Abstract
Objective:
Thrombocytopenia is common in critically ill patients treated with continuous renal replacement therapy (CRRT) and decreases in platelets following CRRT initiation have been associated with increased mortality. Platelets play a role in innate and adaptive immunity, making it plausible that decreases in platelets following CRRT initiation predispose patients to development of infection. Our objective was to determine if greater decreases in platelets following CRRT correlate with increased rates of secondary infection.
Design:
Retrospective cohort analysis
Setting:
This study utilizes a CRRT database from Mayo Clinic (Rochester), a tertiary academic center.
Participants:
Adult patients who survived until ICU discharge and were on CRRT for <30 days were included. A sub-group analysis was also performed in patients with thrombocytopenia (platelets <100×103/μL) at CRRT initiation.
Measurements and Main Results:
The primary predictor variable was a decrease in platelets from pre-CRRT levels of >40% or ≤40%, although multiple cut-points were analyzed. The primary outcome was infection after ICU discharge, and secondary endpoints included post-ICU septic shock and post-ICU mortality. Univariable, multivariable, and propensity-adjusted analyses were used to determine associations between the predictor variable and outcomes.
Results:
Among 797 eligible patients, 253 had thrombocytopenia at CRRT initiation. A >40% decrease in platelets after CRRT initiation was associated in the multivariable-adjusted models with increased odds of post-ICU infection in the full cohort (OR 1.49, CI 1.02–2.16) and in the thrombocytopenia cohort (OR 2.63, CI 1.35–5.15) cohorts.
Conclusion:
Platelet count drop by > 40% following CRRT initiation is associated with an increased risk of secondary infection, particularly in patients with thrombocytopenia at the time of CRRT initiation. Further research is needed to evaluate the impact of both CRRT and platelet loss on subsequent infection risk.
Keywords: Thrombocytopenia, Acute Kidney Injury, Continuous Renal Replacement Therapy, Sepsis, Secondary Infection, Platelets
Introduction
Severe acute kidney injury (AKI) requiring renal replacement therapy (AKI-RRT) is common in critically ill patients in the intensive care unit (ICU), complicating 13.5% of all ICU cases[1]. Continuous renal replacement therapy (CRRT) is commonly used in the ICU setting because it is associated with less hemodynamic instability than intermittent hemodialysis[2]. AKI requiring CRRT has an in-hospital mortality rate >50%, making it one of the deadliest conditions commonly seen in US hospitals[3, 4].
Sepsis and infection are the leading causes of death of patients with AKI [5]. Both are also common conditions requiring CRRT, and also complications of CRRT. In an analysis of the program to improve care in acute renal disease (PICARD) study, 54% of patients who were initiated on CRRT for a reason other than sepsis went on to develop a secondary infection, which doubled the risk of mortality[6]. Sepsis and AKI-CRRT are now thought to be bidirectional[7], with relative immunosuppression as a proposed mechanism of sepsis following AKI[8]. One possible cause of immunosuppression in this patient population is immunosuppression related to loss of platelets during CRRT. Platelets have many roles beyond thrombus formation and are important to both innate and adaptive immunity[9]. It is plausible that thrombocytopenia in the critically ill is a marker of platelet dysregulation which can directly contribute to increased infection, and that further platelet losses temporally related to CRRT can exacerbate the risk.
RRT initiation has been associated with worsening thrombocytopenia in multiple studies. A secondary analysis of the Acute Renal Failure Trial Network (ATN) study showed that nearly two-thirds of RRT patients had some amount of platelet decrease within 72 hours of RRT initiation, with an average platelet decrease of 33%[10]. A platelet decrease following RRT initiation was associated with increased mortality as well as a lack of renal recovery in survivors.
It is possible that the loss of platelets, which have a role in innate and adaptive immunity, may predispose to subsequent infection. In cardiac surgery patients who have undergone cardiopulmonary bypass, another membrane-based extracorporeal therapy, platelet loss occurs in 90% of patients following cardiac surgery, and nearly half of patients have a platelet nadir <100×103/μL. The degree of platelet decrease from baseline, as well as the absolute platelet nadir, have both been shown to correlate with postoperative infection as well as mortality[11].
Our overall hypothesis is that platelet decreases following CRRT initiation will be associated with subsequent infection even after adjustments for demographic data, comorbidities, and illness severity.
Methods
Study population:
After obtaining local institutional review board approval for this retrospective cohort analysis, a de-identified CRRT database from the Mayo Clinic at Rochester, MN, was obtained. Patients in the database had previously provided authorization to have their data stored in the database and used for analysis. The database contained adult patients with severe AKI or end stage kidney disease (ESKD) who were treated with CRRT between 2007 and 2015. The standard prescription at Mayo is CVVH with 50% pre-filter and 50% post-filter fluid replacement, with citrate, at a blood flow rate of 200 mL/hr. Patients were excluded if they did not survive until ICU discharge, spent more than 30 days on CRRT, or had extreme outliers in body mass index (BMI) (nine patients had BMI values < 16 or >150). Patients were also excluded if the ICU platelet nadir did not occur while on CRRT. A small number of cases had a platelet value at CRRT initiation that was below the overall nadir for the ICU stay, likely indicating a data entry error, and were excluded (Figure 1). Patients with a platelet count <100×103/μL were pre-specified for subgroup analysis considering that patients with reduced platelets at the time of CRRT initiation are at higher risk for mortality[10].
Figure 1.
Patient exclusion diagram
Abbreviations: AKI; acute kidney injury, BMI; body mass index, CRRT; continuous renal replacement therapy, ESKD; end-stage kidney disease, ICU; intensive care unit
Predictor:
The percentage difference in platelets between the CRRT initiation value and the platelet nadir value was the predictor variable. Cut-points of 30% - 60% by tens were evaluated to determine the point at which the difference between unadjusted rates of secondary infection was greatest between the above and below cut-point groups, which was 40%. For the purposes of Table 1, patients were therefore categorized based on a percent change in platelet count of ≤40% vs. those with >40% change; however, given uncertainty about the most clinically appropriate endpoint, platelet change of >30%, >40%, >50%, >60%, and platelet change as a continuous variable were all analyzed as predictors in separate models.
Table 1.
Comparison of baseline characteristics of 797 patients on CRRT who met inclusion criteria and survived until ICU discharge.
| Characteristics | Platelet Change ≤40% (n=288) | Platelet Change >40% (n=509) | |
|---|---|---|---|
| N (%)* | N (%)* | P value† | |
|
| |||
| Age, years, mean (SD) | 61.8 (15.7) | 61.2 (14.0) | .61 |
|
| |||
| Female | 104 (36.1) | 209 (41.1) | .17 |
|
| |||
| Body mass index, mean (SD) | 32.4 (8.8) | 30.4 (9.6) | .005 |
|
| |||
| Comorbidities | |||
| Myocardial Infarction | 40 (13.9) | 54 (10.6) | .17 |
| Diabetes Mellitus | 94 (32.6) | 128 (25.1) | .02 |
| Mod. or severe Renal disease | 93 (32.3) | 134 (26.3) | .07 |
| Chronic pulmonary disease | 56 (19.4) | 79 (15.5) | .16 |
| Cirrhotic liver disease | 20 (6.9) | 50 (9.8) | .17 |
| Congestive heart failure | 55 (19.1) | 74 (14.5) | .09 |
| Malignancy | 52 (18.1) | 119 (23.4) | .08 |
| Peripheral vascular disease | 12 (4.2) | 18 (3.5) | .65 |
| Cerebrovascular disease | 25 (8.7) | 28 (5.5) | .08 |
| Kidney Transplant | 23 (8.0) | 47 (9.2) | .55 |
|
| |||
| Charlson Comorbidity Index, mean (SD) | 5.4 (3.2) | 5.2 (3.0) | .35 |
|
| |||
| Platelets at ICU admission, mean (SD) | 184 (100) | 180 (115) | .62 |
|
| |||
| Platelets at CRRT initiation, mean (SD) | 156 (95) | 167 (107) | .17 |
|
| |||
| Platelet nadir, mean (SD) | 119 (73) | 55 (44) | <.001 |
|
| |||
| INR at CRRT initiation, mean (SD) | 1.7 (0.7) | 1.8 (1.0) | .06 |
|
| |||
| Albumin at CRRT initiation, mean (SD) | 2.9 (0.6) | 3.1 (0.8) | .02 |
|
| |||
| Hemoglobin at CRRT initiation, mean (SD) | 9.9 (1.9) | 9.8 (1.7) | .59 |
|
| |||
| WBC at CRRT initiation, mean (SD) | 14.3 (20.5) | 14.8 (10.8) | .29 |
|
| |||
| SOFA Score at CRRT initiation, mean (SD) | 10.0 (3.3) | 11.8 (3.3) | <.001 |
|
| |||
| Number of days on CRRT, mean (SD) | 4.2 (3.9) | 7.3 (5.5) | <.001 |
|
| |||
| ICU day of CRRT initiation, mean (SD) | 1.1 (1.5) | 1.3 (2.7) | .30 |
|
| |||
| CRRT end to ICU discharge, days, mean (SD) | 3.6 (6.6) | 5.9 (10.5) | .001 |
|
| |||
| CRRT Initiation to Platelet Nadir, days, mean (SD) | 3.0 (3.2) | 5.0 (5.6) | <.001 |
|
| |||
| Ventilator Days, mean (SD) | 4.9 (7.0) | 9.1 (12.4) | <.001 |
|
| |||
| Citrate Use | 271 (94.1) | 479 (94.1) | .99 |
|
| |||
| Average Delivered Day 1 CRRT dose, mean (SD), mL/kg/hr | 19.7 (9.1) | 19.3 (8.6) | .60 |
Abbreviations: CRRT, continuous renal replacement therapy; INR, international normalization ratio; WBC, white blood count; SOFA, sequential organ failure assessment
P-values less than 0.05 are bolded.
Outcomes:
The primary outcome was secondary infection following ICU discharge when CRRT was no longer being used. Since sepsis is a common cause of AKI[7], we chose to use only new infections that were diagnosed after discharge from the ICU to ensure that CRRT initiation preceded infection rather than vice-versa. This approach also allowed us to include patients on CRRT due to sepsis-associated AKI, since the new, post-ICU infections could be clearly delineated from the initial infection. Infection in the post-ICU period was defined as a new antibiotic prescription. A stricter endpoint evaluated as a secondary outcomes was post-ICU development of septic shock, defined as meeting at least 2 SIRS criteria in the setting of suspected infection, with hypotension that was refractory to fluid resuscitation or with hyperlactatemia[12]. Finally, post-ICU mortality, which was defined as surviving until ICU discharge, but not to hospital discharge, was evaluated as an additional secondary endpoint.
Covariates:
Covariates in the fully adjusted models included age, gender, body mass index (BMI), sequential organ failure assessment (SOFA) score on the day of CRRT initiation, number of days on mechanical ventilation, Charlson Comorbidity Index, history of kidney transplant, baseline platelet value at CRRT initiation, number of days on CRRT, ICU day of CRRT initiation, time from CRRT initiation to platelet nadir, and time from CRRT discontinuation to ICU discharge.
Statistical Analysis:
Mean ± standard deviation or counts and percentages were used to describe the distribution of continuous and categorical variables, respectively. ANOVA, Kruskal Wallis, or Chi-Square tests were used as appropriate across categories.
Patients were categorized based on a percent change in platelets of ≤40% or >40%, which was the point between 30% and 60% by tens that gave the greatest difference between unadjusted secondary infection rates, as outlined above. However, all cut-points 30–60% and change in platelets a continuous variable were analyzed separately. The unadjusted analysis was performed initially, followed by the addition of the covariates listed above into a multivariable model. Because there were more covariates than could be supported by the number of events, especially for secondary outcomes, we used Akaike information criterion (AIC) to choose the best model with a maximum of 8 covariates included. Finally, a propensity-adjusted analysis was performed. A propensity score for each patient was calculated using a multivariable logistic regression model in which the dependent variable was a percent change in platelets for the cut-off values used in the model, and the independent variables were the covariates. For the propensity model, the β-coefficients were combined with the patients’ values for each covariate to generate individual propensity scores. All covariates listed were included in the creation of the propensity score. A propensity-adjusted analysis was chosen instead of propensity-score matching to ensure that all eligible patients were included in the analysis.
Results
Patient Characteristics:
Among 1,759 screened patients, 797 patients entered the final analysis after meeting all eligibility criteria (Figure 1). Table 1 shows the baseline characteristics of the full cohort. The mean platelet decrease was 51% overall after CRRT initiation in the full cohort, and 50% in the thrombocytopenia cohort.
Patients in the ≤40% group were similar to those with in the >40% group in terms of age and gender. The overall Charlson Comorbidity Index value was equivalent between groups, although the >40% group had a lower proportion of diabetes mellitus. The >40% group had a higher SOFA score at the time of CRRT initiation, spent more days on CRRT, spent more time on mechanical ventilation, and had lower BMI. Patients with thrombocytopenia at the time of CRRT initiation were then analyzed separately (Supplemental Table S1). Similar to the full cohort, thrombocytopenic patients in the >40% group spent more time on CRRT and mechanical ventilation, had higher SOFA scores at CRRT initiation, and had higher rates of malignancy and congestive heart failure.
Primary and Secondary Outcomes:
Table 2 demonstrates the raw outcomes data for the full cohort of patients with a platelet decrease >40%. For the primary outcome of post-ICU infection, the unadjusted analysis showed an association between platelet decrease 40% and 50% and post-ICU infection. In the multivariable and propensity adjusted models, odds ratios were 1.49 (CI 1.02 – 2.16) and 1.48 (CI 1.03 – 2.14) respectively, with both results achieving statistical significance. The development of post-ICU septic shock was associated with >40% platelet decrease in the multivariable adjusted model (OR 1.55, 95% CI 1.00–2.42), but missed significance in the propensity-adjusted model (p=0.053). Interestingly, Septic shock was associated with a >30% decrease in all models, even though total infection was negative for that cut-point. Post-ICU mortality was not associated with any cut point in any model (Table 3).
Table 2.
Comparison of primary and secondary outcomes in all patients and in patients who had thrombocytopenia at CRRT initiation.
| All Patients (N = 797) | TP at CRRT Initiation (N = 253) | |||||
|---|---|---|---|---|---|---|
|
| ||||||
| Outcomes | Platelet Change ≤40% (n=288) | Platelet Change >40% (n=509) | Platelet Change ≤40% (n=97) | Platelet Change >40% (n=56) | ||
| N (%)* | N (%)* | P-value | N (%)* | N (%)* | P-value | |
|
| ||||||
| Post-ICU Infection | 67 (23.3) | 155 (30.5) | .03 | 20 (20.6) | 59 (37.8) | .005 |
|
| ||||||
| Post-ICU Septic Shock | 40 (13.9) | 103 (20.2) | .02 | 13 (13.4) | 37 (23.7) | .04 |
|
| ||||||
| Post-ICU Mortality | 41 (14.2) | 89 (17.5) | .19 | 12 (12.4) | 34 (21.8) | .004 |
Abbreviations: TP, thrombocytopenia; CRRT, continuous renal replacement therapy; acute kidney injury; OR, odds ratio; CI, confidence interval.
P-values less than 0.05 are bolded.
Table 3.
Unadjusted and adjusted primary and secondary outcomes by percent change in platelets in the full patient cohort of 797 patients on CRRT who survived to ICU discharge.
| Outcome | Continuous | p-value | Platelet decrease >30% | p-value | Platelet decrease >40% | p-value | Platelet decrease >50% | p-value | Platelet decrease >60% | p-value |
|---|---|---|---|---|---|---|---|---|---|---|
| Infection | ||||||||||
|
| ||||||||||
| Unadjusted | 1.006 (0.999–1.013) | .06 | 1.33 (0.91–1.95) | .14 | 1.46 (1.04–2.04) | .03 | 1.39 (1.01–1.90) | .04 | 1.18 (0.87–1.62) | .28 |
| Multivariable Adjusted | 0.997 (0.990–1.004) | .33 | 1.43 (0.94–2.18) | .09 | 1.49 (1.02–2.16) | .04 | 1.41 (0.99–2.01) | .06 | 1.14 (0.80–1.62) | .47 |
| Propensity Adjusted | 1.39 (0.92–2.10) | .12 | 1.48 (1.03–2.14) | .04 | 1.41 (1.00–2.00) | .051 | 1.15 (0.81–1.63) | .43 | ||
|
| ||||||||||
| Septic Shock | ||||||||||
|
| ||||||||||
| Unadjusted | 1.010 (1.002–1.018) | .01 | 1.74 (1.07–2.80) | .02 | 1.61 (1.08–2.41) | .02 | 1.48 (1.02–2.15) | .04 | 1.25 (0.87–1.81) | .22 |
| Multivariable Adjusted | 0.993 (0.984–1.001) | .07 | 1.75 (1.04–2.94) | .03 | 1.55 (1.00–2.42) | .049 | 1.40 (0.93–2.12) | .11 | 1.12 (0.75–1.69) | .58 |
| Propensity Adjusted | 1.74 (1.04–2.91) | .04 | 1.54 (0.99–2.40) | .053 | 1.41 (0.93–2.12) | .10 | 1.13 (0.75–1.69) | .56 | ||
|
| ||||||||||
| Death | ||||||||||
|
| ||||||||||
| Unadjusted | 1.006 (.999–1.013) | .15 | 1.20 (0.76–1.90) | .43 | 1.31 (0.87–1.96) | .19 | 1.26 (0.86–1.85) | .23 | 1.21 (0.83–1.76) | .33 |
| Multivariable Adjusted | 0.996 (0.987–1.005) | .35 | 1.22 (0.73–2.04) | .45 | 1.24 (0.78–1.96) | .36 | 1.18 (.76–1.82) | .46 | 1.12 (0.72–1.72) | .62 |
| Propensity Adjusted | 1.18 (0.71–1.96) | .52 | 1.21 (0.77–1.89) | .41 | 1.14 (0.74–1.74) | .56 | 1.09 (0.71–1.66) | .22 | ||
P-values less than 0.05 are bolded.
Primary and Secondary Outcomes in the Thrombocytopenia Cohort:
For the primary outcome of post-ICU infection, the unadjusted, multivariable-adjusted, and propensity adjusted analyses all showed a statistically significant association between platelet decreases at all cut-points and post-ICU infection. There were statistically significant associations with septic shock in the continuous, 30%, and 50% models, though not in the 40% or 60% models. Similarly, there was higher post-ICU mortality in the continuous, 30%, and 50% models, though not in the 40% or 60% models (Table 4).
Table 4.
Unadjusted and adjusted primary and secondary outcomes by percent change in platelets in patients with a platelet value of <100×103/μL at the time of CRRT initiation, in 253 patients who survived to ICU discharge.
| Outcome | Continuous | p-value | Platelet decrease >30% | p-value | Platelet decrease >40% | p-value | Platelet decrease >50% | p-value | Platelet decrease >60% | p-value |
|---|---|---|---|---|---|---|---|---|---|---|
| Infection | ||||||||||
|
| ||||||||||
| Unadjusted | 1.020 (1.008–1.031) | .006 | 2.27 (1.15–4.45) | .02 | 2.37 (1.30–4.30) | .005 | 2.37 (1.36–4.12) | .002 | 1.87 (1.09–3.20) | .02 |
| Multivariable Adjusted | 1.021 (1.009–1.035) | .001 | 2.81 (1.32–5.98) | .007 | 2.63 (1.35–5.15) | .005 | 2.51 (1.36–4.63) | .003 | 1.84 (1.01–3.34) | .047 |
| Propensity Adjusted | 2.55 (1.24–5.24) | .01 | 2.56 (1.33–4.91) | .005 | 2.49 (1.36–4.56) | .003 | 1.89 (1.06–3.36) | .03 | ||
|
| ||||||||||
| Septic Shock | ||||||||||
|
| ||||||||||
| Unadjusted | 1.019 (1.005–1.033) | .007 | 2.56 (1.09–6.01) | .03 | 2.13 (1.05–4.32) | .04 | 2.21 (1.15–4.25) | .02 | 1.82 (0.98–3.41) | .06 |
| Multivariable Adjusted | 1.021 (1.005–1.036) | .008 | 2.82 (1.12–7.11) | .03 | 2.08 (0.95–4.55) | .07 | 2.25 (1.09–4.63) | .03 | 1.76 (0.87–3.55) | .11 |
| Propensity Adjusted | 2.72 (1.11–6.68) | .03 | 2.11 (0.98–4.53) | .06 | 2.19 (1.08–4.44) | .03 | 1.79 (0.92–3.48) | .09 | ||
|
| ||||||||||
| Death | ||||||||||
|
| ||||||||||
| Unadjusted | 1.019 (1.004–1.033) | .01 | 2.77 (1.12–6.86) | .04 | 2.11 (1.02–4.39) | .004 | 2.38 (1.20–4.71) | .002 | 1.82 (0.96–3.46) | .07 |
| Multivariable Adjusted | 1.023 (1.006–1.040) | .009 | 3.17 (1.11–9.07) | .03 | 1.97 (.85 (4.56) | .11 | 2.51 (1.15–5.48) | .02 | 1.96 (0.93–4.11) | .08 |
| Propensity Adjusted | 3.13 (1.14–8.60) | .03 | 2.01 (0.89–4.51) | .09 | 2.25 (1.07–4.76) | .03 | 1.75 (0.88–3.49) | .11 | ||
P-values less than 0.05 are bolded.
Discussion
Our results show that in patients with severe acute kidney injury (AKI) initiated on CRRT and surviving to ICU discharge, a platelet decrease while on CRRT is associated with increased odds of post-ICU infection and post-ICU septic shock. In patients with a platelet count of <100×103/μL at the time of CRRT initiation, patients with a >40% decrease in platelets had a doubling in the odds of developing post-ICU infections. While thrombocytopenia is known to associate with an increased risk of infection, this is the first study to show that the platelet decreases commonly observed following CRRT initiation are associated with subsequent infection regardless of the presence of thrombocytopenia.
Previous studies have shown that a decrease in platelets following CRRT initiation is common and associated with increased mortality. In a secondary analysis of the ATN trial, a platelet decline from baseline after CRRT initiation was associated with a 55% increase in the odds of 28-day mortality (OR 1.55, CI 1.19–2.02)[10], independent of the presence of thrombocytopenia at baseline. Nearly two-thirds of RRT patients had some amount of platelet decrease within 72 hours of RRT initiation, with an average platelet decrease of 33%. A study from the Mayo Clinic retrospectively evaluating 541 patients showed that platelet count before initiation of CRRT was a strong predictor of mortality, even after adjustments for cofounding variables[13]. While platelet decline following CRRT initiation was not directly evaluated, the authors further found that in the subset of patients who had a normal platelet count of >150,000×109 at CRRT initiation, those with a severe decrease to <50,000×109 platelets had higher mortality. Another retrospective study evaluated 125 patients who received CRRT for more than 72 hours and reported that a platelet decline of >50% was associated with higher mortality rates[14]. Finally, a secondary analysis of the randomized evaluation of normal versus augmented level (RENAL) study showed that a 60% decrease in platelets following CRRT initiation in patients who survived beyond four days was associated with increased 90-day mortality [15]. While the association between platelet decrease and death is consistently established, its association with secondary infection has not been previously reported.
It is unclear whether platelet decreases are due to CRRT factors such as shear force or platelet-membrane interactions, or if platelet decline is a marker of more severe illness. One study that measured platelet levels before and after the CRRT filter reported an estimated 24-hour loss of 625×10^9 platelets across the diafilter[16]. In patients on cardiopulmonary bypass, another membrane-based extracorporeal therapy, platelet loss occurs in 90% of patients following cardiac surgery, and nearly half of patients have a platelet nadir <100×103/μL despite having normal platelets at the time of surgery[11]. Similar findings have been observed in extracorporeal membrane oxygenation (ECMO)[17], plasma exchange[18], and hemodialysis (HD)[19–21], making it plausible that machine-related factors are causally related to the observed platelet loss in these modalities; however, further research into the mechanism of platelet loss is needed.
The relationship between AKI and infection is complex, and the subject of ongoing research. Sepsis is the leading cause of death in patients with AKI[5], and AKI is a risk factor for the development of subsequent infection[7, 22, 23]. In a study of 1,620 patients undergoing cardiac surgery, patients who developed stage 2 or 3 AKI were 3.5 times more likely to develop a later postoperative infection, and even stage 1 AKI was associated with a doubling in the odds of developing postoperative infection[24]. Secondary infection is also a major source of morbidity and mortality in patients treated with CRRT, and CRRT also appears to have a bidirectional association with sepsis. In a secondary analysis of the PICARD study, 54% of patients who were initiated on CRRT for a reason other than sepsis went on to develop a secondary infection, and the patients with secondary infection saw a doubling in their risk of mortality[6]. Postulated mechanisms include the presence of indwelling dialysis catheters or immunosuppression in the setting of critical illness.
One possible source of immunosuppression that has not been well explored in this patient population is relative immunosuppression related to loss of platelets during CRRT. Platelets have many roles beyond thrombus formation and are important to both innate and adaptive immunity[9]. Platelets mediate the recruitment of leukocytes and facilitate their translocation to areas of inflammation. Platelets also release cytokines and other molecular mediators such as serotonin, epinephrine, norepinephrine, nitric oxide, and vascular endothelial growth factor, contributing directly to the inflammatory response in the critically ill. Mouse models of infection with pseudomonas pneumonia demonstrate that platelet depletion increases the rate of pneumonia development[25, 26], and in patients with community-acquired pneumonia, thrombocytopenia is recognized as a risk factor for prolonged illness and mortality[27]. An association between platelet loss and infection has also been observed in cardiac surgery patients treated with cardiopulmonary bypass[11]. It is, therefore, plausible that platelet loss following CRRT is a contributor to the high rates of secondary infection noted in this population.
If platelet loss is a contributor to poor outcomes including infection, mitigation of platelet loss may be a strategy to improve patient outcomes. For instance, a pilot study of 40 patients previously demonstrated that Tirofiban, an anti-platelet IIb/IIIa receptor inhibitor, can prevent platelet loss and preserve platelet function in patients on CRRT[28]. Whether prevention of platelet loss using antiplatelet agents improves outcomes should be explored in future research.
Notably, our models produced higher odds ratios and more highly significant results in the sub-analysis of patients with a platelet-value of <100×103/μL at the time of CRRT initiation. Despite the low baseline platelet values, the mean platelet decrease in these patients following CRRT initiation was still 50%, and higher platelet decreases were strongly associated with subsequent infection and death. This finding suggests that platelet decrease adds additional value beyond baseline platelet value alone.
This study has multiple limitations that need to be considered when interpreting these results. Only new infectious episodes that occurred after CRRT initiation were considered in the analysis, and so secondary infections occurring in the ICU itself were not investigated. Secondly, time to infection was not available, which limited some of the statistical methods that could be utilized. Tests for important potential sources of thrombocytopenia, such as heparin-induced thrombocytopenia and disseminated intravascular coagulation, were not universally available. While we attempted to control for confounding using multivariable and propensity-adjusted models, residual confounding may be present. Finally, we do not have mechanistic data such as platelet activation status over time, and a causal relationship between platelet decreases and subsequent infection cannot be established.
Our study has several notable strengths. The population of 1,759 patients, including 797 patients who survived to ICU discharge, is among the largest CRRT studies published to date. By using only new post-ICU infections, there was less difficulty in diagnosing secondary infections, and the temporal relationship between CRRT and subsequent sepsis was clear.
Conclusion
Platelet decline following CRRT initiation is common and known to associate with increased mortality. In this study of 797 CRRT patients who survived to ICU discharge, a platelet decrease of >40% from the pre-CRRT value was associated with an increase in post-ICU infections, regardless of the presence of baseline thrombocytopenia. Further research is needed to evaluate the impact of both CRRT and platelet loss on immune function and the risk of subsequent infection.
Supplementary Material
Acknowledgments
Support
This study was supported in part by The University of Iowa Clinical and Translational Science Award granted with funds from the NIH (UL1TR002537).
This funding sources had no role in study design, data collection, analysis, reporting, or the decision to submit for publication.
Footnotes
Financial Disclosures
The authors declare that they have no financial conflicts of interest.
References
- 1.Hoste EA, Bagshaw SM, Bellomo R, Cely CM, Colman R, Cruz DN, Edipidis K, Forni LG, Gomersall CD, Govil D et al. : Epidemiology of acute kidney injury in critically ill patients: the multinational AKI-EPI study. Intensive Care Med 2015, 41(8):1411–1423. [DOI] [PubMed] [Google Scholar]
- 2.Palevsky PM, Liu KD, Brophy PD, Chawla LS, Parikh CR, Thakar CV, Tolwani AJ, Waikar SS, Weisbord SD: KDOQI US commentary on the 2012 KDIGO clinical practice guideline for acute kidney injury. Am J Kidney Dis 2013, 61(5):649–672. [DOI] [PubMed] [Google Scholar]
- 3.Kao CC, Yang JY, Chen L, Chao CT, Peng YS, Chiang CK, Huang JW, Hung KY: Factors associated with poor outcomes of continuous renal replacement therapy. PLoS One 2017, 12(5):e0177759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Griffin BR, Thomson A, Yoder M, Francis I, Ambruso S, Bregman A, Feller M, Johnson-Bortolotto S, King C, Bonnes D et al. : Continuous Renal Replacement Therapy Dosing in Critically Ill Patients: A Quality Improvement Initiative. Am J Kidney Dis 2019, 74(6):727–735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Selby NM, Kolhe NV, McIntyre CW, Monaghan J, Lawson N, Elliott D, Packington R, Fluck RJ: Defining the cause of death in hospitalised patients with acute kidney injury. PLoS One 2012, 7(11):e48580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Mehta RL, Bouchard J, Soroko SB, Ikizler TA, Paganini EP, Chertow GM, Himmelfarb J: Sepsis as a cause and consequence of acute kidney injury: Program to Improve Care in Acute Renal Disease. Intensive Care Med 2011, 37(2):241–248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Matejovic M, Chvojka J, Radej J, Ledvinova L, Karvunidis T, Krouzecky A, Novak I: Sepsis and acute kidney injury are bidirectional. Contrib Nephrol 2011, 174:78–88. [DOI] [PubMed] [Google Scholar]
- 8.Faubel S, Shah PB: Immediate Consequences of Acute Kidney Injury: The Impact of Traditional and Nontraditional Complications on Mortality in Acute Kidney Injury. Adv Chronic Kidney Dis 2016, 23(3):179–185. [DOI] [PubMed] [Google Scholar]
- 9.Semple JW, Italiano JE Jr., Freedman J: Platelets and the immune continuum. Nat Rev Immunol 2011, 11(4):264–274. [DOI] [PubMed] [Google Scholar]
- 10.Griffin BR, Jovanovich A, You Z, Palevsky P, Faubel S, Jalal D: Effects of Baseline Thrombocytopenia and Platelet Decrease Following Renal Replacement Therapy Initiation in Patients With Severe Acute Kidney Injury. Crit Care Med 2019, 47(4):e325–e331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Griffin BR, Bronsert M, Reece TB, Pal JD, Cleveland JC, Fullerton DA, Gist KM, Jovanovich A, Jalal D, Faubel S et al. : Thrombocytopenia After Cardiopulmonary Bypass Is Associated With Increased Morbidity and Mortality. Ann Thorac Surg 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Cawcutt KA, Peters SG: Severe sepsis and septic shock: clinical overview and update on management. Mayo Clin Proc 2014, 89(11):1572–1578. [DOI] [PubMed] [Google Scholar]
- 13.Guru PK, Singh TD, Akhoundi A, Kashani KB: Association of Thrombocytopenia and Mortality in Critically Ill Patients on Continuous Renal Replacement Therapy. Nephron 2016, 133(3):175–182. [DOI] [PubMed] [Google Scholar]
- 14.Wu B, Gong D, Xu B, He Q, Liu Z, Ji D: Decreased platelet count in patients receiving continuous veno-venous hemofiltration: a single-center retrospective study. PLoS One 2014, 9(5):e97286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lin J, Gallagher M, Bellomo R, Duan M, Trongtrakul K, Wang AY: SOFA coagulation score and changes in platelet counts in severe acute kidney injury: Analysis from the randomized evaluation of normal versus augmented level (RENAL) study. Nephrology (Carlton) 2019, 24(5):518–525. [DOI] [PubMed] [Google Scholar]
- 16.Mulder J, Tan HK, Bellomo R, Silvester W: Platelet loss across the hemofilter during continuous hemofiltration. Int J Artif Organs 2003, 26(10):906–912. [DOI] [PubMed] [Google Scholar]
- 17.Cheung PY, Sawicki G, Salas E, Etches PC, Schulz R, Radomski MW: The mechanisms of platelet dysfunction during extracorporeal membrane oxygenation in critically ill neonates. Critical care medicine 2000, 28(7):2584–2590. [DOI] [PubMed] [Google Scholar]
- 18.Abdullah S, Ayub Q, Saboor M: Pre and post apheresis platelet CD markers evaluation using flow cytometry. JPMA The Journal of the Pakistan Medical Association 2016, 66(11):1440–1443. [PubMed] [Google Scholar]
- 19.Daugirdas JT, Bernardo AA: Hemodialysis effect on platelet count and function and hemodialysis-associated thrombocytopenia. Kidney international 2012, 82(2):147–157. [DOI] [PubMed] [Google Scholar]
- 20.Cases A, Reverter JC, Escolar G, Sanz C, Sorribes J, Ordinas A: In vivo evaluation of platelet activation by different cellulosic membranes. Artificial organs 1997, 21(4):330–334. [DOI] [PubMed] [Google Scholar]
- 21.Kawabata K, Nakai S, Miwa M, Sugiura T, Otsuka Y, Shinzato T, Hiki Y, Tomimatsu I, Ushida Y, Hosono F et al. : Platelet GPIIb/IIIa is activated and platelet-leukocyte coaggregates formed in vivo during hemodialysis. Nephron 2002, 90(4):391–400. [DOI] [PubMed] [Google Scholar]
- 22.Griffin BR, You Z, Holmen J, SooHoo M, Gist KM, Colbert JF, Chonchol M, Faubel S, Jovanovich A: Incident infection following acute kidney injury with recovery to baseline creatinine: A propensity score matched analysis. PLoS One 2019, 14(6):e0217935. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.SooHoo M, Griffin B, Jovanovich A, Soranno DE, Mack E, Patel SS, Faubel S, Gist KM: Acute kidney injury is associated with subsequent infection in neonates after the Norwood procedure: a retrospective chart review. Pediatr Nephrol 2018, 33(7):1235–1242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Griffin BR, Teixeira JP, Ambruso S, Bronsert M, Pal JD, Cleveland JC, Reece TB, Fullerton DA, Faubel S, Aftab M: Stage 1 acute kidney injury is independently associated with infection following cardiac surgery. J Thorac Cardiovasc Surg 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Amison RT, O’Shaughnessy BG, Arnold S, Cleary SJ, Nandi M, Pitchford SC, Bragonzi A, Page CP: Platelet Depletion Impairs Host Defense to Pulmonary Infection with Pseudomonas aeruginosa in Mice. American journal of respiratory cell and molecular biology 2018, 58(3):331–340. [DOI] [PubMed] [Google Scholar]
- 26.de Stoppelaar SF, van ‘t Veer C, Claushuis TA, Albersen BJ, Roelofs JJ, van der Poll T: Thrombocytopenia impairs host defense in gram-negative pneumonia-derived sepsis in mice. Blood 2014, 124(25):3781–3790. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Anderson R, Feldman C: Review manuscript: Mechanisms of platelet activation by the pneumococcus and the role of platelets in community-acquired pneumonia. The Journal of infection 2017, 75(6):473–485. [DOI] [PubMed] [Google Scholar]
- 28.Link A, Girndt M, Selejan S, Rbah R, Böhm M: Tirofiban preserves platelet loss during continuous renal replacement therapy in a randomised prospective open-blinded pilot study. Crit Care 2008, 12(4):R111. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.