RR:C19 Evidence Scale rating by reviewer:
Potentially informative. The main claims made are not strongly justified by the methods and data, but may yield some insight. The results and conclusions of the study may resemble those from the hypothetical ideal study, but there is substantial room for doubt. Decision-makers should consider this evidence only with a thorough understanding of its weaknesses, alongside other evidence and theory. Decision-makers should not consider this actionable, unless the weaknesses are clearly understood and there is other theory and evidence to further support it.
***************************************
Review:
There is a recognized global need for improved non-sputum TB
diagnostics. LAM detection in urine (uLAM) offers potential as a non-
sputum test; however, a widely available commercial uLAM diagnostic
lacks sensitivity and is recommended for use in TB suspects with
advanced HIV-immune suppression only. Recent studies have explored use
of high affinity antibodies against LAM epitopes coupled with increased
sensitivity of detection and have demonstrated ability to measure uLAM
down to low femtomolar concentrations.
This study used two LAM epitope antibodies S/4-20 and F/28 as capture
antibodies together with a detection antibody A194-01 in a sensitive
electrochemiluminescence (ECL) assay achieving lower limits of detection
of uLAM between 10pg/ml and 63 pg/ml. The S/4-20 antibody binds to the
to a 5-methylthioxylosyl-mannan (MTX-Man) motif which is present in the
pathogenic mycobacteria M.tb and M.bovis whereas F/28 binds to tetra-
arabinofuranoside (Ara4) present in a wide spectrum of mycobacteria. The
detection antibody A194-01 binds to both Ara4 and hexa-arabinofuranosee
(Ara6) which is not limited to mycobacteria.
Many prior high sensitivity studies of uLAM detection have been small and
case-controlled, whereas the current study utilized a large pragmatically
selected cohort of 745 HIV-uninfected TB suspects. The study explored the
diagnostic utility of the urinary LAM assay for identification of GeneXpert
Ultra or MGIT culture confirmed TB cases from among those attending
outpatient and inpatient TB clinics in a Hanoi hospital. TB diagnosis was
confirmed in 341 of TB suspects (46%) with only 6 confirmed with extra-
pulmonary disease.
The performance of the ECL assay utilizing this combination of capture
and detection antibodies in this “real-world clinical setting” did not reach
the levels reported from smaller earlier case-controlled studies. Clinical
sensitivity was also below the WHO minimum target product profile (TPP)
of 80% for sputum-negative TB and 95% for sputum-positive TB. S/A
uLAM testing identified 132 (38.7%) and F/A 140 (41%) of 341 MRS-positive TB cases with false positive rates of 3.7% and 22.3% respectively.
Receiver operator curve of sensitivity plotted against 1-specificity gave
AUC values for the ECL-S/A combination of antibodies of 74.4% and 75.5%
for identification of PCR or culture-proven pulmonary and extrapulmonary
TB disease. The ROC for this antibody combination in earlier small case-
controlled studies was reported at a higher figure of 95% (CI 87%-100%).
Despite a higher sensitivity the ECL-F/A antibody combination, the AUCs
for the F/A antibody combination were 62.8% and 53.9% for proven
pulmonary and extrapulmonary TB disease. Interestingly the ROC curves
of the F/A antibody assay were not dissimilar from earlier presented case-
controlled study data.
Lack of sensitivity, specificity and concordance severely limits the
development potential of these assays for accurate identification of
culture confirmed pulmonary and extra-pulmonary tuberculosis cases fin
TB suspects in Hanoi, Vietnam. The technical complexity of the assays will
also limit their use in combination with routine assays to improve their
discriminating power. However, the study did demonstrate the presence of
femtomolar concentrations of LAM in this HIV-uninfected TB cohort, a level
of detection approximately 50-fold lower than the current commercial
lateral flow assay. Surprisingly, the marked increase in test sensitivity of
the S/A TB specific capture assay enabled identification of femtomolar
levels of TB specific uLAM in both PCR and culture-negative TB suspects, a
finding requiring confirmation in other studies.
The poorer performance of the ECL assays with the chosen antibodies
might be attributed to either clinical, specimen handling or laboratory
assay factors. The study was performed in a pragmatically identified
cohort of pulmonary and extrapulmonary TB suspects that was recruited
solely from a Hanoi hospital. Earlier studies have indicated that there may
be geographic variability in uLAM assay performance raising a hypothesis
that variability of LAM structure may be TB strain related. The presence of
host derived neutralizing antibodies in urine might require antibody
antigen dissociation by urine heating which was not performed. It has also
been noted that glycosuria has impacts on LAM detection, which was not
monitored in this study. However, it is unlikely that unrecognized diabetes
reached a level to seriously impact LAM determination in this TB suspect
cohort. The urine samples were frozen and the bench time prior to
freezing was not recorded, both factors that could negatively impact test
performance. The very poor concordance between the two antibody tests
is of major concern. The F/A antibody binds to Ara4 and Ara6 epitopes that
are present in a wider spectrum of mycobacteria, including slow growers.
It has been proposed that biofilms containing environmental mycobacteria
can become a source of laboratory contamination. In summary this paper
demonstrated the presence of femtomolar concentrations of uLAM in a
cohort of Vietnamese HIV-uninfected TB suspects. The sensitivity of
detection with S/A and F/a antibody combinations in a ECL sandwich assay
failed to reach the criteria required for the WHO TPP in a large clinical population representing a typical clinical scenario in which such testing
will be required.
Creative Commons Attribution 4.0 International License (CC-BY 4.0)
