Platelet and Red Blood Cell Utilization and Transfusion Independence in Umbilical Cord Blood and Allogeneic Peripheral Blood Hematopoietic Cell Transplants

Abstract

Allogeneic hematopoietic cell transplantation (HCT) recipients have substantial transfusion requirements. Factors associated with increased transfusions and the extent of blood product use in umbilical cord blood (UCB) recipients are uncertain. We reviewed blood product use in 229 consecutive adult recipients of allogeneic HCT at the University of Minnesota: 147 with leukemia, 82 lymphoma or myeloma; 58% received unrelated UCB and 43% sibling donor peripheral blood stem cell (PBSC) grafts. Although neutrophil recovery was prompt (UCB median 17, range 2–45 days, and PBSC 14, range 3–34 days), only 135 of 229 (59% cumulative incidence, CI) achieved RBC independence and 157 (69%) achieved platelet independence by 6 months. Time to platelet independence was prolonged in UCB recipients (median UCB 41 vs. PBSC 14 days) and in patients who had received a prior transplant (median 48 vs. 32 days). Patients who received UCB grafts required more RBC through day 60 post HCT (mean UCB 7.8 (95% CI 6.7–8.9) vs. PBSC 5.2 (3.7–6.7) transfusions, p=0.04), and more platelet transfusions (mean 25.2 (95% CI 22.1–28.2) vs. 12.9 (9.4–16.4), p<0.01) compared to PBSC recipients. Patient receiving myeloablative (MA) conditioning required more RBC and platelet transfusions during the first 2 months post HCT compared to reduced intensity conditioning (RIC) (7.4 vs. 6.2, p=0.3 for RBC; 23.2 vs 17.5, p=0.07 for platelets). Despite prompt neutrophil engraftment, UCB recipients had delayed platelet recovery as well as more prolonged and costly blood product requirements. Enhanced approaches to accelerate multilineage engraftment could limit the transfusion-associated morbidity and costs accompanying UCB allotransplantation.

INTRODUCTION

Over the past decade there has been an increase in the use of allogeneic hematopoietic stem cell transplants (HCT) to treat hematologic malignancies and life-threatening non-malignant hematologic disorders. This increase is attributable to newer techniques, safer preparative regimens, and expanded use of peripheral blood stems cells (PBSC) and umbilical cord blood (UCB) as graft sources [1]. Even using newer, less intensive regimens, patients undergoing HCT will receive frequent blood products as part of their supportive therapy [2]. In light of substantial cost and some risk associated with blood product use, we pursued a detailed study of red blood cell (RBC) and platelet (PLT) transfusion needs during a HCT course. Limited available data on the extent of blood product utilization after HCT includes Osterwalder et al reporting 39 HCT patients needing a range of 1–32 RBC and 1–11 PLT transfusions [2] and Pihlstedt et al describing 182 allogeneic transplants patients receiving 1–63 RBCs and 2–394 PLT transfusions [3]. We reviewed the blood product use in 229 consecutive patients at our institution and assessed factors associated with greater transfusion needs.

MATERIALS AND METHODS

Patients and methods

Using prospectively collected data from the University of Minnesota Blood and Marrow Transplant Database we reviewed demographics and HCT complications in 229 consecutive adult patients who underwent allogeneic HCT. This was supplemented with complete and detailed transfusion data for the initial 3 months after HCT. We also detailed the preparative regimen, gender, race, number of prior transplants to determine associated factors which may have altered their blood product utilization, RBC and PLT engraftment and time to transfusion independence. The patients were divided into four groups based on the underlying disease: leukemia (n=147), lymphoma/myeloma (n=82), non-malignant (43), and Fanconi anemia (7).

The formal analysis was limited to the more homogeneous subset of patients with either leukemia or lymphoma/myeloma (N=229). The patients received a preparative regimen classified as either myeloablative (MA) plus filgrastim-mobilized sibling donor PBSC; reduced intensity conditioning (RIC) plus sibling donor PBSC; Umbilical cord blood (UCB) MA or UCB RIC. Patients were given prophylaxis for graft-versus-host disease (GVHD) and supportive care based on the defined institutional guidelines and active investigational protocols.

Transfusion Support and Endpoints

Prophylactic platelet transfusions were administered to non-bleeding patients at a pre-transfusion trigger of less than 10,000/μl unless additional risks of bleeding (e.g. mucositis, plasma coagulation disorder, sepsis) were present. Treatment for therapeutic bleeding followed WHO guidelines. Red blood cell transfusions were administered for Hgb < 8 g/dL. All products were irradiated in vitro with 25 Gy prior to transfusion.

Transfusion independence was defined as the day of the last, post-HCT PLT transfusion with no PLT transfusions in the following 7 days. Red blood cell transfusion independence was defined as the day of the last RBC transfusion with no transfusion in the following 30 days. The total number of transfusion episodes and the total transfused PLT and red blood cell units during the transplant period were determined. Incidence and time to hematologic and red blood cell independence was determined for the total (N=229) and separately for the leukemia (N=147) and lymphoma/myeloma subgroup (N=82).

Statistical analysis

Comparison of patients’ characteristics used the X² test or Fisher’s exact test for categorical variables, and Wilcoxon’s rank-sum test for continuous variables. The cumulative incidence of RBC and PLT independence was used by treating deaths from other causes as competing risks. Univariate analysis was performed for differences in PLT independence and RBC independence while risk factor analysis of time to RBC transfusion independence and time to PLT transfusion independence was performed using competing risk regression analysis. Risk factors considered included: donor and graft source, conditioning regimen (myeloablative versus RIC), race, gender and HCT number for those with a prior HCT. All p values reported were 2-sided and p values <0.05 were considered statistically significant. Marginal mean was used to report on the monthly and total number of PLT and RBC units used. Marginal mean is defined as the weighted average of the conditional means, with weights equal to the probability of being in the subgroup determined by the corresponding value of the conditioning variable. Data analysis was performed using SAS software, version 8.0. The University of Minnesota IRB approved the data collection and analysis.

RESULTS

Patient characteristics

We studied 229 consecutive allogeneic adult HCT recipients with hematologic malignancies receiving either PBSC or UCB grafts (Table 1). Of these, 140 were males (61%), 196 were Caucasian (86%), 131 (57%) received UCB and 98 (43%) PBSC grafts. One hundred and eighteen (52%) received RIC and 111 (48%) MA conditioning. Thirty-nine (17%) had a previous (38 autologous; 1 allogeneic) HCT. In the entire group, hematopoietic reconstitution manifest as sustained neutrophil recovery was prompt (UCB median 17, range 2–45 days, and PBSC median 14, range 3–34 days).

Table 1.

Patient Demographics and Graft Source

	N	Percent
Total	229	100%
Graft Source & Conditioning Intensity
PBSC RIC	46	20%
PBSC Myeloablative	52	23%
UCB RIC	72	31%
UCB Myeloablative	59	26%
Sex
Male/Female	140/89	61/39%
Race
White/Other	196/31	86/14%
Transplant Number
1	190	83%
>1	39	17%
Diagnosis
Leukemia	147	64%
Lymphoma/Myeloma	82	36%

PBSC: peripheral blood stem cell; RIC: reduced intensity conditioning;

UCB: umbilical cord blood

Transfusion independence

Time to transfusion independence by donor type and conditioning regimen subgroups is shown in Table 2. RBC independence occurred at a median of 44 (range 7–141) days and PLT independence at a median of 33 (range 11–172) days after HCT. While median time to RBC independence did not differ based upon donor source or HCT number, it was somewhat longer after myeloablative conditioning (median 55 (range 9–141) days vs. 42 (7–127) days, p=0.43), and among women (56 days vs. 42 days, p=0.26). The time to PLT independence was prolonged in UCB recipients compared to PBSC, though it did not reach statistical significance (41 (range 18–172) days vs. 14 (11–161) days, p=0.52) and in recipients of second HCT (48 days vs. 32 days, p=0.09). By 6 months after HCT, only 76 of 131 (58%) UCB recipients vs. 59 of 98(60%) PBSC recipients were independent of RBC transfusions and 92 of 131(70%) UCB and 65 of 98(66%) PBSC were independent of PLT support.

Table 2.

Time to RBC and Platelet Transfusion Independence

		RBC Independence		Platelet Independence
	N	RBC Independent by 6 months (N)	Median time to independence (range, days)	Platelet Independent by 6 months (N)	Median time to independence (range, days)
Total	229	135	44 (7–141)	157	33 (11–172)
Graft Source
UCB	131	76	44.5	92	41
PBSC	98	59	44	65	14
Conditioning
RIC	118	68	41.5	80	29
Myeloablative	111	67	55	77	35
Transplant number
>1	39	17	44	21	48
1	190	118	44.5	136	32

RBC: red blood cells; PBSC: peripheral blood stem cells; UCB: umbilical cord blood

Transfusion independence within Disease Subgroups

Because either diagnosis or prior therapies confounded the assessment of transfusion needs, we analyzed the cumulative transfusion requirements of patients with leukemia and with myeloma/lymphoma separately. Of 147 patients with leukemia, 98 (67%) received UCB grafts, 90 (62%) received myeloablative conditioning regimen, and 126 patients (86%) had no prior transplant. Eighty-three patients (56%) achieved RBC independence within 6 months of transplant with a median time to RBC independence of 49 (range 11–160) days. Time to RBC independence was somewhat longer in females (63 vs. 43 days, p=0.27) and after myeloablative conditioning (56 vs. 43 days, p=0.21) but did not reach statistical significance. Ninety-five patients (65%) achieved PLT independence by 6 months with a median time of 35 days. Time to PLT independence was longer yet not significant after UCB (43 versus 25 days, p=0.43) and in recipients of second transplants (57 days vs. 34 days, p=0.16). However, within the leukemia group we observed no statistically significant differences in the time to PLT transfusion independence based on donor source (Figure 1), conditioning regimen, race, gender or number of transplants. Among the patients who received RIC conditioning, median times to RBC and PLT transfusion independence in the UCB subgroup were 41.5 and 39 days and were not different from the PBSC subgroup (58 and 12 days).

Figure 1.

Time to RBC and Platelet independence

In leukemia patients, shown is the cumulative incidence of the time to transfusion independence (untransfused Hb>8 g/dl; PLT >20,000/μl) in recipients of UCB or sibling donor PBSC grafts.

Of 82 patients with lymphoma/myeloma 32 (39%) had UCB grafts, 21 (26%) had myeloablative conditioning and 18 (22%) had received a prior transplant. Fifty-two patients (63%) achieved RBC independence by 6 months with a median time to independence of 43 days. The median time to RBC independence was somewhat longer in females (52 vs. 35 days, p=0.27), recipients of UCB (40 vs. 10 days, p=0.10), those with prior HCT (54 vs. 35 days, p=0.02) and following myeloablative conditioning (47 vs. 41 days, p=0.03). 62 patients (75%) achieved PLT independence by 6 months with a median time of 21 days. The time to PLT independence was longer yet not statistically significant in recipients of UCB vs. PBSC, (40 vs 10 days, p=0.16) and also in those with a prior transplant (48 vs. 15 days, p=0.01). There were no differences in time to PLT independence based on conditioning regimen (21 days for myeloablative vs. 20 days for RIC) or other clinical factors.

Multivariate Regression Analysis of Transfusion Independence

Using competing risk regression analysis we evaluated factors influencing the time to platelet or RBC independence in the different disease subgroups. In the lymphoma/myeloma group, there was a statistically significant longer time to RBC independence using myeloablative conditioning (p=0.03) and following second HCTs (p=0.02). Similarly, time to PLT independence was significantly longer after MA conditioning (p=0.01) and second HCT (p=0.009). There was no significant difference based on the graft source (UCB vs. PBSC, p=0.16). However, within the larger subgroup of patients with leukemia, no factors were independently associated with the time to achieve either RBC or PLT independence.

Transfusion Requirements in Patients with Leukemia

Patients in the larger leukemia subgroup were further studied to detail the number of transfusions required during each of the first 3 months after HCT. The marginal mean number of RBC and PLT transfusions received in each month was compared in cohorts defined by the graft source (UCB vs. PBSC) and conditioning regimen intensity (Myeloablative vs RIC) (Table 3).

Table 3.

Marginal Mean RBC and Platelet Transfusions (HCT for Leukemia)

			MONTH 1			MONTH 2			MONTH 3
	N		Mean	95% CI	P*	Mean	95% CI	P	Mean	95% CI	P*
All	146	RBC	4.5	(4.0–5.0)		7.0	(6.0–7.9)		7.7	(6.6–8.8)
		Platelets	14.5	(13.6–15.7)		21.2	(18.7–23.8)		23.7	(20.2–27.2)
Graft Source
UCB	98	RBC	4.8	(4.2–5.4)	0.18*	7.8	(6.7–8.9)	0.04	8.9	(7.5–10.2)	0.02
PBSC	48		3.9	(3.1–4.8)		5.2	(3.7–6.7)		5.4	(3.7–7.0)
UCB		Platelets	16.1	(14.7–17.6)	<0.01	25.2	(22.1–28.2)	<0.01	28.5	(24.1–32.8)	<0.01
PBSC			11.0	(8.7–13.2)		12.9	(9.4–16.4)		13.6	(9.3–17.9)
Conditioning Intensity
Myeloablative	90	RBC	4.8	(4.3–5.4)	0.16	7.4	(6.3–8.5)	0.3	8.4	(7.0–9.8)	0.17
RIC	56		4.0	(3.1–4.9)		6.2	(4.6–7.8)		6.4	(4.6–8.3)
Myeloablative		Platelets	15.8	(14.5–17.2)	<0.01	23.2	(20–26)	0.07	26.4	(22.2–30.6)	0.07
RIC			12.1	(9.6–14.6)		17.5	(12.9–22.2)		18.4	(12.5–24.2)

^*p values represent each time point comparison of the marginal mean of RBC and Platelet transfusions for graft source and conditioning intensity

As compared to PBSC, UCB recipients required a greater number of RBC transfusions during each of the first 3 months post transplantation and this reached statistical significance for months 2 and 3. Patients who received MA conditioning also had a greater need for PLT transfusions (Table 3). Similarly, during each of the first 3 months post HCT, UCB recipients required significantly more PLT transfusions than PBSC (p<0.01). The marginal mean and total number of both PLT and RBC transfusions was significantly higher in UCB recipients during the first 3 months post HCT (Figure 2). Myeloablative conditioning was associated with significantly increased PLT requirements during month 1, and a similar trend in months 2 and 3, but RBC requirements were similar in both conditioning groups.

Figure 2.

Marginal Mean of Platelet (A) and RBC transfusion (B) over time

Shown are the marginal mean of cumulative PLT (panel A) and RBC (B) transfusions over the first 6 months post HCT for patients with leukemia. UCB recipients have significantly greater mean PLT (p<0.01) and RBC (p=0.02) transfusion needs compared to PBSC.

DISCUSSION

Recipients of HCT often have substantial transfusion requirements, particularly in the minority with either delayed recovery of hematopoiesis, alloimmunization to PLTs or GVHD. Even using standard thresholds for blood product administration such as RBC for Hb <8 gm/dl or PLT for a count <10,000/μl, transfusion needs vary substantially. The PLADO trial questioned the necessity of PLT prophylactic transfusions by noting that major bleeding episodes are infrequent and not clearly associated with PLT counts, even in the severely thrombocytopenic patient [4]. The majority of hospitalized allogeneic HCT recipients are vulnerable to infections or febrile, have added risks bleeding due to mucositis, plasma coagulation abnormalities or similar complications that suggest a need for a higher PLT count trigger for transfusion. We restricted our detailed analysis of total blood product usage and time to transfusion independence in allogeneic HCT patients with leukemia because they are usually multiply transfused and possibly alloimmunized prior to HCT.

Transfusion guidelines in HCT patients are traditionally extrapolated from other subgroups of pancytopenic patients, as specific data from HCT are less well defined [5–8]. The threshold for PLT transfusion in HCT recipients has changed over time, with earlier guidelines suggesting 20,000/μl as an acceptable threshold for prophylaxis [9]. Several reports showed that a platelet threshold of less than 20,000/μl can be safely used in acute leukemia and HCT recipients [10–12]. In some reports, lowering the transfusion threshold from 20,000 to 10,000/μl in stable marrow transplant recipients did not increase severe bleeding or death due to bleeding, but reduced PLT use by about 25% [13]. The current threshold of 10,000/μl is used by most centers to initiate prophylactic PLT transfusions [5, 8, 14].

The demand for RBC transfusion in HCT has decreased in recent years. This could be related to the use of less intensive regimens and implementing G-CSF into the transplant process, hence decreasing oral or gastrointestinal bleeding episodes related to prolonged mucositis[15]. A transfusion trigger of Hb < 8g/dL is used at our and most centers although higher thresholds are indicated for patients with cardiovascular co-morbidities[16]. Higher thresholds have also been suggested for transplant recipients with acute GVHD, delayed engraftment, major ABO incompatibility, and high risk malignancies [3].

Reduced intensity conditioning (RIC) is more widely used in recent years due to the decreased risk of associated toxicities and the reliance on a more pronounced graft versus tumor effect for cancer control [17]. Prebet et al reported data on PLT recovery and transfusion needs in 145 adult allogeneic sibling donor HCT recipients with RIC[18]. PLT recovery (>20,000/μl) was seen at a median of 9 days with 68% of the patients demonstrating a PLT count >100,000/μl by day +100. A lower PLT count prior to conditioning and the presence of acute GVHD (grade III–IV) significantly influenced day +100 PLT recovery and PLT transfusion needs. Similarly, the median time to PLT independence in our RIC cohort is 14 days, even with 61% receiving UCB grafts which may delay engraftment compared to other donor sources [19]. Other unrelated donor sources and MA conditioning are also associated with delayed PLT recovery. Median time to more robust PLT recovery (>50,000/μl) in UCB HCT has been reported to be as late as day +90; and the success was related to the patient’s age, diagnosis, graft cell dose infused, post-HCT infection, and the development of GVHD[20]. Although the incidence of PLT transfusion independence at 6 months did not differ, we observed that UCB grafts led to delayed PLT independence compared to other donor sources [21] and an overall greater transfusion burden. Delayed engraftment, a precursor to PLT recovery in UCB recipients has been associated with a lower CD34+ cell dose, recipient weight, and HLA disparity [22].

We also observed less frequent and later transfusion independence in recipients having a prior HCT. In our cohort of 39 repeat HCT recipients, this delay could be due to the graft source, to prior marrow stromal damage from extensive prior treatment or to their specific diagnoses (most often secondary AML/MDS) although we could not fully characterize the pathogenesis of delayed PLT recovery in this small subgroup. In comparison, Baron et al reported on 147 patients receiving a very low intensity, non-myelablative regimen followed by either matched related (N=62) or unrelated (N=85) allogeneic RIC HCT due to disease relapse after a prior autologous or allogeneic transplant[23] who needed only a median of zero and 4 for PLT and RBC transfusions, respectively. Martino et al reported a median of 11 days to obtain a PLT count of >20,000/μL in a similar group of sibling donor RIC HCT recipients who had relapsed after a prior autologous transplant [24].

RBC transfusion needs in the first 60 days after allogeneic HCT were observed at a median of 4 RBC units for RIC and 12 units for MA conditioning [25]; slightly more in patients with leukemia (6 units for RIC and 13 units for MA). Another report described infusion of only 2 RBC units after RIC [26]. Factors associated with increased RBC transfusion requirements include acute GVHD, ABO incompatible donor, and high risk malignancy [3]. Our data shows that both MA conditioning and UCB grafts are associated with higher transfusion requirements.

We previously analyzed the increased cost for UCB compared to matched related or unrelated donor HCT with 11–14% of the costs in different HCT settings attributable to transfusion product needs [27]. Our current data shows that patients receiving UCB grafts required a mean of 15 extra platelet transfusions and 4 extra RBC transfusions during the first 3 months post HCT. Based on costs from the Centers for Medicare and Medicaid services 2007 reimbursement report on blood transfusion (Leukoreduced Irradiated RBC unit: $227; Apheresis PLT irradiated leukoreduced: $614), the increase in transfusion requirements seen with UCB grafts augmented transfusion-associated costs by approximately $11,000 during the first 3 months post HCT, equating to $122 per day, plus additional costs for pre-medications, infusion supplies and the nursing effort for transfusion and monitoring.

Our observations suggest that RBC and PLT transfusion requirements are somewhat increased with UCB grafts, though are moderated by RIC HCT. The use of UCB as a donor source may delay transfusion independence and modestly increase the overall costs of HCT as well. Additional efforts to improve transfusion practices and augment hematopoietic recovery may limit both the transfusion burden and the costs of prolonged RBC and PLT transfusions.