The main findings of this study are: (i) citrate-based anticoagulation does not impact thrombin generation, platelet function or fibrinolysis during CRRT; (ii) critically ill patients with severe AKI are hypercoagulable with laboratory evidence of coagulopathies that were not known to the clinical team, and (iii) there was no significant difference in the coagulation results between systemic blood samples and samples drawn directly from the circuit (Fig. 4).
These results are important for clinical practice for the following reasons: filter life during citrate-based CRRT can be variable even when post-filter [Ca2+] is in target range. It is reassuring to know that this is not due to effects of citrate on thrombin generation or platelet function or inhibition of fibrinolysis. The fact that there was no significant difference between systemic and circuit samples provides re-assurance, too. Our data confirm that critical illness associated hypercoagulability is very common.
Despite the fact that we excluded all patients with a known bleeding or clotting diathesis and also patients with premature filter clotting, we found markedly raised factor VIII concentrations and VWF in all patients. TAT and prothrombin fragment 1.2 were increased in the majority of patients. Most patients also had reduced levels of protein C which most likely reflects the severity of critical illness rather than hereditary deficiencies. Some patients had positive lupus anticoagulant and mildly raised anticardiolipin and anti-β2 glycoprotein I antibody levels. Without checking for persistence of the antibodies, it is not possible to suggest the possibility of pre-event antiphospholipid syndrome. Two of five patients had low classic APCR ratio and only slightly reduced Modified APCR ratio, which reflects acquired APCR, most likely due to the grossly elevated FVIII levels and/or the lupus anticoagulants. One patient had reduced Classic APCR ratio accompanied by markedly reduced Modified APCR ratio, indicating hereditary APCR, although genetic analysis for FV Leiden, the most common F5 variant conferring APCR, was not performed [19].
Thrombin, a key protein in the regulation of haemostatic processes, has both pro- and anticoagulant properties. TGAs evaluate thrombin generation and decay. The peak represents the highest thrombin concentration that can be generated, and the time to reach the peak represents the velocity of thrombin generation. In contrast, PT and APTT indicate whether there is a coagulation deficiency of one or more procoagulant factors, but not whether this deficiency is counterbalanced by a concomitant deficiency of anticoagulant factors. Our results showed that pre-CRRT, thrombin generation was significantly higher and faster in patients than in healthy volunteers, supporting the concept of critical illness associated hypercoagulability. Data on TGA in critically ill patients have previously been reported in the context of liver failure, trauma, sepsis, severe burns and extracorporeal membrane oxygenation, but to the best of our knowledge, the report by Wiegele et al. and our data are the first in patients with severe AKI requiring RRT [18, 20,21,22].
During CRRT with RCA, there was a normalisation of INR and APTTr from baseline but no significant change in D-dimer, fibrinogen, and peak and time to peak thrombin generation. It is unlikely that the normalisation of INR and APTTr reflects increased activation of coagulation as it would have been accompanied by a decrease in time to peak, an increase in peak thrombin, and possibly further D-dimer elevation.
To test platelet function, we used the PFA-100 analyser which is based on the property of platelets to adhere to collagen via VWF under conditions of shear stress and to aggregate in the presence of agonists [23]. The method involves the drawing of citrate anticoagulated blood through a narrow bore capillary that has at its end a collagen-coated membrane in which a defined microscopic aperture (147 μm) is present. Platelets adhere to the collagen which is infused with either adenosine diphosphate (ADP) or epinephrine to stimulate aggregation. A platelet clot occurs because of shear stress and agonists. The time taken by platelets to occlude the orifice and to block the flow is defined as the closure time (CT). During the 72-h period on CRRT, we observed no significant change in CT using ADP or collagen/epinephrine. We also did not detect any significant changes in systemic or intra-circuit D-dimer concentrations which suggests that fibrinolysis was not affected by citrate either.
This study adds to the existing data on the mechanisms of citrate and complements the findings by Wiegele et al. who studied 24 critically ill surgical patients and showed that thrombin generation did not change during citrate-based CVVHD [18]. Using multiple electrode aggregometry, they also demonstrated decreased platelet function at baseline and during CVVHD but citrate had no impact. A different study focused on the mechanisms of early filter clotting and compared the effects of citrate, heparin, and no anticoagulation strategies on TAT, activated protein C-protein C inhibitor (APC-PCI), and PAI-1 [14]. It showed that in case of early filter failure (< 24 h), inlet concentrations of TAT and APC-PCI were higher, irrespective of anticoagulation. In the heparin group, there was more production of APC-PCI and platelet-derived PAI-1 in the filter after 10 min than in patients who received citrate. Another study explored coagulation parameters in critically ill patients with AKI receiving CRRT with heparin or RCA [24]. Patients with active bleeding had RCA, whereas those without bleeding received heparin anticoagulation. Pre-existing coagulopathy or bleeding disorder was not an exclusion criterion. The study demonstrated no changes in platelets, TAT complexes, beta-thromboglobulin, and VWF during RCA.
An important strength of our analysis is the application of strict eligibility criteria to minimise the impact of confounding factors and the exclusion of participants who needed blood products, had CRRT for less than 48 h or developed premature filter clotting whilst in the study. This allowed us to investigate the direct impact of citrate as best as possible. In addition, citrate-based CVVHD was delivered by an experienced clinical team according to an established protocol [13]. Finally, the analyses were undertaken in a Centre for Haemostasis and Thrombosis at a tertiary care centre using established techniques. The data will serve to underpin future research studies and quality improvement projects in critically ill patients [25].
It is important to also acknowledge some potential limitations. Firstly, our patient cohort was heterogenous despite strict eligibility criteria. Although we excluded all patients with known pro- or anti-thrombotic conditions, we acknowledge that thrombin generation at baseline pre-CRRT ranged from 7.8 nM to 1255.17 nM, all patients had factor VIII levels above the reference range, and protein C concentrations were below the reference range in the majority of patients. This is likely a reflection of critical illness and multi-organ failure, illustrating the under-recognised prevalence of hypercoagulability in critically ill patients with AKI. However, we only included patients with AKI who needed CRRT and cannot comment on coagulation abnormalities in critically ill patients without AKI. Second, TGA still lacks defined reference values [18]. Consequently, reference ranges for TGA parameters specific to the reagents/analyser/analytical protocol were locally derived from healthy volunteer donors to which patient results were compared. Third, the TGA failed to return a result in 6.8% of samples (8/117). Similarly, the PFA failed to return a result in 26.5% of samples (62/234) due to the test exceeding the maximum permitted run time (300 s). This may be related to the fact that several patients developed thrombocytopenia during the study (4/11) which returns elevated PFA-100 CTs purely as a function of reduced platelet numbers and not necessarily altered function [26]. Although thrombocytopenia was an exclusion criterion, it can obviously develop during the course of an illness. In our study, there was a significantly lower platelet count in those tests that timed out compared to those that returned a result (88 vs. 216, p < 0.0001). Fourth, all samples for haemostasis testing were necessarily taken in vacutainers containing 3.2% citrate which leads to citrate concentrations 10.9–12.9 mmol/L and a fall in [Ca2+] in the vacutainer to ~ 0. Where analytically necessary, haemostasis assay design principles incorporated replenishment of [Ca2+] with 25 mmol/L CaCl2 in coagulation screening tests and 15 mmol/L CaCl2 in TGA. Consequently, INR, APTT and TGA results were possibly confounded by variations in the final [Ca2+]. This was not an issue for specific-component assays that employ calcium replenishment as they were performed to merely assess levels of each parameter and not effects of citrate anticoagulation on their function. Calcium chloride is not replenished in PFA-100 analysis. The citrate anticoagulated samples would therefore not assess any direct effects of citrate in the circulation, but reflect changes in primary haemostasis resulting from the CRRT treatment itself. Fifth, we acknowledge that D-dimer and PFA-100 tests do not reflect the full spectrum of fibrinolysis or platelet function. Global fibrinolysis assays, such as euglobulin clot lysis and dilute clot lysis were not available. Performing specific-component assays like tissue plasminogen activator (t-PA), urokinase-type plasminogen activator (u-PA), plasminogen activator inhibitor-1 (PAI-1), plasminogen, thrombin-activatable fibrinolysis inhibitor (TAFI) or histidine-rich glycoprotein (HRG) was beyond the resources available. However, markers such as t-PA/PAI-1 complexes and plasmin/antiplasmin complexes give broadly the same information as D-dimer. Also, we used widely available assays that could have been applied in this context if our results had suggested that monitoring citrate anticoagulation with them could be valuable. Lastly, the study took 2.5 years to complete and not all planned investigations were performed in all patients [17]. This reflects our very tight inclusion criteria and the challenges of studying coagulation biochemistry in critically ill patients.
Our results have implications for clinical practice and future research. More in-depth investigations of patients with repeated episodes of premature filter clotting are warranted to identify potential contributing factors and to rule out an underlying acquired coagulopathy. This is particularly relevant since filter life has been proposed as a potential marker for quality assessment and performance measurement during CRRT [25]. Finally, the fact that all patients exhibited raised thrombin generation potential should prompt research studies exploring the role of thrombin inhibitors in CRRT.