Acetylsalicylic acid in critically ill patients: a cross‐sectional and a randomized trial

Abstract Background Despite decades of clinical use, the pharmacokinetics and the effects of acetylsalicylic acid (ASA) in critically ill patients remain ill‐defined. We aimed to investigate the pharmacokinetics and the effects of different ASA formulations during critical illness. Design A cross‐sectional study and a randomized, parallel‐group trial were performed. Critically ill patients under chronic oral ASA treatment (100 mg enteric‐coated) were screened for high ‘on‐treatment’ platelet reactivity (HTPR) according to arachidonic acid‐induced whole‐blood aggregometry. Thirty patients with HTPR were randomized to receive 100 mg ASA intravenously, 100 mg enteric‐coated ASA bid (bis in die) or 81 mg chewable ASA (n = 10 per group). Serum thromboxane B2 (TXB2) levels, ASA and salicylic acid levels were quantified. Results Of 66 patients, 85% (95% confidence intervals 74–93%) had HTPR. Compared to baseline infusion of 100 mg, ASA significantly reduced platelet aggregation after 24 h to median 80% (Quartiles: 66–84%). Intake of 81 mg chewable ASA significantly reduced platelet aggregation to 75% (54–86%) after four hours, but increased it to 117% after 24 h (81–163%). Treatment with 100 mg enteric‐coated ASA bid decreased platelet aggregation after 24 h to median 56% (52–113%). Baseline TXB2 levels were median 0·35 ng/mL (0·07–0·94). Infusion of ASA or intake of 100 mg ASA bid reduced TXB2 levels to 0·07–0·18 ng/mL after 24 h, respectively. Chewable ASA reduced TXB2 levels only transiently. Pharmacokinetic analysis revealed highly variable absorption patterns of oral ASA formulations. Conclusion There is a very high prevalence of HTPR in critically ill patients on peroral ASA therapy, caused by an incomplete suppression of TXB2 and/or by impaired absorption of ASA.


Introduction
For decades, acetylsalicylic acid (ASA) has been a mainstay in the secondary prophylaxis of cardiovascular diseases [1], whereas its role in primary prevention remains debatable [2]. Intake of low-dose ASA was associated with improved outcome in critically ill patients [3], in patients with Staphylococcus aureus bloodstream infections [4] or in patients with septic shock from community-acquired pneumonia [5].
High on-treatment platelet reactivity (HTPR) is defined as an insufficient inhibition of platelets despite antiplatelet therapy that may be associated with an increased risk of experiencing cardiovascular events [6]. In contrast to P2Y12 inhibitors, no well-defined thresholds exist for treatment with ASA [6]. Depending on cut-offs and platelet function tests, the prevalence of HTPR ranges between~20% and 30% [7,8].
Surprisingly, no data are available on the prevalence of HTPR or on the pharmacokinetics of ASA in critically ill patients. However, the pharmacological properties of drugs during critical illness may be significantly altered, that is due to impaired gastric motility, altered bioavailability after enteral administration, altered drug distribution, organ dysfunction or altered enzyme activity [9][10][11]. Furthermore, critically ill patients may have an increased platelet turnover with increased risk of HTPR [12].

Patients
Inclusion criteria requested patients > 18 years of age, admitted to a medical ICU with pre-existent low-dose ASA treatment [oral 100 mg/day enteric-coated ASA (Thrombo-ASS, Gerot Lannach Pharma, Lannach, Austria)]. Exclusion criteria included allergies or hypersensitivities to the trial drugs, active bleeding, known coagulation disorders or intake of other antiplatelet drugs.

Noninterventional trial
Blood samples were drawn before the daily dose of ASA was administered, orally or via a nasogastric tube, 2 and 24 h thereafter. ASA tablets had to be crushed and/or dissolved in 0Á9% sodium chloride solution, if they were administered via a nasogastric tube.

Platelet function assays
To assess platelet function, we performed arachidonic acid (AA)-induced whole-blood aggregometry and measured platelet function under high shear rates, both standard methods to assess responsiveness to ASA treatment [14].

Pharmacokinetics
Plasma concentrations of ASA and salicylic acid were determined by liquid chromatography tandem mass spectrometry (LC-MS/MS) based on a published procedure (Appendix S1) [21].

Disease scores
The Sequential Organ Failure Assessment (SOFA) score and the simplified acute physiology score (SAPS) III were calculated on trial day 1 [22,23].

Randomization
Physicians enrolled patients, who were assigned continuous identification numbers. Patients with HTPR on trial day 1 were available for randomization. Physicians or a trained study nurse randomized patients through an online randomization programme (https://www.meduniwien.ac.at/randomizer/ web/login.php) using permuted block randomization.

Statistics
No data were available on the prevalence of HTPR in ICU patients. In stable patients, a HTPR rate of 20-30% was reported [7,8]. To randomize 30 patients, we assumed that we need to screen a multiple of that and initially planned to include at least 100 patients. However, due to the high rate of HTPR in European Journal of Clinical Investigation Vol 47 505 critically ill patients we finished the trial after inclusion of 66 patients.
We estimated a mean area under the curve (AUC) of 30 U in AA-induced platelet aggregation in patients with HTPR [18]. Thus, a sample of n = 10 per group allowed us to detect a significant reduction in platelet function from an AUC of 30 to 10 with a power of > 90% and a corrected alpha error of 1Á6%. HTPR rates are presented as percentage AE 95% confidence intervals (CI).
The primary endpoint was AA-induced platelet aggregation. Secondary endpoints included drug concentrations, platelet function under high shear rates, TXB2 concentrations and correlations.
A repeated-measures ANOVA was performed to compare repeatedly measured parameters (in detail: Appendix S1). To adjust for multiple comparisons, we performed the Bonferroni procedure.
For exploratory analysis, pairwise comparisons were performed by the Wilcoxon test, and unpaired comparisons with the Kruskal-Wallis ANOVA followed by a Mann-Whitney U-test. Due to the exploratory character of these analyses, no corrections for multiple testing were applied. Correlations were calculated using the nonparametric Spearman test.

Results
Sixty-six critically ill patients were recruited and finished the trial between 13 November 2012 and 25 March 2016 (Fig. 1). The trial ended after the last visit of the last patient. Patient demographics and baseline data are presented in Table 1. The 28-day mortality was 38%, which corresponded well with the predicted mortality from the simplified acute physiology score 3 [22]. All randomized patients completed the trial per protocol, and data were available for all analyses, while in the observational part, 24-h values of nine subjects were missing (Fig. 1).
Platelet aggregation between the three groups did not differ significantly. However, platelet aggregation changed significantly over time (P < 0Á001). This indicates the impact of each intervention on platelet aggregation during 24 h.
Infusion of 100 mg ASA i.v. and intake of 81 mg chewable ASA significantly reduced AA-induced platelet aggregation 66 patients with ASA treatment screened for high on-treatment platelet reactivity high on-treatment platelet reactivity by whole-blood aggregometry 30 patients randomized  Table S1). However, intake of 100 mg enteric-coated ASA bid did not significantly reduce platelet aggregation in univariate analysis; numerically MEA values after 24 h were the lowest of all groups (Table S1). The impact of the alternative treatments on HTPR is presented in Table S3.

Pharmacokinetics
There were no significant differences between the three treatment groups in multivariate analysis, although drug concentrations changed significantly over time (P < 0Á001).
Peak concentrations of ASA (P = 0Á013) and salicylic acid (P = 0Á005) were higher after infusion of 100 mg ASA compared to 100 mg enteric-coated ASA and maximum salicylic acid levels compared to 81 mg chewable ASA (P = 0Á023). Plasma concentrations of ASA and salicylic acid were similar between 81 mg chewable ASA and 100 mg enteric-coated ASA.
Plasma concentrations of ASA and salicylic acid levels are presented in Table S2 and Fig. 4.
The coefficient of variation of salicylic acid concentrations was 80% 2 h after administration of 100 mg enteric-coated ASA, 70% after intake of 81 mg chewable ASA and 45% after infusion of 100 mg ASA.
The T 1/2 of salicylic after infusion of ASA was approximately 102 and 96 min after intake of chewable ASA. The T 1/2 of entericcoated ASA could not be determined because of slow absorption.

Adverse events
The trial lasted 24 h in 36 patients and 48 h in 30 patients. No drug-related adverse events occurred in the interventional trial. Two patients experienced major bleeding and one patient acute cerebral infarction in the noninterventional study. One patient died during the noninterventional trial.

Discussion
Our main findings are that (i) the prevalence of HTPR is much higher in critically ill patients than reported previously [7,8], (ii) the plasma concentrations of ASA and salicylic acid are highly variable, and (iii) TXB2 levels, although suppressed to a large extent, show a remaining potentially pathophysiologically relevant background activity in many patients.
In patients with acute myocardial infarction with chronic ASA therapy, TXB2 levels were 0Á09 ng/mL (quartiles: 0Á03-0Á43) [24]. Additional infusion of 250 mg ASA decreased TXB2 levels to 0Á04 ng/mL (0Á01-0Á04 ng/mL) and significantly lowered HTPR [24]. This indicates that thromboxane levels < 0Á43 ng/mL are still relevant. Median TXB2 concentrations in our population were 0Á35 ng/mL and thus~fourfold higher [24]. Furthermore, a discrepancy between the measurement of serum TXB2 ex vivo and thromboxane generation in vivo, assessed by excretion of 2,3-dinor-thromboxane, was reported [25]. In healthy volunteers, a 94 AE 1% inhibition of serum TXB2 ex vivo by intake of aspirin only translated to a 28 AE 8% inhibition of urinary 2,3-dinor-thromboxane in vivo [25]. However, the in vivo generation of thromboxane was reduced by increasing the degree of serum TXB2 inhibition to ≥ 95% [25]. Thus, ≥ 95% inhibition of TXB2 could be necessary to achieve sufficient platelet inhibition.
Surprisingly, even after the infusion of ASA, TXB2 was not fully suppressed. Although thromboxane is generally regarded as COX-1 dependent [26], other sources of thromboxane generation have been reported [27,28] including COX-2-dependent thromboxane generation during inflammatory states [29] or in reticulated platelets [30]. Moreover, nucleated cells may regenerate COX-1 [31]. Treatment of septic patients with ibuprofen had beneficial effects on metabolic biomarkers, such as oxygen consumption or blood lactate levels, and it reduced the excretion of thromboxane metabolites, but the mortality benefit did not reach significance (37% vs. 40%) [32]. However, this trial included 455 patients and was only powered to detect a 35% difference in mortality between groups. The observed difference would have required a sample size of approximately > 8Á000 patients.
We detected a surprisingly high HTPR rate. Most studies report a HTPR rate of approximately 20% using MEA [33]. In patients undergoing haemodialysis, 50% of patients had HTPR [34]. Interestingly, in patients undergoing coronary artery bypass-graft surgery, the HTPR rate increased postoperatively [17,35]. Various factors, such as platelet count, haematocrit, leucocyte count, red blood cell count, C-reactive protein levels or fibrinogen, correlated with the results of MEA [16,36]. In our trial, only platelet count correlated well with MEA results. However, our population was not homogenous and probably too small to find such associations. The extraordinary high rate of HTPR in our trial may be explained by the residual thromboxane activity, a higher platelet turnover and possibly an impaired absorption after oral ASA treatment.
The PFA-100 classified 44% of patients with HTPR, which is in line with other reports [37]. The closure time detected by this system is affected by haematocrit, platelet count and the von Willebrand factor activity [38]. There is limited knowledge on its applicability in critically ill patients. Platelet function measurements with CADP cartridges showed a generalized platelet defect in 24 of 66 patients. In these patients, detection of HTPR may not have been possible, and the prevalence of HTPR may therefore be underestimated.
Crushing and/or dissolving tablets affects the pharmacokinetics with a shorter T max compared to intake of whole tablets [39]. Individual plasma concentration-time curves after intake of oral ASA formulations suggest highly variable absorption patterns in critically ill patients ( Figure S1), which is also reflected by high coefficients of variation for salicylic acid levels ranging between 45% and 80% 2 h after intake. In some patients, drug absorption seems to be delayed possibly due to reduced gastrointestinal motility. Compared to healthy volunteers, the salicylic acid concentrations after intake of ASA were markedly lower in our patients, regardless of the formulation [18,39]. Thus, absorption of orally administered drugs is severely impaired in some critically ill patients. Therefore, especially in high-risk patients, intravenous formulations should be considered to optimize bioavailability.
The half-life of salicylic acid on day 2 for i.v. and chewable ASA was approximately 100 min. This is in line with other studies [18] and demonstrates that metabolism and excretion of ASA is not altered in ICU patients.
Taken together, these data indicate that in order to ensure adequate platelet inhibition in critically ill patients, clinicians should consider using intravenous formulations, and/or possibly two daily doses. Treatment with oral formulations bears the risk of insufficient absorption of ASA and consequently reduced antiplatelet effects.

Limitations
The included population was quite inhomogeneous. However, all patients were treated with ASA and we aimed to define the prevalence of HTPR in ICU patients regardless of the underlying disease. Secondly, this study was powered to show improvements in platelet function assays by alternative treatments and to show the significantly increased rate of HTPR in critically ill patients, but not to assess the clinical impact of HTPR in these patients. Thirdly, we were only able to conduct sparse pharmacokinetic analyses due to the severe condition of patients. Finally, this trial did not interfere with regular treatment of patients and the intake of nonsteroidal antiphlogistic drugs was therefore not forbidden.

Conclusion
The measured TXB2 levels and the high HTPR rate suggest insufficient platelet inhibition by ASA in critically ill patients. This may in part be caused by the impaired absorption of orally administered drugs or by COX-2-dependent thromboxane generation during systemic inflammation. Intravenous formulations reduce the substantial variability of ASA/salicylic acid concentrations and could be a better choice for high-risk patients in the ICU.

Supporting Information
Additional Supporting Information may be found online in the supporting information tab for this article: Appendix S1. Methods  Table S1. Arachidonic Acid-induced platelet aggregation using whole blood aggregometry after alternative ASA treatments. Table S2. Acetylsalicylic acid (ASA) and Salicylic acid (SA) plasma concentrations after different ASA treatments.