Osteoarthritis (OA) is a major source of pain, disability, and socioeconomic costs worldwide and commonly affects athletes (Hunter and Bierma-Zeinstra, 2019). Total knee arthroplasty (TKA) is recommended for end-stage knee OA patients, but its safety concerns may outweigh the benefits (Juni et al., 2006). TKA surgery has many complications, especially hemorrhage, which can be controlled via the application of a pneumatic tourniquet, allogeneic transfusion, and antifibrinolytic therapy (Pawaskar et al., 2017; Arthur and Spangehl, 2019; Helito et al., 2019). Considering that tourniquet application during TKA is related to ischemic injury (Cao et al., 2018; Lei et al., 2019) and allogeneic blood transfusion has been associated with a poor postoperative outcome (Spahn, 2010), antifibrinolytic therapy may be a better choice.

Tranexamic acid (TXA) has attracted great interest in the past decade under the advantages, including ease of administration, low expense, and excellent hemostatic efficacy (Good et al., 2003). TXA is a synthetic lysin-analog that inhibits fibrinolysis by blocking the lysine-binding sites on plasminogen. Previous studies have identified that intravenous (IV) (Kuo et al., 2018; Zhang S. et al., 2019), intra-articular (IA) (Guzel et al., 2016; Pinsornsak et al., 2016), oral (Alipour et al., 2013; Cao et al., 2018), and combined TXA administration (Lin et al., 2015; King et al., 2019; Wang et al., 2019) can successfully reduce blood loss and transfusions in primary TKA without increasing the risk of thrombosis. The most common form of TXA administration during TKA is through IV administration, and IV administration is getting seriously due to reducing systemic exposure.

A recent network meta-analysis (Fillingham et al., 2018) showed that regardless of the formulation of TXA used, patients undergoing TKA showed a significant reduction in blood loss and risk of transfusion compared to placebo without clear difference between different formulations of TXA administration. However, this network meta-analysis did not involve specific intervention plans, and the study did not pay attention to TXA security issues. TKA is a risk factor for deep venous thrombosis (DVT), coupled with the anti-fibrinolysis effect of TXA, so it should be more cautious in clinical application. According to the guidelines for the prevention of DVT in Chinese orthopedic surgery, Doppler ultrasound has become the preferred imaging method for DVT as a noninvasive angiographic technique. Currently, numerous findings regarding TXA therapies have been generated, but no consensus has been reached on the optimal route and dosage of TXA administration. We undertook this systematic review and network meta-analysis of RCTs and ranked TXA therapies to compare the hemorrhage control and DVT rate of different TXA therapies in TKA.

This is the first network meta-analysis to take all available evidence into account from RCTs directly or indirectly comparing specific TXA therapies for TKA use, thereby increasing the power of the study. The main findings are: 1) TXA therapies are effective for the blood management after TKA; 2) The highest probability of being the best intervention for total blood loss in TKA is probably M therapy (IV/IV infusion + oral TXA > 3 g) (SUCRA = 91.3%); 3) E therapy (IV TXA ≤ 10 mg/kg or 1 g three times), F therapy (IV TXA ≥ 15 mg/kg or 1 g three times), L therapy (IV/IV infusion + IA TXA > 3 g), and M therapy (IV/IV infusion + oral TXA > 3 g) are significantly more effective than A therapy, B therapy (IV TXA ≤ 10 mg/kg or 1 g once and IV TXA ≥ 15 mg/kg or 1 g once) and G therapy (IA TXA < 2 g) for blood loss reduction; 4) L therapy (IV/IV infusion + IA TXA > 3 g) is significantly more effective than G therapy (IA TXA < 2 g), H therapy (IA TXA ≥ 2 g) therapy (IA TXA only) and I therapy (oral TXA ≤ 2 g) for total blood loss reduction; 5) All the TXA therapies in this study are independent with DVT risk; 6) Based on the clustergram of total blood loss and DVT risk (shown in Figure 4), F therapy (IV TXA ≥ 15 mg/kg or 1 g three times) is suggested for TKA according to the surface under SUCRAs considering the hemorrhage control and DVT rate simultaneously.

Our NMA of TXA therapies for patients with TKA demonstrated that all available therapies of TXA were significantly effective for the treatment of total blood loss. This evidence is reinforced by a previous network meta-analysis (Fillingham et al., 2018), as well as many conventional meta-analyses (Liu et al., 2017; Guo et al., 2018; Moskal and Capps, 2018; Meng et al., 2019). The highest probability of being the best intervention as for total blood loss control in TKA is probably M therapy (IV/IV infusion + oral TXA > 3 g), but little data regarding M therapy (IV/IV infusion + oral TXA > 3 g) exist. Only one double-blinded trial, M therapy (IV/IV infusion + oral TXA > 3 g) was IV TXA 20 mg/kg twice, and then oral 1 g TXA from postoperative day (POD) 1 to POD 14, which showed that IV and subsequent long-term oral TXA produced less blood loss compared with short-term TXA without increasing the risk of complications (Wang et al., 2019). The second highest probability of being the best intervention for total blood loss control in TKA is F therapy (IV TXA ≥ 15 mg/kg or 1 g three times). Only one trial in our study associated with F therapy (IV TXA ≥ 15 mg/kg or 1 g three times) in TKA was included, in which three doses of IV TXA can effectively and safely reduce TKA-induced total blood loss (Tzatzairis et al., 2019). In terms of dosage, total dose of two therapies is more than 3 g. And the higher the total dose of TXA, the better it performed. Therefore, M and F therapies are effective for hemorrhage control. In terms of delivery method, both M and F therapies include IV administration. And oral administration is also included in M therapy. IV or oral TXA is both absorbed through the systemic circulation, and exposure level is very high, which is the material basis for the efficacy.

E therapy (IV TXA ≤ 10 mg/kg or 1 g three times) is also three doses of IV TXA, and the single dose of TXA as for E therapy (IV TXA ≤ 10 mg/kg or 1 g three times) is less than or equal to 1 g or 10 mg/kg. Five trials (Maniar et al., 2012; Shinde et al., 2015; Song et al., 2017; Sun et al., 2017; Adravanti et al., 2018) in our study associated with E therapy (IV TXA ≤ 10 mg/kg or 1 g three times) in TKA from 2012 to 2018 were included. Both E therapy (IV TXA ≤ 10 mg/kg or 1 g three times) and F therapy (IV TXA ≥ 15 mg/kg or 1 g three times) were significantly more effective than A therapy (IV TXA ≤ 10 mg/kg or 1 g once), B therapy (IV TXA ≥ 15 mg/kg or 1 g once) and G therapy (IA TXA < 2 g) in reducing blood loss. A clinical trial (Sun et al., 2017) comparing E therapy (IV TXA ≤ 10 mg/kg or 1 g three times) and B therapy (IV TXA ≥ 15 mg/kg or 1 g once) proved that E therapy (IV TXA ≤ 10 mg/kg or 1 g three times) was superior to B therapy (IV TXA ≥ 15 mg/kg or 1 g once) in reducing blood loss in TKA. Based on the available literature, there is no statistical difference between E therapy (IV TXA ≤ 10 mg/kg or 1 g three times) and F therapy (IV TXA ≥ 15 mg/kg or 1 g three times) in terms of reducing blood loss, and we find that three doses of IV TXA may be better than a single dose of TXA for reducing blood loss reduction. However, this conclusion is inconsistent with the previous NMA results (Fillingham et al., 2018), in which higher doses and multiple doses of TXA are not necessary to reduce blood loss. Lei et al. (2018) found that multiple doses of IV TXA could further diminish hidden blood loss, and decrease maximum hemoglobin drop following total hip arthroplasty (THA), similar to our study result that multiple doses of IV TXA may necessary in TKA.

IV injection for patients undergoing TKA is the best method for rapidly raising and maintaining the therapeutic concentration of TXA in the knee and the time taken for maximum plasma levels of TXA to be reached has been reported to be 5–15 min for IV administration (Pilbrant et al., 1981; Svahn et al., 1986). Many studies have evaluated the timing of intravenous TXA administration in TKA, and several clinical studies have shown the efficacy of TXA when first or only given before surgery (Castro-Menéndez et al., 2016; Seviciu et al., 2016), during or before deflation of the tourniquet (Volquind et al., 2016; Prakash et al., 2017), or at the end of the surgery (Veien et al., 2002; Keyhani et al., 2016). It seems that it is better when TXA is first or only administered before operation (Tanaka et al., 2001; Fillingham et al., 2018).

Evidence including a growing number of well-conducted RCTs and summarized clinical guidelines have demonstrated that combined TXA plays a potentially important role in supporting the effectiveness of blood loss reduction in patients with TKA (Ou et al., 2018; Zhang Y. M. et al., 2019). L therapy (IV/IV infusion + IA TXA > 3 g) and M therapy (IV/IV infusion + oral TXA > 3 g) are both combined TXA and their total TXA dose is more than 3 g, in which the former is IV/IV infusion combined with IA TXA and the latter is IV/IV infusion combined with oral TXA. There was no statistical difference between L therapy (IV/IV infusion + IA TXA > 3 g) and M therapy (IV/IV infusion + oral TXA > 3 g) in terms of blood loss reduction. Thus, combined TXA (total dose 3>g) may be better than A and B therapies (IV TXA once), similar to E therapy (IV TXA ≤ 10 mg/kg or 1 g three times) and F therapy (IV TXA ≥ 15 mg/kg or 1 g three times). Only one trial in our study is associated with M therapy (IV/IV infusion + oral TXA > 3 g) in TKA, in which three doses of IV TXA could effectively and safely reduce blood loss undergoing TKA. We find that IV/IV infusion combined with IA TXA maybe have a better effect on blood loss reduction than IA TXA only. There is a trend toward better efficacy in reducing blood loss using combined TXA.

H therapy (IA TXA ≥ 2 g) had the largest probability of having the least DVT risk (SUCRA = 79.20%), which reminded that physicians may consider IA TXA in patients with a higher risk of DVT. This was supported by a recent NMA (Xu et al., 2019), which concluded that physicians may consider topical alone in patients with higher risk of thrombosis for its best safety profile. Continuous and high doses of administration of TXA could conceivably increase the rate of thromboembolic disease and IA isolated injection of TXA combined with drain clamping was reported first in the English literature (Ishida et al., 2011). The use of IA TXA injection has many theoretical and practical advantages. IA TXA may be a more specific method than a systemic route and using only a small amount of TXA in limited knee joint volume could create a high TXA concentration inside the knee joint. An animal study showed topical delivery of TXA had very low systemic absorption, which resulted in avoiding systemic side effects (Damji et al., 1998).

According to our study, low dose IA TXA does not associate with lower DVT risk, hence a high dose of TXA with no less than 2 g is suggested for TKA. These TXA therapies have shown their differential impacts based on their characteristics or their specific techniques. When it comes to IV TXA only, three doses of IV TXA and a single dose of 15 mg/kg TXA may be reasonable choices. When it occurs to combined TXA, IV/IV infusion combined with IA TXA and oral TXA are both fine and suggest their dose is more than 3 g.

Rather than only considering blood loss and DVT risk simultaneously, as the biggest strength, our NMA was pioneered to assess each specific TXA dosing schedule individually and compare all TXA therapies simultaneously for patients with TKA. Besides, the TXA therapies are complex and multifaceted and this is the first network meta-analysis associating with specific TXA therapies, which proves the particular significance of our NMA.

The limitations of our study also need to be acknowledged. Firstly, there was significant heterogeneity among studies. In a small number of articles included, drainage tube among the trials may have accounted for such heterogeneity. Secondly, the duration of the follow-up of the included studies was variable. Thirdly, none analysis for hemoglobin drop, drain blood loss, or hidden blood loss was conducted in this study. Fourthly, there was only one study involving M therapy (IV/IV infusion + oral TXA > 3 g), and the same for F therapy (IV TXA ≥ 15 mg/kg or 1 g three times). Based on the results of this study, more attention needs to be paid to M therapy (IV/IV infusion + oral TXA > 3 g) and F therapy (IV TXA ≥ 15 mg/kg or 1 g three times) in future studies.

To summarize, our network meta-analysis indicates all thirteen TXA therapies are effective for hemorrhage control in TKA and no DVT risk is increased. The study provides some references that IV TXA ≥ 15 mg/kg or 1 g three times may be the optimal intervention for future studies.