Is There Evidence of Benefit of Therapeutic Drug Monitoring for Direct Oral Anticoagulants?—Spinning Down the Centrifuge of Enthusiasm

Direct oral anticoagulants (DOACs) have become the first-line therapy for stroke prevention in nonvalvular atrial fibrillation (NVAF), because of their higher net clinical benefit over warfarin, mainly because of lower rates of major bleeding (MB) and intracranial hemorrhages (ICHs).1,2 Because of their relatively predictable pharmacokinetics, they do not require laboratory monitoring and have fixed clinical criteria for dose reduction of proven safety and efficacy.1,2 These advantages have increased the number of NVAF patients receiving recommended oral anticoagulant therapy (OAT).3,4 Still, OAT underprescription and DOAC underdosing are common in older patients, because of the fear of bleeding complications.5,6 Secondary analyses of Phase III randomized controlled trials (RCTs) have reported a significant interpatient variability of peak and trough DOAC plasma concentrations (DPCs) that were associated with efficacy and safety outcomes. DOAC exposure showed an inverse correlation with stroke risk that peaked at lower DPC percentiles and a steeper and more linear direct correlation with bleeding; risk of ICH remained low throughout a wide range of DPCs.7,8 These observations and real-world reports of MBs in patients with extremely high DPCs have posed the question whether the efficacy and safety profile of DOACs could be improved by concentration-driven dose-adjustment.7–9 The work by Bernier and colleagues published in a recent issue of the Journal of Cardiovascular Pharmacology10 adds to the number of studies on the subject. In this cross-sectional study, the authors measured DPCs by high-pressure liquid chromatography/tandem mass spectrometry on blood samples of 322 acute patients receiving rivaroxaban (74%) or dabigatran (26%), in 82% of cases for NVAF. The subjects were labeled as “out-of-range” when their DPCs fell above the 95th percentile (70 patients, 21.7%) or below the 5th percentile (49 patients, 15.2%) of published distributions. At univariate analysis, patients with DPCs above the 95th percentile had higher blood creatinine levels and were more frequently women. No significant difference was found between patients with DPCs below and above the 5th percentile. The authors investigated the association of out-of-range DPCs with outcomes adjudicated by standard clinical practice. It is important to note that these patients had a high pretest probability of thrombotic or hemorrhagic events, both because DPC testing was driven by clinical suspicion and because of high thromboembolic risk (mean CHA2DS2-VASc score 4.4 ± 1.7). A bleeding event was recorded in 62 patients (19.3%), in 71% of cases of major clinical relevance and in 8 cases fatal, whereas 26 patients (8.1%) experienced a thrombotic event, more often intracranial (53.8%). At univariate analysis, bleeding was associated with higher HAS-BLED scores and DPCs above the 95th percentile, whereas thrombotic complications with DPCs below the 5th percentile.10 The study results are of interest, but in our opinion, do not support the conclusion drawn in the title that “Therapeutic Drug Monitoring of Direct Oral Anticoagulants May Increase Their Benefit-Risk Ratio.10” This caveat is crucial for clinicians, who may be tempted to implement this message in their practice. We would like to make some considerations to put these interesting findings in an appropriate clinical context. First, some methodological aspects of this study should be underscored. This is not a large prospective cohort of consecutive outpatients, but a small cross-sectional convenience sample of acute patients, with limited information about therapeutic adherence, clinical indication for DPC evaluation, sampling time, and diagnosis at discharge (72.7% of these patients with suspected bleeding or thrombosis were not diagnosed either condition). Also, the study provides no information about the number of patients receiving a full-dose or a reduced-dose DOAC and a clear reporting of in-label or off-label DOAC dose use; the potential interaction of clinical variables with both DPCs and outcomes cannot be evaluated in the absence of a multivariate analysis. Even if these data were available and consistent with published results, the conclusion that “patients over 75 years old could benefit of a single residual measurement at steady state”10 would still be misleading11; this is an hypothesis-generating study that neither applied a standard DPC evaluation protocol nor assessed its possible clinical benefit. Second, the authors should at least have discussed the extensive research evidence required for the implementation of a pharmacokinetic/pharmacodynamic-guided (ie, DPC-guided) DOAC dosing strategy. This includes for each molecule: (1) proof that multiple DPC determinations in the same patient are consistent and recognition of optimal sampling time; (2) identification of a DPC desirable range, if any; (3) development of a DPC-guided dose adjustment protocol yielding in-range DPCs; (4) evidence of better safety and at least equal efficacy of this protocol; and (5) definition of a subset of patients for which it may be cost-effective and implementable.7–9 Eikelboom et al7 have estimated that an RCT on DPC-guided dose-adjustment would require at least 10,000 patient/yr to detect a 20% relative risk reduction in MB compared with approved clinical dose reduction criteria and higher sample sizes to demonstrate noninferior efficacy. A study on DPC-guided off-label underdosing of apixaban and rivaroxaban in Japanese NVAF patients has been recently published,12 and, to the best of our knowledge, it is the only one that has developed and evaluated the clinical implications of such a strategy. Suwa and colleagues measured DPCs on a first sample of 255 outpatients and identified DPC cutoffs to discriminate patients with major/clinically relevant nonmajor bleeding (23/255, 9.0% of the sample) with a 100% sensitivity. On an extended sample of 348 subjects, 34 of them (9.8%) had history of previous bleeding or presented DPCs above the identified cutoffs; in those receiving full doses, an off-label reduced dose was prescribed. After dose-adjustment, all patients showed DPCs above the 5th percentile and below the identified cutoff; no bleeding or thromboembolic events were recorded during a mean follow-up of 17 ± 6 months.12 Third, several physiopathological reasons make the clinical implications of DPC-guided dosing uncertain in real-world clinical practice. A “single residual measurement” of DPC may not be sufficient to grant a dose adjustment because a large proportion of patients with dabigatran concentrations in the upper and lower quintiles at baseline had midrange DPCs after 6 months.9 Moreover, the arbitrary use of the 5th and 95th percentiles as cutoffs would automatically label 10% of subjects in a normally distributed population as out-of-range, with a risk of over adjustment. The theoretical model suggested by Bernier et al implies a relatively simple and consistent relationship between a single DPC measurement and ischemic and hemorrhagic adverse events. However, reality is less straightforward because both pharmacokinetics (ie, DPCs) and pharmacodynamics (ie, clinical outcomes) are influenced by a host of intrinsic and extrinsic factors. The first include age, body weight, renal function, and evolving physiological and pathological conditions, whereas the second regard concomitant antithrombotic and interacting drugs.7,9 Relying on a “single residual measurement” of DPC to take long-term clinical decisions might thus seem rash. Moreover, DPC desirable ranges may vary according to the individual's thromboembolic and bleeding risks, other relevant clinical characteristics, and preference to favor stroke prevention or the reduction of MBs. In conclusion, even if DPC-driven dose adjustment is an attractive and evolving field of research, proof of concept of this strategy is still missing. This requires a more balanced interpretation of the present findings within the framework of current clinical recommendations. Physicians should keep in mind that, despite interesting speculations on DPC-guided dose adjustment, on-label DOAC dosing has consistently shown a higher net clinical benefit in RCTs and observational studies than warfarin, whereas off-label dose-adjustment on the basis of DPC is not evidence-based at the moment.