Chapter 10 Diagnosis of TB: state of the art Jonathan G. Peter*, Richard N. van Zyl-Smit*, Claudia M. Denkinger # and Madhukar Pai ",+ SUMMARY: Rapid, affordable and accurate tuberculosis (TB) diagnosis is key to effective patient management and global TB control. Effective clinical screening and optimised sample acquisition methods remain the first steps in the diagnostic process. Smear microscopy, despite optimisation, remains widely used even though its sensitivity is poor. Mycobacterial liquid culture is accurate but poorly accessible. The use of novel molecular tools, such as Xpert1 MTB/RIF (Cepheid, Sunnyvale, CA, USA) or GenoType1 MTBDRplus (Hain Lifescience GmbH, Nehren, Germany) assays, which offer superior diagnostic accuracy and decreased time-to-diagnosis for drug-sensitive and/or -resistant TB, is increasing following World Health Organization (WHO) endorsement and, in some countries, na- tional roll-out is underway. In contrast, both serology (antibody- detection tests) and interferon-c release assays (IGRAs) have been found to offer little diagnostic utility for active TB diagnosis and have been discouraged by WHO. IGRAs and the tuberculin skin test (TST) remain important tools for latent TB infection (LTBI) diagnosis. Other novel, simple technologies, such as the point-of-care (POC) urine lipoarabinomannan strip test and the visually read loop isothermal amplification PCR nucleic acid amplification technique (NAAT), although of uncertain and restricted clinical utility, highlight the progression toward an inexpensive, instrument-free, laboratory-free POC diagnostic technology for TB in the future. KEYWORDS: Diagnosis, tuberculosis *Lung Infection and Immunity Unit, Division of Pulmonology and UCT Lung Institute, Dept of Medicine, University of Cape Town, Rondebosch, South Africa. # Division of Infectious Diseases, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA. " Dept of Epidemiology and Biostatistics, McGill University, and + Respiratory Epidemiology and Clinical Research Unit, Montreal Chest Institute, Montreal, Canada. Correspondence: M. Pai, McGill University, Dept of Epidemiology and Biostatistics, 1020 Pine Ave West, Montreal, QC H3A 1A2, Canada. Email: [email protected]Eur Respir Monogr 2012; 58: 124–143. Copyright ERS 2012. DOI: 10.1183/1025448x.10023211 Print ISBN: 978-1-84984-027-9 Online ISBN: 978-1-84984-028-6 Print ISSN: 1025-448x Online ISSN: 2075-6674 A ffordable, accurate and rapid diagnosis followed by effective therapy is the cornerstone of tuberculosis (TB) control. TB is curable in over 95% of cases but the same diagnostic standard remains elusive to many who need it most. Multiple social, host and pathogen factors, as depicted in figure 1, intersect to produce and worsen TB diagnostic delay or failure. Fortunately, thanks to renewed global awareness, financial investment and international collaboration, several new diagnostic options have been developed. Old tools continue to be optimised and several new tools are now commercially available and being scaled up by national TB programmes. 124 TB DIAGNOSIS
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Chapter 10
Diagnosis of TB: state ofthe artJonathan G. Peter*, Richard N. van Zyl-Smit*,Claudia M. Denkinger# and Madhukar Pai",+
SUMMARY: Rapid, affordable and accurate tuberculosis (TB)diagnosis is key to effective patient management and globalTB control. Effective clinical screening and optimised sampleacquisition methods remain the first steps in the diagnosticprocess. Smear microscopy, despite optimisation, remains widelyused even though its sensitivity is poor. Mycobacterial liquidculture is accurate but poorly accessible. The use of novelmolecular tools, such as Xpert1 MTB/RIF (Cepheid, Sunnyvale,CA, USA) or GenoType1 MTBDRplus (Hain Lifescience GmbH,Nehren, Germany) assays, which offer superior diagnosticaccuracy and decreased time-to-diagnosis for drug-sensitiveand/or -resistant TB, is increasing following World HealthOrganization (WHO) endorsement and, in some countries, na-tional roll-out is underway. In contrast, both serology (antibody-detection tests) and interferon-c release assays (IGRAs) havebeen found to offer little diagnostic utility for active TB diagnosisand have been discouraged by WHO. IGRAs and the tuberculinskin test (TST) remain important tools for latent TB infection(LTBI) diagnosis. Other novel, simple technologies, such as thepoint-of-care (POC) urine lipoarabinomannan strip test and thevisually read loop isothermal amplification PCR nucleic acidamplification technique (NAAT), although of uncertain andrestricted clinical utility, highlight the progression toward aninexpensive, instrument-free, laboratory-free POC diagnostictechnology for TB in the future.
KEYWORDS: Diagnosis, tuberculosis
*Lung Infection and Immunity Unit,Division of Pulmonology and UCTLung Institute, Dept of Medicine,University of Cape Town,Rondebosch, South Africa.#Division of Infectious Diseases, BethIsrael Deaconess Medical Center,Harvard Medical School, Boston,MA, USA."Dept of Epidemiology andBiostatistics, McGill University, and+Respiratory Epidemiology andClinical Research Unit, MontrealChest Institute, Montreal, Canada.
Correspondence: M. Pai, McGillUniversity, Dept of Epidemiology andBiostatistics, 1020 Pine Ave West,Montreal, QC H3A 1A2, Canada.Email: [email protected]
Affordable, accurate and rapid diagnosis followed by effective therapy is the cornerstone oftuberculosis (TB) control. TB is curable in over 95% of cases but the same diagnostic standard
remains elusive to many who need it most. Multiple social, host and pathogen factors, as depictedin figure 1, intersect to produce and worsen TB diagnostic delay or failure. Fortunately, thanks torenewed global awareness, financial investment and international collaboration, several newdiagnostic options have been developed. Old tools continue to be optimised and several new toolsare now commercially available and being scaled up by national TB programmes.
12
4T
BD
IAG
NO
SIS
This chapter will discuss the optimisation of old tools, the development and integration of newdiagnostic technologies for active TB (including smear-negative TB), and drug-susceptibilitytesting (DST), extrapulmonary TB (EPTB) and latent TB infection (LTBI). It also aims tohighlight the progress towards novel point-of-care (POC) TB diagnostics. In particular, diagnostictest performance characteristics and optimal settings for use will be discussed, finally emphasisingresearch ‘‘gaps’’ and the ongoing unmet diagnostic needs.
Diagnosis of active pulmonary TB and DST
Table 1 describes current, commercially available TB diagnostics for active TB and DST stratifiedby the test’s ability to provide diagnosis and DST alone or combined. In addition, to bettercontextualise available old and new tests, figure 2 visually contrasts the test performance andtime-to-result of commercially available diagnostics for active TB and DST.
Improving the old
Clinical case definitions and symptom screening
Clinical screening, diagnosis and case definitions continue to guide treatment decisions for activeTB, and form part of the composite diagnostic reference standards. In recent years, researchers havebegun to objectively evaluate the performance of some of the available clinical and radiology-baseddiagnostic guidelines, reach consensus on clinical case definitions for certain TB forms withsuboptimal reference standards (e.g. TB meningitis) and use screening to strengthen World HealthOrganization (WHO) guidelines.
HIV
Pathogen
Children
Limited healthcare accessAsymptomatic, EPTB and atypical disease
Poor laboratory services
DR strains
Slow growth
Unique clinical presentation
Difficult sample aquisition
Subclinical latent disease
Variable humoral and cellular immune responses
Performance of traditional diagnostics
Poverty
Delayed/misdiagnosisTB diagnosis
Host
Figure 1. The multiple inter-related factors driving misdiagnosis or delayed diagnosis of tuberculosis (TB). Thediameters of the larger circles indicate the relative impacts on delayed/misdiagnosis of TB and the overlappingcircles indicate relatedness. EPTB: extrapulmonary tuberculosis; DR: drug-resistant.
12
5J.
G.
PE
TE
RE
TA
L.
Ta
ble
1.
Com
merc
ially
ava
ilab
led
iagnost
ics
for
activ
etu
berc
ulo
sis
(TB
)and
dru
g-s
usc
ep
tibility
test
ing
(DS
T)
Te
st
typ
eo
rp
latf
orm
De
sc
rip
tio
no
fte
st
Cu
rre
nt
va
lid
ate
dc
om
me
rcia
lve
rsio
ns
Lo
w/h
igh
se
nsit
ivit
y(e
xp
ec
ted
)L
ow
/hig
hsp
eci-
ficit
y(e
xp
ecte
d)
WH
Oe
nd
ors
em
en
t
Co
mm
en
ts
De
tec
tio
no
fa
cti
ve
TB
Flu
ore
scent
mic
rosc
op
yusi
ng
LE
D
Aura
min
eO
-sta
ined
smea
rre
adby
fluore
scen
tm
icro
scopy
usi
ng
LE
Dlig
ht
sourc
e
Prim
oS
tar
iLE
DTM
(Carl
Zeis
s,O
berk
ochen,
Germ
any)
Lum
inTM
(LW
Scie
ntif
ic,
Law
rencevi
lle,
GA
,U
SA
)and
oth
ers
56–8
0%
(concentr
ate
d,
dire
ct
sam
ple
scom
pare
dw
ithcultu
re)
[1–4
]
92–9
8%
(com
pare
dw
ithcultu
re)
[2–4
]Y
es
Ap
pro
xim
ate
ly10%
gre
ate
rse
nsi
tivity
vers
us
ZN
light
mic
rosc
op
yN
ore
ductio
nin
perf
orm
ance
inH
IVco-i
nfe
ctio
n[1
]W
HO
end
ors
ed
Sem
iauto
mate
d,
nonin
tegra
ted
NA
AT
Am
plif
icatio
nand
dete
ctio
nof
myc
ob
acte
rialrR
NA
or
DN
Ad
irect
from
clin
icalsa
mp
les
Am
plif
ied
MTD
1(G
enP
rob
e,
San
Die
go,
CA
,U
SA
)P
rob
eTec
ET
(BD
,Fra
nkl
inLake
s,N
J,U
SA
)C
ob
as
Taq
man
MTB
(Roche
Mole
cula
rD
iagnost
ics,
Ple
asa
nto
n,
CA
,U
SA
)
36–1
00%
(poole
dap
pro
xim
ate
ly66–9
6%
)[5
–7]
54–1
00%
(poole
d85–9
8%
)[5
–7]
No
Sensi
tivity
insm
ear-
negativ
ep
atie
nts
:50–8
0%
[6,
7]
Op
en
syst
em
at
risk
of
DN
Aconta
min
atio
nand
specifi
city
affecte
db
yla
bora
tory
qualit
ycontr
ol
Sim
plif
ied
,m
anualN
AA
TIs
oth
erm
alam
plif
icatio
nw
ithvi
sualre
ad
out
tod
ete
ct
myc
ob
acte
rialD
NA
dire
ct
from
clin
icalsa
mp
les
Eik
en
LA
MP
1(E
iken,
Toky
o,
Jap
an)
Ove
rall:
83%
;sm
ear-
posi
tive
patie
nts
:.
95%
;sm
ear-
negativ
e:
41–5
2%
[8,
9]
.97%
[8]
No
This
test
isund
erg
oin
gla
rge-s
cale
eva
luatio
nand
dem
onst
ratio
nst
ud
ies
by
FIN
Dso
should
not
be
consi
dere
dfu
llyva
lidate
d;
evi
dence
tob
ere
view
ed
by
WH
Oexp
ert
gro
up
inearly
2012
Sero
logic
al
(antib
od
y)d
ete
ctio
nte
st
Imm
unolo
gic
alte
st:
dete
ctio
nof
antib
od
ies
toTB
antig
ens
by
ELIS
Aor
rap
idla
tera
lflo
wfo
rmat
Alth
ough
seve
ralass
ays
are
on
the
mark
et,
no
curr
ently
ava
ilab
lete
sthas
been
valid
ate
dand
pro
ven
tob
eclin
ically
use
ful
0–1
00%
[10]
And
a-T
B(A
nd
aB
iolo
gic
als
,S
trasb
ourg
,Fra
nce)
IgG
:p
oole
dest
imate
sin
smear-
posi
tive
patie
nts
,76%
;sm
ear-
negativ
e,
59%
31–1
00%
[10];
And
a-T
BIg
Gp
oole
dest
imate
:92%
No
WH
Om
ad
enegativ
ere
com
mend
atio
nin
2011
Antig
en
dete
ctio
nte
stTB
antig
ens
dete
cte
db
yE
LIS
Aor
late
ralflo
wte
stfo
rmat
TB
LA
ME
LIS
A(A
lere
,W
alth
am
,M
A,
US
A)
Ove
rall:
18–5
9%
;H
IVonly
:20–6
7%
[11]
88–1
00%
[11]
No
Only
offers
clin
icalutil
ityin
HIV
-infe
cte
dp
atie
nts
with
ad
vanced
imm
unosu
pp
ress
ion
Valid
atio
nof
late
ralflo
wst
ripte
st(D
ete
rmin
eTB
1;
Ale
re)
ongoin
gD
ete
cti
on
of
ac
tive
TB
an
dD
ST
Fully
auto
mate
d,
inte
gra
ted
NA
AT
Fully
auto
mate
d,
self-
conta
ined
pla
tform
inte
gra
ting
sputu
mp
rocess
ing,
myc
ob
acte
rialD
NA
ext
ractio
nand
am
plif
icatio
n
Xp
ert1
MTB
/RIF
(Cep
heid
,S
unnyv
ale
,C
A,
US
A)
M.
tub
ercu
losi
sd
ete
ctio
n:
ove
rall,
90%
;sm
ear-
posi
tive
patie
nts
,94–1
00%
;sm
ear-
negativ
ep
atie
nts
,46–8
3%
(poole
dse
nsi
tivity
75%
);rif
am
pic
inre
sist
ance,
98–9
9%
[12–1
5]
M.
tub
ercu
losi
sd
ete
ctio
n:
.98%
;rif
am
pic
inre
sist
ance:
.98%
[12]
Yes
WH
Ost
rong
recom
mend
atio
nfo
rfr
ontli
ne
TB
dia
gnosi
sin
HIV
-infe
cte
dand
MD
R-T
Bsu
spects
Auto
mate
dliq
uid
cultu
rew
ithin
dire
ct
DS
T
Auto
mate
dsy
stem
for
myc
ob
acte
rialliq
uid
cultu
reand
sub
seq
uent
DS
T
Bact
ecTM
MG
ITTM
960
(BD
Dia
gnost
ics,
Spar
ks,
MD
,U
SA
)B
acT/A
LE
RT1
3D
(bio
Merie
ux,
Marc
yl’E
toile
,Fra
nce)
Sm
ear-
posi
tive
patie
nts
:100%
;sm
ear-
negativ
ep
atie
nts
:.
75%
.99%
Yes
Ap
pro
xim
ate
ly10%
hig
her
dia
gnost
icyi
eld
scom
pare
dw
ithso
lid-m
ed
iacultu
reC
onta
min
atio
ncan
be
10–2
0%
inla
bora
torie
sw
ithp
oor
qualit
yass
ura
nce
Ind
irect
DS
Tcan
take
3–4
month
sto
pro
vid
ere
sults
Nonauto
mate
dliq
uid
cultu
rew
ithd
irect
DS
T
Sim
plif
ied
syst
em
sfo
rm
ycob
acte
rialliq
uid
cultu
rew
ithre
duced
lab
ora
tory
eq
uip
ment
for
MTB
dete
ctio
nand
dire
ct
DS
T
TB
MO
DS
Kit1
(Hard
yD
iagnost
ics,
Santa
Maria
,C
A,
US
A)
96%
(com
pare
dw
ithtr
ad
itional
auto
mate
dliq
uid
cultu
re)
[16]
96%
(com
pare
dw
ithtr
ad
itional
auto
mate
dliq
uid
cultu
re)
[16]
Yes
Low
eq
uip
ment
req
uire
ments
offse
tb
yhig
hla
bour
need
sD
irect
DS
Tp
rovi
des
resu
ltsin
10–1
4d
ays
12
6T
BD
IAG
NO
SIS
WHO recently used the findings of a largemeta-analysis of symptom screening to informthe intensified case finding and isoniazidpreventive therapy guideline for persons livingwith HIV [21]. This study found that, at TBprevalence rates of 5 and 20%, the absence ofcurrent cough, fever, night sweats or weightloss reliably excluded active TB in 98% and90% of patients, respectively [22]. The 2006WHO smear-negative TB diagnostic algorithmwas recently evaluated in ambulatory patientsattending an outpatient clinic, and was foundto have a sensitivity and specificity of only 80%and 44%, respectively [23]. However, despitethis modest diagnostic accuracy, another studyin hospitalised patients suggested that the strictuse of these clinical guidelines could reduce 8-week mortality and hospital length of stay [24].Finally, concerted efforts are underway to unifyclinicoradiological case definitions for differentforms of TB (e.g. TB meningitis) to allow forbetter comparative assessment across studiesand evaluate diagnostic performance moreconsistently across settings [25].
Chest radiology
Radiology is widely used in both high- andlow-burden settings for both TB screening inasymptomatic patients and the diagnosis ofactive disease [26]. Used alone, chest radiologyhas only moderate specificity and, in settingsof high HIV prevalence, moderate sensitivity[27], with 10–71% of HIV co-infected TBpatients having an entirely normal chestradiogram despite culture-positive disease[28–31]. However, when chest radiology isused in conjunction with other simple diag-nostic tools, such as symptom screening and/or smear microscopy, it can offer bothdiagnostic utility and cost-efficacy, particu-larly for ruling out active TB disease [32–34].Two South African studies found that thecombination of symptoms with or withoutsputum smear microscopy followed by chestradiology offered a negative predictive value ofmore than 95% for active TB in a high-burdensetting [35–37]. Unfortunately, optimal chestradiology utility requires interpretation bytrained, skilled observers, which are not alwaysavailable. Interobserver variability of chestradiology has been shown to be poor,irrespective of reader skill [27, 38, 39]. Toovercome this drawback, several radiological
Ta
ble
1.
Contin
ued
.
Te
st
typ
eo
rp
latf
orm
De
sc
rip
tio
no
fte
st
Cu
rre
nt
va
lid
ate
dc
om
me
rcia
lve
rsio
ns
Lo
w/h
igh
se
nsit
ivit
y(e
xp
ec
ted
)L
ow
/hig
hsp
eci-
ficit
y(e
xp
ecte
d)
WH
Oe
nd
ors
em
en
t
Co
mm
en
ts
Phage-b
ase
dd
ete
ctio
nB
acte
riop
hage
viru
ses
infe
ct
and
dete
ct
the
pre
sence
of
viab
leM
.tu
ber
culo
sis
FA
STP
laq
ue
TM
(Bio
tec
Lab
ora
torie
sLtd
,Ip
swic
h,
UK
)81–1
00%
[17,
18]
73–1
00%
[17]
No
3–3
6%
ind
ete
rmin
ate
rate
limits
use
DS
Ta
nd
/or
sp
ec
iati
on
Manual
am
plif
icatio
nand
hyb
ridis
atio
n(L
PA
)
NA
AT
with
hyb
ridis
atio
nof
am
plif
ied
pro
duct
tost
ripte
stallo
win
gfo
rid
entif
icatio
nof
M.
tub
ercu
losi
sand
com
mon
muta
tions
causi
ng
resi
stance
torif
am
pic
inand
isonia
zid
GenoTyp
e1
MTB
DR
plu
s(H
ain
Life
scie
nce
Gm
bH
,N
ehre
n,
Germ
any)
INN
O1
-LiP
AR
if.TB
(Innogenetic
s,G
hent,
Belg
ium
)
For
rifam
pic
inre
sist
ance:
.98%
;fo
ris
onia
zid
:.
84%
[19,
20]
For
rifam
pic
inre
sist
ance:
.98%
;fo
ris
onia
zid
:.
99%
Yes
Pro
vid
es
DS
Tre
sults
on
cultu
reis
ola
tes
and
smear-
posi
tive
clin
icalsp
ecim
ens
in1–2
days
Und
erg
oin
gw
idesp
read
scale
-up
inN
TP
sw
ithth
ehelp
of
the
EX
PA
ND
-TB
pro
gra
mm
eR
ap
idsp
ecia
tion
ass
ay
Rap
idim
munochro
mato
gra
phic
(late
ralflo
w)
test
for
identif
icatio
nof
M.
tub
ercu
losi
scom
ple
xin
cultu
reis
ola
tes
Cap
illa
TB
-Neo1
(Tauns,
Toky
o,
Jap
an)
TB
cID
1(B
DD
iagnost
ics)
SD
Bio
line
Ag
MP
T64
Rap
id1
(Sta
nd
ard
Dia
gnost
ics
Inc.,
Yongin
,S
outh
Kore
a)
.98%
.99%
Yes
Sim
ple
,re
liab
lete
sts
for
use
esp
ecia
llyin
sett
ings
with
hig
hra
tes
of
NTM
s
WH
O:W
orld
Health
Org
aniz
atio
n;LE
D:lig
ht-
em
ittin
gd
iod
e;N
AA
T:nucle
icacid
am
plif
icatio
nte
chniq
ue;LP
A:lin
ep
rob
eass
ay;
rRN
A:rib
oso
malR
NA
;M
.tu
ber
culo
sis:
Myc
ob
acte
rium
tub
ercu
losi
s;Ig
:im
munoglo
bulin
;Z
N:Z
iehl–
Neels
en;
FIN
D:
Found
atio
nfo
rIn
nova
tive
New
Dia
gnost
ics;
MD
R:
multi
dru
g-r
esi
stant;
NTP
:natio
nalTB
pro
gra
mm
e;
EX
PA
ND
-TB
:E
xpand
ing
Access
toN
ew
Dia
gnost
ics
for
TB
;N
TM
:non-t
ub
erc
ulo
us
myc
ob
acte
rium
.
12
7J.
G.
PE
TE
RE
TA
L.
scoring systems, such as the Chest Radiograph Reading System, have been developed to improveinterobserver variability [31]. Furthermore, automated computer systems to interpret and reportdigital chest radiograms are currently in development [40].
Sample acquisition technology
Definitive TB diagnosis relies on the demonstration of TB organisms, or TB-specific antigens orgenetic material. For this, an appropriate and sufficient biological sample is essential. Attainingadequate samples can, however, be challenging and a major obstacle to diagnosis. A number ofstrategies and techniques have been evaluated to improve sputum expectoration, to induce sputum orto attain an alternative pulmonary sample suitable for laboratory TB diagnosis (table 2).
Of these techniques, sputum induction is emerging as the optimal technique given its safety, efficacyand feasibility even in resource-limited settings. The challenge lies in successfully integrating sputuminduction into busy, routine clinical practice settings with limited resources.
Smear microscopy
Although widely used, the sensitivity of smear microscopy is highly variable, ranging between 20%and 80% [62], performing poorest in HIV-infected patients [27] and children [63]. Additionally,smear microscopy relies on well-trained microscopists, and sensitivities between field andreference laboratories can vary by as much as 28% [14].
The most important developments in optimising smear microscopy and associated WHO policychanges are outlined in figure 3a. In addition, several innovative approaches to further improvesmear microscopy are under development, including improved concentration techniques usingnanobeads, fluorescence in situ hybridisation, automated and computer-assisted smear readingtechnologies, and use of mobile phones for microscopy [64, 65].
The expansion and development of culture-based techniques
Mycobacterium tuberculosis culture remains the clinical and research diagnostic gold standard for allforms of active TB. Figure 3b outlines the progress and associated WHO policy changes formycobacterial culture techniques for both diagnosis and DST. Traditional solid culture methods aretedious, time-consuming and have limited clinical impact. Automated liquid culture systems, withapproximately 10% higher yields and a decreased time to diagnosis [66], have largely replaced solidculture. However, automated liquid cultures are expensive, prone to contamination, and requireconsiderable laboratory infrastructure and expertise. Thus, despite WHO endorsement in 2007, theyremain inaccessible to populations where they are most needed. In 2009, the WHO endorsed the use ofalternative, simpler, less expensive noncommercial culture and DST technologies, as an interim measurewhile the capacity for genotypic testing is scaled up [67]. Details of the microscopy observed drugsusceptibility (MODS) method are shown in table 1, while other endorsed noncommercial methodsinclude the colorimetric redox indicator and the nitrate reductase assay. Under controlled laboratoryconditions, these noncommercial culture methods are inexpensive and can provide culture and DSTresults in 7–14 days [67, 68]. Lack of standardisation and local variations in methodology remainprogrammatic concerns and have thus far limited scale-up. Phage-based methods have not been WHO-endorsed due to insufficient evidence, variable specificity and high rates of invalid results [69].
Conventional, nonintegrated nucleic acid amplification techniques (NAATs) (table 1) have beenfound to offer high specificity (85–98%) and sensitivity for smear-positive TB (,96%), but poorer
12
8T
BD
IAG
NO
SIS
sensitivity (,60%) and specifi-city for smear-negative TB [5–7].Compared with smears and culture,these assays are expensive, requiringspecialised laboratory infrastructureand expertise, while, being opensystems, they are at risk for cross-contamination in settings with sub-optimal laboratory quality. Thesefactors have limited their widespreaduptake in high-burden, resource-limited settings. Simplified, manualNAATs, such as LAMP1 (loop iso-thermal amplification PCR; Eiken,Tokyo, Japan) using isothermalamplification and a visual readouthave been developed as more afford-able options where laboratory in-frastructure is limited [8]. Earlyevaluation suggested similar perfor-mance to other commercial NAATswith a sensitivity of ,40% in smear-negative TB [8]. The Foundation forInnovative New Diagnostics (FIND)is currently conducting large-scaleevaluation and demonstration stu-dies of LAMP, with promisingpreliminary results [9]. However,despite the assay’s simplicity, therisk of cross-contamination duringmanual DNA extraction and needfor laboratory training and technicalskill remain, and may prevent wide-spread application.
An integrated, fullyautomated NAAT: Xpert1
MTB/RIF
In December 2010, the WHOannounced the endorsement ofthe novel Xpert1 MTB/RIF assay(Cepheid, Sunnyvale, CA, USA) [70]. Xpert1 MTB/RIF is a fully automated and integrated DNAextraction and amplification system, thereby addressing many limitations of existing commercialNAATs [71]. Furthermore, Xpert1 MTB/RIF has the potential to be performed in decentralisedlocations outside of reference laboratories by staff with minimal laboratory training (1–2 days).
To date, the Xpert1 MTB/RIF assay has undergone evaluation in sputum samples from more than11,000 patients in 19 countries [12–15, 61, 72–74], although these studies have performed Xpert1
MTB/RIF in a laboratory, rather than at the POC. A meta-analysis of these 18 published studiesshowed a sensitivity and specificity of a single sputum-based Xpert1 MTB/RIF for culture-positiveTB of 90.4% (95% CI 89.2–91.4%) and 98.4% (95% CI 98.0–98.7%), respectively, and 75% forsmear-negative, culture-positive pulmonary TB [12]. Performing a second and third MTB/RIFincreases sensitivity by approximately 13% and 5%, respectively [15], while indeterminate rates of
Old tests New testsPOC antigen detection test¶ e.g. LAM lateral flow assay
Sensitivity key:
Automated integrated NAATs; Xpert® MTB/RIF assay#
Overall
Smear negative 70%
Simplified, manual NAAT; LAMP
Overall
Smear negative 60%
Automated nonintegrated NAAT e.g. Amplified MTD®
Overall
Manual amplification and hybridisation (LPA)# GenoType® MTBDRplus; main use DST
Smear positive
Smear negative 60%
Direct, noncommercial culture tools e.g. MODS#
Children
95%
Figure 2. Comparison of the sensitivity and time to diagnosis foractive pulmonary tuberculosis (TB) and drug-susceptibility diagnos-tic tools indicating areas of reduced performance in children andHIV-TB co-infected patients. Only tests commercially availableand with a specificity of .95% for the diagnosis of active TB areincluded. BactecTM MGITTM 960 is manufactured by BD Diagnostics(Sparks, MD, USA). Xpert1 MTB/RIF is manufactured by Cepheid(Sunnyvale, CA, USA). Amplified MTD1 is manufactured by GenProbe(San Diego, CA, USA). GenoType1 MTBDRplus is manufactured byHain Lifescience GmbH (Nehren, Germany). POC: point-of-care;LAM: lipoarabinomannan; NAAT: nucleic acid amplification techni-que; LAMP: loop isothermal amplification PCR; LPA: line probe assay;DST: drug-susceptibility testing; MODS: microscopy observed drugsusceptibility. #: diagnostic test that can be used for bothMycobacterium tuberculosis detection and DST; ": LAM lateral flowstrip test is restricted to use in HIV-infected patients and will only becommercially available in the fourth quarter of 2012.
12
9J.
G.
PE
TE
RE
TA
L.
Ta
ble
2.
Str
ate
gie
s/te
chniq
ues
for
the
imp
rove
dacq
uis
ition
of
pulm
onary
sam
ple
sfo
rtu
berc
ulo
sis
(TB
)d
iagnosi
s
Te
ch
niq
ue
Dia
gn
osti
cp
erf
orm
an
ce
ran
ge
s#
Ad
va
nta
ge
sD
isa
dva
nta
ge
sK
no
wle
dg
eg
ap
s
Exp
ec
tora
ted
sp
utu
ma
ssis
tan
ce
tec
hn
iqu
es
Pro
vid
er
train
ing
and
ob
serv
ed
sputu
mcolle
ctio
nM
ala
wia
nS
N-T
Bsu
spects
[41]:
39
out
of
46
(85%
)d
efin
iteTB
case
sd
ete
cte
d
Min
imalst
aff
train
ing
req
uire
ments
Inexp
ensi
veW
idely
ap
plic
ab
lein
all
sett
ings
Tim
e-c
onsu
min
gIn
fectio
ncontr
olris
kN
ot
ap
plic
ab
lefo
rchild
ren
Pro
gra
mm
atic
rese
arc
hon
imp
lem
enta
tion,
up
take
and
effic
acy
Sp
utu
msu
bm
issi
on
inst
ructio
ns/
train
ing
Paki
stanife
male
s[4
2]:
qsm
ear-
posi
tive
case
dete
ctio
n;
Qsp
ot-
sputu
msa
liva
sub
mis
sion;
qfe
male
sre
turn
ing
with
sputu
mIn
donesi
an
male
s/fe
male
s[4
3]:
15%
hig
her
case
dete
ctio
n
Min
imalst
aff
train
ing
req
uire
ments
Inexp
ensi
veW
idely
ap
plic
ab
lein
all
sett
ings
No
infe
ctio
ncontr
olris
k
Tim
e-c
onsu
min
gN
ot
ap
plic
ab
lefo
rchild
ren
Pro
gra
mm
atic
rese
arc
hon
imp
lem
enta
tion,
up
take
and
effic
acy
Sp
utu
min
du
cti
on
tec
hn
iqu
es
Phys
icalm
anoeuvr
es
( e.g
.chest
phys
ioth
era
py)
Dia
gnost
icyi
eld
(TB
cultu
re):
ad
ults
,5–2
6%
[41,
44];
smear
sensi
tivity
(TB
cultu
rere
fere
nce
stand
ard
):ad
ults
,50–5
3%
[41,
44]
Safe
pro
ced
ure
Min
imaltr
ain
ing/n
oeq
uip
ment
req
uire
ments
Low
dia
gnost
icyi
eld
Hig
hin
fectio
nris
kfo
rhealth
work
er
Curr
ently
rest
ricte
dto
hosp
itals
with
train
ed
phys
ioth
era
pis
ts
No
stud
ies
inp
rimary
care
sett
ings
and
child
ren
Few
com
paris
on
stud
ies
with
oth
er
meth
od
sof
sputu
min
ductio
n
Ultr
aso
nic
neb
ulis
atio
n"
Dia
gnost
icyi
eld
(TB
cultu
re):
ad
ults
,8–3
4%
[41,
45];
child
ren,
10–3
0%
[46,
47]
Sm
ear
sensi
tivity
(TB
cultu
rere
fere
nce
stand
ard
):ad
ults
,37–7
8%
[45,
48];
child
ren
+ ,20–5
7%
[49,
50]
Safe
pro
ced
ure
Nonin
vasi
veFeasi
ble
inre
sourc
e-p
oor
sett
ings
Good
yield
inad
ults
and
child
ren
Sim
ple
rp
erf
orm
ance
inH
IV-i
nfe
cte
dand
-unin
fecte
dp
atie
nts
Eq
uip
ment
and
consu
mab
lecost
sH
igh
infe
ctio
nris
kfo
rhealth
work
er
Curr
ently
rest
ricte
dto
dis
tric
thosp
itals
with
infe
ctio
ncontr
olfa
cilitie
s
Few
stud
ies
inp
rimary
care
sett
ings
No
stud
ies
of
imp
act
on
patie
nt-
imp
ort
ant
outc
om
es,
posi
tionin
gin
dia
gnost
icalg
orit
hm
sand
use
of
nove
ld
iagnost
icto
ols
on
ind
uced
sputu
msa
mp
les
Oth
er
devi
ces
( e.g
.vi
bra
tion
toolsu
ch
as
lung
flute
)43%
smear
mic
rosc
op
yse
nsi
tivity
[51]
(sm
all
stud
yof
15
patie
nts
)S
afe
pro
ced
ure
Nonin
vasi
veD
isp
osa
ble
/self-
exp
lanato
ryeq
uip
ment
decre
ase
sin
fectio
nris
k
Hig
hcost
sand
wast
eof
devi
ce
Feasi
bility
ind
iffere
nt
sett
ings
Infe
ctio
nris
kTools
still
ind
eve
lop
ment
Pro
spectiv
est
ud
ies
req
uire
din
clin
ically
ap
pro
pria
tese
ttin
gs
Alt
ern
ati
ve
resp
ira
tory
sa
mp
lea
cq
uis
itio
nte
ch
niq
ue
sG
ast
ricw
ash
ings"
Dia
gnost
icyi
eld
(TB
cultu
re):
ad
ults
,11–3
0%
[52,
53];
child
ren,
5–1
7%
[46,
54]
Sm
ear
sensi
tivity
(TB
cultu
rere
fere
nce
stand
ard
):ad
ults
,30–3
7%
[41,
53];
child
ren,
18–5
3%
[46,
54]
Safe
and
effectiv
ep
roced
ure
esp
ecia
llyfo
rchild
ren
Min
imalin
fectio
nris
k
Inva
sive
pro
ced
ure
Req
uire
sfa
stin
gS
am
ple
colle
ctio
nad
vise
don
3conse
cutiv
ed
ays
Not
feasi
ble
inm
any
pub
lichealth
facilitie
sC
urr
ently
rest
ricte
dto
dis
tric
t-le
vel
hosp
italse
ttin
gs
Pro
gra
mm
atic
stud
ies
of
yield
from
reso
urc
e-p
oor
sett
ings
13
0T
BD
IAG
NO
SIS
only 1–3%, decreasing to ,1% after repeattesting, have been found across settings [14,15]. Importantly, the use of MTB/RIFdecreased the mean time to treatmentinitiation amongst smear-negative, culture-positive TB patients from 56 to 5 days,similar to that of smear-positive patients[14]. For the detection of rifampicinresistance, the meta-analysis data show asensitivity and specificity of 94.1% and97.0%, respectively [12]. On this evidencebase, WHO has made a strong recommen-dation for the use of frontline Xpert1 MTB/RIF in all patients with suspected drug-resistant (DR)-TB and/or co-infected withHIV, and a conditional recommendation,in acknowledgment of resource implica-tions, for the use of Xpert1 MTB/RIF as afollow-on test to microscopy in settingswhere multidrug-resistant (MDR)-TB orHIV is of lesser concern, especially forsmear-negative TB [70].
Undoubtedly, a number of unansweredquestions and concerns surrounding theuse of the Xpert1 MTB/RIF assay remain.First, although specificity for detectingrifampicin resistance remains .98%, stu-dies continue to find false-positive rifam-picin resistance results [75]. In areas of lowMDR-TB prevalence, given the significantdecrease in positive predictive value asso-ciated with small reductions in specificity, alarge number of false-positive rifampicinresistance results may occur with wide-spread routine use. Despite the develop-ment of updated versions of both theGeneXpert1 cartridge and software tofurther improve assay specificity, thisremains an important concern. Secondly,given the reduced ability of a single Xpert1
MTB/RIF test to rule out TB in HIV-infected patients [13], the role of additionalXpert1 MTB/RIF tests and alternativeinvestigations in HIV-infected patientswith ongoing symptoms needs to be betterdefined. Thirdly, given that Xpert1 MTB/RIF detects both viable and non-viable M.tuberculosis, the interpretation of a positiveXpert1 MTB/RIF in patients not respond-ing to TB therapy and the use, if any, ofXpert1 MTB/RIF for treatment monitoringrequires urgent clarification. Finally, anumber of operational challenges and
Ta
ble
2.
Contin
ued
.
Te
ch
niq
ue
Dia
gn
osti
cp
erf
orm
an
ce
ran
ge
s#
Ad
va
nta
ge
sD
isa
dva
nta
ge
sK
no
wle
dg
eg
ap
s
Naso
phary
ngealasp
irate
"D
iagnost
icyi
eld
(TB
cultu
re):
child
ren,
7–9
%[5
4,
55]
Sm
ear
sensi
tivity
(TB
cultu
rere
fere
nce
stand
ard
):child
ren,
58–7
1%
[54,
56]
Safe
Feasi
ble
acro
ssse
ttin
gs
Use
fulfo
rd
iagnosi
sof
oth
er
resp
irato
ryp
ath
ogens
( e.g
.vi
ruse
s),
esp
ecia
llyin
child
ren
Low
dia
gnost
icyi
eld
for
TB
Infe
ctio
nris
kC
urr
ently
rest
ricte
dto
dis
tric
t-le
vel
hosp
ital
Pro
gra
mm
atic
stud
ies
of
dia
gnost
icutil
ityin
routin
eclin
icse
ttin
gs
Bro
nchosc
op
y"D
iagnost
icyi
eld
(TB
cultu
re):
ad
ults
,9–4
6%
[41,
57]
Sm
ear
sensi
tivity
(TB
cultu
rere
fere
nce
stand
ard
):ad
ults
,27–6
3%
[58,
59]
Eq
uiv
ale
nt
dia
gnost
icyi
eld
tooth
er
sam
ple
acq
uis
ition
meth
od
sb
ut
allo
ws
dire
ct
visu
alis
atio
nof
resp
irato
rytr
act
¡b
iop
syA
llow
sd
iagnosi
sof
oth
er
resp
irato
ryp
ath
ogens
( e.g
.P
neu
mocy
stis
)
Inva
sive
Req
uire
ssp
ecia
lised
eq
uip
ment
and
staff
Rest
ricte
dto
tert
iary
and
dis
tric
thosp
itals
Exp
ensi
veIn
fectio
nris
kfo
rhealth
work
er
Stu
die
son
the
perf
orm
ance
of
nove
ld
iagnost
ics
usi
ng
bro
nchoalv
eola
rla
vage
fluid
tod
iagnosi
sor
exc
lud
eactiv
eTB
SN
:sm
ear-
negat
ive.
#:
dia
gnost
icperform
ance
chara
cte
ristic
sare
from
pro
spectiv
est
udie
spre
dom
inantly
inhig
h-b
urd
en
countr
ies.
Wid
eva
riatio
nand
hete
rogeneity
pre
dom
inantly
acco
unte
dfo
rby
diff
ere
nces
inin
clu
ded
study
pop
ula
tions
( e.g
.S
N-T
Bsu
spect
sve
rsus
SN
-TB
susp
ects
with
chest
radio
gra
phy
sugges
tive
of
TB
)and
backg
round
TB
pre
vale
nce.
Inin
duced
sputu
mst
udie
susi
ng
ultr
aso
nic
nebulis
atio
n,
ad
ult
studie
sin
clu
de
only
SN
/sputu
m-s
car
ce
TB
susp
ect
s,w
hile
child
studie
sare
of
TB
susp
ects
with
out
prio
rdia
gnost
icte
stin
g.
Perform
ance
outc
om
es
for
nove
ldia
gnost
ics
applie
dto
acquire
dpulm
onary
sam
ple
sare
noti
nclu
ded."
:com
para
tive
stud
ies
ofs
om
e/a
lloft
hese
sam
ple
acq
uis
ition
tech
niq
ues
have
been
perform
edfo
radults
[41,4
4,5
3,5
8]a
nd
child
ren
[46,5
4,6
0].
Forboth
ad
ults
and
child
ren,s
putu
min
ductio
nusi
ng
ultr
aso
nic
nebulis
atio
nis
eq
uiv
ale
nto
rsu
perio
rto
the
more
inva
sive
tech
niq
ues
ofg
astr
icw
ash
ing
and
bro
nchosc
op
y.+ :a
recen
tst
udy
of4
52
child
ren
with
susp
ect
ed
TB
found
dia
gnost
icyi
eld
ofin
duced
sputu
mto
be
15%
and
the
sensi
tivity
oft
he
nove
ldia
gnost
ic,Xp
ert1
MTB
/RIF
(Cepheid
,S
unnyv
ale
,C
A,U
SA
),to
be
83%
(58
outof7
0patie
nts
)[61].
13
1J.
G.
PE
TE
RE
TA
L.
•
•
•
•
•
• MODS• NRA
•
•
•
Impacts•
•
•
•
•
Impacts•
•
•
•
•
•
•
•
•
•
•
2006 2007 2009 2011
<2000 2004 2007 2009
Systematic review/clinical
trial data
WHO policy endorsements
WHO policy endorsements
Systematic review/clinical
trial data
Reduced workloads for microscopy centres
Improved diagnosisDecreased diagnostic drop-out
Reality check
High variability in smear microscopy sensitivity between primary clinics and reference laboratories Limited uptake of LED microscopy by NTPs in HBCS
More sensitive and rapid culture tools availableImproved time to DSTAlternative culture techniques developed, tested and endorsed for HBC with limited laboratory infrastructureGood BactecTM
MGITTM uptake with WHO endorsement
Reality checkLaboratory capacity-building making slow progress
High contamination rates in poor laboratories
Best culture time to diagnosis 4_7 days with consequent diagnostic delay
•
•
•
Sputum processing methods (e.g. bleach and centrifugation) improves sensitivity by average of 18%FM 10% better sensitivity versus LM3rd sputum yield 2_5% only
LED microscopy non-inferior to MVLP FM even in HIV
Front-loaded smear microscopy (spot-spot versus spot-morning) offers equivalent diagnostic yield
RCT demonstrates non-inferiority of spot-spot versus spot-morning with significantly less diagnostic drop-out2 smears from single sample equivalent to 2 smears on 2 samples
•
•
Revised case definition for TB number of smears required for diagnosis from 3 to only 2
LED microscopy for FM
•
•
Solid culture method (Lowenstein–Jensen or Ogawa media) standard technique taking 4_8 weeks
Agar proportion and indirect method used for DST (Middlebrook 7H10 media recommended)
Liquid TB culture (manual or automated) ~10% improved diagnostic yield and time-to-diagnosis by 10_15 days
Indirect methods for DST using automated liquid culture (e.g. BactecTM MGITTM SIRE)
Liquid media for culture and DST in middle- and lower-income countries Laboratory capacity building initiative (FIND/WHO)
Development and evaluation studies of alternative non-commercial culture techniques allowing for direct DST
Use of noncommercial culture techniques (MODS, NRA or CRI) for rapid direct DST
a)
b)
Figure 3. Progress in optimising and streamlining a) sputum smear microscopy and b) tuberculosis (TB) culture.BactecTM MGITTM and SIRE are manufactured by BD Diagnostics (Sparks, MD, USA). FM: fluorescencemicroscopy; LM: light microscopy; LED: light-emitting diode; MVLP: mercury vapour lamp; RCT: randomisedcontrolled trial; WHO: World Health Organization; NTP: national TB programme; HBC: high-burden country;DST: drug-susceptibility testing; MODS: microscopy observed drug susceptibility; NRA: nitrate reductase assay;CRI: colorimetric redox indicator; FIND: Foundation for Innovative New Diagnostics.
13
2T
BD
IAG
NO
SIS
research questions associated withthe national and international scale-up of Xpert1 MTB/RIF and othernew TB diagnostic technologiesremain and are outlined in figure 4.
Line probe assays
As a rapid alternative to phenotypicDST, specialised NAATs using man-ual amplification and hybridisationtechniques, known as line probeassays (LPAs) (table 1), offer M.tuberculosis speciation and genotypicDST with results in 1–2 days. LPAsreceived WHO endorsement in 2008as the test of choice for rapidgenotypic rifampicin and isoniazidDST [76]. LPAs offer sensitivities.98% for rifampicin resistance, butonly ,85% for isoniazid resistance due to the presence of resistance coding mutations outside theregions of the inhA and katG genes detected by the assays [19, 77]. LPAs are now routinely available andare being scaled-up in certain national TB programmes, e.g. South Africa and India by the ExpandingAccess to New Diagnostics for TB (EXPAND-TB) project for MDR-TB suspects with smear- or culture-positive samples. Recently, an assay has been developed and is now available for rapid genotypic second-line DST and extensively drug-resistant (XDR)-TB diagnosis (GenoType1 MTBDRsl; Hain LifescienceGmbH, Nehren, Germany). Sensitivities vary across initial studies depending on the specific drug tested[78–80] and clinical utility is restricted to rapidly ruling in XDR-TB at this stage.
Antigen detection
The detection of circulating TB antigens for diagnosis using different biological samples has beenextensively studied [11]. A recent meta-analysis evaluated 47 studies using 12 different single orcombinations of TB antigen [11]. Lipoarabinomannan (LAM), a 17.3-kDa immunogenic glycolipidcomponent of the mycobacterial cell wall, is most extensively evaluated and, using the urine TB LAMELISA (Alere, Waltham, MA, USA) (table 1), has shown promising utility for HIV-infected patientswith advanced immunosuppression [81–84]. In HIV-infected patients, urine TB LAM ELISA has anoverall sensitivity of ,50%, increasing to 67% and 85% in HIV-infected patients with CD4 counts,50 cells?mL-1 from out- and in-patient settings, respectively [82, 85], and an overall specificity of83–100% [83, 86, 87]. Cross-reactivity with nontuberculous mycobacteria or other commensalorganisms (e.g. Candida [83]) remains a concern. Given these performance characteristics, urine TBLAM ELISA, despite commercial availability, is not yet widely used or approved by WHO. However,urine LAM positivity has been correlated with bacterial burden [88] and may identify TB HIV co-infected patients with the highest mortality [89]. These factors, together with recent progression ofthe TB LAM ELISA into a POC lateral flow strip test, means that urine LAM, if used for rapiddiagnosis to guide the early initiation of TB treatment in high-risk HIV/TB co-infected patients, mayoffer important clinical utility, and impact patient mortality and morbidity. Further research isnecessary to confirm these hypothesised clinical benefits of POC LAM testing.
Immunodiagnosis of active TB
Numerous serological tests to detect TB-specific antibodies are available in many developingcountries [90]. Updated meta-analyses show current serological assays are of no clinical value,with high variability in both sensitivity and specificity [10], and poor cost-effectiveness [91].
1
6
32
5
4
71)
2)3) Limited evidence around use at point-of-treatment4) 5)6)7)
High cost (both machine and cartridge) of Xpert® MTB/RIF, aggravated by supplier monopolyLimited evidence around use in special patient groups, e.g. HIV+, children
No single test fulfils all requirements of TB management, e.g. diagnosis and treatment monitoringEthical concerns around diagnosing DR-TB without available second-line therapeuticsLack of clear government regulation to prevent unproven tests from gaining market shareRapid technical innovation flooding the market with new tests with competing evidence
Figure 4. The challenges of scale-up and implementation of Xpert1MTB/RIF (Cepheid, Sunnyvale, CA, USA) and other novel tubercu-losis (TB) diagnostic technologies. DR-TB: drug-resistant TB.
13
3J.
G.
PE
TE
RE
TA
L.
Despite the demonstrated lack of either accuracy or cost-efficacy, these tests continue to be sold in17 out of 22 high-burden settings [90]. In India alone, an estimated US$15 million per annum isspent on performing serological tests for TB in the private sector [90]. In 2011, in response to this,WHO issued a negative policy advising against the use of any of the numerous available bloodserological assays for the diagnosis of TB [92], and countries such as India have banned the clinicaluse of these tests. However, this policy does not discourage ongoing research into serological tests forTB diagnosis.
Blood-based interferon-c release assays (IGRAs) in high-burden settings have also been extensivelyevaluated and found to offer little, if any, clinical utility as a frontline diagnostic tool for active TBin either HIV-infected or -uninfected patients [36, 93–95]. A recent WHO policy (2011)discourages use of IGRAs for active TB diagnosis in low- and middle-income countries [96]. Thetuberculin skin test (TST) remains a useful tool for the diagnosis of active TB in young children andIGRAs offer equivalent, but not superior, performance [97, 98]. It is clear that immunodiagnosis is not asubstitute for molecular or microbiological site-of-disease diagnosis, although its use for investigatingLTBI remains important and is discussed later.
Diagnosis of TB infection
There is a growing recognition that LTBI is a spectrum that is poorly understood [99]. Both theTST and IGRAs are widely used as surrogate markers for TB infection and, consequently, for thediagnosis of LTBI. These assays have been extensively studied and systematically reviewed in anumber of settings [94, 100–102].
IGRAs were developed to improve the specificity of TST, as they are not affected by bacilleCalmette–Guerin (BCG) vaccination status. Hence, they are useful in the evaluation of LTBI
Table 3. Diagnostic accuracy of old and novel tests for common forms of extrapulmonary tuberculosis (EPTB)
Type of EPTB Traditional diagnostic testperformance
Commercial NAAT: no dataXpert1 MTB/RIF [112]: sensitivity,
29–100%
Novel diagnostics poorly studied in bothascitic fluid and histopathological samples
Xpert1 MTB/RIF is manufactured by Cepheid (Sunnyvale, CA, USA). NAAT: nucleic acid amplification technique; FNA: fine-needle aspiration; IFN:interferon; POC: point-of-care; ADA: adenosine deaminase.
13
4T
BD
IAG
NO
SIS
amongst BCG-vaccinated individuals, especially those who received BCG after infancy or multipleBCG vaccinations. Given the wide variations in BCG policies, online resources have beendeveloped to guide practice by helping clinicians and public health practitioners review variationand their potential impacts on TST performance. These resources include a world atlas of BCGpolicy and practices [103], and a web-based algorithm for interpreting TST and IGRAs [104].
The use of IGRAs in clinical practice for diagnosing and managing LTBI shows considerablediversity. A recent survey of 33 IGRA guidelines and position papers from 25 countries and twosupranational organisations has been published [101]. Four diagnostic approaches were commonlyproposed: 1) a two-step approach of TST first, followed by IGRA either when the TST is negative (toincrease sensitivity, mainly in immunocompromised individuals), or when the TST is positive (toincrease specificity, mainly in BCG-vaccinated individuals); 2) either TST or IGRA, but not both; 3)IGRA and TST together (to increase sensitivity); and 4) IGRA only, replacing the TST. Overall, inlow-burden settings, IGRAs are increasingly recommended to guide the use of preventative therapy.However, this survey suggests that most current guidelines do not use objective, transparentmethods to grade evidence and recommendations, and do not disclose conflicts of interest [105].
Increasingly, it is becoming evident that neither IGRAs nor TST can adequately define or resolvethe various stages of TB infection [99, 106]. A growing proportion of IGRA and TST studiesshow that both tests have limited prognostic value. For instance, a large proportion (.95%) ofTST- or IGRA-positive individuals will not progress to active TB disease [107]. These findingsindicate that the existing diagnosis of LTBI using IGRAs or TST may not be ideal to guide theuse of preventive therapy to the subgroup of individuals who are most likely to benefit from it.Novel, highly predictive biomarkers, or combinations of biomarkers and risk factors (i.e. acomposite risk prediction model), allowing for accurate prediction of patients with the highestrisk of progression to active disease are urgently required [108]. For example, efforts areunderway to develop a PCR-based test, as a follow-up test to IGRAs, for detecting transcriptionalprofiles of immune cells circulat-ing in the blood, which mighthelp predict risk of disease pro-gression [109].
Diagnosis of EPTB
EPTB is diagnostically challengingand composite reference standardsare the norm rather than the excep-tion. Obtaining samples for diag-nosis often requires specialised skillsand equipment (e.g. biopsy andlumbar puncture), and the tradi-tional diagnostics of smear micro-scopy and culture perform poorlyon many of the paucibacillary,nonsputum biological samples.
Table 3 compares the perfor-mance of old and novel tools forthe most common forms of EPTB.Overall, body cavity fluids (e.g.pleural, pericardial and cerebrosp-inal fluids) are paucibacillary,smear microscopy performance isdismal and liquid culture performsvariably. Biomarkers, such as
Multiplex assaysfor associated disease
in simple format
C
CHBVHIV
HCV
HBVHIV
HCV
FluorescenceLED microscopy
9A
Hand-held PCRassays for robust
POC useAutomated,
intergrated NAAT
Immunochromatographicstrip test
GeneXpert®GX-l
Figure 5. Current progress and future evolution of tuberculosisdiagnosis from smear microscopy to molecular methods andonwards towards simple, affordable point-of-care (POC) testformats. GeneXpert1 is manufactured by Cepheid (Sunnyvale, CA,USA). LED: light-emitting diode; NAAT: nucleic acid amplificationtechnique; HBV: hepatitis B virus; HCV: hepatitis C virus.
13
5J.
G.
PE
TE
RE
TA
L.
unstimulated interferon (IFN)-c and adenosine deaminase, appear to be useful but underused[119, 120]. Xpert1 MTB/RIF looks promising for EPTB in initial studies with few patients and thebenefit of concentrating larger samples volumes seem similar to culture, with minimal PCRinhibition ([111] and unpublished data).
Towards POC technology for active TB
Despite the advancement in the molecular diagnosis of TB and drug resistance, the need for a simple,instrument-free, laboratory-free POC test continues to be articulated by both research groups andcivil societies [121–123]. Required minimum specifications for the ideal POC TB test have beendefined by a number of groups and recently published [124]. Mathematical models suggest a hugepotential impact of POC TB diagnosis on both case detection and overall TB incidence [125, 126].
Some commercially available novel diagnostic tests already come close to meeting theserequirements and may offer important POC utility. The Xpert1 MTB/RIF assay easily meets thespecifications for diagnostic test accuracy (sensitivity .95% for smear-positive, culture-positivepatients and 60–80% for smear-negative, culture-positive patients; specificity .95%) and time toresult (,3 hours), but falls short as an ideal decentralised POC test because of its cost and thespecialised equipment needed. Feasibility and impact studies of point-of-treatment, clinic-basedXpert1 MTB/RIF are nearing completion and will provide insights on its POC utility.
For TB/HIV co-infected patients with advanced immunosuppression, the Determine1 TB LAM Agstrip test (Alere) offers POC potential. It is an instrument-free, laboratory-free, affordable(,US$3.50) POC test producing results within 25 minutes and using an easily obtainable urine
Table 4. Important current unmet tuberculosis (TB) diagnostic needs and research gaps
Research gap and/or unmetdiagnostic need
Rationale for need and/or research question(s)
1) Development of a simple, affordable, field-friendlyPOC for active TB using sputum samples
High-burden countries, severely limited resourcesPoor laboratory infrastructure and technical skillsPatients have difficulty accessing health services and default prior to diagnosis
2) Impact evaluations of different simple and safe sampleacquisition techniques e.g. sputum induction forsputum-scarce, smear-negative and childhood TB inprimary care settings
Up to one-third of patients in high HIV and TB prevalence settings are unableto produce sputum
All TB diagnosis relies on an adequate sampleSputum induction is simple and feasible yet carries high infection risk and
moderate cost3) Impact evaluations of Xpert1 MTB/RIF at different
healthcare levels, operational research and costefficacy evaluations of Xpert1 MTB/RIF, and optimalpositioning of Xpert1 MTB/RIF in diagnostic algorithms
Rapid WHO endorsement and plan for global implementationBenefits of a 2-hour test result may be lost if not used at point of treatment and
rapidly available to patientsOperational performance and actual cost efficacy unknown
4) Development of rapid, non-sputum based POC test forthe diagnosis of EPTB and childhood TB
EPTB and children most often unable to produce sputumBiological samples (e.g. urine) readily availableCertain forms of EPTB (e.g. TB meningitis) carry very high mortality and rapid
diagnosis would save lives5) Development of a rapid rule-out test for TB HIV
co-infection for use in high-burden settingsTB can be clinically atypical in HIV co-infection but progresses rapidly with high
mortality rateHigh TB drug-related morbidity in HIV-infected patientsOther pathogens can mimic TB presentation and cause mortality if untreated
6) Further studies and impact evaluation of availablePOC urine LAM strip test for HIV-infected patientswith advanced immunosuppression
First simple, affordable, rapid, non-sputum based TB diagnostic availableTargets HIV co-infected patients with advanced immunosuppression and highest
TB-related mortalityLack of clarity about test specificity, cut-point selection and test patient impact
6) Development of simple-to-perform, improved rapidmolecular assays for first and second-line drugresistance
Growing epidemic of MDR- and XDR-TBAll phenotypic DST methods require at least 10–14 days to provide results
7) Predictive biomarker(s) to identify latently infectedpeople likely to progress to active TB and who willbenefit most from preventive therapy
IGRA and TST predict progression to active TB suboptimallyIsoniazid preventative therapy can cause significant individual morbidity and require
large public health expenditure
Xpert1 MTB/RIF is manufactured by Cepheid (Sunnyvale, CA, USA). POC: point-of-care; EPTB: extrapulmonary TB; LAM: lipoarabinomannan;WHO: World Health Organization; MDR: multidrug-resistant; XDR: extensively drug-resistant; DST: drug-susceptibility testing; IGRA: interferon-crelease assay; TST: tuberculin skin test.
13
6T
BD
IAG
NO
SIS
sample with low infectious risk [82]. Unfortunately, diagnostic accuracy is dismal in unselected TBpatients, and its use is restricted to HIV-infected patients with advanced immunosuppression [82].Nevertheless, two initial evaluations in HIV-infected out- and in-patients showed similar diagnosticaccuracy measures to the preceding TB LAM ELISA, and improved sensitivity when combined withsputum smear microscopy [127, 128]. Issues of test specificity and inter-reader agreement wheninterpreting the faintest bands still requires further study, while the measurable impact of a POC testwith modest sensitivity on morbidity, mortality and hospital length of stay needs to be demonstratedprior to widespread uptake and WHO endorsement. Neither of these tests provide the ideal POC TBtest, yet they indicate that the development of such a test may be within reach [121].
The ongoing progress towards POC TB diagnosis is outlined in figure 5. The diagnostic pipelinehas many promising molecular POC tests and platforms under development. Hand-held orportable platforms using DNA chips and/or disposable cartridges are being evaluated for POC,simplified NAATs [122], while technologies to transition ELISA assays into simplified lateral flowPOC test formats are well established and are being increasingly exploited. Although currentlycommercial serological tests are inaccurate for TB diagnosis, the detection of individual orcombinations of TB-specific antibodies, antigens and or immune markers using lateral flow assays ormicrofluidic technologies still seems most likely to provide a field-friendly POC tool [122, 129]. Inaddition, both platforms seem to be evolving toward simultaneous detection and diagnosis ofdifferent infectious disease. Finally, electronic nose technology allowing analysis of breathcondensates and the detection of distinct profiles of volatile organic compounds offers anotherpossibility for POC TB diagnosis [130, 131].
Unmet needs andresearch priorities
Social, environmental, host andpathogen-specific factors continueto create distinct diagnostic chal-lenges and settings (fig. 1), both atindividual patient and public healthlevels. No single test has yet met, orperhaps will ever meet, all diagnos-tic requirements across resource,healthcare and clinical settings.Integration of old and novel tech-nologies, and continued tailoring oftechnology to individual high- andlow-burden, local and national set-tings is essential to optimise TBdiagnosis. Table 4 highlights manyof the current unmet diagnosticneeds and research gaps. Ongoingbasic and clinical research, as well asincreased operational research, willbe required to address these gaps. Inparticular, research moving beyondthe simple assessment of diagnosticaccuracy towards impact evalua-tions of novel tools and integratedalgorithms for important patientand public health outcomes, such asmorbidity, mortality, case detection
lndividual country decision
Technicalpolicy recommendationGlobal level
Country level
Before policy During scale-upBefore scale-up
Programmaticpolicy recommendation
to adopt and scale up#
Test accuracy● Effectiveness● Test and resourceutilisation
●
Epidemiological impactchanges in TB and DR-TB case notifications, treatment delay,treatment outcomes
●
Economic impact●
Health system impact●
Cost of diagnostic process and treatment; patient costs
●
Operational data●
Surrogate patient-important outcomes
●
Ease of use●
Basic cost comparisons●
outcomes,case detection
Patient-important
practical constraintsresource requirements,infrastructural and human
Figure 6. Proposed new value chain for phased evaluation oftuberculosis (TB) diagnostics, from accuracy to impact assessment.Grey arrows: stages in the evaluation pathway; coloured boxes: policydecisions at the global level (red) and the country level (blue). In thestages before scale-up and during and after scale-up, evaluation datawould be collected on diagnostic algorithms incorporating the newtest. DR: drug-resistant. #: countries would adopt implementation atdifferent points and should provide feedback about their experiences.Reproduced from [132] with permission from the publisher.
13
7J.
G.
PE
TE
RE
TA
L.
rates and/or default rates, and hospital length of stay [132], are required to best develop and guidepolicy. Recently, COBELENS et al. [132] proposed a new phased evaluation pathway for TB diagnostics(fig. 6).
Conclusion
The armamentarium of diagnostics tests for TB has never been greater. Nevertheless, manydiagnostic challenges on both an individual patient level, such as for smear-negative or sputum-scarce TB, EPTB, TB/HIV co-infection and childhood TB, and on a larger public health levelremain suboptimally addressed.
In addition, it is becoming increasingly evident that simply developing new tests will beinsufficient to ensure successful scale-up and/or guarantee impact for either individual patients orthe global TB epidemic [132]. Countries vary in their ability to embrace and in their interest inadopting new technologies, and it is clear that the willingness of national TB control programmesand private sector clinicians to use and invest in new TB diagnostics is fundamental to thesuccessful widespread implementation of a novel technology. Even tools with excellent diagnosticaccuracy, such as Xpert1 MTB/RIF, may have little impact unless widely and appropriately used.
Thus, is it important that high TB-burden countries, especially the emerging economies (Brazil,Russia, India, China and South Africa), all with large TB and drug-resistance problems, drive theearly adoption and scale-up of new technologies as well as lead the next wave of TB diagnosticinnovation towards an affordable, simple POC test. In fact, only the combination of these effortswill allow advancements in TB diagnosis to significantly impact the global TB epidemic.
Support StatementJ.G. Peter is supported by Carnegie, NIH Fogarty and SAMA Fellowships. J.G. Peter, R.N. van Zyl-Smitand M. Pai are supported by EDCTP (TB-NEAT). M. Pai is supported by the Canadian Insti-tutes of Health Research, Grand Challenges Canada and Fonds de recherche du Quebec – Sante. R.N.van Zyl-Smit is supported in part by a USA CRDF fellowship. C.M. Denkinger is supported by aRichard Tomlinson Fellowship. These agencies had no involvement in this publication.
Statement of InterestM. Pai is a consultant for the Bill & Melinda Gates Foundation (BMGF). The BMGF had noinvolvement in this publication.
AcknowledgementsFigures were produced with the help of K. Adamson (Chocolate Cow Design, Cape Town, SouthAfrica).
References1. Whitelaw A, Peter J, Sohn H, et al. Comparative cost and performance of light-emitting diode microscopy in