Effect of Aspirin Coadministration on the Safety of Celecoxib, Naproxen, or Ibuprofen - C5Research

JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY                                                                                 VOL. 71, NO. 16, 2018

                              ª 2018 THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION.

                              PUBLISHED BY ELSEVIER. ALL RIGHTS RESERVED.

                              Effect of Aspirin Coadministration
                              on the Safety of Celecoxib,
                              Naproxen, or Ibuprofen
                              Grant W. Reed, MD, MSC,a,b Mouin S. Abdallah, MD, MSC,a,b Mingyuan Shao, MS,a Kathy Wolski, MPH,a
                              Lisa Wisniewski, RN,a Neville Yeomans, MD,c Thomas F. Lüscher, MD,d Jeffrey S. Borer, MD,e David Y. Graham, MD,f
                              M. Elaine Husni, MD, MPH,g Daniel H. Solomon, MD, MPH,h Peter Libby, MD,h Venu Menon, MD,a,b
                              A. Michael Lincoff, MD,a,b Steven E. Nissen, MDa,b


                                    BACKGROUND The safety of nonsteroidal anti-inflammatory drug (NSAID) and aspirin coadministration is uncertain.

                                    OBJECTIVES The aim of this study was to compare the safety of combining NSAIDs with low-dose aspirin.

                                    METHODS This analysis of the PRECISION (Prospective Randomized Evaluation of Celecoxib Integrated Safety Versus
                                    Ibuprofen or Naproxen) trial included 23,953 patients with osteoarthritis or rheumatoid arthritis at increased cardio-
                                    vascular risk randomized to celecoxib, ibuprofen, or naproxen. The on-treatment population was used for this study.
                                    Outcomes included composite major adverse cardiovascular events, noncardiovascular death, gastrointestinal or renal
                                    events, and components of the composite. Cox proportional hazards models compared outcomes among NSAIDs strat-
                                    ified by aspirin use following propensity score adjustment. Kaplan-Meier analysis was used to compare the cumulative
                                    probability of events.

                                    RESULTS When taken without aspirin, naproxen or ibuprofen had greater risk for the primary composite endpoint
                                    compared with celecoxib (hazard ratio [HR]: 1.52; 95% confidence interval [CI]: 1.22 to 1.90, p
1742         Reed et al.                                                                                                                                JACC VOL. 71, NO. 16, 2018

             Safety of Combined Aspirin and NSAID Use                                                                                                        APRIL 24, 2018:1741–51

ABBREVIATIONS                                       oncomitant         use     of    nonsteroidal          Randomized           Evaluation        of    Celecoxib     Integrated
AND ACRONYMS                                        anti-inflammatory drugs (NSAIDs)                        Safety Versus Ibuprofen or Naproxen) trial to evaluate
                                                    and aspirin is widespread. In 2010,                    the safety of combining aspirin (a selective COX-1
APTC = Antiplatelet Trialists’
                                       >43 million U.S. adults were regular aspirin                        inhibitor) with celecoxib (a selective COX-2 inhibi-
                                       users, and >28 million took NSAIDs regularly                        tor), naproxen (a nonselective COX-1 > COX-2 inhib-
CAD = coronary artery disease
                                       (1,2). An estimated 20% to 50% of patients                          itor), or ibuprofen (a nonselective COX-2 > COX-1
CI = confidence interval
                                       with osteoarthritis or rheumatoid arthritis                         inhibitor).
COX = cyclooxygenase
                                       take aspirin daily (3,4). The coadministration
GI = gastrointestinal
                                       of NSAIDs and aspirin has raised safety con-
HR = hazard ratio
                                       cerns, because both inhibit synthesis of pros-
IPTW = inverse probability of          tanoids in tissues in which these lipid
treatment weights                                                                                          STUDY POPULATION. This was a post hoc analysis of
                                       mediators        may      have      protective       effects.       the PRECISION trial. The design, rationale, and pri-
MACE = major adverse
                                       Furthermore,          nonselective           NSAIDs       may       mary results of PRECISION have been previously
cardiovascular event(s)
                                       compete with aspirin for its binding site on                        published (3,21). Briefly, PRECISION was a multi-
NSAID = nonsteroidal anti-
inflammatory drug                       cyclooxygenase (COX)–1 (5,6), blocking aspi-                        center, randomized controlled trial of patients with
                                       rin’s ability to acetylate a serine residue on                      osteoarthritis or rheumatoid arthritis on long-term
                           COX-1 necessary for platelet inhibition (6,7).                                  NSAIDs at increased cardiovascular risk that demon-
                                                        SEE PAGE 1752                                      strated the noninferiority of moderate-dose celecoxib
                                                                                                           (100 to 200 mg twice daily) to naproxen (375 to
                                 Observational studies and meta-analyses suggest                           500 mg twice daily) and ibuprofen (600 to 800 mg 3
                           that both NSAIDs nonselective for COX-1 or COX-2                                times daily) with regard to cardiovascular safety. The
                           and NSAIDs selective for COX-2 have adverse ef-                                 PRECISION trial protocol pre-specified an analysis
                           fects, including cardiovascular, gastrointestinal (GI),                         stratified by aspirin use a priori. However, the sta-
                           and renal events, compared with placebo (8–13).                                 tistical methodology used to accomplish this was
                           Studies have reported conflicting results with regard                            established post hoc.
                           to whether adding aspirin to an NSAID modifies these                                 As previously reported, in PRECISION, patients
                           risks (14–19). Likewise, there have been reports of                             were randomized according to their primary diag-
                           aspirin failure in patients with inflammatory disor-                             nosis (osteoarthritis or rheumatoid arthritis), and
                           ders, particularly among patients on NSAIDs (20). The                           stratified by low-dose aspirin (#325 mg) use at base-
                           relative safety of combining aspirin and NSAIDs with                            line to ensure equal distribution among NSAIDs. The
                           various degrees of COX-1 and COX-2 inhibition is not                            study protocol did not permit doses of aspirin
                           well understood, presenting challenges to providing                             >325 mg. All subjects were provided with once-daily
                           patient recommendations on this subject.                                        esomeprazole 20 to 40 mg as a gastroprotective
                                 Accordingly, we performed a propensity score–                             agent regardless of aspirin use. All patients gave
                           adjusted substudy of the PRECISION (Prospective                                 informed consent before enrollment in the study.

                           membership on the adjudication committee in the ARIVE trial. Dr. Borer served as chair of a data and safety monitoring board for
                           an unrelated product being developed by Pfizer; has served as a consultant, trial executive committee member, data and safety
                           monitoring board member, or cardiac event adjudication committee member for unrelated products for Amgen, Novartis,
                           AstraZeneca, Takeda, Biotronik, Servier, GlaxoSmithKline, Gilead, and ARMGO; and owns stock in BioMarin and ARMGO. Dr.
                           Graham is a consultant for RedHill Biopharma regarding novel Helicobacter pylori therapies; has received research support from
                           RedHill Biopharma for culture of H. pylori; is the principal investigator of an international study of the use of antimycobacterial
                           therapy for Crohn’s disease; and is a consultant for BioGaia in relation to probiotic therapy for H. pylori infection and for Takeda in
                           relation to H. pylori therapies. Dr. Husni has received an institution grant to perform the PRECISION trial; has received a Sanofi
                           Genzyme investigator grant; and has served on advisory boards for AbbVie, Bristol-Myers Squibb, Amgen, UCB, Regeneron, and
                           Janssen. Dr. Solomon has received a research grant from Pfizer for unrelated work; and has received royalties from UpToDate for a
                           chapter about NSAIDs. Dr. Libby has been an unpaid consultant to or has been involved in clinical trials for Amgen, AstraZeneca,
                           Esperion Therapeutics, Ionis Pharmaceuticals, Kowa Pharmaceuticals, Merck, Novartis, Pfizer, Sanofi-Regeneron, Takeda Phar-
                           maceuticals, and XBiotech; has served as a member of scientific advisory boards for Amgen, Corvidia Therapeutics, DalCor
                           Pharmaceuticals, Kowa Pharmaceuticals, Olatec Therapeutics, Medimmune, and Novartis; and his laboratory has received
                           research funding in the past 2 years from Novartis. Dr. Menon has received grant support to the institution to perform the
                           PRECISION trial. Dr. Lincoff has received grant support to the institution to perform the PRECISION trial; is a consultant for
                           Amgen, Novo, Nordisk, Sanofi, Abbott, Sarpeta, and Sermonix; and has received research grants to his institution from Pfizer,
                           AstraZeneca, Esperion, AbbVie, Eli Lilly, and Roche. Dr. Nissen has received grant support to the institution to perform the
                           PRECISION trial. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.

                           Manuscript received January 15, 2018; accepted February 6, 2018.
JACC VOL. 71, NO. 16, 2018                                                                                                                           Reed et al.   1743
APRIL 24, 2018:1741–51                                                                                                  Safety of Combined Aspirin and NSAID Use

                                                                                    myocardial infarction, nonfatal stroke, coronary
 T A B L E 1 Patient Characteristics Stratified by
 Aspirin Use at Randomization
                                                                                    revascularization, or hospitalization for unstable
                                                                                    angina or transient ischemic attack. Clinically signif-
                                   No Aspirin           Aspirin
                                  (n ¼ 12,935)       (n ¼ 11,018)       p Value
                                                                                    icant GI events included acute gastric or intestinal
 Age, yrs                          61.6  9.6         65.1  8.8
1744   Reed et al.                                                                                                  JACC VOL. 71, NO. 16, 2018

       Safety of Combined Aspirin and NSAID Use                                                                          APRIL 24, 2018:1741–51

                     from the imputed and nonimputed datasets were                  package in R version 3.4.0 (R Foundation for Statis-
                     very similar.                                                  tical Computing, Vienna, Austria). The plot of abso-
                       To adjust for differences between patients taking            lute   standardized    differences   was   made     using
                     aspirin and not, propensity scores for aspirin treat-          SigmaPlot version 11.0 (Systat, San Jose, California).
                     ment and corresponding stabilized inverse probabil-
                     ity of treatment weights (IPTWs) were calculated
                     using a universal binary logistic regression model
                                                                                    PATIENT CHARACTERISTICS. This study included
                     including all of the baseline characteristics as cova-
                                                                                    23,953 patients (24,222 patients were randomized in
                     riates (23), in each of the imputed datasets for each
                                                                                    PRECISION, but 141 patients were excluded because
                     endpoint. Improvement in the balance of baseline
                                                                                    of enrollment errors and 128 patients did not have
                     characteristics was assessed by evaluating a plot of
                                                                                    data on aspirin use); 8,030 were assigned to celecoxib,
                     the absolute standardized differences with and
                                                                                    7,933 to naproxen, and 7,990 to ibuprofen. Aspirin was
                     without IPTW (Online Figure 1). An absolute value in
                                                                                    used by 11,018 patients (46.0%) at randomization;
                     standardized differences of
JACC VOL. 71, NO. 16, 2018                                                                                                                          Reed et al.   1745
APRIL 24, 2018:1741–51                                                                                                 Safety of Combined Aspirin and NSAID Use

   F I G U R E 1 Outcomes in Non–Aspirin Users on Naproxen or Ibuprofen Compared With Celecoxib

                                                                                                                             (N = 4,239)
                                                                                                             Celecoxib        Ibuprofen
           Endpoints                       Adjusted HR (95% CI)                                             (N = 4,347)      (N = 4,295)

           Composite Safety Endpoint
            Naproxen vs. Celecoxib                                                    1.52 (1.22, 1.90)      132 (3.0)        195 (4.5)
            Ibuprofen vs. Celecoxib                                                   1.81 (1.46, 2.26)      132 (3.0)        213 (5.0)

           Extended MACE
            Naproxen vs. Celecoxib                                                    1.11 (0.83, 1.49)       90 (2.1)         96 (2.2)
            Ibuprofen vs. Celecoxib                                                   1.35 (1.02, 1.78)       90 (2.1)        109 (2.5)

           APTC MACE
            Naproxen vs. Celecoxib                                                    1.20 (0.84, 1.72)        57 (1.3)         67 (1.6)
            Ibuprofen vs. Celecoxib                                                   1.47 (1.04, 2.08)        57 (1.3)         74 (1.7)

           GI Events
            Naproxen vs. Celecoxib                                                    2.60 (1.59, 4.27)       22 (0.5)          56 (1.3)
            Ibuprofen vs. Celecoxib                                                   3.20 (1.97, 5.22)       22 (0.5)          64 (1.5)

           Renal Events
            Naproxen vs. Celecoxib                                                    2.09 (1.10, 3.96)       19 (0.4)         33 (0.8)
            Ibuprofen vs. Celecoxib                                                   1.93 (1.00, 3.73)       19 (0.4)         28 (0.7)

                                         0.71 1.0     1.41   2.0                6.0

                                      Favors        Favors
                          Naproxen/Ibuprofen        Celecoxib

   The number of events (percentage of total) is reported. APTC ¼ Antiplatelet Trialists’ Collaboration; CI ¼ confidence interval;
   GI ¼ gastrointestinal; HR ¼ hazard ratio; MACE ¼ major adverse cardiovascular event.

and 5.0% vs. 3.0%, respectively; naproxen vs. cele-                        common with celecoxib compared with the other
coxib adjusted HR: 1.52; 95% CI: 1.22 to 1.90;                             NSAIDs (Figures 2A and 2C).
p < 0.001; ibuprofen vs. celecoxib HR: 1.81; 95% CI:                       OUTCOMES           OF     CELECOXIB,           NAPROXEN,         AND
1.46 to 2.26; p < 0.001) (Figure 1). These results                         IBUPROFEN          COMBINED          WITH       ASPIRIN. Patients
derived primarily from a greater risk for GI events                        taking ibuprofen plus aspirin had greater risk for the
with either naproxen             compared with celecoxib                   composite safety endpoint compared with celecoxib
(1.3% vs. 0.5%; adjusted HR: 2.60; 95% CI: 1.59 to                         plus aspirin (7.1% vs. 6.0%; adjusted HR: 1.27; 95% CI:
4.27; p < 0.001) and ibuprofen compared with cele-                         1.06 to 1.51; p ¼ 0.01) (Figure 3). However, there was
coxib (1.5% vs. 0.5%; adjusted HR: 3.20; 95% CI: 1.97                      no difference in extended MACE or APTC-defined
to 5.22; p < 0.001). However, when comparing                               MACE between NSAIDs with the addition of aspirin.
ibuprofen with celecoxib, there was also slightly                          The difference in the composite safety endpoint was
higher risk for extended MACE (adjusted HR: 1.35;                          driven by an increased risk for GI events with
95% CI: 1.02 to 1.78; p ¼ 0.039) and APTC-defined                           ibuprofen plus aspirin compared with celecoxib plus
MACE (adjusted HR: 1.47; 95% CI: 1.04 to 2.08;                             aspirin (1.4% vs. 0.9%; adjusted HR: 1.71; 95% CI: 1.10
p ¼ 0.031), and compared with celecoxib, naproxen                          to 2.67; p ¼ 0.017), which was also observed with
was associated with a greater risk for renal events                        naproxen plus aspirin compared with celecoxib plus
(adjusted HR: 2.09; 95% CI: 1.10 to 3.96; p ¼ 0.024),                      aspirin (1.6% vs. 0.9%; adjusted HR: 1.91; 95% CI: 1.24
while excess risk with ibuprofen was borderline sig-                       to 2.94; p ¼ 0.003). Likewise, the hazard of renal
nificant (adjusted HR: 1.93; 95% CI: 1.00 to 3.73;                          events was greater with ibuprofen plus aspirin
p ¼ 0.052).                                                                compared with celecoxib plus aspirin (1.2% vs. 0.6%;
   On adjusted Kaplan-Meier analysis, the composite                        adjusted HR: 2.01; 95% CI: 1.23 to 3.30; p ¼ 0.005), but
safety endpoint and extended MACEs occurred least                          there was no difference with naproxen plus aspirin
frequently with celecoxib compared with naproxen or                        compared with celecoxib plus aspirin. Event rates
ibuprofen (Central Illustration) (log-rank p < 0.0001                      tended to be higher for all endpoints with aspirin
and p ¼ 0.0012, respectively). Similarly, GI events                        addition, regardless of which NSAID was used
(p < 0.0001) and renal events (p ¼ 0.0005) were least                      (Figures 1 and 3).
1746      Reed et al.                                                                                                                                               JACC VOL. 71, NO. 16, 2018

          Safety of Combined Aspirin and NSAID Use                                                                                                                      APRIL 24, 2018:1741–51

   C E NT R AL IL L U STR AT IO N Safety of Combined Aspirin and Nonsteroidal Anti-Inflammatory Drug Use

                                                                        No Aspirin                                                                            Aspirin
                          A                                                                                   B
                                              1.00                                                                                1.00
                                                                                 Log rank p < 0.0001                                                                Log rank p < 0.0001
                                              0.98                                                                                0.98
                        Event-Free Survival

                                                                                                            Event-Free Survival
                                              0.96                                                                                0.96

       Composite                              0.94                                                                                0.94
         Safety                               0.92                                                                                0.92
        Endpoint                              0.90                                                                                0.90
                                              0.88                                                                                0.88
                                              0.86                                                                                0.86
                                              0.84                                                                                0.84
                                                     0      250         500     750    1000       1250                                   0        250      500     750    1000     1250
                                                                     Time to Event (Days)                                                               Time to Event (Days)

                          C                                                                                   D
                                              1.00                                                                                1.00
                                                                                 Log rank p = 0.0012                                                                Log rank p = 0.5506

                                              0.98                                                                                0.98
                        Event-Free Survival

                                                                                                            Event-Free Survival

                                              0.96                                                                                0.96
                                              0.94                                                                                0.94

                                              0.92                                                                                0.92

                                              0.90                                                                                0.90
                                                     0      250         500     750    1000       1250                                   0        250      500     750    1000     1250
                                                                     Time to Event (Days)                                                               Time to Event (Days)

                                                                                         Celecoxib              Naproxen                          Ibuprofen

   Reed, G.W. et al. J Am Coll Cardiol. 2018;71(16):1741–51.

   Adjusted Kaplan-Meier curves for the composite safety endpoint and extended major adverse cardiovascular event (MACE), stratified by aspirin use. Among patients
   not on aspirin, celecoxib had the lowest incidence of composite safety events and extended MACE. The addition of aspirin attenuated the relative safety benefit of

                                              Furthermore,       adjusted     Kaplan-Meier       analysis                    CAD AND ASPIRIN COMPLIANCE SENSITIVITY ANALYSES.
                        demonstrated that the addition of aspirin attenuated                                                 In a pre-specified sensitivity analysis, comparing pa-
                        the benefit of celecoxib compared with naproxen and                                                   tients with CAD with those without, the relative
                        ibuprofen with regard to the composite safety                                                        hazard          of   the   composite   safety   endpoint     was
                        endpoint and extended MACE (Central Illustration).                                                   approximately the same whether aspirin was also
                        Likewise, in the presence of aspirin, there was less                                                 used (adjusted HR: 2.23; 95% CI: 1.90 to 2.60;
                        differentiation                         in   renal   events   among      NSAIDs                      p < 0.001) or aspirin was not used (adjusted HR: 2.01;
                        (Figure 2D). However, GI events remained least                                                       95% CI: 1.56 to 2.59; p < 0.001), demonstrating no
                        frequent                         with   celecoxib    (log-rank   p   ¼    0.004)                     significant interaction between CAD and aspirin use
                        (Figure 2B).                                                                                         (Online Figure 3) (interaction p ¼ 0.35).
JACC VOL. 71, NO. 16, 2018                                                                                                                                                       Reed et al.     1747
APRIL 24, 2018:1741–51                                                                                                                           Safety of Combined Aspirin and NSAID Use

   F I G U R E 2 Adjusted Kaplan-Meier Curves for the Cumulative Incidence of Gastrointestinal and Renal Events, Stratified by Aspirin Use

                   A                                           No Aspirin                                    B                                                 Aspirin
                                         1.00                                                                                      1.00
                                                                         Log rank p < 0.0001                                                                          Log rank p = 0.0040

                                         0.99                                                                                      0.99
                   Event-Free Survival

                                                                                                             Event-Free Survival
                                         0.98                                                                                      0.98
      GI Events
                                         0.97                                                                                      0.97

                                         0.96                                                                                      0.96

                                         0.95                                                                                      0.95
                                                0   250       500     750      1000        1250                                           0       250       500      750     1000         1250
                                                           Time to Event (Days)                                                                           Time to Event (Days)

          Celecoxib                         4410    3075      2336    1794        1299     997        Celecoxib                       3670       2561      1993     1529     1139          872
          Ibuprofen                         4283    2830      2081    1573        1187     906        Ibuprofen                       3674       2492      1853     1406     1049          795
          Naproxen                          4312    2972      2208    1681        1279     971        Naproxen                        3623       2542      1956     1453     1059          819

                   C                                                                                         D
                                         1.00                                                                                      1.00
                                                                       Log rank p = 0.0005                                                                             Log rank p = 0.0546
                   Event-Free Survival

                                                                                                             Event-Free Survival

                                         0.99                                                                                      0.99


                                         0.98                                                                                      0.98

                                         0.97                                                                                      0.97
                                                0   250       500     750      1000        1250                                           0       250       500      750     1000         1250
                                                           Time to Event (Days)                                                                           Time to Event (Days)

          Celecoxib                         4410    3077      2345   1799         1302     1000       Celecoxib                       3670       2561      1993      1528     1142         875
          Ibuprofen                         4283    2830      2087   1586         1195      907       Ibuprofen                       3674       2491      1858      1410     1057         804
          Naproxen                          4312    2933      2212   1685         1282      974       Naproxen                        3623       2547      1963      1463     1065         826

                                                                                          Celecoxib      Naproxen                             Ibuprofen

   The addition of aspirin attenuated the gastrointestinal (GI) and renal safety of celecoxib, although GI events were still less frequent with celecoxib than ibuprofen
   or naproxen.

   Only 304 patients (1.3%) discontinued aspirin after                                   modified by concomitant aspirin use. Specifically,
randomization, and 964 patients (4.0%) who were not                                      celecoxib was associated with a more favorable
on aspirin started aspirin during the study (this was                                    overall safety profile than naproxen or ibuprofen
equally    distributed                          among   the   NSAIDs).      When         among regular NSAID users not taking aspirin.
excluding these patients, the results for the primary                                    However, the addition of aspirin attenuated the
endpoint were not significantly different.                                                relatively favorable safety profile of celecoxib, and
                                                                                         although there was still a slight advantage of cele-
DISCUSSION                                                                               coxib over ibuprofen with regard to composite
                                                                                         safety events, there were no longer differences in
This study demonstrates that the relative cardio-                                        extended MACE or APTC-defined MACE. However,
vascular and overall safety of NSAID therapy is                                          celecoxib was still associated with fewer GI events
1748   Reed et al.                                                                                                                       JACC VOL. 71, NO. 16, 2018

       Safety of Combined Aspirin and NSAID Use                                                                                                APRIL 24, 2018:1741–51

                        F I G U R E 3 Outcomes in Patients on Naproxen or Ibuprofen Compared With Celecoxib, With Combined Aspirin

                                                                                                                                                (N = 3,640)
                                                                                                                                 Celecoxib       Ibuprofen
                               Endpoints                       Adjusted HR (95% CI)                                             (N = 3,683)     (N = 3,695)

                               Composite Safety Endpoint
                                Naproxen vs. Celecoxib                                                     1.18 (0.98, 1.41)    222 (6.0)        249 (6.8)
                                Ibuprofen vs. Celecoxib                                                    1.27 (1.06, 1.51)    222 (6.0)         264 (7.1)

                               Extended MACE
                                Naproxen vs. Celecoxib                                                     1.05 (0.84, 1.31)     157 (4.3)        157 (4.3)
                                Ibuprofen vs. Celecoxib                                                    1.19 (0.96, 1.47)     157 (4.3)        175 (4.7)

                               APTC MACE
                                Naproxen vs. Celecoxib                                                     1.04 (0.76, 1.43)      77 (2.1)         77 (2.1)
                                Ibuprofen vs. Celecoxib                                                     1.10 (0.81, 1.51)     77 (2.1)         81 (2.2)

                               GI Events
                                Naproxen vs. Celecoxib                                                     1.91 (1.24, 2.94)      32 (0.9)         59 (1.6)
                                Ibuprofen vs. Celecoxib                                                     1.71 (1.10, 2.67)     32 (0.9)         51 (1.4)

                               Renal Events
                                Naproxen vs. Celecoxib                                                     1.30 (0.76, 2.23)     23 (0.6)         29 (0.8)
                                Ibuprofen vs. Celecoxib                                                    2.01 (1.23, 3.30)     23 (0.6)          45 (1.2)

                                                             0.71     1.0     1.41                   3.5

                                                           Favors        Favors
                                               Naproxen/Ibuprofen        Celecoxib

                        The number of events (percentage of total) is reported. Abbreviations as in Figure 1.

                     than ibuprofen or naproxen and fewer renal events                          by aspirin use, that analysis was in the intention-to-
                     than ibuprofen.                                                            treat population. The present study was better
                       The primary findings of PRECISION were reported                           equipped to assess the effect of aspirin on outcomes,
                     in the intention-to-treat population, in which pa-                         as it was a propensity score–adjusted analysis of
                     tients were grouped on the basis of the study drug to                      overall harms in the on-treatment population and
                     which they were assigned at randomization. In that                         thus more relevant for the reasons described.
                     analysis, events were counted regardless of whether                            Our findings do not support the premise that se-
                     the study drug was stopped during the study. How-                          lective COX-2 inhibitors as a class increase cardio-
                     ever, in studies of drug safety, the on-treatment                          vascular risk compared with nonselective COX-1 and
                     population can be more informative, as it considers                        COX-2 inhibitors (27). On the contrary, in the pri-
                     events only while patients are actually taking the                         mary results from the PRECISION trial, selective COX-2
                     drug, and patients are censored a period of time after                     inhibition        with     celecoxib     was     noninferior     for
                     the study drug is stopped (30 days in the present                          cardiovascular safety to nonselective COX-2 > COX-1
                     analysis). This consideration has particular relevance                     inhibition with ibuprofen or COX-1 > COX-2 inhibi-
                     to studies of patients with pain, as these patients                        tion with naproxen in the intention-to-treat popula-
                     discontinue medications frequently in favor of trying                      tion. The present analysis of the on-treatment
                     other analgesic drugs. In this case, the intention-to-                     population is clinically more relevant with regard to
                     treat analysis actually reflects the effects of these                       safety endpoints and showed the most favorable car-
                     other medications, while effects of the study drug are                     diovascular safety profile in patients with the selective
                     better reflected in an on-treatment analysis. Indeed,                       COX-2 inhibitor celecoxib alone. Adding COX-1 inhi-
                     in PRECISION, study drug discontinuation was more                          bition with aspirin attenuated the cardiovascular
                     common than expected, as previously reported (see                          safety advantage of celecoxib and rendered the rela-
                     PRECISION Trial Supplemental Appendix Figure S2)                           tive cardiovascular safety profiles of the NSAIDs
                     (3). Furthermore, although the main findings of                             approximately equivalent. These findings support the
                     PRECISION did contain a subgroup analysis stratified                        hypothesis and main findings of the PRECISION trial
JACC VOL. 71, NO. 16, 2018                                                                                                Reed et al.   1749
APRIL 24, 2018:1741–51                                                                      Safety of Combined Aspirin and NSAID Use

that the increased cardiovascular risk observed with       adjust for baseline characteristics (Online Figure 1).
rofecoxib is not a COX-2 inhibition class effect.          Our use of this technique allowed adjustment for
Furthermore, our results suggest that celecoxib has a      baseline characteristics, creating a “pseudo-random-
more favorable cardiovascular safety profile than           ization” result, increasing the validity to our results.
ibuprofen or naproxen among patients not on aspirin        To the best of our knowledge, ours is the only study
and that the cardiovascular safety profile of celecoxib     on this topic to use such a technique. We further
is noninferior to ibuprofen or naproxen among aspirin      incorporated multivariate adjustment into propensity
users.                                                     score–weighted survival regression analysis, a so-
   Our findings underscore the importance of appro-         called doubly robust adjustment, to reduce residual
priate patient counseling on the relative safety profile    confounding as much as possible (33).
of NSAIDs when initiating therapy. Although short-
                                                           STUDY LIMITATIONS. The strengths and limitations
term NSAID use is likely safe (28), long-term use of
                                                           of the PRECISION trial have been previously dis-
any NSAID has been associated with increased car-
                                                           cussed (3). Although the analysis was pre-specified in
diovascular, GI, and renal risk compared with placebo
                                                           the trial protocol, the study was not designed to
in observational studies (9,29,30). However, if an
                                                           detect an interaction between the study NSAIDs and
NSAID is required, the relative safety of the various
                                                           aspirin. Accordingly, the present analysis should be
NSAIDs appears to be modified by concomitant
                                                           considered hypothesis generating and needs to be
aspirin use. Physicians administering any NSAID
                                                           confirmed by other studies. There were no compari-
should consider the potential GI and renal hazards of
                                                           sons of ibuprofen and naproxen in this study, though
using combined NSAID and aspirin therapy. It would
                                                           the relative safety of these agents to each other can be
be reasonable to consider gastric protection with an
                                                           inferred via the Kaplan-Meier curves. In the present
H2 blocker or proton pump inhibitor if aspirin is also
                                                           study we did not evaluate outcomes stratified by
used (29).
                                                           NSAID dose, given power limitations. Furthermore,
   Most studies on the comparative safety of NSAIDs
                                                           the NSAID doses in our study were moderate, in
have not investigated a possible interaction between
                                                           accordance with approved labeling in the countries
aspirin and NSAID use and have revealed conflicting
                                                           where the trial was conducted. In addition, data on
results, likely because of their heterogeneous study
                                                           erythrocyte sedimentation rate was not collected, and
designs and potential selection bias (16,17,31). A key
                                                           although baseline levels of C-reactive protein were
meta-analysis of 39 trials including more than 41,000
                                                           reported, this variable was included post hoc and not
patients prior to PRECISION suggested that the inci-
                                                           adjusted for. The present study was not designed to
dence of APTC-defined MACE did not differ in pa-
                                                           evaluate outcomes on the basis of arthritis type
tients   treated    with     celecoxib   compared   with
                                                           (osteoarthritis or rheumatoid arthritis), though a
nonselective NSAIDs regardless of aspirin use. How-
                                                           separate analysis of PRECISION was recently pub-
ever, this analysis did find a lower cardiovascular
                                                           lished addressing this (34).
death rate with celecoxib compared with nonselective
NSAIDs (relative risk: 0.43; 95% CI: 0.19 to 0.95;         CONCLUSIONS
p ¼ 0.04) (14). Other observational studies have
revealed increased cardiovascular risk with ibuprofen      Among patients not taking aspirin, moderate-dose
plus aspirin compared with ibuprofen alone (15,16),        celecoxib is associated with a more favorable overall
while others have shown reduced risk (18) or no            safety profile compared with naproxen or ibuprofen.
change at all (17,32). Our study aimed to illuminate       Combination with aspirin attenuates the safety
these differences. Results of our study support the        advantage    of   celecoxib,   although    celecoxib      is
relative safety of celecoxib compared with naproxen        still associated with fewer GI events than ibuprofen
or ibuprofen with regard to cardiovascular, GI, and        or naproxen and fewer renal events than ibuprofen.
renal events.                                              These results suggest that in many cases, celecoxib
   The present study is a post hoc analysis of a ran-      would be preferred to naproxen or ibuprofen,
domized controlled trial, and as such there were dif-      especially if the patient is not required to take
ferences in patient characteristics when stratified by      aspirin.
aspirin use (Table 1). This is a common issue seen in
most other observational studies of NSAID and aspirin      ADDRESS FOR CORRESPONDENCE: Dr. Steven E.
use and introduces bias that can be addressed only by      Nissen, Department of Cardiovascular Medicine,
a properly designed trial. A strength of the present       Cleveland Clinic, 9500 Euclid Avenue, Desk J2-3,
study is the use of propensity score weighting to          Cleveland, Ohio 44195. E-mail: nissens@ccf.org.
1750   Reed et al.                                                                                                                                      JACC VOL. 71, NO. 16, 2018

       Safety of Combined Aspirin and NSAID Use                                                                                                              APRIL 24, 2018:1741–51


                         COMPETENCY IN PATIENT CARE AND PROCEDURAL                                       TRANSLATIONAL OUTLOOK: The mechanisms by
                         SKILLS: All NSAIDs increase cardiovascular, GI, and renal                       which aspirin and NSAIDs interact with respect to safety
                         risk, but the risk profiles of the various NSAIDs differ and                     and efficacy are complex and probably not confined to
                         are modified by concomitant aspirin use. Celecoxib ex-                           COX specificity. Additional research is needed to elucidate
                         hibits a more favorable overall safety profile than                              these other contributing factors and delineate pharma-
                         ibuprofen or naproxen and is not associated with                                cological strategies that maximize analgesic and cardio-
                         increased cardiovascular risk, whether or not patients                          vascular benefit while minimizing risk.
                         take aspirin concurrently.


                     1. Kaufman    DW,     Kelly   JP,   Rosenberg      L,   anti-inflammatory drugs: nested case-control             antiinflammatory agents in patients with arthritis.
                     Anderson TE, Mitchell AA. Recent patterns of            study. Lancet 2005;365:475–81.                          Am Heart J 2009;157:606–12.
                     medication use in the ambulatory adult population
                                                                             12. McGettigan P, Henry D. Cardiovascular risk and      22. Antiplatelet Trialists’ Collaboration. Collabo-
                     of the United States: the Slone survey. JAMA
                                                                             inhibition of cyclooxygenase: a systematic review       rative overview of randomised trials of antiplatelet
                                                                             of the observational studies of selective and           therapy—I: Prevention of death, myocardial
                     2. Zhou Y, Boudreau DM, Freedman AN. Trends in          nonselective inhibitors of cyclooxygenase 2. JAMA       infarction, and stroke by prolonged antiplatelet
                     the use of aspirin and nonsteroidal anti-               2006;296:1633–44.                                       therapy in various categories of patients. BMJ
                     inflammatory drugs in the general U.S. popula-                                                                   1994;308:81–106.
                                                                             13. Kearney PM, Baigent C, Godwin J, Halls H,
                     tion. Pharmacoepidemiol Drug Saf 2014;23:43–50.
                                                                             Emberson JR, Patrono C. Do selective cyclo-             23. Wyss    R,   Girman  CJ,    LoCasale    RJ,
                     3. Nissen SE, Yeomans ND, Solomon DH, et al.            oxygenase-2 inhibitors and traditional non-             Brookhart AM, Sturmer T. Variable selection for
                     Cardiovascular safety of celecoxib, naproxen, or        steroidal anti-inflammatory drugs increase the           propensity score models when estimating treat-
                     ibuprofen for arthritis. N Engl J Med 2016;375:         risk of atherothrombosis? Meta-analysis of rand-        ment effects on multiple outcomes: a simulation
                     2519–29.                                                omised trials. BMJ 2006;332:1302–8.                     study. Pharmacoepidemiol Drug Saf 2013;22:
                     4. Sopena F, Lanas A. How to advise aspirin use in      14. White WB, West CR, Borer JS, et al. Risk of
                     patients who need NSAIDs. Curr Pharm Des 2007;          cardiovascular events in patients receiving cele-       24. Normand ST, Landrum MB, Guadagnoli E,
                     13:2248–60.                                             coxib: a meta-analysis of randomized clinical trials.   et al. Validating recommendations for coronary
                                                                             Am J Cardiol 2007;99:91–8.                              angiography following acute myocardial infarction
                     5. Loll PJ, Picot D, Garavito RM. The structural
                                                                                                                                     in the elderly: a matched analysis using propensity
                     basis of aspirin activity inferred from the crystal     15. Kimmel SE, Berlin JA, Reilly M, et al. The ef-
                                                                                                                                     scores. J Clin Epidemiol 2001;54:387–98.
                     structure of inactivated prostaglandin H2 syn-          fects of nonselective non-aspirin non-steroidal
                     thase. Nat Struct Biol 1995;2:637–43.                   anti-inflammatory medications on the risk of             25. Austin PC, Stuart EA. Moving towards best
                                                                             nonfatal myocardial infarction and their interac-       practice when using inverse probability of treat-
                     6. Patrono C, Garcia Rodriguez LA, Landolfi R,
                                                                             tion with aspirin. J Am Coll Cardiol 2004;43:           ment weighting (IPTW) using the propensity score
                     Baigent C. Low-dose aspirin for the prevention of
                                                                             985–90.                                                 to estimate causal treatment effects in observa-
                     atherothrombosis. N Engl J Med 2005;353:
                                                                                                                                     tional studies. Stat Med 2015;34:3661–79.
                     2373–83.                                                16. MacDonald TM, Wei L. Effect of ibuprofen on
                                                                             cardioprotective effect of aspirin. Lancet 2003;        26. Austin PC. Variance estimation when using
                     7. Gladding PA, Webster MW, Farrell HB, Zeng IS,
                                                                             361:573–4.                                              inverse probability of treatment weighting (IPTW)
                     Park R, Ruijne N. The antiplatelet effect of six non-
                                                                                                                                     with survival analysis. Stat Med 2016;35:5642–55.
                     steroidal anti-inflammatory drugs and their phar-        17. Curtis JP, Wang Y, Portnay EL, Masoudi FA,
                     macodynamic interaction with aspirin in healthy         Havranek EP, Krumholz HM. Aspirin, ibuprofen,           27. FitzGerald GA, Patrono C. The coxibs, selective
                     volunteers. Am J Cardiol 2008;101:1060–3.               and mortality after myocardial infarction: retro-       inhibitors of cyclooxygenase-2. N Engl J Med
                                                                             spective cohort study. BMJ 2003;327:1322–3.             2001;345:433–42.
                     8. Solomon SD, McMurray JJ, Pfeffer MA, et al.
                     Cardiovascular risk associated with celecoxib in a      18. Patel TN, Goldberg KC. Use of aspirin and           28. Moore RA, Derry S, McQuay HJ. Discontinua-
                     clinical trial for colorectal adenoma prevention.       ibuprofen compared with aspirin alone and the risk      tion rates in clinical trials in musculoskeletal pain:
                     N Engl J Med 2005;352:1071–80.                          of myocardial infarction. Arch Intern Med 2004;         meta-analysis from etoricoxib clinical trial reports.
                                                                             164:852–6.                                              Arthritis Res Ther 2008;10:R53.
                     9. Trelle S, Reichenbach S, Wandel S, et al. Car-
                     diovascular safety of non-steroidal anti-inflam-         19. Hudson M, Baron M, Rahme E, Pilote L.               29. Scarpignato C, Lanas A, Blandizzi C, et al. Safe
                     matory drugs: network meta-analysis. BMJ 2011;          Ibuprofen may abrogate the benefits of aspirin           prescribing of non-steroidal anti-inflammatory
                     342:c7086.                                              when used for secondary prevention of myocardial        drugs in patients with osteoarthritis—an expert
                                                                             infarction. J Rheumatol 2005;32:1589–93.                consensus addressing benefits as well as gastro-
                     10. Coxib and Traditional NSAID Trialists’ (CNT)
                                                                                                                                     intestinal and cardiovascular risks. BMC Med 2015;
                     Collaboration, Bhala N, Emberson J, et al. Vascular     20. Gasparyan AY, Watson T, Lip GY. The role of
                     and upper gastrointestinal effects of non-steroidal     aspirin in cardiovascular prevention: implications
                     anti-inflammatory drugs: meta-analyses of indi-          of aspirin resistance. J Am Coll Cardiol 2008;51:       30. Bavry AA, Thomas F, Allison M, et al.
                     vidual participant data from randomised trials.         1829–43.                                                Nonsteroidal anti-inflammatory drugs and cardio-
                     Lancet 2013;382:769–79.                                                                                         vascular outcomes in women: results from the
                                                                             21. Becker MC, Wang TH, Wisniewski L, et al.
                                                                                                                                     women’s health initiative. Circ Cardiovasc Qual
                     11. Graham DJ, Campen D, Hui R, et al. Risk of          Rationale, design, and governance of Prospective
                                                                                                                                     Outcomes 2014;7:603–10.
                     acute myocardial infarction and sudden cardiac          Randomized Evaluation of Celecoxib Integrated
                     death in patients treated with cyclo-oxygenase          Safety Versus Ibuprofen or Naproxen (PRECISION),        31. Silverstein FE, Faich G, Goldstein JL, et al.
                     2 selective and non-selective non-steroidal             a cardiovascular end point trial of nonsteroidal        Gastrointestinal toxicity with celecoxib vs
JACC VOL. 71, NO. 16, 2018                                                                                                                                  Reed et al.   1751
APRIL 24, 2018:1741–51                                                                                                        Safety of Combined Aspirin and NSAID Use

nonsteroidal anti-inflammatory drugs for osteoar-       33. Funk MJ, Westreich D, Wiesen C, Sturmer T,
thritis and rheumatoid arthritis: the CLASS study: a   Brookhart MA, Davidian M. Doubly robust estimation   KEY WORDS aspirin, celecoxib, ibuprofen,
randomized controlled trial. Celecoxib Long-Term       of causal effects. Am J Epidemiol 2011;173:761–7.    naproxen, nonsteroidal anti-inflammatory
Arthritis Safety Study. JAMA 2000;284:1247–55.                                                              drugs
                                                       34. Solomon DH, Husni ME, Wolski KE, et al. Dif-
32. Garcia Rodriguez LA, Varas-Lorenzo C,              ferences in safety of non-steroidal anti-inflam-
Maguire A, Gonzalez-Perez A. Nonsteroidal anti-        matory drugs in patients with osteoarthritis and
inflammatory drugs and the risk of myocardial           rheumatoid arthritis: a randomized clinical trial.   A PP END IX For a supplemental table and
infarction in the general population. Circulation      Arthritis Rheumatol 2017 Dec 20 [E-pub ahead of      figures, please see the online version of this
2004;109:3000–6.                                       print].                                              paper.
You can also read
NEXT SLIDES ... Cancel