Abstract
Background
Thrombocytopenia is a clinically relevant finding, particularly in patients with critical illness. 1 While it is unclear if this impacts mortality, other downstream treatment effects occur in thrombocytopenic patients. 1 These patients are more likely to receive platelet transfusions and have anticoagulation held due to a presumed increased risk of bleeding. 1 Critically ill patients have numerous risk factors that may predispose them to the development of thrombocytopenia including organ dysfunction, sepsis, renal failure, and medication exposure.1,2 Given the multitude of risk factors, thrombocytopenia is observed in various intensive care unit (ICU) patient populations. Additional confounders in cardiovascular ICUs include cardiogenic shock and mechanical circulatory support with a reported incidence near 20% in patients admitted to these units.3,4 One medication cause that is frequently explored is heparin-induced thrombocytopenia (HIT).5,6 This work-up includes testing for heparin-platelet factor 4 complex antibodies and changing to a non-heparinoid anticoagulant such as a direct thrombin inhibitor.
Another important factor associated with thrombocytopenia is continuous renal replacement therapy (CRRT). In recent studies anywhere from 20%–50% of patients exposed to CRRT developed thrombocytopenia depending on the definition used.2,5,7,8 Thrombocytopenia associated with CRRT appears to occur within five days of initiation and platelet counts recover after cessation.2,9 The mechanism of thrombocytopenia with CRRT is likely multifactorial including loss of platelets via hemofiltration and via platelet activation and consumption.10,11 Various CRRT modalities can be utilized for both solute clearance and volume removal. The impact of different CRRT modalities or systems on the incidence of thrombocytopenia is unknown. Existing case reports have implicated NxStage hemodialysis systems specifically as an etiology for thrombocytopenia, and others have suggested differing membranes and sterilization techniques may impact development of thrombocytopenia.12-14
To address this research gap, this study was designed to compare the incidence of thrombocytopenia when using continuous venovenous hemodialysis (CVVHD) with the NxStage system to continuous venovenous hemodiafiltration (CVVHDF) with the Prismaflex system in a cardiac ICU. We hypothesized that more patients treated with the NxStage system would develop thrombocytopenia.
Methods
Study Design and Participants
This was a single-center retrospective cohort study approved by The Ohio State University Institutional Review Board on December 12, 2022 (project number 2022H0421). The project was designed and conducted in accordance with the amended Declaration of Helsinki. The electronic health record (EHR) was queried to identify patients between June 1, 2016 and September 30, 2022 who had orders for continuous renal replacement therapy and were admitted to the cardiovascular ICU within a large academic medical center. In June 2020 the institution switched from using CVVHDF with the Prismaflex system using the M100 hemofilter set and AN 69 membrane to CVVHD with the NxStage system for CRRT.
Patients had to receive CRRT for at least 48 hours to be included. Exclusion criteria included mechanical circulatory support (ex. temporary or durable ventricular assist devices, intraaortic balloon pump, extracorporeal membrane oxygenation); surgical operation during admission prior to CRRT; history of transplant or receiving antiproliferative immunosuppression; history of HIT; known history of immune thrombocytopenia, lymphoproliferative disorder, cirrhosis, active chemotherapy treatment or myelodysplastic syndrome; platelet count less than 100 000 platelets per microliter prior to CRRT; positive COVID-19 PCR or antigen test during index admission; active bleeding defined as a hemoglobin drop >2 g/dL or receipt of 2 units pRBC in a 24-hour period prior to or at the time of developing thrombocytopenia; and patients treated with both NxStage and Prismaflex CRRT systems. For patients who received CRRT multiple times during the study period, only the first therapy was analyzed. Select medication use prior to or during CRRT was collected including heparin infusions. All heparin infusions followed a nurse driven institutional protocol titrated every 6-12 hours to a goal aPTT of 72-95 seconds.
Outcomes
The primary outcome was the incidence of thrombocytopenia during CRRT treatment. Thrombocytopenia was defined as a ≥ 50% drop to a platelet count <150 000 per microliter OR a ≥ 30% drop to a platelet count <100 000 per microliter from baseline. Baseline platelet count was defined as the last available platelet count prior to CRRT initiation.
Secondary outcomes included the proportion of patients with thrombocytopenia who had HIT testing ordered, were treated with a direct thrombin inhibitor, and had positive HIT testing results; the proportion of patients with thrombocytopenia at 48 hours after CRRT initiation; the proportion of patients reaching platelet thresholds of <150 000 per microliter, <100 000 per microliter, <50 000 per microliter, or had a 50% platelet count drop while on CRRT. Other outcomes included the difference in the mean change in platelet count from baseline to each of the following timepoints: 24, 48, and 72 hours after CRRT treatment initiation, the end of CRRT treatment, and 48 hours after treatment ended. Finally, the difference in the mean change in platelet count from baseline to platelet nadir was assessed.
Statistical Analysis
Descriptive statistics were used for baseline characteristics, with means and standard deviations for continuous variables and counts and proportions for categorical variables. Between group comparisons were made using students t-test or Wilcoxon rank-sum as appropriate or with Chi-square or Fischer’s exact. For the primary outcome of the development of thrombocytopenia, unadjusted and adjusted odds ratios were constructed using logistic regression models. Covariates determined a priori for adjustment were time on CRRT, baseline platelet count, SOFA score, treatment dose heparin use, aspirin use, and P2Y12 inhibitor use. Assuming a 25% incidence rate of thrombocytopenia in patients treated with the Prismaflex system, 120 patients (60 in each arm) would provide 80% power with an alpha level of 0.05 to detect a 25% difference between groups. Data were collected through retrospective chart review of the electronic medical record and were managed using REDCap (Research Electronic Data Capture) hosted by the Clinical and Translational Science (CCTS) at The Ohio State University. REDCap is a secure, web-based software platform designed to support data capture for research studies. 15
Results
Patient Characteristics
Between June 1, 2016 and September 30, 2022, 876 unique patients were admitted to the cardiovascular ICU and received continuous renal replacement therapy for at least 48 consecutive hours. A total of 729 patients were excluded from further analysis (Figure 1). Of the 147 patients included, 86 received CVVHDF with the Prismaflex system, while 61 received CVVHD with the NxStage system. The mean age was 63.7 years and 68% of patients were male (Table 1). Baseline platelet counts and sequential organ failure assessment (SOFA) scores were similar between groups. Patients treated with the NxStage system had higher initial blood flow rates (255 mL/hr vs 200 mL/hr; P < 0.001). While there was similar use of heparin infusions and subcutaneous heparin between groups, more patients treated with the Prismaflex system received aspirin (80.2% vs 52.5%; P < 0.001). Enrollment CVVHDF = continuous venovenous hemodiafiltration, CVVHD = continuous venovenous hemodialysis, HIT = heparin induced thrombocytopenia. Baseline Characteristics. CRRT = continuous renal replacement therapy, CVVHDF = continuous venovenous hemodiafiltration, CVVHD = continuous venovenous hemodialysis, SD = standard deviation, Hgb = hemoglobin, SCr = serum creatinine, SOFA = sequential organ failure assessment score.
Baseline platelet count defined as the last available platelet count prior to CRRT initiation.
Active bleeding was defined as a hemoglobin drop greater than 2 g/dL or receipt of 2 units packed red blood cells in a 24-hour period prior to or at the time of developing thrombocytopenia.
Primary Outcome: Thrombocytopenia.
Thrombocytopenia was defined as a ≥ 50% drop platelet drop to a platelet count <150 K/μL OR a ≥ 30% drop to a platelet count <100 K/μL from baseline. Baseline platelet count was defined as the last available platelet count prior to CRRT initiation.
Primary and Secondary Outcomes
Thrombocytopenia at 48 hours and Platelet Counts After CRRT Initiation.
CRRT = continuous renal replacement therapy, CVVHDF = continuous venovenous hemodiafiltration, CVVHD = continuous venovenous hemodialysis, OR = odds ratio, SD = standard deviation.
Thrombocytopenia was defined as a ≥ 50% drop platelet drop to a platelet count <150 K/μL OR a ≥ 30% drop to a platelet count <100 K/μL from baseline. Baseline platelet count was defined as the last available platelet count prior to CRRT initiation.
Platelet count over time CRRT = continuous renal replacement therapy. Platelet counts expressed in thousandths per microliter.
HIT Testing, DTI Utilization, and Transfusions.
CVVHDF = continuous venovenous hemodiafiltration, CVVHD = continuous venovenous hemodialysis, HIT = heparin induced thrombocytopenia, DTI = direct thrombin inhibitor, SD = standard deviation.
HIT positivity was defined as either heparin platelet factor 4 antibody >2.00 or a positive serotonin release assay.
Discussion
In this single center retrospective cohort study, a larger proportion of patients treated with CVVHD using the NxStage CRRT system as compared to CVVHDF with the Prismaflex system developed thrombocytopenia. This led to many important downstream effects such as increased HIT testing, direct thrombin inhibitor use, and blood transfusions. Collectively these results suggest management strategies for, and outcomes associated with, thrombocytopenia were similar regardless of CRRT modality; however, the roughly 3-fold higher incidence of thrombocytopenia in the NxStage/CVVHD group translated to an analogously higher rate of each of these measures when compared to Prismaflex/CVVHDF. The time to thrombocytopenia was not different between groups; however, patients treated with the NxStage system had a more pronounced decline as evidenced by a lower platelet nadir and more patients reaching prespecified platelet count thresholds of <150 and <100 platelets/microliter.
Previous studies have demonstrated an association between thrombocytopenia and CRRT therapy, with the incidence ranging from 20%–50% depending on how thrombocytopenia is defined.2,5,7 In this study we defined thrombocytopenia as either 1) a ≥ 50% drop to a platelet count <150 000 per microliter, or 2) a ≥ 30% drop to a platelet count <100 000 per microliter from baseline. This definition was chosen as it is the most commonly accepted definition when assessing HIT, and an attempt to not capture patients with mild thrombocytopenia at baseline and a minimal platelet fall, which may occur with simple count-based thresholds alone. 16 Our study population was limited to those receiving care in a cardiovascular ICU, which may explain why we observed a higher rate of thrombocytopenia. In a previous study, the NxStage system was used for CRRT and the incidence of thrombocytopenia was 22.5% in cardiovascular ICU patients. 4 This is lower than the 57.9% incidence seen in our study which was likely due to the different definition of thrombocytopenia used along with a 72-hour cutoff from CRRT initiation. 4
Despite numerous trials characterizing the association of thrombocytopenia with CRRT, the precise etiology remains unknown.2,5,7 A small observational study by Mulder and colleagues demonstrated reduction in platelet counts across the hemofilter. 10 Further, there appeared to be an association with the blood flow rate and degree of platelet loss across the filter, which aligns with our current study, as patients treated with the NxStage system had higher initial blood flow rates. Previous research has suggested the association is multifactorial though with other components such as consumption and the sterilization techniques contributing to thrombocytopenia.11,12,14 A case report by Parmar outlined the possibility of thrombocytopenia induced by the sterilization technique in a patient receiving conventional hemodialysis with the NxStage system. 14 Standard NxStage system dialyzers consist of gamma sterilized polyethersulfone membranes which have been hypothesized to lead to increased platelet activation.14,17,18 A larger proportion of patients treated with the Prismaflex system in our study received aspirin (80.2% vs 52.5%). Among all patients treated with aspirin (n = 101), 22.8% developed thrombocytopenia, while among those not exposed to aspirin (n = 46), 63.0% developed thrombocytopenia. While this difference was not observed with P2Y12 inhibitors, it remains intriguing whether antiplatelet medications may have any utility in preventing CRRT-associated platelet loss. Although patients in our study treated with the NxStage system more often developed thrombocytopenia, it must be recognized that different modalities (CVVHDF vs CVVHD) were being utilized as well. It seems unlikely that CVVHD, which removes waste via diffusion and does not utilize replacement fluid, would contribute to a difference in thrombocytopenia; however, this possibility cannot be excluded.
Determining the cause of thrombocytopenia in critically ill patients is complex given there are often multiple contributing factors. Studies have shown that heparin-induced thrombocytopenia is often explored as a possible cause.5,6 HIT testing was performed in 20 patients (13.6%) with a statistically significant difference between groups in the overall population (5.8% vs 24.6%). An important finding of our trial was that despite this testing, no patients were found to be HIT positive. Our study was conducted exclusively in a cardiac intensive care unit with 72.8% of patients treated with a heparin infusion; therefore, the amount of testing is not surprising as this potential etiology was being ruled out. Despite this, our results highlight the importance of appropriately weighing CRRT as a risk factor when assessing a patient’s HIT probability score. Inappropriate HIT testing leads to not only increased health care expenditure, but also increased exposure to non-heparinoid anticoagulants as demonstrated in our study. While not assessed in our study, this may lead to increased periods of subtherapeutic anticoagulation given the necessity to switch between agents.
Our study has important limitations, some of which are inherent to its retrospective design. As mentioned above, 2 different modalities were compared in addition to different CRRT systems therefore we cannot evaluate the extent to which each of these contributed to the observed results. Additionally, the study looked at a pre- and post- institutional change regarding CRRT system; therefore, the 2 groups consisted of patients in 2 different time periods. Given that the change occurred in 2020, only the NxStage cohort were included throughout the COVID-19 pandemic. To minimize potential confounding, we excluded patients who tested positive for COVID-19 at any point of the index admission. Potential confounders that we were unable to control include glycoprotein IIb/IIIa inhibitor use, heparin dosing including assessment of activated partial thromboplastin time (aPTT) levels. Another consideration is that while overall CRRT interruptions for filter changes or clotting were similar between groups, the exact cause for interruption, amount of blood returned to the patient and timing of changes relative to development of thrombocytopenia were unable to be captured.
Conclusions
While the etiology of CRRT-induced thrombocytopenia remains unknown, a greater proportion of cardiovascular ICU patients developed thrombocytopenia when treated with CVVHD using the NxStage system compared to being treated with CVVHDF using the Prismaflex system. This difference was associated with increased HIT testing, direct thrombin inhibitor use, and blood transfusion. Institutions may need to consider these outcomes collectively with other factors when evaluating changes to CRRT modalities. Future studies assessing differences in CRRT modalities and systems may help elucidate contributing factors of CRRT-induced thrombocytopenia.
Footnotes
Acknowledgements
This publication was supported, in part, by the National Center for Advancing Translational Sciences of the National Institutes of Health under Grant Number UL1TR002733. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health under Grant Number (UL1TR002733).
