Detection of lipoarabinomannan in urine and serum of ... - Tuberculosis

Download PDF
8 downloads 0 Views 2MB Size Report
Comment
Anita G. Amina, Prithwiraj Dea, John S. Spencera, Patrick J. Brennana, Joshua ... Maju Joeb, Yu Baib, Lars Laurentiusc, Marc D. Porterc,d,e, William J. Honnenf,.
Tuberculosis 111 (2018) 178–187

Contents lists available at ScienceDirect

Tuberculosis journal homepage: www.elsevier.com/locate/tube

Detection of lipoarabinomannan in urine and serum of HIV-positive and HIV-negative TB suspects using an improved capture-enzyme linked immuno absorbent assay and gas chromatography/mass spectrometry

T

Anita G. Amina, Prithwiraj Dea, John S. Spencera, Patrick J. Brennana, Joshua Dauma, Barbara G. Andrea, Maju Joeb, Yu Baib, Lars Laurentiusc, Marc D. Porterc,d,e, William J. Honnenf, Alok Choudharyf, Todd L. Lowaryb, Abraham Pinterf, Delphi Chatterjeea,∗ a Mycobacteria Research Laboratories, Department of Microbiology, Immunology and Pathology, Colorado State University, 1682 Campus Delivery, Fort Collins, CO, 80523, USA b Alberta Glycomics Centre and Department of Chemistry, University of Alberta, Edmonton, AB, T6G 2G2, Canada c Nano Institute of Utah, University of Utah, Salt Lake City, UT, 84112, USA d Department of Chemical Engineering, University of Utah, Salt Lake City, UT, 84112, USA e Department of Chemistry, University of Utah, Salt Lake City, UT, 84112, USA f Public Health Research Institute, New Jersey Medical School, Rutgers, The State University of New Jersey, Newark, NJ, USA

A R T I C LE I N FO

A B S T R A C T

Keywords: LAM TB HIV GC/MS Capture immunoassay LAM monoclonal antibodies

TB diagnosis and treatment monitoring in resource limited regions rely heavily on serial sputum smear microscopy and bacterial culture. These microbiological methods are time-consuming, expensive and lack adequate sensitivity. The WHO states that improved TB diagnosis and treatment is imperative to achieve an end to the TB epidemic by 2030. Commercially available lipoarabinomannan (LAM) detection tools perform at low sensitivity that are highly dependent on the underlying immunological status of the patient; those with advanced HIV infection perform well. In this study, we have applied two novel strategies towards the sensitive diagnosis of TB infection based on LAM: Capture ELISA to detect LAM in paired urine and serum samples using murine and human monoclonal antibodies, essentially relying on LAM as an ‘immuno-marker’; and, secondly, detection of αD-arabinofuranose and tuberculostearic acid (TBSA)- ‘chemical-markers’ unique to mycobacterial cell wall polysaccharides/lipoglycans by our recently developed gas chromatography/mass spectrometry (GC/MS) method. Blinded urine specimens, with microbiologically confirmed active pulmonary TB or non TB (HIV +/HIV–) were tested by the aforementioned assays. LAM in patient urine was detected in a concentration range of 3–28 ng/mL based on GC/MS detection of the two LAM-surrogates, D-arabinose and tuberculostearic acid (TBSA) correctly classifying TB status with sensitivity > 99% and specificity = 84%. The ELISA assay had high sensitivity (98%) and specificity (92%) and the results were in agreement with GC/MS analysis. Both tests performed well in their present form particularly for HIV-negative/TB-positive urine samples. Among the HIV +/TB+ samples, 52% were found to have > 10 ng/mL urinary LAM. The detected amounts of LAM present in the urine samples also appears to be associated with the gradation of the sputum smear, linking elevated LAM levels with higher mycobacterial burden (odds ratio = 1.08–1.43; p = 0.002). In this small set, ELISA was also applied to parallel serum samples confirming that serum could be an additional reservoir for developing a LAMbased immunoassay for diagnosis of TB.

1. Introduction A recent report by the World Health Organization (WHO) estimates

∗

there were 1.37 million deaths associated with tuberculosis (TB) infections in 2015 [1]. TB is treatable and while timely intervention often usually results in positive patient outcomes (83% success rate for drug-

Corresponding author. E-mail addresses: [email protected] (A.G. Amin), [email protected] (P. De), [email protected] (J.S. Spencer), [email protected] (P.J. Brennan), [email protected] (J. Daum), [email protected] (B.G. Andre), [email protected] (M. Joe), [email protected] (Y. Bai), [email protected] (L. Laurentius), [email protected] (M.D. Porter), [email protected] (W.J. Honnen), [email protected] (A. Choudhary), [email protected] (T.L. Lowary), [email protected] (A. Pinter), [email protected] (D. Chatterjee). https://doi.org/10.1016/j.tube.2018.06.004 Received 23 February 2018; Received in revised form 25 May 2018; Accepted 5 June 2018 1472-9792/ © 2018 Published by Elsevier Ltd.

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

LAM spiked in both urine and serum. We then examined a subset of 100 clinically characterized patient samples to determine the potential correlation between HIV co-infection and test performance. Almost all of the HIV-negative clinically positive TB samples were positive by ELISA. Moreover, our capture ELISA with specific antibodies appeared to be more sensitive than the commercially available TB Test and meets the WHO determined sensitivity (> 95%) and specificity (≥87%) criteria [32].

susceptible TB) [2], the early and accurate diagnosis of TB remains a significant barrier to reducing mortality. The conventional diagnosis of TB is by detecting Mycobacterium tuberculosis (Mtb) either through sputum smear microscopy or by culturing methods and the “gold standard” for TB diagnosis is repeat bacterial culture of sputum samples. While having a high clinical accuracy, this method has a turnaround time of a few weeks, requires a sophisticated laboratory infrastructure for effective implementation, often misses many presumptive extrapulmonary TB cases (which accounts for ∼10% of all TB cases) [3–5], and many patients, particularly children, are unable to produce sputum. New nucleic acid tests also have a high level of diagnostic accuracy, but high cost and limited ease of use continues to hinder their widespread usage in Point-Of-Care (POC) applications [6–8]. A number of recent studies have focused on the development of tests for the direct detection of primary antigenic markers of Mtb, in serum, urine, and other body fluids [9,10]. One marker in particular, mannose-capped lipoarabinomannan (ManLAM), is shed into the urine during active infection with pulmonary TB, and has been the subject of several investigations regarding its utility as a clinical biomarker for infection [11–18]. ManLAM, and its structural variants generally known as LAM are immunodominant multiglycosylated lipoglycans present in copious amounts in all mycobacteria [19–22]. While ELISA and lateral flow tests have been developed to detect LAM, their sensitivity is limited to approximately 1 ng/mL. These tests can detect LAM with high specificity in patients with pulmonary TB who are co-infected with HIV, but not in those who are HIV- negative [12,23–25]. The failure to detect LAM in the urine of HIV-negative patients limits the global utility of these assays, as the majority of TB patients are HIV-negative (85% of the 9.6 million patients worldwide) [26]. It has been argued that in HIV-negative patients with active TB, LAM cannot be detected because of its low abundance, possible formation of complexes with proteins present in the blood or urine or because LAM is shed into the urine of patients with active TB only when there is extrapulmonary renal tract involvement, such that the antigen can enter the urine directly from infected tissue as in the case of whole bacteria being shed due to colonization [26]. A recent study reported that using a novel pre-concentration of urine that LAM was present, albeit in low concentrations, in HIV-negative patients with active TB infections [27]. This suggests that an assay that utilizes antibodies with increased affinities for capture and detection of LAM in urine could improve the sensitivity of these assays. In our ongoing efforts to validate LAM as a TB-biomarker in clinical (serum and urine) samples from both HIV-positive and negative subjects with known sputum smear grades and culture results, we have shown that LAM can be detected using immunoassays once urine is treated with chaotropic reagents [28,29]. Moreover, recently we reported the detection of urinary LAM in patient samples in a concentration range of 3–40 ng/mL using GC/MS, correctly classifying TB status with a sensitivity > 99% and specificity = 84% [17]. The suboptimal quality and incomplete characterization of antiLAM antibodies available in the past has been a major obstacle to the development of sensitive, specific, robust diagnostic assays suitable for clinical use. Improvements in sensitivity are required to maintain assay performance as it is translated to an ‘easy-to-use’ lateral flow format, and to optimize LAM detection in patients with low antigen burden in serum and/or urine, so that a robust assay is applicable to a broad spectrum of TB patients (e.g. HIV-infected, HIV-non-infected, children, and those with extrapulmonary TB). In this study, we systematically evaluated the binding affinity of seven anti-LAM monoclonal antibodies (mAb) available to us at the Colorado State University (CSU) and from New Jersey Medical School (NJMS), Rutgers University. Based upon these results, we selected a pair that have the potential advantage of recognizing a large range of epitopes, and that are not dependent on the presence of any specific structural motifs in the arabinan component of LAM [30,31]. Using this pair of mAbs, we achieved a limit of detection (LOD) of 100 pg/mL of

2. Methods and materials 2.1. LAM from M. tuberculosis CDC 1551 for assay standardization The LAM used in the study was isolated and purified from Mycobacterium tuberculosis CDC1551 in vitro culture in our laboratory at CSU as described in Mycobacterial Protocols, IInd ED [33]. 2.2. LAM-epitope directed synthetic glycoconjugates A panel of twelve glycoconjugates of oligosaccharides reflective of the termini of LAM were synthesized as described previously [34]. 2.3. Generation of monoclonal antibodies for ELISA development 2.3.1. Mouse mAbs to LAM Mouse mAbs CS35 IgG3, 906.7 IgG3, 908.1 IgG3 and 906.41 IgG3 all raised against Mycobacterium leprae LAM [30,35] were isolated in our laboratory from the hybridoma cell lines generated by the fusion of myeloma cells with immunized mouse splenocytes. CS40 IgG1 mAb was generated against purified ManLAM from M. tuberculosis Erdman strain [31] as a hybridoma. Its variable VH and VL chain sequences were then cloned from these hybridoma cells and expressed as a mouse/human chimera (CS40hu) generating a IgG1 Fc domain. The same method was used to generate a chimera for CS35 as (CS35hu) also expressing a human IgG1 Fc domain. 2.3.2. Human mAb to LAM A novel human IgG1mAb, A194-01 was molecularly cloned from a patient diagnosed with pulmonary TB who had already started on drug treatment for a month before screening the culture supernatant against ManLAM in an ELISA assay using a high throughput in vitro B cell culture method [36]. VH and VL gene sequences were amplified using RT-PCR followed by nested PCR with different VH–VL specific primers and cloned into an IgG expression vector and the antibody was then expressed by transient transfection of 293 T cells [37–39]. 2.4. Clinical and molecular parameters in selecting sample cohort 2.4.1. Study cohort Anonymized archived urine and serum samples used in our study were provided by the Foundation for Innovative New Diagnostic (FIND, Geneva). The study samples were collected from patients with symptoms of pulmonary tuberculosis presenting prior to the initiation of treatment to clinics in Vietnam, South Africa and Peru. All human urine specimens were collected from adult participants of both sexes suspected of pulmonary TB, with and without HIV co-infection. Urine specimens were sedimented by centrifugation and the supernatant was stored at −80 °C within a few hours of collection. Final diagnosis (TB vs. non TB) was established on the basis of microscopy plus > 2 sputum cultures and clinical and radiologic examinations. TB was defined as being culture positive from at least one sample. Non TB was defined as being smear and culture negative on all samples and having improved clinically/radiologically without TB-specific therapy. Patients without a firm final diagnosis (e.g. contaminated culture, persistent symptoms despite repeated negative TB cultures, or treatment for TB without culture-confirmation) were excluded from study. 179

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

2.4.5. Capture ELISA The polystyrene microplate (Corning Costar) was coated with 100 μL of a capture antibody at 10 μg/mL concentration in PBS and incubated at 4 °C overnight. Urine and serum control samples were spiked with known amounts of LAM at different concentrations and incubated at 4 °C overnight to allow for the complexation of LAM and protein/s. Control and clinical samples were pretreated with Proteinase K and the supernatant used for ELISA. After overnight incubation, the antibody-coated plates and the LAM samples were brought to room temperature and the plates were blocked for 1 h. The plates were washed with the wash buffer (200 μL x 10) after which the control and the clinical samples were added to the appropriate wells (100 μL) and incubated for 2 h at room temperature. Following a second wash, the plates were incubated for 2 h at room temperature with the biotinylated detection antibody at a final concentration of 250 ng/mL in wash buffer. Biotinylation of the antibody was carried out using EZ-Link Sulfo NHS-LC Biotin (Thermofisher Scientific) following the kit protocol and the labelled antibody was desalted on Zeba spin desalting columns, 7 K MWCO (ThermoFisher Scientific) as per the kit protocol. Following a third wash, 100 μL of 1:200 dilution of Streptavidin–Horseradish Peroxidase (HRP) (R & D Systems) was added to the plates and incubated as per the kit protocol. After the final wash, 100 μL Ultra TMBELISA chromogenic substrate (ThermoFisher Scientific) was added to the plates and incubated for at least 30 min and observed for color development. The reaction was stopped by addition of sulphuric acid (Fisher Scientific) and the optical density was read at 450 nm. All the controls were run in triplicate and reported as the mean ± standard deviation. The samples were run in duplicate and plotted against the standard curve generated by spiking the urine and/or serum with known amounts of LAM and serially diluted. Limits of blank (LoB) and limits of detection (LoD) were generated from the standard curve using Clinical and Laboratory Standards Institute (CLSI) standard LoB = mean blank +1.645(SDblank); LoD = LoB + 1.645 (SD low concentration sample) [41].

2.4.2. Ethics statement Anonymized archived urine samples used in our study were provided by the Foundation for Innovative New Diagnostics (FIND, Geneva) http://www.finddiagnostics.org/programs/tb/find_activities/ tb_specimen_bank.html. The sample numbers presented here are arbitrary and cannot be traced back to individual patients. The study was approved by the local IRB and all subjects providing samples signed an informed consent at the time of enrollment and before sample collection. Participants were informed that the samples were going to be stored at FIND repository and will only be used for the development of new TB diagnostics. None of the authors have access to identifying patient information. The sample cohort was divided into four subgroups(25 TB-positive HIV-positive, 25 TB-positive HIV-negative, 25 TB- negative HIV-positive and 25 TB-negative HIV-negative). On receipt, the samples were aliquoted into 0.5 mL aliquots and stored frozen at −80 °C. Control samples were obtained from healthy volunteers from a TB non-endemic region; these were spiked with known amounts of LAM to derive an assay standard curve compared to the unspiked sample used as a negative control. This study conforms to the Declaration of Helsinki and was approved by the CSU Institutional Biosafety Committee (IBC) and Integrity and Compliance Review Board under approval human IRB protocol number 09-006B (Lipoarabinomannan Analysis in Urine and Serum) valid till 2019-03-06 and renewed annually. To prevent contamination, all samples were processed in the sterile biosafety cabinet by trained personnel. All the buffers and reagents used were sterile. Glassware for chemical derivatization were washed and baked at 550 °F (overnight) before use. 2.4.3. Pretreatment of samples Urine samples were thawed on ice. A 500 μL aliquot of urine sample and 250 μL aliquot of serum sample were transferred to centrifuge tubes. All urine and serum samples were pretreated with Proteinase K (ThermoFisher Scientific) at a final concentration of 200 μg/mL at 55 °C for 30 min followed by inactivation by boiling the samples at 100 °C for 10 min. This inactivation step also helped in denaturation of the cleaved protein fragments, allowing for easy removal by centrifugation at 12,000×g for 10 min. The supernatant from the centrifugation in case of urine samples was then concentrated five-fold using a vacuum concentrator (LabConco) for ELISA. For serum samples no concentration was applied. All spiked samples were run in triplicate and the clinical samples were run in duplicate in ELISA.

2.5. D-arabinose and tuberculostearic acid (TBSA) analysis by GC/MS All urine samples were separated by hydrophobic interaction chromatography (HIC) over Octyl Sepharose (OS)-CL 4B. The 40% and 65% n-propanol in 0.1 M NH4OAc eluents off the HIC column were processed for GC/MS analysis. For D-arabinose estimation, acid hydrolysis (2 M trifluoroacetic acid) was carried out to release D-arabinose in the first step. 2- O-octylarabinoside was prepared using acidified R-(2)-octanol which was then converted to 1-(α/β-O-(R)-2-octyl)- 2,3,5 triO- trifluoroacetyl-D-arabinofurano/pyranoside using trifluoroacetic anhydride in acetonitrile containing 10% trifluoroacetic acid. For Darabinose quantitation, D-13C5-UL-arabinose (200 ng) was used as an internal standard to compare to the diagnostic four peaks arising due to the formation of α/β anomers of the D-arabinopyranosyl and D-arabinofuranosyl ring conformers during derivatization. The amount of LAM-equivalent was calculated using the previously reported formulation (see equations in the Supplemental Files). The D-arabinose derivatives were then analyzed by GC/MS using MS/MS. The ions m/z 420.9 (parent ion) to 192.9 (daughter ion), and m/z 425.9 (parent ion) to 197.9 (daughter ion) were monitored for D-arabinose and D-UL-13C5arabinose (internal standard, Cambridge Isotope Laboratories Inc.) respectively, as reported earlier [17]. For TBSA, the purified LAM from urine was subjected to alkaline hydrolysis and subsequently the corresponding pentafluorobenzyl tuberculostearate derivative was made. D2-palmitic acid was used as the internal standard. The GC/MS analysis of the pentafluorobenzoate ester was carried out using selective ion monitoring in negative ion chemical ionization mode whereby the characteristic free fatty acyl anion at m/z 293.7 was monitored. A comparison between TBSA and D2-palmitic acid (internal standard; 20 ng; m/z 257.3) yields the TBSA content in the sample, which was then used to calculate the LAM-equivalent from

2.4.4. Development of immunoassays To optimize the concentration of the mAbs used, indirect ELISA (which measures binding of the antibody to the immobilized antigens) was carried out as previously described [40] with some modifications. Antigen solutions were prepared in phosphate buffered saline (PBS, pH 7.4) and applied to the 96-well plate (Corning Costar) and incubated at 4 °C overnight. Non-specific antibody binding sites were blocked with 1% Bovine Serum Albumin (BSA) (Sigma-Aldrich) in PBS containing 0.05% Tween-80 (Sigma-Aldrich) (blocking buffer) after washing the wells briefly with the same. Purified murine (CS35, 906.7, 908.1 and 906.41) and human (A194-01hu hu, CS35hu ms and CS40hu hu) antibodies were used at a concentration of 2 μg/mL (100 μL volume) and added to all the wells and incubated for 2 h at room temperature. The plates were washed and then incubated for 1.5 h with anti-mouse IgG alkaline phosphatase conjugate (Sigma) for all murine primary mAbs and anti-human IgG alkaline phosphatase conjugate (Sigma) for the human primary mAbs, diluted 1:2500 in wash buffer. The plates were again washed and the alkaline phosphatase activity measured by addition of 100 μL of p-nitrophenyl phosphate (pNPP) (Kirkegard and Perry Laboratories) as a substrate. The optical density was measured at 405 nm. All standards were run in duplicate and the absorbance plotted to determine the binding activity of the antibodies to the LAM or the LAM glycoconjugates. 180

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

the formulations in the Supplemental files. GC/MS analyses were carried out using a CP 3800 gas chromatograph (Varian) equipped with an MS320 mass spectrometer. Chromatograms with respective peaks were integrated manually (i.e., peak areas were defined manually and integrated areas were generated by the computer software) for the estimation of total D-arabinose and TBSA content.

motifs present in LAM and a broad range of selectivity among these Ara6/Ara4 conjugates. The focus on the known non-reducing termini of in vitro grown ManLAM/LAM was based on the fact that non-reducing end arabinan terminus of urinary LAM is not known. Six anti-LAM mAbs, namely, CS35 ms, 906.7 ms, A194-01hu and CS40hu, 908.1 ms and 906.41 ms were first screened by ELISA against the synthetic glycoconjugates. These six mAbs bind all the twelve synthetic Ara4 and Ara6 LAM conjugates with varied affinity and were independent of mannose capping. The mAb CS40hu was found to have high affinity towards the Man1Ara4 motif. The mAb 906.7 ms showed binding affinity only towards nonmannosylated Ara4/Ara6 (Fig. 1). A similar pattern of binding affinity was observed with two other mAbs, namely 906.41 ms and 908.1 ms (data not shown). Based on the broad range of binding to all of these structurally diverse synthetic glycoconjugates, CS35 ms was chosen as the capture antibody and A19401hu as the detection antibody, respectively, for the development of a capture ELISA. A checkerboard analysis was performed (See Supplemental File, Table SM-4) to determine the optimal concentration of the antibodies and assess cross reactivity.

2.6. Statistical analysis Samples in which TBSA or D-arabinose was detected were classified as having active TB infections (TB-positive). These classifications were tabulated against the reference of clinical diagnosis of TB-positive (sputum smear/culture positive) or TB-negative in a 2 × 2 matrix following Standards for Reporting Diagnostic Accuracy Studies (STARD) [42]. All data analyses were conducted using R open source software version 3.4.0 [43]. Sensitivity (proportion of TB-positive patients who were urinary LAM positive) and specificity (proportion of TB-negative patients who were urinary LAM negative) were evaluated using the R package caret version 6.0.76 [44]. The Receiver Operating Curve (ROC) was created using the ROCR package 1.0.7 [45] URL:http://rocr.bioinf. mpi.sb.mpg.de; optimal cutoff values and AUC were also calculated with this package. Figures were created using the ggplot2 package version 2.2.1 [46]. The relationship of LAM to gradation of sputum smear in samples with positive smear results was evaluated with an ordinal logistic regression model in which the response was the ordered level of smear results (from low to high: ‘scanty’/‘+’/‘1+’/‘2 + 3+’), using the R package MASS [47]. The relationship of LAM to HIV status was evaluated with a logistic regression model in base R. The computing platform was Windows 7 × 64 (build 7601) Service Pack 1.

3.3. Development of capture ELISA method CS35 ms mAb, when used as a capture along with biotinylated A194-01 mAb as a detection antibody gave the best limit of detection with ManLAM spiked in PBS at ∼100 pg/mL (Fig. 2) compared to other mAbs (CS35hu, CS40hu, 906.7 ms, 908.1 ms and 906.41 ms). However, when reversed i.e.; A194-01 mAb as the capture and biotinylated CS35 ms mAb as the detection antibody, the assay gave a very low absorbance value, indicating that this reverse pair capture/detection system was not optimal. At this point, we examined the dynamic range of mAb concentrations that could be used in the capture ELISA to have minimal cross reactivity responsible for background signal or falsepositivity in downstream assays. We tested six mAbs (CS35 ms, CS35hu, CS40hu, 906.7, 908.1, 906.41) as capture antibodies at a concentration of 10 μg/mL paired with biotinylated A194-01 as a detection antibody at 250 ng/mL, on control urine and serum spiked with various amounts of LAM. CS35ms paired with biotinylated A194-01 showed an LoD of ∼100 pg/mL for both urine and serum spiked samples in contrast to CS40hu and biotinylated A194-01 which had an LoD of ∼400 pg/mL. The remaining four pairs (CS35hu/A194-01; 906.7/A194-01; 908.1/ A194-01; 906.41/A194-01) showed an LoD of ∼1 ng/mL in both urine and serum spiked samples (Fig. 3) indicating that these capture/detection pairs were suboptimal.

3. Results and discussion 3.1. Rationale for use of D-Arabinose and TBSA as surrogates of LAM in biological fluids In the context of detecting LAM by GC/MS in biofluids such as urine, and TBSA were the chosen structural surrogates as the “fulllength” LAM is difficult to analyze due to its size (Fig. 1). These monomeric LAM units are rare and mycobacteria specific. The application of hydrophobic interaction chromatography, an integral part of our GC/MS protocol, ensures elimination of D-arabinose as a component of similar neutral polysaccharides derived from vegetables and other food [48]. D-arabinose

3.4. Urine pretreatment optimization for LAM ELISA 3.2. Structure of LAM: terminal arabinan affinity to mAbs The detection of LAM spiked in buffers (PBS, ammonium bicarbonate, carbonate/bicarbonate buffer) in a direct ELISA using any of the anti-LAM specific mAb in our collection was found to have a sensitivity of ∼100 pg/mL. It was, therefore, surprising to find that LAM spiked in urine was not detectable, even at 1 μg/well (Fig. 4). We initially suspected that the inhibition in signal was due to the acidity or high salt content of the urine samples or low molecular weight components in the urine samples nonspecifically adsorbing on the capture surface. However, dialysis of urine against PBS did not remove the inhibitory activity. Incubating urine with Proteinase K partially restored the response for the ELISA test. As a result, we hypothesized that protein/s in urine were responsible for this inhibition [52], as treatment with the Proteinase K partially restored the ability to detect LAM spiked in urine in a direct binding ELISA. Thus, a sample pretreatment with Proteinase K was incorporated prior to ELISA.

The arabinan domain confers immunogenicity to LAM. For LAMELISA, the non-reducing end arabinan motifs of LAM play crucial roles in Ab recognition. Years ago, these arabinan termini [49,50] were defined for in vitro grown Mtb LAM. It was established that non-reducing terminal segments of LAM consist of two major arrangements, a branched β-D-Araf-(1 → 2)-α-D-Araf-[β-D-Araf-(1 → 2)-α-D-Araf-(1 → 3)](1 → 5)-α-D-Araf- (1 → 5)- α-D-Araf (Ara6) and a linear β-D-Araf-(1 → 2)-α-D-Araf-(1 → 5)-α-D-Araf-(1 → 5)- α-D-Araf (Ara4) (Fig. 1). Moreover, in Mtb LAM there is frequent “capping” with short α (1 → 2) mannopyranosyl (Manp) chains [51]. Furthermore, one 5-methylthio-Dxylofuranose (MTX) unit, linked α (1 → 4) to the terminal Manp cap, is also found in Mtb LAM. Based on these data, a series of glycoconjugates, mapping the terminal end of LAM, were synthesized [34] (see Supplemental Files Fig. SM-1). These synthesized substrates were conjugated to BSA (8–15 Ara6/ Ara4 units per BSA molecule) using 3-amino-4-(8-hexa/tetraarabinofuranosyl octylamino)cyclobut-3-ene-1,2-dione as a linker, and used to screen our collection of anti-LAM mAbs (Supplemental Files Table SM-1) to determine the binding efficiency for the various glycan

3.5. Detection of LAM in clinical samples 3.5.1. Use of ELISA on urine samples Equipped with an efficient urine pretreatment protocol and a pair of 181

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

Fig. 1. Schematic representation of ManLAM structure incorporating symbolic nomenclature recommended by the Consortium for Functional Glycomics. The yellow shaded areas represent different non-reducing end arabinan epitopes present in in vitro grown LAM. The antiLAM mAbs CS35 (red), A194-01(light green), 906.7 (teal) and CS40hu (purple) had been shown according to their respective structural epitope selectivity. Weak affinity was defined as low O.D. values: LAM control > observed < 0.9. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

50 non TB (TB-negative) urine samples, only 4 (2 HIV-positive and 2 HIV-negative) were found to be LAM positive by ELISA. Classifying patients at or above that value as ‘TB-positive’ and below that value as ‘non-TB’ resulted in a sensitivity of 1.00, a specificity of 0.92, and overall accuracy of 0.96. A ROC curve shows the area under the curve (AUC) to be 0.98 (Fig. 5B).

mAbs (CS35 ms as capture and biotinylated A194-01hu as detection) with a broad range of LAM-structure selectivity, we tested a small cohort of 100 clinical samples by capture ELISA. An optimal cut-off absorbance value of 0.186 was applied to exclude any background signal as determined by simultaneous ELISA using non-endemic urine as true negative control. Patients at or above that value were classified as ‘TBpositive’ and below that value as ‘non TB’. Among the 50 TB-positive (25 each HIV-positive/HIV-negative) urine samples, LAM was detected in all 50 samples. Significantly, all the 25 HIV-negative/TB-positive samples (a group where LAM-detection immunoassays have constantly failed) and 25 HIV-positive/TB-positive samples gave signals above the background and were regarded as LAM positive (Fig. 5A). Among the

3.5.2. GC/MS analysis of D-Arabinose and tuberculostearic acid in urinary LAM For an alternate validation of the ELISA results, we analyzed the same 100 urine samples using an antibody independent approach. The 40% and 65% n-propanol in 0.1 M NH4OAc eluents off the HIC column Fig. 2. Capture ELISA to detect LAM in buffer. mAb CS35 ms was used as a capture antibody to detect LAM in PBS. Biotinylated A194-01hu was used as the detection antibody. The mAb pair showed LoD of ∼100 pg/mL. Conversely, CS35 ms when biotinylated and used as a detection antibody against A194-01 (as capture Ab) did not show the similar detection sensitivity.

182

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

formulation [17]. For estimation of TBSA by GC/MS, the HIC-purified urinary LAM was subjected to alkaline hydrolysis followed by the generation of the pentafluorobenzyl tuberculostearate ester [17]. This derivative was analyzed by GC/MS and the characteristic mass fragment ion at m/z 297.3 (arising from TBSA:C19H37O2−) was monitored using chemical ionization in the negative mode. A comparison between TBSA and D2palmitic acid (internal standard; 20 ng; m/z 257.3; C16H31O2-) yields the TBSA content in the sample, which was then used to calculate the LAM-equivalent using the formulations in Supplemental Files. D-Arabinose (a representative chromatogram Fig. 6B) was detected in 41 of the 50 (82%) TB-positive samples whereas all 50 samples had measurable levels of TBSA. These data (Tables SM-2 and SM-3 in the Supplemental Files) indicated that the quantities of urinary LAM determined by GC/MS using the two different surrogates were in strong agreement demonstrating the ability to detect LAM in HIV-negative/TBpositive samples. Based on the TBSA analysis (a representative chromatogram is shown in Fig. 6A and the chromatograms for representative 68 urine samples are presented in the Supplemental Files Figs. SM2-5), 52% (13/ 25) of the HIV-positive/TB-positive samples contained more than 10 ng/mL urinary LAM (Fig. 7A). Moreover, 79% (27/34) of TB-positive urine samples, which were classified as low TB smear grade (i.e., +Scanty,+, 1+), were found to have LAM levels ≤10 ng/mL (Fig. 7B). Although the small number of samples prohibits a robust statistical assessment, a preliminary ordinal logistic regression suggests the odds for higher TB smear grade is multiplied by 1.24 (95% CL = 1.08–1.44) for every unit increase in LAM (p = 0.002). Among the 50 TB-negative urine samples, 8 (2 HIV-positive and 6 HIV-negative) contained TBSA; of these, all but 2 of the HIV-negative samples were also D-arabinose positive. Overall, TBSA analysis had no false negatives and D-arabinose had a negative predictive value of 83%. The GC/MS analysis of urinary LAM surrogates (i.e. D-arabinose and TBSA) was undertaken to address three objectives. The first objective was to validate the ELISA results and quantify the LAM equivalent in clinical urine samples based on D-arabinose and on TBSA. Because the relative quantitation is based on LAM from Mtb grown in vitro, the comparable values of the LAM equivalents as estimated from TBSA and D-arabinose may be indicative of a structural similarity of the arabinan domain as well as presence of intact LAM molecule as in the known structure [35,49,53]. The second objective was to determine if there was a correlation between the quantitatively measured amounts of LAM and HIV co-infection. The third objective was to determine if the urinary LAM amounts correlates with the TB smear designation, which is considered as an indicator of the bacterial load in pulmonary TB patients [54]. The cumulative result from GC/MS points to a possible link between HIV co-infection and higher levels of urinary LAM, indeed, a logistic regression with TB-positive samples revealed an increase in

Fig. 3. Capture ELISA for LAM detection using specific mAbs humanized (hu), mouse (ms) as Capture Abs against the biotinylated A194-01 as a detection antibody. BSA was used as the ELISA negative control and LAM (ng/ml) spiked in (A) Urine and (B) Serum were analyzed at various (serially diluted two-fold) concentrations to obtain the LoD for each antibody pair. ELISA was done in triplicate and an average of each value, with the standard error removed, is represented above.

was processed for GC/MS analysis downstream. For D-arabinose estimation, 1-(α/β-O-(R)-2-octyl)-2,3,5 tri-O-trifluoroactyl-D-arabinofurano/pyranoside derivative was generated as described before [17]. For D-arabinose quantitation, D-13C5-UL-arabinose (200 ng) was used as an internal standard to compare to the diagnostic four peaks arising due to the formation of α/β anomers of the D-arabinopyranosyl and D-arabinofuranosyl ring conformers during derivatization. The amount of LAM-equivalent was calculated using the previously reported

Fig. 4. ELISA assay to detect spiked LAM in buffer or urine. (healthy control or FIND TB patient sample) using mAb CS35 ms specific for LAM. Untreated urine samples showed lack of signal which was restored by Proteinase K treatment. 183

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

Fig. 5. Urine LAM ELISA on TB- positive samples and related ROC curve: A) The O.D. values have been shown for HIV-positive/TB-positive (solid red) and HIV-negative/TB-positive (pattern red). B) The area under the curve (AUC) is 0.98. The results of the culture and smear are taken to represent the true status of TB infection. The optimal cutoff value for LAM in urine as measured by Capture ELISA (based on these 100 samples) is 0.186. Classifying patients at or above that value as 'TB-Positive' and below that value as ‘non TB’ results in sensitivity of 1.00, a specificity of 0.92 (95% CI), and overall accuracy of 0.96 (95% CI). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 6. Chromatograms showing presence and absence of LAM in urine using GC/MS method. A) Representative chromatogram for TBSA. B) Representative chromatogram for D-Arabinose. Supplemental Files showing GC/MS chromatograms of 68 samples.

status of the entire patient not just the lungs. This hypothesis is supported by the fact that 16% of urine samples clinically negative were found to have LAM [17,26,55] by GC/MS (and 8% by ELISA), which we suspect arises from the extra-pulmonary TB or renal TB in the case of HIV-positive patients. Moreover, infection with non-tuberculous mycobacteria (NTMs) is often encountered among hospitalized patients. Anti-LAM antibodies can cross react with NTM LAM. Notably, GC/MS assay can resolve the issue of any false positivity because it is based on the principle of chemical characterization unlike antibody based immunoassays. GC/MS analysis can only corroborate true LAM positive within the LoD when the sample is not contaminated. Thorough precautions were taken in sample handling to avoid any possible contamination. A side-by-side comparison of the three approaches for urine LAM detection for the 100 samples analyzed is presented in Fig. 8.

odds for HIV infection of 1.38 (95% CL = 1.16–1.70) for each ng/mL increase of LAM (p = 0.0007). Interestingly 27 out of 34 (79%) urine samples belonging to low smear grade (i.e., +Scanty, +, and 1+) were found to have ≤10 ng/mL of urinary LAM equivalent, which points to a possible correlation (Fig. 7B). These observations are in agreement with previous reports [55,56]. Six of the nine TB-positive urine samples that had undetectable levels of D-arabinose belong to the sputum smear gradation of +Scanty, 1+. These same samples also showed the presence of LAM equivalent at < 9 ng/mL as determined with the more sensitive TBSA assay. We feel that this disparity between D-arabinose and TBSA detected is due to the fact that quantification of D-arabinose requires four well-resolved peaks with specific retention times and these peaks must be present in the correct ratio. The smallest of the four peaks is sometimes below the LoD of GC/MS thus making the overall assignment negative. The four peaks from D-arabinose arise due to the mixture of stereoisomers formed during chiral octanolysis followed by trifluoroacetylation. We argue that in any non-sputum based LAM assay development, there should be a clear and reliable correlation in LAM quantification with an independent approach. Moreover, urinary LAM has the potential to represent both pulmonary and extra-pulmonary TB infection because this excreted material is more representative of the health

3.5.3. Feasibility of capture ELISA on serum Based on the results obtained from the urine samples we analyzed paired serum samples collected from the same patients. An optimal cutoff O.D. value of 0.7 was applied to exclude any background signal as determined by simultaneous ELISA using non-endemic serum as true negative control. Among the 50 TB-positive (HIV-positive/HIV-negative, 25 each) serum samples, LAM was detected in all 50 samples (25 184

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

Fig. 7. Distribution of TBSA in HIV-positive/TBpositive and HIV-negative/TB-positive in GC/MS. Each dot represents TBSA detected LAM equivalent by GC/MS in one urine sample. (A) Relative distribution of LAM equivalents among HIV-positive and HIV-negative TB-positive urine samples: 52% of HIV-positive/TB-positive (orange dots) have > 10 ng/mL LAM equivalent; 84% of HIV-negative/TBpositive (blue dots) have < 10 ng/mL LAM equivalent. Logistic regression revealed an odds increase of 1.38 for each unit increase of LAM amount (95% CL = 1.16–1.70) (p = 0.0007). (B) Relative distribution of LAM-equivalents in TB-positive urine samples with respect to the designated smear category: TB sputum smear grade: 2+,3+ (purple dots, 62%, > 10 ng/mL); +, 1+, Scanty+ (green dots, 76%, < 10 ng/mL). Logistic regression revealed an odds increase of higher smear grade (95% CL = 1.08–1.43) (p = 0.002). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

among HIV-positive or HIV-negative populations, we applied three independent strategies to validate the presence or absence of LAM. The GC/MS detection of LAM, which is independent of application of any antibodies, had a significantly higher sensitivity and specificity and conclusively showed that LAM is present in HIV-negative/TB- positive patients albeit in lower amounts than HIV-positive/TB-positive. The estimated LAM amounts in the urine samples is proportional to the smear gradation indicative of bacterial load in TB patients, and in accord with published reports. We also show that with the proper selection of mAbs, a capture ELISA can compare well with the GC/MS results

HIV-positive, 25 HIV-negative) (sensitivity = 1.00). Among the 50 non TB (TB-negative) serum samples, only 2 (1 HIV-positive and 1 HIVnegative) were also found to be LAM-positive by ELISA (specificity = 0.96) (Fig. 8). The ELISA test had an overall accuracy of 0.98. A ROC curve shows the AUC for the serum ELISA results to be 0.99 (Fig. 9). 4. Conclusion In a small set of clinical sample replicates representing TB or non-TB

Fig. 8. Capture ELISA using CS35 as the capture and biotinylated A194-01 IgG as the detection antibody. Urine clinical status and GC-MS designation of the samples for LAM presence or absence were taken as the standard for the ELISA. However, there were four clinically negative samples that were positive for LAM by ELISA. All the samples were pretreated with Proteinase K. GC/MS quantification of LAM in serum is under development. 185

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

Competing interests The authors declare that they have no competing interests. AP and AC are co-inventors on a patent describing the A194-01 antibody. Consent for publication Not applicable. Acknowledgement We gratefully acknowledge the Foundation for Innovative New Diagnostics (FIND) of Geneva, Switzerland for providing the clinically characterized patient urine and serum samples. Abbreviations WHO World Health Organization LAM lipoarabinomannan ManLAM mannose capped LAM TB Tuberculosis TB-positive TB sputum serial smear and culture positive; TB-negative sputum smear and culture negative; GC/MS Gas Chromatography/Mass Spectrometry TBSA tuberculostearic acid mAb monoclonal antibody LoD limit of detection LoB limit of blank ELISA enzyme linked immune absorbent assay CL confidence level

Fig. 9. ROC curve for LAM classification compared to culture results in serum. The blue dot represents the optimal cutoff value. The area under the ROC curve (AUC) is 0.99. The optimal cutoff for LAM in serum as measured by Capture ELISA (based on 100 samples) is 0.7. Classifying patients at or above that value as ‘TB-positive’ and below that value as ‘non TB’ results in sensitivity of 1.00, a specificity of 0.96 and overall accuracy of 0.98 (95% CI). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

giving us the confidence that a true PoC test can be developed. Both tests performed well in their present form including for HIV-negative/ TB-positive urine samples, which will target a majority of the TB infected population. This study establishes clearly that LAM is present in detectable amounts in HIV-negative TB patients and can be detected both by chemical and immunoassay methods. In this cohort, we also show that measuring serum LAM has great potential to be developed into a diagnostic tool. Although with limitations such as sample concentration was required to achieve the sensitivity, this will be the first report of analysis on paired urine and serum samples. We believe that the clinical accuracy of the assays is related to the fact that a better combination of antibodies was used. Future work will include introducing simpler method of sample pretreatment and determinations of the epitopes of urinary LAM and selecting more specific antibodies that bind to in vivo LAM.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx. doi.org/10.1016/j.tube.2018.06.004. References [1] WHO. Global tuberculosis report Geneva: WHO978 92 4 156539 4; 2016. Contract No.: WHO/HTM/TB/2016.13. [2] WHO. Multi-drug resistant tuberculosis (MDR-TB). 2013. [3] Perkins MD, Cunningham J. Facing the crisis: improving the diagnosis of tuberculosis in the HIV era. J Infect Dis 2007;196(Suppl 1):S15–27. http://dx.doi.org/10. 1086/518656. PubMed PMID: 17624822. [4] Norbis L, Alagna R, Tortoli E, Codecasa LR, Migliori GB, Cirillo DM. Challenges and perspectives in the diagnosis of extrapulmonary tuberculosis. Expert Rev Anti Infect Ther 2014;12(5):633–47. http://dx.doi.org/10.1586/14787210.2014.899900. PubMed PMID: 24717112. [5] Piccini P, Chiappini E, Tortoli E, de Martino M, Galli L. Clinical peculiarities of tuberculosis. BMC Infect Dis 2014;14(Suppl 1):S4. http://dx.doi.org/10.1186/ 1471-2334-14-S1-S4. PubMed PMID: 24564419; PubMed Central PMCID: PMCPMC4015485. [6] Theron G, Pooran A, Peter J, van Zyl-Smit R, Kumar Mishra H, Meldau R, et al. Do adjunct tuberculosis tests, when combined with Xpert MTB/RIF, improve accuracy and the cost of diagnosis in a resource-poor setting? Eur Respir J 2012;40(1):161–8. http://dx.doi.org/10.1183/09031936.00145511. PubMed PMID: 22075479; PubMed Central PMCID: PMCPMC5523948. [7] Wang S, Lifson MA, Inci F, Liang LG, Sheng YF, Demirci U. Advances in addressing technical challenges of point-of-care diagnostics in resource-limited settings. Expert Rev Mol Diagn 2016;16(4):449–59. http://dx.doi.org/10.1586/14737159.2016. 1142877. PubMed PMID: 26777725; PubMed Central PMCID: PMCPMC4943866. [8] McNerney R, Daley P. Towards a point-of-care test for active tuberculosis: obstacles and opportunities. Nat Rev Microbiol 2011;9(3):204–13. http://dx.doi.org/10. 1038/nrmicro2521. PubMed PMID: 21326275. [9] Reither K, Saathoff E, Jung J, Minja LT, Kroidl I, Saad E, et al. Low sensitivity of a urine LAM-ELISA in the diagnosis of pulmonary tuberculosis. BMC Infect Dis 2009;9:141. http://dx.doi.org/10.1186/1471-2334-9-141. Epub 2009/09/01 PubMed PMID: 19715562; PubMed Central PMCID: PMC2741465, doi: 1471-23349-141 [pii]. [10] Tucci P, Gonzalez-Sapienza G, Marin M. Pathogen-derived biomarkers for active tuberculosis diagnosis. Front Microbiol 2014;5:549. http://dx.doi.org/10.3389/ fmicb.2014.00549. PubMed PMID: 25368609; PubMed Central PMCID: PMCPMC4202705. [11] Chatterjee D, Khoo KH. Mycobacterial lipoarabinomannan: an extraordinary lipoheteroglycan with profound physiological effects. Glycobiology 1998;8(2):113–20. [12] Lawn SD. Point-of-care detection of lipoarabinomannan (LAM) in urine for

Author's contributions The work was conceived by DC, AGA, PD. Manuscript was written by AGA, PD and DC. Critical inputs were provided by DC, JSS, MDP, LL, PD and AGA. Statistical calculations were done by BA. GC/MS was performed by PD, ELISA was performed by AGA, all mouse antibodies were purified and characterized by AGA and JD (JD, summer intern). All the mouse mAbs were obtained from an in-house collection at CSU (PJB and JSS). Human A194-01 was isolated, characterized and provided by AC, WH and AP. The manuscript was critically reviewed by all the authors. TL, MJ and YB were solely responsible for synthesizing and providing the 12 LAM glycoconjugates.

Funding Bill and Melinda Gates Foundation (OPP#1039621 to DC) and (OPP#1039665 to AP) for generous funding for this work and the Department of MIP for supporting AGA. 186

Tuberculosis 111 (2018) 178–187

A.G. Amin et al.

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20] [21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33]

diagnosis of HIV-associated tuberculosis: a state of the art review. BMC Infect Dis 2012;12:103. http://dx.doi.org/10.1186/1471-2334-12-103. PubMed PMID: 22536883; PubMed Central PMCID: PMC3423001. Hamasur B, Bruchfeld J, van Helden P, Kallenius G, Svenson S. A sensitive urinary lipoarabinomannan test for tuberculosis. PLoS One 2015;10(4):e0123457http://dx. doi.org/10.1371/journal.pone.0123457. PubMed PMID: 25905641; PubMed Central PMCID: PMCPMC4408114. Drain PK, Losina E, Coleman SM, Giddy J, Ross D, Katz JN, et al. Rapid urine lipoarabinomannan assay as a clinic-based screening test for active tuberculosis at HIV diagnosis. BMC Pulm Med 2016;16(1):147. http://dx.doi.org/10.1186/ s12890-016-0316-z. PubMed PMID: 27842535; PubMed Central PMCID: PMCPMC5109839. Drain PK, Rousseau C. Point-of-care diagnostics: extending the laboratory network to reach the last mile. Curr Opin HIV AIDS 2017;12(2):175–81. http://dx.doi.org/ 10.1097/COH.0000000000000351. PubMed PMID: 28079591; PubMed Central PMCID: PMCPMC5287256. Kerkhoff AD, Wood R, Vogt M, Lawn SD. Prognostic value of a quantitative analysis of lipoarabinomannan in urine from patients with HIV-associated tuberculosis. PLoS One 2014;9(7):e103285http://dx.doi.org/10.1371/journal.pone.0103285. PubMed PMID: 25075867; PubMed Central PMCID: PMC4116167. De P, Amin AG, Valli E, Perkins MD, McNeil M, Chatterjee D. Estimation of Darabinose by gas chromatography/mass spectrometry as surrogate for mycobacterial lipoarabinomannan in human urine. PLoS One 2015;10(12):e0144088http://dx.doi.org/10.1371/journal.pone.0144088. PubMed PMID: 26633829; PubMed Central PMCID: PMCPMC4669150. Gupta-Wright A, Peters JA, Flach C, Lawn SD. Detection of lipoarabinomannan (LAM) in urine is an independent predictor of mortality risk in patients receiving treatment for HIV-associated tuberculosis in sub-Saharan Africa: a systematic review and meta-analysis. BMC Med 2016;14:53. http://dx.doi.org/10.1186/s12916016-0603-9. PubMed PMID: 27007773; PubMed Central PMCID: PMCPMC4804532. Khoo K-H, Tang J-B, Chatterjee D. Variation in mannose-capped terminal arabinan motifs of lipoarabinomannans from clinical isolates of Mycobacterium tuberculosis and Mycobacterium avium complex. J Biol Chem 2001;276(6):3863–71. Venisse A, Fournié J-J, Puzo G. Mannosylated lipoarabinomannan interacts with phagocytes. EJB 1995;231:440–7. Chan J, Fan X, Hunter SW, Brennan PJ, Bloom BR. Lipoarabinomannan, a possible virulence factor involved in persistence of Mycobacterium tuberculosis within macrophages. Infect Immun 1991;59. 1775-61. Juffermans NP, Verbon A, Belisle JT, Hill PJ, Speelman P, van DS, et al. Mycobacterial lipoarabinomannan induces an inflammatory response in the mouse lung. A role for interleukin-1 [In Process Citation]. Am J Respir Crit Care Med 2000;162(2 Pt 1):486–9. Hanifa Y, Telisinghe L, Fielding KL, Malden JL, Churchyard GJ, Grant AD, et al. The diagnostic accuracy of urine lipoarabinomannan test for tuberculosis screening in a South African correctional facility. PLoS One 2015;10(5):e0127956http://dx.doi. org/10.1371/journal.pone.0127956. PubMed PMID: 26010840; PubMed Central PMCID: PMCPMC4444353. Lawn SD, Dheda K, Kerkhoff AD, Peter JG, Dorman S, Boehme CC, et al. Determine TB-LAM lateral flow urine antigen assay for HIV-associated tuberculosis: recommendations on the design and reporting of clinical studies. BMC Infect Dis 2013;13:407. http://dx.doi.org/10.1186/1471-2334-13-407. PubMed PMID: 24004840; PubMed Central PMCID: PMC3846798. Achkar JM, Lawn SD, Moosa MY, Wright CA, Kasprowicz VO. Adjunctive tests for diagnosis of tuberculosis: serology, ELISPOT for site-specific lymphocytes, urinary lipoarabinomannan, string test, and fine needle aspiration. J Infect Dis 2011;204(Suppl 4):S1130–41. http://dx.doi.org/10.1093/infdis/jir450. PubMed PMID: 21996695; PubMed Central PMCID: PMC3192548. Lawn SD, Gupta-Wright A. Detection of lipoarabinomannan (LAM) in urine is indicative of disseminated TB with renal involvement in patients living with HIV and advanced immunodeficiency: evidence and implications. Trans R Soc Trop Med Hyg 2016;110(3):180–5. http://dx.doi.org/10.1093/trstmh/trw008. PubMed PMID: 26884498; PubMed Central PMCID: PMCPMC4755427. Paris L, Magni R, Zaidi F, Araujo R, Saini N, Harpole M, et al. Urine lipoarabinomannan glycan in HIV-negative patients with pulmonary tuberculosis correlates with disease severity. Sci Transl Med 2017;9(420). http://dx.doi.org/10. 1126/scitranslmed.aal2807. PubMed PMID: 29237757. Laurentius LB, Crawford AC, Mulvihill TS, Granger JH, Robinson R, Spencer JS, et al. Importance of specimen pretreatment for the low-level detection of mycobacterial lipoarabinomannan in human serum. Analyst 2016;142(1):177–85. http://dx.doi.org/10.1039/C6AN02109C. Crawford AL L, Mulvihill T, Granger J, Spencer J, Chatterjee D, Hanson K, Porter MD. Detection of the tuberculosis antigenic marker mannose-capped lipoarabinomannan in pretreated serum by surface enhanced Raman scattering. Analyst 2016;142(1):186–96. http://dx.doi.org/10.1039/c6an02110g PubMed Central PMCID: PMCPMID: 27924983 , Accessed date: 9 December 2016. Gaylord H, Brennan PJ, Young DB, Buchanan TM. Most Mycobacterium leprae carbohydrate-reactive monoclonal antibodies are directed to lipoarabinomannan. Infect Immun 1987;55:2860–3. Rivoire B, Ranchoff B, Chatterjee D, Gaylord H, Tsang A, Kolk AHJ, et al. Generation of monoclonal antibodies to the specific sugar epitopes of Mycobacterium avium complex serovars. Infect Immun 1989;57:3147–58. WHO. High-priority target product profi les for new tuberculosis diagnostics: report of a consensus meeting Geneva, Switzerland 2014 28-29 April Report No.: Contract No.: WHO/HTM/TB/2014.18. Shi L, Torrelles JB, Chatterjee D. Lipoglycans of Mycobacterium tuberculosis: isolation, purification, and characterization. second ed.Brown TPaACeditor.

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41] [42]

[43] [44] [45] [46] [47] [48]

[49]

[50]

[51]

[52]

[53]

[54]

[55]

[56]

187

Mycobacteria protocols, vol. 465. Barts and The London, London, UK: Humana Press; 2009. p. 23–45. Zheng RB, Jegouzo SAF, Joe M, Bai Y, Tran HA, Shen K, et al. Insights into interactions of mycobacteria with the host innate immune system from a novel array of synthetic mycobacterial glycans. ACS Chem Biol 2017;12(12):2990–3002. http:// dx.doi.org/10.1021/acschembio.7b00797. PubMed PMID: 29048873; PubMed Central PMCID: PMCPMC5735379. Hunter SW, Gaylord H, Brennan PJ. Structure and antigenicity of the phosphorylated lipopolysaccharide antigens from the leprosy and tubercle bacilli. J Biol Chem 1986;261:12345–51. Choudhary A, Patel D, Honnen W, Lai Z, Prattipati RS, Zheng RB, et al. Characterization of the antigenic heterogeneity of lipoarabinomannan, the major surface glycolipid of Mycobacterium tuberculosis, and complexity of antibody specificities toward this antigen. J Immunol 2018;200(9):3053–66. http://dx.doi. org/10.4049/jimmunol.1701673. PubMed PMID: 29610143; PubMed Central PMCID: PMCPMC5911930. Lou XM, Maarschalk E, O'Connell RM, Wang P, Yang L, D. B. Engineering human hematopoietic stem/progenitor cells to produce a broadly neutralizing anti-HIV antibody after in vitro maturation to human B lymphocytes. Blood 2009;113(7):1422–31. Huang J, Doria-Rose NA, Longo NS, Laub L, Lin CL, Turk E, et al. Isolation of human monoclonal antibodies from peripheral blood B cells. Nat Protoc 2013;8(10):1907–15. http://dx.doi.org/10.1038/nprot.2013.117. PubMed PMID: 24030440; PubMed Central PMCID: PMCPMC4844175. Scheid JF, Mouquet H, Feldhahn N, Walker BD, Pereyra F, Cutrell E, et al. A method for identification of HIV gp140 binding memory B cells in human blood. J Immunol Meth 2009;343(2):65–7. http://dx.doi.org/10.1016/j.jim.2008.11.012. PubMed PMID: 19100741; PubMed Central PMCID: PMCPMC2754789. Britton WJ, Hellqvist L, Basten A, Raison RL. Mycobacterium leprae antigens involved in human immune responses. I. Identification of four antigens by monoclonal antibodies. J Immunol 1985;135(6):4171–7. PubMed PMID: 2415603. Armbruster DA, Pry T. Limit of blank, limit of detection and limit of quantitation. Clin Biochem Rev 2008;29(Supplement 1):S49–52. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig L, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. Bmj 2015;351:h5527. http://dx.doi.org/10.1136/bmj.h5527. PubMed PMID: 26511519; PubMed Central PMCID: PMCPMC4623764. R Core Team. R. A language and environment for statistical computing. R Foundation for Statistical Computing. Computing RFfS. 2017. Vienna, Austria. Kuhn M. R Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation of Statistical Computing; 2017. Sing T, Sander O, Beerenwinkel N, Langauer T. ROCR : visualizing classifier performance in R. Bioinformatics 2005;21(20):7881. Wickham H. ggplot2: elegant graphics for data analysis. New York: Springer-Verlag; 2009. Venables WN, Ripley BD. Modern applied statistics with S. fourth ed. ed. New York: Springer; 2002. Lindqvist L, Jansson PE. Determination of the absolute configuration of sugar residues using gas chromatography. Method with potential for elimination of references. J Chromatogr A 1997;769(2):253–61. PubMed PMID: 9188183. Chatterjee D, Hunter SW, McNeil M, Brennan PJ. Lipoarabinomannan. Multiglycosylated form of the mycobacterial mannosylphophatidylinositols. J Biol Chem 1992;267:6228–33. Chatterjee D, Khoo K-H, McNeil MR, Dell A, Morris HR, Brennan PJ. Structural definition of the non-reducing termini of mannose-capped LAM from Mycobacterium tuberculosis through selective enzymatic degradation and fast atom bombardmentmass spectrometry. Glycobiology 1993;3:497–506. Kaur D, Obregon-Henao A, Pham H, Chatterjee D, Brennan PJ, Jackson M. Lipoarabinomannan of Mycobacterium: mannose capping by a multifunctional terminal mannosyltransferase. Proc Natl Acad Sci U S A 2008;105(46):17973–7. Epub 2008/11/14. doi: 0807761105 [pii], 10.1073/pnas.0807761105. PubMed PMID: 19004785. Sakamuri RM, Price DN, Lee M, Cho SN, Barry 3rd CE, Via LE, et al. Association of lipoarabinomannan with high density lipoprotein in blood: implications for diagnostics. Tuberculosis (Edinb) 2013;93(3):301–7. http://dx.doi.org/10.1016/j.tube. 2013.02.015. PubMed PMID: 23507184; PubMed Central PMCID: PMCPMC3807251. Torrelles JB, Sieling PA, Zhang N, Keen MA, McNeil MR, Belisle JT, et al. Isolation of a distinct Mycobacterium tuberculosis mannose-capped lipoarabinomannan isoform responsible for recognition by CD1b-restricted T cells. Glycobiology 2012;22(8):1118–27. http://dx.doi.org/10.1093/glycob/cws078. PubMed PMID: 22534567; PubMed Central PMCID: PMC3382347. Olaru ID, Heyckendorf J, Grossmann S, Lange C. Time to culture positivity and sputum smear microscopy during tuberculosis therapy. PLoS One 2014;9(8):e106075http://dx.doi.org/10.1371/journal.pone.0106075. PubMed PMID: 25171337; PubMed Central PMCID: PMCPMC4149502. Cox JA, Lukande RL, Kalungi S, Van Marck E, Van de Vijver K, Kambugu A, et al. Is urinary lipoarabinomannan the result of renal Tuberculosis? Assessment of the renal histology in an autopsy cohort of ugandan HIV-infected adults. PLoS One 2015;10(4):e0123323http://dx.doi.org/10.1371/journal.pone.0123323. PubMed PMID: 25897661; PubMed Central PMCID: PMCPMC4405591. Shah M, Variava E, Holmes CB, Coppin A, Golub JE, McCallum J, et al. Diagnostic accuracy of a urine lipoarabinomannan test for tuberculosis in hospitalized patients in a high HIV prevalence setting. J Acquir Immune Defic Syndr 2009. http://dx.doi. org/10.1097/QAI.0b013e3181b98430 PubMed PMID: 19692904 , Accessed date: 21 August 2009.