A Service Evaluation of Anti-Xa Measurements in Patients with Kidney Impairment in a Tertiary Centre

Abstract

Background

Pre-emptive dose reduction of low-molecular-weight heparins (LMWHs) is often utilised in those with chronic kidney disease (CKD) to prevent bioaccumulation. We report on the association of a pre-emptive dose reduction on anti-Xa range and its correlation with clinical outcomes.

Methods

We undertook a retrospective service evaluation of patients with CKD (eGFR < 30 mL/min/1.73 m²) receiving therapeutic dose dalteparin. The primary exposure was dalteparin anti-Xa trough and peak. Primary outcomes were ISTH-defined clinically relevant bleeding, thrombosis, and all-cause mortality within 90 days of LMWH initiation. A multivariate Cox proportional hazards model was employed to assess the relationship between anti-Xa levels and the incidence of bleeding and mortality.

Results

A total of 103 patients were identified over a 2-year period. Seventy-eight (75.7%) had anti-Xa monitoring done. Trough anti-Xa distribution was within-target in 58 (75.6%). Patients on dialysis had a higher incidence of bleeding (19 vs. 12, p < 0.05). Patients with bleeding had significantly higher median anti-Xa trough (0.26 vs. 0.13 U/mL, p < 0.01). The median time to bioaccumulation was 19 days. In multivariate Cox models, only anti-Xa trough remained an independent association with bleeding (HR: 1.47 per 0.1 U/mL, 95% CI: 1.05–2.15; p < 0.05). No associations with mortality were identified.

Conclusion

In this report, trough anti-Xa measurement of dalteparin is independently associated with bleeding in patients with CKD. Further prospective, larger studies are warranted to validate these results before it can be universally recommended in clinical practice.

Introduction

Chronic kidney disease (CKD), defined as eGFR < 90 mL/min/1.73 m², affects an estimated 22% of UK adult population (greater than 35 years old) [1]. Approximately 48% of individuals aged greater than 75 are affected by CKD stages 1–5, with 36% having CKD at stage 3a or above [2].

Patients with kidney impairment are at increased risk for both arterial and venous thromboembolic events. Reduced eGFR independently increases venous thromboembolic risk by over two-fold [3–5]. The mechanisms behind this are heterogenous and not fully understood; it is thought that CKD promotes a prothrombotic state through endothelial dysfunction, elevated fibrinogen and von Willebrand factor, and impaired fibrinolysis while concomitantly promoting bleeding via uraemia-induced platelet dysfunction [6–9]. Atrial fibrillation is also more prevalent as kidney function declines, further complicating anticoagulant management [10].

Low-molecular-weight heparins (LMWHs) such as dalteparin offer predictable pharmacokinetics, superior anti-Xa:IIa specificity, and reduced binding to plasma proteins compared with unfractionated heparin [11–13]. Their predominant kidney clearance, however, leads to prolonged half-life and risk of accumulation in those kidneys impaired [14]. Several studies correlate anti-Xa levels with kidney function, suggesting monitoring in severe impairment [15–17]. One non-randomised cohort study demonstrated that pre-emptive 25–50% dose reductions of therapeutic LMWH yield target anti-Xa peak in only 19% of patients without clear outcome benefits in bleeding and mortality prediction [18].

Current UK practice of LMWH prescription in those with kidney impairment varies widely. A survey of nephrology and haematology clinicians found diverse use of LMWH dosing and monitoring strategies in CKD stage 3–5, including dialysis patients [19]. National guidance advises selecting kidney function estimates carefully to avoid adverse bleeding events [20], yet the utility of anti-Xa monitoring remains controversial [21]. We therefore undertook a retrospective service evaluation to determine whether our local protocol of dalteparin prescription in this population correlates with anti-Xa levels being in range and if these levels are associated with bleeding or 3-month all-cause mortality in those with severe kidney impairment.

Methods

Study Design and Patient Selection

We retrospectively interrogated electronic health records to identify adults with eGFR <30 mL/min/1.73 m² by CKD-EPI [22] (including those with acute kidney injury or on dialysis) who had received therapeutic dose dalteparin between September 2022 and September 2024. Patients who had less than three doses of dalteparin or lost to follow-up <3 months were excluded. This study was not deemed as research via the Health Regulatory Authority research tool. It was registered as a clinical audit with Manchester University NHS Foundation Trust Clinical Audit Department.

Dalteparin Prescription and Anti-Xa Monitoring

In patients with kidney impairment, a pre-emptive dose reduction is applied according to our local protocol. In line with the UK Kidney Association national guidelines, creatinine clearance is used to guide therapeutic dosing of LMWH [23]. When creatinine clearance is below 30 mL/min, the dalteparin dose is reduced by one-third and administered as a split dose twice daily. Anti-Xa monitoring includes both peak and trough levels. Trough anti-Xa levels are at a target of less than 0.25 U/mL for twice daily dosing. Peak levels are measured 4 h after the third dose, in accordance with standard pharmacokinetic monitoring practice.

Laboratory Assay

Anti-Xa activity was measured using a chromogenic assay (Hyphen Biomed) based on the inhibition of factor Xa by heparin-antithrombin complexes. All assays were performed on the Sysmex CN series analyser. In this amidolytic method, factor Xa hydrolyses a chromogenic substrate unless inhibited by the heparin-antithrombin complex. The resulting release of para-nitroaniline is inversely proportional to heparin concentration.

Venous blood was collected in 0.109 m trisodium citrate tubes (9:1 ratio). Plasma was separated by centrifugation at 3,500 rpm for 10 min. Only properly filled, non-haemolysed samples were accepted per SOP criteria. Reagents included ready-to-use chromogenic substrate and factor Xa (Hyphen Biomed). Calibration was performed with five-point calibrators and controls (C1 and C3), reconstituted in sterile distilled water. Calibrations were verified with every new reagent batch. Interferences and cross-reactivity potential interferences (e.g., lipaemia, haemolysis, rivaroxaban, apixaban) were managed per laboratory SOP. Elevated results triggered cross-checks for DOAC use.

Demographic and Clinical Data

Demographic and clinical variables such as age, sex, weight, creatinine clearance (Cockcroft-Gault [24]), dalteparin dose and frequency were recorded (see online supplementary data S1; for all online suppl. material, see https://doi.org/10.1159/000551515). Bioaccumulation of dalteparin was defined as a subsequent trough anti-Xa level exceeding 0.25 U/mL in patients whose initial trough level was ≤0.25 U/mL. Major bleeding (fatal bleeding, critical-site bleeding, Hb drop >20 g/L, or transfusion of ≥2 units) and clinically relevant non-major bleeding (CRNMB, not meeting the major bleeding definition but requiring medical intervention by a healthcare professional, leading to increased level of care, or prompting an evaluation) were defined by ISTH criteria [25] All-cause mortality was assessed through 3 months post-initiation. Bleeding and thrombosis events were noted while the patient was on dalteparin.

The primary exposure was anti-Xa level peak and trough measured within five doses of LMWH initiation as per UK Kidney Association guidelines [23]. Trough levels are drawn ≤1 h before the next dose (pre-dose) within the first five doses of starting or after any dose change. Peak levels are drawn 3–5 h after a subcutaneous dose once at steady state (after ≥3 doses and trough levels). We analysed anti-Xa levels as [1] a continuous variable (per 0.1 U/mL increase) and [2] categorically relative to protocol targets (within-target vs. above-target as defined above).

Outcomes

The primary endpoints were ISTH-defined clinically relevant bleeding, thrombosis, and all-cause mortality within 90 days of dalteparin initiation. Our primary analytical aim was to estimate the association between anti-Xa trough/peak and these endpoints.

Statistical Analysis

Multivariate Cox proportional hazards models, adjusted for age, sex, weight, creatinine clearance, anti-Xa trough and peak (U/mL), tested factors associated with bleeding and mortality. We adjusted for body weight rather than BMI because therapeutic dalteparin is weight-based. Anti-Xa trough was modelled per 0.1 U/mL (continuous) and as above-target vs. within-target (categorical). Anti-Xa peak was modelled analogously. Discriminatory ability of anti-Xa for bleeding was assessed by area under the receiver operating characteristic (ROC) curve. Statistical analyses were done on R, with two-sided p < 0.05 considered significant.

Results

Baseline Demographic Data

A total of 103 patients with eGFR <30 mL/min/1.73 m² on dalteparin were identified (Table 1). Median age was 67 years (IQR 53–73), 60 (58.3%) were male, median weight was 80.0 kg (IQR 67.5–98.0), and median creatinine clearance was 21.6 mL/min (IQR 13.1–30.2).

Table 1.

Baseline demographics data of the cohort (n = 103)

Gender, n (%)	N = 103
Male	60 (58.3%)
Female	43 (41.7%)
Renal impairment, n (%)
CKD stage 4	23 (22.3)
CKD stage 5	9 (8.7)
Acute kidney injury	18 (17.5)
Dialysis	53 (51.5)
Body mass index, n (%)
BMI <18.5	5 (4.9)
BMI 18.5–24.9	16 (15.5)
BMI 25–29.9	39 (37.9)
BMI 30–30.9	32 (31.1)
BMI >40	11 (10.7)
Indications for anticoagulation, n (%)
Therapeutic for thrombosis	79 (76.7)
Peri-procedural bridging	12 (11.7)
Temporary substitute in non-procedural settings	10 (9.7)
Acute coronary syndrome	2 (1.9)

Anti-Xa Monitoring

Of the 103 patients, 78 (75.7%) had anti-Xa measurements and comprised the analysis cohort (Fig. 1). A bleeding event occurred in 23 individuals (29.5%) and 14 patients (17.9%) died within 3 months. There was no significant difference in the incidence of bleeding between those who had anti-Xa monitoring and those who did not (23 vs. 8, p = 0.811).

Fig. 1.

Flowchart of patient inclusion in the report.

All 78 patients with monitoring had an anti-Xa trough within five doses from initiation, with 58 (75.6%) within-target and 19 (24.4%) above-target. Anti-Xa peak was done in 65 (83.3%), with 21 (32.3%) within range, 41 (63.1%) below, and 3 (4.6%) above. Twelve individuals had anti-Xa accumulation. Of which, half were on dialysis. The median time to bioaccumulation was 19 days (95% CI: 11–26). The median duration of dalteparin therapy in this group was 12 days (95% CI: 9.5–16.5).

Bleeding and Thrombotic Events

In the entire cohort, thirty-one individuals (30.1%) reported a bleeding event. Of which, 18 (58.1%) were major bleeding and 13 (41.9%) were CRNMB events. Patients on dialysis had a significantly greater incidence of bleeding events compared to those without (19 vs. 12, p < 0.05). Weight ≤55 kg vs. >55 kg (p = 0.729) and BMI ≤18 vs. > 18 (p = 0.519) were not correlated with bleeding events. None had thrombotic event while on dalteparin.

For those with anti-Xa monitoring, patients with bleeding had significantly higher median anti-Xa trough (0.26 vs. 0.13 U/mL, p < 0.01). The same was not found in median anti-Xa peak (0.39 vs. 0.36 U/mL, p = 0.436). Patients who died within 3 months of dalteparin initiation also had higher median anti-Xa trough, although this was not significant (0.14 vs. 0.23 U/mL, p = 0.171).

Multivariate and Statistical Analyses

Anti-Xa trough level ≥0.25 U/mL was associated with bleeding risk on unadjusted analysis (HR = 4.34, 95% CI 1.38–14.38; p < 0.01) and remained significant after multivariable adjustment for age, BMI, creatinine clearance, peak anti-Xa, and sex (HR = 6.18, 95% CI: 1.39–35.1; p < 0.05) (Fig. 2). Dialysis was also a factor associated with bleeding (HR = 6.29, 95% CI: 1.63–31.0; p < 0.05). The ROC-AUC for anti-Xa’s association with bleeding was 0.79. In contrast, anti-Xa trough level ≥ 0.25 U/mL was not associated with mortality in unadjusted (HR = 2.65, 95% CI 0.78–9.13; p = 0.088) or adjusted models (HR = 0.99, 95% CI 0.63–1.56; p = 0.955).

Fig. 2.

Multivariable associations of anti-Xa measurements with bleeding and mortality in patients with kidney impairment receiving dalteparin. Forest plots show adjusted hazard ratios for bleeding (a) and 3 month all-cause mortality (b). Receiver operating characteristic curves for prediction of bleeding (c) and mortality (d). e Predicted bleeding probability by anti-Xa trough level per 0.1IU/mL increase.

Youden’s J statistic was utilised to explore if there was an optimal cutoff point of anti-Xa trough that correlated with bleeding. It determined a cutoff of 0.21 U/mL (Fig. 3). However, this had moderate ROC-AUC performance of 0.71 (Youden’s J Index: 0.43, sensitivity: 0.74; specificity: 0.66).

Fig. 3.

Youden’s J-derived cutoff for anti-Xa trough and risk of bleeding.

Discussion

In this real-world analysis, we observed a 75.7% rate of anti-Xa monitoring in those with kidney impairment. A substantial number of individuals within our institute therefore did not receive anti-Xa monitoring in line with protocol. The majority of individuals with anti-Xa monitoring had trough values within range. Additionally, we report a low rate of thrombotic events. This suggests that our local dalteparin prescribing protocol is effectively prescribed in this population. Our findings suggest that those with kidney impairment may benefit from anti-Xa monitoring, and as such, further efforts should be undertaken to encourage this.

We report a median time to accumulation of 19 days. Patients receiving dalteparin for longer durations, particularly beyond this threshold, may be at increased risk of accumulation-related bleeding. As such, clinicians should consider repeat trough anti-Xa monitoring beyond 2 weeks of therapy, especially in high-risk individuals such as those on dialysis or with fluctuating kidney function, to guide ongoing dosing and minimise bleeding complications.

Our analysis suggests that higher trough anti-Xa levels are independently associated with major bleeding but not 3-month mortality. Our findings align with similar pharmacokinetic studies showing LMWH accumulation with impaired kidney function [14, 16, 17]. A meta-analysis exploring risk of bleeding of LMWH use (predominantly enoxaparin) in a similar patient population found that elevated anti-Xa levels peak increased the risk of major bleeding [17]. However, several studies exploring LMWH and bleeding outcomes found no relationship [26, 27]. One study on dalteparin (n = 445) in a kidney-impaired population similarly did not correlate anti-Xa peak with bleeding, thrombosis, and mortality [18]. There is therefore a degree of uncertainty on the correlation between bleeding risk and anti-Xa peak levels in LMWH use in this population specifically.

In this report, a Youden’s J-derived anti-Xa trough cutoff of 0.21 U/mL (AUC: 0.71) was identified. This indicates that even modestly elevated trough levels within range are associated with increased bleeding risk. The finding aligns with the multivariable Cox findings, which demonstrated a significant association between higher trough levels and bleeding risk. Patients with CKD or on dialysis are predisposed to an elevated baseline bleeding risk due to heterogeneous and poorly understood pathophysiological mechanisms, with anticoagulation potentially exacerbating this risk [6–9]. However, the moderate ROC-AUC (0.70) of the multivariable model and the Youden’s J-derived anti-Xa trough cutoff (AUC: 0.71) underscores that anti-Xa trough levels alone are not a robust association of bleeding. It highlights the influence of other factors that contribute to bleeding not captured in our model. We therefore cannot recommend this cutoff value for use in clinical practice given the low statistical power. Instead, clinicians may consider closer monitoring or dose adjustments for patients with trough levels ≥0.21 U/mL, particularly those high-risk.

Multidisciplinary approaches have been advocated to optimise anticoagulation in kidney impairment [28], and national guidance underscores the importance of selecting the correct kidney function estimate when dosing anticoagulants [20]. Our results suggest that, alongside careful dose selection using Cockcroft-Gault, trough anti-Xa monitoring can refine bleeding risk stratification.

Limitations include the single-centre, retrospective design. This limited our sample size. Consequently, our multivariate model was built on a small sample size and limited bleeding events; model overfitting therefore cannot be excluded. We also did not capture comorbid conditions, CKD aetiology, or albuminuria/proteinuria beyond dialysis status and LMWH indication. These are important potential confounders. Consequently, residual confounding may remain and should be addressed in future analyses. Prospective studies comparing trough anti-Xa-guided dosing are needed to refine guidelines.

Conclusion

Trough anti-Xa measurement of dalteparin therapy is shown to be independently associated with bleeding in patients with CKD. However, the moderate statistical power suggests that other confounding factors may contribute to bleeding in this high-risk group. Incorporating routine anti-Xa monitoring may enhance safety by guiding dose adjustments in this population. However, larger cohort analyses are needed to refine its role in clinical use.

Statement of Ethics

Ethics approval was not required as patients are consented with institutional consent to anonymised data use in service and outcome evaluations. This report was registered as a clinical audit with Manchester University NHS Foundation Trust Clinical Audit Department.

Conflict of Interest Statement

The authors declare no conflicts of interest.

Funding Sources

The authors declare that no funding was received for this manuscript.

Author Contributions

G.G., R.H., M.L., and N.A.: data collection, data analysis, and writing – original draft. J.K.: data analysis and writing – review and editing. Y.B.: writing – review and editing. K.P. and J.T.: writing – review and editing and conceptualisation.

Funding Statement

The authors declare that no funding was received for this manuscript.

Data Availability Statement

All relevant data are supplied within the manuscript and its accompanying online supplementary material.