Keywords apixaban - atrial fibrillation - elderly - rivaroxaban - thrombin generation
Introduction
Nonvalvular atrial fibrillation (NVAF) is the most common heart rhythm disorder with
a prevalence exceeding 10% in octogenarians.[1 ] Current guidelines recommend the use of direct oral anticoagulants (DOACs) targeting
either factor Xa (FXa; such as rivaroxaban or apixaban) or thrombin (dabigatran) as
first-line treatment for the prevention of stroke and systemic thromboembolism in
patients with NVAF.[2 ]
[3 ] DOACs are characterized by predictable pharmacokinetics (PK) and pharmacodynamics
(PD) with a wide therapeutic index: therefore, they are administered at fixed doses
without laboratory monitoring.[4 ]
[5 ]
[6 ]
[7 ]
[8 ]
[9 ]
[10 ]
[11 ] However, the management of very elderly patients receiving anticoagulant treatment
may be challenging because they are at both high risk of thrombotic and bleeding complications
even though meta-analysis and real-life studies provided reassuring data regarding
the clinical benefit–risk balance of DOACs.[12 ]
[13 ]
[14 ]
[15 ] Phase II/III clinical trials and phase IV studies evaluating the PK/PD profiles
of DOACs in healthy volunteers and NVAF patients have provided consistent PK and PD
data with substantial inter-individual variability.[16 ]
[17 ]
[18 ]
[19 ]
[20 ]
[21 ]
[22 ]
[23 ]
[24 ]
[25 ]
[26 ] However, little is known regarding specific concentration–time profiles of rivaroxaban
and apixaban in polymedicated NVAF patients aged ≥80 years.[27 ]
[28 ]
[29 ] This age group is characterized by the presence of multiple comorbidities (including
renal impairment) and polypharmacy, potentially altering both drug exposure and effect.
Both xabans are metabolized via cytochrome P450 (CYP) 3A4/5 and CYP2J2, and substrates
of the drug-efflux transporter P-glycoprotein (P-gp). Several studies have evaluated
the effect of moderate CYP3A4/P-gp inhibitor concomitant use on rivaroxaban or apixaban
exposure, especially in healthy volunteers.[4 ]
[20 ] One question is whether drug–drug interactions could influence PK response variability
in the very elderly.
Regarding the assessment of DOAC PD, the study of thrombin generation (TG) has been
shown as a suitable and reliable assay to assess in vitro or ex vivo anticoagulant
effect of direct FXa inhibitors in contrast to prothrombin time.[30 ]
[31 ]
[32 ]
[33 ]
[34 ]
[35 ]
[36 ] Nevertheless, to the best of our knowledge, no specific TG data regarding elderly
patients aged 80 years and over receiving rivaroxaban or apixaban have been published
yet.
Hence, we conducted an academic prospective exploratory study in a cohort of NVAF-hospitalized
patients aged ≥80 years receiving rivaroxaban or apixaban. The primary objectives
were to characterize rivaroxaban and apixaban concentration–time profiles and determine
the potential impact of demographic, clinical, therapeutic, and pharmacogenetic factors
on their peak and trough concentrations. The secondary objectives were to perform
TG assays in a subset of patients under several experimental conditions and evaluate
the potential influence of individual characteristics on TG parameter variability
at peak and trough levels. The tertiary objectives were to report clinical outcomes
(bleeding and thromboembolic complications, death) at 6 months.
Methods
Adage Study Design and Inclusion Criteria
The Assessment of Direct oral Anticoagulants in GEriatrics
(Adage) study is an academic multicenter prospective observational study (www.clinicaltrials.gov ; NCT02464488). We recruited hospitalized patients in university hospital geriatric
departments (Assistance Publique-Hôpitaux de Paris AP-HP, France; CHU-UCL Namur-site-Godinne,
Belgium) (January 2015 to June 2019). Patients aged ≥80 years and receiving rivaroxaban
or apixaban for NVAF since at least 4 days were eligible. The DOAC dose regimen was
at the discretion of the physician. Exclusion criteria included any acute unstable
comorbid condition or an estimated life expectancy of a few weeks. The study was approved
by the local ethics committee (Comité consultatif de protection des personnes pour
la recherche biomédicale – Île-de-France-6). All participants gave their written informed
consent to participate in the study.
Physicians prospectively collected demographic data, laboratory data (including creatinine
clearance [CrCl] calculated with the Cockcroft–Gault formula), comorbid conditions
allowing the calculation of the CHA2 DS2 VASc, HEMORR2 HAGES, and the 13-item Cumulative Illness Rating Scale-Geriatrics (CIRS-G) scores.[37 ] We recorded DOAC regimen and concomitant medications (excluding vitamins). None
of the patients had received heparin derivatives in the last 5 days prior the first
sampling. Drugs were considered as modulators (substrate or moderate/strong inhibitors)
of P-gp/CYP3A4/5 according to Geneva University Hospitals classification.[38 ] Finally, geriatricians were responsible for the clinical follow-up of the patients
at 6 months through a phone call and/or the use of medical records, and recorded bleedings
(site, severity), thromboembolic events (site), and deaths. Two geriatricians (F.K.,
L.R.) blinded to the PK/PD study results adjudicated bleeding events as major or nonmajor
clinically relevant bleedings according to the International Society on Thrombosis
and Haemostasis criteria[39 ]; minor bleedings were not systematically recorded.
Sample Collection in Adage Patients
Blood samples were collected by venipuncture into a citrate tube (buffered trisodium
citrate 0.109 M, 9 vol/1vol) in addition to other samples taken as part of usual care
between 7.00 a.m. and 7.00 p.m. No additional venipuncture was specifically performed
for our study purpose. Sampling and last DOAC intake times were systematically recorded
on a standardized request form. Sampling time points were at the discretion of the
physicians. At most five samples per patient were collected over a maximal 20-day
period following inclusion. Time points were classified according to the time elapsed
from the last oral intake (T , hours). Especially, for rivaroxaban patients, T1 to T4 samples corresponded to time
to peak (TTP) level (namely T
max ) and T20 to T24 corresponded to time to trough level (T
min ); for patients receiving apixaban (twice daily [b.i.d.]), T1 and T4 samples corresponded
to T
max , and T10 to T12 samples to T
min .
Tubes were sent to each center's laboratory within 2 hours after sampling. They were
immediately double-centrifuged for 15 minutes at 2,000 g at 15 to 22°C to obtain platelet-poor plasma (PPP) distributed in 500 μL aliquots;
pellets were kept for genotyping. All samples were stored at −80°C, prior to onward
shipment on dry ice for central analysis (Haematology laboratory, Hôpital Lariboisière
AP-HP, Paris, France). PPP samples were thawed for 3 to 5 minutes in a 37°C water-bath
just before use.
Rivaroxaban and Apixaban Concentrations and Thrombin Generation Assessment
Anti-Xa chromogenic assays (STA-Liquid-Anti-Xa, Stago, Asnières-sur-Seine, France)
were used to determine rivaroxaban and apixaban plasma concentrations using dedicated
calibrators and controls on STA-R-Evolution (Stago) (lower limit of quantification:
20 ng/mL). The maximal observed plasma DOAC concentration (C
max ) was defined as the level measured at T
max (T1–T4), whereas the minimal one (C
min ) was defined as the level measured at T
min (i.e., T20–T24 for rivaroxaban, T10–T12 for apixaban).
TG was studied using the ST-Genesia system (Stago) whenever sufficient plasma was
available. ST-Genesia is a fully automated system enabling quantitative standardized
assessment of TG derived from Hemker's fluorescence method, using dedicated reagents,
calibrator, and quality controls.[35 ] Three experimental conditions (as a function of the reagent used) were tested: STG-DrugScreen
and STG-ThromboScreen, the latter in presence/absence of thrombomodulin (+TM/− TM)
according to the manufacturer's instructions. STG-DrugScreen contains a mixture of
phospholipids and recombinant human tissue factor (TF) at a high pM concentration;
STG-ThromboScreen contains TF at a lower, intermediate, pM concentration.[35 ] TM is contained in the reagent at a concentration chosen to inhibit 50% of the endogenous
thrombin potential (ETP) assessed in normal pooled plasma. Four parameters were analyzed:
lag time (LT; in minutes, time from test triggering to signal detection), TTP (in
minutes, time necessary for thrombin concentration to reach its maximal value), peak
height (PH; in nM: maximal thrombin concentration), and ETP (in nM·min: area under
the thrombin time–concentration curve).[35 ] Results are presented as absolute values and inhibition percentages in presence
of TM.
TG in Plasma Samples Spiked with Xabans
Frozen normal pooled plasma samples (Cryocheck, Montpellier, France) were spiked with
increasing concentrations of rivaroxaban or apixaban ranging from 0 to 450 ng/mL as
previously described.[35 ] We measured effective plasma concentrations in each sample with specific anti-Xa
assays.
ABCB1 , CYP3A5 , and CYP2J2 Genotyping
DNA was extracted from peripheral blood leukocytes using the kits E.Z.N.A. SQ (Omega
Bio-tek, Norcross, Georgia, United States). We analyzed four loss-of-function variants
in three genes potentially involved in the DOAC metabolism (CYP2J2, CYP3A5 ) or transport (ABCB1 encoding P-gp) variability: CYP2J2 -76G > T (rs890293;*7), CYP3A5 6981A > G (rs776746;*3),
ABCB1 rs2032582 (2677G > T/A), and ABCB1 rs1045642 (3435C > T) (PharmGKB; www.pharmvar.org ). They were analyzed using real-time polymerase chain reaction with TaqMan Assay-Reagents
for allelic discrimination (Applied Biosystems, Courtabœuf, France).
Statistical Analysis
All statistical analyses were run on R-software (version 3.5.1) using the nlme package.[40 ]
[41 ] Quantitative data were described as median (interquartile range [IQR]) or mean values
(±standard deviation), and minimal and maximal values. Genetic variants were coded
as 0 (wild type), 1 (heterozygous), or 2 (homozygous) to model additive allelic effects.
Frequencies of genotypes were compared to the Hardy–Weinberg equilibrium expected
frequencies using the χ2 test. The association between concentrations (at T
max and T
min ) or PD parameters and each covariate was tested using either Spearman's rank correlation
(quantitative covariate) or the Kruskal–Wallis test (qualitative covariate). We prespecified
that PH measured using STG-ThromboScreen without TM would be chosen as the PD parameter
for statistical analyses (given its optimal sensitivity to a wide range of DOAC concentrations
according to previous data of our group).[35 ] Multivariate analysis was done using multiple linear regression, after log-transformation
of the predicted parameter (concentrations or PH). Predictors included in this analysis
were selected based on either their association with the studied parameter (p- value less than 0.10 in the univariate analysis) or their clinical pertinence. Results
are given as 95% confidence interval of the parameters in the model. Model quality
and assumptions were graphically checked. A p -value less than 0.05 was considered statistically significant. All p -values are given uncorrected.
Results
Patient Characteristics
From November 2015 to July 2019, 215 in-patients (153 females, 62 males) with NVAF
were included in the Adage study, with a mean age of 86.6 ± 4.3 years (min–max: 80–100). Patient characteristics
are summarized in [Table 1 ]. The rivaroxaban group comprised 104 patients: 18 (17.3%) received the 20-mg full
dose o.d., 83 (79.8%) the 15-mg dose, and 3 (2.9%) the 10-mg dose. The apixaban group
comprised 111 patients: 26 (23.4%) received the 5-mg full dose b.i.d. and 85 (76.6%)
the 2.5-mg dose b.i.d.; 72 (33.5%) patients did not receive the dose recommended according
to the Summary of Product Characteristics[10 ]
[11 ] (i.e., 15-mg rivaroxaban in patients with CrCl ≤50 mL/min, and 2.5-mg apixaban in
patients ≥80 years with creatinine level ≥133 µM or body weight ≤60 kg): 62 patients
received a lower dose regimen than recommended (28 on rivaroxaban and 34 on apixaban),
whereas 10 patients received a higher one (8 on rivaroxaban and 2 on apixaban).
Table 1
Characteristics of Adage patients at inclusion
Total
Rivaroxaban
Apixaban
p -Value
n = 215
n = 104
n = 111
Demographic characteristics
Age (y)
86.6 ± 4.3
86.4 ± 4.4
86.8 ± 4.3
0.491
Females (%)
71.2
68.3
73.9
0.371
Body weight (kg)
65.6 ± 14.7
64.9 ± 12.6
66.4 ± 16.5
0.543
DOAC regimen
Full dose (%)[a ]
20.5
17.3
23.4
0.312
Reduced dose (%)[b ]
79.5
82.7
76.6
0.312
Clinical characteristics
Type of AF
Persistent (%)
9.1
8.7
9.4
0.978
Paroxysmal (%)
34.0
35.0
33.0
0.978
Permanent (%)
56.9
56.3
57.6
0.978
Previous stroke or TIA (%)
24.9
21.4
28.2
0.250
Diabetes mellitus (%)††
19.0
10.9
26.6
0.005
Hypertension (%)
76.1
72.8
79.1
0.336
Heart failure (%)
49.8
46.6
52.7
0.412
Dyslipidemia (%)
26.4
24.3
28.4
0.535
Clinical scores
CHA2 DS2 VASc
5.0 ± 1.4
4.9 ± 1.4
5.2 ± 1.4
0.122
HEMORR2HAGES†
2.3 ± 1.0
2.1 ± 1.0
2.5 ± 1.0
0.022
CIRS-G†
10.4 ± 4.5
9.3 ± 3.7
11.4 ± 4.8
0.007
Laboratory data
Creatinine clearance (mL/min)[c ]
49.1 ± 16.5
50.5 ± 17.1
47.9 ± 15.9
0.314
Albumin (g/L)
33.2 ± 5.2
33.4 ± 4.2
33.0 ± 5.9
0.756
C-reactive protein (mg/L)
28.8 ± 39.8
28.6 ± 41.9
28.9 ± 38.0
0.374
Hemoglobin (g/dL)
12.1 ± 1.7
12.2 ± 1.7
12.0 ± 1.8
0.400
Fibrinogen (g/L)
5.0 ± 1.5
4.9 ± 1.6
5.0 ± 1.4
0.309
Therapeutic data
Number of co-medications
6.2 ± 2.7
5.9 ± 2.5
6.5 ± 2.9
0.256
≥1 P-gp substrate/inhibitor including amio. (%)
48.3
44.6
51.8
0.292
≥1 CYP3A4/5 inhibitor including amio. (%)
22.3
17.8
26.4
0.136
Amiodarone (%)
21.8
17.8
25.5
0.187
Antiplatelet drugs (%)
15.0
10.0
20.0
0.055
Abbreviations: AF, atrial fibrillation; Amio, amiodarone; TIA, transient ischemic
attack.
Note: Data shown as percentage or mean ± standard deviation. Statistically significantly
different values between the rivaroxaban and apixaban groups are indicated: †
p < 0.05 or ††
p < 0.01.
a 20 mg o.d. (rivaroxaban) or 5 mg b.i.d. (apixaban).
b 15 or 10 mg o.d. (rivaroxaban) or 2.5 mg b.i.d. (apixaban).
c Estimated using the Cockcroft–Gault formula.
Most patients had multiple comorbidities (mean CIRS-G score: 10.4 ± 4.5); 68.8% had
moderate renal impairment (CrCl: 30–59 mL/min). Most patients were polymedicated (mean
number of comedications: 6.2 ± 2.7 per patient). Overall, patient characteristics
did not significantly differ between the two DOAC groups, except for CIRS-G and HEMORR2 HAGE scores which were higher in the apixaban group (p = 0.0023 and p = 0.0218, respectively). ABCB1 2677G > T/A , ABCB1 3435C > T , CYP2J2*7 , and CYP3A5*3 allelic frequencies in Adage patients are shown in [Supplementary Table S1 ] (available in the online version).
Concentration–Time Profiles of Rivaroxaban and Apixaban in Adage Patients
A total of 456 blood samples were collected (average 2.1 samples per patient): 232
from 104 rivaroxaban patients and 224 from 111 apixaban patients. [Fig. 1 ] shows DOAC plasma concentration–time profiles. Among Adage patients receiving 15-mg rivaroxaban o.d., median C
max was 251 ng/mL (IQR: 176–377; range: 64–676); median C
min was 42 ng/mL (IQR: 28–70, range: <20–241). Among patients receiving 2.5-mg apixaban
b.i.d., the median C
max was 180 ng/mL (IQR: 130–227; range: 69–469) and the median C
min was 74 ng/mL (IQR: 49–116; range: <20–251). At T
max , 30.6% of patients on rivaroxaban 15 mg o.d. and 31.8% of patients on 2.5-mg apixaban
b.i.d. had plasma concentrations above the 95th percentile of values reported in the
Summary of Product Characteristics obtained in younger patients included in pivotal
clinical trials.[10 ]
[11 ]
[21 ] At T
min , only 3.1 and 9.7% were above the 95th percentile, whereas 21.5 and 9.7% were below
the 5th percentile, respectively.
Fig. 1 Plasma concentrations of rivaroxaban (A) and apixaban (B) in Adage patients as a function of time since the last oral intake. (A ) Patients receiving rivaroxaban 10, 15, or 20 mg o.d. are represented with yellow,
red, and grey circles, respectively (noteworthy, 10 mg off-labeled in AF). In order
to facilitate interpretation, we represented the geometric mean C
max and C
min and the 5th–95th percentiles, i.e., 229 ng/mL (5th–95th: 178–313) and 57 ng/mL (5th–95th: 18–136), as reported in ROCKET-AF
patients (median age: 73 years) receiving 15 mg according to Girgis et al[21 ] with solid and dashed lines, respectively. (B ) Patients receiving apixaban 2.5 or 5 mg b.i.d. are represented with blue and grey
circles, respectively. We represented the median C
max and C
min and 5th–95th percentiles, i.e., 123 ng/mL (5th–95th: 69–221) and 79 ng/mL (5th–95th: 34–162), as reported in patients
receiving apixaban 2.5 mg b.i.d. in ARISTOTLE trial[11 ] (median age: 70 years), with solid and dashed lines, respectively. AF, atrial fibrillation;
b.i.d., twice daily; o.d., once daily.
Association of Covariates with Rivaroxaban Plasma Concentration at C
max and C
min
In the rivaroxaban group, we observed an important inter-individual variability of
C
max (coefficients of variation [CVs] 49 and 47%) and C
min (75 and 37%) for 15- and 20-mg regimens, respectively. No covariates were found significantly
(p < 0.05) associated with C
max in univariate analysis ([Table 2 ]).
Table 2
Influence of Adage patients' characteristics on xaban plasma concentrations at T
max and T
min
Covariates
Univariate analysis
Multivariate analysis
C
max
C
min
C
max
C
min
Effect
%
(explained variability)
p
Effect
%
(explained variability)
p
Effect
p
Effect
p
Rivaroxaban
Dose regimen
+20.3%[a ]
[−16.7; +73.5]
2.23%
0.346
−3.8%[a ]
[−35.1; +42.6]
0.72%
0.974
29.4%[a ]
[−14.2; 95.2]
0.210
−3.2%[a ]
[−35.1; 44.6]
0.873
Female gender
+12.2%
[−17.7; 53.0]
4.09%
0.28
−14.2%
[−36.0; 15.1]
3.34%
0.424
+11.8%
[−20.8; 57.9]
0.515
−10.1%
[−35.2; 22.3]
0.466
Creatinine clearance
+1.89%[b ]
[−8.2; 13.1]
0.72%
0.83
+8.1%[b ]
[−3.9; 15.8]
0.57%
0.093
+3.4%[b ]
[−7.6; 15.6]
0.553
+5.0%[b ]
[−4.9; 16.0]
0.325
Amiodarone
+4.73%
[−31.5; 60.2]
0.07%
0.99
+47.8%
[−2.8; 109.4]
2.23%
0.0315
−26.1%[b ]
[−75.8; 125.5]
0.345
99.8%[c ]
[10.3; 262.0]
0.023
P-gp inhibitor[c ]
−9.95%
[−35.3; 25.4]
1.42%
0.60
−15.9%
[−37.6; 13.5]
1.47%
0.264
−6.31%
[−35.8; 36.6]
0.200
−8.7%
[−34.2; 26.8]
0.162
CYP3A4/5 modulator[c ]
+9.5%
[−21.0; 51.6]
1.20%
0.52
+31.1%
[−24.7; 72.2]
3.45%
0.0361
+10.6%
[−23.1; 58.9]
0.417
+43.1%
[5.3; 94.3]
0.023
ABCB1 3435C > T
+18.3%[d ]
[−5.3; 36.6]
4.25%
0.20
+16.9%[d ]
[−6.6; 31.0]
1.39%
0.0788
+18.3%[d ]
[−10.9; 39.8]
0.188
+10.4%[d ]
[−14.9; 30.1]
0.381
Apixaban
Dose regimen
+44.9%
[e ]
[7.14; 96.1]
12.20%
0.025
+45.6%
[e ]
[4.7; 102.6]
7.45%
0.0317
+64.8%
[e ]
[16.6; 132.9]
0.006
+54.8%
[e ]
[6.65; 124.6]
0.022
Female gender
+44.0%
[7.4; 93.1]
9.80%
0.017
+15.7%
[−14.9; 57.1]
1.12%
0.348
+28.7%
[−4.5; 73.4]
0.095
14.5%
[−17.3; 58.6]
0.410
Creatinine clearance
+3.3%[b ]
[−4.28; 10.31]
0.98%
0.289
+4.6%[b ]
[−4.2; 12.7]
3.23%
0.166
+4.5%[b ]
[−3.0; 11.4]
0.223
+4.5%[b ]
[−4.9; 13.1]
0.332
Amiodarone
+43.5%
[7.9; 90.9]
10.25%
0.0098
+14.4%
[−18.5; 60.5]
0.58%
0.433
+79.3%
[5.5; 204.7]
0.032
+44.1%
[−31.3; 202.2]
0.492
P-gp inhibitor[c ]
−12.7%
[−37.1; 21.0]
2.12%
0.408
+5.7%
[−26.6; 52.1]
0.23%
0.764
−4.0%
[−36.3; 44.8]
0.187
−10.7%
[−43.7; 41.6]
0.808
CYP3A4/5 modulator[c ]
+4.0%
[−20.4; 35.9]
0.23%
0.769
+20.8%
[−9.8; 61.7]
2.42%
0.203
+12.1%
[−17.3; 51.9]
0.284
+23.4%
[−11.1; 71.3]
0.330
ABCB1 3435C > T
+2.1%[d ]
[−18.7; 19.3]
0.13%
0.938
+14.8%[d ]
[−4.1; 30.3]
1.70%
0.087
+10.1%[d ]
[−8.3; 25.3]
0.256
+15.2%[d ]
[−4.1; 30.9]
0.113
Note: significance threshold p < 0.05 (bold values)
a 15 mg o.d. vs. 20 mg o.d.
b Per decrease of 10 mL/min.
c Patients who received a P-gp or CYP3A4/5 modulator in addition to amiodarone.
d Per mutated allele. The 3 patients who received 10 mg o.d. were excluded from the
statistical analysis.
e 2.5 mg b.i.d. vs. 5 mg b.i.d.
Rivaroxaban concentrations slightly increased with the daily dose (+20.3% in 20-mg
patients compared to 15 mg, p = 0.346). At T
min , patients on amiodarone displayed significantly higher rivaroxaban levels compared
to the other patients (60 vs. 41 ng/mL; +47.8%; p = 0.0315). The intake of a CYP3A4/5 modulator other than amiodarone also contributed
to increase C
min (p = 0.0361). C
min increased when renal function worsened (+8.1% by 10 mL/min CrCl decrease, p = 0.093). Finally, there was a trend to higher concentrations in ABCB13435C > T carriers, without reaching significance (+16.9% per mutated allele, p = 0.0788). The multivariate analysis included gender, amiodarone intake, CYP3A4/5-
or P-gp-modulator intake in addition to amiodarone, DOAC daily dose (20 vs. 15 mg),
CrCl, and ABCB1 3435C > T genotype. The model also included interaction terms between amiodarone and other
modulators intake. This analysis showed significantly higher C
min in patients receiving either amiodarone (p = 0.0232) or a CYP3A4/5 modulator other than amiodarone (p = 0.0228).
Association of Covariates with Apixaban Plasma Concentrations at C
max and C
min
We observed a substantial inter-individual variability in Adage patients on apixaban with CVs of 46 and 47% at T
max and 61 and 68% at T
min for 2.5 and 5.0 mg b.i.d. regimens, respectively. In univariate analysis, the apixaban
dose regimen was significantly associated with apixaban concentrations, which were
44.9% higher (p = 0.0247) at T
max and 45.6% higher at T
min (p = 0.0317) in patients receiving the full dose ([Table 2 ]). Two other variables were associated with higher apixaban C
max : amiodarone therapy (mean C
max 251 vs. 175 ng/mL, +43.5%, p = 0.0098) and female gender (mean C
max 208 ng/mL vs. 145, +44.0% in women, p = 0.0165). At T
min , higher concentrations were observed in ABCB1 3435C > T carriers, without reaching statistical significance (+14.8% per mutated
allele, p = 0.0867). No other variables apparently explained the apixaban concentration variability
at T
max or T
min . The multivariate analysis included the same predictors than for rivaroxaban. The
dose regimen was significantly associated with apixaban concentrations at both T
max (+64.8%, p = 0.0058) and T
min (+54.8%, p = 0.0222). Amiodarone might also contribute to higher concentrations at T
max (p = 0.0319).
Thrombin Generation Profiles of Rivaroxaban and Apixaban in Adage Patients
TG parameters were measured in 250 plasma samples from 128 Adage patients: 128 samples from 62 rivaroxaban patients and 122 from 66 apixaban patients.
Patient characteristics did not differ significantly between both groups (not shown).
Temporal parameters (LT and TTP) were prolonged and PHs were decreased in a DOAC concentration-dependent
manner (p < 10−4 ), using either STG-DrugScreen ([Supplementary Fig. S1 ], available in the online version) or intermediate pM TF concentration (STG-ThromboScreen
without TM [[Fig. 2 ]], LT and TTP not shown); ETP was associated with apixaban concentrations under these
both experimental conditions (p = 0.0102 and p < 10−4 , respectively), but not with rivaroxaban concentrations (p = 0.1028 and 0.1846, respectively).
Fig. 2 Peak height and endogenous thrombin potential as a function of DOAC concentrations
in Adage patients and in normal pooled plasma samples spiked with DOAC, using STG-Thrombo-Screen
reagent in the absence of thrombomodulin, and inhibition percentage in the presence
of thrombomodulin. Absolute values of TG parameters in the absence of thrombomodulin
(TM) (left) and percentage of inhibition in the presence of TM (right). Samples from
patients receiving 10, 15, or 20 mg of rivaroxaban o.d. are represented with yellow,
red, and grey circles, respectively. Samples from patients receiving 2.5 or 5 mg of
apixaban b.i.d. are represented with blue and grey circles, respectively. Samples
from pooled normal plasma spiked with DOAC at varying concentrations are represented
with green diamonds.[35 ] b.i.d., twice daily; DOAC, direct oral anticoagulant; o.d., once daily; TG, thrombin
generation.
Interestingly, Adage patients' PH and ETP values presented substantial variability with both xabans for
a given plasma concentration, independently of the dose regimen and the experimental
conditions ([Supplementary Fig. S1 ] [available in the online version] and [Fig. 2 ]). Adding TM to TG triggered with intermediate TF concentration markedly reduced
ETP and PH, with strong associations with rivaroxaban and apixaban concentrations
(p < 10−4 ) ([Fig. 2 ]). A marked variability in ETP and PH percentages of inhibition was also observed,
reflecting the important variability in the protein C-based negative feedback in very
elderly patients.
Association of Covariates with DOAC TG Peak Height/ETP at T
max and T
min
[Supplementary Table S2 ] (available in the online version) summarizes TG results focused on patient variability
at T
max and at T
min . PH CV confirmed the substantial variability, both at T
max and T
min . We sought to identify covariates, namely DOAC plasma concentrations together with
individual characteristics, potentially associated with PH and ETP measured using
STG-ThromboScreen in the absence of TM at T
max and T
min ([Table 3 ]). In the univariate analysis performed in the rivaroxaban group at T
max , two variables were significantly associated with PH decrease: renal failure assessed
using CrCl (−16.5% per 10 mL/min decrease, p = 0.0016) and heart failure (−36.6%, p = 0.0414). In multivariate analysis, CrCl was the only variable associated with PH
at T
max (p = 0.0085). At T
min , amiodarone intake was associated with PH decrease (−52.2%, p = 0.0116) as well as rivaroxaban concentrations (p = 0.0005), the latter explaining 18.1% of PH variance in univariate analysis.
Table 3
Association of Adage patients' characteristics including xaban plasma concentration and peak height (using
STG-ThromboScreen reagent without thrombomodulin)
Covariates
Univariate
Multivariate
T
max
T
min
T
max
T
min
Effect
%
p
Effect
%
p
Effect
p
Effect
p
Rivaroxaban
Plasma concentrations
+23.9%[a ]
[−11.4; 73.2]
8.4%
0.537
+38.7%[a ]
[12.4;71.1]
22.1%
0.0018
–5.0%[a ]
[−27.9; 25.2]
0.700
+21.0%[a ]
[−4.0; 52.4]
0.103
Creatinine clearance
−12.8%[b ]
[−26.3; 3.0]
22.9%
0.027
−6.1%[b ]
[−16.2; 5.2 ]
3.2%
0.184
−16.4%[b ]
[−26.3; −5.0]
0.0085
−6.76%[b ]
[−15.5; 2.8]
0.156
Heart failure
−37.5%
[−59.8; −2.7]
19.34%
0.041
−31.4%
[−52.2; −1.7]
10.8%
0.055
−22.3%
[−45.0; 9.8]
0.143
−12.7%
[−37.7; 22.4]
0.422
Amiodarone[c ]
N.A
[c ]
N.A
[c ]
.
N.A.
−51.5%
[−70.1; −21.2]
12.7%
0.014
N.A.
[c ]
N.A.
[c ]
−44.6%**
[−75.1; 23.2]
0.228
Apixaban
Plasma concentrations
+80.2%[a ]
10.0; 195.0]
22.6%
0.022
37.1%[a ]
13.1; 66.2]
20.8%
0.0005
+86.8%[a ]
[−0.48; 250.5]
0.052
+32.7%[a ]
[8.28; 62.59]
0.0074
Creatinine clearance
+0.42%[b ]
[4.70; 36.9]
0.08%
0.903
1.7%[b ]
[−10.7; 15.8]
0.49%
0.566
+2.4%[b ]
[−17.9; 27.8]
0.825
−1.9%[b ]
[−13.4; 11.1]
0.753
Heart failure
−26.7%
[−66.7; 61.4]
1.8%
0.426
3.2%
[−30.5; 53.3]
0.01%
0.935
−9.2%
[−62.9; 122.6]
0.826
+0.84%
[−30.8; 46.9]
0.965
Amiodarone[c ]
+3.7%
[−58.7; 160.1]
0.5%
0.936
−37.2%
[−60.0; −1.5]
6.30%
0.043
−26.0%
[−91.8; 564.1]
0.759
−57.47%
[−83.2; 7.9]
0.129
Note: Significance threshold: p < 0.05, bold values.
a For halved concentrations.
b Per decrease of 10 mL/min.
c N.A = not available (number of values from patients on amiodarone at T
max < 5).
The univariate analysis performed in the apixaban group showed a PH decrease significantly
associated with increasing concentrations of apixaban, both at T
max and T
min (p = 0.0224 and p = 0.0005) ([Table 3 ]), accounting for 12.4 and 16.6% of the observed variance, respectively. In addition,
patients receiving amiodarone had a lower PH at T
min than those without amiodarone intake (−37.2%, p = 0.0426). The multivariate analysis confirmed that apixaban concentrations were the
only determinant of the PH variability at T
min (p = 0.0074) and T
max (p = 0.0516) ([Table 3 ]).
Six-Month Follow-Up
We could not obtain data on bleedings and thrombosis in 11 patients (5 on rivaroxaban,
6 on apixaban). The overall rate of major bleedings was 6.0% (rivaroxaban: n = 7 [6.7%]; apixaban: n = 6 [5.4%]); the rate of clinically relevant nonmajor bleeding was 3.3% (rivaroxaban:
n = 2 [1.9%]; apixaban: n = 5 [4.5%]). The main bleeding sites were gastrointestinal (n = 7), intracranial (n = 3), and genitourinary tract (n = 3). Five patients (2.3%) (rivaroxaban: n = 1 [1.0%], apixaban: n = 4 [3.6%]) had a thromboembolic event. Finally, 39 patients died (18.1%), 18 on
rivaroxaban (17.3%) and 21 on apixaban (18.9%), with a median 3.4 month delay (1.8–4.6)
after inclusion. No relationships were found between clinical events and PK/PD parameters.
Discussion
Adage is, to the best of our knowledge, the first study specifically focused on the assessment
of both rivaroxaban/apixaban concentrations and effects evaluated by TG in a real-life
cohort of very elderly NVAF patients. Patients were very old (mean age, 87 years),
had substantial comorbidities and polypharmacy, frequently including the intake of
amiodarone and/or another CYP3A4/5 or P-gp inhibitor. Thus, we believe that the characteristics
of the participants in Adage study are representative of the complex medical issues associated with anticoagulant
treatment in NVAF patients of 80 years and over.
One strength of the Adage study was to provide DOAC PK data which are scarce in this specific setting of hospitalized
very elderly patients.[28 ]
[29 ] Indeed, knowing the extent of DOAC concentrations in octo- and nonagenarians should
facilitate the interpretation of DOAC concentrations when monitoring such patients
in specific situations. Mean C
max and C
min of Adage patients receiving rivaroxaban 15 mg were remarkably close to those reported in Rocket-AF patients (mean age 73 years) receiving 20 mg,[7 ] whereas mean concentrations of Adage patients receiving 20 mg o.d. were higher, especially at T
max . Moreover, we observed a larger inter-individual variability of C
max and C
min for both regimens compared to Mueck et al's data, i.e., around 30%.[7 ] Our data show similar or even greater CVs compared to those of previous observational
studies conducted in patients rather younger than ours.[21 ]
[22 ]
[23 ]
[25 ]
[26 ]
[27 ]
[42 ]
[43 ]
Interestingly, we found that the intake of amiodarone, which is both a CYP3A4/5 and
a P-gp inhibitor, as well as the additional intake of another CYP3A4/5 modulator,
significantly contributed to the increase of rivaroxaban trough levels, explaining
part of the concentration variability consistently with data from pivotal trials.[20 ] Besides, renal function is known to influence the clearance of rivaroxaban, justifying
dose adjustment in case of moderate renal impairment: we only observed a trend towards
higher concentrations at T
min with low CrCl, suggesting that the 25% rivaroxaban dose reduction was appropriate
in our patients.
Regarding apixaban, Adage patients receiving 2.5 mg b.i.d. had a median concentration at T
max close to that of patients receiving 5 mg b.i.d., while the residual concentration
in Adage patients was close to that reported in the Summary of Product Characteristics for
2.5 mg b.i.d.; patients receiving 5 mg b.i.d. had higher median C
max and C
min .[11 ] In addition, inter-individual variabilities were higher than those reported in SmPC
(i.e., around 30% from the ARISTOTLE data set, mean age 70 years) and in previous
studies.[11 ]
[18 ]
[19 ]
[23 ]
[24 ]
[27 ]
[28 ]
[29 ]
[42 ]
[43 ] Interestingly, the apixaban dose regimen was the only covariate that was significantly
associated with plasma C
max and C
min in multivariate analysis, in contrast to rivaroxaban; moreover, the apixaban regimen
subsequently significantly impacted TG peak height. If the apixaban dose regimen had
been given according to the patient characteristics, one would have expected similar
concentrations as for rivaroxaban. One explanation is that the apixaban reduced dose
(2.5 mg b.i.d.) corresponds to half of the full dose (5 mg bid), while the rivaroxaban
reduced dose (15 mg) corresponds to three-quarters of the full dose (20 mg). It is
important to note that a substantial proportion of patients were theoretically underdosed
for both xabans (31% of patients on apixaban, 27% on rivaroxaban). The fear of hemorrhagic
complications probably explains the higher proportion of off-label reduced dose among
Adage patients compared with those described in real-life registries.[44 ]
[45 ]
[46 ]
[47 ] Herein, we showed that the impact of the DOAC regimen choice, reduced versus full
dose, on plasma concentrations and PD parameters was higher with apixaban than with
rivaroxaban. However, our study was underpowered to conclude whether dose reduction
impacts clinical outcomes: this deserves specific future investigations in geriatrics.
Among covariates, female sex was significantly associated with high apixaban C
max in univariate analysis, as already reported by Frost et al in a PK study[19 ] and real-life studies.[48 ] Moreover, we also showed that the intake of amiodarone was a significant determinant
of the DOAC variability of C
max in univariate analysis. However, although Adage patients were highly polymedicated with several co-prescribed CYP3A4/5 and/or P-gp
inhibitors, no strong association was shown with DOAC concentrations in multivariate
analysis within the limits of the study. Our results reinforce the literature data
suggesting that rivaroxaban and apixaban do not interact with those drugs to the extent
of a clinically relevant impact justifying a subsequent dose reduction. It should
be noted that no patients were taking strong CYP3A4/5/P-gp inhibitors or inducers,
highlighting the compliance with DOAC contraindications in this cohort.
One of the original aims of our work was to explore the potential influence of pharmacogenetic
factors on DOAC concentrations. We showed that the presence of CYP3A5*3 and CYP2J2*7 variants was not significantly associated with the variability of rivaroxaban or
apixaban, in agreement with previous studies.[48 ]
[49 ]
[50 ]
[51 ] We observed a trend to higher concentrations in ABCB1-3435T carriers: interestingly, this variant has been shown to be associated with a reduced
risk of thromboembolic outcomes in a large Finnish cohort study.[52 ] Interestingly, an impact of an ABCG2 variant coding breast resistance coding protein was reported in a younger cohort
and could be investigated in Adage patients.[50 ]
[53 ]
Another strength of our study was to evaluate the TG profile of rivaroxaban and apixaban
in a subset of Adage patients using a standardized TG analyzer (ST-Genesia). We confirmed that both xabans
exerted a concentration-dependent effect on TG parameters using both TF conditions;
PHs were consistently more affected than ETP in agreement with previous findings in
various series of much younger patients.[31 ]
[35 ]
[54 ]
[55 ]
[56 ]
[57 ] The optimal choice between intermediate (STG-ThromboScreen) and high (STG-DrugScreen)
pM-TF concentrations may depend on DOAC concentration, preferentially intermediate
up to 300 ng/mL and high above 300 ng/mL.[57 ] Remarkably, we were able to evidence a high inter-individual variability of TG parameters
among patients receiving both full and reduced-dose xabans. Spiking pooled normal
plasma with increasing concentrations of xabans allowed a better understanding of
the inter-individual variability and highlighted an underlying hypercoagulable state
in some Adage patients, leading to shorter LT and TTP combined with higher PH and ETP. We had primarily
checked that the pooled normal plasma contained normal procoagulant factors and natural
coagulation inhibitor levels.[35 ] Moreover, for a given concentration, we observed under the same conditions higher
PH and ETP results in Adage rivaroxaban patients when compared to those obtained in Driving rivaroxaban healthy volunteers, aged 18 to 45 years, thus supporting this hypothesis.[35 ]
[57 ] It is well known that advancing age and chronic comorbid conditions lead to an imbalance
between procoagulant factors and coagulation natural inhibitors, potentially causes
of hypercoagulability.[58 ] This issue regarding TG variability deserves further studies in geriatric setting.
The overall major bleeding rate of 6.0% at 6 months was not surprisingly higher than
that reported in phase III trials in patients >75 years given the characteristics
of Adage patients.[8 ]
[9 ] The predominance of gastrointestinal bleedings could be expected in these patients,
as shown in previous studies and meta-analysis.[13 ]
[14 ]
[15 ]
[59 ] The mortality rate (18.1%) was similar in patients treated with rivaroxaban and
apixaban. The recruitment of Adage patients in geriatrics units probably reflects a high degree of frailty in Adage patients.
Our study has several limitations. First, although one to five samples per patient
were planned over a 20-day period with an average of three samples per patient, a
smaller number of blood samples were drawn ( i.e., 2.1) because of short hospital stays. Moreover, the sample size clearly limits
analyses of associations of outcomes with measures. Second, the results regarding
the influence of individual factors on concentrations at T
max and T
min and PD obtained in this exploratory study must be confirmed in an external cohort
of hospitalized patients. Third, most patients were recruited during hospitalization:
these results cannot be extrapolated to stable outpatients of similar age even though
CIRS-scores took into account both chronic and acute comorbid conditions. Finally,
our study was not designed to detect possible associations between PK/PD parameters
measured only over a short period at inclusion and clinical events at 6 months.
In conclusion, our study provides new real-life data regarding DOAC concentrations
and TG inter-individual variability in elderly in-patients with NVAF, multiple comorbidities,
and medications. The DOAC regimen choice (reduced vs. full dose) was significantly
associated with plasma concentrations and TG PHs for apixaban, but not for rivaroxaban.
The underlying coagulation status of elderly patients could affect TG profiles. Further
research is needed to better understand the substantial PK and PD variability and
its potential clinical impact in the rapidly growing population of very old patients
with multiple conditions who need long-term anticoagulant treatment.
What is known about this topic?
Very elderly patients (≥80 years) are characterized by numerous comorbid conditions
(e.g., renal impairment) and polypharmacy, potentially altering drug PK/PD.
Few specific DOAC pharmacokinetics (PK) and pharmacodynamics (PD) data are available
in this age group with numerous comorbid conditions and polypharmacy.
What does this paper add?
Our academic prospective study provides original data in very elderly in-patients
receiving xaban in real-life setting, showing great inter-individual variability in
plasma concentrations and PD parameters.
Substantial variability of thrombin peak-heights was noticed at a given plasma concentration
for both xabans, suggesting an impact of the underlying coagulation status of elderly
in-patients.
The dose regimen (reduced dose vs. full dose) had a significant impact on plasma concentrations
at T
max and T
min in apixaban, but not in rivaroxaban patients.
