Tranexamic acid is associated with selective increase in inflammatory markers following total knee arthroplasty (TKA): a pilot study

Informed consent was obtained prospectively from all participants, and the study was approved by the Institutional Human Research Ethics Committee (MHS20140812-03). The research undertaken strictly adhered to the Code of Ethics (Declaration of Helsinki) of the World Medical Association for trials involving humans. This study was an analytic, prospective, observational cohort (level II) investigation in which patient groups were separated non-randomly by treatment, with exposure occurring after the initiation of anesthesia.

Twenty-three patients (6 male, 17 female) undergoing TKA across three private practices were recruited to participate in the study (Table 1). The inclusion criteria were patients who were diagnosed with primary knee osteoarthritis (OA). The exclusion criteria were patients with (1) rheumatoid OA, (2) autoimmune disorders, (3) recent or recurrent infections with antibiotic treatment, or (4) contraindication for TXA use (thrombotic disorder and hematuria). Two of the three surgeons routinely use TXA perioperatively, and the remaining surgeon performed all surgeries for the non-TXA group.

All patients with exception of one TXA patient received both spinal and general anesthesia (Table 1). The anesthetic procedure included intravenous administration midazolam (0.02 mg/kg) and propofol target-controlled infusion (2–5 mg/kg/h) with fentanyl (75–85 μg) as required. A tourniquet was inflated prior to the first midline skin incision and a medial parapatellar access was used to expose the joint capsule. After the bone cut, an intra-articular cocktail comprising Ropivacaine (400 mg), Ketorolac (30 mg), Adrenaline (1 mg), and Methylprednisolone (40 mg) in a total volume of 200 ml of saline was injected into the sub-synovial space, ligaments and muscles around the knee. All unilateral TKA surgery was assisted with Precision Computer Navigation (Stryker®). Implanted prosthesis (cruciate retaining) were cemented (PALACOS® R+G), Heraeus Medical, Germany). After implantation and tourniquet deflation, the capsule was closed.

Following induction of anesthesia and prior to skin incision, an intravenous bolus injection of TXA (1.2 g per 90-kg patient) was administered (Fig. 1, Table 1). After the operation and before skin closure, a second TXA bolus (15.5 g/kg body wt.) was injected in the intra-articular space in saline (i.e., 1.0 g TXA/10 ml saline).

Patients were assessed preoperatively and followed up at 6 weeks postoperative for clinical assessments (goniometry for range of movement (ROM) as well as pain scores), and patient reported outcome measures, which included the Knee injury and Osteoarthritis Outcomes Score (KOOS), Oxford Knee Score (OKS), EuroQol (EQ-5D 3L) and Forgotten Joint Score (FJS) (Table 1). Practice nurses used the Angulus ROM iPhone app, which provides flexion and extension values by recording and measuring movement in both a horizontal and vertical plane.

Peripheral venous blood was collected from patients at three time points: (1) baseline, prior to anesthesia, (2) ~ 10 min after the first bone cut, and (3) ~ 30 min following skin closure. Blood was also collected from patients on days 1 and 2 postoperative. Whole blood (1.8 ml) was collected in 3.2% sodium-citrate vacutainers (BD Australia) for coagulation assessment, and 4 ml was collected into K₂EDTA vacutainers (BD Australia) and centrifuged (1500 rpm, 15 min, 4 °C). Plasma was removed and snap-frozen in liquid nitrogen and stored at − 80 °C for cytokine measurements.

Rotational thromboelastometry (ROTEM®, Tem International, Munich, Germany) was performed on ROTEM® delta according to the manufacturer’s instructions and described in Letson and Dobson [20] and Solomon and colleagues [21] (Fig. 2). Assays were run for 60 min. Hyperfibrinolysis was defined as a maximum lysis index greater than or equal to 15% [20, 22].

A priori power analysis to determine sample sizes was conducted using G-power3 program to minimize type 1 errors and was based on differences between coagulation parameters prior to and at surgery end [16]. A sample size of 10 patients in each group was sufficient for statistically valid comparisons to be made with respect to TXA vs non-TXA treatments with the power set at 0.8 and alpha level at 0.05. SPSS Statistics 24.0 was used for all data analysis (IBM, Armonk, NY). Data normality was assessed using Shapiro-Wilks test, with Levene’s test used to determine equality of variances. Independent samples t tests were used for between-groups comparison for normally distributed data. Within group differences were analysed with paired samples t tests. Non-normally distributed data was compared using a Mann-Whitney U test. MILLIPLEX Analyst 5.1 software (Luminex Corporation, Austin, Texas, USA), which analyses data with a 5 parametric logistic weighted curve fit, was used to determine cytokine concentrations. Area under the curve (AUC) was determined for changes in plasma cytokine levels across each of the five time points assessed. The mean AUC for each cytokine was compared for non-TXA and TXA patients using an independent-t test with Welch’s correction. All values are expressed as mean ± standard error of the mean (SEM) with significance set at p < 0.05.

There were no significant preoperative differences in patient demographic or clinical parameters (Table 1). Tourniquet time was significantly less in TXA patients compared to non-TXA patients, with no significant differences in surgical times (Table 1). Lower tourniquet times may be due to differences in surgical procedures among surgeons; however, it is important to note that possible longer ischemic times in non-TXA patients may exacerbate postoperative inflammation; however, it was less than TXA-treated patients in our cohort (see below). No differences in preoperative knee biomechanic measures were observed, with the exception of significantly higher extension (5° vs. 2°) in the non-TXA group (Table 2). At 6 weeks postoperative, patients within each group demonstrated significant improvements in KOOS measures compared to baseline, with no significant differences between scores for non-TXA and TXA patients (Table 2).

There were no significant differences in baseline plasma inflammatory mediators between non-TXA and TXA patients (Fig. 3). At surgery end and postoperative days 1 and 2, patients that received TXA had significantly higher plasma levels of MCP-1 compared to non-TXA patients (Fig. 3). Area-under-the-curve (AUC) analysis over 3 days supported this finding (p = 0.013) (Fig. 4). TNF-α was also significantly higher in TXA patients at each of the time points assessed (Fig. 3), and supported by AUC analysis (p = 0.010). IL-6 was significantly higher immediately after surgery and 1.8-times higher than non-TXA group on day 2 postoperative, but did not reach significance. Similarly, IL-8, a chemokine attractant for neutrophils and lymphocytes, and inducible by TNF-α and IL-1β [23], was 1.8 times higher on day 2 (p = 0.085) in TXA versus non-TXA patients. Levels of IL-1β, an inflammation amplifier, were also elevated in plasma of TXA patients after the first bone cut and at surgery end (Fig. 3). However, despite a ninefold increase in IL-1β in TXA patients compared to the non-TXA group at day 2 postoperative (Fig. 3) and a threefold higher AUC value (Fig. 4; p = 0.064), these differences were not significant. Plasma levels of IL-1RA remained unchanged throughout surgery through to postoperative day 2 (Fig. 3). Plasma IL-4 levels were higher in TXA patients compared to non-TXA patients after the first bone cut and at surgery end, with levels continuing to increase on day 2. The AUC for IL-4 was significantly higher (8 times) for TXA than non-TXA patients (p = 0.042, Fig. 4). After surgery, the anti-inflammatory cytokine IL-10 peaked in both patient groups, and then decreased on days 1 and 2, with a trend toward higher values of IL-10 in plasma from non-TXA compared to TXA patients (1.45 to 2.2 times higher) (Fig. 3).

In chronic OA patients, joint inflammation appears to be expressed systemically as a low-grade inflammatory state [24‐29]. We found that baseline levels of plasma MCP-1 (CCL2) were up to four times higher than in aged-matched healthy individuals (95–168 pg/ml) [30, 31], and TNF-α levels almost three times higher than in normal humans (~ 5 pg/ml) [31, 32] (Fig. 3). MCP-1 is a chemokine that regulates recruitment of immune cell traffic from the circulation to sites of inflammation in OA patients [33, 34], and has been implicated in articular cartilage degradation and pain [27, 35]. TNF-α is another potent inflammatory mediator involved in OA progression [36, 37], contributing to cartilage loss through its suppression of collagen and proteoglycan synthesis [24, 36, 38]. Notwithstanding the difficulty of finding aged-matched healthy human data, our data suggest the OA patients in the current study presented with a low-grade systemic inflammation.

We report significant increases in plasma IL-1β and TNF-α in TXA patients after the first bone cut and at surgery end compared to patients that did not receive TXA (Fig. 3). At the end of surgery, TNF-α and IL-1β continued to increase in TXA patients and were accompanied by significantly higher IL-6 and MCP-1 compared to non-TXA patients (Fig. 3). Although the cytokine increases were small, they indicate a heightened inflammatory state in the TXA patients, and heightened surgical stress response [39‐41]. During this early period in knee or hip surgery, Hall and colleagues have confirmed activation of the surgical stress response involving the hypothalamic–pituitary–adrenal (HPA) axis, with concomitant increases in plasma cortisol and catecholamines [42]. In TKA, the stress response is most likely activated from multiple neural, hormonal and metabolic inputs including danger signals (e.g. alarmins) from, soft tissue and bone resection, and firing of afferent nerves, that are detected by resident and circulating immune cells, and the brain respectively [40].

In addition to TXA exacerbating the inflammatory response during surgery, another key finding was the apparent amplifying effect of TXA on inflammatory cytokine levels over the first two postoperative days (Fig. 3). We found increased concentrations of plasma MCP-1, TNF-α, IL-1β, IL-6, IL-8, and IL-4 and decreased IL-10 levels in patients that received TXA compared to those that did not. AUC analysis from baseline to postoperative day 2 showed significantly higher levels of MCP-1, TNF-α and IL-4 in plasma of TXA than in non-TXA patients (Fig. 4). The differences in IL-4 are of particular interest since it is generally regarded as an anti-inflammatory cytokine, similar to IL-10 and IL-13 [43]. In this role, IL-4 is known to inhibit TNF-α production and IL-1β synthesis and to increase IL-1RA [43, 44]. However, the opposite occurred in TXA patients in our study. At postoperative day 2, plasma TNF-α levels were twofold higher, and IL-1β was fivefold higher compared to non-TXA patients, with no change in IL-1RA (Fig. 2).

Recently, Major and colleagues also reported that IL-4 was not purely an anti-inflammatory cytokine, but could prime macrophages, increase TNF-α and increase inflammation [44]. IL-4, in combination with GM-CSF, can further promote inflammation by increasing differentiation of monocytes into dendritic cells [44]. Bellini and colleagues also showed that IL-4 can stimulate a unique circulating leukocyte subpopulation (0.1–0.5%) of bone marrow-derived stem cells known as fibrocytes that leave the blood and enter the site of healing and differentiate into fibroblasts/myofibroblasts with increased production of cell matrix components, growth factors, and inflammatory cytokines [45‐47]. Therefore, in the current study, it is possible that IL-4 contributes to a heightened systemic inflammatory response observed in TXA patients (Figs 3, 4).

In our study, baseline ROTEM clotting parameters for OA patients were similar to normal healthy individuals [22, 31, 32, 48]. In contrast to a low-grade inflammatory state at baseline in our OA groups, it appears that there were no apparent coagulation defects. However, after the first bone cut and surgery end, the non-TXA patients had decreased EXTEM, FIBTEM and INTEM clot times (9 to 21% falls relative to baseline) (Table 3), indicating increased thrombin availability. This was further supported by 22 and 34% decreases in EXTEM CFT and INTEM CFT, respectively, with little or no change in α-angles (Table 3). The shift in CT and CFT was associated with no effect on clot amplitude or strength (Fig. 3) but a twofold increase (3.1 to 6.3%) in maximum lysis in EXTEM (p = 0.021) and 1.4 times (4.6 to 6.3%) in INTEM after the bone cut (Table 3). Notably, the increases in fibrinolysis in non-TXA patients were within the range of normal values, with 15% often used as a guide to indicate hyperfibrinolysis [20, 22]. Thus, in non-TXA patients, surgical stress appeared to decrease clot times and thrombin availability without changes in other ROTEM parameters.

However, in TXA patients, EXTEM CT at surgery end was significantly higher (1.3-fold) than non-TXA patients, indicating that TXA during surgical stress has a thrombin-slowing effect (Table 3). We reported a similar finding in cardiac surgery after a sternotomy with TXA having a twofold increase in CT (all tests) [16]. In that study, we speculated that TXA may (1) reduce the rate of prothrombin-thrombin conversion, or (2) inhibit one or more of the polypeptide cleavage reactions, and thereby slow the fibrinogen to fibrin conversion [16]. Interestingly, since TXA is a lysine analogue, reducing the prothrombin-thrombin conversion is possible since the kringle-2 domain of the prothrombin complex is rich in lysine residues, and TXA may partially block these sites thus reducing thrombin production. In contrast to our cardiac surgery study, both TXA and non-TXA patients had significantly decreased INTEM CT during surgery in the current study (Table 3), highlighting differences in TXA with clotting factors or pathway selection. Importantly, and in agreement with our previous study, we found no difference in EXTEM or INTEM clot lysis in TXA and non-TXA patients. This finding suggests that perhaps the beneficial effect of TXA published in a large number of randomized controlled trials involving nearly 1 million patients [10] might not be reflected by the absence of evidence of hyperfibrinolysis with ROTEM (and TEG). In addition, we found no difference in hemoglobin levels between the groups postoperatively (Table 1), suggesting blood loss was similar for both TXA and non-TXA patients after TKA surgery.

Another interesting trend observed in the present study was that TXA increased maximum lysis in FIBTEM after the bone cut (5-fold higher) and the end of surgery (~ 3-fold higher) compared to non-TXA patients (Table 3). The FIBTEM test is EXTEM with platelet inhibition (see Fig. 2), and these paradoxical results suggest that TXA weakens, not strengthens, the fibrin network in the absence of platelet contribution. This pro-fibrinolytic effect of TXA was not due to falling levels of fibrinogen because there was no change in FIBTEM amplitudes (Fig. 5). Currently, we do not know the underlying mechanisms for this effect of TXA on maximum lysis. Platelets normally support the formation of a dense, stable fibrin network from αIIbβ3 integrin interactions and the fibrin network [49]. In the absence of platelets, it appears that lysine residues play a role in securing fibrin density and stability in the FIBTEM clot, which is decreased in the presence of TXA. While this observation may be clinically silent under normal hematological conditions, it has the potential to become a significant problem in major surgery or various trauma states, where platelet numbers may decrease or platelet activation is impaired numbers or function.

An important finding in the present study was that there appeared to be no clinical advantage of using TXA in our patient group undergoing elective TKA. Without evidence of hyperfibrinolysis (Table 3), there is no clinical justification for TXA use because there is no excessive bleeding [4]. In addition, TXA administration appeared to have a potentially untoward pro-inflammatory effect during and after surgery, which may be linked to a more pronounced TXA-induced stress response to the trauma of surgery. This may be clinically significant since Galvez and colleagues recently demonstrated that patients with OA already have a diminished ability to tolerate surgical stress [29], which the authors associated with a pre-existing low-grade chronic systemic inflammation [40, 42]. Our study further underscores a number of outstanding questions on TXA use in major surgery: (1) Would a single dose administration of TXA have less effect to increase inflammation and stress response to surgery? (2) What laboratory tests should be used to drive TXA use in elective or emergency surgery? and (3) What is the scientific basis for using TXA in orthopaedic surgery? In our view, TXA should not be viewed as a one-size-fits-all approach to elective surgery; rather it should be incorporated into a more precision-based set of guidelines [4, 50, 51]. The potential harmful effects of TXA on promoting inflammation warrant further investigation.

A major limitation of our pilot study was its lack of randomization, blinding and small patient numbers. Our postoperative period was also limited and requires extension beyond day 2 when joint swelling is at a maximum, and 10–14 days when adhesions begin to form. Given surgically induced inflammation is also linked to postoperative pain and fragmented sleep patterns following TKA [42, 52, 53], these additional metrics should be included in future studies. Another limitation was that we did not measure plasma stress hormones, which may be higher in patients with higher inflammatory status during and following surgical stress [39‐42]. We also do not know the effect of the cocktail components that were injected around the knee after the bone cut on TXA’s effect to change some ROTEM parameters and/or inflammatory markers. In vitro studies are also required to examine TXA’s effect on fibrinogen with and without platelets, and role of lysine residues using rapid-kinetic monitoring, X-ray crystallography, nuclear magnetic resonance and electron microscopy techniques. Notwithstanding these limitations, our study provides a springboard for a larger prospective, randomized trial to further elucidate the effects of TXA on inflammation and the surgical stress response to TKA, the outcomes of which may have implications for other pediatric and adult elective and emergency surgeries.