Efficacy of tranexamic acid in paediatric cardiac surgery: a systematic review and meta-analysis^†

Abstract

The benefit-to-risk ratio of using tranexamic acid (TXA) in paediatric cardiac surgery has not yet been determined. This systematic review evaluated studies that compared TXA to placebo in children undergoing cardiac surgery. A systematic search was conducted in all relevant randomized controlled trials. The following information was extracted from the studies and analysed if relevant: demographic data, TXA dose and regimen of administration, cardiopulmonary bypass time, blood loss and blood product transfusion at 24 h. From the studies screened, only 8 (848 patients) were included in the analysis. Most data were heterogeneously distributed and could not be analysed. Further, transfusion policies were not well defined for each study. TXA reduced the need for red blood cell transfusion by 6.4 ml kg⁻¹ day⁻¹ (I² = 0%, P = 0.45), platelet transfusion by 3.7 ml kg⁻¹ day⁻¹ (I² = 0%, P = 0.46) and fresh frozen plasma transfusion by 5.4 ml kg⁻¹ day⁻¹ (I² = 0%, P = 0.53). The number of children who avoided all blood product transfusions was not reported in most of the studies. Evaluation of the side effects associated with TXA use and the effects of the agent on postoperative morbidity and mortality was not possible from the data. There was marked variability in the dosage and infusion schemes used in different studies. This systematic review showed that in paediatric cardiac surgery, the benefit-to-risk ratio associated with the use of TXA cannot be adequately defined. Evidence supporting the routine use of TXA in paediatric cardiac surgery remains weak.

Antifibrinolytic, Tranexamic acid, Paediatric, Blood loss reduction, Cardiac surgery

INTRODUCTION

The paediatric cardiac surgical population is at high risk of bleeding and transfusion. Bleeding may adversely affect outcomes as it is associated with haemodynamic instability, prolonged surgical time, risk of reoperation and increased need for allogeneic transfusion [1]. In infants undergoing congenital heart surgery, administration of intra- and postoperative blood products is an important risk factor for longer duration of mechanical ventilation [2] and increased length of hospitalization [3].

Haemodilution at the initiation of cardiopulmonary bypass (CPB) is more pronounced in children than in adults [4] and can be associated with impaired haemostasis [5]. Fibrinolysis is enhanced by CPB, probably through activation of the kallikrein pathway [6]. Antifibrinolytic agents are therefore used routinely in paediatric surgical patients in an attempt to decrease bleeding and, consequently, the need for transfusion.

In 2008, the BART (Blood Conservation using Antifibrinolytics in a Randomized Trial) study showed that in comparison to tranexamic acid (TXA) and εaminocaproic acid (EACA), the use of aprotinin was associated with an increased postoperative mortality [7]. As a result, aprotinin was withdrawn from the market, leaving TXA as the sole alternative in Europe. However, the risks, benefits and most appropriate way of administering TXA in the paediatric population remain unclear. The available data show large variability in patient population, type of procedure and administration schemes [8], precluding a reliable comparison of the different studies. We therefore performed a systematic review with meta-analysis to evaluate the available literature on the efficacy of TXA compared with placebo in reducing blood loss and allogeneic blood transfusion exposure in children undergoing cardiac surgery.

MATERIALS AND METHODS

This systematic review of randomized trials was performed in accordance with the Quality of Reporting of Meta-analyses (QUOROM) consensus [9]. We performed a systematic search of the electronic databases PubMed and EMBASE using the following keywords: TXA, antifibrinolytic agents, cardiac surgery, paediatrics, children, infants, cardiovascular procedures, CPB and bleeding reduction. The Boolean operator ‘AND’ was inserted between the keywords. The following inclusion criteria were chosen: children under 18 years old, cardiac surgery, administration of TXA, randomized controlled trial and English language publications. We excluded meta-analyses, review articles and studies that used other antifibrinolytic therapy. After screening the titles, two authors (David Faraoni and Philippe Van der Linden) examined abstracts and articles, and articles that met inclusion criteria were selected (Fig. 1).

Figure 1:

A flow chart of the meta-analysis study selection process diagram. Number of studies found. The number of studies excluded at each stage and the reason are listed in the corresponding box.

The following data were extracted from the studies for analysis: number of patients, age, weight, type of surgery, TXA dose and regimen of administration, CPB time, blood loss, transfusion of packed red blood cells (RBCs), platelets (PLT) and fresh frozen plasma (FFP) at 24 h. Postoperative adverse effects, mechanical ventilation duration, length of stay in the intensive care unit and mortality rate were also assessed.

When relevant data were available, statistical analysis was performed using Review Manager 5 software (REVMAN 5; The Cochrane Collaboration, Oxford, UK). When original data were expressed as continuous variables, fixed and random effects models were used to compute the mean difference. All outcomes were noted as differences compared with the placebo. Heterogeneity was assessed using I²-statistics. According to the Cochrane review guidelines, I² > 40% and P < 0.01 were considered threshold values for heterogeneity, indicating the presence of a random effects mean difference computation. In studies with more than one dosage scheme, each one was considered as a separate study in the meta-analysis and compared with the control group. A sensitivity analysis was performed for included study to test the robustness of the results obtained by our meta-analysis.

RESULTS

The initial search for studies that compared TXA to placebo in children undergoing cardiac surgery retrieved 184 articles, from which 24 were fully screened. Only eight matched the inclusion criteria and were included for the analysis [10–17]. Table 1 reports the systematic review quality checklist. For each study included, it reports the level of evidence and grades of recommendations. The dose regimens used in the studies are summarized in Table 2. Patient characteristics and surgical data are presented in Table 3.

Table 1:

Quality and methodology of included studies

Study	Study design	Rand	Blind	TP	Primary outcome	Secondary outcome	LOE	GOR
Bulutcu et al. [14]	Prospective	Y	Y	Hct: 30% Qtt: N Clinics: N	BL: Y T: Y	Thr: Y Kid: Y Morbi: Y Morta: Y	1b	A
Chauhan et al. [10]	Prospective	Y	Y	Hct: 33% Qtt: N Clinics: N	BL: Y T: Y	ND	1b	A
Chauhan et al. [12]	Prospective	Y	Y	ND	BL: Y T: Y	Thr: Y Kid: Y Morbi: N Morta: N Re-expl: Y	1b	A
Chauhan et al. [11]	Prospective	Y	Y	ND	BL: Y T: Y	ND	1b	A
Reid et al. [13]	Prospective	Y	Y	ND	BL: Y T: Y	Thr: Y Kid: Y Morbi: Y Morta: Y	1b	A
Zonis et al. [15]	Prospective	Y	Y	ND	BL: Y T: Y	ND	1b	A
Levin et al. [16]	Prospective	Y	Y	ND	BL: Y T: Y	Coagulation Fibrinolysis	1b	A
Shimizu et al. [17]	Prospective	Y	Y	Hb: 12–15 Qtt: N Clinics: N	BL: Y T: Y	Thr: Y Kid: N Morbi: Y Morta: Y	1b	A

Rand: randomization; Blind: double blind; TP: transfusion policy; LOE: level of recommendation; GOR: grades of recommendations; Y: yes; N: no; ND: not defined; Hb: haemoglobin level; Hct: haematocrit; Qtt: quantity transfused defined in the protocol; Clinics: clinical evaluation before transfusion; BL: blood loss; T: transfusion requirement; Thr: thrombosis; Kid: kidney failure; Morbi: morbidity; Morta: mortality.

Table 2:

TXA dose and number of patients in each group of the studies

Study	Year	Placebo, n	TXA group, n	ID (mg kg−1)	ECC (mg kg−1)	End ECC (mg kg−1)	CI (mg kg−1 h−1)
Bulutcu et al. [14]	2005	24	25	100	100	100	0
Chauhan et al. [10]	2004	30	30	50	0	0	0
Chauhan et al. [10]	2004	30	30	10	0	0	1 (8 h)
Chauhan et al. [10]	2004	30	30	10	10	10	0
Chauhan et al. [10]	2004	30	30	20	0	20	0
Chauhan et al. [12]	2003	24	96	10	10	10	0
Chauhan et al. [11]	2004	50	50	10	10	10	0
Reid et al. [13]	1997	21	20	100	0	100	10
Zonis et al. [15]	1996	42	40	50	0	0	0
Levin et al. [16]	2000	28	28	50	0	0	0
Shimizu et al. [17]	2011	79	81	50	50	0	15

ID: bolus received at induction; ECC: bolus administered in the ECC prime; End ECC: bolus received at the end of ECC; CI: continuous infusion.

Table 3:

Patient characteristics and surgery data

Study	TXA group, n	Placebo, n	Age (months)	Weight (kg)	TOF	Fontan	Senning	ASD	VSD	PAC	Glenn	TAVR	TGV	Other
Bulutcu et al. [14]	25	24	47.4	6.0	27	0	0	0	0	0	10	10	3	0
Chauhan et al. [10]	30	30	45.6	6.6	46	8	6	0	0	0	0	0	0	0
Chauhan et al. [10]	30	30	51.0	6.1	42	11	7	0	0	0	0	0	0	0
Chauhan et al. [10]	30	30	43.8	6.7	43	10	7	0	0	0	0	0	0	0
Chauhan et al. [10]	30	30	43.2	6.8	42	10	8	0	0	0	0	0	0	0
Chauhan et al. [10]	96	24	51.6	8.0	83	25	12	0	0	0	0	0	0	0
Chauhan et al. [11]	50	50	49.8	7.0	61	19	20	0	0	0	0	0	0	0
Reid et al. [13]	20	21	37.8	12.4	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Zonis et al. [15]	40	42	57.7	18.8	6	6	0	26	16	5	3	0	0	20
Levin et al. [16]	28	28	45.8	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Shimizu et al. [17]	81	79	32.5	10.6	23	0	0	54	With ASD	0	9	29	0	7

TOF: tetralogy of Fallot; ASD: atrial septal defect; VSD: ventricular septal defect; PAC: pulmonary artery conduit; TAVR: total anomalous venous return; TGV: transposition of great vessels.

Postoperative blood losses (ml kg⁻¹) at 24 h were heterogeneously distributed (I² = 82%, P < 0.0001), precluding further analysis (Fig. 2). TXA reduced the need for RBC transfusion (Fig. 3A) by 6.4 ml kg⁻¹ day⁻¹ (I² = 0%, P = 0.45), PLT transfusion (Fig. 3B) by 3.7 ml kg⁻¹ day⁻¹ (I² = 0%, P = 0.46) and FFP transfusion (Fig. 3C) by 5.4 ml kg⁻¹ day⁻¹ (I² = 0%, P = 0.53). Compared with the placebo, CPB time with TXA was 2.11 min shorter (95% CI: −4.13 to −0.09; I² = 14%, P = 0.31).

Figure 2:

Differences in blood loss (ml kg⁻¹) at 24 h after surgery for TXA versus placebo. 95% CI: 95% confidence interval.

Figure 3:

(A) RBC transfusion, (B) PLT transfusion and (C) FFP transfusion expressed in ml kg⁻¹ day⁻¹ for TXA versus placebo. 95% CI: 95% confidence interval.

However, this analysis can be biased by two factors. First, four included studies were published by the same team. Second, one of the studies evolved different TXA schemes compared with a same control group and each TXA scheme was considered as a single study in the meta-analysis. We therefore performed a sensitivity analysis to take into account these possible biases (Table 4). This sensitivity analysis did not change the results of our first statistical analysis except for blood losses that were significantly reduced by TXA (−5.22 ml kg⁻¹; 95% CI: −8.16 to −2.28) when the studies of Chauhan et al. were excluded [10–12].

Table 4:

Sensitivity analysis performed after exclusion of the Ref. [10] and after exclusion of the studies published by Chauhan et al.

Subgroup	Studies, (n)	Patients (n)	Random	95% CI	I2 (%)	Fixed	95% CI	I2 (%)
Blood losses (ml kg−1)
Overall	11	848	−3.61	(−8.08, −0.85)	82	−3.79	(−5.67, −1.92)	82
Without Ref. [10]	7	608	−7.82	(−11.54, −4.10)	57	−8.04	(−10.41, −5.68)	57
Without Chauhan et al.	5	388	−5.22	(−8.16, −2.28)	0	−5.22	(−8.16, −2.28)	0
RBC transfusion (ml kg−1)
Overall	9	710	−6.38	(−8.28, −4.47)	0	−6.38	(−8.28, −4.47)	0
Without Ref. [10]	5	470	−7.57	(−10.17, −4.98)	0	−7.57	(−10.17, −4.98)	0
Without Chauhan et al.	3	250	−9.08	(−15.04, −3.12)	39	−8.83	(−13.48, −4.19)	39
FFP transfusion (ml kg−1)
Overall	8	669	−5.52	(−7.54, −3.50)	0	−5.52	(−7.54, −3.50)	0
Without Ref. [10]	4	429	−6.19	(−8.93, −3.45)	4	−6.19	(−8.87, −3.52)	4
Without Chauhan et al.	2	209	−4.48	(−10.27, 1.31)	40	−4.13	(−8.49, 0.23)	40
PLT transfusion (ml kg−1)
Overall	7	520	−3.7	(−5.40, −2.00)	0	−3.7	(−5.40, −2.00)	0
Without Ref. [10]	3	180	−3.12	(−7.09, 0.96)	53	−3.55	(−6.277, −0.83)	53
Without Chauhan et al.	1	160	NA	NA	NA	NA	NA	NA

A subgroup analysis between cyanotic and non-cyanotic patients could only be performed for three studies [15–17]. However, the distribution of results was heterogeneous in cyanotic patients, precluding further analysis (I² = 70%, P = 0.07). In non-cyanotic patients, TXA was not associated with a significant reduction in blood loss compared with placebo (P = 0.47). The effect of TXA on transfusion requirements could not be analysed because of insufficient data.

The possible effects of TXA on postoperative complications (seizure, renal failure, thrombosis, re-exploration for bleeding) and mortality could not be assessed because these data were not presented in the majority of the included studies.

DISCUSSION

A systematic search of the literature addressing the efficacy of TXA compared with placebo to reduce blood loss and allogeneic transfusion requirement lead to the disappointing conclusion that the available literature is weak. For this reason, the present work is rather a systematic review of the literature in which appropriate statistical analysis was performed when data were available.

The results of our analysis indicate that TXA marginally decreased blood product transfusion compared with the placebo. Postoperative blood loss and other outcomes, such as re-exploration, were too heterogeneously distributed to be interpreted despite the inclusion of more than 300 patients in each group.

Several factors may explain the high degree of heterogeneity observed between the included studies. Notably, the variability in the dosage schemes used is striking. After anaesthetic induction, some studies used a single bolus that ranged from 10 to 100 mg kg⁻¹ [10, 15, 16]. Others used several boluses: at anaesthetic induction, in the CPB priming and after protamine administration [10–12, 14]. Still other studies used continuous infusion during and after CPB [10, 13, 17]. In the included studies, the choice of the TXA dosage was not based on pharmacodynamic data about the fibrinolytic inhibiting activity of the drug. Instead, dosage choice was empirical, based on its effects on blood loss. Chauhan et al. [10] performed the only study comparing different TXA dosage schemes in a paediatric population. They found that the maximum reduction in blood loss was obtained with the following scheme: a bolus of 10 mg kg⁻¹ TXA after anaesthetic induction, repeated at initiation and after weaning from CPB.

In fact, no pharmacological data on TXA are available in the paediatric cardiac surgery population. Pharmacokinetic data in adults suggested the administration of a loading dose of 12.5 mg kg⁻¹ given over 30 min, a maintenance infusion of 6.5 mg kg⁻¹ h⁻¹ and a CPB priming dose of 1 mg kg⁻¹ is needed to maintain blood TXA concentration higher than 345 µM, which is considered to be the lowest concentration to fully inhibit fibrinolysis. With this scheme, the authors also found that the TXA plasma concentration was more stable in comparison to repeated bolus administration schemes [18]. However, these pharmacokinetic data cannot unequivocally be applied to a paediatric population. Of note, the EACA pharmacokinetic profile has been found to be different in the adult and the paediatric population [19].

Although TXA reduces the volume of transfused allogeneic blood products, several concerns remain. First, most included studies did not describe the transfusion policies that have been used. It has been demonstrated that the efficacy of blood sparing techniques is highly dependent on the presence of well-defined transfusion policies [20]. In the absence of this information, the efficacy of TXA is difficult to evaluate. Second, reports on the mean volume of transfused blood products do not inform about the number of children avoiding blood product transfusion or about the number of donors to which the children were exposed. Information on both these variables is essential for evaluating the clinical efficacy of agents such as TXA. Third, several of the published studies are drawn from the same team. For this reason, we performed a sensitivity analysis to evaluate the impact on the results. The exclusion of these studies reduces the number of patients included in our analysis but did not change the overall results.

Most of the studies included both cyanotic and non-cyanotic patients. However, cyanotic patients have more pronounced bleeding tendencies, thrombotic complications and a higher postoperative mortality rate [21]. The mechanisms of the haemostatic disorders observed in these patients are complex and can be related to the degree of hypoxia and hyperviscosity [22], to complement [23] and contact activation, and to inflammation [24]. Only three studies performed a subgroup analysis of cyanotic and non-cyanotic patients [15–17]. Interestingly, these data showed that TXA does not seem to decrease blood loss at 24 h in non-cyanotic patients. Results for cyanotic patients could not be interpreted as they were too heterogeneously distributed.

Of note, data for other outcomes related to the use of TXA, such as re-exploration for bleeding, time to extubation, length of ICU and hospital stay, and mortality, were not available for the cardiac paediatric population.

Adverse effects associated with the use of TXA were not adequately reported in the included studies. In fact, these trials were not designed to address this question [25]. Recent studies compared the safety of TXA with aprotinin [26] and EACA [27]. Incidences of 9.6% renal injury, 1.8% renal failure, 3.5% seizures and 2.6% other neurological events were reported for children receiving TXA. These numbers were not different from those observed with aprotinin and EACA. As reported recently, the use of TXA and EACA is associated with side effects that are dose-dependent [28].

Finally, if aprotinin was known to exert an anti-inflammatory effect by inhibiting the kallikrein pathway, the situation seems to be less evident with TXA [29]. Recently, Weber et al. hypothesized that TXA [28] could play a beneficial role in PLT function for patients exposed to antiplatelet therapy [30].

Little information is available on the use of antifibrinolytics in paediatric cardiac surgery. In a recent meta-analysis, Schouten et al. [31] concluded that there is no evidence to suggest that TXA is less effective than aprotinin in reducing blood loss in major paediatric surgery. The results of the current systematic review confirm the conclusions of the qualitative literature review published by Eaton in 2008 [32], emphasizing the lack of consistent data on the topic and the need for further large randomized trials to evaluate the benefit-to-risk ratio of TXA in paediatric cardiac surgery.

In conclusion, the present systematic review shows that TXA reduces blood transfusion requirements significantly in paediatric cardiac surgery although the clinical relevance of these results is not clear. As data on postoperative morbidity and mortality and on TXA-related side effects could not be evaluated in the available studies, we conclude that the evidence for the routine use of TXA in paediatric cardiac surgery remains weak. Further studies are needed to assess the potential beneficial effects of TXA on postoperative outcomes and to define the optimal dosage scheme for TXA.

Conflict of interest: none declared.

Author notes

†

The abstract was presented at the annual symposium of the NATA (Network for Advancement of Transfusion Alternatives), Dublin, Ireland, April 2011.