Nejmoa 2104535

new england
The
journal of medicine
established in 1812 March 10, 2022 vol. 386 no. 10
Shorter Treatment for Nonsevere Tuberculosis in African

and Indian Children
A. Turkova, G.H. Wills, E. Wobudeya, C. Chabala, M. Palmer, A. Kinikar, S. Hissar, L. Choo, P. Musoke, V. Mulenga,
V. Mave, B. Joseph, K. LeBeau, M.J. Thomason, R.B. Mboizi, M. Kapasa, M.M. van der Zalm, P. Raichur,
P.K. Bhavani, H. McIlleron, A.-M. Demers, R. Aarnoutse, J. Love‑Koh, J.A. Seddon, S.B. Welch, S.M. Graham,
A.C. Hesseling, D.M. Gibb, and A.M. Crook, for the SHINE Trial Team*
a bs t r ac t
BACKGROUND
Two thirds of children with tuberculosis have nonsevere disease, which may be The authors’ full names, academic de-
treatable with a shorter regimen than the current 6-month regimen. grees, and affiliations are listed in the
Appendix. Dr. Turkova can be contacted
METHODS at a.turkova@ucl.ac.uk or at the Medical
We conducted an open-label, treatment-shortening, noninferiority trial involving Research Council Clinical Trials Unit, Uni-
versity College London, 90 High Holborn,
children with nonsevere, symptomatic, presumably drug-susceptible, smear-negative London WC1V 6LJ, United Kingdom.
tuberculosis in Uganda, Zambia, South Africa, and India. Children younger than
*The members of the SHINE Trial Team
16 years of age were randomly assigned to 4 months (16 weeks) or 6 months (24 are listed in the Supplementary Appen-
weeks) of standard first-line antituberculosis treatment with pediatric fixed-dose dix, available at NEJM.org.
combinations as recommended by the World Health Organization. The primary
Drs. Gibb and Crook contributed equally
efficacy outcome was unfavorable status (composite of treatment failure [exten- to this article.
sion, change, or restart of treatment or tuberculosis recurrence], loss to follow-up
This article was updated on March 18,
during treatment, or death) by 72 weeks, with the exclusion of participants who 2022, at NEJM.org.
did not complete 4 months of treatment (modified intention-to-treat population).
This is the New England Journal of Medi-
A noninferiority margin of 6 percentage points was used. The primary safety out- cine version of record, which includes all
come was an adverse event of grade 3 or higher during treatment and up to 30 days Journal editing and enhancements. The
after treatment. Author Final Manuscript, which is the au-
thor’s version after external peer review
RESULTS and before publication in the Journal, is
From July 2016 through July 2018, a total of 1204 children underwent randomiza- available under a CC BY license at
PMC7612496.
tion (602 in each group). The median age of the participants was 3.5 years (range,
2 months to 15 years), 52% were male, 11% had human immunodeficiency virus N Engl J Med 2022;386:911-22.
infection, and 14% had bacteriologically confirmed tuberculosis. Retention by 72 DOI: 10.1056/NEJMoa2104535
Copyright © 2022 Massachusetts Medical Society.
weeks was 95%, and adherence to the assigned treatment was 94%. A total of 16
participants (3%) in the 4-month group had a primary-outcome event, as com- CME
pared with 18 (3%) in the 6-month group (adjusted difference, −0.4 percentage at NEJM.org
points; 95% confidence interval, −2.2 to 1.5). The noninferiority of 4 months of
treatment was consistent across the intention-to-treat, per-protocol, and key sec-
ondary analyses, including when the analysis was restricted to the 958 participants
(80%) independently adjudicated to have tuberculosis at baseline. A total of 95 par-
ticipants (8%) had an adverse event of grade 3 or higher, including 15 adverse drug
reactions (11 hepatic events, all but 2 of which occurred within the first 8 weeks,
when the treatments were the same in the two groups).
CONCLUSIONS
Four months of antituberculosis treatment was noninferior to 6 months of treat-
ment in children with drug-susceptible, nonsevere, smear-negative tuberculosis.
(Funded by the U.K. Medical Research Council and others; SHINE ISRCTN number,
ISRCTN63579542.)
n engl j med 386;10 nejm.org March 10, 2022 911
The New England Journal of Medicine
Downloaded from nejm.org on February 8, 2023. For personal use only. No other uses without permission.
Copyright © 2022 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
M
ore than 1 million children be- comparing 4 months (16 weeks) of antitubercu-
come ill with tuberculosis annually, losis treatment with the standard 6 months (24
and almost 20% of them die,1,2 but weeks) of treatment using WHO-recommended
children have historically been excluded from pediatric doses.10 All the relevant national and
clinical efficacy trials of antituberculosis treat- local ethics committees and the University Col-
A Quick Take
is available at ment. This situation is due in part to low rates lege London research ethics committee approved
NEJM.org of bacteriologic confirmation of disease among the trial protocol, which is available with the full
children, given high rates of paucibacillary dis- text of this article at NEJM.org. Caregivers pro-
ease and difficulties in obtaining respiratory vided written informed consent, and children
specimens. Treatment recommendations for chil- gave assent as appropriate. The authors vouch
dren are therefore extrapolated from trials in- for the completeness and accuracy of the data
volving adults for which the criteria for treat- and for the fidelity of the trial to the protocol.
ment entry have often included smear-positive
respiratory disease. Participants
In contrast to adults, most children have non- Children younger than 16 years of age who had
severe, smear-negative tuberculosis.3,4 Although symptomatic nonsevere tuberculosis that was
spontaneous resolution has been described,5 it is smear-negative on a respiratory sample and who
generally agreed that treatment is appropriate in were due to start first-line antituberculosis treat-
children with mild forms of tuberculosis be- ment were eligible for enrollment. Nonsevere tu-
cause of the risk of disease progression and dis- berculosis included respiratory tuberculosis con-
semination, particularly among the youngest fined to one lobe (opacification of <1 lobe) with
children or those with concurrent human immu- no cavities, no signs of miliary tuberculosis, no
nodeficiency virus (HIV) infection.6,7 It is likely complex pleural effusion, and no clinically sig-
that nonsevere forms of tuberculosis could be nificant airway obstruction or peripheral lymph-
treated with shorter durations of therapy, but node tuberculosis (see the protocol).11
data are limited regarding the shortening of treat-
ment for drug-susceptible tuberculosis in children. Trial Procedures
Current international guidelines recommend Children were seen at screening, at enrollment
6 months of antituberculosis treatment in chil- (randomization), and at weeks 2, 4, 8, 12, 16, 20,
dren, which is the same duration as in adults. 24, 28, 36, 48, 60, and 72. Screening procedures
Early pharmacokinetic studies of first-line anti- included history taking to identify contacts with
tuberculosis treatment showed lower drug expo- persons with tuberculosis and an evaluation of
sures in young children than in adults and led to symptoms associated with tuberculosis; perfor-
recommendations of increased drug doses from mance of a Mantoux tuberculin skin test or inter-
the World Health Organization (WHO) in 2010.8 feron γ–release assay, where available; radiogra-
New dispersible-formulation tablets with a fixed- phy of the chest; and obtaining at least two
dose combination were developed to enable the use respiratory samples (gastric aspirate, expectorated
of revised doses and became available in 2015.9,10 sputum, or induced sputum) for smear micros-
In the SHINE trial, we investigated whether copy, Xpert MTB/RIF assay (Xpert, Cepheid),
4 months of antituberculosis treatment would mycobacterial culture (Löwenstein–Jensen solid-
be as good as 6 months of treatment in children culture medium or a mycobacteria growth indi-
with nonsevere, smear-negative, presumably drug- cator tube with liquid culture system), and drug-
susceptible tuberculosis, using the new fixed- susceptibility testing. In children with peripheral
dose combination formulations. In addition, we lymph-node tuberculosis, a fine-needle aspirate
evaluated the cost-effectiveness of the 4-month was obtained. A baseline radiograph of the chest
treatment approach. was assessed by site clinicians for severe respira-
tory tuberculosis.3 Blood samples for biochemi-
cal and hematologic testing (in all children) and
Me thods
for HIV type 1 viral load and CD4 count (in
Trial Design and Oversight children with HIV infection) were obtained at
We conducted an international, open-label, parallel- screening and at scheduled follow-up visits.
group, randomized, controlled, noninferiority trial Children with confirmed drug resistance or with
912 n engl j med 386;10 nejm.org March 10, 2022

Shorter TB Treatment in Children
known exposure to an adult with any drug-resis- and pyrazinamide (fixed-dose combination for-
tant tuberculosis were excluded from the trial. mulation), with or without ethambutol accord-
A symptom checklist was completed and a ing to local guidelines (intensive phase). This
clinical examination was performed at each visit treatment was followed by standard therapy with
to detect tuberculosis-associated symptoms and isoniazid and rifampin (continuation phase) in a
adverse events. Repeat respiratory samples were fixed-dose combination for either 8 weeks in the
obtained if previous respiratory specimens were 4-month group (intervention) or 16 weeks in the
positive on microbiologic testing, if such assess- 6-month group (control). All antituberculosis
ment was clinically indicated to assess recurrence treatment was administered 7 days per week, on
or treatment failure, or if a new contact with the basis of WHO weight bands for tuberculosis
drug-resistant tuberculosis was identified. treatment, with the use of child-friendly formu-
Radiographs of the chest were retrospectively lations10 that have been found to be acceptable
reviewed centrally by two independent experts. by trial participants and caregivers.16 Directly
Radiographs with discordant interpretations at observed treatment by health care workers was
the primary reading were reviewed by a third not used.
expert, and the majority opinion was used. The
radiographic image review was conducted in a Primary and Secondary Outcomes
blinded manner with the use of a standardized The primary efficacy outcome was unfavorable
approach (see the protocol). Tuberculosis status status by 72 weeks. Unfavorable status was de-
at enrollment (confirmed, unconfirmed, or un- fined as a composite of tuberculosis events
likely) was adjudicated by an independent clini- (treatment failure, including treatment extension
cal expert committee on the basis of all available beyond the replacement of missed doses, anti-
clinical, radiologic, and laboratory data.13,14 An tuberculosis-treatment drug change or restart
end-point review committee, whose members due to suspected treatment failure, and tubercu-
were unaware of the treatment assignments, re- losis recurrence as adjudicated by the end-point
viewed clinical events suggestive of treatment review committee), loss to follow-up during
failure or tuberculosis recurrence and all deaths. treatment, or death from any cause. The primary
Clinical and laboratory adverse events of grade 3 safety outcome was an adverse event of grade 3
or higher were defined with the use of the Divi- or higher during treatment and up to 30 days
sion of AIDS Table for Grading the Severity of after treatment. The key secondary efficacy out-
Adult and Pediatric Adverse Events.15 Additional come was unfavorable status at 72 weeks in
notable events to be reported included suspected participants who were adjudicated by the end-
bacterial infection leading to hospitalization, point review committee as having tuberculosis
ocular toxic effects, and pregnancy. at baseline. Other secondary outcomes were
Adherence was assessed by means of pill death; adverse drug events that were considered
counts at each visit during treatment and by the by the site investigators to be possibly, probably,
administration of adherence questionnaires at or definitely related to a trial drug (adverse drug
the end of the intensive phase (first 8 weeks of reactions); bacterial infection leading to hospi-
treatment) and at the end of treatment. Treat- talization; adherence to the treatment regimen;
ment extensions due to excessive missed doses and acceptability of treatment as determined by
were reconciled against pill-count data. the caregiver (or by the child, when appropriate).
Randomization and Treatment Statistical Analysis

Eligible children were randomly assigned in a We determined the statistical power of the trial
1:1 ratio to receive 4 months or 6 months of on the basis of a key subgroup analysis involving
antituberculosis treatment. Randomization was children who were found on independent adjudi-
conducted with the use of minimization (with a cation to have tuberculosis at enrollment (80%
random element) according to trial center, age of the intention-to-treat population). Assuming
(<3 years or ≥3 years), HIV status, and ethambu- a 10% loss to follow-up (after treatment), an
tol use. unfavorable status by 72 weeks (primary efficacy
All the participants initially received 8 weeks outcome) in 8% of the participants in the control
of standard therapy with isoniazid, rifampin, group,17,18 and a noninferiority margin of 6 per-

centage points, we calculated that the enrollment R e sult s

of 1200 children would provide the trial with
90% power to detect noninferiority at a two- Participants
sided significance level of 5% (see the statistical Of the 1461 children who were screened, 1204
analysis plan, available with the protocol). underwent randomization between July 2016 and
The primary modified intention-to-treat pop- July 2018. A total of 376 participants were in
ulation included all the children who had under- Uganda, 364 in Zambia, 315 in South Africa, and
gone randomization except those who did not 149 in India. The principal reasons for ineligibil-
complete 4 months of treatment (when the treat- ity were smear-positive respiratory samples and
ment in the two groups was the same) or had a the presence of severe tuberculosis on chest ra-
late exclusion (on the basis of data collected diography (Fig. 1).
before randomization) and those who were clini- A total of 59 children (30 in the 4-month
cally well after the completion of treatment but group and 29 in the 6-month group) were ex-
were subsequently lost to follow-up. For the per- cluded from the modified intention-to-treat
protocol population, an additional exclusion analysis. A total of 36 participants (18 in the
criterion was the receipt of trial medications at 4-month group and 18 in the 6-month group)
less than 80% of the daily doses within 120% of did not complete 4 months of treatment (i.e.,
the assigned treatment duration (prespecified had a protocol-defined unfavorable status before
in the statistical analysis plan). month 4), and 14 participants (8 in the 4-month
The primary efficacy analysis was based on group and 6 in the 6-month group) were lost to
the absolute difference between the 4-month follow-up after successfully completing treat-
and 6-month strategies in the percentages of ment. The most frequent reason for further ex-
participants with an unfavorable status in the clusion from the per-protocol analysis was non-
modified intention-to-treat population, with ad- adherence to the assigned treatment strategy (in
justment for minimization factors with Cochran– 9 participants in the 4-month group and 15 in
Mantel–Haenszel weights. Time-to-event analyses the 6-month group) (Fig. 1).
of unfavorable status and death were conducted The demographic and clinical characteristics
with the use of log-rank tests and Cox propor- of the participating children were similar in the
tional-hazards models. Analyses were performed two groups (Table 1). The median age of the
with the use of Stata software, version 15.1 or participants was 3.5 years (range, 2 months to
later (StataCorp), and SAS software, version 10.1 15 years). A total of 52% of the participants were
(SAS Institute). An independent data monitoring male, 88% were African, and 12% were Indian;
committee reviewed data according to treatment 11% of the participants had HIV infection. A
group four times during the trial. total of 67% of the participants had respiratory
We performed economic analyses to estimate tuberculosis, 3% had peripheral lymph-node tu-
costs and health outcomes in terms of life-years berculosis, and 29% had mixed tuberculosis
and quality-adjusted life-years during the 72-week (respiratory and peripheral lymph-node disease).
trial. In these analyses, we used data on re- A total of 14% of the participants had tuberculo-
sources that had been used in the trial and the sis that was bacteriologically confirmed as positive
unit costs obtained from published sources for for Mycobacterium tuberculosis on culture or Xpert
each country. Costs were estimated from a health- testing. Xpert semiquantitative results showed
sector perspective to capture a range of health that all positive Xpert values were low or very
care resource use, including hospitalization, med- low (Table S1 in the Supplementary Appendix).
ication, and testing costs. We estimated quality-
adjusted life-years by combining health-related Retention and Treatment Adherence
quality-of-life scores, which were estimated with Retention, as assessed on the basis of atten-
the use of the European Quality of Life–5 Dimen- dance at the final trial visit at week 72, was 95%
sions questionnaire, and survival. Costs and out- among those expected (excluding formal with-
comes were discounted at 3% per annum, in line drawals [8 participants] and deaths [31]). Reten-
with international recommendations12 (see Sec- tion was similar in the two groups (Table S2).
tion S7 in the Supplementary Appendix, available Adherence to the assigned treatment duration
at NEJM.org). was similar in the two groups, with 94% of the

1461 Patients were assessed for eligibility
257 Had screening failure and did not

undergo randomization
144 Did not have minimal TB
31 Had positive smear and started
TB treatment in the National TB
Program
22 Did not have informed consent
provided by caregiver
21 Were too sick to participate
17 Were not willing or able to attend
follow-up visits
7 Did not return
6 Had MDR-TB
2 Had drug-resistant TB
7 Had other reason
1 Had weight issue
6 Had unknown reason
1204 Underwent randomization
602 Were assigned to 4-mo regimen and were 602 Were assigned to 6-mo regimen and were
included in the intention-to-treat population included in the intention-to-treat population
30 Were excluded from the 29 Were excluded from the

modified intention-to- modified intention-to-
treat population treat population
4 Had late screening 5 Had late screening
failure failure
18 Did not reach wk 16 18 Did not reach wk 16
5 Died from any cause 6 Died from any cause
3 Had TB progression 2 Had TB progression
4 Were withdrawn 4 Were withdrawn
3 Were lost to follow-up 4 Were lost to follow-up
3 Had treatment change 2 Had treatment change
in intensive phase in intensive phase
8 Were clinically well on 6 Were clinically well on
completing treatment completing treatment
but were lost to but were lost to
follow-up follow-up
572 Were included in the modified intention- 573 Were included in the modified intention-
to-treat population to-treat population
9 Were excluded from the 15 Were excluded from the

per-protocol population per-protocol population
owing to not adhering owing to not adhering
to trial treatment to trial treatment
563 Were included in the per-protocol 558 Were included in the per-protocol
population population
Figure 1. Randomization and Treatment of the Patients.

All the eligible participants with tuberculosis (TB) initially received 8 weeks of standard therapy with isoniazid, rifam
pin, and pyrazinamide (fixed-dose combination formulation), with or without ethambutol according to local guide-
lines (intensive phase). This treatment was followed by standard therapy with isoniazid and rifampin (continuation
phase) in a fixed-dose combination for either 8 weeks in the 4-month group (intervention) or 16 weeks in the 6-month
group (control). MDR-TB denotes multidrug-resistant TB.

Table 1. Demographic and Clinical Characteristics of the Participants at Baseline.*
4-Month Treatment 6-Month Treatment Total

Characteristic (N = 602) (N = 602) (N = 1204)
Age
Median (interquartile range) — yr 3.4 (1.5 to 6.9) 3.5 (1.5 to 7.1) 3.5 (1.5 to 7.0)
Range 2 mo to 15 yr 2 mo to 15 yr 2 mo to 15 yr
Female sex — no. (%) 297 (49) 286 (48) 583 (48)
Site country — no. (%)
Uganda 188 (31) 188 (31) 376 (31)
Zambia 183 (30) 181 (30) 364 (30)
South Africa 156 (26) 159 (26) 315 (26)
India 75 (12) 74 (12) 149 (12)
HIV-positive status — no. (%) 65 (11) 62 (10) 127 (11)
WHO weight band — no. (%)
3–3.9 kg 0 3 (<1) 3 (<1)
4–7.9 kg 86 (14) 92 (15) 178 (15)
8–11.9 kg 162 (27) 152 (25) 314 (26)
12–15.9 kg 126 (21) 116 (19) 242 (20)
16–24.9 kg 142 (24) 153 (25) 295 (25)
≥25 kg 86 (14) 86 (14) 172 (14)
Clinical presentation — no. (%)
Respiratory tuberculosis 398 (66) 406 (67) 804 (67)
Mixed respiratory and peripheral lymph- 182 (30) 171 (28) 353 (29)
node tuberculosis
Peripheral lymph-node tuberculosis 19 (3) 21 (3) 40 (3)
Other† 3 (<1) 4 (1) 7 (1)
M. tuberculosis culture and Xpert MTB/RIF test-
ing results — no. (%)‡
All positive results 85 (14) 80 (13) 165 (14)
Tuberculosis culture–positive only 40 (7) 40 (7) 80 (7)
Xpert MTB/RIF–positive only 14 (2) 5 (1) 19 (2)
Tuberculosis culture–positive and Xpert 31 (5) 35 (6) 66 (5)
MTB/RIF–positive
* HIV denotes human immunodeficiency virus, and WHO World Health Organization.
† These participants did not have a cough for more than 2 weeks or one or more peripheral lymph nodes suggestive of
tuberculosis.
‡ Microbiologic confirmation was from respiratory samples (gastric aspirate or washing or induced or expectorated spu-
tum) and fine-needle aspiration of enlarged lymph nodes and was defined as positive for Mycobacterium tuberculosis on
culture or the Xpert MTB/RIF assay. Cultures were assessed with Löwenstein–Jensen solid-culture medium or a myco-
bacteria growth indicator tube with liquid culture system.
children taking at least 80% of daily doses within Primary Outcome

120% of the assigned days (Fig. S1). Radiographs In the modified intention-to-treat analysis, 21
of the chest with discordant interpretations at events occurring before month 4 were excluded
the primary (baseline) reading were reviewed by (11 deaths, 5 events of tuberculosis progression,
a third expert for 435 of 1174 participants (37%). and 5 treatment extensions or drug changes)
Chest radiographs that had been obtained at (Fig. 1). In the primary modified intention-to-
baseline were missing for 30 participants. treat analysis, an unfavorable status was observed

No. of 4-Month 6-Month

Patients Treatment Treatment Risk Difference (95% CI)
no. of participants with
event/total no. (%) percentage points
Primary outcome
Modified intention-to-treat population 1145 16/572 (3) 18/573 (3) −0.3 (−2.3 to 1.6)
Per-protocol population 1121 14/563 (2) 17/558 (3) −0.6 (−2.5 to 1.4)
Intention-to-treat population 1204 44/602 (7) 44/602 (7) 0 (−2.9 to 2.9)
Key secondary outcome
Modified intention-to-treat population 910 10/450 (2) 13/460 (3) −0.6 (−2.6 to 1.4)
Per-protocol population 895 8/445 (2) 13/450 (3) −1.1 (−3.1 to 0.9)
−8 −4 0 4 8
4-Month 6-Month
Treatment Treatment
Better Better
Figure 2. Unadjusted Analysis of the Primary Efficacy and Key Secondary Outcomes in the Trial Populations.
The primary efficacy outcome was unfavorable status by 72 weeks, which was defined as a composite of treatment
failure (treatment extension, change, or restart or tuberculosis recurrence), loss to follow-up during treatment, or
death, with the exclusion of all the participants who had undergone randomization but did not complete 4 months
of treatment (modified intention-to-treat population). The per-protocol population included all the participants in
the modified intention-to-treat population except those who had not adhered to the trial regimen. The intention-to-
treat population included all the participants who had undergone randomization. Differences have been carried
to one decimal place because of the small values. The prespecified margin for noninferiority in the primary efficacy
analysis was 6 percentage points (dashed line). The key secondary analysis was unfavorable status at 72 weeks as
assessed among the 958 participants who had been independently adjudicated as having tuberculosis at baseline.
in 16 participants (3%) in the 4-month group Key Secondary Analysis

and in 18 participants (3%) in the 6-month The key secondary analysis included 958 partici-
group (unadjusted difference, −0.3 percentage pants who had been independently adjudicated
points; 95% confidence interval [CI], −2.3 to 1.6; to have tuberculosis at baseline (80% of the en-
adjusted difference, −0.4 percentage points; 95% rolled population); 910 of these participants
CI, −2.2 to 1.5) (Fig. 2 and Table 2). were included in the modified intention-to-treat
Results in the intention-to-treat and per-pro- population. Of these 910 participants, 440 of
tocol populations were similar to those of the 450 (98%) in the 4-month group and 447 of 460
primary analysis (Fig. 2 and Table S9). In time- (97%) in the 6-month group had a favorable
to-event analyses, there were no significant be- outcome. The adjusted absolute difference in the
tween-group differences in the risks of an unfa- risk of an unfavorable status in the 4-month
vorable status (hazard ratio, 0.88; 95% CI, 0.45 group as compared with the 6-month group was
to 1.74) or death (hazard ratio, 0.63; 95% CI, −0.6 percentage points (95% CI, −2.6 to 1.4). In
0.31 to 1.30) (Figs. S2 and S3). the per-protocol analysis, the difference was
The most common reasons for an unfavor- −1.1 percentage points (95% CI, −3.1 to 0.9)
able status (after month 4) were death from any (Fig. 2). Results for the modified intention-to-
cause (in 7 participants in the 4-month group treat efficacy outcome according to prespecified
and 12 in the 6-month group) and treatment subgroup analyses (HIV status, region, sex, age,
failure (in 9 and 5 participants, respectively). weight band, tuberculosis type, bacteriologic
Among participants with treatment failure, 2 confirmation, and ethambutol given at baseline)
(both in the 4-month group) had a treatment were all consistent with the primary result (Fig.
extension, 2 (1 in each group) stopped treatment S4 and Table S8).
and had their treatment course restarted from
the beginning during the first 8 weeks, and 10 Adverse Events
(6 in the 4-month group and 4 in the 6-month A total of 115 adverse events of grade 3 or
group) had recurrence of tuberculosis (Table 2). higher occurred during treatment or up to 30

Table 2. Primary Efficacy Analysis (Modified Intention-to-Treat Population).*
4-Month Treatment 6-Month Treatment

Outcome (N = 572) (N = 573) Difference (95% CI)
Adjusted Unadjusted
Analysis† Analysis
percentage points
Unfavorable status — no. (%) 16 (3) 18 (3) −0.4 −0.3
(−2.2 to 1.5) (−2.3 to 1.6)
Death from any cause after 4 mo 7 (1) 12 (2)
Loss to follow-up after 4 mo but 0‡ 1 (<1)
during treatment period
Treatment failure
Tuberculosis recurrence 6 (1) 4 (1)
Extension of treatment 2 (<1) 0
Restart of treatment§ 1 (<1) 1 (<1)
Favorable status — no. (%) 556 (97) 555 (97)
* The primary efficacy outcome was unfavorable status by 72 weeks. Unfavorable status was defined as a tuberculosis
event (treatment failure, including treatment extension beyond the replacement of missed doses, drug changes or re-
starts in antituberculosis treatment due to suspected treatment failure, and tuberculosis recurrence as adjudicated by
the end-point review committee, whose members were unaware of the treatment assignments), loss to follow-up dur-
ing treatment, or death from any cause. Favorable status was defined as the completion of treatment and a status of
being clinically well, without having had retreatment or another unfavorable outcome. The primary modified intention-
to-treat population included all the participants who had undergone randomization except those who did not complete
4 months of treatment or who had a late exclusion (on the basis of data collated before randomization) and those who
were clinically well after the completion of treatment but were subsequently lost to follow-up.
† The analysis was adjusted with Cochran–Mantel–Haenszel weighting for trial center, participant age (<3 years or ≥3 years),
HIV status, and ethambutol use.
‡ None of the participants extended their treatment beyond 4 months and were subsequently lost to follow-up.
§ Two participants stopped treatment and had their treatment course restarted from the beginning during the first 8 weeks.
days after treatment in 95 participants (8% of all pants (5%), with 45 participants (26 in the
the children enrolled in the trial; 49 events in 47 4-month group and 19 in the 6-month group)
participants in the 4-month group and 66 events being hospitalized after month 4 (Table 3).
in 48 participants in the 6-month group). Of A total of 15 adverse drug reactions of grade 3
these 115 adverse events, the most common or 4 were considered by the investigators to be
were pneumonia or other chest infection (29 possibly, probably, or definitely related to trial
events [25%]) or liver-related events (11 [10%]); drugs. These adverse reactions included 11 hepatic
the incidences were similar in the two groups events. All the adverse reactions except 2 oc-
(Tables 3 and S3). curred during the first 8 weeks of therapy (Table
A total of 192 serious adverse events occurred S4). Two participants permanently discontinued
in 150 participants (12%) in the trial, including therapy after treatment interruption for an ad-
31 deaths (12 in the 4-month group and 19 in verse reaction. None of the adverse reactions led
the 6-month group). Twenty deaths (7 in the to death.
4-month group and 13 in the 6-month group)
occurred after month 4, and 13 deaths (5 in the Cost-Effectiveness Analysis
4-month group and 8 in the 6-month group) A cost-effectiveness analysis showed that at 72
were considered by the end-point review com- weeks, participants who had been treated for
mittee to be related to tuberculosis. Of the 31 4 months had similar health outcomes as those
deaths, 25 were in children younger than 2 years who had been treated for 6 months but with
of age (Table S5). Hospitalization for a respira- lower health care costs. A regression analysis
tory bacterial infection occurred in 66 partici- controlling for chance differences in demographic

Table 3. Primary Safety Outcome, Serious Adverse Events, Deaths, Adverse Drug Reactions, and Suspected Bacterial
Infections Leading to Hospitalization.*
4-Month Treatment 6-Month Treatment Total

Event (N = 602) (N = 602) (N = 1204)
Primary safety outcome — no. of events 49 66 115
No. of participants with ≥1 event (%)† 47 (8) 48 (8) 95 (8)
At ≤4 mo
No. of adverse events of grade ≥3 35 52 87
No. of participants with ≥1 event (%) 33 (5) 40 (7) 73 (6)
At >4 mo
No. of adverse events of grade ≥3 14 14 28
Serious adverse event — no. of events 88 104 192
No. of participants with ≥1 serious adverse event (%)† 75 (12) 75 (12) 150 (12)
At ≤4 mo
No. of serious adverse events 35 50 85
No. of participants with ≥1 serious adverse 33 (5) 40 (7) 73 (6)
event (%)
At >4 mo
No. of serious adverse events 53 54 107
No. of participants with ≥1 serious adverse 47 (8) 44 (7) 91 (8)
event (%)
Death — no. (%) 12 (2) 19 (3) 31 (3)
At ≤4 mo
No. of deaths (%) 5 (1) 6 (1) 11 (1)
No. of deaths considered to be related to 3 (<1) 2 (<1) 5 (<1)
tuberculosis (%)
At >4 mo
No. of deaths (%) 7 (1) 13 (2) 20 (2)
No. of deaths considered to be related to 2 (<1) 6 (1) 8 (1)
tuberculosis (%)
Adverse drug reaction — no. of participants (%)‡ 6 (1) 11 (2) 17 (1)
Bacterial infection leading to hospitalization 40 40 80
— no. of events
* The Division of AIDS Table for Grading the Severity of Adult and Pediatric Adverse Events was used for grading the
severity of adverse events. The primary safety outcome was an adverse event of grade 3 or higher that occurred during
treatment and up to 30 days after the last dose of a trial drug. Serious adverse events were defined with the use of
International Council for Harmonisation and Good Clinical Practice definitions as an adverse event that resulted in death,
was life-threatening, led to hospitalization or prolonged existing hospitalization, resulted in persistent or clinically
significant disability or incapacity, consisted of a congenital anomaly or birth defect, or was considered to be another
important medical condition.
† Participants may have had events both before and after the 4-month breakdown.
‡ Adverse drug reactions during treatment and within 30 days after completion of treatment were assessed by the site
investigator as being possibly, probably, or definitely related to the trial drugs. Further information on the adverse drug
reactions, including event types and information on treatment discontinuations, is provided in Table S4.
characteristics and symptom severity estimated that CI, −0.009 to 0.014) and that health care costs (as-
the number of quality-adjusted life-years was simi- sessed in 2019) were $17.34 (95% CI, 3.77 to 30.91)
lar in the two groups (mean difference, 0.003; 95% lower per child in the 4-month group (Table S7).

Discussion matic approach by following routine screening

procedures and reviewing chest radiographs to
The SHINE trial evaluated the duration of anti- assess the severity of respiratory tuberculosis.
tuberculosis treatment in children with non Despite perceived difficulties of obtaining respi-
severe, drug-susceptible tuberculosis who were ratory samples in children, such challenges were
living in countries with a high burden of tuber- overcome with appropriate training, and sam-
culosis, where nearly 90% of cases of tubercu- ples were successfully obtained for tuberculosis
losis in children occur.19 The trial showed the testing in all 1204 children who underwent
noninferiority of 4 months as compared with the randomization. The trial included children with
standard 6 months of treatment, with the upper HIV infection as well as those without HIV in-
boundary of the 95% confidence interval being fection, with consistent results.
below the prespecified margin of 6 percentage Most children with tuberculosis have pauci-
points. Consistency of results across all the bacillary and nonsevere disease with low rates of
analyses, including a key secondary analysis in a microbiologic tuberculosis confirmation in rou-
subgroup of children who were adjudicated to tine care. To ensure the applicability of our re-
have tuberculosis at baseline, was observed. Par- sults to clinical practice and the spectrum of
ticipants had a good response to treatment with disease that is most prevalent in children, we did
few adverse drug reactions, most of which oc- not limit the trial to bacteriologically confirmed
curred before 4 months, during the period when tuberculosis. We adapted the pediatric consen-
the two trial groups had the same treatment sus algorithm for diagnosis of intrathoracic tuber-
regimen. culosis13 to both intrathoracic tuberculosis and
Shortening treatment for drug-susceptible peripheral lymph-node tuberculosis and used
tuberculosis is a key goal for both adults and independent expert review and central reading
children. Early trials showed that it was possible of chest radiographs, with blinding to the treat-
to shorten the treatment duration in adults with ment assignments, to ensure objective categori-
culture-negative disease.20-22 A meta-analysis of zation of tuberculosis status. The end-point re-
treatment-duration trials involving adults showed view committee, whose members were unaware
that 4-month drug regimens were efficacious in of the randomized group assignments, reviewed
adults with paucibacillary tuberculosis who had tuberculosis outcomes to minimize the effect of
disease with a sputum-smear grade of less than treatment assignment on adjudication.
2+ (<1 acid-fast bacillus per field) or noncavity Our trial had several strengths. It was well-
disease.23 Recently, the Tuberculosis Trials Con- powered, and we observed 94% adherence (to
sortium Study 31/AIDS Clinical Trials Group the receipt of ≥80% of the doses) in the assigned
A5349 trial showed the noninferiority of a groups and 95% retention of trial participants,
4-month rifapentine-based regimen containing findings that increase confidence in the results.
moxifloxacin, as compared with the 6-month We assumed in the sample-size calculation that
standard regimen, for all forms of drug-suscep- 8% of the participants in the 6-month group
tible tuberculosis (including cavity disease) in would have an unfavorable status, and we ob-
adults and adolescents.24 Challenges remain in served this result in 7% of the participants over-
terms of the availability of child-friendly formu- all in the trial. However, events in 3% of the
lations of rifapentine and moxifloxacin, data on participants occurred after month 4, when the
doses in children, and cost. However, our results trial groups were receiving different durations
show that a new regimen with new drugs is not of the treatment, which we had not anticipated
necessary for the shortening of treatment in the when we designed the trial.
majority of children with drug-susceptible tuber- The trial also has several limitations. One
culosis, since such treatment shortening can be limitation is that this trial was open-label, which
accomplished with the affordable, child-friendly, had the potential to result in more frequent
fixed-dose combinations that are already avail- treatment extensions in the 4-month group, con-
able.25 tributing to more unfavorable outcomes in this
This trial showed the feasibility of identifying group. Despite this possible disadvantage in the
children with nonsevere disease. We used a prag- 4-month group, the results showed consistently

that the 4-month regimen was as good as the In this trial, we found that 4 months of anti-
6-month regimen. Another limitation relates to tuberculosis treatment was noninferior to 6 months
the generalizability of our results to settings of therapy in children with drug-susceptible, non-
where chest radiographs are not available for severe, smear-negative tuberculosis. The results
characterizing nonsevere tuberculosis. suggest that a stratified medicine approach as
Our inclusion criteria for the trial required an alternative to the one-size-fits-all strategy of
smear microscopy to be undertaken to rule out treatment for presumptive drug-susceptible tu-
more severe forms of respiratory tuberculosis. berculosis could be implemented in children
With the current rollout of rapid molecular diag- with nonsevere tuberculosis.
nostic tests replacing smear microscopy,26 this
may pose a challenge to implementation on the Supported by a grant (MR/L004445/1) from the U.K. Medical
basis of the trial results. However, smear-grade Research Council (MRC) and the Department for International
Development (DFID) Wellcome NIHR Joint Global Health Trials,
and Xpert semiquantitative results have been by a U.K. Research and Innovation Covid-19 Grant Extension
shown to be correlated.27 In our trial, most Xpert Allocation Award, by a U.K. Research and Innovation MRC grant
results from respiratory samples were negative (MC_UU_00004/04, to Drs. Turkova and Crook), and by a Clini-
cian Scientist Fellowship (to Dr. Seddon), jointly funded by the
and the few positive Xpert samples had low or MRC and the Department for International Development under
very low semiquantitative results, which suggest an MRC–DFID Concordat agreement (MR/R007942/1). TB Alli-
that the trial findings can be extrapolated to ance provided support for drug purchase.
Disclosure forms provided by the authors are available with
settings where Xpert is replacing smear testing the full text of this article at NEJM.org.
and that children with negative, low, or very low A data sharing statement provided by the authors is available
positive values on Xpert testing can be catego- with the full text of this article at NEJM.org.
We thank all the children who participated in the SHINE
rized as having nonsevere tuberculosis. It will be trial and their caregivers; the nursing, pharmacy, and laboratory
useful in future implementation studies to ex- staff; all those who advised, volunteered, or otherwise support-
plore treatment shortening in all children who ed community engagement at the trial sites; and the members
of the trial steering committee, the end-point review committee,
are treated for nonsevere, drug-susceptible tuber- and the independent data monitoring committee for their con-
culosis, regardless of smear or Xpert results. tributions, including oversight of the safety of the trial.
Appendix
The authors’ full names and academic degrees are as follows: Anna Turkova, M.R.C.P.C.H., Genevieve H. Wills, M.Sc., Eric Wobudeya,
M.Med., Chishala Chabala, M.Med., Megan Palmer, M.B., Ch.B., M.Med., Aarti Kinikar, M.D., Syed Hissar, M.D., M.P.H., Louise Choo,
Ph.D., Philippa Musoke, Ph.D., Veronica Mulenga, M.Med., M.Sc., Vidya Mave, M.D., M.P.H.&T.M., Bency Joseph, M.B., B.S., M.P.H.,
Kristen LeBeau, M.Sc., Margaret J. Thomason, Ph.D., Robert B. Mboizi, M.Sc., Monica Kapasa, M.Med., Marieke M. van der Zalm,
Ph.D., Priyanka Raichur, M.B., B.S., Perumal K. Bhavani, M.B., B.S., M.P.H., Helen McIlleron, Ph.D., Anne‑Marie Demers, M.D., Rob
Aarnoutse, Ph.D., James Love‑Koh, Ph.D., James A. Seddon, Ph.D., Steven B. Welch, F.R.C.P.H., Stephen M. Graham, Ph.D., Anneke C.
Hesseling, Ph.D., Diana M. Gibb, M.B., Ch.B., M.D., and Angela M. Crook, Ph.D., for the SHINE Trial Team
The authors’ affiliations are as follows: the Medical Research Council Clinical Trials Unit, University College London (A.T., G.H.W.,
L.C., K.L., M.J.T., D.M.G., A.M.C.), and the Department of Infectious Diseases, Imperial College London (J.A.S.), London, the Centre
for Health Economics, University of York, York (J.L.-K.), and the Department of Paediatrics, Birmingham Chest Clinic and Heartlands
Hospital, University Hospitals Birmingham, Birmingham (S.B.W.) — all in the United Kingdom; Makerere University–Johns Hopkins
University Research Collaboration, Kampala, Uganda (E.W., P.M., R.B.M.); University Teaching Hospital, Lusaka, Zambia (C.C., V.
Mulenga, M.K.); Desmond Tutu TB Centre, the Department of Paediatrics and Child Health, Stellenbosch University, Stellenbosch
(M.P., M.M.Z., A.-M.D., J.A.S., A.C.H.), and the Division of Clinical Pharmacology, University of Cape Town, Cape Town (H.M.) — both
in South Africa; B.J. Medical College, Pune (A.K., V. Mave, P.R.), and the National Institute for Research in Tuberculosis, Chennai (S.H.,
B.J., P.K.B.) — both in India; Radboud University Medical Center, Nijmegen, the Netherlands (R.A.); the Centre for International Child
Health, Department of Paediatrics, University of Melbourne, and Murdoch Children’s Research Institute, Royal Children’s Hospital —
both in Melbourne, VIC, Australia (S.M.G.); and the International Union against Tuberculosis and Lung Disease, Paris (S.M.G.).
References
1. WHO Global Tuberculosis Programme. a mathematical modelling study. Lancet trum of disease in children treated for
Global tuberculosis report 2021. Geneva: Glob Health 2017;5(9):e898-e906. tuberculosis in a highly endemic area. Int
World Health Organization, 2021 (https:// 3. Wiseman CA, Gie RP, Starke JR, et al. J Tuberc Lung Dis 2006;10:732-8.
www.who.i nt/publications/d igital/g lobal A proposed comprehensive classification 5. Loveday M, Sunkari B, Marais BJ,
-t uberculosis-report-2021). of tuberculosis disease severity in chil- Master I, Brust JC. Dilemma of managing
2. Dodd PJ, Yuen CM, Sismanidis C, Sed- dren. Pediatr Infect Dis J 2012;31:347-52. asymptomatic children referred with ‘cul-
don JA, Jenkins HE. The global burden 4. Marais BJ, Gie RP, Schaaf HS, Hessel- ture-confirmed’ drug-resistant tuberculo-
of tuberculosis mortality in children: ing AC, Enarson DA, Beyers N. The spec- sis. Arch Dis Child 2016;101:608-13.

6. Jaganath D, Mupere E. Childhood tu- in a therapeutic clinical trial: experience mens of chemotherapy for sputum-smear-
berculosis and malnutrition. J Infect Dis from SHINE. Proceedings and abstracts negative pulmonary tuberculosis: results
2012;206:1809-15. of the 51st Union World Conference on at 5 years. Am Rev Respir Dis 1989;139:
7. Martinez L, Cords O, Horsburgh CR, Lung Health, October 20–24, 2020. Paris: 871-6.
Andrews JR;Pediatric TB Contact Stud- the International Union against Tubercu- 22. Hong Kong Chest Service, Madras Tu-
ies Consortium. The risk of tuberculosis losis and Lung Disease, 2020. berculosis Research Centre, British Medi-
in children after close exposure: a sys- 15. Department of Health and Human cal Research Council. A controlled trial of
tematic review and individual-partici- Services. Division of AIDS (DAIDS) table 2-month, 3-month, and 12-month regi-
pant meta-analysis. Lancet 2020;395:973- for grading the severity of adult and pedi- mens of chemotherapy for sputum-smear-
84. atric adverse events. Version 2.0. Novem- negative pulmonary tuberculosis: results
8. Rapid advice: treatment of tuberculo- ber 2014 (https://rsc.niaid.nih.gov/sites/ at 60 months. Am Rev Respir Dis 1984;
sis in children. Geneva:World Health Or- default/f iles/d aids-ae-g rading-t able-v2 130:23-8.
ganization, 2010. -nov2014.pdf). 23. Imperial MZ, Nahid P, Phillips PPJ, et al.
9. Graham SM, Grzemska M, Gie RP. 16. Wademan DT, Busakwe L, Nicholson A patient-level pooled analysis of treat-
The background and rationale for a new TJ, et al. Acceptability of a first-line anti- ment-shortening regimens for drug-sus-
fixed-dose combination for first-line treat- tuberculosis formulation for children: ceptible pulmonary tuberculosis. Nat Med
ment of tuberculosis in children. Int J Tu- qualitative data from the SHINE trial. Int 2018;24:1708-15.
berc Lung Dis 2015;19:Suppl 1:3-8. J Tuberc Lung Dis 2019;23:1263-8. 24. Dorman SE, Nahid P, Kurbatova EV,
10. Guidance for national tuberculosis 17. Floyd K, Glaziou P, Sismanidis C, et al. et al. Four-month rifapentine regimens
programmes on the management of tu- Global tuberculosis report 2012. Geneva: with or without moxifloxacin for tubercu-
berculosis in children. 2nd ed. Geneva: World Health Organization, 2012. losis. N Engl J Med 2021;384:1705-18.
World Health Organization, 2014. 18. Kekitiinwa A, Cook A, Nathoo K, et al. 25. Global Alliance for TB Drug Devel-
11. Chabala C, Turkova A, Thomason MJ, Routine versus clinically driven laboratory opment. New milestone for children
et al. Shorter treatment for minimal monitoring and first-line antiretroviral with tuberculosis reached as one million
tuberculosis (TB) in children (SHINE): therapy strategies in African children with treatments of child-friendly medicines are
a study protocol for a randomised con- HIV (ARROW): a 5-year open-label ran- ordered. June 18, 2019 (https://www
trolled trial. Trials 2018;19:237. domised factorial trial. Lancet 2013;381: .t balliance.org/news/one-m illion-child
12. Wilkinson T, Sculpher MJ, Claxton K, 1391-403. -friendly-t uberculosis-medicines).
et al. The International Decision Support 19. WHO Global Tuberculosis Programme. 26. World Health Organization. WHO
Initiative Reference Case for economic Global tuberculosis report 2019. Geneva: consolidated guidelines on tuberculosis.
evaluation: an aid to thought. Value Health World Health Organization, 2019. Module 3: diagnosis — rapid diagnostics
2016;19:921-8. 20. Teo SK, Tan KK, Khoo TK. Four- for tuberculosis detection. July 7, 2021
13. Graham SM, Cuevas LE, Jean-Philippe month chemotherapy in the treatment of (https://www.who.int/publications/i/item/
P, et al. Clinical case definitions for clas- smear-negative pulmonary tuberculosis: 9789240029415).
sification of intrathoracic tuberculosis in results at 30 to 60 months. Ann Acad Med 27. Blakemore R, Nabeta P, Davidow AL,
children: an update. Clin Infect Dis 2015; Singap 2002;31:175-81. et al. A multisite assessment of the quan-
61:Suppl 3:S179-S187. 21. Hong Kong Chest Service, Madras Tu- titative capabilities of the Xpert MTB/RIF
14. Wills GH, Graham SM, Seddon JA, berculosis Research Centre, British Medi- assay. Am J Respir Crit Care Med 2011;
et al. A method for baseline adjudica- cal Research Council. A controlled trial of 184:1076-84.
tion of tuberculosis diagnosis in children 3-month, 4-month, and 6-month regi- Copyright © 2022 Massachusetts Medical Society.


Nejmoa 2104535

Uploaded by

Document Informationclick to expand document informationtb

Document Informationclick to expand document information

Copyright:

Available Formats

Nejmoa 2104535

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

Nejmoa 2104535

Uploaded by

Copyright:

Available Formats

new england

Shorter Treatment for Nonsevere Tuberculosis in African

912 n engl j med 386;10 nejm.org March 10, 2022

The New England Journal of Medicine

Randomization and Treatment Statistical Analysis

n engl j med 386;10 nejm.org March 10, 2022 913

centage points, we calculated that the enrollment R e sult s

914 n engl j med 386;10 nejm.org March 10, 2022

The New England Journal of Medicine

1461 Patients were assessed for eligibility

257 Had screening failure and did not

1204 Underwent randomization

30 Were excluded from the 29 Were excluded from the

9 Were excluded from the 15 Were excluded from the

Figure 1. Randomization and Treatment of the Patients.

n engl j med 386;10 nejm.org March 10, 2022 915

Table 1. Demographic and Clinical Characteristics of the Participants at Baseline.*

4-Month Treatment 6-Month Treatment Total

children taking at least 80% of daily doses within Primary Outcome

916 n engl j med 386;10 nejm.org March 10, 2022

The New England Journal of Medicine

No. of 4-Month 6-Month

in 16 participants (3%) in the 4-month group Key Secondary Analysis

n engl j med 386;10 nejm.org March 10, 2022 917

Table 2. Primary Efficacy Analysis (Modified Intention-to-Treat Population).*

4-Month Treatment 6-Month Treatment

918 n engl j med 386;10 nejm.org March 10, 2022

The New England Journal of Medicine

4-Month Treatment 6-Month Treatment Total

n engl j med 386;10 nejm.org March 10, 2022 919

Discussion matic approach by following routine screening

920 n engl j med 386;10 nejm.org March 10, 2022

The New England Journal of Medicine

n engl j med 386;10 nejm.org March 10, 2022 921

922 n engl j med 386;10 nejm.org March 10, 2022

The New England Journal of Medicine

You might also like