Correspondence
Prophylactic Platelet Transfusions in Acute Myeloid Leukemia
N Engl J Med 1998; 338:1468-1470May 14, 1998
- Article
To the Editor:
The December 25 issue of the Journal presents two important prospective, controlled studies on the use of platelet transfusions.1,2 Both studies, as commented on in an editorial in the same issue,3 could change transfusion policies. Leukocyte-depleted or ultraviolet B–irradiated pooled platelet products could replace single-donor–apheresis products because of their lower production costs, and the standard transfusion trigger could be lowered to 10×109 platelets per liter. Though carefully designed, both studies in our view missed some important points.
Alloimmunization, the end point in the Trial to Reduce Alloimmunization to Platelets,1 is only one aspect of a comparison of single-donor and pooled platelet products. Each transfusion carries a risk of transfusion-transmitted diseases. Any six-unit platelet product increases sixfold the risk of transmitting hepatitis C or any other disease as compared with a single-donor product. The trial was not designed to address this issue, but the problems involved should at least be discussed.
Similarly, in the study by Rebulla et al.2 on the risks of a lowered platelet-transfusion trigger, the only end points studied were bleeding and the need for red-cell transfusion. The results were not significantly different in the two treatment groups. Any bleeding into the mucosa or the skin alters the protective barrier and potentially increases the risk of infection. Any bleeding into an organ can lead to organ damage. The number of episodes of infection was increased, and twice as many patients (18 vs. 9) died in the group with the lower platelet threshold. Although the causes of death were not directly related to bleeding, these differences should at least be commented on. On the basis of these thoughts, we consider it premature to postulate the general safety of a lowered platelet trigger for transfusion.
3 ReferencesAlois Gratwohl, M.D.
André Tichelli, M.D.
Kantonsspital, CH-4031 Basel, Switzerland1
The Trial to Reduce Alloimmunization to Platelets Study Group
Full Text | Web of Science | Medline2
Rebulla P ,Finazzi G ,Marangoni F , et al. The threshold for prophylactic platelet transfusions in adults with acute myeloid leukemia. N Engl J Med 1997;337:1870-1875
Full Text | Web of Science | Medline3
Kruskall MS . The perils of platelet transfusions. N Engl J Med 1997;337:1914-1915
Full Text | Web of Science | Medline
To the Editor:
The Trial to Reduce Alloimmunization to Platelets Study Group used the Kaplan–Meier method to estimate event rates. This method requires that censoring be noninformative,1 which means that censoring must be unrelated to the likelihood of an event's developing. This may not hold true in this study, since death was a cause of censoring. If the patients who died were those in whom platelet refractoriness was more likely to have developed had they survived, censoring would have been informative, and the Kaplan–Meier estimates could have been biased. This concern is heightened by the fact that the patients receiving treated platelets had higher rates of death from refractory leukemia and infection than control patients. Thus, they were more likely to die than control patients and therefore more likely to be censored before refractoriness had an opportunity to develop.
The investigators chose an arbitrary cutoff level of 5000 platelets per cubic millimeter for the corrected-count increase to define platelet refractoriness. This unnecessarily reduces the amount of information provided by the study. Instead, the mean corrected-count increase could have been plotted against time or transfusion number for each group. This would have avoided the problem of Kaplan–Meier estimates altogether, and would have provided a fuller picture of the effectiveness of platelet manipulation before transfusion.
This trial measured platelet refractoriness primarily among women with a history of pregnancy. The author notes that platelet refractoriness among men and among women who had never been pregnant was uncommon and did not vary among the four study groups. Thus, platelet manipulation would seem to be unnecessary for men or for women without a history of pregnancy.
1 ReferencesCarl D. Atkins, M.D.
242 Merrick Rd., Rockville Centre, NY 11570Author/Editor Response
The authors reply:
To the Editor: The point raised by Gratwohl and Tichelli, that any bleeding into the mucosa or the skin could alter the protective barrier and increase the risk of infection, was not considered in our original analysis. We investigated the relation between hemorrhage and infection in the 12 of 135 patients (8.8 percent) and the 7 of 120 patients (5.8 percent) in our study who died of infection and were randomly assigned to thresholds of 10,000 and 20,000 platelets per cubic millimeter, respectively. The 3 percent difference in the frequency of deaths caused by infection was not statistically significant (95 percent confidence interval, -3.3 to +9.4 percent). We also examined the frequency of hemorrhage and the type of bleeding in the patients, and found that during the 377 days of hospitalization for the 12 patients with the lower threshold and the 153 days for the 7 patients with the higher threshold, hemorrhage was reported as code 1 (petechiae or mucosal or retinal bleeding that did not require red-cell transfusion) or code 2 (melena, hematemesis, hematuria, or hemoptysis) on 60 days (15.9 percent) for the 12 patients and 8 days (5.2 percent) for the 7 patients, respectively. On the remaining days no hemorrhage was reported. Although the 10.7 percent difference between the subgroups was statistically significant (95 percent confidence interval, 5.6 to 15.8 percent), 55 of the 60 days on which hemorrhage was reported for the group of 12 patients with the lower threshold applied to only 3 patients — a fact that did not allow us to draw any conclusions. Moreover, there was no clear time relation between the onset of hemorrhage and the onset of infection in these three patients.
Because the evidence is inconclusive, we believe that the hypothesis of Gratwohl and Tichelli is worth exploring in a randomized clinical trial, even though this may take some time. Until we have conclusive evidence, we believe that our data support the safety of reducing the threshold to 10,000 platelets per cubic millimeter, because the percentage of patients with major hemorrhage, the actuarial survivals, and the numbers of days of hospitalization per patient did not differ significantly in the two groups. In addition to the safety of this threshold, it was associated in our study with a 21.5 percent reduction in platelet-donor exposure of the patients. This in turn increases the availability of a limited and expensive resource and reduces the small but still measurable risk of acquiring transfusion-transmissible infections.
Paolo Rebulla, M.D.
Francesca Marangoni, M.D.
Girolamo Sirchia, M.D.
Ospedale Maggiore, 20122 Milan, ItalyAuthor/Editor Response
We appreciate Dr. Atkins's comments about the bias that can arise if censoring is informative; in this case, however, we believe that censoring and outcome are independent. There is no reason to believe that patients who are likely to die have an increased or decreased risk of alloimmune platelet refractoriness. Often, patients who die have clinical conditions that may cause platelet refractoriness, but these are usually not linked to alloantibodies.1,2 Furthermore, the perceived association between treatment and mortality, which Dr. Atkins commented on, is not significant (P = 0.42). In addition, only 14 percent of the patients died during the eight-week study period, with no differences among the four groups.
In previously pregnant women, rates of platelet refractoriness were significantly reduced only in the patients who received leukoreduced, pooled, random-donor platelets as compared with the patients in the control group (P = 0.009). There were no differences in rates of platelet refractoriness between any other groups of patients in the study. The overall rate of refractoriness in this trial was only 10 percent, which is extremely low when compared with other studies3-5 and may reflect the relatively restrictive criteria we used to determine platelet refractoriness — i.e., a 1-hour post-transfusion corrected-count increment of <5000 on two sequential occasions that required at least one of these transfusions to be ABO-compatible and <48 hours old. Previous studies have usually required only a single 1-hour post-transfusion corrected-count increment of <7500. Many of the patients in whom antibodies developed were no longer receiving transfusions when their antibodies were detected. Thus, if these antibodies are durable, the benefit of reducing the incidence of antibodies seen in all the patients (both those previously pregnant and those who had never been pregnant) receiving either filtered or ultraviolet B–irradiated platelet transfusions may be apparent during subsequent courses of chemotherapy. The durability of these antibodies is currently being evaluated.
Concerning the comments of Drs. Gratwohl and Tichelli, we agree that the risk of transfusion-transmitted diseases may be increased by greater numbers of donor exposures. However, the Trial to Reduce Alloimmunization to Platelets was not designed to evaluate disease transmission by transfusion. It should be noted, however, that even in the patients receiving single-donor platelets, there was an average of 28 donor exposures during induction. The study results do indicate that there is no advantage to using single-donor platelets over pooled random-donor platelets to prevent platelet alloimmunization.
5 ReferencesSherrill J. Slichter, M.D.
Puget Sound Blood Center and Program, Seattle, WA 98104-1256Charles A. Schiffer, M.D.
Wayne State University School of Medicine, Detroit, MI 48201Kathryn A.B. Davis, Ph.D.
University of Washington, Seattle, WA 98105for the Trial to Reduce Alloimmunization to Platelets Study Group
1
Friedberg RC ,Donnelly SF ,Boyd JC ,Gray LS ,Mintz PD . Clinical and blood bank factors in the management of platelet refractoriness and alloimmunization. Blood 1993;81:3428-3434
Web of Science | Medline2
Bishop JF ,McGrath K ,Wolf MM , et al. Clinical factors influencing the efficacy of pooled platelet transfusions. Blood 1988;71:383-387
Web of Science | Medline3
Sniecinski I ,O'Donnell MR ,Nowicki B ,Hill LR . Prevention of refractoriness and HLA-alloimmunization using filtered blood products. Blood 1988;71:1402-1407
Web of Science | Medline4
van Marwijk Kooy M ,van Prooijen HC ,Moes M ,Bosma-Stants I ,Akkerman JW . Use of leukocyte-depleted platelet concentrates for the prevention of refractoriness and primary HLA alloimmunization: a prospective, randomized trial. Blood 1991;77:201-205
Web of Science | Medline5
Williamson LM ,Wimperis JZ ,Williamson P , et al. Bedside filtration of blood products in the prevention of HLA alloimmunization -- a prospective randomized study. Blood 1994;83:3028-3035
Web of Science | Medline
- Citing Articles (2)
Citing Articles
1
Shoshan Nevo, Alice K. Fuller, Marianna L. Zahurak, Eric Hartley, Mark E. Borinsky, Georgia B. Vogelsang. (2007) Profound thrombocytopenia and survival of hematopoietic stem cell transplant patients without clinically significant bleeding, using prophylactic platelet transfusion triggers of 10
נ10 9 or 20
CrossRef2
Nancy M. Heddle, Richard J. Cook, Kathryn E. Webert, Christopher Sigouin, Paolo Rebulla, . (2003) Methodologic issues in the use of bleeding as an outcome in transfusion medicine studies. Transfusion 43:6, 742-752
CrossRef
