Randomized clinical trial of preoperative intravenous iron sucrose to reduce blood transfusion in anaemic patients after colorectal cancer surgery

Abstract

Background:

The transfusion rate following colorectal cancer resection is between 10 and 30 per cent. Receipt of allogeneic blood is not without risk or cost. A preoperative adjunct that reduced the need for transfusion would mitigate these risks. This study was designed to determine whether iron sucrose reduces the likelihood of postoperative blood transfusion in patients undergoing elective colorectal cancer resection.

Methods:

In this randomized prospective blinded placebo-controlled trial of patients undergoing resectional surgery with a preoperative diagnosis of colorectal cancer, 600 mg iron sucrose or placebo was given intravenously in two divided doses, at least 24 h apart, 14 days before surgery. The primary outcome measures were serum haemoglobin concentration, recorded at recruitment, immediately before surgery and at discharge, and perioperative blood transfusions.

Results:

No difference was demonstrated between treatment groups (iron sucrose, 34 patients; placebo, 26) for any of the primary outcome measures, for either the whole study population or a subgroup of anaemic patients.

Conclusion:

This pilot study provided no support for the use of intravenous iron sucrose as a preoperative adjunct to increase preoperative haemoglobin levels and thereby reduce the likelihood of allogeneic blood transfusion for patients undergoing resectional surgery for colorectal cancer. Registration number: 2005-003608-13UK (Medicines and Healthcare products Regulatory Agency). Copyright © 2009 British Journal of Surgery Society Ltd. Published by John Wiley & Sons, Ltd.

Introduction

Patients with colorectal cancer commonly demonstrate biochemical iron deficiency with or without anaemia at presentation1. The majority proceed to surgery. Colorectal cancer resections are designated as complex major procedures and resulting blood loss may be significant. In the context of a prescribed transfusion protocol, there are three independent risk factors for requiring allogeneic blood as a result of surgery: age, intraoperative blood loss and preoperative serum haemoglobin (Hb) concentration2. With colorectal cancer having a peak incidence in the seventh decade of life and with patients commonly suffering co-morbid iron deficiency and anaemia, it is not surprising that resection is associated with a perioperative allogeneic blood transfusion (ABT) rate of between 10 and 38 per cent3.

ABT is not without risk, despite advances in transfusion medicine. Alloimmunization, incompatibility reactions and transmission of viral load are well recognized4. The potential for transmission of prion disease is unknown5. After resection of colorectal cancer there is an association with an increased infective complication rate6, and after resection there may be increased disease recurrence7 and a reduced 5-year survival rate8. In addition, ABT confers a significant cost and is an increasingly pressured resource.

Any adjunct that effects a reduction in the perioperative ABT rate will mitigate these risks. Of the three principal preoperative predictors for requiring ABT, anaemia is the only practical target for intervention. A number of adjunctive treatments have been tested. Recombinant erythropoietin (rEPO) has been administered before colorectal cancer9, gynaecological10 and orthopaedic11 surgery. The results were equivocal. The efficacy of rEPO is reduced by coexisting iron deficiency. Acute normovolaemic haemodilution and autologous blood donation techniques were found to have limited application in this population owing to pre-existing iron deficiency and anaemia12, 13. Oral iron supplements alone before colorectal cancer resection have been shown to reduce preoperative anaemia and the need for ABT in both anaemic14 and non-anaemic15 patients. Intravenous preoperative iron has also been used. The iron dextran form, although demonstrating efficacy, resulted in an alarming rate of anaphylaxis, thereby precluding its use16. More recently, observational studies have shown a iron sucrose complex to reduce ABT and to correct anaemia following arthroplasty17, 18 and hysterectomy19.

By conducting a prospective randomized controlled study, the present authors set out to verify this observation in the context of colorectal cancer resection. The aim of this pilot study was to test the hypothesis that preoperative intravenous iron sucrose increases the preoperative serum Hb concentration from baseline levels and reduces the rate of blood transfusion compared with placebo in patients undergoing resectional surgery for suspected colorectal cancer.

Methods

A prospective randomized blinded placebo-controlled trial was undertaken with approval from Cornwall and Plymouth Research Ethics Committee. The trial was sponsored by the in-hospital research and development unit and registered with the UK Medicines and Healthcare products Regulatory Agency (registration number: 2005-003 608-13).

The study was based at a single centre, Derriford Hospital, Plymouth, which is part of the Peninsula Medical School in the south-west of the UK. Volunteers were recruited from all patients scheduled to undergo bowel resection for suspected colorectal cancer at the centre during the study period. Patients were excluded if they were under 18 years, were on existing (or had been taking) oral iron supplementation within 6 weeks of the day they were approached, if they had received a blood transfusion within that same period, or if the date of their scheduled surgery fell within 15 days of the date of recruitment.

Participants were allocated to either the treatment (iron) group or a placebo group, based on a computer-generated randomization sequence provided by the Research and Development Support Unit. To ensure equal numbers of anaemic patients in each treatment group, randomization was stratified according to prerecruitment Hb status: normal (Hb level at least 13·5 g/dl in males and 12·5 g/dl in females), anaemic, or unknown (no test within 2 months of recruitment). Block randomization was used to ensure similar numbers in each group for each subset.

Allocation codes were sealed in sequentially numbered opaque envelopes which were secured within a locked store room in a dedicated research unit (this was remote from the clinical areas of the hospital where participants were to undergo outpatient, ward and operative treatment). Only after recruitment was an envelope opened by the investigator administering the infusion, following the inscribed strict numerical order and for the relevant subset appropriate to the Hb status of the participant.

Treatments

Participants were randomized to receive two infusions of either 250 ml intravenous placebo (0·9 per cent saline) or 300 mg iron sucrose made up to 250 ml with 0·9 per cent saline. This was provided as 15 ml ferric hydroxide with sucrose in a 2 per cent solution (Venofer^®; Syner-Med, Purley, UK). Thus, patients in the treatment group received a total dose of 600 mg iron sucrose. The minimum time between each infusion was 24 h and all infusions were completed within a minimum of 14 days before undergoing elective surgery. A test dose of 5 ml was given while monitoring pulse oximetry and blood pressure. If the patient remained well, the remainder was infused over 1 h. Subsequent elective surgical treatment, including neoadjuvant therapy, proceeded unchanged. Notably, allogeneic blood was to be transfused only in accordance with a strict protocol, similar to that described previously15.

Although the investigator administering the infusion was not blinded to the treatment group, this was concealed from the patient by using an opaque sheath to cover the drug-giving set. The chief investigator and clinicians involved in perioperative care also remained blinded to the treatment group for the duration of the trial.

Outcome measures

The primary outcome measure was change in Hb concentration between recruitment and day of admission. Secondary outcomes were transfusion rate, changes in serum iron markers over the same time period, length of hospital stay and adverse perioperative events.

Blood samples were obtained for Hb, haematocrit, mean cell Hb (MCH), reticulocytes, serum iron, ferritin and transferrin saturation at recruitment, before surgery, 1 day after surgery, and on discharge or at day 7 (whichever was earlier). Potential confounding variables, including sex, age, body mass index, overt (or occult) blood loss as a presenting feature, risk factors for occult blood loss (including anticoagulants, aspirin and non-steroidal anti-inflammatory drug treatment), and estimated operative blood loss, were recorded.

Statistical analysis

The study was powered at 80 per cent to detect a difference in the mean change in serum Hb concentration between recruitment and treatment of 0·5 g/dl in anaemic patients. Previous work by this team identified the standard deviation of this difference in Hb to be 0·5. At the 5 per cent significance level (two sided), the number needed in each group was ten. With the prevalence of anaemia within this patient group having previously been identified as 38 per cent, to ensure that at least 20 anaemic patients were recruited, the sample size was set at 60 (n = 30 in each group).

Statistical analyses were performed using SPSS^® version 13·0 (SPSS, Chicago, Illinois, USA). Data analysis was carried out according to a pre-established plan, on an intention-to-treat basis. Normality was assessed using the Kolmogorov–Smirnov test with the Lilliefors correction and bar histograms. Parametric data were analysed with Student's t test and presented as the mean with 95 per cent confidence intervals (c.i.). Non-parametric data were assessed using the Mann–Whitney or Wilcoxon signed rank test as appropriate. The χ² test was used for proportions, and the Fisher correction for small groups. A subgroup analysis for anaemic patients was built into the protocol. No interim analysis was conducted.

Results

Between May 2006 and August 2008, 62 participants were recruited, of whom 11 were known to be anaemic at recruitment, 22 to have a normal Hb level and 29 to have no recent record of anaemia status. Two patients were subsequently found to be unsuitable for surgery and were not included in the analysis (Fig.1). Thirty-four patients received iron and 26 placebo, with a median time interval between infusion and surgery of 17 (range 11–32) days (in two patients surgery had to be performed before the prescribed 14 days owing to constraints of the cancer waiting targets). One patient from each group failed to attend for the second infusion.

Figure 1. CONSORT diagram for the trial. The analysis was performed on an intention-to-treat basis

At baseline there was good balance between the confounding variables in the two groups (Table1). Two patients in the iron group experienced transient symptomatic hypotension immediately after the infusion which did not require treatment.

Table 1. Confounding variables

	Iron (n = 34)	Placebo (n = 26)
NSAID, non-steroidal anti-inflammatory drug.
Median age (years)	67	70
Sex ratio (M : F)	22 : 12	17 : 9
Body mass index (kg/m^2)	26·6	26·3
NSAID, antiplatelet, anticoagulant use	13	9
History of overt or occult bleeding	17	11

There was no significant difference in the primary outcome measures between groups (Table2). The mean change in Hb between recruitment and before surgery was −0·19 g/dl following iron therapy and −0·50 g/dl after placebo (P = 0·355). Five patients in each group required a perioperative transfusion. There was no significant difference between the groups in other secondary outcome measures (Hb, haematocrit, MCH, ferritin, serum iron or transferrin saturations) at recruitment, before surgery, on day 1 after surgery, and at discharge (Table3). Neither was there any significant difference in the mean change between recruitment and surgery for ferritin, serum iron or transferrin saturation for iron versus placebo (Table4).

Table 2. Primary outcome variables for the whole study group

	Iron	Placebo	P
Values in parentheses are * range and † 95 per cent confidence interval. ‡ Mann–Whitney U test (two tailed); § Student's t test (two tailed).
Median haemoglobin (g/dl)*
At recruitment	13·4 (10·8–15·9)	13·7 (9·2–16·8)	0·526‡
Before surgery	13·9 (9·9–16·0)	13·8 (9·2–16·9)	0·957‡
Mean change between recruitment and surgery†	−0·19 (−0·74, 0·36)	−0·50 (−1·00, −0·01)	0·355§

Table 3. Serum iron markers in the two treatment groups at the four observation points

	Haemoglobin (g/dl)	Hct	MCH (pg/cell)	Ferritin (ng/ml)	Serum iron (µmol/l)	Transferrin saturation (%)
Values are mean (95 per cent confidence interval). Hct, haematocrit; MCH, mean cell haemoglobin; ND, not done.
Recruitment
Placebo	13·7 (12·7, 14·7)	0·41	29·5	100·8 (41·1, 160·7)	14·9 (11·6, 18·1)	19·9 (12·7, 27·1)
Iron	13·4 (12·6, 14·1)	0·42	28·6	70·8 (31·6, 110·1)	14·4 (9·5, 19·4)	20·6 (15·7, 25·4)
Before surgery
Placebo	13·6 (12·6, 14·6)	0·41	29·6	172·3 (75·3, 269·3)	16·6 (12·8, 20·4)	21·6 (15·5, 27·7)
Iron	13·4 (12·5, 14·3)	0·40	28·9	226·7 (166·5, 286·9)	17·0 (13·7, 20·3)	23·1 (17·1, 29·1)
Postop. day 1
Placebo	11·3 (10·7, 12·0)	0·35	29·5	ND	ND	ND
Iron	11·2 (10·4, 12·0)	0·35	28·7	ND	ND	ND
Discharge
Placebo	11·7 (10·9, 12·5)	0·35	29·7	302·1 (163·1, 441·1)	7·4 (4·7, 10·2)	12·7 (8·7, 16·8)
Iron	11·2 (10·4, 12·1)	0·35	28·7	286·9 (184·8, 389·1)	5·2 (2·9, 7·4)	10·3 (6·7, 14·0)

Table 4. Intra-subject changes in haemoglobin, ferritin, serum iron and transferrin saturation between recruitment and preoperative evaluation

	Change from recruitment to preop. value		P*
	Iron	Placebo
Values are mean (95 per cent confidence interval). * Student's t test (two tailed).
Haemoglobin (g/dl)	−0·19 (−0·74, 0·36)	−0·50 (−1·00, −0·01)	0·355
Ferritin (ng/ml)	106 (42, 171)	142 (31, 254)	0·581
Serum iron (µmol/l)	−0·5 (−7·3, 6·3)	1·6 (−1·5, 4·7)	0·542
Transferrin saturation (%)	0·5 (−9·4, 10·3)	2·9 (−4·4, 10·2)	0·470

Median (range) intraoperative blood loss was 414 (125–1500) ml following iron therapy and 300 (100–800) ml for placebo (P = 0·683), and median length of hospital stay was 10 versus 8 days respectively (P = 0·273).

Primary outcome measures for the subgroup of patients who were anaemic on recruitment are summarized in Table5, and secondary outcome measures in Table6. There were 18 anaemic patients in total, nine in each group. There was no significant difference between the groups in mean Hb, haematocrit, MCH, ferritin, serum iron or transferrin saturation at any time point. In both groups there was an increase in ferritin concentration and transferrin saturation between recruitment and the preoperative value, which was greatest in the iron group: 109·3 versus 85·3 ng/ml and 4·9 versus 3·2 per cent for the mean change per individual in ferritin and transferrin respectively.

Table 5. Primary outcome variables in the anaemic subgroup

	Iron	Placebo	P
Values in parentheses are * interquartile range and † 95 per cent confidence interval. ‡ Mann–Whitney U test (two tailed); § Student's t test (two tailed).
Median haemoglobin (g/dl)*
Recruitment	11·8 (2·0)	12·4 (2·0)	0·627‡
Before surgery	11·2 (3·0)	12·5 (4·0)	0·427‡
Mean change from recruitment to preop. value (g/dl)†	−0·46 (−0·79, −0·12)	−0·11 (−0·63, 0·41)	0·223§

Table 6. Serum iron markers for anaemic patients in the two treatment groups at the four observation points

	Haemoglobin (g/dl)	Hct	MCH (pg/cell)	Ferritin (ng/ml)	Serum iron (µmol/l)	Transferrin saturation (%)
Values are mean (95 per cent confidence interval). Hct, haematocrit; MCH, mean cell haemoglobin; ND, not done.
Recruitment
Placebo	11·8 (10·2, 13·5)	0·37	26·7	51·5	11·0 (5·7, 16·4)	14·0 (5·4, 22·6)
Iron	11·7 (10·6, 12·7)	0·38	25·8	100·5	8·6 (1·7, 15·6)	11·0 (1·5, 20·5)
Before surgery
Placebo	11·9 (9·9, 14)	0·37	27·3	153·5	13·0 (5·1, 20·9)	18·5 (5·7, 31·2)
Iron	11·2 (9·7, 12·6)	0·35	26·7	200·3	13·0 (5·9, 20·1)	19·6 (8·3, 30·9)
Postop. day 1
Placebo	10·7 (9·8, 11·6)	0·37	27·5	ND	ND	ND
Iron	9·6 (8·9, 10·4)	0·35	25·5	ND	ND	ND
Discharge
Placebo	11·3 (10·4, 12·2)	0·35	28·1	259·5	5·6 (2·2, 9·1)	10·0 (5·2, 14·8)
Iron	10·2 (8·4, 12·1)	0·32	26·1	179·0	4·3 (2·7, 5·8)	7·9 (6·2, 9·6)

Five patients receiving placebo required transfusion, compared with two receiving iron (P = 0·335, Fisher's exact test). The median number of units transfused was 0 (interquartile range (i.q.r.) 1) for iron versus 2 (i.q.r. 3) units for placebo.

Discussion

The intravenous administration of 600 mg iron sucrose at least 2 weeks before resectional surgery for colorectal cancer conferred no benefit in terms of increasing preoperative Hb levels, reducing the rate of ABT, or reducing length of stay in comparison with placebo. In addition, there was little evidence of increased total body iron or iron availability. No benefit was seen either for all patients awaiting surgery or for the subgroup with anaemia.

As the only randomized controlled trial (RCT) to investigate the efficacy of intravenous iron sucrose in the preoperative setting to date, these data must be compared with those previously reported in three observational studies. Those results contradict the present ones: intravenous iron sucrose given 3–4 weeks before arthroplasty17, 20 and hysterectomy19 resulted in significant improvements in Hb and iron stores, and reduced ABT rates compared with selected controls.

Comparisons may also be inferred from the use of intravenous iron sucrose in perioperative and postoperative contexts. Used perioperatively, observational data indicate similar increases in Hb levels and reductions in ABT rates for orthopaedic patients17, 21 as seen following preoperative use. When given in the postoperative setting, the existing data are contradictory. Although observational data derived from treatment following arthroplasty22 support the benefits previously described, observational data following cardiac surgery23 alongside two RCTs, following cardiac24 and spinal25 surgery, failed to demonstrate any increase in Hb levels after treatment compared with placebo.

With these contradictory findings, the true efficacy of intravenous iron sucrose is difficult to reconcile. When only high levels of evidence26 are considered, there is no evidence to support the efficacy of intravenous iron sucrose in any of these contexts.

For patients with colorectal cancer, unlike those awaiting arthroplasty, gynaecological or cardiac surgery, iron deficiency, anaemia and likely ongoing blood loss are an integral part of the underlying pathology. This blood loss, combined with the metabolic changes associated with malignant disease27, may mask any benefit derived from parenteral iron. The trend towards a benefit in terms of reduced ABT rates in the anaemic subgroup that received treatment (despite no demonstrable improvement in Hb or iron stores) may reflect both a true benefit of iron sucrose and the fallibility of serum iron, ferritin and transferrin saturations when used as estimates of body iron stores in the context of malignancy and/or an acute-phase response28.

By excluding patients taking oral iron and those who had recently received an ABT, patients most likely to benefit from the treatment (those with profound anaemia) were systematically excluded. At recruitment, median Hb levels for patients with anaemia were 11·8 g/dl in the iron subgroup and 12·4 g/dl in the placebo subgroup, that is, these patients were only just anaemic. This conferred no bias in itself, but it did remove those physiologically primed to utilize the parenteral iron and therefore more likely to demonstrate a benefit, if one existed. Equally, it left the anaemic subgroups one short of the target recruitment, increasing the probability of a type II error.

The decision to use 600 mg iron sucrose, which is at the lower end of that used in clinical practice for the treatment of anaemic patients, as the treatment dose was taken following guidance from the manufacturer. However, there is evidence that oral iron supplements equivalent to this dose are effective15, so it was reasonable to assume that, with the significantly enhanced bioavailability resulting from parenteral administration, this dose of intravenous iron sucrose should have been sufficient.

It may be argued that giving the infusions a median of 17 days before surgery may have resulted in initial benefits of therapy being missed, as they diminish over time. Previous observational data have demonstrated that the effects seen persist for more than 3 weeks18.

In conclusion, the results of this RCT did not show any improvement in Hb levels or reduction in transfusion rates following the preoperative intravenous administration of 600 mg iron sucrose in all patients, or in a subgroup of anaemic patients, undergoing resection of colorectal cancer. It is possible that this treatment may reduce the likelihood of ABT in anaemic patients, but the trial was not sufficiently powered to detect this. The use of a higher dose of iron sucrose in anaemic patients, as dictated by their Hb level and the manufacturer's treatment algorithm, may augment this benefit. Use of newer intravenous iron preparations that can be given as a single intravenous injection rather than repeated infusions may improve trial recruitment and ultimately simplify clinical use.

Acknowledgements

The authors thank Syner-Med Pharmaceutical Products Limited for providing Venofer^® and for funding the blood tests, and the Department of Colorectal Surgery, Derriford Hospital, for allowing patients to participate in the trial. The authors declare no conflict of interest.