The n e w e ng l a n d j o u r na l of m e dic i n e

Original Article

Efficacy and Safety of Low-Dose Colchicine

after Myocardial Infarction
Jean‑Claude Tardif, M.D., Simon Kouz, M.D., David D. Waters, M.D.,
Olivier F. Bertrand, M.D., Ph.D., Rafael Diaz, M.D., Aldo P. Maggioni, M.D.,
Fausto J. Pinto, M.D., Ph.D., Reda Ibrahim, M.D., Habib Gamra, M.D.,
Ghassan S. Kiwan, M.D., Colin Berry, M.D., Ph.D., José López‑Sendón, M.D.,
Petr Ostadal, M.D., Ph.D., Wolfgang Koenig, M.D., Denis Angoulvant, M.D.,
Jean C. Grégoire, M.D., Marc‑André Lavoie, M.D., Marie‑Pierre Dubé, Ph.D.,
David Rhainds, Ph.D., Mylène Provencher, Ph.D., Lucie Blondeau, M.Sc.,
Andreas Orfanos, M.B., B.Ch., Philippe L. L’Allier, M.D., Marie‑Claude Guertin, Ph.D.,
and François Roubille, M.D., Ph.D.​​

A BS T R AC T

BACKGROUND
Experimental and clinical evidence support the role of inflammation in atheroscle- The authors’ affiliations are listed in the
rosis and its complications. Colchicine is an orally administered, potent antiinflam- Appendix. Address reprint requests to
Dr. Tardif at the Montreal Heart Institute,
matory medication that is indicated for the treatment of gout and pericarditis. 5000 Belanger St., Montreal, PQ H1T 1C8,
Canada, or at ­jean-claude​.­tardif@​­icm-mhi​
METHODS .­org.
We performed a randomized, double-blind trial involving patients recruited within This article was published on November
30 days after a myocardial infarction. The patients were randomly assigned to receive 16, 2019, at NEJM.org.
either low-dose colchicine (0.5 mg once daily) or placebo. The primary efficacy end DOI: 10.1056/NEJMoa1912388
point was a composite of death from cardiovascular causes, resuscitated cardiac ar- Copyright © 2019 Massachusetts Medical Society.
rest, myocardial infarction, stroke, or urgent hospitalization for angina leading to
coronary revascularization. The components of the primary end point and safety were
also assessed.
RESULTS
A total of 4745 patients were enrolled; 2366 patients were assigned to the colchicine
group, and 2379 to the placebo group. Patients were followed for a median of 22.6
months. The primary end point occurred in 5.5% of the patients in the colchicine
group, as compared with 7.1% of those in the placebo group (hazard ratio, 0.77; 95%
confidence interval [CI], 0.61 to 0.96; P = 0.02). The hazard ratios were 0.84 (95% CI,
0.46 to 1.52) for death from cardiovascular causes, 0.83 (95% CI, 0.25 to 2.73) for
resuscitated cardiac arrest, 0.91 (95% CI, 0.68 to 1.21) for myocardial infarction, 0.26
(95% CI, 0.10 to 0.70) for stroke, and 0.50 (95% CI, 0.31 to 0.81) for urgent hospi-
talization for angina leading to coronary revascularization. Diarrhea was reported in
9.7% of the patients in the colchicine group and in 8.9% of those in the placebo
group (P = 0.35). Pneumonia was reported as a serious adverse event in 0.9% of the
patients in the colchicine group and in 0.4% of those in the placebo group (P = 0.03).
CONCLUSIONS
Among patients with a recent myocardial infarction, colchicine at a dose of 0.5 mg
daily led to a significantly lower risk of ischemic cardiovascular events than placebo.
(Funded by the Government of Quebec and others; COLCOT ClinicalTrials.gov num-
ber, NCT02551094.)

n engl j med nejm.org 1

The New England Journal of Medicine
Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

I
nflammation appears to play an im- philanthropic foundations. The trial protocol,
portant role in atherosclerosis.1 Inhibition of available with the full text of this article at NEJM
interleukin-1β by the injectable monoclonal .org, was designed by the trial steering committee.
antibody canakinumab led to a 15% lower risk of The protocol was approved by the institutional
cardiovascular events than was observed with review board at each of the 167 centers in the 12
placebo in the Canakinumab Antiinflammatory countries that participated in the trial (see the
Thrombosis Outcomes Study (CANTOS) but also Supplementary Appendix, available at NEJM.org).
led to a slightly higher incidence of fatal infec- All trial support activities, including project co-
tions.2 In contrast, methotrexate did not affect ordination, medical review, data management, site
cardiovascular outcomes or plasma markers of in- monitoring, and statistical oversight and analyses,
flammation in the Cardiovascular Inflammation were performed at the Montreal Health Innova-
Reduction Trial (CIRT).3 In light of these differing tions Coordinating Center. Potential trial end-point
results and given that canakinumab has not been events were adjudicated by an independent clini-
approved for cardiovascular prevention, the search cal end-point committee composed of experienced
for a widely used alternative antiinflammatory cardiologists and neurologists who were unaware
treatment that may reduce the risk of atheroscle- of the trial-group assignments. The trial was over-
rotic events among patients with coronary artery seen by a data and safety monitoring board of in-
disease continues. dependent experts. The trial medication and
Colchicine is an inexpensive, orally adminis- matching placebo were provided by Pharmasci-
tered, potent antiinflammatory medication that ence, which had no role in the design or conduct
was initially extracted from the autumn crocus and of the trial or in the preparation or review of the
has been used for centuries. Its mechanism of ac- manuscript. The first author and the lead statis-
tion is through the inhibition of tubulin polymer- tician (also an author) prepared the first draft of
ization and microtubule generation and, possibly, the manuscript, had full access to the trial data-
effects on cellular adhesion molecules, inflamma- base, and generated statistical analyses; they
tory chemokines, and the inflammasome.4-6 Col- also made the decision to submit the manuscript
chicine is currently indicated for the treatment of for publication and assume responsibility for the
gout, familial Mediterranean fever, and pericardi- accuracy and completeness of the data and for
tis.7,8 In the Low-Dose Colchicine (LoDoCo) trial, the fidelity of the trial to the protocol.
patients with stable coronary disease treated with
colchicine at a dose of 0.5 mg once daily had Trial Population
fewer cardiovascular events than those not receiv- Adult patients were eligible if they had had a
ing colchicine.9 However, that trial enrolled only myocardial infarction within 30 days before en-
532 patients and was not placebo-controlled. Be- rollment, had completed any planned percutane-
cause acute coronary syndromes are associated ous revascularization procedures, and were treat-
with higher risks of recurrent events and exacer- ed according to national guidelines that included
bated inflammation, we conducted the Colchicine the intensive use of statins. Patients were excluded
Cardiovascular Outcomes Trial (COLCOT) to evalu- if they had severe heart failure, a left ventricular
ate the effects of colchicine on cardiovascular ejection fraction of less than 35%, stroke within
outcomes as well as its long-term safety profile the previous 3 months, a type 2 index myocardial
in patients who had recently had a myocardial infarction, coronary-bypass surgery either within
infarction. the previous 3 years or planned, a history of non-
cutaneous cancer within the previous 3 years,
inflammatory bowel disease or chronic diarrhea,
Me thods
neuromuscular disease or a nontransient creatine
Trial Design and Oversight kinase level that was greater than three times
In this randomized, double-blind, placebo-con- the upper limit of the normal range (unless due
trolled, investigator-initiated trial, we assigned to infarction), clinically significant nontransient
patients in a 1:1 ratio to receive either colchicine hematologic abnormalities, severe renal disease
(at a dose of 0.5 mg once daily) or placebo. The with a serum creatinine level that was greater than
trial was funded by the Government of Quebec, two times the upper limit of the normal range;
the Canadian Institutes of Health Research, and severe hepatic disease, drug or alcohol abuse, cur-

2 n engl j med nejm.org

The New England Journal of Medicine

Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Low-Dose Colchicine after Myocardial Infarction

rent or planned long-term systemic glucocorticoid cebo, indicated by a hazard ratio of 0.724. With
therapy, or a history of clinically significant sen- the use of a two-sided test at the 0.05 significance
sitivity to colchicine. (Details regarding eligibility level, the trial would have 80% power if it con-
criteria are provided in the Supplementary Ap- tinued until 301 positively adjudicated primary
pendix.) events occurred in the combined trial groups.
Written informed consent was obtained from The trial design assumed an event rate of 7% in
all the patients before enrollment. Clinical eval- the placebo group at 24 months, an 18-month
uations occurred at 1 month and 3 months after recruitment period during which patients would
randomization and every 3 months thereafter. be uniformly recruited, a 24-month minimum
follow-up period, and a 1% annual rate of loss to
End Points follow-up or withdrawal of consent.
The primary efficacy end point was a composite The efficacy analyses were conducted with
of death from cardiovascular causes, resuscitated the use of positively adjudicated data and ac-
cardiac arrest, myocardial infarction, stroke, or cording to the intention-to-treat principle. The
urgent hospitalization for angina leading to coro- primary end point was compared between the
nary revascularization in a time-to-event analysis. two trial groups with the use of a log-rank test,
The secondary end points consisted of the com- and the hazard ratio, with a 95% confidence in-
ponents of the primary efficacy end point; a terval, was calculated from a Cox proportional-
composite of death from cardiovascular causes, hazards model. A Cox proportional-hazards model
resuscitated cardiac arrest, myocardial infarc- with adjustment for important baseline charac-
tion, or stroke; and total mortality in time-to- teristics was also used as prespecified in the
event analyses. Coronary revascularization, hos- protocol. The analysis of the primary end point
pitalization for heart failure, atrial fibrillation, was repeated in the per-protocol population (i.e.,
and deep venous thrombosis or pulmonary em- patients without major protocol deviations). Sec-
bolus were prespecified as exploratory end points ondary and exploratory end points expressed as
in the protocol. Additional prespecified explor- time to event were analyzed similarly. The chang-
atory end points included the change from base- es from baseline to follow-up were analyzed with
line to 6 months in the high-sensitivity C-reactive the use of an analysis of covariance model with
protein level and the change from baseline to 12 adjustment for baseline value, and estimates of
months in the white-cell count. The C-reactive treatment effect are presented with 95% confi-
protein biomarker substudy was implemented after dence intervals.
a protocol amendment and was optional for sites The efficacy end points expressed as time to
and for patients; 34 sites chose to participate in event could be assessed in all patients because
this substudy. the event dates and censoring dates were com-
All serious adverse events were recorded. The plete, with the exception of one incomplete event
only other adverse events recorded were those that date for atrial fibrillation; therefore, imputation
were considered to be related to the gastrointes- for missing data was not done. In the analysis of
tinal system, events that were judged by the in- time to event, the following censoring rules were
vestigator to be related to colchicine or placebo, used. For death from any cause and death from
or laboratory abnormalities that had been judged cardiovascular causes, data from event-free pa-
by the investigator to be clinically significant. tients who completed the trial were censored at the
date of trial completion, and data from patients
Statistical Analysis who did not complete the trial, such as those who
In this event-driven trial, it was estimated that a were lost to follow-up or who withdrew consent,
sample of approximately 4500 patients undergo- were censored at the date of last contact or the
ing randomization (with 2250 patients in each date of the assessment of survival status, which-
group) or, in terms of events, a total number of ever was later. For the analysis of death from
301 patients with a first positively adjudicated cardiovascular causes, patients who died from a
primary end-point event would yield adequate noncardiovascular cause had their data censored
power. The sample-size calculation was based at the time of death. For all other end points,
on the primary efficacy end point and assumed including the primary end point, the same cen-
a 27% lower risk with colchicine than with pla- soring rules applied, but the survival status was

n engl j med nejm.org 3

The New England Journal of Medicine
Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

analysis of the primary end point. All other sta-

tistical tests were two-sided and conducted at the
4745 Patients underwent randomization
0.05 significance level. Statistical analyses were
performed with the use of SAS software, version
9.4 (SAS Institute). There was no prespecified plan
to adjust for multiple comparisons across the mul-
2366 Were assigned to receive colchicine 2379 Were assigned to receive placebo tiple methods that were used to analyze the pri-
(36 did not receive colchicine) (33 did not receive placebo) mary and secondary end points; results of these
analyses are reported with point estimates and
95% confidence intervals, without P values. The
2226 Completed the trial 2232 Completed the trial
140 Did not complete the trial 147 Did not complete the trial
95% confidence intervals were not adjusted for
17 Withdrew consent 13 Withdrew consent multiple comparisons, and inferences drawn from
41 Discontinued colchicine and 38 Discontinued placebo and
visits visits
them may not be reproducible. The final amend-
39 Were lost to follow-up 1 Had an adverse event ment to the statistical analysis plan was approved
42 Died 50 Were lost to follow-up
1 Had been hospitalized since 44 Died
on August 28, 2019, before unblinding of the trial-
October 10, 2018 1 Discontinued trial prematurely group assignments occurred.
and did not complete all visits

R e sult s
Survival status at end of trial: Survival status at end of trial: Patients
2309 Were alive 2325 Were alive
44 Had died 44 Had died Trial enrollment began in December 2015 and
13 Had unknown survival status 10 Had unknown survival status was completed in August 2018; the last trial visit
was in July 2019. A total of 4745 patients under-
Figure 1. Randomization and Follow-up of the Patients. went randomization (with 2366 being assigned
to the colchicine group and 2379 to the placebo
group) and were followed for a median of 22.6
not used because no formal assessment of end months. At the time of the database lock on
points was done at the assessment of survival August 28, 2019, and unblinding on August 29,
status. An analysis of the components of the pri- 2019, vital status was available for all except 23
mary end point with death from noncardiovascular patients (99.5%); 89 patients (1.9%) were lost to
causes as a competing event for death from cardio- follow-up, and 30 patients (0.6%) withdrew con-
vascular causes, and with death from any cause as sent. Details regarding the disposition of the pa-
a competing event for the other components, was tients are provided in Figure 1.
conducted with the use of the Fine and Gray sub- The characteristics of the patients at baseline
distribution hazard model.10 No missing data were are shown in Table 1. Patients were enrolled a
imputed except for age in cases in which informa- mean of 13.5 days after myocardial infarction.
tion on the day or the month and day of birth was The mean age of the patients was 60.6 years,
missing. To account for the occurrence of multiple 19.2% of the patients were women, and 20.2%
primary end-point events within patients, recur- had diabetes. Most patients (93.0%) underwent
rent-event analyses were undertaken with the use percutaneous coronary intervention for their index
of negative binomial regression, Andersen–Gill, myocardial infarction. Aspirin, a different anti-
and Wei–Lin–Weissfeld models.11-19 platelet agent, and a statin were taken by 98.8%,
An interim analysis was performed after 50% 97.9% and 99.0% of the patients, respectively.
of the primary end-point events had been posi- At the end of the trial, the trial regimen had
tively adjudicated. The prespecified stopping rule been discontinued in 18.4% of the patients in the
for efficacy was based on the Lan–DeMets pro- colchicine group and in 18.7% of those in the pla-
cedure with the O’Brien–Fleming alpha-spending cebo group. Among the patients who discontinued
function. After review of the interim results, the the trial regimen, the median duration of receipt of
data and safety monitoring board recommended the trial drug was 7.1 months (interquartile range,
that the trial should continue as planned. To ac- 1.9 to 14.6) in the colchicine group, as compared
count for this interim analysis, the statistical with 6.1 months (interquartile range, 1.6 to 14.4)
significance level was set to 0.0490 for the final in the placebo group. Overall, the median dura-

4 n engl j med nejm.org

The New England Journal of Medicine

Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Low-Dose Colchicine after Myocardial Infarction

Table 1. Characteristics of the Patients.*

Colchicine Placebo
Characteristic (N = 2366) (N = 2379)
Age — yr 60.6±10.7 60.5±10.6
Female sex — no. (%) 472 (19.9) 437 (18.4)
White race — no./total no. (%)† 1350/1850 (73.0) 1329/1844 (72.1)
Body-mass index 28.2±4.8 28.4±4.7
Current smoking — no./total no. (%) 708/2366 (29.9) 708/2377 (29.8)
Hypertension — no. (%) 1185 (50.1) 1236 (52.0)
Diabetes — no. (%) 462 (19.5) 497 (20.9)
History of myocardial infarction — no. (%) 370 (15.6) 397 (16.7)
History of PCI — no. (%) 392 (16.6) 406 (17.1)
History of CABG — no. (%) 69 (2.9) 81 (3.4)
History of heart failure — no. (%) 48 (2.0) 42 (1.8)
History of stroke or TIA — no. (%) 55 (2.3) 67 (2.8)
Time from index myocardial infarction to randomization — days 13.4±10.2 13.5±10.1
PCI for index myocardial infarction — no./total no. (%) 2192/2364 (92.7) 2216/2375 (93.3)
Medication use — no. (%)
Aspirin 2334 (98.6) 2352 (98.9)
Other antiplatelet agent 2310 (97.6) 2337 (98.2)
Statin 2339 (98.9) 2357 (99.1)
Beta-blocker 2116 (89.4) 2101 (88.3)

* Plus–minus values are means ±SD. Data were missing on the following characteristics: age (assessed according to
date of birth; see below) for 435 patients (215 in the colchicine group and 220 in the placebo group), body-mass index
(the weight in kilograms divided by the square of the height in meters) for 5 (1 and 4 patients, respectively), and infor-
mation about the index myocardial infarction for 6 (2 and 4 patients, respectively). Date of birth and race were not re-
quired fields because both were considered in some countries to be sensitive data that could allow for the identification
of patients. For statistical reporting, missing information regarding the day of birth was replaced by 15, and missing in-
formation regarding the month and day of birth was replaced by July 1. CABG denotes coronary-artery bypass graft sur-
gery, PCI percutaneous coronary intervention, and TIA transient ischemic attack.
† Race was reported by the patient.

tion of receipt of the trial drug was 19.6 months cebo group (hazard ratio, 0.71; 95% CI, 0.56 to
in the colchicine group and 19.5 months in the 0.90) (Table S2). Table 2 shows the percentages
placebo group. of patients with events and the hazard ratios for
the components of the primary end point, includ-
Clinical Efficacy End Points ing death from cardiovascular causes (hazard ra-
A primary end-point event occurred in 5.5% of the tio, 0.84; 95% CI, 0.46 to 1.52), resuscitated car-
patients in the colchicine group, as compared with diac arrest (hazard ratio, 0.83; 95% CI, 0.25 to
7.1% of those in the placebo group (hazard ratio, 2.73), myocardial infarction (hazard ratio, 0.91;
0.77; 95% confidence interval [CI], 0.61 to 0.96; 95% CI, 0.68 to 1.21), stroke (hazard ratio, 0.26;
P = 0.02 by the log-rank test). A multivariable Cox 95% CI, 0.10 to 0.70), and urgent hospitalization
regression model with adjustment for baseline for angina leading to coronary revascularization
covariates yielded a similar result (Table S1 in the (hazard ratio, 0.50; 95% CI, 0.31 to 0.81). The
Supplementary Appendix). The event curves that hazard ratios remained unchanged in the analy-
were based on a Kaplan–Meier analysis of the sis that took competing events into account.
primary efficacy end point are shown in Figure 2. The secondary efficacy end point consisting
In the prespecified per-protocol analysis involving of a composite of death from cardiovascular
patients who adhered to the protocol, the primary causes, cardiac arrest, myocardial infarction, or
end point occurred in 5.1% of the patients in the stroke occurred in 4.7% of the patients in the
colchicine group and in 7.1% of those in the pla- colchicine group and in 5.5% of those in the

n engl j med nejm.org 5

The New England Journal of Medicine
Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Table 2. Major Clinical End Points (Intention-to-Treat Population).*

Colchicine Placebo Hazard Ratio

End Point (N = 2366) (N = 2379) (95% CI) P Value

number (percent)
Primary composite end point 131 (5.5) 170 (7.1) 0.77 (0.61–0.96) 0.02†
Components of primary end point
Death from cardiovascular causes 20 (0.8) 24 (1.0) 0.84 (0.46–1.52)
Resuscitated cardiac arrest 5 (0.2) 6 (0.3) 0.83 (0.25–2.73)
Myocardial infarction 89 (3.8) 98 (4.1) 0.91 (0.68–1.21)
Stroke 5 (0.2) 19 (0.8) 0.26 (0.10–0.70)
Urgent hospitalization for angina lead- 25 (1.1) 50 (2.1) 0.50 (0.31–0.81)
ing to revascularization
Secondary composite end point‡ 111 (4.7) 130 (5.5) 0.85 (0.66–1.10)
Death 43 (1.8) 44 (1.8) 0.98 (0.64–1.49)
Deep venous thrombosis or pulmonary 10 (0.4) 7 (0.3) 1.43 (0.54–3.75)
embolus
Atrial fibrillation 36 (1.5) 40 (1.7) 0.93 (0.59–1.46)

* Only the initial event was counted in the analyses of time to first event for the primary composite end point and for the
secondary composite end point. In the component analysis, the different types of events were counted separately.
† The log-rank test and the multivariable Cox proportional-hazards model including age, history of diabetes, previous cor-
onary revascularization, and previous heart failure yielded similar P values.
‡ The secondary composite end point included death from cardiovascular causes, resuscitated cardiac arrest, myocardial
infarction, and stroke.

placebo group (hazard ratio, 0.85; 95% CI, 0.66

100 15
Hazard ratio, 0.77 (95% CI, 0.61–0.96)
to 1.10). Data on the primary, secondary, and
90 P=0.02 exploratory efficacy end points are provided in
Cumulative Incidence (%)

80 10 Placebo Table 2. Two patients had a first positively adju-

70 dicated event of urgent hospitalization for angi-
60 Colchicine
5 na leading to coronary revascularization within
50
14 days after randomization. The median time
40
0 to this clinical end point was 258 days. Efficacy
30 0 7 14 21 28 35 42 results in prespecified subgroups are shown in
20
10
Table S3. The total number of primary end-point
0
events (first and recurrent) was 154 in the col-
0 7 14 21 28 35 42 chicine group and 223 in the placebo group, over
Months since Randomization periods of 52,949 and 53,060 patient-months of
No. at Risk follow-up, respectively. Thus, the primary end-
Placebo 2379 2261 1854 1224 622 144 0 point event rates per 100 patient-months were
Colchicine 2366 2284 1868 1230 628 153 0
0.29 in the colchicine group and 0.42 in the pla-
Figure 2. Cumulative Incidence of Cardiovascular Events (Intention-to-Treat cebo group (rate ratio, 0.66; 95% CI, 0.51 to 0.86)
Population). (Table S4).
Shown are the Kaplan–Meier event curves for the primary efficacy compos-
ite end point of death from cardiovascular causes, resuscitated cardiac ar- Biomarkers of Inflammation
rest, myocardial infarction, stroke, or urgent hospitalization for angina High-sensitivity C-reactive protein was measured
leading to coronary revascularization in the colchicine group and the place-
in a subgroup of only 207 patients at the time of
bo group in a time-to-event analysis. The inset shows the same data on an
enlarged y axis. randomization and 6 months later, and the me-
dian concentration at trial entry was 4.28 mg per

6 n engl j med nejm.org

The New England Journal of Medicine

Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Low-Dose Colchicine after Myocardial Infarction

liter. The baseline characteristics of these patients Table 3. Adverse Events (Safety Population).*
were similar to those of the overall population
(Table S5), but the small and selected subgroup Colchicine Placebo
Event (N = 2330) (N = 2346) P Value
with these data limits the interpretation of these
analyses. The adjusted geometric mean percent number of patients (percent)
changes in the high-sensitivity C-reactive protein
Any related adverse event† 372 (16.0) 371 (15.8) 0.89
level at 6 months after myocardial infarction
Adverse events
were −70.0% in the colchicine group and −66.6%
in the placebo group, and the placebo-adjusted Gastrointestinal event 408 (17.5) 414 (17.6) 0.90
geometric mean percent change was −10.1 per- Diarrhea 225 (9.7) 208 (8.9) 0.35
centage points in the colchicine group (95% CI, Nausea 43 (1.8) 24 (1.0) 0.02
−28.6 to 13.4) (Table S6). Flatulence 15 (0.6) 5 (0.2) 0.02
Information about white-cell counts at baseline Gastrointestinal hemorrhage 7 (0.3) 5 (0.2) 0.56
and at the 12-month follow-up were also available
Anemia 14 (0.6) 10 (0.4) 0.40
for a relatively small subgroup of 1972 patients.
The adjusted geometric mean percent changes Leukopenia 2 (0.1) 3 (0.1) 0.66
from baseline to 1 year in the total white-cell count Thrombocytopenia 3 (0.1) 7 (0.3) 0.21
were −18.8% in the colchicine group and −19.0% Serious adverse events
in the placebo group, with no significant dif- Any serious adverse event‡ 383 (16.4) 404 (17.2) 0.47
ference between groups (0.3 percentage points; Gastrointestinal event 46 (2.0) 36 (1.5) 0.25
95% CI, −2.2 to 2.7).
Infection 51 (2.2) 38 (1.6) 0.15

Safety and Adverse Events Pneumonia 21 (0.9) 9 (0.4) 0.03

The incidence of adverse events that were consid- Septic shock 2 (0.1) 2 (0.1) 0.99
ered to be related to the active drug or placebo was Hospitalization for heart 25 (1.1) 17 (0.7) 0.21
failure
16.0% in the colchicine group and 15.8% in the
placebo group, and the overall incidence of serious Cancer§ 43 (1.8) 46 (2.0) 0.77
adverse events was 16.4% and 17.2%, respectively
* The safety population was defined as patients who took at least one dose of
(Table 3). At least one gastrointestinal adverse colchicine or placebo. All serious adverse events were recorded, and the only
event during the double-blind period occurred in other adverse events recorded were those that were related to the gastrointes-
17.5% of the patients in the colchicine group, as tinal system, events that were judged by the investigator to be related to col-
chicine or placebo, or laboratory abnormalities that were judged by the inves-
compared with 17.6% of those in the placebo tigator to be clinically significant. This table lists serious adverse events that
group. Diarrhea was reported in 9.7% of the pa- were present in more than 2% of the patients in either trial group, adverse
tients in the colchicine group and in 8.9% of events that were considered to be related to colchicine or placebo in more
than 5% of the patients in either trial group, and any other safety events of
those in the placebo group (P = 0.35), and nausea special interest. Chi-square tests were conducted to compare the incidence of
was more common in the colchicine group than adverse events between the trial groups.
in the placebo group (1.8% vs. 1.0%, P = 0.02). † These adverse events were considered to be related to colchicine or placebo
by the physician in charge of the participant.
Pneumonia was reported as a serious adverse event ‡ There was one serious adverse event of myopathy, which was attributed to
in 0.9% of the patients in the colchicine group, high-dose statin therapy (rosuvastatin at a dose of 40 mg daily) by the local
as compared with 0.4% of those in the placebo investigator and academic sponsor, in a man of short stature (165 cm, 68 kg)
with normal renal function in the colchicine group who had received colchi-
group (P = 0.03). cine for 8 days 3 months before the adverse event.
§ Cancers, excluding nonmelanoma skin cancers, occurred in 42 patients
(1.8%) in the colchicine group and in 44 (1.9%) in the placebo group.
Discussion
In COLCOT, the risk of the primary composite
efficacy end point of death from cardiovascular once daily than among those who received place-
causes, resuscitated cardiac arrest, myocardial bo. This result was due predominantly to a lower
infarction, stroke, or urgent hospitalization for incidence of strokes and urgent hospitalizations
angina leading to coronary revascularization, as for angina leading to coronary revascularization.
assessed in a time-to-event analysis, was signifi- These results were observed against a back-
cantly lower among the patients who were ran- ground of appropriate medications, which includ-
domly assigned to receive 0.5 mg of colchicine ed aspirin, a different antiplatelet agent, and a

n engl j med nejm.org 7

The New England Journal of Medicine
Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

statin in 98 to 99% of the patients. In addition, These differences in the incidence of infections
percutaneous coronary intervention was performed could be due to the play of chance or could re-
in 93% of the patients for their index myocar- flect altered immunologic responses. In contrast
dial infarction. The benefits of colchicine with to canakinumab,2 colchicine did not increase the
regard to cardiovascular end points in COLCOT incidence of septic shock in our trial. Infections
were at least as large as those of canakinumab have previously been described in patients who
in CANTOS.2 In the small subgroup of patients have attempted suicide by taking an overdose of
with available data, as expected, a large (>65%) colchicine.20 There was no serious adverse event
reduction in the C-reactive protein level occurred of myopathy linked to colchicine despite the use
over the first 6 months after myocardial infarc- of statins in 99% of the patients in the trial.
tion in both trial groups in COLCOT, but the dif- Our trial has certain limitations. The duration
ference between the changes in the groups was of follow-up was relatively short at approximately
not significant. These findings must be interpreted 23 months. The risks and benefits of longer-term
cautiously given that this was a small subgroup treatment with colchicine were not evaluated.
that was not randomly selected from the full trial Although the inclusion of 4745 patients was suf-
sample. A similar observation was made with ficient for the trial to show a significant benefit
white-cell counts. The different patient popula- with regard to the primary composite efficacy end
tions involved in the two trials — early after point, a larger trial could have allowed a better
myocardial infarction in COLCOT and stable assessment of individual end points and subgroups
coronary disease in CANTOS — may also have and the risks associated with colchicine. Finally,
affected the relationship between biomarkers of our results apply only to patients who have re-
inflammation and the effects of treatments on cently had a myocardial infarction.
ischemic end points. In conclusion, among patients with a recent
The known benefits of colchicine in the treat- myocardial infarction, colchicine at a dose of
ment of pericarditis were not at play in COLCOT. 0.5 mg daily led to a significantly lower percent-
Postinfarction pericarditis typically occurs with- age of patients with ischemic cardiovascular events
in the first few days after the injury, whereas the than placebo.
mean time from the index myocardial infarction
to randomization was 13.5 days. There were only Supported by the Government of Quebec, the Canadian Insti-
two patients with a first positively adjudicated tutes of Health Research, and philanthropic foundations, with
the funds administered by the Montreal Heart Institute. The
event of urgent hospitalization for angina lead- Montreal Health Innovations Coordinating Center managed the
ing to coronary revascularization within 14 days trial and is a division of the Montreal Heart Institute.
after randomization, and the median time to this Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
clinical end point was 258 days. A data sharing statement provided by the authors is available
The most common adverse events observed with the full text of this article at NEJM.org.
were gastrointestinal. Diarrhea was reported in We thank the trial investigators and coordinators at all the
centers; trial monitors and staff from the Montreal Health Inno-
9.7% of the patients in the colchicine group and vations Coordinating Center, including Gabriela Stamatescu,
in 8.9% of those in the placebo group, and nausea M.D., Otilia Goga, M.D., and Ourida Mehenni Hadjeres, M.D., for
occurred in 1.8% and 1.0%, respectively. Infection medical review; Ève Roy-Clavel, M.Sc., C.M.C., Andrée Brunelle,
B.A., and Luc Dion, M.Sc., for data management and program-
as a serious adverse event was more frequent in ming; Sylvie Lévesque, M.Sc., Anna Nozza, M.Sc., Mariève Cos-
the colchicine group than in the placebo group sette M.Sc., Annik Fortier, M.Sc., and Daniel Cournoyer, M.Sc.,
(in 2.2% vs. 1.6% of the patients), and pneumonia for assistance with biostatistics; and Randa Zamrini, B.Sc.,
Marianne Rufiange, Ph.D., Andréa Alicia Dumont, B.Sc., and
as a serious adverse event was also more fre- Zohar Bassevitch, B.Sc., for clinical operations; and the partici-
quent in the colchicine group (0.9% vs. 0.4%). pating patients for their contribution to the trial.

Appendix
The authors’ affiliations are as follows: the Montreal Heart Institute (J.-C.T., R.I., J.C.G., M.-A.L., M.-P.D., D.R., P.L.L.) and the Mon-
treal Health Innovations Coordinating Center (M.P., L.B., A.O., M.-C.G.), Montreal, Centre Hospitalier Régional de Lanaudière, Joliette
(S.K.), and Institut de Cardiologie et Pneumologie de Québec, Quebec City (O.F.B.) — all in Canada; San Francisco General Hospital,
San Francisco (D.D.W.); Estudios Clínicos Latinoamérica, Rosario, Argentina (R.D.); Associazione Nazionale Medici Cardiologi Os-
pedalieri Research Center, Florence, Italy (A.P.M.); Santa Maria University Hospital (Centro Hospitalar Universitário Lisboa Norte),
Centro Académico de Medicina de Lisboa, Centro Cardiovascular da Universidade de Lisboa, Faculdade de Medicina da Universidade de
Lisboa, Lisbon, Portugal (F.J.P.); Fattouma Bourguiba University Hospital, Monastir, Tunisia (H.G.); Bellevue Medical Center, Beirut,

8 n engl j med nejm.org

The New England Journal of Medicine

Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Low-Dose Colchicine after Myocardial Infarction

Lebanon (G.S.K.); Institute of Cardiovascular and Medical Sciences, University of Glasgow, and NHS Greater Glasgow and Clyde,
Glasgow, United Kingdom (C.B.); Hospital Universitario La Paz, Universidad Autónoma de Madrid, Instituto de Investigación La Paz,
Centro de Investigación Biomédica en Red–Enfermedades Cardiovasculares, Madrid (J.L.-S.); Cardiovascular Center, Na Homolce Hos-
pital, Prague, Czech Republic (P.O.); Deutsches Herzzentrum München, Technische Universität München, and DZHK (German Center
for Cardiovascular Research), partner site Munich Heart Alliance, Munich, and Institute of Epidemiology and Medical Biometry, Uni-
versity of Ulm, Ulm — all in Germany (W.K.); Centre Hospitalier Universitaire (CHU) de Tours and Équipe d’Accueil 4245 Transplanta-
tion Immunité Inflammation Loire Valley Cardiovascular Collaboration, Tours University, Tours (D.A.), and PhyMedExp (Physiologie et
Médecine Expérimentale du Cœur et des Muscles), Université de Montpellier, INSERM, Centre National de la Recherche Scientifique,
Cardiology Department, CHU de Montpellier, Montpellier (F.R.) — all in France.

References
1. Hansson GK. Inflammation, athero- colchicine or analogues: an update. Curr parametric regression for the mean and
sclerosis, and coronary artery disease. Drug Targets Inflamm Allergy 2005;​ 4:​ rate functions of recurrent events. J R Stat
N Engl J Med 2005;​352:​1685-95. 117-24. Soc 2000;​62:​711-30.
2. Ridker PM, Everett BM, Thuren T, et al. 8. Imazio M, Bobbio M, Cecchi E, et al. 15. Wei LJ, Glidden DV. An overview of
Antiinflammatory therapy with canakin­ Colchicine in addition to conventional statistical methods for multiple failure
umab for atherosclerotic disease. N Engl therapy for acute pericarditis: results of time data in clinical trials. Stat Med 1997;​
J Med 2017;​377:​1119-31. the COlchicine for acute PEricarditis (COPE) 16:​833-9.
3. Ridker PM, Everett BM, Pradhan A, et trial. Circulation 2005;​112:​2012-6. 16. Ghosh D. Methods for analysis of
al. Low-dose methotrexate for the preven- 9. Nidorf SM, Eikelboom JW, Budgeon multiple events in the presence of death.
tion of atherosclerotic events. N Engl J CA, Thompson PL. Low-dose colchicine Control Clin Trials 2000;​21:​115-26.
Med 2019;​380:​752-62. for secondary prevention of cardiovascu- 17. Li QH, Lagakos SW. Use of the Wei-
4. Ravelli RB, Gigant B, Curmi PA, et al. lar disease. J Am Coll Cardiol 2013;​61:​ Lin-Weissfeld method for the analysis of a
Insight into tubulin regulation from a 404-10. recurring and a terminating event. Stat
complex with colchicine and a stathmin- 10. Fine JP, Gray RJ. A proportional haz- Med 1997;​16:​925-40.
like domain. Nature 2004;​428:​198-202. ards model for the subdistribution of a 18. Metcalfe C, Thompson SG. The im-
5. Perico N, Ostermann D, Bontempeill competing risk. J Am Stat Assoc 1999;​94:​ portance of varying the event generation
M, et al. Colchicine interferes with L-selec- 496-509. process in simulation studies of statisti-
tin and leukocyte function-associated an- 11. Rogers JK, Pocock SJ, McMurray JJ, et cal methods for recurrent events. Stat
tigen-1 expression on human T lympho- al. Analysing recurrent hospitalizations in Med 2006;​25:​165-79.
cytes and inhibits T cell activation. J Am heart failure: a review of statistical meth- 19. Jahn-Eimermacher A. Comparison of
Soc Nephrol 1996;​7:​594-601. odology, with application to CHARM-Pre- the Andersen–Gill model with Poisson
6. Pope RM, Tschopp J. The role of inter- served. Eur J Heart Fail 2014;​16:​33-40. and negative binomial regression on re-
leukin-1 and the inflammasome in gout: 12. Andersen PK, Gill RD. Cox’s regres- current event data. Comput Stat Data Anal
implications for therapy. Arthritis Rheum sion model for counting processes: a large 2008;​52:​4989-97.
2007;​56:​3183-8. sample study. Ann Stat 1982;​10:​1100-20. 20. Kocak Z, Akay H, Gucenmez S, Tufan
7. Cerquaglia C, Diaco M, Nucera G, La 13. Lin DY, Wei LJ. The robust inference A, Donderici O. Colchicine intoxication
Regina M, Montalto M, Manna R. Pharma- for the proportional hazards model. J Am and infection risk: a case report. J Clin
cological and clinical basis of treatment of Stat Assoc 1989;​84:​1074-8. Pharm Ther 2008;​33:​451-2.
Familial Mediterranean Fever (FMF) with 14. Lin DY, Wei LJ, Yang I, Ying Z. Semi- Copyright © 2019 Massachusetts Medical Society.

n engl j med nejm.org 9

The New England Journal of Medicine
Downloaded from nejm.org at SANTA BARBARA (UCSB) on November 16, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.