The advent of modern transfusion medicine improved surgical survival and morbidity; however, the legacy of transfusion-related blood-borne infections lingers in public perception. Although the implementation of strategies to avoid transfusion were once an infrequent occurrence limited to populations such as Jehovah’s Witnesses or individuals with rare antibodies, individuals preparing for surgery are increasingly likely to ask to avoid transfusion.Reference Spahn and Goodnough 1
The safety of allogeneic blood products is constantly improving. Yet, despite improved donor screening and testing, the risk of viral infection remains.Reference Spahn and Goodnough 1 The risks also include noninfectious complications such as hemolysis from clerical errors, lung injuries, and volume overload.Reference Goodnough 2 Although transfusions requirements may reflect more challenging surgical cases or an increased burden of disease, they have been shown to be associated with worse outcomes in perioperative patients including postoperative infections, most typically pneumonia and bacteremia,Reference Horvath, Acker and Chang 3 and increased length of stay.Reference Spahn and Goodnough 1 In addition to rigorous transfusion protocols, surgical planning and optimized intraoperative hemostasis, there are several approaches to limit blood loss and avoid allogeneic red blood cell (RBC) transfusion, including diagnosis and management of preoperative anemia, use of hemostatic medications, normovolemic hemodilution, controlled hypotensive anesthesia, salvaging and reinfusing RBCs lost intraoperatively, and preoperative RBC donation through preoperative autologous donation (PAD).Reference Spahn and Goodnough 1
An additional method to manage anemia and hasten erythroid recovery is to use erythroid-stimulating agents such as erythropoietin (EPO). The advent of recombinant erythropoietic-stimulating agents provided a mechanism for improving erythroid status in many settings, including renal disease, myelodysplastic syndromes, and anemia of chronic disease.Reference Bohlius, Schmidlin and Brillant 4 When used to prevent perioperative allogeneic blood products, meta-analyses of randomized controlled trials indicate that recombinant erythropoiesis-stimulating agents reduce allogeneic RBC transfusion in cardiovascularReference Alghamdi, Albanna, Guru and Brister 5 , Reference Laupacis and Fergusson 6 and orthopedic surgeries.Reference Laupacis and Fergusson 6 Furthermore, when used in conjunction with PAD, recombinant erythropoiesis-stimulating agents reduce the risk of allogeneic transfusion in orthopedic and cardiovascular surgeries.Reference Alghamdi, Albanna, Guru and Brister 5 - Reference Shapiro, Boachie-Adjei, Dhawlikar and Maier 7
Early experience in neurological surgery with recombinant EPO to avoid RBC transfusion involved successful use in Jehovah’s Witnesses in both spineReference Joseph, Berekashvili and Mariller 8 and intracranial cases.Reference Schiff and Weinstein 9 , Reference Kantrowitz, Spallone, Taylor, Chi, Strack and Feghali 10 There was a prospective open-label randomized control trial of erythropoiesis-stimulating agents in spinal surgeryReference Colomina, Bago, Pellise, Godet and Villanueva 11 ; however, to our knowledge, there are limited data on its use in nonspinal neurological surgeries.
Although meta-analyses have shown that PAD reduces exposure to allogeneic RBC, it also results in an increased exposure to transfusion when both autologous and allogeneic RBCs are considered,Reference Forgie, Wells, Laupacis and Fergusson 12 most likely because of iatrogenic anemia. It is unknown whether this is a consideration when planning blood conservation efforts in neurosurgery.
The Ontario Transfusion Coordinators (ONTraC) Provincial Blood Conservation program is involved in patient blood management efforts in 27 sites across Ontario, Canada. This program seeks to minimize the use of allogeneic blood products through various techniques, including management of anemia with EPO and PAD. The program specifically targets certain surgical procedures such as coronary artery bypass grafting and arthroplasty, but will accept referrals for patients undergoing other surgeries.
The aim of this study was to characterize RBC transfusion requirements among neurosurgical patients who received erythropoiesis-stimulating agents with PAD and without versus comparison subjects who were matched for age, sex, and preoperative diagnosis.
Methods
This study was approved by the St-Michael’s Hospital Research Ethics board. This was a retrospective case control study. No mention of patient status as a Jehovah’s Witness was identified in any of the cases studied.
Subjects
Cases
The ONTraC program was launched in 2002 with the aim of minimizing unnecessary use of allogeneic blood products. It coordinates patient blood management programs in institutions of all sizes in Ontario. ONTraC’s targeted procedures include total knee and hip arthroplasty/replacement, coronary artery bypass graft surgery, and radical prostatectomy. ONTraC nurses also consider referrals for patients undergoing other surgical procedures.
Between 2002 and 2009, 420 patients received recombinant EPO preoperatively through the ONTraC program at St-Michael’s Hospital. Fifty-seven of these were neurosurgical patients and were retained in the current study. International Classification of Diseases (ICD)-10 codes for the preoperative diagnosis were extracted for all cases to permit comparison subject matching (Table 1). Ultimately, three cases were not included in the final analyses (encephalocele n=1, cerebral cyst n=2) because comparison subjects could not be identified that respected our matching protocol.
Table 1 Preoperative diagnoses
ICD=International Classification of Diseases.
Blood Conservation
Cases received iron supplementation and an average of 2.44±1.00 40,000 IU subcutaneous doses of EPO preoperatively. Recombinant EPO for the purposes of blood conservation is not covered under Canadian socialized health care; therefore, the costs of the treatment were covered by insurance providers (n=27, 47.3%), the patients themselves (n=13, 22.8%), compassionate access program (n=8, 14.0%), or by the Ontario Drug Benefit program (n=9, 15.7%). Concomitant to receiving EPO, 32 (59.2%) cases also elected to participate in a preoperative autologous donation program. The patients donated an average of 1.78±1.13 whole blood units (range 1-3 units); one patient donated 3 units twice because of rescheduling.
Comparison Subjects
Controls were identified through the St-Michael’s Hospital registry. Cases were matched 2:1 to comparison subjects identified among elective neurosurgical patients. To be included, comparison subjects could not be emergency cases. Cases were matched for ICD-10 preoperative diagnosis, sex, age (±3 years), and year of surgery (±3 years). ONTraC files were consulted to ensure that no comparison subject participated in blood conservation programs, including preoperative autologous RBC donation or intraoperative reinfusion. Thus, 108 matched comparison neurosurgical patients were identified and included in this study.
RBC Transfusion
Participation in the ONTraC program does not result in a mandated transfusion guideline. All decisions to proceed to RBC transfusions were managed by the patients’ most responsible physician as clinically indicated. This is unlikely to have resulted in any systematic biases with respect to transfusions.
Data Extraction
Patients’ charts were reviewed for information pertaining to preoperative diagnosis, age, and sex. Preoperative neurosurgical, medical, and anesthesia consultations were reviewed for comorbidities—specifically hematological, renal, hepatic, pulmonary, and cardiovascular—and medications, including antiplatelet and anticoagulant use. Laboratory values, namely hemoglobins and platelet counts, were extracted when available.
The primary outcome for this study was allogeneic RBC transfusion. Secondary outcomes were RBC units transfused and perioperative hemoglobins, particularly day-of-surgery hemoglobins. Additional analyses are presented for relevant laboratory values.
Statistical Analyses
Statistical analyses were performed with SPSS v19 (SPSS Inc., Chicago, IL). For categorical variables, we used chi-square tests (with odds ratio and 95% confidence interval). For continuous variables, we used Student-t test or analysis of variance. Significance was set at α≤0.05.
Results
Demographic and clinical characteristics of EPO cases, subdivided according to PAD, and comparison subjects are presented in Table 2. Cases and controls were of identical sex distribution (24.1% male, 75.9% female) and comparable age (50.31±13.78 vs 50.83±14.15). Some clinical differences emerged: cases who received EPO without PAD were more likely to have comorbid diabetes mellitus, and preoperative hemoglobin was lower for cases that received EPO both with and without autologous predonation. Of note, the difference between patients who did not receive any blood conservation measures versus those who received EPO alone the difference in preoperative hemoglobins was about 10 g/L. Cases were otherwise comparable with respect to medical comorbidities, anticoagulant, and antiplatelet use.
Table 2 Demographic and clinical features
Hb=hemoglobin; M=mean; Plts=platelets; RBC=red blood cell; SD=standard deviation.
Surgical matching was also achieved, with spinal surgeries involving comparable number of levels (2.58±0.79 vs 2.20±0.76 levels, p≥0.19). Similarly, mean tumor volume did not differ between groups (29.16±34.71 vs 28.85±43.17 cm3, p≥0.97).
Allogeneic RBC Transfusion
Cases and controls did not differ with respect to exposure to allogeneic RBCs (Figure 1a), and an identical proportion (18.5%) were exposed to allogeneic RBCs. Similarly, they did not differ with respect to the number of allogeneic units transfused (Figure 2a; F(2, 161)=0.34, p≥0.70). When distinguishing cases that received EPO and predonated blood, no difference in exposure to allogeneic RBCs was noted (Figure 1b; χ2=1.61, df=2, p≥0.44). Cases that required an allogeneic transfusion did not differ from patients who did not with respect to the number of EPO doses received (2.50±1.26 vs 2.43±0.94, p≥0.84).
Figure 1 Proportion of patients exposed to allogeneic RBC units subdivided according to (a) whether or not they received EPO and (b) whether they received EPO in conjunction with PAD or not. In (c), the proportion of patients who engaged in PAD and received EPO who required RBC transfusion (allogeneic and autologous). ***p≤0.001.
Figure 2 Mean allogeneic RBC requirements for (a) patients who received EPO and comparison subjects, (b) subdivided EPO patients into those who also engaged in PAD or did not, and (c) the total RBC transfusion requirements (allogeneic and autologous) for patients with PAD and EPO. *p≤0.05.
Recombinant EPO and PAD
Because 32 (59.2%) of cases concomitantly participated in preoperative EPO and PAD, we also considered autologous RBC transfusion. When considering all units of RBCs transfused, cases who received EPO and predonated autologous RBCs were more likely to have been transfused (Figure 1c; 46.9%; χ2=13.39, df=2, p≤0.001) and have received a greater number of RBC units (Figure 2b; F(2, 161)=3.77, p≤0.05).
We characterized cases’ hemoglobin levels before EPO and on the day of surgery (Figure 3). Analyses revealed a significant interaction between predonation status×time (F(1, 46)=6.49, p≤0.05) after controlling for the main effects of predonation status (F(1, 46)=0.87, p≥0.35) and time (F(1, 46)=3.58, p≥0.06).
Figure 3 Hemoglobin prerecombinant erythropoietin and preoperatively. *p≤0.05.
Cases that concomitantly participated in preoperative autologous donation did not receive a different number of EPO doses than patients who did not (2.56±0.98 vs 2.27±1.03 t(52)=1.04, p=0.30). Of these cases, 15 (46.9%) were transfused autologous RBCs. Ultimately, 58.6% of the autologous units donated (34 units) were not transfused and were discarded.
Transfusion Threshold
Because participation in ONTRaC does not involve mandated transfusion guidelines and PAD RBCs may be viewed as more benign, we examined the final hemoglobins for our transfused cases and comparison subjects as proxies for transfusion threshold. There were no differences between patients who had received transfusions belonging to the EPO group (99.66±13.86 g/L), the EPO with PAD group (98.73±10.57 g/L), or the comparison group (103.33±13.36 g/L; F(2, 35)=.59, p≥0.55).
Discussion
Using a case-control design, our analyses suggest that PAD does not reduce the risk of allogeneic RBC transfusion in neurological surgery, despite having been used in conjunction with EPO. Indeed, PAD was associated with an increase in transfusion requirements when both allogeneic and autologous RBCs are considered. Our data suggest that PAD resulted in patients going into surgery with lower hemoglobin levels, and therefore that autologous donation rendered patients anemic despite the concomitant use of EPO. Indeed, without reducing the risk of exposure to allogeneic RBC, concomitant EPO and PAD resulted in a greater number of total RBC units transfused. These findings are in keeping with systematic reviews and meta-analyses demonstrating that PAD is associated with lower preoperative hemoglobin levelsReference Henry, Carless and Moxey 13 and increases the risk of any transfusion.Reference Forgie, Wells, Laupacis and Fergusson 12 , Reference Henry, Carless and Moxey 13
Although our analyses cast doubt on the utility of PAD, our data cannot determine whether EPO used in isolation is effective at reducing allogeneic RBC exposure in neurosurgery. We recommend careful selection of patients for patient blood management; these should include patients with preoperative anemia and cases associated with a significant risk of blood loss and thus allogeneic transfusion, patients objecting to transfusions, and perhaps those for whom it may be difficult to find compatible RBCs.
Further, there are counterbalanced data suggesting that recombinant EPO is not without risk. The oncology literature suggests that chronic use of erythropoiesis-stimulating agents is associated with poorer long-term survivalReference Bohlius, Schmidlin and Brillant 4 , Reference Harlaar, Gosselink, Hop, Lange, Busch and Jeekel 14 ; however, this remains controversial.Reference Glaspy, Crawford and Vansteenkiste 15 , Reference Ludwig, Crawford and Osterborg 16 Adverse events associated with chronic use of EPO include hypertension, thromboembolism, and iron deficiency.Reference Singbartl 17 When given acutely, as in blood conservation programs, the risk of each of these drops significantly, yet the risk of thromboembolism remains elevated (3-9%).Reference Laupacis and Fergusson 6 , Reference Singbartl 17 , Reference Stowell, Jones, Enny, Langholff and Leitz 18 As such, the risks may not be counterbalanced in light of limited benefit of EPO, and additional research is required to justify the use of such agents because they may expose neurosurgical patients to unnecessary harm.
Although the individual and societal value of avoiding allogeneic blood products should not be minimized, the utility of PAD and EPO in neurosurgery to prevent allogeneic RBC transfusion should be carefully evaluated. Although efficacy has been well demonstrated in surgeries with a high risk for RBC transfusion—such as cardiac surgery where transfusion rates are approximately 40%Reference Hutton, Fergusson, Tinmouth, McIntyre, Kmetic and Hebert 19 but approach 90% in some samplesReference Bennett-Guerrero, Zhao and O'Brien 20 —rates of transfusion in neurological surgery are far lower, between 2% and 25%.Reference Goodnough, Saha, Hirschler and Yomtovian 21 , Reference Linsler, Ketter, Eichler, Schwerdtfeger, Steudel and Oertel 22 Intrinsically, the possible absolute risk reduction will be constrained by a floor effect in surgeries with a lower risk of RBC transfusion; therefore, the ability to demonstrate the efficacy of blood conservation strategies in neurological surgery will be limited. Our results suggest that PAD and EPO may not be equally effective across surgical interventions. Indeed, EPO has not been demonstrated effective in colorectal cancer surgery,Reference Devon and McLeod 23 and the same may be true in neurological surgery.
Considering the low likelihood of requiring a transfusion in neurosurgery, PAD may not be warranted. PAD may be associated with increased risk of adverse events during donation as compared with allogeneic donation.Reference Popovsky, Whitaker and Arnold 24 PAD is associated with increased risk of requiring any transfusionReference Forgie, Wells, Laupacis and Fergusson 12 , Reference Henry, Carless and Moxey 13 and does not eliminate the risk of transfusion complications such as volume overload or acute hemolysis resulting from a clerical error. The increased transfusion requirements of neurosurgical patients undergoing PAD was demonstrated even in conjunction with EPO, which highlights the need for further research into the benefit of PAD in neurosurgery.
We could not identify the reasons for which patients elected to participate in blood conservation strategies; no cases self-identified as Jehovah’s Witness. Analyses examining pre-EPO hemoglobins suggest that preoperative anemia may have been the primary reason among those who did not participate in PAD, whereas personal preference may explain those who did. Indeed, 17 patients received recombinant EPO that was specially funded by the Ontario Drug Benefit program or through compassionate use to treat low hemoglobin. These same 17 patients did not, then, donate autologous RBC; because of their low hemoglobin levels, they were not deemed to be candidates for PAD. With our data, it is not possible to conclude that preoperative erythropoietin did not optimize erythroid status and hasten postoperative erythroid recovery and thereby prevent blood transfusions in these patients.
Economic Cost-Benefit
The economic cost-benefit analysis of preoperative EPO and autologous donation in neurosurgery deserves discussion. Canadian Blood Services reports the cost of an RBC unit to be $367 CAD, 25 including collection and testing. However, the activity-based cost of transfusing an RBC unit is closer to $760 CAD.Reference Shander, Hofmann, Ozawa, Theusinger, Gombotz and Spahn 26 Autologous blood must undergo the same infectious disease and pretransfusion testing as allogeneic blood. However, about two-thirds of autologous units are routinely discarded (Pavenski, personal communication, 2013) suggesting that the true cost of producing an autologous unit is in fact much higher compared with an allogeneic one. At the same time, the Ontario Drug Benefit formulary lists recombinant EPO alfa 40,000 IU at $450 CAD per dose.
In light of these costs, neurosurgical cases in the current study received recombinant erythropoietin at a mean cost of $1098, before administration costs. Cases participating in autologous predonation did so at a mean cost of $1352.80 CAD. Most important, whether EPO was used in isolation (representing 40.8% of the sample) or in conjunction with PAD (59.2% of the sample; mean per capita cost $2504.80), it did not reduce the absolute risk of receiving allogeneic RBC, nor did it reduce the number of allogeneic RBC units transfused compared with comparison subjects matched for age, sex, and preoperative diagnosis; with comparable tumor burden or number of spinal levels; and medical comorbidities.
Under Canadian transfusion standards, autologous units cannot cross over into the general blood inventory because autologous blood collection does not involve the rigorous screening of donors for potential exposures required for allogeneic blood. If unused, autologous units can only be discarded. Almost 60% of the autologous RBC units in this cohort were ultimately disposed of, at a cost exceeding $25,000.
Limitations
This study is subject to limitations inherent in retrospective case-control designs; nevertheless, we used strict matching criteria and achieved a comparison population with similar demographic and clinical characteristics. Unfortunately, because of our strict matching procedure, three cases were not retained in the final analyses because comparison patients could not be identified. Moreover, we could not, by design, control for the surgeons and anesthetists who were involved in the surgeries. Similarly, we did not capture operative duration and this should be addressed in future studies. Although we matched according to ICD-10 codes and achieved similar tumour burden, we cannot definitely rule out more vascular lesions among our cases, and angiographic characterization should be considered in future studies.
A major limitation of our study is that cases and controls were not matched for hemoglobins, and this methodological shortcoming may have influenced our results. Further, because RBC transfusions were initiated according to the clinical judgment of the patient’s most responsible physician, additional research should include a standardized transfusion protocol that controls for this potential confound. Although the use of final hemoglobin as a proxy for transfusion thresholds suggests the absence of bias and that autologous RBCs were not viewed as a more benign intervention, we cannot definitively exclude this possibility. Another limitation is that our data did not systematically capture adverse events that may have been related to PAD or EPO.
Conclusion
In this cohort of patients undergoing neurological surgery, preoperative EPO did not result in a lower rate of allogeneic RBC transfusion. When used in conjunction with preoperative autologous donation, it similarly did not reduce the risk of allogeneic RBC transfusion or the number of transfused allogeneic units. Our data do not address whether EPO used in isolation reduces the risk of exposure to allogeneic RBCs; analyses revealed that our patients who received EPO but did not engage in PAD nevertheless went into surgery with a 10 g/L hemoglobin deficit compared with controls without a corresponding increase in transfusion requirements. The utility of PAD and EPO in neurological surgery is deserving of additional research using randomized prospective designs, as the non-neglible cost and side-effect profile of these strategies require sound evidence to justify their implementation in neurosurgery.
Acknowledgments
We thank Alanna Howell, Anna Nassis, and Katherine Luke, Ontario Transfusion Coordinators nurses at St-Michael’s Hospital, for their blood conservation efforts and support of this research.
