Evaluation of platelet cross-matching in the management of patients refractory to platelet transfusions

Abstract

Background

Cross-match-compatible platelets are used to support thrombocytopenic patients who are refractory to randomly selected platelets. However, few studies have addressed the efficacy of using this strategy for patients requiring intensive platelet transfusion therapy. The aim of this study was to determine the effectiveness of cross-match-compatible platelets in an unselected group of patients refractory to platelets from random donors.

Materials and methods

A total of 406 cross-match-compatible platelet components were administered to 40 evaluable patients who were refractory to random-donor platelets. A solid-phase red cell adherence method was used for platelet cross-matching. The corrected count increment was used to monitor the effectiveness of each platelet transfusion. Multivariate analysis was performed to detect whether any variables could predict the response to transfusion.

Results

Statistically significant improvements were found in the mean corrected count increment when comparing cross-match-compatible platelets with randomly selected and incompatible platelets (p<0.001 for each). Compatible platelet transfusions were associated with a good response in 72.9% of cases while incompatible platelets were associated with a poor response in 66.7% of transfusion events (p<0.001). In the presence of clinical factors or alloimmunisation, compatible platelets were associated with good responses in 67.9% and 28.0% respectively vs 100% and 93.3% in their absence (p=0.009, p<0.001). Multivariate analysis revealed that cross-matching and alloimmunisation were the strongest predictors of transfusion response at 1 hour, while ABO compatibility, type of units received, followed by alloimmunisation then clinical factors were predictors at 24 hours.

Discussion

Platelet cross-matching using the solid-phase red cell adherence technique is an effective and rapid first-line approach for the management of patients refractory to platelet transfusions.

Introduction

Platelet transfusion is an essential part of the treatment of cancer, haematological malignancies, bone marrow failure, and hematopoietic stem cell transplantation¹. Between 30% and 70% of multiply transfused thrombocytopenic patients become refractory to random-donor platelets². The term “refractoriness”, defined as two consecutive failures to respond to platelet transfusions, is used to indicate that platelets from random donors produce post-transfusion platelet count increments in the recipient which are significantly lower than expected, based on the number of platelets administered and an estimate of the patient’s blood volume³. Platelet transfusion refractoriness may have an alloimmune or non-alloimmune mechanism, with the latter being responsible for 80% of cases⁴. Non-immune mechanisms include sequestration because of splenomegaly, and accelerated consumption or decreased production in states such as fever/infection, disseminated intravascular coagulation, circulating immune complexes, drug-related antibodies or toxicity, thrombotic microangiopathies and, finally, the properties of the platelet unit itself (quantity transfused and duration of storage)^1,5,6. Any of these non-immune insults may manifest with platelet transfusion refractoriness in 50% of patients^4,7. In contrast, alloimmune mechanisms consist chiefly of sensitisation to foreign class I A or B human leucocyte antigens (HLA) through transfusion, pregnancy or transplantation. The attack of variably expressed major-incompatible A, B antigens on donor platelets by pre-formed recipient ABO-isogglutinins also plays a role in this category, and least often, sensitisation to polymorphic antigens in the human platelet alloantigen (HPA) system may occur^1,4,5,7–9. Two main strategies have been used to transfuse alloimmunised patients: matching donor-recipient HLA antigens and cross-matching platelets¹. HLA-matching is one of the most frequently used modern methods. While this method does reliably improve platelet increments in patients with alloimmune refractoriness, some studies have found that up to 40% of HLA-matched platelet transfusions remain unsuccessful^10,11. HLA typing of patients as well as platelet donors is expensive and the long turn-around time decreases its utility in some clinical situations. In addition to these drawbacks, HLA matching requires the availability of large numbers of HLA-typed donors¹. Even large blood suppliers periodically have difficulty identifying HLA-matched donors for some patients¹². As a result, alternative strategies have been developed to obtain HLA-compatible, if not fully HLA-matched platelets. Platelet cross-matching assays are a relatively low-cost and rapid alternative to the HLA-matched approach to the management of platelet refractoriness^3,13,14. Cross-matching assays have been used for the identification of candidate platelet donors and may be beneficial for patients in whom refractoriness is due to HPA alloimmunisation, so the HLA-matched platelet transfusion has no value¹⁵. Different cross-matching methods have, therefore, been developed to identify compatible platelet donors, including radioactive techniques, flow cytometry, enzyme-linked immunosorbent assay (ELISA) and solid-phase procedures^16–23. In a common version of this assay, a solid-phase capture method is used to screen patients’ plasma for platelet antibodies directed against HLA or other antigens on platelets. Typically, a given patient’s plasma is tested against platelet samples. Donor platelets lacking reactivity in the assay are considered to be “cross-match-compatible” and are selected for transfusion support²². Selection of products based on platelet cross-matching has been shown to improve post-transfusion platelet increments in refractory patients^11,23–25. Despite the routine use of platelet cross-matching at many institutions, little has been published about the safety or effectiveness of this strategy in the mid-term (several weeks-months) management of refractory patients. Here, we present transfusion-related and clinical outcomes observed at our institution, which primarily uses cross-matched platelets for managing platelet refractoriness to determine whether platelet cross-matching can effectively identify platelet units that will improve the post-transfusion platelet counts in all patients who are refractory to randomly selected platelets.

Materials and methods

Patients

This study was performed on 40 randomly selected patients (16 males and 24 females) with a mean age of 37.6±13.35 years who were identified as refractory after receiving random-donor platelet transfusions; all presented to the Oncology Centre of Mansoura University between May 2011 and December 2102. All patients were receiving treatment for haematological disorders. They received a total of 120 platelet transfusions (range, 1 to 4 per patient). Platelets were stored at 20–24 °C with continuous agitation for a maximum of 5 days (mean 2.1±1.27 days). In all transfusions, patients were monitored for the presence of splenomegaly, active bleeding, sepsis, drugs and fever higher than 38.5 °C.

This study was approved by the Committee on Human Research at the University of Mansoura, Egypt. Our standard protocol for platelet cross-matching was performed only for patients who had demonstrated a 1-hour or 24-hour post-transfusion corrected count increment (CCI) of less than 5,000 or 2,500, respectively, after at least two consecutive transfusions. The CCI was calculated using the following formula:

$$\frac{[\text{post}-\text{transfusion\hspace{0.17em}platelet\hspace{0.17em}count\hspace{0.17em}}({10}^9/\text{L})-\text{pre}-\text{transfusion\hspace{0.17em}platelet\hspace{0.17em}count\hspace{0.17em}}({10}^9/\text{L})]\times [\text{body\hspace{0.17em}surface\hspace{0.17em}area}\hspace{0.17em}({\text{m}}^2)]}{[\text{platelet\hspace{0.17em}dose\hspace{0.17em}transfused\hspace{0.17em}}({10}^{11})]}$$

Preparation of indicator cells

Whole blood samples from healthy donors with blood group O, Rh+ (R1r) were drawn into acid-citrate-dextrose. The red cells were washed three times with normal saline solution. Anti-D IgG (Diagast, Parc Eurasanté, France) was mixed with an equal volume of washed packed red blood cells. The mixture was incubated in a water bath at 37 °C for 45 minutes and mixed every 5 minutes to enhance the reaction. The anti-D IgG-coated red blood cells were then washed three times with normal saline solution. Sensitised red blood cells were prepared as a 5% suspension in normal saline solution, and a direct antiglobulin test was performed. The optimal indicator cell was the dilution giving a score of 4+ in the direct antiglobulin test. The working dilution of the indicator cells was 0.2% in Alsever solution²⁶.

Platelet cross-match assays

Platelet cross-match assays were performed using a solid phase red cell adherence assay (SPRCA) technique with the MASPAT kit (Sanquin, Amsterdam, The Netherlands) for the detection of IgG antibodies to the HLA-A and HLA-B antigens found on platelets and to platelet-specific antigens. Briefly, donor platelets are first bound to the surface of polystyrene microplate wells. The patient’s serum is incubated in these wells; unbound immunoglobulin is then washed away and replaced by anti-IgG-coated “indicator” red blood cells. In the case of a positive reaction the anti-human IgG and MASPAT indicator red cells bind to the IgG-antibodies on the platelet monolayer. Positive reactions are thus characterised by adherence of MASPAT indicator red cells all over the surface of the wells. If there is no reaction (a negative test), the indicator cells do not bind to the platelets attached to the wall of the microtitre well, so the cells pellet to the bottom of the well, forming a red button. Results are reported as the number of incompatible (reactive) donor units present in the entire donor pool²⁷.

Statistical analysis

The statistical analysis of data was done by using the Excel and Statistical Package for Social Sciences (SPSS) version 16 programmes (SPSS Inc., Chicago, IL, USA). The Kolmogorov-Smirnov test was applied to test the normality of the data distribution. Significant data were considered to be non-parametric. Qualitative data are presented as frequencies and percentages. The chi-square test was used to compare groups. Quantitative data are presented as means and standard deviations. For comparison between two groups, the Student’s t-test, and Mann-Whitney test (for non-parametric data) were used. For comparison between more than two groups, ANOVA and the Kruskal-Wallis test (for non-parametric data) were used. Multivariate analysis was performed by unconditional logistic regression analysis. The associated risk is expressed as an odds ratio (with a 95% confidence interval). All p values are two-sided and p values <0.05 are considered statistically significant.

Results

Patients’ characteristics

The patients’ characteristics are summarised in Table I. Sixty percent (24/40) of refractory patients were females. Refractory women were younger, had a shorter interval from first transfusion to detection of refractoriness and had higher cross-match reactivity when compared with men (p=0.334, p=0.124 and p=0.015, respectively). Out of the 40 refractory patients, 35 (87.5%) had clinically detrimental factors and 16 (40%) had HLA/HPA alloimmunisation. Sixteen (40%) refractory patients died during the study period: causes of death were infection, disease progression and bleeding.

Table I.

Patients’ demographic and clinical data.

		Patients (n=40)
Age (years)		37.6±13.35
Sex (males/females)		16/24 (40.0/60.0%)
Diagnosis	Acute myeloid leukaemia	20 (50%)
	Acute lymphoblastic leukaemia	10 (25%)
	Chronic myeloid leukaemia	3 (7.5%)
	Myelodysplastic syndrome	1 (2.5%)
	Myelofibrosis	1 (2.5%)
	Aplastic anaemia	5 (12.5%)
Drugs	Chemotherapy	31 (77.5%)
	Antibiotics	38 (95.0%)
	Others	40 (100.0%)
Weight (kg)		76.67±18.69
Height (cm)		157.47±38.88
Body surface area (m2)		1.84±0.24
Number of units previously transfused	RBC (total=38 patients)	12.8±7.94
	Plasma (total=14 patients)	4.1±3.54
	Platelets (total=40 patients)	14.2±16.18
Time to detection of refractoriness (months)		1–14 (3.05±3.20)
Outcome (alive/died)		24/16 (60.0/40.0%)

Age, weight, height, body surface area, and history of previous transfusion are represented by mean±SD. Time to refractoriness detection is represented by range (mean±SD). Sex, diagnosis, drugs, and outcomes are represented by numbers and percentages.

Platelet cross-matching reactivity

A median of one cross-match assay was performed per patient (range, 1–4). The percent reactivity in each cross-match assay was defined as the number of incompatible donor units divided by the total number of donor units tested. The cross-match assays showed positive reactivity in 44 (36.7%) transfusion events and negative reactivity in 76 (63.3%). Patients were stratified into three groups based on initial cross-match reactivity: group 1 (n=4; 10%): 67% to 100% reactivity; group 2 (n=3; 7.5%): 34% to 66% reactivity; and group 3 (n=33; 82.5%): 0% to 33% reactivity (Table II). This stratification means that patients in group 1 had the fewest cross-match-compatible units identified among the tested donor pool, while patients in group 3 were compatible with the large majority of platelet units. A total of 406 donor units were tested during the cross-match procedures of which 52 were apheresis units collected for 15 patients from blood donors. Overall, 83.7% of tested units (340/406) were compatible and 16.3% (66/406) were incompatible. All compatible platelets (96 transfusions) and 66 incompatible platelets (24 transfusions) were given to our patients (Table III).

Table II.

Overall cross-match assay compatibility at first transfusion with patients subdivided by reactivity on cross-match assay.

	Reactivity (%)	Total	Group 1	Group 2	Group 3	P
		0–100 (%)	>66–100 (%)	>33–66 (%)	0–33 (%)
1st transfusion (n=40)	Number (%)	40 (100%)	4 (10.0%)	3 (7.5%)	33 (82.5%)	<0.0001
	Median	0	96.75	46.30	0
	Mean ± SD	16.61±29.14	92.98±10.19	43.21±8.75	4.94±9.19	<0.0001

Table III.

Characteristics of transfusion events.

Data	Total (n=120)	Cross-match compatible platelets (n=96)	Cross-match incompatible platelets (n=24)
N. of transfusions
SDP	52 (43.3%)	38 (39.6%)	14 (58.3%)
RDP	68 (56.7%)	58 (60.4%)	10 (41.7%)
N. of units	406	340	66
SDP	52 (12.8%)	38 (11.2%)	14 (21.2%)
RDP	354 (87.7%)	302 (88.8%)	52 (81.8%)
Platelet count (×1011)/transfusion
SDP	3.0±0.64	3.0±0.91	3.1±0.75
RDP	2.07±0.73	2.04±0.74	2.22±0.68
N. of transfusions/patient (median, range)	3 (1–4)	3.5 (1–4)	3 (1–4)
(mean±SD)	(3.0±1.11)	(3.06±1.13)	(2.67±1.03)
N. of units of platelets/patient (median, range)	6 (3–20)	6 (3–20)	4 (3–12)
(mean± SD)	7.05±4.35	7.32±4.50	5.71±3.32
N. of platelet units/transfusion	3.38±2.26	3.54±2.24	2.75±2.29
ABO compatibility status
ABO identical	78 (65.0%)	68 (70.8%)	10 (41.7%)
ABO minor incompatible	12 (10.0%)	8 (8.3%)	4 (16.7%)
ABO major incompatible	30 (25.0%)	20 (20.8%)	10 (41.7%)
Storage time (days)	2.10±1.27	2.21±1.30	1. 67±1.05
1 day	58 (48.3%)	42 (43.8%)	16 (66.7%)
2 days	18 (15.0%)	16 (16.7%)	2 (8.3%)
3 days	24 (20.0%)	20 (20.8%)	4 (16.7%)
4 days	14 (11.7%)	12 (12.5%)	2 (8.3%)
5 days	6 (5.0%)	6 (6.2%)	0 (0%)

SDP: single donor platelets; RDP: random donor platelets.

Platelet transfusion outcomes

The characteristics of all transfusion events are detailed in Table III. The mean ± standard deviation (SD) of the post-transfusion count observed after transfusion of compatible platelet products were significantly higher (p<0.001; 23.28±15.87×10⁹/L) than those observed in the same patients given 532 random platelet pools before cross-matching assays, when the post-transfusion count was 11.98±7.51×10⁹/L. The mean CCI for the study population after receiving random-donor platelet concentrates was −2.010±3.080×10³ (range, −10.000–1.800×10³. The mean CCI improved to 13.98±15.26×10³ (range, −7.97–86.14×10³ and 10.46±15.46×10³ (range, −8.46–80.62×10³ at 1 hour and 24 hours, respectively, when the same patients received cross-match-compatible platelets. Thus, there was a significant differences between the mean CCI following the administration of cross-match-compatible platelets and the mean CCI following administration of randomly selected platelets (p<0.001). There were also statistically significant increases in CCI at 1 hour and at 24 hours in patients received compatible versus incompatible platelet transfusions (p<0.001 for each) (Table IV).

Table IV.

Laboratory data of all transfusion events according to cross-match compatibility status.

	Total (n=120)		Cross-match compatible platelets (n=96)		Cross-match incompatible platelets (n=24)		P
	mean	SD	mean	SD	mean	SD
Pre-transfusion platelet count (×109/L)	12.94	9.94	13.48	10.62	10.76	6.67	0.123
1-hour post-transfusion platelet count (×109/L)	25.63	15.75	28.33	16.30	14.80	5.82	<0.001
24-hour post-transfusion platelet count (×109/L)	21.02	15.07	23.28	15.87	11.97	5.28	<0.001
CCI at 1 hour post-transfusion (×103)	11.72	14.52	13.98	15.26	2.72	4.79	<0.001
CCI at 24 hours post-transfusion (×103)	8.66	14.37	10.46	15.46	1.47	3.67	<0.001

CCI: corrected count increment.

Value of platelet cross-matching in predicting responses to platelet transfusion

Compatible and incompatible platelets were transfused in 96 (80%) and 24 (20%) transfusions, respectively. Transfusion response was evaluated at 1 hour and 24 hours after transfusion. A good response (CCI >5,000 at 1 hour and >2,500 at 24 hours) was reported in 72.9% (70/96) and 58.3% (56/96) of compatible transfusions at 1 and 24 hours, respectively. There were good responses to incompatible transfusions in 33.3% (8/24) and 50% (12/24) of cases at 1 and 24 hours, respectively. Additionally, incompatible platelets were predictive of poor platelet recovery at 1 hour (OR: 5.38; 95% CI: 2.060–14.073; p<0.001), but not at 24 hours (OR:1.400; CI: 0.572–3.434; p=0.46) (Table V).

Table V.

Role of platelet cross-matching in the prediction of response to platelet transfusions.

CCI			Cross-match compatible platelets (n=96)		Cross-match incompatible platelets (n=24)		P	OR	95% CI
At	CCI (×103)	Transfusion response	N.	%	N.	%
1h	>5	Good (n=78)	70	72.9	8	33.3	<0.001	5.385	2.060	14.073
	<5	Poor (n=42)	26	27.1	16	66.7
24h	>2.5	Good (n=68)	56	58.3	12	50.0	0.461	1.400	0.571	3.434
	<2.5	Poor (n=52)	40	41.7	12	50.0

CCI: corrected count increment; 1h: at 1 hour post-transfusion; 24 h: at 24 hours post-transfusion; OR: odds ratio; CI: confidence interval.

Influence of clinical factors on the prediction of platelet transfusion response by cross-match assay

Compatible platelet transfusions (n=96) were given to 81 patients with clinical factors that might affect the outcome of the transfusion and to 15 without such factors. The mean ± SD 1- and 24-hour post-transfusion counts observed in the presence of clinical factors (26.71± 17.17 and 22.04±16.48, respectively) were lower than those observed in the absence of such factors (37.09 ± 4.44 and 29.96 ± 10.08, respectively) with the differences being statistically significant (p<0.001 and p=0.019, respectively). Additionally, good platelet responses were observed in 55 (67.9%) and 41 (50.6%) compatible transfusions at 1 and 24 hours in the presence of clinical factors that might affect the transfusion outcome and in all 15 (100%) transfusions to patients without such factors: the differences were statistically significant (p=0.009 and p<0.001, respectively) (Table VI).

Table VI.

Influence of clinical factors on response to platelet transfusions.

	Transfusions (n=96)	Presence of non-immune factors (n=81)	Absence of non-immune factors (n=15)	P
		81 (84.4%)	15 (15.6%)
1h CCI	Good response	55 (67.9%)	15 (100.0%)	0.009
	Poor response	26 (32.1%)	0 (0%)
24h CCI	Good response	41 (50.6%)	15 (100.0%)	<0.001
	Poor response	40 (49.4%)	0 (0%)
Pre-transfusion platelet count (×109/L)		13.51±10.90	13.32±8.84	0.949
1-hour post-transfusion platelet count (×109/L)		26.71±17.17	37.09±4.44	<0.001
24-hour post-transfusion platelet count (×109/L)		22.04±16.48	29.96±10.08	0.019
CCI at 1 hour post-transfusion (×103)		12.08±15.63	24.25±7.00	0.004
CCI at 24 hours post- transfusion (×103)		9.46±16.47	15.86±5.91	0.141

1h CCI and 24h CCI are presented as numbers and percentages. Pre-transfusion platelet count, 24-hour post-transfusion platelet count, CCI at 1 hour post-transfusion and CCI at 24 hours post-transfusion are presented as means±SD.

Influence of alloimmunisation on the prediction of platelet transfusion response by cross-match assay

Sixteen patients were alloimmunised to HLA/HPA based on reactivity observed in the cross-match assay. Thirty-six compatible transfusions were given in the presence of alloimmunisation and 60 in the absence of alloimmunisation. The mean ± SD 1- and 24-hour post-transfusion counts in the presence of alloimmunisation (24.51±18.0 and 21.57±17.93, respectively) were lower than those observed in the absence of alloimmunisation (32.44±13.19 and 25.14±13.23, respectively) (p=0.015 and p=0.274, respectively). Good platelet recovery was observed in 14 (38.9%) compatible transfusions in the presence and in 56 (93.3%) and 42 (70%) transfusions in the absence of alloimmunisation at 1 and 24 hours with the differences being statistically significant (p<0.001 and p=0.003, respectively) (Table VII).

Table VII.

Influence of alloimmunisation on response to platelet transfusions.

Transfusions (n=96)		Presence of alloimmunisation (n=36)	Absence of alloimmunisation (n=60)	P
		36 (37.5%)	60 (62.5%)
1h CCI	Good response	14 (38.9%)	56 (93.3%)	<0.001
	Poor response	22 (61.1 %)	4 (6.7%)
24h CCI	Good response	14 (38.9%)	42 (70.0%)	0.003
	Poor response	22 (61.1 %)	18 (30.0%)
Pre-transfusion platelet count (×109/L)		13.43±11.64	13.53±9.38	0.964
1 hour post-transfusion platelet count (×109/L)		24.51±18.00	32.49±13.19	0.015
24 hour post-transfusion platelet count (×109/L)		21.57±17.93	25.14±13.23	0.274
CCI at 1 hour post-transfusion (×103)		10.77±15.97	17.47±13.77	0.031
CCI at 24 hours post- transfusion (×103)		9.81±19.08	11.16±10.36	0.670

1h CCI and 24h CCI are presented as numbers and percentages. Pre-transfusion platelet count, 24-hour post-transfusion platelet count, CCI at 1 hour post-transfusion and CCI at 24 hours post-transfusion are presented as means±SD.

Multivariate analysis

Multivariate analysis was done to detect which variables (platelet cross-matching, clinical factors, alloimmunisation, ABO compatibility and type of received units) can predict transfusion response. Cross-matching and alloimmunisation were the best predictors of transfusion response at 1 hour (p=0.005 and p<0.001, respectively), while ABO compatibility, type of received units, followed by alloimmunisation then clinical factors were predictors at 24 hours (p=0.004, p=0.016, p=0.043 and p=0.045, respectively).

Correlation between mean corrected count increment after transfusion of cross-matched platelets and the degree of reactivity

We investigated the correlation between the response to transfusion of cross-matched platelets, in terms of mean CCI, and the degree of reactivity observed in the initial cross-match assay. In patients with a low degree of cross-match reactivity, the large majority of random platelet units would actually be cross-match-compatible; thus, obtaining cross-matched platelets would not lead to much improvement in CCI over prior ineffective response to random units. Patients with higher degrees of average cross-match reactivity (>60% reactivity) have a larger mean CCI benefit in response to cross-matched platelets. There was, however, no correlation between mean CCI response and mean degree of cross-match reactivity at 1 hour and 24 hours (p=0.41 and p=0.82, respectively). Likewise, cross-match reactivity was not related to ABO compatibility, type of received units, storage time days, and clinical factors (p>0.05 for each)

Discussion

A number of approaches have been developed to address the problem of platelet transfusion refractoriness. One of the most frequently used methods is HLA matching which can be highly effective and is the routine approach to the management of refractory patients in a number of institutions²⁵. A commonly used alternative to HLA-matched platelets is the transfusion of cross-match-compatible platelets^3,14. Given the widespread use of cross-matched platelets, there are surprisingly few reports describing the benefit obtained from using a SPRCA assay to identify cross-match-compatible platelets^{3,11,19,20,23,24,28}. We decided to use the SPRCA technique because it was simpler, less time-consuming and cheaper than other methods²⁹. Consequently we examined the usefulness of the SPRCA technique in assisting the blood bank in providing platelet components for patients refractory to platelet transfusions. We did not specifically exclude patients with fever, splenomegaly, coagulopathy, or other potential clinical causes of refractoriness, because such patients reflect the reality of platelet management. Thirty-six of 40 (90%) patients included in the study were transfused with compatible platelets during the course of their illness. The mean CCI of 13,980 achieved with compatible platelets in this study corresponds to a platelet count increase of 28.33×10⁹/L, which is sufficient to avoid significant spontaneous bleeding. This CCI response to cross-matched units was significantly higher than that to comparable random platelet units for these patients, demonstrating benefits from cross-match compatibility. The response to compatible platelets seen in our study is also consistent with that in prior studies^3,24,30 which demonstrated a significant improvement in CCI by using the SPRCA method to cross-match platelets. Platelet cross-matching was found to be a good predictor of transfusion response. Transfusion of compatible platelets was more successful in 72.9% of transfusion than incompatible platelets (33.3%) (p<0.001). This good response to compatible platelet transfusions is consistent with the results of a previous study by Rebulla et al.,³ who reported good platelet recovery in 68% of evaluable transfusions, although using an automated SPRCA technique. Sayed et al.³¹, however, reported a good response in 57.7% of compatible transfusion events, which may be due to the use of flow-cytometric platelet cross-matching, a more sensitive method for cross-matching. Our threshold for good response (CCI>5000), which was consistent with the thresholds described by Rebulla¹⁴ and Sayed et al.³¹, is lower than the commonly used threshold (CCI>7500) for successful transfusions. A better prediction of transfusion response was observed when clinical factors were absent than when they were present. Many previous studies^24,32,33 have found that the ability of cross-matching to predict response to transfusions may be lower in unselected patients with refractoriness to platelet transfusion than in those without associated clinical factors but their patients were did not come from specific age groups. Sayed et al. noted that the predictive role of cross-matched platelets on transfusion response is more affected by the presence of clinical factors in adults than in children. The same results were obtained when alloimmunisation is absent. The better transfusion response after exclusion of clinical factors and HLA alloimmunisation reflects the importance of transfusing cross-matched platelets to all patients, especially to those who have neither clinical factors nor HLA alloimmunisation³¹. The incidence of alloimmmunisation in our study was 40% (16/40), which was higher than the 3–4% and 13–14% recently documented in large groups of chronic recipients of leucoreduced and non-leucoreduced platelets^33,34. This likely points to the use of non-leucoreduced random platelet concentrates in our institution.

Only 18 patients had multiple cross-match assays (4 assays). The CCI response to cross-matched platelets averaged across those patients remained relatively constant indicating that there was no trend towards increasing alloimmunisation during management with cross-matched platelets. So, patients with consistent CCI (or compatible cross-match) will benefit from leucoreduced single donor apheresis. Leucoreduction decreases alloimmune platelet refractoriness by decreasing exposure to donor antigen-presenting cells. As in our study, other studies also found no evidence of a significant increase in alloimmunisation to unmatched HLA antigens over time but the selection of compatible platelets was based on the patient’s HLA antibody specificity, which is an alternative to the gold standard of matching for HLA antigens³⁰.

In our study, women were more frequently refractory than men (60%), refractory female patients were younger and the interval between first transfusion and development of refractoriness was shorter than that in refractory males although the difference was not statistically significant. These findings may be attributed to the development of HLA alloimmunisation during pregnancy. The proportion of refractory women was lower than that found by Rebulla et al.³, who reported that more than 80% of the refractory patients in their study were women. Further data from larger studies may help to determine whether HLA primary alloimmunisation developed at pregnancy may have a causative role in these differences.

By using multivariate analysis, platelet cross-matching was found to be the best predictor of transfusion response, followed by clinical factors and alloimmunisation, at 1 hour. Meanwhile, the most important predictor of platelet survival at 24 hours was ABO compatibility, followed by type of received units, alloimmunisation and then clinical factors. To the best of our knowledge, multivariate analysis was used rarely in previous studies. Heal et al.³⁵ stated that ABO antigens have independent predictive value on platelet survival. Sayed et al.³¹ also reported that platelet cross-matching, clinical factors and HLA alloimmunisation are predictors of transfusion response. In conclusion, our study suggests that platelet cross-matching using a commercially available immunoadherence assay may be a useful way of rapidly selecting effective platelets from the local inventory for transfusion support of patients with platelet refractoriness.