The Role of the Plasma from Platelet Concentrates in Transfusion Reactions

Clinical Protocol

Patients were eligible for the study if they were over 17 years of age, had thrombocytopenia, and were expected to require a minimum of six platelet transfusions over a one-to-two-month period. Since our hypothesis stated that the plasma supernatant, rather than a factor in the recipient (i.e., antileukocyte alloantibodies) was responsible for these reactions, a previous platelet-associated reaction was not a criterion for entry. Each platelet transfusion consisted of a pool of concentrates prepared from whole blood from five random donors that had been collected in bags containing CP2D anticoagulant (Miles Laboratories, Berkeley, Calif.). Whenever possible, four- or five-day-old platelets were transfused to increase the likelihood of a reaction². The pooled platelet concentrate was centrifuged at 2000 × g for 10 minutes, and 150 to 180 ml of the plasma supernatant was removed to a sterile transfer bag. The plasma supernatant contained low numbers of platelets and leukocytes (less than 10⁶ per product)¹⁰. Approximately 180 ml of previously prepared fresh-frozen plasma from one of the platelet donors was added to the pellet of platelets and leukocytes remaining in the bag, and the product was allowed to rest at room temperature for 40 to 60 minutes before resuspension.

Plasma samples taken from the supernatant and from the product after resuspension of the cellular component were assayed to determine the leukocyte count, platelet count, interleukin-1β concentration, and interleukin-6 concentration. The age of the platelet product was also recorded. The components of the platelet product (plasma supernatant and cells) were transfused in random order, with a two-hour interval between infusions. The two-hour interval was selected as a washout period because our previous results indicated that most reactions occur during or just after transfusion and that symptoms subside within two hours². Neither the patients nor the nursing staff were told which component was being transfused.

Before each transfusion, a questionnaire was administered to the patient to document base-line evidence of chills, cold, or discomfort on a seven-point Likert scale on which 1 indicated “no symptoms” and 7 “severe symptoms”². Scores of 2 and 3 were considered to indicate mild symptoms, 4 and 5 moderate symptoms, and 6 and 7 severe symptoms. The questionnaire was readministered immediately after the transfusion and again one hour later. A reaction was defined as an increase in the severity of symptoms over the base-line level. The patient's temperature was also recorded each time. Standardized pretransfusion medication consisted of 25 mg of intravenous diphenhydramine hydrochloride (Benadryl, Parke-Davis, Scarborough, Ont.) and two 325-mg acetaminophen tablets (Tylenol, McNeil, Guelph, Ont.). Some patients also received 50 mg of meperidine hydrochloride (Demerol, Sanofi Winthrop, Markham, Ont.). Serum from the patients was tested for lymphocytotoxic antibodies (GenTrak, Plymouth Meeting, Pa.). The study was approved by the institutional ethics committee, and informed consent was obtained from each eligible patient before enrollment.

Laboratory Studies

Ten platelet concentrates were prepared from blood from random donors that had been collected in bags containing CP2D anticoagulant. Each concentrate was separated into two equal portions, one of which was filtered (Pall PL50, Pall Biomedical Products, Toronto) to remove contaminating leukocytes. The concentrates were then stored at 22 °C with horizontal agitation for 10 days. Samples were taken from each concentrate (filtered and unfiltered) on days 1 through 5, day 7, and day 10 for cytokine measurement and bacterial culture. In the unfiltered portion of the concentrates, leukocyte and platelet counts were performed with a Coulter counter (Coulter Electronics, Hialeah, Fla.). In the filtered platelets, leukocyte counts were performed manually with a Hausser counting chamber¹⁰. The threshold of detection with this method is 10⁶ leukocytes per product. Interleukin-1β was measured with an enzyme-linked immunosorbent assay (R and D Systems, Minneapolis). Interleukin-6 was measured with a standard bioassay. Proliferation of the interleukin-6-dependent B9 hybridoma cell line was used to assay samples, with recombinant human interleukin-6 used as a standard¹¹.

Statistical Analysis

We analyzed the association between characteristics of the platelet concentrate (age, leukocyte count, and cytokine concentrations) and the likelihood of reaction using the Wilcoxon test. We compared rates of reaction and the severity of reactions associated with the plasma and cellular components of the platelet products using McNemar's chi-square test. The correlation of interleukin-1β and interleukin-6 concentrations with the leukocyte count during storage was determined with the Spearman rank-correlation coefficient.

Clinical Study

Twelve patients received a total of 64 pairs of transfusion components (range, 3 to 10) according to the protocol described. Seven of the patients had newly diagnosed acute myelogenous leukemia (AML) and were receiving induction chemotherapy. The remaining five patients had AML (relapsed or in remission), myelodysplastic syndrome, chronic myelogenous leukemia, or chronic idiopathic thrombocytopenic purpura with gastrointestinal hemorrhage. For 32 transfusion pairs, the plasma was administered first, and the cellular component two hours later. For the remaining 32 pairs, the cellular component was administered first.

Table 1. Table 1. Number of Adverse Reactions, According to the Type of Product.

The frequency of adverse reactions to the plasma supernatant, the cellular component, and both products is shown in Table 1. There were 28 reactions to the plasma supernatant and 14 reactions to the cellular component. In 20 transfusion pairs there were reactions to the plasma supernatant but not to the cellular component, and in 6 transfusion pairs there were reactions to the cellular component but not to the supernatant (chi-square = 6.50, P = 0.009).

Nine patients had more than one adverse reaction: seven had more reactions to the plasma supernatant than to the cellular component, one had more reactions to the cellular component, and one had an equal number of reactions to both components. With eight transfusion pairs, there were reactions to both the plasma and the cellular component. Two patients, who received three and four pairs of transfusion components, had no reactions. Six of the seven platelet products these two patients received were stored for four days or less, had leukocyte counts of less than 600 per cubic millimeter, and had low concentrations of interleukin-1β (≤ 12 pg per milliliter) and interleukin-6 (≤ 373 pg per milliliter). One five-day-old platelet concentrate had an absolute leukocyte count of 1800 per cubic millimeter and interleukin concentrations of 36 and 475 pg per milliliter, respectively.

The severity of reactions to the plasma supernatant was significantly greater than the severity of reactions to the cellular component. Forty-three percent of the reactions to the plasma (12 of 28) were graded severe. Thirty-two percent (9 of 28) were mild, and 25 percent (7 of 28) were moderate. In contrast, only 21 percent of the reactions to the cellular component (3 of 14) were severe, 21 percent (3 of 14) were moderate, and 57 percent (8 of 14) were mild. Thus, for the 34 pairs of transfusion components that were associated with reactions, severity was greater with the plasma supernatant in 24 cases, was greater with the cellular component in 6, and did not differ between components in 4 (chi-square = 9.6, P<0.01). An increase in temperature of more than 1 degree C was observed with only 7 of the 34 transfusion pairs associated with reactions. For all seven transfusions that produced a febrile reaction, the reaction was caused by the plasma alone.

The order of infusion did not have any effect on the likelihood or severity of reaction. The plasma supernatant was transfused first in 9 of 20 cases of reactions associated with the plasma alone, 4 of 6 associated with the cellular component alone, and 4 of 8 associated with both products.

Table 2. Table 2. Correlation of the Likelihood of a Reaction with the Age, Leukocyte Count, and Cytokine Concentrations of the Platelet Product.

Reactions to the plasma supernatant were associated with higher concentrations of interleukin-1β (P<0.001 by the Wilcoxon test) and interleukin-6 (P = 0.034 by the Wilcoxon test) (Table 2). Reactions to the plasma supernatant were also associated with older platelet concentrates (P = 0.007) and higher leukocyte counts in the original platelet pool (P = 0.03). The likelihood of a reaction to the cellular component of the platelet concentrate was not associated with the age of the product (P = 0.87), the leukocyte count (P = 0.32), the interleukin-1β concentration (P = 0.18), or the interleukin-6 concentration (P = 0.16) (Table 2). The mean (±SD) leukocyte count in the cellular component was 1760 ±1710 per cubic millimeter. The mean leukocyte count in the plasma supernatant was 30 ±20 per cubic millimeter.

Serum from nine of the patients was available for testing for lymphocytotoxic anti-HLA antibodies. Five of these patients had reactions to the cellular component of the product, but only one of the five had detectable lymphocytotoxic anti-HLA antibodies. This patient, who had idiopathic thrombocytopenic purpura, had two reactions to the plasma supernatant, one reaction to the cellular component, and two reactions to both components. Lymphocytotoxic antibodies were not detected in the serum of four patients who did not have reactions to the cellular component of the platelet product.

In Vitro Studies

Table 3. Table 3. Concentration of Cytokines in Platelet Concentrates during Storage, According to Whether They Had Been Depleted of Leukocytes.

The concentrations of both interleukin-1β and interleukin-6 rose progressively during storage of the platelet concentrates that had not been depleted of leukocytes (Table 3). Interleukin-1β, which was undetectable (<0.3 pg per milliliter) in all platelet concentrates on day 1, was detected in 9 of 10 concentrates by day 5 and in all concentrates by day 10. Interleukin-6 was detectable in 9 of 10 platelet concentrates by day 5 and in all concentrates by day 10. The correlation of cytokine concentrations with the duration of storage was 0.94 for interleukin-1β and 0.95 for interleukin-6 (P<0.01). When the platelet concentrates were depleted of white cells before storage, the concentrations of interleukin-1β and interleukin-6 remained undetectable or very low throughout the period of storage. The concentrations of interleukin-1β and interleukin-6 in the plasma correlated positively with the leukocyte count in the platelet product (r = 0.73 and r = 0.71, respectively). Bacterial cultures of the platelet concentrates showed no growth on day 10 of sampling.

We found that patients reacted significantly more often and more severely to the plasma from the platelet concentrate than to the cells (platelets and leukocytes). As in our previous study,² reactions to the plasma supernatant were most frequent with supernatants from older platelet products and from those with higher leukocyte counts. In contrast, adverse reactions to the cells, which were resuspended in fresh-frozen plasma, did not correlate with the age of the product or the leukocyte count. Our study supports the hypothesis that bioreactive substances released into the supernatant plasma during storage cause platelet-associated transfusion reactions^2,12.

Since interleukins (particularly interleukin-1β and interleukin-6) are potent pyrogens, we measured the concentrations of interleukin-1β and interleukin-6 in samples taken from the platelet products just before infusion. Only with plasma supernatant was a strong correlation observed between the cytokine concentrations and transfusion reactions.

As in previous studies,^12-14 interleukin-1β and interleukin-6 were undetectable in the platelet product at the time of its collection but increased progressively during storage (Table 3). An important determinant of the concentration of these cytokines was the leukocyte count in the platelet product during storage. However, cytokines may not be the only mediators of platelet-associated reactions. It has recently been reported that lipid compounds that increase the activity of neutrophil oxidase also accumulate in platelet concentrates during storage, and these agents may have a role in acute transfusion-associated lung injury¹⁵.

The outcome measured in this study was subjective, and the clinical relevance of mild reactions, which were usually characterized by a cold feeling or discomfort, can be questioned. However, the severe reactions, which included rigors, chills, and nausea, could be documented objectively by the nursing staff. They were also significantly more likely to be caused by plasma than by cells.

Several of our patients reacted only to the cellular component of the platelet product, suggesting that in some patients mechanisms other than cytokines may contribute to or be responsible for these reactions. It is possible that these reactions involve an immune-mediated event^4-6; however, only one of these patients had detectable lymphocytotoxic anti-HLA antibodies (Table 1).

The generalizability of these results is limited, since we studied only 12 patients; however, a significant statistical and clinical difference was observed in spite of the small number of patients. The overall frequency of reactions in this study was higher than previously reported^1,2. This was not unexpected, because four- and five-day-old platelets were selected for transfusion whenever possible, since we had previously identified the age of the platelet concentrate as the most significant predictor of a reaction².

Our observations could explain why removing leukocytes from the platelet product just before transfusion may not prevent reactions in some patients. However, our study does suggest that depleting platelet concentrates of leukocytes before storage may be an effective way of preventing febrile, nonhemolytic transfusion reactions. Although removing the plasma supernatant just before infusion was efficacious in preventing reactions to the platelet product in this small group of patients, investigations in a more diverse population are needed to determine the overall effectiveness of this approach.