Mitigating Lipemia Interference in Anti-Xa Activity Measurement Through High-Speed Centrifugation

Lipemic samples are challenging for diagnostic laboratories, as lipids are likely to interfere in spectrophotometric methods by altering sample turbidity. Such samples can occur when blood is taken after a meal or following the infusion of parenteral lipid emulsions. In these cases, resampling may resolve this issue [1]. Nevertheless, lipemia can result from certain conditions (familial hypertriglyceridemia, pancreatitis, treatment with protease inhibitors, etc.), where resampling is ineffective.

Some authors have suggested that high speed centrifugation (over 10 000 g) could mitigate this interference [2, 3]. One recent study examined whether such a method could be used for coagulation tests such as PT, aPTT, fibrinogen [4], for instance. Although most suppliers indicate that lipids are likely to interfere with anti-Xa (AXA) activity, the effect of such a process on this assay has not been evaluated yet to our best knowledge.

In this study, approved by the institutional ethics board (CE-2024-34), samples from inpatients at University Hospital of Strasbourg (France) were included between December 2023 and March 2024. Inclusion criteria were as follows: patients over 18 years old anticoagulated with unfractioned heparin (UFH), low molecular weight heparin (LMWH), apixaban or rivaroxaban. Underaged patients and those with visibly lipemic samples were excluded, to ensure that only non lipemic reference samples were included. Patient blood was drawn into Vacutainer glass citrated tubes (Becton Dickinson, Franklin Lakes, USA) or Vacutest PET citrated tubes (Kima, Padua, Italy), with 0.109 of trisodium citrate. Platelet-poor plasma (PPP) was prepared by centrifugation at 2500 g for 10 min at 20°C. For high-speed centrifugation (HSC), aliquots of 1.5 mL of PPP were centrifuged at 11 000 g for 10 min in a high speed centrifuge Mikro 200R (Andreas Hettich, Tuttlingen, Germany). As the lipid fraction was found in the supernatant, the aqueous phase of PPP was carefully collected at the bottom of the tube for measurements.

First, to ensure that HSC would not interfere on this assay, AXA activity was measured on PPP of patients anticoagulated with UFH, LMWH, apixaban or rivaroxaban before and after HSC, on a STA-R Max analyzer with STA Liquid anti-Xa (both Diagnostica Stago, Asnières-sur-Seine, France).

Second, to determine whether such a process could effectively mitigate the interference due to lipemia, aliquots of 1.5 mL of PPP were spiked with lipids using 25 μL of Smoflipid 200 mg/mL (Fresenius Kabi, Sèvres, France) to achieve TG levels ≥ 6.9 g/L (manufacturer's established cut-off for TG interference). AXA was then measured before spiking, after spiking and after HSC. Triglycerids (TG) were measured after spiking and after HSC on a Dimension Vista 1500 analyzer with Trig Flex (both Siemens, Erlangen, Germany).

Wilcoxon tests were carried out to compare AXA and triglyceridemia. A p value ≤ 0.05 was considered statistically significant. Analyses were performed with Prism v6.05 (GraphPad Software).

One hundred and six patients were included. The impact of high-speed centrifugation on AXA was assessed on 26 patients anticoagulated with UFH (n = 8; AXA: 0.13–1.43 IU/mL), LMWH (n = 6; AXA: 0.2–1.56 IU/mL), apixaban (n = 8; AXA: 35–130 ng/mL), and rivaroxaban (n = 4; AXA: 28–118 ng/mL). No significant differences were found in AXA before and after HSC (see Figure 1), indicating that HSC did not significantly impact this assay.

Spiking experiments were carried out on samples from patients anticoagulated with UFH (n = 30), LMWH (n = 15), apixaban (n = 26) and rivaroxaban (n = 9). Mean TG after spiking was 9.2 g/L, while mean TG after HSC was 6.1 g/L (p < 0.05, see Figure 2E). In the UFH, LMWH, apixaban, and rivaroxaban subgroups, mean TG after spiking was 9.2 g/L, 8.9 g/L, 9.3 g/L, and 9.5 g/L, respectively, while mean TG after HSC was 6.3 g/L, 6.0 g/L, 6.0 g/L, and 6.0 g/L, respectively.

Regarding UFH, AXA before spiking (mean = 0.35 IU/mL) and after spiking (mean = 0.29 IU/mL) were significantly different (p < 0.05), while AXA after HSC (mean = 0.34 IU/mL) did not significantly differ from AXA before spiking (p = 0.22, see Figure 2C).

Similar results were observed for LMWH, as AXA before spiking (mean = 0.63 IU/mL) and after spiking (mean = 0.58 IU/mL) were significantly different (p < 0.05). Meanwhile, AXA after HSC (mean = 0.62 IU/mL) was not statistically different from AXA before spiking (p = 0.11, see Figure 2D).

Finally, spiking with lipids (TG up to 11.8 g/L) did not seem to interfere significantly on the measurement of rivaroxaban or apixaban (see Figure 2A,B). Indeed, mean rivaroxaban (77 ng/mL) before spiking was not different from mean rivaroxaban after spiking (73 ng/mL; p = 0.18) or after HSC (77 ng/mL; p = 0.88). Similarly, mean apixaban before spiking (84 ng/mL) was not different from mean apixaban after spiking (81 ng/mL; p = 0.08) or after HSC (84 ng/mL; p = 0.98).

Notably, triglyceride levels remained above 6.9 g/L after HSC in three samples from patients anticoagulated with UFH and three from patients anticoagulated with apixaban. Despite this, no significant differences were observed between AXA levels before spiking and after HSC (p = 0.5 and > 0.99, respectively).

In conclusion, in this study on 106 patients anticoagulated with AXA drugs, we assessed whether HSC could efficiently mitigate the interference due to lipemia for AXA measurement. We first demonstrated that HSC did not affect AXA measurement, and second that HSC was efficient for overcoming lipemia for UFH and LMWH AXA. Indeed, we noticed a significant difference in AXA before and after spiking, after spiking and after HSC, while no difference was observed between AXA before spiking and after HSC.

In the meantime, lipemia (TG up to 11.8 g/L) did not interfere on apixaban and rivaroxaban AXA. This is most likely because, on STA-R Max analyzers, the final dilution of plasma within the reaction medium is roughly 8/100 for UFH or LMWH AXA, while it is 2/100 for rivaroxaban or apixaban.

This study has several limitations: the experiment was conducted using a single type of analyzer (STA-R Max) and a single type of reagent (STA Liquid anti-Xa). As edoxaban and betrixaban are not commercially available in France, they could not be evaluated. Additionally, because the interference depends on the type of lipids, the lipid composition used in this study might not reflect accurately that observed under “physiological” conditions. Therefore, it is still necessary to perform such studies on native lipemic plasmas.

Despite these limitations, HSC is a simple and effective method to mitigate lipemia interference on UFH and LMWH AXA measurements, suitable for most diagnostic laboratories.

A.H. designed the study, analyzed data and wrote the manuscript. C.O. and C.L. collected and analyzed data. J.W., L.S. and L.M. revised intellectual content.

This study was approved by the institution's ethics board (CE-2024-34).

The authors have nothing to report.

The authors declare no conflicts of interest.