Autologous blood transfusion reduces the requirement for perioperative allogenic blood transfusion in patients undergoing major hepatopancreatobiliary surgery: a retrospective cohort study

Abstract

Introduction:

Major hepatopancreatobiliary surgery is associated with a risk of major blood loss. The authors aimed to assess whether autologous transfusion of blood salvaged intraoperatively reduces the requirement for postoperative allogenic transfusion in this patient cohort.

Materials and methods:

In this single centre study, information from a prospective database of 501 patients undergoing major hepatopancreatobiliary resection (2015–2022) was analysed. Patients who received cell salvage (n=264) were compared with those who did not (n=237). Nonautologous (allogenic) transfusion was assessed from the time of surgery to 5 days postsurgery, and blood loss tolerance was calculated using the Lemmens–Bernstein–Brodosky formula. Multivariate analysis was used to identify factors associated with allogenic blood transfusion avoidance.

Results:

32% of the lost blood volume was replaced through autologous transfusion in patients receiving cell salvage. Although the cell salvage group experienced significantly higher intraoperative blood loss compared with the noncell salvage group (1360 ml vs. 971 ml, P=0.0005), they received significantly less allogenic red blood cell units (1.5 vs. 0.92 units/patient, P=0.03). Correction of blood loss tolerance in patients who underwent cell salvage was independently associated with avoidance of allogenic transfusion (Odds ratio 0.05 (0.006–0.38) P=0.005). In a subgroup analysis, cell salvage use was associated with a significant reduction in 30-day mortality in patients undergoing major hepatectomy (6 vs. 1%, P=0.04).

Conclusion:

Cell salvage use was associated with a reduction in allogenic blood transfusion and a reduction in 30-day mortality in patients undergoing major hepatectomy. Prospective trials are warranted to understand whether the use of cell salvage should be routinely utilised for major hepatectomy.

Introduction

Highlights

• Major hepatopancreatobiliary (HPB) surgery is associated with a risk of major blood loss.

• We analysed the use of cell salvage in a cohort of 501 patients undergoing major HPB surgery.

• The use of cell salvage reduced the absolute number of patients needing allogenic transfusion as well as the average number of units transfused per patient.

• Correction of blood loss tolerance was independently associated with reduced allogenic transfusion.

• In a subgroup analysis limited to hepatectomy alone, there was a significant reduction in 30-day mortality.

• This is the largest analysis of cell salvage use in the HPB field and supports routine use in major HPB resection.

Major hepatopancreatobiliary (HPB) surgery is associated with a risk of significant blood loss, with volumes ranging from 200 ml to 2000 ml dependent on the surgical approach and size of the operation^1–4. Intraoperative blood loss is directly correlated with major morbidity^5,6, and allogenic transfusion is associated with the potential for worse oncological outcomes in patients undergoing cancer surgery^4,7–11. Blood is a scarce resource and allogenic transfusion is associated with complications inclusive of anaphylaxis, iatrogenic microbial infection, transfusion associated circulatory overload or lung injury and febrile nonhaemolytic reactions^12,13. The three pillars approach to patient blood management inclusive of optimisation of endogenous red cell mass, bleeding and blood loss minimisation and optimisation of physiological tolerance to anaemia has^14,15 become central to the improving perioperative outcomes and reducing the financial costs of major surgery¹⁶.

Various methods have been developed to manage blood loss in HPB surgery. In major liver resection, the pringle manoeuvre significantly reduces intraoperative blood loss¹⁷ and is frequently used. For liver parenchymal transection, many technologies have been developed, although none demonstrably reduce blood loss when compared to the standard clamp-crush technique^18–21. In pancreatic surgery, minimally invasive approaches appear to reduce blood loss, with robotic surgery particularly favourable in nonrandomised studies^22,23, although there is no equivalent data for liver surgery. Haemostatic agents applied intraoperatively can be effective for the control of minor bleeding^24–26, but are unlikely to substantially influence transfusion requirement or postoperative outcomes. Finally, anaesthetic techniques to lower central venous pressure intraoperatively are associated with a reduction in estimated blood loss and transfusion during liver resection^27–29.

Intraoperative cell salvage and autologous transfusion of spilled blood is an additional, cost-effective approach to intraoperative blood management³⁰. It aims to replace circulating volume as it is lost, maintaining perfuson, avoiding coagulopathy if major haemorrhage ensues and providing blood with potentially better oxygen carrying capacity than cold-stored allogenic blood. This has been of demonstrable benefit for the reduction of allogenic transfusion in randomised trials of cardiac^31,32, obstetric³³ and orthopaedic surgery³⁴ as well as in small, retrospective liver resection and transplantation series^35–37. In addition, we recently demonstrated complete avoidance of allogenic transfusion in patients who lost at least 50% of their circulating blood volume during aortic aneurysm surgery³⁸. There is comparatively little evidence supporting the use of cell salvage in HPB surgery. To address this, we aim to assess whether the use of intraoperative cell salvage reduces allogenic transfusion in patients undergoing major HPB surgery.

Materials and methods

Surgical setting and patients

A retrospective analysis of a prospectively-maintained database was performed to identify consecutive patients who underwent major HPB resection in our institution from January 2015 to June 2022. Our HPB unit within the Oxford University Hospitals NHS Trust is a tertiary referral unit in the UK, serving a population of 2.4 million and performing over 200 HPB resectional procedures per year. Five HPB consultant surgeons and five dedicated HPB anaesthetists cover elective major HPB surgery in Oxford.

Operative inclusion criteria for this study were pancreaticoduodenectomy, total pancreatectomy or major liver resection, where major liver resection involved resection of greater than or equal to 3 vascular segments. Patients with incomplete data were excluded, as were those who refused transfusion on religious or other grounds. Internal ethics approval was sought and granted by the Clinical Improvements Division of Oxford University Hospitals NHS Foundation Trust. The study was retrospectively registered in the Australian New Zealand Clinical Trials Registry (ANZCTR).

Inclusion criteria for the use of cell salvage in our centre are any patient undergoing major hepatectomy (≥3 liver segments), and pancreatic resections predicted to require vascular reconstruction. Because cell salvage is a shared service between surgical specialties in our hospital, there was no guarantee that it could be available and used on a given operating day. Hence, for some operations where cell salvage was felt to be indicated by the operating team, it was not available. Such patients are included in the ‘no cell salvage’ group in the data analysis as this study set out to understand the potential benefit of cell salvage use should it be available whenever clinically required. Blood loss was calculated through the measurement of fluid volumes aspirated into suction bottles during surgery, minus wash volumes used and with consideration of fluid soaked into gauze estimated by postoperative minus preoperative gauze weight. Cumulative blood loss was recorded on charts prospectively on an hourly basis throughout the operation.

Perioperative transfusion policy

Perioperative allogenic transfusion was defined as the transfusion of blood products during surgery or within the first 5 postoperative days. We routinely utilised restrictive transfusion in the postoperative period and have a threshold of less than 80 g/l serum haemoglobin concentration before packed red cells are administered, although in higher-risk patients such as those with a history of ischaemic heart disease this threshold may be elevated.

Intraoperative cell salvage and calculation of physiological volumes

One of five cell salvage practitioners attend HPB elective operations with the sole role of perioperative patient blood management. The practitioners complete a teaching programme, including educational and competency assessment workbooks in line with UK Cell Salvage Action Group [UKCSAG] recommendations³⁸ together with practice-based education in thromboelastography (TEG) and other, near-patient testing. There is a close working relationship between the anaesthetic nurse, cell salvage practitioner, anaesthetist and surgeon. During the preoperative Team Brief, we discuss specific aspects of the case including the autologous transfusion trigger appropriate for that patient (dependent on comorbidity and preoperative plasma haematological variables) and potential coagulation issues. The patient’s measured weight is used to determine the estimated circulating volume using the Lemmens–Bernstein–Brodsky formula corrected for sex and BMI³⁹. The acceptable blood loss (blood loss tolerance) was calculated using the following formula:

$$\mathrm{Blood}-\mathrm{losstolerance}=\mathrm{ECV}\left(\left(\mathit{x}-\mathit{y}\right)/\left(\mathit{x}\right)\right),$$

where ECV is the estimated circulating volume, x is the preoperative haemoglobin concentration and y is the transfusion threshold of 80 g/l.

Process management

Blood was collected by Intraoperative Cell Salvage [(ICS), Sorin Electa and XTRA Autotransfusion systems]. The cell saver was managed on a manual setting using optimal wash quality to achieve the maximum-possible return rate. All swabs were weighed, which contributed to blood loss calculations. Blood-soaked surgical swabs were washed in a bowl containing 1000 ml normal saline with the addition of 20 000 iu heparin. This swab wash was processed together with blood suctioned from the operating field. Each litre of swab wash when processed contributed 225 ml autologous blood returned to the patient. A minimum of 1 l of swab wash was processed for each patient. This continuous ICS processing allowed blood to be returned intermittently in response to haemodynamic changes. All ICS returns were used in conjunction with leucodepletion filters as standard operating procedure where appropriate.

This multiple re-infusion approach is unusual in the UK but is a part of our Trust’s standard operating procedure document in the management of ICS. This helps to maintain haemoglobin and haematocrit levels within the tolerances agreed at the preoperative Team Brief. A continuous volume-in and volume-out record was maintained for blood loss, blood salvage recovery, swab weights, surgical irrigation and anticoagulated saline. Together with blood loss and measured haemoglobin/haematocrit levels, these data assisted the anaesthetist with overall fluid management. Thromboprophylactic subcutaneous low-molecular-weight heparin was administered 6 h postoperatively unless there were coagulation abnormalities identified by TEG, in which case an appropriate alternative thromboprophylaxis plan was implemented. Tranexamic acid was used at the discretion of the operating surgeon and anaesthetist rather than routinely.

Statistical analysis

The Strengthening the reporting of cohort, cross-sectional and case-control studies in surgery (STROCSS) criteria was followed when reporting this study⁴⁰. Patients were assigned to either ‘cell saver’ or ‘no cell saver’ groups depending on whether or not the cell salvage system was utilised intraoperatively. Patients who had access to cell salvage but did not receive autologous blood intraoperatively because of insufficient intraoperative losses were included in the cell saver group as an intention-to-treat approach. Continuous variables were compared using students t-test or Mann–Whitney U-test for normally and non-normally distributed data, respectively. Fisher’s exact test was used to compare categorical variables. The correlation between EBL and blood returned in the cell salvage group was performed using the Pearson correlation coefficient. Multivariate logistic regression was performed to assess the relationship between allogenic PRC transfusion (absolute requirement) and various explanatory variables in patients who breached their blood loss tolerance, to assess the ability of cell salvage to reduce postoperative allogenic transfusion by reinfusing autologous blood. Explanatory variables were chosen based on the rational that they could be linked to transfusion requirement on clinical grounds. Data were checked for multicollinearity with the Belsley-Kuh-Welsch technique. Heteroskedasticity and normality of residuals were assessed respectively by the Breusch–Pagan test and the Shapiro–Wilk test. Statistical analysis are performed in GraphPad Prism V.2.0 and a P-value <0.05 was considered statistically significant.

Results

Demographics

Five hundred and one patients met the inclusion criteria, of whom 264 underwent surgery with cell salvage and 237 without. The demographic features of these patients are presented in Table 1.

Table 1.

Patient demographics.

	All patients (501)	No cell saver (237)	Cell saver (264)	P
Age (years)b	62 (53–71)	64 (55–73)	60 (51–69)	0.01
Male, n (%)	255 (51)	113 (48)	142 (54)	0.2
BMIa	26±0.2	25.6±0.3	26.4±0.3	0.1
Operation (%)
Pancreaticoduodenectomy	143 (29)	135 (57)	8 (3)
+ venous reconstruction	40 (28)	34 (25)	6 (75)
+ arterial reconstruction	7 (5)	5 (4)	2 (25)
Total pancreatectomy	44 (9)	33 (14)	11 (4)
+ venous reconstruction	12 (27)	7 (21)	5 (45)
+ arterial reconstruction	2 (5)	1 (3)	1 (9)
+ islet transplant	9 (20)	5 (15)	4 (36)
Other pancreatectomy	3 (0.6)	1 (0.4)	2 (1)
Liver resection	311 (62)	68 (29)	243 (92)
Trisectionectomy	65 (21)	5 (7)	59 (24)
Hemi-hepatecomy	122 (39)	22 (32)	100 (41)
Central hepatectomy	20 (6)	6 (9)	14 (6)
ALPPS	13 (4)	3 (4)	10 (4)
Major segmental	92 (30)	32 (47)	60 (25)	<0.0001
+Biliary reconstruction	28 (6)	0 (0)	28 (11)
+portal vein reconstruction	10 (2)	0 (0)	10 (4)
Any vascular reconstruction	71 (14)	47 (20)	24 (9)	0.001
Cancer resection, n (%)	439 (88)	211 (89)	228 (86)	0.4
Preoperative haemoglobin (g/l)a	131±0.6	130±0.9	131±0.9	0.2
Total blood volume (ml)a	4686±35	4650±56	4717±45	0.3
Blood loss tolerance (ml)a	1789±23	1752±32	1822±31	0.1

Statistically significant P <0.05 values are in bold.

^amean ±standard error of the mean.

^bmedian (Inter-quartile range).

Patients undergoing surgery with cell salvage were significantly younger, and a significantly higher proportion underwent major hepatectomy as opposed to major pancreatic resection. Vascular reconstructions were also significantly more common in patients in the noncell salvage group as a result of the fact that the groups were unbalanced with regard to the operation type between groups. Other parameters of relevance to perioperative blood loss including preoperative haemoglobin concentration, total blood volume and blood loss tolerance were well balanced between the two groups.

Outcomes

Outcome data is presented in Tables 2–4. There was a significantly higher mean estimated blood loss in the patients receiving cell salvage compared to patients who underwent surgery without cell salvage. Although the number of patients experiencing a blood loss tolerance breach (blood loss greater than blood loss tolerance) was also greater in the cell salvage group, this was not significantly different to the noncell salvage group. Of patients in the cell salvage group, 224 (85%) received autologous blood transfusion intraoperatively. In these patients, an average of 32% of lost blood volume was replaced as an autologous transfusion and the volume replaced correlated closely with estimated blood loss (Fig. 1). In 52% of the cell salvage patients whose blood loss tolerance was breached, this volume was completely corrected for by re-infusion of autologous blood.

Table 2.

Intraoperative blood loss and perioperative transfusion.

	All patients (501)	No cell saver (237)	Cell saver (264)	P
Estimated blood loss (ml)a	1178±56	971±84	1360±74	0.0005
Patients receiving reinfused (autologous) blood, n (%)	n/a	n/a	224 (85)	n/a
% of blood loss returned*	n/a	0±0	32±1.0	n/a
Tolerance breached, n (%)	110 (22)	44 (19)	66 (25)	0.09
Breach corrected for blood returned, n (%)		0 (0)	30 (52)	n/a
Patients requiring allogenic transfusion
PRC	151 (30)	83 (35)	68 (26)	0.03
FFP	30 (6)	12 (5)	18 (7)	0.5
Plt	12 (2)	7 (3)	5 (2)	0.6
Total allogenic units transfused
PRC	609	365	244
FFP	190	103	87
Plt	29	19	10	n/a
Allogenic PRC units per patienta	1.2±0.1	1.5±0.3	0.92±0.2	0.03
Allogenic FFP units per patienta	0.38±0.1	0.43±0.2	0.33±0.1	0.3
Allogenic Plt units per patienta	0.06±0.02	0.08±0.03	0.05±0.02	0.4
Hospital stay (days)b	8 (6–14)	11 (7–16)	7 (5–13)	0.1
Major complication, n (%)	60 (12)	41 (17)	36 (14)	0.3
Intervention for postoperative bleeding	7 (1)	3 (1)	4 (2)	1.0
30-day mortality, n (%)	8 (2)	5 (2)	3 (1)	0.5
90-day mortality, n (%)	15 (3)	7 (3)	8 (3)	1.0

Statistically significant P <0.05 values are in bold.

^aMean±SEM.

^bmedian (IQR).

FFP, fresh-frozen plasma; Plt, platelets; PRC, packed red cells.

Table 4.

Subgroup analysis of outcomes limited to major pancreatectomy.

	All patients (190)	No cell saver (169)	Cell saver (21)	P
Estimated blood loss (ml)a	1028±95	941±95	1760±371	0.008
% of blood loss returneda	n/a	0±0	22±3.6	n/a
Tolerance breached, n (%)	44 (23.2)	34 (20.0)	10 (47.6)	0.01
Breach corrected for blood returned, n (%)		0 (0)	2 (20)	n/a
Patients requiring allogenic transfusion
PRC	63 (33.1)	53 (31.4)	10 (47.6)	0.2
FFP	10 (5.2)	8 (4.7)	2 (9.5)	0.3
Plt	5 (2.6)	4 (2.4)	1 (4.8)	0.4
Total allogenic units transfused
PRC	277	236	41
FFP	70	60	10
Plt	14	12	2	n/a
Allogenic PRC units per patienta	1.5±0.2	1.4±0.2	2.1±0.8	0.4
Allogenic FFP units per patienta	0.4±0.1	0.4±0.2	0.5±0.4	0.8
Allogenic Plt units per patienta	0.07±0.04	0.07±0.04	0.1±0.1	0.8
Hospital stay (days)b	12 (9–18)	12 (9–18)	12 (7–20)	0.8
Major complication, n (%)	21 (11.0)	17 (10.0)	4 (19.0)	0.8
Intervention for postoperative bleeding, n (%)	4 (2.1)	2 (1.1)	2 (9.5)	0.06
30-day mortality, n (%)	1 (0.5)	1 (0.6)	0 (0)	1.0
90-day mortality, n (%)	2 (1.1)	2 (1.2)	0 (0)	1.0

Statistically significant P <0.05 values are in bold.

^aMean±Standard Error of the Mean.

^bmedian (Inter-Quartile Range).

FFP, fresh-frozen plasma; Plt, platelets; PRC, packed red cells.

Figure 1.

Correlation between volume of blood lost and volume of autologous blood returned in patients receiving cell salvage.

Importantly, the absolute proportion of patients who required transfusion of allogenic PRC in the perioperative period was significantly higher in the patients who did not receive cell salvage (35 vs. 26% in the cell salvage group, P=0.03). This equates to a number needed to treat of 12; that is, 12 patients need to be managed with cell salvage to prevent one patient needing PRC transfusion. Similarly, the mean number of allogenic PRC units transfused per patient was significantly higher in the noncell-salvage group (1.5±0.3 units vs. 0.92±1.6 in the cell salvage group, P=0.03). Comparison of cell salvage and noncell salvage groups indicates that more of the noncell salvage patients required low-level transfusion (≤3 units) (Fig. 2).

Figure 2.

Histogram of transfusion requirements (number of allogenic packed red cells transfused) in patients undergoing major hepatopancreatobiliary resection.

Subgroup analysis

Given the selection bias identified when using cell salvage for pancreatic versus hepatic resections, we also analysed outcomes in these operations separately. There was a significant reduction in both the absolute number of allogenic PRC units transfused and in the average number of PRC units per person in the patients undergoing major hepatectomy with cell salvage when compared to those without cell salvage. There was a also a significant reduction in 30-day mortality in major hepatectomy patients in the cell saver group when compared with the noncell saver group (Table 3). When considering major pancreatectomy alone, there was no difference in rate or number of transfused PRC units when comparing patients receiving cell salvage, with those who did not receive it (Table 4).

Table 3.

Subgroup analysis of outcomes limited to major hepatectomy.

	All patients (311)	No cell saver (68)	Cell saver (243)	P
Estimated blood loss (ml)a	1270±69	1054±172	1327±75	0.1
% of blood loss returneda	n/a	0±0	33±1.0	n/a
Tolerance breached, n (%)	58 (18.6)	10 (14.7)	48 (19.5)	0.4
Breach corrected for blood returned, n (%)		0 (0)	28 (58.3)	n/a
Patients requiring allogenic transfusion
PRC	87 (27.8)	29 (42.6)	58 (23.9)	0.004
FFP	19 (6.1)	4 (5.9)	15 (6.2)	1.0
Plt	8 (2.6)	3 (4.4)	5 (2.1)	0.4
Total allogenic units transfused
PRC	332	129	203
FFP	120	43	77
Plt	17	7	10	n/a
Allogenic PRC units per patienta	1.1±0.18	1.9±0.61	0.8±0.16	0.01
Allogenic FFP units per patienta	0.4±0.1	0.6±0.4	0.3±0.1	0.2
Allogenic Plt units per patienta	0.05±0.02	0.1±0.06	0.04±0.02	0.2
Hospital stay (days)b	6 (5–11)	6 (5–9)	7 (5–13)	0.6
Major complication, n (%)	42 (13.5)	10 (14.7)	32 (13.2)	0.7
Intervention for postoperative bleeding, n (%)	4 (1)	2 (3)	2 (1)	0.2
30-day mortality, n (%)	7 (2)	4 (6)	3 (1)	0.04
90-day mortality, n (%)	13 (4)	5 (7)	8 (3)	0.2

Statistically significant P <0.05 values are in bold.

^aMean±SEM.

^bmedian (IQR).

FFP, fresh-frozen plasma; Plt, platelets; PRC, packed red cells.

Multivariate analysis

Finally, a multivariate analysis was performed to identify factors independently associated with the absolute need for PRC transfusion, limiting the analysis to patients who breached their blood loss tolerance (Table 5). This demonstrated that correction of blood loss tolerance breach is independently associated with avoidance of allogenic PRC transfusion.

Table 5.

Multivariate analysis of factors predicting the need for allogenic packed red cell transfusion in patients who breached their blood loss tolerance.

	Multivariate OR (95% CI)	P
Operation (pancreas versus liver)	1.19 [0.24–6.22]	0.83
Venous reconstruction	0.57 [0.02–6.76]	0.66
Operative blood loss	1.00 [0.99–1.00]	0.82
Total blood volume	1.00 [0.99–1.00]	0.99
Cell salvage use	0.60 [0.05–6.42]	0.66
Correction of blood loss tolerance breach	0.05 [0.006–0.38]	0.005

Stactistically significant P <0.5 values are in bold.

OR, odds ratio.

Discussion

In this retrospective, single-institution study of patients undergoing HPB operations at risk of major haemorrhage, we have demonstrated significant benefit in the use of cell salvage for reducing perioperative allogenic blood transfusion in addition to a potential benefit in reducing postoperative mortality in major hepatectomy patients. Cell salvage appears particularly beneficial at preventing the need for low-level (<3 units PRC) allogenic transfusion, and the benefit was primarily attributable to the replacement of blood volume and thereby, the correction of the blood loss tolerance breach. This appears to be especially the case in major hepatectomy, although the number of pancreatectomy patients receiving cell salvage were small. Whilst there are other potential benefits of cell salvage use, including prevention of coagulopathy through the provision of warmed autologous blood and serial TEG monitoring, better maintenance of physiological oxygen delivery and more nuanced benefits that come through the process of integrating the cell salvage practitioners within the surgical team, we were not able to measure these benefits due to the retrospective nature of our study, although each may have contributed to reduced mortality in hepatectomy patients.

The prevention of low-level transfusion is likely to be particularly important at a population level, when it is considered that we identified an average reduction of just over 0.5 units of PRC per patient, equivalent to a total saving of greater than 100 units in the entire cell salvage population when compared to the noncell salvage group. This is likely to result from our relatively unique practice of performing multiple re-infusion and ultimately, a close attention to intraoperative patient blood management through the engagement of specialist cell salvage practitioners. Were this to be replicated in every major HPB centre through routine cell salvage use, it would have significant implications for the reduction of blood product usage. Given that blood products are a scare resource, our findings support the practice of more routine cell salvage usage for major HPB surgery.

Through subgroup analysis of major hepatectomy and pancreatectomy, we identified subtle differences in the outcomes of these patient groups when the use of cell salvage was considered. It is difficult to reach strong conclusions due to the relatively small population of patients in whom cell salvage was used in the pancreatectomy group. However, the data indicates that cell salvage is primarily effective when blood loss tolerance breaches are corrected for. In patients undergoing pancreatectomy, the rate of correction was only 20% compared with 58% in those undergoing major hepatectomy which may be why the use of cell salvage was not associated with a reduction in allogenic transfusion rates in pancreatectomy patients; a finding supported by the multivariate analysis, where correction of breach was more important than the type of operation for preventing allogenic transfusion. Whilst our analysis is not able to investigate why cell salvage may be less effective at correcting blood loss tolerance breaches under certain conditions, such information would be useful for planning better utilisation of cell salvage.

Whilst it may appear obvious that autologous blood returned intraoperatively can reduce the need for perioperative transfusion in those undergoing major HPB surgery, this is one of the first and certainly the largest study to definitively document the benefit of cell salvage use in HPB surgery outside of the field of liver transplantation. By quantifying the degree to which autologous transfusion is prevented through cell salvage use, we further highlight the benefit of intraoperative cell salvage in this patient cohort and present data that can be used to power subsequent randomised trials. Equally, we are only able to conclude that correction of blood loss through autologous return is responsible at least in part for reducing perioperative transfusion, through the calculation of blood volume and blood loss tolerance for every patient in the study. This novel strategy indicates that the calculation of such parameters could be considered a mandatory step in the perioperative blood management of patients undergoing major liver resection or pancreatectomy as it supports the practice of the three pillars approach. Indeed, routinely calculating the blood loss threshold pre-operatively and noting it at the WHO checklist could help identify patients in whom cell salvage is particularly beneficial, aid goal-directed blood management and increase awareness amongst the whole surgical team of blood loss and the importance of its minimisation.

Of course, it is possible that the differences we identify in allogenic transfusion practice are not related to autologous transfusion or to the use of cell salvage. A significant potential weakness in our analysis is the selection bias in our approach to cell salvage, with a far greater number of major liver cases selected for cell salvage than for major pancreatectomy. We chose to include pancreatectomy patients rather than limiting our analysis to liver resection, as we concluded that both sets of patients are at risk of major intraoperative haemorrhage and therefore both stand to benefit from cell salvage use. Whilst an alternative approach would have been to limit our analysis to patients undergoing hepatectomy, this would likely also have introduced selection bias, as we primarily reserve cell salvage for major hepatectomy. Equally, use of a historical hepatectomy cohort would have introduced further bias as cell salvage was introduced in our unit at a similar time to enhanced recovery, intraoperative central venous pressure monitoring and preoperative cardiopulmonary exercise testing; all factors which have the potential to impact upon intraoperative blood loss.

Our selection methods for the use of cell salvage were clearly justified, because we experienced higher blood loss in the cell salvage group. But it should also be noted that the noncell-salvage cohort experienced high levels of intraoperative blood loss too, making them a good control for the cell salvage population when attempting to understand how autologous transfusion in patients suffering major blood loss may be linked to perioperative allogenic transfusion. However, it is possible that patients recovering from pancreatic surgery are inherently more likely to require allogenic transfusion in the postoperative period for reasons other than intraoperative blood loss.

Although there is historic concern regarding tumour cell dissemination from autologous blood transfusion⁴¹, there is a paucity of evidence to support this^42,43. We routinely utilise leucocyte depleting filters, which have been demonstrated to filter tumour cells in in-vitro models⁴⁴. Leucocyte filters are also effective at removing tumour cells in patients undergoing gynae-oncological cancer surgery⁴⁵ and in patients undergoing sarcoma surgery⁴⁶. Importantly, there is an association between metastasis or recurrence in patients receiving leucocyte-deplete red cells⁴⁷. Several studies have demonstrated a reduction in cancer recurrence and metastasis in patients receiving cell salvage⁴², potentially because allogenic transfusion is associated with impaired cancer immunity and infusion requirements are reduced through cell salvage usage. In summary, we routinely use cell salvage in cancer cases based on current literature demonstrating no clear adverse effect on oncological outcome and recognise that more work is needed to understand the potentially for improving oncological outcome through the use of cell salvage.

We chose a 5-day postoperative period in which to record allogenic transfusion events (and no longer), specifically with this issue in mind. This period should capture transfusion events in patients who are initially hemoconcentrated as a result of permissive hypotension to reduce intraoperative blood loss and whose serum haemoglobin concentration subsequently falls below the transfusion threshold with the filling of crystalloid in the early postoperative period. Equally, whilst it is recognised that a significant proportion of patients who undergo pancreatic surgery may require transfusion because of sepsis and malnutrition following an anastomotic leak, this is unlikely to have declared itself in the first 5 days following surgery. It should also be considered that the type of surgery (pancreatectomy vs. hepatectomy) was not independently predictive of transfusion requirement, whereas correction of blood loss through the use of cell salvage was, further supporting the positive role for this practice with regards mitigating allogenic transfusion. Nonetheless, it is recognised that the retrospective approach and selection bias remain the major limiting factors in our study.

Our sample size was large, and the lack of information bias, considering that we limited inclusion to those patients with complete data collection, ensure that our data is robust. Our blood loss volumes are consistent with those reported by others considering the significance of the surgery undertaken^1,48,49. Equally, the patient groups were well matched with respect to factors associated with the requirement for transfusion inclusive of circulating blood volume, blood loss tolerance, preoperative haemoglobin concentration and malignant indication. Importantly, the recruitment period covers a time where prehabilitation and Enhanced Recovery After Surgery were routinely used in our centre. We believe that this has contributed to our relatively low 30 and 90-day mortality rates, as well as major complication rates, which are comparable with other large series, despite the inclusion of many patients who underwent major venous reconstruction^50–52. It is also notable that our patient cohort includes those undergoing a wide variety of HPB procedures. This real-world view of our practice is in contrast to a trial setting, which, given sample size constraints, is likely to be capable of analysing the effect of cell salvage in a far narrower clinical setting.

We recognise that there are considerable challenges involved in establishing the high level of cell salvage infrastructure that we present here, particularly within lower economic settings. Expensive equipment, highly trained cell salvage practitioners, anaesthetists and surgeons working as a team to deliver this capability are a valuable resource that our surgical service has strategically invested in. This investment has led to an almost unique service within our institution, where intraoperative cell salvage is managed by practitioners with specialist expertise and training, whose sole responsibility is patient blood management. This is at odds with most centres throughout the UK, where management of cell salvage is undertaken by a member of the scrub nursing team. However, once a cell salvage system is established, the costs per case ranges from £70–£190, depending on caseload, with a higher caseload per annum proving more fiscally rewarding⁵³. Data from obstetric surgery indicates a cost of $34 881 per quality-adjusted life-year gained when cell salvage is selectively used for high-risk cases. However, this cost climbs steeply if cell salvage were to be used routinely⁵⁴ indicating the potential value of careful case selection in HPB surgery. Whilst we select our higher-risk patients for cell salvage use, our data indicate that we can improve our ability to identify patients who are more likely to benefit from cell salvage use, and in particular, more work is required to identify, which patients undergoing pancreatectomy are most benefited.

Conclusion

This is the largest study of cell salvage use in the HPB surgery field. Our findings show that the use of cell salvage reduces the absolute number of patients needing allogenic transfusion, as well as the number of units transfused per patient. Correction of blood loss tolerance breach is independently associated with avoidance of allogenic transfusion, and use of cell salvage is associated with a significant reduction in 30-day mortality in hepatectomy patients. Further work is required to understand whether the cell salvage should be routinely used for major liver resection, considering economic factors in addition to effects on perioperative transfusion burden.

Ethical approval

Internal ethics approval sought and obtained via Trust Ulysses Clinical Improvements Division of Oxford University Hospitals NHS Foundation Trust. Reference number 7809 Ver 1 Type 03 Approved Local Audit.

Sources of funding

This study/project is funded by the NIHR BRC 4 SITE - NIHR203311. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Carea.

Author contribution

A.S.L.: first author, study design, data collections, data analysis, writing; A.G.W.: lead investigator, study design, data collections, data analysis, writing; R.C.: study design, data collections, data analysis, writing; remaining authors: study design, data analysis, writing.

Conflicts of interest disclosure

The authors declare that they have no financial conflicts of interest with regard to the content of this report.

Guarantor

MR Alex Gordon-Weeks, Consultant HPB surgeon and lead investigator for the study Department of Hepatobiliary Surgery, Churchill Hospital Oxford United Kingdom. E-mail: alex.gordonweeks@gmail.com.

Data availability statement

The full anonymised raw data set is available on request in writing to the lead investigator for the study, Mr Alex Gordon-Weeks (consultant HPB surgeon).

Provenance and peer review

Not commissioned, externally peer-reviewed.