New England Journal Medicine: The of

new england
The
journal of medicine
established in 1812 october 23, 2014 vol. 371 no. 17
Four-Month Moxifloxacin-Based Regimens

for Drug-Sensitive Tuberculosis
Stephen H. Gillespie, M.D., D.Sc., Angela M. Crook, Ph.D., Timothy D. McHugh, Ph.D., Carl M. Mendel, M.D.,
Sarah K. Meredith, M.B., B.S., Stephen R. Murray, M.D., Ph.D., Frances Pappas, M.A., Patrick P.J. Phillips, Ph.D.,
and Andrew J. Nunn, M.Sc., for the REMoxTB Consortium*
A bs t r ac t
Background
Early-phase and preclinical studies suggest that moxifloxacin-containing regimens From the University of St. Andrews Medi-
could allow for effective 4-month treatment of uncomplicated, smear-positive pul- cal School, St. Andrews (S.H.G.), and the
Medical Research Council Clinical Trials
monary tuberculosis. Unit at University College London
(A.M.C., S.K.M., P.P.J.P., A.J.N.) and the
Methods Division of Infection and Immunity, Uni-
We conducted a randomized, double-blind, placebo-controlled, phase 3 trial to test versity College London (T.D.M.), London
the noninferiority of two moxifloxacin-containing regimens as compared with a both in the United Kingdom; and the
TB Alliance, New York (C.M.M., S.R.M.,
control regimen. One group of patients received isoniazid, rifampin, pyrazinamide, F.P.). Address reprint requests to Dr. Gil-
and ethambutol for 8 weeks, followed by 18 weeks of isoniazid and rifampin (con- lespie at the School of Medicine, Univer-
trol group). In the second group, we replaced ethambutol with moxifloxacin for 17 sity of St. Andrews, North Haugh, St.
Andrews KY16 9TF, United Kingdom, or
weeks, followed by 9 weeks of placebo (isoniazid group), and in the third group, we at [email protected].
replaced isoniazid with moxifloxacin for 17 weeks, followed by 9 weeks of placebo
(ethambutol group). The primary end point was treatment failure or relapse within *
A complete list of investigators and
committee members in the Rapid Eval-
18 months after randomization. uation of Moxifloxacin in Tuberculosis
(REMoxTB) study is provided in the
Results
Supplementary Appendix, available at
Of the 1931 patients who underwent randomization, in the per-protocol analysis, a NEJM.org.
favorable outcome was reported in fewer patients in the isoniazid group (85%) and
This article was published on September 7,
the ethambutol group (80%) than in the control group (92%), for a difference favor- 2014, at NEJM.org.
ing the control group of 6.1 percentage points (97.5% confidence interval [CI], 1.7 to
10.5) versus the isoniazid group and 11.4 percentage points (97.5% CI, 6.7 to 16.1) N Engl J Med 2014;371:1577-87.
DOI: 10.1056/NEJMoa1407426
versus the ethambutol group. Results were consistent in the modified intention-to- Copyright 2014 Massachusetts Medical Society.
treat analysis and all sensitivity analyses. The hazard ratios for the time to culture
negativity in both solid and liquid mediums for the isoniazid and ethambutol
groups, as compared with the control group, ranged from 1.17 to 1.25, indicating
a shorter duration, with the lower bounds of the 95% confidence intervals exceed-
ing 1.00 in all cases. There was no significant difference in the incidence of grade
3 or 4 adverse events, with events reported in 127 patients (19%) in the isoniazid
group, 111 (17%) in the ethambutol group, and 123 (19%) in the control group.
Conclusions
The two moxifloxacin-containing regimens produced a more rapid initial decline in
bacterial load, as compared with the control group. However, noninferiority for these
regimens was not shown, which indicates that shortening treatment to 4 months
was not effective in this setting. (Funded by the Global Alliance for TB Drug De-
velopment and others; REMoxTB ClinicalTrials.gov number, NCT00864383.)
n engl j med 371;17nejm.orgoctober 23, 2014 1577

The New England Journal of Medicine
Downloaded from nejm.org on September 21, 2015. For personal use only. No other uses without permission.
Copyright 2014 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
A
short-term tuberculosis treat- regimen (Fig. S1 in the Supplementary Appendix,
ment regimen could improve rates of ad- available with the full text of this article at NEJM
herence, reduce rates of adverse events, and .org). The full trial protocol and statistical analy-
lower costs. Fluoroquinolones have shown prom- sis plan are also available at NEJM.org.
ising activity against mycobacteria1 and are es- A trial steering committee with an indepen-
tablished as a critical component of the treatment dent chair supervised the conduct of the trial. An
of multidrug-resistant tuberculosis,2,3 with later independent data and safety monitoring commit-
fluoroquinolones recognized as having a more po- tee with access to unblinded data oversaw the
tent effect. It has been proposed that these drugs safety of the study patients. The ethics commit-
may have a role in reducing the duration of tuber- tee at University College London and all national
culosis treatment.4 and local ethics committees approved the study.
Moxifloxacin has been approved for a range The Food and Drug Administration, the Federal
of indications globally.5 It has favorable pharma- Institute for Drugs and Medical Devices (Bundes-
cokinetics, a large volume of distribution, and institut fr Arzneimittel und Medizinprodukte),
penetration into epithelial-lining fluid and mac- and the national regulatory authorities of the
rophages.6-8 The activity of moxifloxacin in vitro countries in which the trial was conducted re-
against Mycobacterium tuberculosis, which has been viewed and approved the protocol.
confirmed in murine models9 and in clinical Bayer Healthcare donated moxifloxacin, and
monotherapy studies,10,11 has raised the prospect Sanofi donated rifampin. Neither company had
that the drug could be used as part of an improved any role in the study design, data accrual, data
regimen.1 Subsequent studies in mice showed analysis, or manuscript preparation. Representa-
that combination regimens that included moxi- tives of Bayer Healthcare reviewed the manuscript
floxacin had greater bactericidal activity than stan- but did not suggest revisions. All the authors
dard treatment and could produce cure without vouch for the completeness and accuracy of the
relapse after a shorter treatment duration.12,13 data and analyses presented.
When different fluoroquinolones were substi-
tuted for ethambutol in a clinical trial, the moxi- Study Patients
floxacin-containing regimen produced the most Patients were adults (18 years of age) who had
rapid decline in bacterial load and in the propor- newly diagnosed, previously untreated M. tubercu-
tion of patients with culture negativity at 8 weeks.14 losis infection, as determined by positive results
These findings were confirmed by investigators in on sputum smears on two occasions, with cul-
Brazil.15 In contrast, substituting moxifloxacin for ture-confirmed susceptibility to rifampin and fluo-
isoniazid in an 8-week study resulted in a non- roquinolones. Patients who were coinfected with
significant enhancement in bactericidal effect.16 the human immunodeficiency virus (HIV) were
On the basis of supportive evidence from eligible to participate in the study if the CD4+
phase 2 studies and the uncertain relationships count was at least 250 cells per cubic millimeter
between 8-week bacteriologic data and the du- and they were not already receiving antiretroviral
ration of effective therapy, we designed the Ra therapy. Detailed inclusion and exclusion criteria
pid Evaluation of Moxifloxacin in Tuberculosis are provided in the Supplementary Appendix. All
(REMoxTB) study to determine whether the re- patients provided written or witnessed oral in-
placement of either isoniazid or ethambutol with formed consent.
moxifloxacin would provide effective tuberculo-
sis treatment in 4 months, as compared with the Randomization and Study Treatments
standard 6-month regimen. Randomization was performed with the use of
lists with blocks of variable sizes that were strat-
Me thods ified according to the patient weight group and
study center. During randomization, patients were
Study Design and Oversight assigned a unique study number selected sequen-
REMoxTB was a placebo-controlled, randomized, tially from the appropriate randomization list that
double-blind, phase 3 trial to test the noninferi- corresponded to the treatment pack allocated. Eli-
ority of two moxifloxacin-containing 4-month reg- gible patients were assigned in a 1:1:1 ratio to one
imens, as compared with the standard 6-month of the following daily regimens: a control regimen,
1578 n engl j med 371;17nejm.orgoctober 23, 2014

Moxifloxacin-based Regimens for Tuberculosis
which consisted of isoniazid, rifampin, pyrazin- tions. We tested the susceptibility to moxifloxa-
amide, and ethambutol for 8 weeks, followed by cin using a breakpoint of 0.125 mg per liter. In
18 weeks of isoniazid and rifampin (control group); countries with a high rate of multidrug-resistant
a regimen in which we replaced ethambutol with tuberculosis or quinolone resistance (>5%), initial
moxifloxacin for 17 weeks, followed by 9 weeks sputum samples were tested for rifampin resis-
of placebo (isoniazid group); and a regimen in tance with the use of the GenoType MTBDRplus
which we replaced isoniazid with moxifloxacin assay and GenoType MTBDRsl assay, respectively
for 17 weeks, followed by 9 weeks of placebo (eth- (Hain Lifescience). We used 24-locus mycobacte-
ambutol group). Details about the regimens are rial-interspersed-repetitive-unit (MIRU) analysis
provided in Figure S1 in the Supplementary Ap- to compare the initial strains with the recurrence
pendix. strains.17
In all three groups, drug doses were adjusted
according to patient weight, as described in Ta- Study Outcomes
ble S2 in the Supplementary Appendix. Only stat- The primary efficacy outcome was the propor-
isticians who were responsible for preparing the tion of patients who had bacteriologically or clini-
reports for the independent data and safety moni- cally defined failure or relapse within 18 months
toring committee and essential manufacturing and after randomization (a composite unfavorable
distribution staff members had access to the list outcome). Culture-negative status was defined as
of identifiers matched to the intervention. two negative-culture results at different visits
without an intervening positive result. The date
Study Procedures of culture-negative status was defined as the date
After initial screening and baseline visits, patients of the first negative-culture result. This status
were scheduled for eight weekly visits, which continued until there were two positive cultures,
were followed by eight visits until 18 months af- without an intervening negative culture, or until
ter randomization (Fig. S1 in the Supplementary there was a single positive culture that was not
Appendix). All patients underwent a baseline clin- followed by two negative cultures. Relapse strains
ical examination that included posteroanterior were those shown to be identical on 24-locus
chest radiography, pregnancy testing if relevant, MIRU analysis.
collection of two sputum specimens for micro- The primary safety outcome was the propor-
biologic examination, physical examination, tests tion of patients with grade 3 or 4 adverse events
of visual acuity (Ishihara and Snellen), and uri- that were graded according to a modified version
nalysis. Safety monitoring which included test- of the toxicity criteria of the Division of AIDS of
ing of hepatic function (aspartate aminotransfer- the National Institute of Allergy and Infectious
ase, alanine aminotransferase, and bilirubin), Diseases.
vitamin K, prothrombin time, partial thrombo-
plastin time, blood count (hemoglobin and plate- Statistical Analysis
let count), urea, electrolytes, and creatinine was We determined that a sample size of 633 patients
performed at screening and at weeks 2, 8, 12, per group would provide a power of 85% to show
and 17, with additional liver-function testing at noninferiority of the two moxifloxacin interven-
week 4. tions to the control regimen with a margin of
Sputum was decontaminated with acetylcys- 6 percentage points, assuming a one-sided type I
teinesodium hydroxide, examined microscopi- error of 0.0125 (Bonferroni correction). We esti-
cally, and cultured on LowensteinJensen solid mated that 10% of the patients in each study group
medium and in liquid medium in a Mycobacteria would have a unfavorable outcome and that 15%
Growth Indicator Tube (MGIT) (Becton Dickin- would have outcomes that could not be evaluat-
son). All analyses were performed according to ed. (All definitions are provided in the Supple-
the REMoxTB laboratory and quality manuals mentary Appendix.) This margin of 6 percentage
(available on request). We performed mycobacte- points reflected consultation with clinicians in
rial speciation using the AccuProbe assay (Gen- high-burden countries and reanalysis of previous
Probe), and determined the susceptibility of strains trials showing the effect of shortening treatment
to streptomycin, isoniazid, rifampin, and pyrazin- to 4 months without substituting a new drug.
amide using the MGIT manufacturers instruc- Noninferiority was defined as a between-group

difference of less than 6 percentage points in the lar in the three study groups (Table 1, and Tables
upper boundary of the two-sided 97.5% Wald con- S5 and S6 in the Supplementary Appendix).
fidence interval for the proportion of patients with The most common reason that patients were
an unfavorable outcome. We used a generalized excluded from the modified intention-to-treat
linear model with identity-link function with ad- analysis was that they were found to be ineligible
justment for stratification variables (weight group on the basis of data that were collected before
and study center). We performed both modified randomization (e.g., lack of confirmation of the
intention-to-treat and per-protocol analyses, with diagnosis of tuberculosis or confirmed multidrug-
the latter considered to be the primary analysis. resistant tuberculosis). The most common reasons
In the modified intention-to-treat analysis, we for exclusion from the per-protocol analysis were
excluded patients with resistance to moxifloxa- a change of treatment for reasons other than
cin or rifampin at baseline and those in whom treatment failure and a loss to follow-up (Fig. 1).
the outcome could not be assessed (e.g., patients Of the 1931 patients who underwent randomiza-
who had reinfection). (Detailed definitions are tion, 89% in the isoniazid group, 92% in the
provided in Section 2 in the Supplementary Ap- ethambutol group, and 89% in the control group
pendix.) We also performed a number of sensi- met the requirements for treatment adherence,
tivity and secondary analyses of the primary out- which was based on receipt of approximately 80%
come to test the robustness of the results (Tables of the assigned regimen (see the Supplementary
S3A and S3B in the Supplementary Appendix). Appendix for details).
We used the chi-square test to compare the
patients sputum-culture status at the end of Primary Outcome
8 weeks (intensive phase) across treatment groups In the per-protocol analysis, a favorable outcome
and the log-rank test to compare the time to was reported in 436 patients (85%) in the isonia-
culture-negative status. We used similar methods zid group, as compared with 467 patients (92%)
to analyze other secondary outcomes, including in the control group, for an adjusted absolute dif-
the time to an unfavorable outcome, the status ference of 6.1 percentage points (97.5% confidence
at the end of treatment, the status at 12 and 18 interval [CI], 1.7 to 10.5) favoring the control
months among patients with a favorable outcome group (Table 2, and Fig. S2 in the Supplementary
at end of treatment, and the status at 18 months Appendix). A favorable outcome was reported in
according to a blinded clinical review of the data. 419 patients (80%) in the ethambutol group, for
All patients who received at least one dose of an adjusted absolute difference of 11.4 percent-
a study medication were included in the safety age points (97.5% CI, 6.7 to 16.1), as compared
analysis. The proportions of patients who had at with the control group.
least one grade 3 or 4 adverse event were com- In the modified intention-to-treat analysis, the
pared across treatment groups with the use of the corresponding values also favored the control
chi-square test. group, with a favorable outcome reported in 436
patients (77%) in the isoniazid group, as com-
R e sult s pared with 468 (84%) in the control group, for
an adjusted absolute difference of 7.8 percentage
Study Patients points (97.5% CI, 2.7 to 13.0), and in 419 patients
A total of 2763 patients were screened and 1931 (76%) in the ethambutol group, for an adjusted
underwent randomization: 909 in South Africa, absolute difference of 9.0 percentage points
376 in India, 212 in Tanzania, 136 in Kenya, 119 (97.5% CI, 3.8 to 14.2) (Table 2, and Fig. S2 in the
in Thailand, 69 in Malaysia, 66 in Zambia, 22 in Supplementary Appendix). Results of all sensitiv-
China, and 22 in Mexico (Table S4 in the Supple- ity analyses were consistent with those in the per-
mentary Appendix). The principal reasons for protocol and modified intention-to-treat analy-
ineligibility were a lack of confirmation of smear ses (Table S3A in the Supplementary Appendix).
positivity in the study laboratory, a CD4+ count The most common reason for an unfavorable
of less than 250 cells per cubic millimeter, or outcome was relapse after conversion to culture-
multidrug-resistant disease, as detected by means negative status after the end of active treatment
of the Hain test (Fig. 1). The demographic and (in 46 patients in the isoniazid group, 64 in the
clinical characteristics of the patients were simi- ethambutol group, and 12 patients in the control

2763 Patients were screened
832 Had screening failure

290 Did not have positive smear
196 Had CD4+ count <250/mm3
116 Had initial isolate MDR
230 Had other reason
110 Had late-screening failure
61 Had MDR
20 Had protocol violation
29 Did not have confirmed
tuberculosis
1931 Underwent randomization
640 Were assigned to control group 655 Were assigned to isoniazid group 636 Were assigned to ethambutol group
85 Were excluded 87 Were excluded 85 Were excluded

40 Had late-screening 38 Had late-screening 32 Had late-screening
failures failures failures
during treatment during treatment during treatment
10 Had reinfections 13 Had reinfections 19 Had reinfections
during follow-up during follow-up during follow-up
phase phase phase
555 Were included in modified intention- 568 Were included in modified intention- 551 Were included in modified intention-
to-treat analysis to-treat analysis to-treat analysis
45 Were excluded
54 Were excluded
30 Changed treatment
42 Changed treatment 27 Were excluded
(not failure)
(not failure) 21 Changed treatment
13 Were lost to follow-
10 Were lost to follow- (not failure)
up before 6 mo
up before 6 mo 6 Were lost to follow-
2 Had additional
2 Had inadequate up before 6 mo
major protocol
treatment
violations
510 Were included in per-protocol 514 Were included in per-protocol 524 Were included in per-protocol
analysis analysis analysis
Figure 1. Enrollment and Outcomes.

MDR denotes multidrug resistance.
group). A similar pattern of results was seen in the Subgroup Analyses

modified intention-to-treat analysis (Table 2). There There was no evidence that between-group differ-
were no unequivocal cases of acquired resistance, ences in the primary outcome varied according
but there were four cases of possible resistance to HIV status, region, recruitment site, age group,
one in the ethambutol group (for moxifloxacin) isoniazid susceptibility, or cavitation. The propor-
and three in the control group (two for rifampin tion of unfavorable outcomes among female pa-
and one for isoniazid) which require future tients, as compared with male patients, was simi-
whole-genome sequencing for interpretation. lar in the three study groups (test of interaction,

Table 1. Baseline Characteristics of Patients in the Per-Protocol Population.*
Control Isoniazid Ethambutol

Group Group Group All Patients
Characteristic (N=510) (N=514) (N=524) (N=1548)
number of patients (percent)

Male sex 356 (70) 351 (68) 369 (70) 1076 (70)
Weight group
<40 kg 50 (10) 44 (9) 58 (11) 152 (10)
4045 kg 80 (16) 90 (18) 82 (16) 252 (16)
>4555 kg 206 (40) 210 (41) 204 (39) 620 (40)
>5575 kg 161 (32) 158 (31) 174 (33) 493 (32)
>75 kg 13 (3) 12 (2) 6 (1) 31 (2)
Age group
<25 yr 160 (31) 162 (32) 146 (28) 468 (30)
2535 yr 145 (28) 162 (32) 175 (33) 482 (31)
>35 yr 205 (40) 190 (37) 203 (39) 598 (39)
Race or ethnic group
Black 238 (47) 210 (41) 237 (45) 685 (44)
Asian 160 (31) 154 (30) 161 (31) 475 (31)
Mixed race 111 (22) 148 (29) 126 (24) 385 (25)
Other 1 (<1) 2 (<1) 0 3 (<1)
Smoking status
Never 246 (48) 231 (45) 230 (44) 707 (46)
Past 119 (23) 111 (22) 134 (26) 364 (24)
Current 145 (28) 172 (33) 160 (31) 477 (31)
HIV positivity 38 (7) 37 (7) 35 (7) 110 (7)
Drug resistance
Isoniazid 29 (6) 34 (7) 39 (7) 102 (7)
Pyrazinamide 14 (3) 7 (1) 6 (1) 27 (2)
Cavitation 368 (72) 357 (69) 367 (70) 1092 (71)
Time to positivity on MGIT sputum culture
5 days 266 (52) 263 (51) 258 (49) 787 (51)
<5 days 229 (45) 239 (46) 254 (48) 722 (47)
Not available 15 (3) 12 (2) 12 (2) 39 (3)
* There were no significant differences between the study groups. HIV denotes human immunodeficiency virus, and
MGIT Mycobacteria Growth Indicator Tube.
The median body-mass index (the weight in kilograms divided by the square of the height in meters) was 18.4 (range,
12.1 to 50.9) in the control group, 18.3 (range, 12.0 to 33.1) in the isoniazid group, 18.4 (range, 12.2 to 32.6) in the eth-
ambutol group, and 18.3 (range, 12.0 to 50.9) for all patients.
Race or ethnic group was reported by the investigator. Asian category included both South Asians and East Asians.
A single patient had missing HIV status.
Resistance results were missing for isoniazid in 24 patients and for pyrazinamide in 27 patients.
Cavitation status was missing for 148 patients.

Table 2. Primary Efficacy Analysis in Per-Protocol and Modified Intention-to-Treat Populations.*
Variable Per-Protocol Analysis Modified Intention-to-Treat Analysis

Control Isoniazid Ethambutol All Control Isoniazid Ethambutol All
Group Group Group Patients Group Group Group Patients
(N=510) (N=514) (N=524) (N=1548) (N=555) (N=568) (N=551) (N=1674)
Favorable outcome no. (%)
Patients with outcome 467 (92) 436 (85) 419 (80) 1322 (85) 468 (84) 436 (77) 419 (76) 1323 (79)
Culture-negative status at 18 mo 409 (80) 389 (76) 367 (70) 1165 (75) 410 (74) 389 (68) 367 (67) 1166 (70)
Unable to produce sputum 0 2 (<1) 0 2 (<1) 0 2 (<1) 0 2 (<1)
Unable to produce sputum at 49 (10) 31 (6) 35 (7) 115 (7) 49 (9) 31 (5) 35 (6) 115 (7)
18 mo but culture-
negative status earlier
Missing data on LJ culture at 9 (2) 14 (3) 17 (3) 40 (3) 9 (2) 14 (2) 17 (3) 40 (2)
18 mo and MGIT
negative
Unfavorable outcome no. (%)
Patients with outcome 43 (8) 78 (15) 105 (20) 226 (15) 87 (16) 132 (23) 132 (24) 351 (21)
6-Mo treatment phase
Nonviolent death 5 (1) 6 (1) 7 (1) 18 (1) 5 (1) 6 (1) 7 (1) 18 (1)
Treatment failure
Culture-confirmed 3 (1) 4 (1) 1 (<1) 8 (1) 3 (1) 4 (1) 1 (<1) 8 (<1)
Not culture-confirmed 4 (1) 1 (<1) 4 (1) 9 (1) 4 (1) 1 (<1) 4 (1) 9 (1)
Adverse reaction NA NA NA NA 18 (3) 15 (3) 9 (2) 42 (3)
Withdrawal of consent NA NA NA NA 8 (1) 18 (3) 8 (1) 34 (2)
Relocation NA NA NA NA 2 (<1) 4 (1) 4 (1) 10 (1)
Other investigator decision NA NA NA NA 2 (<1) 5 (1) 0 7 (<1)
No completion of treatment NA NA NA NA 13 (2) 10 (2) 6 (1) 29 (2)
Follow-up
Relapse after culture-negative 12 (2) 46 (9) 64 (12) 122 (8) 13 (2) 46 (8) 64 (12) 123 (7)
status
Retreated for tuberculosis 14 (3) 17 (3) 27 (5) 58 (4) 14 (3) 18 (3) 27 (5) 59 (4)
Death from tuberculosis or 2 (<1) 0 0 2 (<1) 2 (<1) 0 0 2 (<1)
respiratory distress
No culture-negative status
Ever 1 (<1) 1 (<1) 0 2 (<1) 1 (<1) 2 (<1) 0 3 (<1)
At last visit 2 (<1) 3 (1) 2 (<1) 7 (<1) 2 (<1) 3 (1) 2 (<1) 7 (<1)
Adjusted difference from control NA 6.1 11.4 NA NA 7.8 9.0 NA
in rate of unfavorable (1.710.5) (6.716.1) (2.713.0) (3.814.2)
outcome percent
age points (97.5% CI)
* The treatment phase was defined as any time from randomization to 32 weeks after randomization (26 weeks plus 6-week window). LJ de-
notes LowensteinJensen solid medium, and NA not applicable.
During follow-up, the relapse and retreatment categories include patients during the scheduled end of active treatment (after month 4 for
the moxifloxacin-containing groups and month 6 for the control group). In the per-protocol analysis, data from 24-locus mycobacterial-inter-
spersed-repetitive-unit analysis were missing for 9 of 17 patients with treatment failure, 42 of 122 patients with relapse, and 38 of 58 pa-
tients who were retreated for tuberculosis.
Listed are patients who were receiving active treatment in whom treatment failed.

Time to Culture-Negative Status

A Time to Unfavorable Outcome
In KaplanMeier analyses, patients in the isonia-
1.0
zid group and the ethambutol group had conver-
Probability of Favorable Outcome
0.9
sion to culture-negative status sooner than those
0.8
in the control group in sputum analyses with the
0.7
use of LowensteinJensen solid medium (Fig. 2B)
0.6
0.5
and MGIT medium (Fig. S3 and Table S7 in the
0.4
Supplementary Appendix) (P<0.01 for both analy
0.3
ses). More patients receiving the moxifloxacin-
Control group
0.2
containing regimens had culture-negative status
Isoniazid group
0.1 Ethambutol group
at 8 weeks, but the difference was not significant
0.0
(Table S8 in the Supplementary Appendix).
0 4 8 17 26 39 52 65 78
Weeks since Randomization Time to an Unfavorable Outcome
No. at Risk In the per-protocol analyses, the time to an unfa-
Control 600 563 533 493 472 vorable outcome was shorter in the isoniazid group
Isoniazid 617 570 522 459 439
Ethambutol 604 568 523 445 425 than in the control group (hazard ratio, 1.87;
97.5% CI, 1.07 to 2.67) and was further reduced
B Time to Culture-Negative Status in the ethambutol group (hazard ratio, 2.56;
1.0 97.5% CI, 1.51 to 3.60) (Fig. 2A, and Table S9 in
Probability of Culture-Positive Status
Control group
0.9 the Supplementary Appendix).
Isoniazid group
0.8
Ethambutol group
0.7 Adverse Events
0.6 There were no significant between-group differ-
0.5 ences in the incidence of grade 3 or 4 adverse
0.4 events, with reports of events in 127 patients (19%)
0.3 in the isoniazid group and 111 patients (17%) in
0.2 the ethambutol group, as compared with 123 pa-
0.1 tients (19%) in the control group (Table 3). A to-
0.0 tal of 349 serious adverse events occurred in 173
0 4 8 12 17 22 26
patients, with 246 events occurring during the
Weeks since Randomization
treatment period and 103 during follow-up. There
No. at Risk were 43 deaths (16 during the treatment period
Control 600 465 183 122 64 19
Isoniazid 617 459 154 76 21 9 and 27 during follow-up) during the study, 30 of
Ethambutol 604 449 141 79 30 9 which were deemed to be tuberculosis-related
(Table S10 in the Supplementary Appendix). Over-
Figure 2. KaplanMeier Estimates of the Time to an Unfavorable Outcome all, the numbers of serious adverse events, types
and Conversion to Culture-Negative Status.
of events, and numbers of patients with events
Panel A shows that the time until patients had an unfavorable outcome was
(including the number of deaths) were similar in
shorter in the isoniazid group than in the control group (hazard ratio, 1.25
[97.5% CI, 1.08 to 1.42]) and was further reduced in the ethambutol group the three study groups during both the treatment
(hazard ratio, 1.21 [97.5% CI, 1.05 to 1.37]). Panel B shows the time until period and the follow-up period.
conversion to culture-negative status, which occurred sooner in the isonia- There were no significant between-group dif-
zid group and the ethambutol group than in the control group, according to ferences in the incidence of adverse events of
analyses of sputum samples cultured in LowensteinJensen solid medium.
special interest, including tendinopathy, seizure,
Patients who were excluded from the primary per-protocol analysis were in-
cluded in this analysis, but data were censored at the time of exclusion clinically significant cardiac toxicity, hypoglyce-
from the per-protocol analysis. mia or hyperglycemia, and peripheral neuropathy.
The proportions of events were similar in the
study groups when all adverse events were con-
P=0.004 for the isoniazid group and P=0.02 for sidered. There were no significant differences in
the ethambutol group) (Table S3B in the Supple- any measures of biochemical, hematologic, or
mentary Appendix). hepatic safety.

Table 3. Safety Analysis.*
Control Group Isoniazid Group Ethambutol Group All Patients

Adverse Event (N=639) (N=655) (N=636) (N=1930)
number of patients (percent)
During treatment phase
or follow-up
Any 123 (19) 127 (19) 111 (17) 361 (19)
Grade 3 only 83 (13) 90 (14) 82 (13) 255 (13)
Grade 4 40 (6) 37 (6) 29 (5) 106 (5)
Serious adverse event 59 (9) 62 (9) 52 (8) 173 (9)
Death
Any 16 (3) 15 (2) 12 (2) 43 (2)
Tuberculosis-related 11 (2) 10 (2) 9 (1) 30 (2)
During treatment phase only
Any 111 (17) 105 (16) 99 (16) 315 (16)
Grade 3 only 76 (12) 71 (11) 73 (11) 220 (11)
Grade 4 35 (5) 34 (5) 26 (4) 95 (5)
Serious adverse event 46 (7) 40 (6) 35 (6) 121 (6)
Death
Any 5 (1) 6 (1) 5 (1) 16 (1)
Tuberculosis-related 4 (1) 6 (1) 5 (1) 15 (1)
* Listed are all patients who had at least one grade 3 or 4 adverse event. The safety population includes all patients who
underwent randomization and who received at least one dose of a study drug. One patient who underwent randomiza-
tion but did not receive a study drug was excluded from the safety analysis. A detailed list of serious adverse events is
provided in Table S10 in the Supplementary Appendix.
Discussion over a 4-month period. The similarity in outcome

among women in the isoniazid group and the
In this phase 3 trial, we aimed to determine control group may represent a chance finding
whether the promising data that were observed but merits further investigation.
for moxifloxacin in studies in animals and phase It has been previously suggested that Asian
2 studies translated into an effective reduction in patients often have a more chronic form of tu-
the duration of the standard tuberculosis treat- berculosis with a different clinical course than
ment regimen. The trial showed that the substi- that in African patients,18,19 but we did not see
tution of moxifloxacin in 4-month regimens based any evidence of variation in clinical-disease out-
on either isoniazid or ethambutol did not meet come in the different racial groups. Our ap-
the margin for noninferiority, as compared with proach in the conduct of this trial, including
the 6-month control regimen. The same conclu- standardized laboratory methods and clinical
sions were reached when the outcome was deter- management, has resulted in consistent results
mined with the use of MGIT cultures of sputum across more than 20 sensitivity analyses, with
samples. Among patients receiving the two moxi- minimal variation among study centers on dif-
floxacin-containing regimens, a small number ferent continents.
had treatment failures, but a larger number had In comparison with other trials that used
a relapse after the end of active treatment. The fluoroquinolones in a 4-month regimen, the rates
difference between the isoniazid group and the of an unfavorable outcome in the experimental
ethambutol group may be due to the bactericidal groups in our study are lower than those in the
effect of isoniazid or the presence of three drugs RIFAQUIN regimen20 and similar to those found

in the OFLOTUB trial.21 In trials evaluating sibility that a 4-month regimen might be effec-
4-month streptomycin-containing regimens that tive, although the 95% confidence interval ranged
were performed in the 1970s in East Africa and from 1.15 to 2.60. In our study, with more than
Singapore, rates of relapse ranged from 11 to 40% 600 patients in each group, we found a more
after 2 years of follow-up.18,19 precise estimate of the hazard ratio to be 1.25
In our study, a daily regimen of moxifloxacin (95% CI, 1.10 to 1.40), a result that is within the
in combination with standard antituberculosis confidence interval found previously14 but with a
agents for 4 months had an acceptable side-effect smaller effect, which would seem unlikely to
profile. We did not find any evidence of either merit progression to a phase 3 trial. Thus, such
hypoglycemia or hyperglycemia or tendinopathies short trials may correlate with long-term out-
that have been associated with fluoroquino- comes, but the small sample size and resulting
lones,22,23 nor did we find evidence of increased wide confidence intervals limit their ability to
hepatic dysfunction, a potential concern in regi- predict treatment shortening.
mens containing moxifloxacin or lacking isonia- This limitation suggests that efficient drug
zid.24 There was no clinical evidence of cardiac development for tuberculosis may require a dif-
toxicity, although electrocardiography was not ferent approach. Instead of relying on the results
performed systematically. These are important of 2-month phase 2 trials to select candidate regi-
findings for future regimens that may use moxi- mens for phase 3 studies, investigators might find
floxacin in combination with other agents in tu- that the most efficient approach is to conduct
berculosis treatment.25 phase 3 trials as quickly as possible while estab-
Our findings raise questions about progres- lishing more feasible and less costly approaches
sion decisions throughout the development path- to performing these studies. Possible improve-
way for tuberculosis drugs. Data from studies in ments could include larger noninferiority margins,
mice predicted that the inclusion of moxifloxa- permitting smaller sample sizes, and building
cin would result in a reduction of 1 to 2 months multiple treatment durations into each study.
in the treatment duration, as compared with stan- In conclusion, in patients with uncomplicated,
dard therapy.12,13 In our study of such treatment smear-positive tuberculosis, the noninferiority of
shortening, the moxifloxacin-containing regimens the moxifloxacin-containing regimens was not
did not work adequately, suggesting that the mu- shown, despite the fact that these regimens had
rine model may have overpredicted the sterilizing better bactericidal activity than the standard con-
potency of moxifloxacin in this regimen. trol regimen.30
More important is the observed poor predict-
ability of culture conversion for long-term out- Supported by the Global Alliance for TB Drug Development
comes. Although 2-month culture conversion is with support from the Bill and Melinda Gates Foundation, the
European and Developing Countries Clinical Trials Partnership,
associated with relapse-free cure, this observed U.S. Agency for International Development, U.K. Department
correlation in populations is not strong enough to for International Development, Directorate General for Interna-
reliably predict outcomes for individual patients or tional Cooperation of the Netherlands, Irish Aid, Australia De-
partment of Foreign Affairs and Trade, and National Institutes
definitively guide the selection of regimen in drug of Health, AIDS Clinical Trials Group and by grants from the
development.26,27 This finding underlines the im- National Institute of Allergy and Infectious Diseases (NIAID)
portance of the content and duration of treatment (UM1AI068634, UM1 AI068636, and UM1AI106701) and by NIAID
grants to the University of KwaZulu Natal, South Africa,AIDS
in the following weeks.28 Four 2-month studies Clinical Trials Group (ACTG) site 31422 (1U01AI069469); to the
of the inclusion of moxifloxacin in the standard Perinatal HIV Research Unit, Chris Hani Baragwanath Hospital,
regimen have been reported, with variable re- South Africa, ACTG site 12301 (1U01AI069453);and to the Dur-
ban International Clinical Trials Unit, South Africa, ACTG site
sults.14-16,29 The only study to report a hazard 11201 (1U01AI069426); Bayer Healthcare for the donation of
ratio for the time to culture conversion was that moxifloxacin; and Sanofi for the donation of rifampin.
of Rustomjee et al.,14 who, in a study involving Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
approximately 50 patients per group, found that We thank all the patients for their participation in the study;
the hazard ratio for the time to culture conver- the nursing and laboratory staff; all those who advised, volun-
sion for the moxifloxacin-containing regimen, as teered, or otherwise supported community engagement around
the REMoxTB clinical trial sites; and Sangita Patel, Marie Mes-
compared with the standard regimen, was 1.73, sina, Melanie Barry, Derek Ambrosino, and Joanna Breitstein for
indicating a shorter duration. This raised the pos- providing administrative support.

references
1. Gillespie SH, Billington O. Activity of of pulmonary tuberculosis: a prospective, susceptible pulmonary tuberculosis: main
moxifloxacin against mycobacteria. J An- randomized study. Antimicrob Agents efficacy and safety results of the OFLOTUB
timicrob Chemother 1999;44:393-5. Chemother 2004;48:780-2. Trial. Presented at the 41st World Confer-
2. Falzon D, Jaramillo E, Schnemann HJ, 12. Nuermberger EL, Yoshimatsu T, Tyagi ence on Lung Health of the International
et al. WHO guidelines for the program- S, et al. Moxifloxacin-containing regimen Union Against Tuberculosis and Lung Dis-
matic management of drug-resistant tu- greatly reduces time to culture conversion ease, Paris, October 30November 3, 2013.
berculosis: 2011 update. Eur Respir J 2011; in murine tuberculosis. Am J Respir Crit abstract.
38:516-28. Care Med 2004;169:421-6. 22. Park-Wyllie LY, Juurlink DN, Kopp A,
3. Johnston JC, Shahidi NC, Sadatsafavi 13. Nuermberger EL, Yoshimatsu T, Tyagi et al. Outpatient gatifloxacin therapy and
M, Fitzgerald JM. Treatment outcomes of S, et al. Moxifloxacin-containing regimens dysglycemia in older adults. N Engl J Med
multidrug-resistant tuberculosis: a system- of reduced duration produce a stable cure 2006;354:1352-61.
atic review and meta-analysis. PLoS One in murine tuberculosis. Am J Respir Crit 23. Khaliq Y, Zhanel GG. Fluoroquino-
2009;4(9):e6914. Care Med 2004;170:1131-4. lone-associated tendinopathy: a critical
4. Ginsberg AM. Tuberculosis drug de- 14. Rustomjee R, Lienhardt C, Kanyok T, review of the literature. Clin Infect Dis
velopment: progress, challenges, and the et al. A phase II study of the sterilising 2003;36:1404-10.
road ahead. Tuberculosis (Edinb) 2010;90: activities of ofloxacin, gatifloxacin and 24. Update: fatal and severe liver injuries
162-7. moxifloxacin in pulmonary tuberculosis. associated with rifampin and pyrazin-
5. Grossman RF, Hsueh P-R, Gillespie Int J Tuberc Lung Dis 2008;12:128-38. amide for latent tuberculosis infection,
SH, Blasi F. Community-acquired pneu- 15. Conde MB, Efron A, Loredo C, et al. and revisions in American Thoracic Soci-
monia and tuberculosis: differential diag- Moxifloxacin versus ethambutol in the ety/CDC recommendations United
nosis and the use of fluoroquinolones. Int initial treatment of tuberculosis: a double- States, 2001. MMWR Morb Mortal Wkly
J Infect Dis 2014;18:14-21. blind, randomised, controlled phase II tri- Rep 2001;50:733-5.
6. Mller M, Stass H, Brunner M, Mller al. Lancet 2009;373:1183-9. 25. Diacon AH, Dawson R, von Groote-
JG, Lackner E, Eichler HG. Penetration of 16. Dorman SE, Johnson JL, Goldberg S, Bidlingmaier F, et al. 14-Day bactericidal
moxifloxacin into peripheral compart- et al. Substitution of moxifloxacin for iso- activity of PA-824, bedaquiline, pyrazin-
ments in humans. Antimicrob Agents niazid during intensive phase treatment amide, and moxifloxacin combinations:
Chemother 1999;43:2345-9. of pulmonary tuberculosis. Am J Respir a randomised trial. Lancet 2012;380:986-
7. Lubasch A, Keller I, Borner K, Koeppe Crit Care Med 2009;180:273-80. 93.
P, Lode H. Comparative pharmacokinet- 17. Bryant JM, Harris SR, Parkhill J, et al. 26. Aber VR, Nunn AJ. Short term chemo-
ics of ciprofloxacin, gatifloxacin, grepa- Whole-genome sequencing to establish therapy of tuberculosis: factors affecting
floxacin, levofloxacin, trovafloxacin, and relapse or re-infection with Mycobacterium relapse following short term chemothera-
moxifloxacin after single oral administra- tuberculosis: a retrospective observational py. Bull Int Union Tuberc 1978;53:276-80.
tion in healthy volunteers. Antimicrob study. Lancet Respir Med 2013;1:786-92. (In French.)
Agents Chemother 2000;44:2600-3. 18. East African/British Medical Research 27. Phillips PPJ, Fielding K, Nunn AJ. An
8. Sullivan JT, Woodruff M, Lettieri J, et Council. Controlled clinical trial of five evaluation of culture results during treat-
al. Pharmacokinetics of a once-daily oral short-course (4-month) chemotherapy reg- ment for tuberculosis as surrogate end-
dose of moxifloxacin (Bay 12-8039), a new imens in pulmonary tuberculosis: second points for treatment failure and relapse.
enantiomerically pure 8-methoxy quino- report of the 4th study. Am Rev Respir Dis PLoS One 2013;8(5):e63840.
lone. Antimicrob Agents Chemother 1999; 1981;123:165-70. 28. Nunn AJ, Jindani A, Enarson DA. Re-
43:2793-7. 19. Singapore Tuberculosis Service/Brit- sults at 30 months of a randomised trial
9. Miyazaki E, Miyazaki M, Chen JM, ish Medical Research Council. Clinical of two 8-month regimens for the treat-
Chaisson RE, Bishai WR. Moxifloxacin trial of six-month and four-month regi- ment of tuberculosis. Int J Tuberc Lung
(BAY12-8039), a new 8-methoxyquinolone, mens of chemotherapy in the treatment of Dis 2011;15:741-5.
is active in a mouse model of tuberculosis. pulmonary tuberculosis: the results up to 29. Burman WJ, Goldberg S, Johnson JL,
Antimicrob Agents Chemother 1999;43: 30 months. Tubercle 1981;62:95-102. et al. Moxifloxacin versus ethambutol in
85-9. 20. Jindani A. Recent progress in TB clin- the first 2 months of treatment for pul-
10. Gosling RD, Uiso LO, Sam NE, et al. ical trials: results of the RIFAQUIN trial. monary tuberculosis. Am J Respir Crit
The bactericidal activity of moxifloxacin Presented at the 41st World Conference Care Med 2006;174:331-8.
in patients with pulmonary tuberculosis. on Lung Health of the International 30. Nunn AJ, Phillips PPJ, Mitchison DA.
Am J Respir Crit Care Med 2003;168:1342- Union Against Tuberculosis and Lung Timing of relapse in short-course chemo-
5. Disease, Paris, October 30November 3, therapy trials for tuberculosis. Int J Tuberc
11. Pletz MWR, De Roux A, Roth A, Neu- 2013. abstract. Lung Dis 2010;14:241-2.
mann K-H, Mauch H, Lode H. Early bacteri- 21. Merle CS, Fielding KL, Lapujade O, et Copyright 2014 Massachusetts Medical Society.
cidal activity of moxifloxacin in treatment al. 4-Month regimen for treating drug-
receive immediate notification when an article

is published online first
To be notified by e-mail when Journal articles

are published Online First, sign up at NEJM.org.


New England Journal Medicine: The of

Uploaded by

Copyright:

Available Formats

New England Journal Medicine: The of

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

New England Journal Medicine: The of

Uploaded by

Copyright:

Available Formats

new england

Four-Month Moxifloxacin-Based Regimens

n engl j med 371;17nejm.orgoctober 23, 2014 1577

1578 n engl j med 371;17nejm.orgoctober 23, 2014

The New England Journal of Medicine

n engl j med 371;17nejm.orgoctober 23, 2014 1579

1580 n engl j med 371;17nejm.orgoctober 23, 2014

The New England Journal of Medicine

2763 Patients were screened

832 Had screening failure

1931 Underwent randomization

85 Were excluded 87 Were excluded 85 Were excluded

Figure 1. Enrollment and Outcomes.

group). A similar pattern of results was seen in the Subgroup Analyses

n engl j med 371;17nejm.orgoctober 23, 2014 1581

Table 1. Baseline Characteristics of Patients in the Per-Protocol Population.*

Control Isoniazid Ethambutol

number of patients (percent)

1582 n engl j med 371;17nejm.orgoctober 23, 2014

The New England Journal of Medicine

Table 2. Primary Efficacy Analysis in Per-Protocol and Modified Intention-to-Treat Populations.*

Variable Per-Protocol Analysis Modified Intention-to-Treat Analysis

n engl j med 371;17nejm.orgoctober 23, 2014 1583

Time to Culture-Negative Status

1584 n engl j med 371;17nejm.orgoctober 23, 2014

The New England Journal of Medicine

Table 3. Safety Analysis.*

Control Group Isoniazid Group Ethambutol Group All Patients

Discussion over a 4-month period. The similarity in outcome

n engl j med 371;17nejm.orgoctober 23, 2014 1585

1586 n engl j med 371;17nejm.orgoctober 23, 2014

The New England Journal of Medicine

receive immediate notification when an article

To be notified by e-mail when Journal articles

n engl j med 371;17nejm.orgoctober 23, 2014 1587

You might also like