Safety of autologous drainage blood reinfusion following total knee arthroplasty prepared with hydrogen peroxide☆

1. Introduction 

The most common late complication following cemented total joint arthroplasty surgery is aseptic loosening, which accounts for almost one-half of knee revisions in Sweden [1]. A factor in the loosening of arthroplasty components is the mechanical integrity of the interface between cement and bone. Research has been directed toward enhancing the bone-cement interface, in an attempt to reduce this complication. An ideally prepared bony surface is one which is dry, clean of marrow and tissue debris, and free from active bleeding. Bleeding from cancellous bone causes lamination within bone cement and at its prosthetic interfaces, and weakens the fixation of joint replacements. Haemostatic agents have been popularized as effective means of retarding the development of potentially harmful debris interposition adjacent to, and blood lamination patterns within, the methylmethacrylate [2]. One such agent is hydrogen peroxide. In an investigational study hydrogen peroxide reduced bleeding in human cancellous bone by 14% [3].

Howells et al. demonstrated that bone surface preparation with hydrogen peroxide irrigation provided statistically superior cement fixation, compared with normal saline or povidone iodine, in an arthroplasty model [4]. The authors postulated that hydrogen peroxide produces an effervescence that mechanically removes fat, blood, marrow and tissue debris from both surface interstices, as well as normally inaccessible regions of the bony microstructure. The resultant improved mechanical interlocking increases the strength of the bone-cement interface.

Hydrogen peroxide is inexpensive, readily available and also has mild antiseptic properties. In Australia, on the strengths of this local research, hydrogen peroxide irrigation became a popular adjunct to conventional cementing techniques. The paucity of published scientific literature regarding hydrogen peroxide irrigation in total knee arthroplasty, suggests this may not be common practice internationally.

Hydrogen peroxide decomposes by the action of catalases, releasing oxygen and water. The rate kinetics of this reaction is not described in surgical literature. Following implantation the knee joint cavity is commonly rinsed and drained prior to wound closure. Residual hydrogen peroxide substrate is therefore further diluted. The quantity of active substrate remaining upon tourniquet release has not been investigated, but is assumed to be extremely small.

The majority of blood loss occurs after tourniquet release following total knee arthroplasty. Autologous drainage blood reinfusion systems utilize this phenomenon to salvage and transfuse shed blood back to the patient. Randomized controlled trials have shown a 64–86% reduction in units of allogenic blood transfused and a reduction in the number of patients requiring allogenic blood by 62–75% [5], [6], [7], [8]. However, “presence in the blood of a substance unsuitable for reinfusion such as peroxide of hydrogen” is currently a contraindication for the use of the Bellovac ABT (Astra Tech AB, Mölndal, Sweden) autologous drainage blood reinfusion system [9].

The effects of residual hydrogen peroxide on shed blood destined to be collected in a reinfusion drainage system have not been reported. The aim of this study was to test the hypothesis that drainage blood is not significantly altered in this scenario in a prospective controlled trial.

3. Results 

Results for thirty-eight patients (twenty-one in the ‘peroxide’ group and seventeen in the ‘control’ group) were available for analysis. Seven patients (15%) were excluded post-operatively: three due to incomplete data collection, two due to equipment failure, with the drain tube becoming disconnected from the patient and two because of breach of protocol. In one such case the drain tube was not unclamped after the wound dressing and bandage was applied. In the other, the drain blood specimen was left at room temperature overnight. Blood film examination of this specimen revealed cellular changes consistent with storage.

Pre-operative patient demographics and characteristics are presented in Table 2. The patient groups were statistically similar for age, pre-operative haemoglobin and MCV. Although there were a higher proportion of female patients in the ‘control’ group, analysis of outcomes by gender showed no significant difference.

Table 2. Pre-operative patient characteristics

		Peroxide	Control	p
		(n=21)	(n=17)
	Mean (range) age (yr)	69.0	69.3
		(48.8–88.0)	(51.4–84.8)	0.9015
	Gender
	Male	10 (48%)	2 (12%)
	Female	11 (52%)	15 (88%)
	Mean (S.D.) pre-operative haemoglobin (g/L)	137.7 (15.08)	133.8 (10.50)	0.3109
	Mean (S.D.) pre-operative MCV (fL)	92.6 (4.68)	90.6 (4.86)	0.1953

Parameters generated intra- and post-operatively are presented in Table 3. The majority of ‘peroxide’ group patients had cemented arthroplasty, compared to the majority of ‘control’ group receiving uncemented components. The ‘peroxide’ group had a mean peroxide time of 19.4 min. No statistically significant differences were observed between the groups in tourniquet time, collection time, handling time or warm time. A significant difference was shown for the time taken for wound closure and the volume of drained blood. Analysis by mode of implant fixation revealed no significant difference for volume of drained blood between the operation types.

Table 3. Operative and post-operative parameters

		Peroxide	Control	p
		(n=21)	(n=17)
	Mode of implant fixation
	Cemented	17	1
	Hybrid	4	3
	Uncemented	0	13
	Mean (S.D.) tourniquet time (min)	73.7 (13.69)	67.9 (7.79)	0.1761
	Mean (S.D.) peroxide time (min)	19.4 (3.93)	–
	Mean (S.D.) closing time (min)	27.9 (6.12)	35.5 (7.80)	0.0018
	Mean (S.D.) collection time (min)	224.1 (101.1)	279.9 (111.1)	0.1129
	Mean (S.D.) handling time (min)	46.6 (58.82)	66.5 (73.87)	0.0789
	Mean (S.D.) warm time (min)	270.7 (112.5)	341.4 (157.5)	0.2145
	Mean (S.D.) volume of drained blood (mL)	513.4 (86.71)	422.0 (141.6)	0.0191

The outcome measures are presented in Table 4. The p-values are greater than 0.05 for all variables therefore there is no statistical difference between the groups. There was no correlation between the time taken for wound closure and any other outcome. Furthermore, there was no correlation between volume of drained blood and potassium, plasma free haemoglobin or methaemoglobin.

Table 4. Outcome measures from collected drain blood

		Peroxide	Control	p
		(n=21)	(n=17)
	Mean (S.D.) haemoglobin (g/L)	98.6 (20.76)	100.9 (13.12)	0.7221
	Mean (S.D.) change from pre-operative haemoglobin (g/L)	−39.1 (13.18)	−32.9 (11.11)	0.2117
	Mean (S.D.) MCV (fL)	94.6 (4.46)	93.1 (5.23)	0.2897
	Mean (S.D.) change from pre-operative MCV (fL)	2.0 (2.29)	2.5 (1.90)	0.6417
	Mean (S.D.) potassium (mmol/L)	4.5 (0.44)	4.6 (0.50)	0.8212
	Mean (S.D.) potassium dose (mmol)	2.3 (0.49)	1.9 (0.73)	0.0945
	Mean (S.D.) plasma free haemoglobin (g/L)	1.2 (0.6)	1.3 (0.71)	0.4367
	Mean (S.D.) plasma free haemoglobin dose (g)	0.6 (0.31)	0.6 (0.43)	0.4282
	Mean (S.D.) methaemoglobin (%)	0.2 (0.21)	0.2 (0.14)	0.8112
	Samples with echinocytes present on blood film	3 (14%)	3 (18%)	1.0000

4. Discussion 

The study compared parameters in drained blood following total knee arthroplasty prepared using hydrogen peroxide with controls. Outcomes represented surrogate measures of potential toxicity rather than direct measurement of the presence of residual hydrogen peroxide. Direct or indirect hydrogen peroxide assay was not available at the hospital's pathology laboratory. Detrimental effects of hydrogen peroxide on shed blood destined to be reinfused relate to its reaction with haemoglobin, causing oxidative stress [15].

A review of the literature for the effect of incubation of erythrocytes in the presence of hydrogen peroxide provided consistent outcomes. Oxidation of haemoglobin forms methaemoglobin at a low concentration of peroxide [16]. Erythrocyte morphologic changes occur in a dose-dependent manner with increased echinocyte formation [16] Assessment of these parameters provides a reliable index of adverse oxidative effects. Dose-dependent decrease in cell deformability with increase in membrane rigidity was found [17], [18], [19]. The shape of the erythrocyte during echinocyte transformation influences its deformability by the fact that while the surface area increases, the cell volume remains essentially constant. The normal discocyte represents an optimum shape for the flow in vivo and an echinocytic transformation compromises rheologic behavior and could impair the flow in larger vessels, by increasing blood viscosity [20].

van den Berg et al. observed that peroxide-induced oxidative stress is at a maximum immediately after addition of this oxidant and decreases rapidly to zero in a short time [21]. Applied to the arthroplasty scenario, the effects of hydrogen peroxide added to the knee joint would be short-lived and may be approaching zero by the time the tourniquet is released. Any unaffected erythrocytes entering the joint would not be oxidized. This would likely explain why there was no significant increase in methaemoglobin or echinocytes in the drain blood of patients from the ‘peroxide’ group. Given this outcome, it is unlikely that cells with compromised rheologic behavior or increased susceptibility to lysis would be reinfused.

Acidic incubation occurs in the drain container because of production of lactate from glucose. In the absence of hydrogen peroxide, Dalen et al. found maximum erythrocyte haemolysis after 24 h of incubation was less than 1% [22]. Extensive cell lysis has been observed after prolonged incubation at high oxidant concentrations [21] Erythrolysis liberates potassium and free haemoglobin.

Potassium is cardiotoxic at a relatively small increased concentration. The protocol advised by Astra Tech for autologous transfusion allows 500 mL of filtered drain blood to be infused within a minimum time of 1 h. The maximum rate for peripheral intravenous potassium administration for unmonitored patients on the general ward must not exceed 10 mmol/h [23]. At 500 mL/h, potassium in drain blood from ‘control’ groups at a mean concentration of 4.6 mmol/L would be infused at a rate of 2.3 mmol/h. This is well within the safe range for potassium administration. Potassium concentration in drain blood would need to be increased at least four-fold before approaching unsafe limits. This was not demonstrated in the ‘peroxide’ group, with a mean drain blood potassium concentration of 4.5 mmol/L.

Free haemoglobin is not directly harmful. It is only a marker that erythrocyte membrane fragments have been released. The toxic dose of free haemoglobin or the degree of haemolysis required to cause nephrotoxicity is unknown. The acceptable level of haemolysis has not been established in North America, but the value of 1% currently is used to assess biocompatibility of blood storage materials, whereas the Council of Europe has set the standard at 0.8% [24]. A mean level of free haemoglobin of 417 g/L at 6–8 h due to haemolysis of drain blood has been previously reported [25]. This was estimated to be a result of haemolysis of only 0.5% of the erythrocytes and there was no documented renal tubular damage. The free haemoglobin levels observed in this study were much lower and not significantly affected by the use of hydrogen peroxide.

The surgical technique prohibits prolonged incubation by methods of irrigation and complete evacuation of the joint when the tourniquet is released. The study results do not demonstrate evidence of increased erythrocyte swelling, lysis or lysis products when hydrogen peroxide was used. Therefore, it would seem to be a safe assumption that there is probably no increased risk of patient morbidity or mortality. Our protocol dictated that the tourniquet is released prior to wound closure and that 1 L of irrigation be used. We accept that surgeons may prefer to release the tourniquet after wound closure or use a smaller volume of irrigation. It is difficult to extrapolate the study results to these scenarios because the rate of decomposition of hydrogen peroxide to oxygen and water is unknown. Late tourniquet release may be of additional benefit if peroxide decomposition is assumed to be slow. The specific volume of irrigation may be irrelevant if the applied peroxide has been quickly exhausted. The important issue is that the joint cavity be thoroughly irrigated to dilute any residual peroxide, to prevent contamination of fresh blood eventually destined for collection in the wound drain and reinfusion.

The main limitation of this study was the small sample size. Post hoc power analysis was performed using the sample size formula presented by Lachin [26] and assuming a 5% significance level. Power for potassium dose was 0.98 (beta error= 0.02) and for haemoglobin change 0.70 (beta error=0.30). High beta errors resulted for the other outcome measures (MCV change 0.92, potassium 0.94, free haemoglobin 0.96, and free haemoglobin dose and methaemoglobin 0.98), thus demonstrating a high likelihood of false-negative results for these parameters. The reason for this is because the observed difference between groups was small, in the context of a limited sample size. We argue that this small difference does not demonstrate a clinically relevant difference with regard to the safety of subsequent blood reinfusion.

The surgical technique was sufficiently controlled to prevent bias, especially as our outcome measures were laboratory-derived parameters of collected drain blood. However, varied modes of implant fixation were used and the groups were different with respect to the proportion of cemented compared to uncemented arthroplasties. The presence of cement is not a contraindication to the use of autologous drain blood reinfusion. Whether or not the cement may have in any way affected the drain blood was not investigated and this issue has not been reported in the literature. Veikkolin et al. observed that post-operative bleeding was more extensive after uncemented arthroplasty [27]. Analysis by operation type in this study revealed that bleeding was slightly increased after cemented arthroplasty, but the difference not significant.

There were more females than males in the ‘control’ group, but we feel this could not have affected the results. The influence of hydrogen peroxide on erythrocytes is independent of gender, particularly as red blood cells contain no nucleus or genetic material. Rates of blood lost following knee surgery are not known to differ between sexes. Furthermore, analysis of outcomes by gender showed no significant difference in results.

The mean handling times were longer than anticipated in the study design. We compensated by measuring warm time, which adjusted for differences in post-operative bleeding rates. Dalen et al. observed that acidic incubation occurs in the drain container because of production of lactate from glucose, with a minimum pH at 5 h of 7.2 [22]. The low pH caused slight but significant erythrocyte swelling. This was observed, with a slight increase in MCV in both groups. The potassium ion concentration in whole blood is known to change during storage at ambient temperature, [28] and centrifugation is recommended within 2 h of sampling [29]. This was achieved in the majority of cases. No reports of significant changes in the other outcome measures when blood is stored at room temperature for short periods have been found.

We acknowledge there is paucity of published scientific literature regarding the use hydrogen peroxide as an irrigating fluid in total knee arthroplasty, suggesting this may not be common practice internationally. Although it is not the intention of the authors to dissuade surgeons from using hydrogen peroxide, concerns have been raised in the literature over the potential effects on biologic tissues, including bone and soft tissues. At a low concentration, hydrogen peroxide was found experimentally to markedly inhibit glucose metabolism and collagen synthesis, and to decrease bone weight and alkaline phosphatase activity [30]. In rat osteoclast cultures hydrogen peroxide stimulated osteoclastic bone resorption [31] Kaysinger et al. advised caution in using irrigation solutions containing hydrogen peroxide on exposed bone tissue, having found these solutions were cytotoxic to both embryonic chick bones and isolated osteoblast cells at concentrations well below those used clinically [32] Similarly, Nicholson et al. found that cell lysis or death occurred after exposing osteoblasts to 30 mmol hydrogen peroxide, concentrations substantially lower than the 3% (880 mmol) hydrogen peroxide commonly used clinically [33].

In a recent report, Guerin et al. postulated that the use of hydrogen peroxide as an irrigation solution in arthroplasty affects the material properties of bone cement and actually contributes to the development of aseptic loosening in the long term. Their preliminary studies indicated that porosity increases and that the tensile strength and yield stresses are reduced by up to a factor of ten by contaminating samples with increasing concentrations of hydrogen peroxide [34]. Exercising caution has also been advised to minimize erosion of prostheses consisting of hydroxyapatite or Ti–6Al–4V alloy when hydrogen peroxide solutions are used [35].

To date, no study has analyzed uncemented components for effects of hydrogen peroxide on bone ingrowth or resultant outcomes of aseptic loosening. However, osteoinductivity of cortical bone allograft is maintained with cleaned in hydrogen peroxide for up to 1 h, and compressive strength, impact strength, and shear strength were demonstrated to be unaffected [36].

Other concerns relate to reports of gas embolism after intraoperative use of hydrogen peroxide. A case is reported in which cardiac arrest immediately followed the use of hydrogen peroxide during preparation of the femoral canal during hip arthroplasty, likely due to oxygen embolism [37] The authors suggested that the use of peroxide in an unvented femoral canal may be hazardous, as it has been shown to be in other closed cavities in the body [38]. However, the risk of this complication occurring during knee surgery, with conditions of an open joint and an inflated tourniquet, is extremely remote.

The use of suction drains in total knee arthroplasty has been common practice, until the last decade. A trend away from routine use has been promoted by recent scientific evidence. A meta-analysis in 2004 by Parker et al. indicated no significant difference in the occurrence of wound infection related to wound drains [39] Other outcome measures, such as reoperation, wound hematoma, dehiscence, later drainage from the wound, limb swelling, deep-vein thrombosis, range of movement, pain, function, return to work, strength, and hospital stay, showed no difference between the groups managed with a drain and those managed without a drain. The only definite advantage of the use of drains was the reduction in the amount of blood leaking through the wound, as demonstrated by the reduced number of dressing reinforcements in the group treated with a drain. A definite advantage for patients managed without a drain was reduced transfusion requirements. With increasing concerns about the risk of transmission of infection and possible immune suppression with blood transfusion, surgeons are citing this finding as an important argument against the use of drains.

Adalberth et al. compared two groups of thirty patients each, who were prospectively randomized to receive either an autotransfusion system or no drain after total knee arthroplasty [40]. No significant differences were seen between the groups in haemoglobin and haematocrit values, drainage volume and transfusions (homologous and autologous), range of knee motion, knee swelling or hospital stay. There was no benefit demonstrated from using an autotransfusion system. Despite this finding, the authors do not intend to influence surgeons' current practice of post-operative drain management or to discourage the use of autologous drainage blood reinfusion systems in selected cases. Such cases may include patients who could not accept allogenic transfusion, either for religious reasons or immunological intolerance, and for whom severe post-operative anaemia would cause unacceptable physiological decompensation.

The clinical benefit of hydrogen peroxide as an irrigation solution during total knee arthroplasty has not been clearly established. There may be adverse effects of peroxide on the local tissues, prostheses or cement, which may be detrimental to long-term outcome. This remains to be proven. The future of closed suction autotransfusion drains also may be in jeopardy, outweighed by the benefits of ‘drainless’ total knee arthroplasty. Despite these issues, we conclude that hydrogen peroxide, under the specific conditions that prevailed in this study, does not result in detrimental alterations of the blood salvaged in the post-operative drain. The results should be considered preliminary because of the limitation of the small study size. In concordance with the study conditions, a surgical technique permitting tourniquet control, thorough lavage and drainage immediately after tourniquet deflation, and a delay in drain unclamping prior to blood collection is recommended. Although hydrogen peroxide is not routinely used for bone preparation by all, blood remains safe for reinfusion for those surgeons who wish to use hydrogen peroxide and this autologous blood saving technique concurrently.