 13 had leucocyte counts greater than 100 x 109/1. 28 were febrile (> 38’0°C), 4 of them with overt septicaemia. Altogether, the patients were treated with 254 cycles of marrow ablative chemotherapy, during which they received a total of 972 platelet transfusions. The entire study time from hospital entry to the recovery of the platelet count to more than 100 x 109/1 or to death or hospital discharge, whichever occurred first, comprised 6002 patient-days.
Adherence to the platelet transfusion protocol The total number of days at different platelet counts, and the number of days with and those without a platelet transfusion on the same day as the low count, are shown in table 11. In accordance with the protocol, a transfusion was withheld on 1471 /2134 (69 %) of days when a morning count was 6-20 x 109/1. Adherence to the protocol was usually strict, with the following exceptions. Although all two hundred and eighty occasions on which morning platelet counts were 5 x 109/1 or less should have been followed by a transfusion on the same day, this procedure was not possible in fourteen instances (5%) because of alloimmunisation and unavailability of an HLA-identical platelet donor. At morning platelet counts of more than 5 x 109/1 platelet transfusions were given on the same day for reasons that were not as specified in the protocol on 191/5722 study days (3’ 3 %). Nearly half these premature transfusions were given to patients with morning platelet counts of 6-10 x 109/1 (89 transfusions, corresponding to 13% of days with a platelet count of 6-10 x 109/1). The reasons for these transfusions were: accelerated platelet consumption after the preceding transfusion (twenty-one instances), extensive minor haemorrhagic manifestations (20 instances), or a rising body temperature (forty-eight instances). At platelet counts of
1225
death is discussed below (patient 1). Epistaxis requiring tamponade and red-cell support occurred 4 times. 3 episodes of haemoptysis, 7 of gastrointestinal haemorrhage, and 3 of macrohaematuria were observed. 6 of these 13 patients needed red-cell transfusions (2-5 units, mean 3-2 units). The remaining 11 major bleeding episodes consisted of retinal haemorrhage with blurred vision that eventually resolved completely in 10. In the 11 th patient it consisted of vitreous body formation and retinal haemorrhage (platelet count 8 x 109/1, DIC, heparin therapy), with permanent impairment of vision.
complications during induction therapy 80 patients (78%) were discharged from the hospital with normal or self-sustained platelet counts of more than 50 x 109/1 (67 in complete and 13 in partial remission). In 7 patients with refractory leukaemia, severe thrombocytopenia persisted but further active therapy was no longer practical; 6 died at home from bleeding and/or infectious complications. The remaining 15 patients died during induction therapy. 12/15 non-bleeding complications resulted in death (septicaemia 2, systemic fungal infection 2, miliary tuberculosis 2, pneumonia 1, pulmonary embolism 2, and cardiovascular failure 3). Although necropsy revealed various bleeding manifestations in all of them, the bleeding Lethal bleeding
Relation of bleeding risk to blood platelet counts: computed per 1000 days at risk.
Open bars=minor bleeding complications Solid bars=major bleeding complications. Numbers in bars are numbers of observed days at risk in a given class The correlation coefficient between class 1-4 and the incidence of bleeding is r= 963, p < 01. Risk category platelets> 20 x 109/1 not included in correlation coefficient because the few bleeding complications in this group were related to factors other than solely the platelet count, and because the broad range of counts in this
11-15 times
x
category.
109/1 a premature platelet transfusion was given 83
(10% of days) because of accelerated platelet consumption after the preceding transfusion (25 instances), extensive minor haemorrhages (24 instances), high fever (20 instances), or no apparent reason (14 instances). At platelet count above 15 x 109/1 the protocol was violated in nineteen instances (0-5% of days).
Haemorrhagic manifestations before study entry Minor bleeding complications at hospital entry and before the first platelet transfusion were commonest at platelet counts of less than 20 x 109/1 (15/19) but occurred also at higher counts (at 21-50 x 109/1,14/28; at > 50 x 109/1, 20/55). Major bleeding complications were present before entry in 6 patients. 3 had retinal haemorrhages causing impaired vision (platelet counts 8, 10, and 15 x 109/1). 2 patients had WHO grade 3-4 vaginal bleeding at platelet counts of 18 and 48 x 109/1, respectively. A bleeding complication, which was eventually lethal, was misinterpreted as torticollis in 1 patient (case 3, see below; platelet count 65 x 109/1). Non-lethalhaemorrhagic episodes during induction chemotherapy 6 major bleeding complications following minor surgical interventions were not judged to be spontaneous haemorrhagic episodes. In 5 patients (platelet counts 36-87 x 109/1) a haemopneumothorax complicated the insertion of a subclavian catheter. Another patient had a haematoma of the scrotum and penis after inguinal lymph node biopsy (platelet count 14 x 109/1). After exclusion of these procedure-related bleeding complications we noted a progressive increase in the frequency of haemorrhage with declining platelet counts (figure). Minor forms of haemorrhage occurred much more frequently than did the more serious forms at all platelet count bands. Overall, 28 major non-lethal bleeding episodes were observed. A case of subdural haematoma complicated by cerebral oedema and jacksonian epilepsy contributing to
did not contribute to death. Death during induction therapy was related to or directly caused by bleeding complications in 3 patients. Patient 1 (51-year-old female, AML FAB M2) had a temporoparietal subdural haematoma after 2 days with a platelet of less than 1 x 109/1 because of refractoriness and lack of an HLA-identical donor. The patient died 7 days later from cardiac arrest due to electrolyte imbalance. Necropsy disclosed a flat (1 mm) residual subdural haematoma. Patient 2 (38-year-old man, AML FAB M5) was admitted with a white-cell count of 430 x 109/1 and platelet count of 55 x 109/1. Because of DIC (fibrinogen 0-3 g/1) he was treated with low doses of heparin and fibrinogen. 9 h later (platelet count 53 x 109/1, fibrinogen 0-4 g/1) he rapidly became comatose, and he died 15 h after admission from a cerebral haematoma with rupture into the ventricular system. Patient 3 (53-year-old female, AML FAB M3) complained of acute torticollis on admission (platelet count of 65 x 109/1, slightly raised concentration of fibrin degradation products). Low-dose heparin therapy was given for 7 days, during which the lowest platelet count was 35 x 109/1. Her neck pain worsened. 11 days after admission and 3 days after the third platelet transfusion (lowest pre-transfusion count 8x 109/1, 20 x 109/1) she suddenly became tetraplegic and died of respiratory arrest. Necropsy disclosed a fresh epidural and subdural haematoma of the spinal cord extending from C2 to C5. count
Clinical characteristics of patients with major haemorrhages The 28 major non-lethal and the 3 lethal bleeding episodes occurred in 23 patients. Their mean age was 42-2 (16 years), compared with 41 ’5(15) for the remainder of the population (p 085, by t-test). 10 were febrile (> 38-0°C) and 20 were on antibiotics; of these 7 also received amphotericin B. The 31 haemorrhagic episodes occurred at platelet counts of 1 x 109/1 to 65 x 109/1, with most being 10 x 109/1 or more (18/31). 4 episodes (1 lethal) occurred in alloimmunised patients left at platelet counts of less than 5 x 109/1 for 2-4 days because of unavailability of HLAcompatible platelet donors. In 4 other episodes endoscopically verified extensive mucous membrane lesions were the source of bleeding (ulcerative colitis and melaena in 2, platelet counts of < 5 x 109/1 in both cases; candida =
1226
oesophagitis and haematemesis in 2, platelet counts of 19 and 38 x 109/1). In patients with DIC and on low-dose heparin at the time of bleeding major bleeding episodes were observed in 3/18 cases with platelet counts of 10 x 109/1 and in 8/13 with higher platelet counts (p == 0-014, Fisher’s exact test). Several patients had more than one risk factor. Thus, additional risk factors for haemorrhage were obviously present in 16/31 of the major bleeding episodes and in 8/13 of serious bleeding complications observed at platelet counts above 10
x
109/1. Discussion
This study on haemorrhagic complications of thrombocytopenia during induction therapy for acute leukaemia shows that a significantly lower threshold than generally used for prophylactic platelet transfusions is safe and not associated with a higher bleeding risk if based on reliable platelet counts and appropriate clinical evaluation of the patient. This more restrained use of platelets results in considerable cost savings (about C350 000 in this study), reduces the rising demand for platelets imposed on blood banks, and prevents excessive exposure of cancer patients to platelet donors. Early retrospective studies in leukaemia patients established a strong relation between manual platelet counts and the frequency of haemorrhagic episodes, with bleeding occurring on 38% of days at counts of 10-20 x z/1,50% at counts of 5-10 x 109/1, and 65-92% at lower counts.1 Our results accord with this observation. However, the incidence of minor and major bleeding episodes was lower at all levels of platelet counts in our study than in earlier ones. The reasons for this difference are difficult to identify. A possible explanation might be that at the time of the older studies the effect of aspirin or other non-steroidal anti-inflammatory drugs on platelet function was not recognised, and so might have been liberally used, whereas they were strictly avoided in our study. The outstanding progress in antimicrobial therapy over the past three decades may also have contributed to the low incidence of bleeding in our patients, since infectious complications and/or high fever are wellaccepted causes of accelerated platelet consumption’,10 and increase the risk of major bleeding complications.3 Our results clearly show that a significantly lower threshold than that generally used is practicable for prophylactic platelet transfusions during induction chemotherapy of acute leukaemias. Minor bleeding episodes and almost all the major episodes did not seem to jeopardise outcome of therapy for our patients since a high response rate (78%) was obtained and lethal haemorrhages (3%) or severe sequelae of bleeding (1%) were infrequent. Many of the major bleeding episodes occurred at platelet above 10 x 109/1 and were related to additional risk factors such as anatomical lesions or DIC and/or heparin therapy. Thus, a liberal transfusion policy would hardly have prevented many of the cases of severe bleeding. With our restrictive transfusion strategy the mean number of platelet transfusions given per chemotherapy course was 3-8 (range 1-22 data not shown), which is appreciably lower than that reported elsewhere.2,11 It is conceivable that the lower number of transfusions (or donors), the lower is the risk of transfusion-transmitted diseases such as cytomegalovirus, hepatitis C, and possibly HIV infection. Whether reducing the number of platelet transfusions has a favourable effect on the incidence of alloimmunisation remains controversial.l1,12 We believe that the low rate of refractoriness observed in our study (10%) is counts
due to the use of single-donor platelets rather than to the lower number of transfusions.7,8 Although there is a tight correlation between platelet counts and the incidence of haemorrhage, other clinical factors that contribute to the risk of bleeding3,10,11 should be taken into account--eg, drugs that may interfere with platelet function. For this reason, clinical findings were always considered when deciding on time of platelet support. Our results show that the low threshold for prophylactic platelet support 5 x 109/1-in the absence of fresh haemorrhagic manifestations or fever (38-0°C), and 10 x 109/1 in the presence of such signs-is safe. To reduce bleeding complications at higher counts, the presence of coagulation disorders, heparin therapy, or major anatomical lesions should also be considered. Finally, although our results have been obtained in a uniform population of leukaemia patients, they might also be relevant for patients with hypoplastic thrombocytopenia due to other causes. .
REFERENCES Gaydos LA, Freireich EJ, Mantel N. The quantitative relation between platelet count and hemorrhage in patients with acute leukaemia. N Engl J Med 1962; 266: 905-09. 2. Silver SS, Rock G, Décary F. Use of platelet concentrates in eastern Ontario. Can Med Assoc J 1987; 137: 128-32. 3. Aderka D, Praff G, Santo M, Weinberger A, Pinkhas J. Bleeding due to thrombocytopenia in acute leukaemias and reevaluation of the prophylactic transfusion policy. Am J Med Sci 1986; 291: 147-51. 4. Slichter SJ, Harker LA. Thrombocytopenia: mechanisms and management of defects in platelet production. Clin Haematol 1978; 7: 523-27. 5. Soloman J, Bofenkamp T, Fahey JL, Chillar RK, Beutler E. Platelet prophylaxis in acute non-lymphoblastic leukaemia. Lancet 1978; i: 267. 6. National Institutes of Health. Consensus development conference on platelet transfusion therapy. JAMA 1987; 257: 1777-80. 7. Gmür J, von Felten A, Osterwalder B, Scali G, Sauter Chr, Frick P. Delayed alloimmunization using random single donor platelet transfusions: a prospective study in thrombocytopenic patients with 1.
acute leukemia. Blood 1983; 61: 473-79. 8. Gmür J, Burger J, Sauter Chr, Oelz O, Frick PG. Alloimmunization by leukocyte-rich or leucocyte-poor random single donor platelet transfusions. Prog Clin Biol Res 1990; 337: 45-48. 9. Murphy S, Litwin S, Herring LM, et al. Indications for platelet transfusion in children with acute leukemia. Am J Hematol 1982; 12: 347-56. 10. Bishop JH, McGrath K, Wolff MM, et al. Clinical factors influencing the efficacy of pooled platelet transfusions. Blood 1987; 71: 383-87. 11. Scott E, Steeger T, Huntington S, et al. Platelet utilization patterns at a large referral hospital. Transfusion 1985; 25: 469. 12. Dutcher JP, Schiffer CA, Aisner J, Wiernik PH. Alloimmunization following platelet transfusions: the absence of dose-response relationship. Blood 1981; 57: 395-99.
From The Lancet Treatment of mad wolves’ bites
Owing to the number of rabid wolves in Romania and the neighbouring countries Professor Babesch, Director of the Bacteriological Institute in Bucharest, and one of the ablest pupils of Pasteur and Koch, has had many opportunities of treating persons bitten by mad wolves by Pasteur’s method, unfortunately, however, without success, for the symptoms of the disease broke out much too quickly to admit of an effective cure by the application of lymph taken from the spinal cord of rabbits that had died of hydrophobia. In view of this sad fact, Professor Babesch has for some time past been making experiments with the purpose of ascertaining whether the immunity from the virus of hydrophobia conferred by Pasteur’s method cannot be transferred by inoculating the blood of animals thus rendered immune to other warm-blooded animals. These experiments showed that such inoculation had a prophylactic effect even in cases in which Pasteur’s spinal cord lymph, under otherwise quite identical conditions, produced no such results whatever.
(July 4,1891)
16 November 1991
No 8777
ORIGINAL ARTICLES
Safety of stringent prophylactic platelet transfusion policy for patients with acute leukaemia
Early studies suggested that the risk of haemorrhagic complications become unacceptable when platelet counts drop below 20 × 109/l. Because there are insufficient data to define 20 x 109/l as the threshold for prophylactic platelet transfusions, the practicability of a more restrictive transfusion policy has been assessed prospectively in 102 consecutive patients being treated for acute leukaemia. Besides platelet count, the transfusion protocol took into consideration factors such as presence of bleeding, fever, coagulation disorders, and intention to do therapeutic procedures. 31 major bleeding episodes occurred on 1·9% of the study days when platelet counts were 10 × 109/l or less and on 0·07% of study days when counts were 10-20 × 109/l. The findings indicate that the threshold for prophylactic transfusions can safely be set at 5 × 109/l in patients without fever or bleeding manifestations and at 10 × 109/l in patients with such For patients with coagulation disorders or anatomical lesions, or for those on heparin, the threshold should be at least 20 × 109/l. Such a restrictive platelet transfusion policy, which is applicable not only to thrombocytopenia associated with acute leukaemia but also to other forms of reduces hypoproliferative thrombocytopenia, of such to blood donors and exposure patients results in substantial health-care savings.
signs.
Introduction Data on the incidence of haemorrhagic manifestations of in relation to blood platelet counts in patients with acute leukaemia were first published in the early 1960s.1 That study suggested a substantial increase in
thrombocytopenia
haemorrhagic episodes at platelet counts below 20 x 109/1. With the development of aggressive chemotherapeutic
regimens for neoplasias the common practice of transfusing platelets at this level has increased the demand for platelets enormously and exposes patients to a large number of donors.2 However, recent studies do not indicate that this platelet level is critica1.3-5 Accordingly, in 1986 the US National Institutes of Health Consensus Development Conference on Platelet Transfusion Therapy recommended that the critical platelet level "might safely be below 20 x 109/1 for some patients".’’ Several uncertainties-due to scarcity of data on the safety of lower platelet thresholds in leukaemic patients, especially when they are receiving drugs that may interfere with the survival and/or function of transfused platelets, and on reproducibility of platelet counts at very low levels-may have contributed to the fact that in many institutions the platelet transfusion policy has not changed. Our adoption of a more restrictive platelet transfusion strategy as a basis for platelet alloimmunisation studies in leukaemia patients7,8 gave us the opportunity to examine prospectively the frequency of bleeding complications at various platelet levels among a representative population of patients being treated for acute leukaemia according to a stringent platelet transfusion
protocol. Methods
Study design In 1983 we started a prospective study on platelet transfusions in with newly diagnosed acute leukaemia.8 A standardised transfusion platelet protocol was used (table i). The analysis includes data on consecutive patients followed-up to Dec 31,1990. The patients were examined daily for evidence of haemorrhage as long as the platelet count was below 50 x 109/1; fundoscopy was a routine part of the examination. In addition all patients were scrutinised at least twice a week by one of us (J. G., J. B., U. S., or A. S.). Special attention was paid to the avoidance of aspirin or non-steroidal anti-inflammatory drugs affecting platelet function.
patients
ADDRESS: Department of Internal Medicine, University Hospital of Zürich, Switzerland (Prof J. Gmur, MD, J. Burger, MD, U. Schanz, MD, J Fehr, MD, A. Schaffner, MD). Correspondence to Prof J. Gmur, MD, Department of Internal Medicine, University Hospital, Ramistrasse 100, CH-8091 Zürich, Switzerland.
1224
TABLE I-PLATELET TRANSFUSION PROTOCOL
Platelet counting The platelet counting procedure was strictly adhered to. Platelets counted daily from entry until discharge, death, or attainment of stable values of > 50 x 109/1, whichever occurred first. The counts were made on edetic acid (EDTA) anticoagulated blood in a ’H*I’ flow cytometer (Technicon Instruments Corporation, Tarrytown, New York, USA). If the automated platelet count was below 50 x 10*’/I a manual platelet count was done on EDTA-blood diluted 1/10 in TIaxan’ (Sorinter SA, 1410 Waterloo, Belgium) and mixed for 3-5 min on a Coulter-roller (Coulter Electronics, Luton, UK). Platelets were counted manually at a magnification of 400 x in an improved Neubauer haematocytometer (5 x 16 fields) after all had been allowed to sediment for 20 min in a humidified petri dish. At counter values below 50 x 109/1 the manual count was always applied. If the first manual count and the automated count deviated by more than 20%, the manual value was verified by a second manual count. were
Paracetamol was used as an antipyretic. Standard antimicrobial therapy consisted of netilmicin and piperacillin; patients who did not respond were given vancomycin with ceftazidime or ciprofloxacin, and if there was still no response, amphotericin B was added. Haemoglobin concentrations were kept at 60-80 g/l by leucocyte-poor packed-red-cell transfusions. Induction chemotherapy consisted of 2-4 marrow-ablative courses according to national therapy protocols. Patients were followed up until hospital discharge, recovery of platelet counts to more than 100 x 109/1, or death, whichever occurred first. All 15 patients who died underwent necropsy.
Platelet transfusion protocol For platelet replacement, single-donor platelets collected on a blood-cell separator (Model V50, Haemonetics, Braintree, Massachusetts, or CS 3000, Baxter, Deerfield, Illinois, USA) were given. Random (non-HLA-typed) donors were used unless alloimmunisation necessitated HLA-matching (11/102 patients, 45/972 transfusions). Most of the 972 platelet preparations were given within 6 h of collection; after extended storage platelet packs became available. Storage for at most 72 h at room temperature on a continuous motion shaker was allowed (about 220/972 transfusions). The criteria for a platelet transfusion on the same day were based on the morning platelet count plus findings on careful clinical examination of the patient. 1 h posttransfusion platelet counts of less than 10 x 109/1 (35/972 transfusions 3-7%) were not routinely taken as an indication for further platelet transfusion on the same day (except for 13 instances in which such transfusions were given). Additional platelet counts were not done, nor were minor bleeding episodes taken into account until the next
morning. Definition of bleeding manifestations Minor
bleeding episodes (WHO grade 1) included any mucocutaneous haemorrhages or haematomas not requiring redcell transfusions, as well as retinal haemorrhages without impairment of vision. Major bleeding episodes included (1) melaena, haematemesis, haematuria, and haemoptysis whether or not blood transfusions were required (WHO grade >_ 2); (2) epistaxis and vaginal or soft tissue bleeding requiring red-cell transfusions (WHO grade 3-4); (3) fresh retinal haemorrhages with impairment of vision. TABLE II-FREQUENCY OF PLATELET TRANSFUSIONS ON SAMEE DAY AS PLATELET COUNT
Tx
=
platelet transfusion
Statistics Mean values were compared by Student’st test, and frequencies by X2 test or, for numbers below 75, by Fisher’s exact test. Linear regression was computed with a computer-based program (’Graph PAD’, San Diego, California, USA). Results are presented as mean values (standard deviation).
Results The study included 102 patients (mean age 41-8, range 15-71 years) with acute non-lymphocytic (n = 87) or acute lymphocytic (n = 15) leukaemia. All 7 patients with acute promyelocytic leukaemia (FAB M3) and 4 of the 80 with
other AML FAB subtypes presented with disseminated intravascular coagulation (DIC). 11 patients had haemoglobin concentrations of less than 6 g/dl and 13 had leucocyte counts greater than 100 x 109/1. 28 were febrile (> 38’0°C), 4 of them with overt septicaemia. Altogether, the patients were treated with 254 cycles of marrow ablative chemotherapy, during which they received a total of 972 platelet transfusions. The entire study time from hospital entry to the recovery of the platelet count to more than 100 x 109/1 or to death or hospital discharge, whichever occurred first, comprised 6002 patient-days.
Adherence to the platelet transfusion protocol The total number of days at different platelet counts, and the number of days with and those without a platelet transfusion on the same day as the low count, are shown in table 11. In accordance with the protocol, a transfusion was withheld on 1471 /2134 (69 %) of days when a morning count was 6-20 x 109/1. Adherence to the protocol was usually strict, with the following exceptions. Although all two hundred and eighty occasions on which morning platelet counts were 5 x 109/1 or less should have been followed by a transfusion on the same day, this procedure was not possible in fourteen instances (5%) because of alloimmunisation and unavailability of an HLA-identical platelet donor. At morning platelet counts of more than 5 x 109/1 platelet transfusions were given on the same day for reasons that were not as specified in the protocol on 191/5722 study days (3’ 3 %). Nearly half these premature transfusions were given to patients with morning platelet counts of 6-10 x 109/1 (89 transfusions, corresponding to 13% of days with a platelet count of 6-10 x 109/1). The reasons for these transfusions were: accelerated platelet consumption after the preceding transfusion (twenty-one instances), extensive minor haemorrhagic manifestations (20 instances), or a rising body temperature (forty-eight instances). At platelet counts of
1225
death is discussed below (patient 1). Epistaxis requiring tamponade and red-cell support occurred 4 times. 3 episodes of haemoptysis, 7 of gastrointestinal haemorrhage, and 3 of macrohaematuria were observed. 6 of these 13 patients needed red-cell transfusions (2-5 units, mean 3-2 units). The remaining 11 major bleeding episodes consisted of retinal haemorrhage with blurred vision that eventually resolved completely in 10. In the 11 th patient it consisted of vitreous body formation and retinal haemorrhage (platelet count 8 x 109/1, DIC, heparin therapy), with permanent impairment of vision.
complications during induction therapy 80 patients (78%) were discharged from the hospital with normal or self-sustained platelet counts of more than 50 x 109/1 (67 in complete and 13 in partial remission). In 7 patients with refractory leukaemia, severe thrombocytopenia persisted but further active therapy was no longer practical; 6 died at home from bleeding and/or infectious complications. The remaining 15 patients died during induction therapy. 12/15 non-bleeding complications resulted in death (septicaemia 2, systemic fungal infection 2, miliary tuberculosis 2, pneumonia 1, pulmonary embolism 2, and cardiovascular failure 3). Although necropsy revealed various bleeding manifestations in all of them, the bleeding Lethal bleeding
Relation of bleeding risk to blood platelet counts: computed per 1000 days at risk.
Open bars=minor bleeding complications Solid bars=major bleeding complications. Numbers in bars are numbers of observed days at risk in a given class The correlation coefficient between class 1-4 and the incidence of bleeding is r= 963, p < 01. Risk category platelets> 20 x 109/1 not included in correlation coefficient because the few bleeding complications in this group were related to factors other than solely the platelet count, and because the broad range of counts in this
11-15 times
x
category.
109/1 a premature platelet transfusion was given 83
(10% of days) because of accelerated platelet consumption after the preceding transfusion (25 instances), extensive minor haemorrhages (24 instances), high fever (20 instances), or no apparent reason (14 instances). At platelet count above 15 x 109/1 the protocol was violated in nineteen instances (0-5% of days).
Haemorrhagic manifestations before study entry Minor bleeding complications at hospital entry and before the first platelet transfusion were commonest at platelet counts of less than 20 x 109/1 (15/19) but occurred also at higher counts (at 21-50 x 109/1,14/28; at > 50 x 109/1, 20/55). Major bleeding complications were present before entry in 6 patients. 3 had retinal haemorrhages causing impaired vision (platelet counts 8, 10, and 15 x 109/1). 2 patients had WHO grade 3-4 vaginal bleeding at platelet counts of 18 and 48 x 109/1, respectively. A bleeding complication, which was eventually lethal, was misinterpreted as torticollis in 1 patient (case 3, see below; platelet count 65 x 109/1). Non-lethalhaemorrhagic episodes during induction chemotherapy 6 major bleeding complications following minor surgical interventions were not judged to be spontaneous haemorrhagic episodes. In 5 patients (platelet counts 36-87 x 109/1) a haemopneumothorax complicated the insertion of a subclavian catheter. Another patient had a haematoma of the scrotum and penis after inguinal lymph node biopsy (platelet count 14 x 109/1). After exclusion of these procedure-related bleeding complications we noted a progressive increase in the frequency of haemorrhage with declining platelet counts (figure). Minor forms of haemorrhage occurred much more frequently than did the more serious forms at all platelet count bands. Overall, 28 major non-lethal bleeding episodes were observed. A case of subdural haematoma complicated by cerebral oedema and jacksonian epilepsy contributing to
did not contribute to death. Death during induction therapy was related to or directly caused by bleeding complications in 3 patients. Patient 1 (51-year-old female, AML FAB M2) had a temporoparietal subdural haematoma after 2 days with a platelet of less than 1 x 109/1 because of refractoriness and lack of an HLA-identical donor. The patient died 7 days later from cardiac arrest due to electrolyte imbalance. Necropsy disclosed a flat (1 mm) residual subdural haematoma. Patient 2 (38-year-old man, AML FAB M5) was admitted with a white-cell count of 430 x 109/1 and platelet count of 55 x 109/1. Because of DIC (fibrinogen 0-3 g/1) he was treated with low doses of heparin and fibrinogen. 9 h later (platelet count 53 x 109/1, fibrinogen 0-4 g/1) he rapidly became comatose, and he died 15 h after admission from a cerebral haematoma with rupture into the ventricular system. Patient 3 (53-year-old female, AML FAB M3) complained of acute torticollis on admission (platelet count of 65 x 109/1, slightly raised concentration of fibrin degradation products). Low-dose heparin therapy was given for 7 days, during which the lowest platelet count was 35 x 109/1. Her neck pain worsened. 11 days after admission and 3 days after the third platelet transfusion (lowest pre-transfusion count 8x 109/1, 20 x 109/1) she suddenly became tetraplegic and died of respiratory arrest. Necropsy disclosed a fresh epidural and subdural haematoma of the spinal cord extending from C2 to C5. count
Clinical characteristics of patients with major haemorrhages The 28 major non-lethal and the 3 lethal bleeding episodes occurred in 23 patients. Their mean age was 42-2 (16 years), compared with 41 ’5(15) for the remainder of the population (p 085, by t-test). 10 were febrile (> 38-0°C) and 20 were on antibiotics; of these 7 also received amphotericin B. The 31 haemorrhagic episodes occurred at platelet counts of 1 x 109/1 to 65 x 109/1, with most being 10 x 109/1 or more (18/31). 4 episodes (1 lethal) occurred in alloimmunised patients left at platelet counts of less than 5 x 109/1 for 2-4 days because of unavailability of HLAcompatible platelet donors. In 4 other episodes endoscopically verified extensive mucous membrane lesions were the source of bleeding (ulcerative colitis and melaena in 2, platelet counts of < 5 x 109/1 in both cases; candida =
1226
oesophagitis and haematemesis in 2, platelet counts of 19 and 38 x 109/1). In patients with DIC and on low-dose heparin at the time of bleeding major bleeding episodes were observed in 3/18 cases with platelet counts of 10 x 109/1 and in 8/13 with higher platelet counts (p == 0-014, Fisher’s exact test). Several patients had more than one risk factor. Thus, additional risk factors for haemorrhage were obviously present in 16/31 of the major bleeding episodes and in 8/13 of serious bleeding complications observed at platelet counts above 10
x
109/1. Discussion
This study on haemorrhagic complications of thrombocytopenia during induction therapy for acute leukaemia shows that a significantly lower threshold than generally used for prophylactic platelet transfusions is safe and not associated with a higher bleeding risk if based on reliable platelet counts and appropriate clinical evaluation of the patient. This more restrained use of platelets results in considerable cost savings (about C350 000 in this study), reduces the rising demand for platelets imposed on blood banks, and prevents excessive exposure of cancer patients to platelet donors. Early retrospective studies in leukaemia patients established a strong relation between manual platelet counts and the frequency of haemorrhagic episodes, with bleeding occurring on 38% of days at counts of 10-20 x z/1,50% at counts of 5-10 x 109/1, and 65-92% at lower counts.1 Our results accord with this observation. However, the incidence of minor and major bleeding episodes was lower at all levels of platelet counts in our study than in earlier ones. The reasons for this difference are difficult to identify. A possible explanation might be that at the time of the older studies the effect of aspirin or other non-steroidal anti-inflammatory drugs on platelet function was not recognised, and so might have been liberally used, whereas they were strictly avoided in our study. The outstanding progress in antimicrobial therapy over the past three decades may also have contributed to the low incidence of bleeding in our patients, since infectious complications and/or high fever are wellaccepted causes of accelerated platelet consumption’,10 and increase the risk of major bleeding complications.3 Our results clearly show that a significantly lower threshold than that generally used is practicable for prophylactic platelet transfusions during induction chemotherapy of acute leukaemias. Minor bleeding episodes and almost all the major episodes did not seem to jeopardise outcome of therapy for our patients since a high response rate (78%) was obtained and lethal haemorrhages (3%) or severe sequelae of bleeding (1%) were infrequent. Many of the major bleeding episodes occurred at platelet above 10 x 109/1 and were related to additional risk factors such as anatomical lesions or DIC and/or heparin therapy. Thus, a liberal transfusion policy would hardly have prevented many of the cases of severe bleeding. With our restrictive transfusion strategy the mean number of platelet transfusions given per chemotherapy course was 3-8 (range 1-22 data not shown), which is appreciably lower than that reported elsewhere.2,11 It is conceivable that the lower number of transfusions (or donors), the lower is the risk of transfusion-transmitted diseases such as cytomegalovirus, hepatitis C, and possibly HIV infection. Whether reducing the number of platelet transfusions has a favourable effect on the incidence of alloimmunisation remains controversial.l1,12 We believe that the low rate of refractoriness observed in our study (10%) is counts
due to the use of single-donor platelets rather than to the lower number of transfusions.7,8 Although there is a tight correlation between platelet counts and the incidence of haemorrhage, other clinical factors that contribute to the risk of bleeding3,10,11 should be taken into account--eg, drugs that may interfere with platelet function. For this reason, clinical findings were always considered when deciding on time of platelet support. Our results show that the low threshold for prophylactic platelet support 5 x 109/1-in the absence of fresh haemorrhagic manifestations or fever (38-0°C), and 10 x 109/1 in the presence of such signs-is safe. To reduce bleeding complications at higher counts, the presence of coagulation disorders, heparin therapy, or major anatomical lesions should also be considered. Finally, although our results have been obtained in a uniform population of leukaemia patients, they might also be relevant for patients with hypoplastic thrombocytopenia due to other causes. .
REFERENCES Gaydos LA, Freireich EJ, Mantel N. The quantitative relation between platelet count and hemorrhage in patients with acute leukaemia. N Engl J Med 1962; 266: 905-09. 2. Silver SS, Rock G, Décary F. Use of platelet concentrates in eastern Ontario. Can Med Assoc J 1987; 137: 128-32. 3. Aderka D, Praff G, Santo M, Weinberger A, Pinkhas J. Bleeding due to thrombocytopenia in acute leukaemias and reevaluation of the prophylactic transfusion policy. Am J Med Sci 1986; 291: 147-51. 4. Slichter SJ, Harker LA. Thrombocytopenia: mechanisms and management of defects in platelet production. Clin Haematol 1978; 7: 523-27. 5. Soloman J, Bofenkamp T, Fahey JL, Chillar RK, Beutler E. Platelet prophylaxis in acute non-lymphoblastic leukaemia. Lancet 1978; i: 267. 6. National Institutes of Health. Consensus development conference on platelet transfusion therapy. JAMA 1987; 257: 1777-80. 7. Gmür J, von Felten A, Osterwalder B, Scali G, Sauter Chr, Frick P. Delayed alloimmunization using random single donor platelet transfusions: a prospective study in thrombocytopenic patients with 1.
acute leukemia. Blood 1983; 61: 473-79. 8. Gmür J, Burger J, Sauter Chr, Oelz O, Frick PG. Alloimmunization by leukocyte-rich or leucocyte-poor random single donor platelet transfusions. Prog Clin Biol Res 1990; 337: 45-48. 9. Murphy S, Litwin S, Herring LM, et al. Indications for platelet transfusion in children with acute leukemia. Am J Hematol 1982; 12: 347-56. 10. Bishop JH, McGrath K, Wolff MM, et al. Clinical factors influencing the efficacy of pooled platelet transfusions. Blood 1987; 71: 383-87. 11. Scott E, Steeger T, Huntington S, et al. Platelet utilization patterns at a large referral hospital. Transfusion 1985; 25: 469. 12. Dutcher JP, Schiffer CA, Aisner J, Wiernik PH. Alloimmunization following platelet transfusions: the absence of dose-response relationship. Blood 1981; 57: 395-99.
From The Lancet Treatment of mad wolves’ bites
Owing to the number of rabid wolves in Romania and the neighbouring countries Professor Babesch, Director of the Bacteriological Institute in Bucharest, and one of the ablest pupils of Pasteur and Koch, has had many opportunities of treating persons bitten by mad wolves by Pasteur’s method, unfortunately, however, without success, for the symptoms of the disease broke out much too quickly to admit of an effective cure by the application of lymph taken from the spinal cord of rabbits that had died of hydrophobia. In view of this sad fact, Professor Babesch has for some time past been making experiments with the purpose of ascertaining whether the immunity from the virus of hydrophobia conferred by Pasteur’s method cannot be transferred by inoculating the blood of animals thus rendered immune to other warm-blooded animals. These experiments showed that such inoculation had a prophylactic effect even in cases in which Pasteur’s spinal cord lymph, under otherwise quite identical conditions, produced no such results whatever.
(July 4,1891)