Quantitative lateral flow strip assays as User-Friendly Tools To Detect Biomarker Profiles For Leprosy

Leprosy is a debilitating, infectious disease caused by Mycobacterium leprae. Despite the availability of multidrug therapy, transmission is unremitting. Thus, early identification of M. leprae infection is essential to reduce transmission. The immune response to M. leprae is determined by host genetics, resulting in paucibacillary (PB) and multibacillary (MB) leprosy associated with dominant cellular or humoral immunity, respectively. This spectral pathology of leprosy compels detection of immunity to M. leprae to be based on multiple, diverse biomarkers. In this study we have applied quantitative user friendly lateral flow assays (LFAs) for four immune markers (anti-PGL-I antibodies, IL-10, CCL4 and IP-10) for whole blood samples from a longitudinal BCG vaccination field-trial in Bangladesh. Different biomarker profiles, in contrast to single markers, distinguished M. leprae infected from non-infected test groups, patients from household contacts (HHC) and endemic controls (EC), or MB from PB patients. The test protocol presented in this study merging detection of innate, adaptive cellular as well as humoral immunity, thus provides a convenient tool to measure specific biomarker profiles for M. leprae infection and leprosy utilizing a field-friendly technology.

levels of pro-inflammatory cytokines and develop localized granulomatous disease with bacilli scarcely detectable in their lesions. At the opposite pole of the spectrum are LL patients who predominantly generate Th2 and anti-inflammatory cytokines such as interleukin-10 (IL-10) resulting in disseminating, progressive infections 15 . In between these two opposite poles of the leprosy spectrum, borderline states of leprosy [borderline tuberculoid (BT), borderline (BB) and borderline lepromatous (BL)] are positioned. Due to the diverse disease spectrum, detection of M. leprae infection in diagnostic tests requires multiple, diverse biomarkers specific for both cellular and humoral mediated immunity. In previous studies we have shown that IFN-γ -inducible protein 10 (IP- 10) in response to a M. leprae-specific antigen (ML2478) correlates with M. leprae exposure and thereby the risk of infection and its subsequent transmission 16 . Additionally, we demonstrated that chemokine (C-C motif) ligand 4 (CCL4), a component of the innate immunity, can be used to identify pathogenic immunity against M. leprae since it was increased in patients, partly in household contacts but not in endemic controls 16 . IL-10, on the other hand, is associated with suppression of Th1 cells in leprosy [17][18][19] . Moreover, most lepromatous patients with high bacillary loads produce antibodies against the M. leprae specific phenolic glycolipid I (PGL-I) 20,21 , which are hardly detected in PB 22 . Hence, sensitive tests that can simultaneously quantitate multiple analytes in one sample provide apt tools to characterize different clinical leprosy types. In particular, tests based on multicomponent host biomarker profiles that can identify M. leprae infected individuals (yet) without clinical symptoms of leprosy, will be useful for guidance of prophylactic treatment, thereby contributing to reduction of M. leprae transmission as well as prevention of disabilities.
Inherent to the situation in leprosy endemic areas is the absence of sophisticated laboratories. It is therefore imperative that new diagnostic tests are facilitated for application in the field. Up-converting phoshor lateral flow assays (UCP-LFAs) have previously shown to be robust, low-complexity assays, representing a field-friendly alternative for common laboratory-based ELISAs 23,24 , applicable for detection of multiple pathogens including food-borne pathogenic strains and potential biowarfare/bioterrorism agents [25][26][27] . Field evaluation of UCP-LFAs for detection of IL-10, IP-10, CCL4 and anti-PGL-I IgM demonstrated high correlation with ELISAs using samples from cohorts of limited numbers of leprosy-or TB patients 28,29 .
In the current study UCP-LFAs were applied to a more extensive (five-fold) sample size compared to our previous studies, derived from a randomized BCG vaccination field trial in Bangladesh 30 . Six test groups were included: MB patients, PB patients, healthy household contacts (HHC), HHC vaccinated with Bacillus Calmette-Guérin (BCG) (HHC&BCG), HHC who developed leprosy after BCG vaccination (new cases; NC) and endemic controls (EC) from the same area without known contact with leprosy patients. This extended cohort study allowed exploratory identification of biomarker profiles for M. leprae infection, leprosy disease per se, the type of leprosy and BCG vaccination as determined with UCP-LFAs for the above indicated targets.

Results
Performance of the UCP-LFA versus ELISA. Whole blood samples (n = 726) from all individuals were analysed using ELISA, as well as the field-friendly UCP-LFAs for IL-10, IP-10, CCL4 and anti-PGL-I antibodies. Comparison of UCP-LFA and ELISA results demonstrated significant correlation for all four biomarkers (p < 0.0001), confirming earlier observations 28,29 .
The diagnostic performance of the UCP-LFA in comparison to ELISA was further assessed through AUCs for the two most distinct phenotypes: MB patients (n = 34) and EC (n = 51). IL-10, IP-10 and CCL4 levels were determined in Nil, WCS and Mlep samples, as well as anti-PGL-I IgM levels (Fig. 1). The IL-10 and IP-10 UCP-LFAs outperformed the corresponding ELISAs, whereas the CCL4 and anti-PGL-I IgM tests performed equally. For discrimination of MB patients from EC, the proposed diagnostic field-tool UCP-LFA provides an equally well or even better alternative for the conventional ELISAs.
M. leprae specific responses based on single analyte UCP-LFA measurements. In order to put the quantitative test results obtained with the four single UCP-LFAs in the context of their biomarker potential, we assessed each analyte/stimulus combination by comparing median group levels. As indicated by the AUCs in Fig. 1, MB patients can be distinguished from EC based on IP-10 and CCL4 (irrespective of stimulus), anti-PGL-I IgM and IL-10 WCS, therefore also showing significantly different median levels (Fig. 2, Table 1). Moreover, median levels of anti-PGL-I IgM and IP-10 Nil differed between MB patients and (BCG-vaccinated) HHC, whereas median levels of IP-10 WCS and CCL4 WCS only distinguished the non-vaccinated HHC from MB patients. BCG vaccination therefore affects the immune response in HHC, as reflected by the significant difference in IP-10 WCS levels between HHC and HHC&BCG (p = 0.018) (Fig. 2b). Furthermore, IL-10 WCS levels differed between HHC and EC, while median levels of IP-10 WCS and CCL4 WCS differed between HHC&BCG and EC as well. PB patients and EC showed significantly different median CCL4 Nil and CCL4 WCS levels (Fig. 2c), as well as borderline significant different levels for IL-10 WCS (p = 0.07) and IP-10 WCS (p = 0.06).
The ability of each analyte/stimulus combination to distinguish between two groups is summarized in Table 1, thereby reviewing the biomarker potential of the four individual host immune markers. Remarkably, the levels of IP-10 (p < 0, 0001), IL-10 (p = 0.003) and CCL4 (p < 0.0001) in WCS stimulated samples were more significantly different for MB and EC than anti-PGL-I IgM levels (p = 0.0042). Moreover, anti-PGL-I IgM levels could not be used to discriminate PB patients or (BCG-vaccinated) HHC from EC, which clearly demonstrates the added value of IP-10, IL-10 and CCL4 in leprosy diagnostics.
Biomarker signatures to specify M. leprae infection, leprosy or disease classification. Diagnostic tests that allow detection of M. leprae infection, leprosy per se and leprosy classification would be of great benefit to the general healthcare in leprosy endemic areas. The four host immune markers allowed distinction between two groups (Table 1). However, to distinguish M. leprae infected from non-infected individuals or patients from healthy contacts, we compared host immune markers for multiple groups (Fig. 3).  First, in order to combine immune markers into multicomponent host biomarker profiles, positive UCP-LFA results for each analyte/stimulus combination were collectively specified ( Fig. 3; Supplementary Table S3). Second, analyte/stimulus combinations were selected such that they optimally distinguished individuals with a specified disease-or infection state (Fig. 3), considering all HHC as M. leprae infected. This resulted in three specific profiles:  Table S4A).

MB vs HHC MB vs HHC&BCG MB vs EC PB vs EC HHC vs EC HHC&BCG vs EC HHC&BCG vs HHC
anti-PGL-I IgM   Table S3) per analyte/stimulus combination the percentage of individuals with a positive test result per group is shown. Based on these data the optimal analyte/stimulus combination to differentiate either infected from non-infected groups, patients and non-patients groups or MB and PB patients were selected to construct the potential profiles described.
(II) To detect leprosy patients from healthy, though possibly M. leprae infected individuals, CCL4 WCS and IP-10 WCS were selected as immune markers since these single tests were more frequently positive in patients (MB and PB) compared to contacts (HCC and HCC&BCG) and are therefore associated with pathogenic immunity to M. leprae. The combination of CCL4 WCS and IP-10 WCS indeed demonstrated a positive test result more often in patients than in HHC or EC (Fig. 4b), whereas the related AUCs were ≥ 0,66 thus confirming leprosy disease-specificity (Supplementary Table S4B). (III) For classification of leprosy a signature consisting of anti-PGL-I IgM, IL-10 WCS and IP-10 Nil was applied, as each of these markers individually showed more positive test results in MB patients compared to PB patients (Fig. 3). This profile proved to be specific for MB patients (Fig. 4c) and thereby allowed the differentiation of MB and PB patients (AUC = 0.73, Supplementary Table S4C).
Ideally only one multicomponent host biomarker profile for diagnosis of M. leprae infection, leprosy per se and leprosy classification would be more suitable for field-use. In this exploratory study, a 4 marker profile of IL-10 WCS , IP-10 Mlep , CCL4 WCS and anti-PGL-I IgM was selected for this purpose, enabling distinction of infected and non-infected individuals by IL-10 WCS , IP-10 Mlep and CCL4 WCS , MB and PB patients from HHC and EC by CCL4 WCS and MB from PB patients by anti-PGL-I IgM and IL-10 WCS (Fig. 4d; Supplementary Table  S4D). However, to distinguish MB from PB patients or PB patients from HHC profile III for leprosy classification showed a higher AUC compared to the 4 marker profile (0.73 vs. 0.65 and 0.66 vs. 0.62 respectively, Supplementary Table S4). These data indicate the importance of distinct phase-specific profiles, the application of which will depend on the nature of the diagnosis to be made.
Nonetheless, application of the 4 marker profile demonstrated the influence of multicomponent host biomarker profiles on test accuracy, showing increased AUCs compared to individual markers (Supplementary Figure S1). The added value of using various analytes indicates the potential of multicomponent host biomarker profiles for leprosy diagnostics to detect M. leprae infection, leprosy disease or disease classification.

Discussion
The obvious incessant transmission of M. leprae has brought about increased focus in leprosy research on discovery of biomarkers to improve diagnosis. Nevertheless, thus far only few biomarkers for leprosy are recommended by expert panels 11 . Consequently, there is a growing need for new and sensitive diagnostic tools based on specific biomarkers which should, ideally, allow straightforward translation into field-friendly tests.
In this exploratory study, we aimed to provide several multicomponent host immune-biomarker profiles which distinguish between distinct stages of M. leprae infection. In this process we also emphasized the challenges that need to be tackled to allow application of these biomarkers in the field. As high-tech laboratories are often lacking in leprosy endemic areas, we examined the diagnostic potential of earlier developed field-friendly UCP-LFAs for detection of anti-PGL-I IgM antibodies and cyto/chemokines IP-10, IL-10 and CCL4 24,28,31 , in an extensive cohort in Bangladesh.
We demonstrated the biomarker potential of IP-10, IL-10, CCL4 and anti-PGL-I IgM measured by UCP-LFAs in whole blood, either in response to M. leprae specific stimuli or without stimulus. Moreover, multicomponent host biomarker profiles including selected analyte/stimulus combinations could indicate M. leprae infection, leprosy per se or be used for classification of leprosy subtypes. A biomarker profile of IP-10 Mlep , CCL4 WCS and IL-10 WCS was highly indicative of M. leprae infection, consistent with our previous finding that the IP-10 response to M. leprae specific proteins indicates exposure to M. leprae 16,32 .
Leprosy per se, on the other hand, was indicated by CCL4 WCS and IP-10 WCS , showing the potential to identify pathogenic immunity against M. leprae and confirming earlier observations on CCL4 16 . As current diagnostic assays for leprosy are antibody-based and only facilitate the diagnosis of MB cases [33][34][35] , inclusion of the host immune markers CCL4 and IP-10 in the profile shows promise for diagnosis of PB patients and indicates the importance of measuring cellular markers simultaneously with humoral markers.
For leprosy classification, the combination of anti-PGL-I IgM, IL-10 WCS and IP-10 Nil was indicative for MB patients, enabling the distinction between MB and PB patients. Although IL-10 and particularly anti-PGL-I IgM have been identified as characteristic markers for MB leprosy 36,37 , we also identified IP-10 as a, seemingly counterintuitive, host immune marker for patients at this side of the spectrum who usually display decreased pro-inflammatory immunity. However, since T-cells are not the exclusive source of IP-10 38 , IP-10 may still be produced in MB patients by monocytes and neutrophils 39 , as described for HIV-infected TB patients 40 .
To detect M. leprae infection, leprosy per se, as well as leprosy classification simultaneously with only one biomarker profile, IL-10 WCS , IP-10 Mlep , CCL4 WCS and anti-PGL-I IgM demonstrated the most optimal 4 marker profile performance. However, it performed less optimal for the distinct stages of M. leprae infection than the phase-specific profiles. Other cyto-/chemokines to identify pathogenic immunity to M. leprae (e.g. MCP-1 and IL-1β 16 ), leprosy classification (e.g. CCL17 and CCL18 41 ) or general mycobacterial infection (EN-RAGE 42,43 ) could therefore be included to achieve more optimal diagnostic accuracy 44 as distinct phase-specific profiles. In a multiplex UCP-LFA format multicomponent host immune biomarker profiles can be measured in one single test. This format therefore provides a field-friendly diagnostic tool, facilitating the diagnosis of leprosy based on biomarker signatures.
Of note is the observation that CCL4 levels in response to M. leprae WCS were elevated for HHC who received BCG vaccination compared to those who did not. Thus, BCG vaccination may also cause increased pro-inflammatory immune responses which renders contacts more prone to development of over-reactive, pathogenic immunity to M. leprae. Indeed, in a recent vaccination study an unexpectedly high proportion of HHC presented with PB leprosy after BCG vaccination supporting this idea 45 . In this respect, this vaccination study also shows the importance of immunomonitoring individuals at high risk to identify and treat patients at an early stage. In addition, since BCG vaccination or boost is a well-accepted prophylaxis against leprosy in contacts of newly diagnosed patients 46 , it is relevant to distinguish BCG-induced immunity in healthy contacts from early stage leprosy in these individuals. To efficiently monitor contacts for this purpose, the different stages of infection and disease of leprosy should be covered in diagnostics tools. Through simultaneous measurement of all analytes of interest on a single lateral flow strip, this format allows assessment of multicomponent host biomarker profiles using a unique field-friendly technology 24,29,31 . Thereby, the UCP-LFA format not only provides diagnostic tools for leprosy but similarly holds promise for TB diagnosis 28 and immunomonitoring of other chronic diseases 31 .

Materials and Methods
Study participants. Participants were recruited on a voluntary basis between January 2013 and December 2014 in leprosy endemic areas in Bangladesh as described previously 30 . Leprosy was diagnosed based on clinical, bacteriological and histological observations and classified by skin smears according to Ridley and Jopling 14 . Clinical and demographic data was collected in a database. Participants were classified into six test groups; MB patients, PB patients, HHC, HHC&BCG, NC and EC. Control individuals from the same leprosy endemic area (EC) were examined for the absence of clinical signs and symptoms of leprosy and TB; staff of leprosy-or TB clinics were excluded.
Test group selection. A randomized sample selection was taken from 1110 participants 30 . Individuals were randomly assigned for sample inclusion using the RAND formula (Excel 2010), aiming for a 50/50 male/female ratio and a 1:1:1 ratio of three age groups: 0-14, 15-29, and 30+ (Supplementary Table S1 Whole blood assay (WBA). Upon recruitment venous, heparinized blood (4 ml) was used directly in whole blood assays (WBA), using microtubes pre-coated with M. leprae whole cell sonicate (designated WCS), ML2478/ ML0840 recombinant proteins (designated Mlep) 16 or without antigen stimulus (designated Nil) 30 . After 24 hour incubation at 37 °C materials were frozen at − 20 °C, shipped on dry ice to the LUMC and stored at − 80 °C until analysis by ELISA or UCP-LFA 24 . UCP-LFA for IL-10, IP-10 and CCL4. UCP-LFAs for CCL4, IL-10 and IP-10 were prepared and performed as described previously 24,28,29 . The same antibody pairs as used for ELISAs were applied, with the non-biotinylated variant of the detection antibodies (non-biotinylated CCL4: AF-271-NA). Briefly, mixtures of 100 ng cytokine-specific UCP reporter conjugate and diluted serum sample (1:4 for IL-10, 1:30 for IP-10 and 1:300 for CCL4) were incubated for 60 min on a thermoshaker at 37 °C and 900 rpm. The mixture was then applied to cytokine-specific LF strips (containing a Test line with an antibody complementary to the antibody on the UCP particles) and immunochromatography was allowed to continue until strips were dry. LF strips were scanned in a Packard FluoroCount microtiterplate reader adapted for measurement of the UCP label (980 nm IR excitation, 550 nm emission). Results are displayed as the ratio value between Test and Flow-Control signal based on relative fluorescence units (RFUs) measured at the respective lines 48 . Ratio values were translated to concentration based on standard curves for each immunemarker. Lower limit of detection was 32 pg/ml for IL-10 and 316 pg/ml for IP-10 and CCL4.

PGL-I and
To determine test positivity, similar wholeblood samples from a set of healthy, non-endemic control individuals (NEC) were analysed and UCP-LFA thresholds were calculated based on the average value of all NEC samples (Supplementary Table S3).

UCP-LFA for anti-PGL-I antibody.
For detection of anti-PGL-I IgM antibodies, the same protocol as used for cytokine detection was applied utilizing 100-fold diluted serum and IgM-specific UCP conjugate (UCP αIgM ). Only unstimulated samples were analysed as the level of antibody levels does not change upon antigen stimulation. The threshold for positivity of 0.29 was determined by computing receiver operating characteristic (ROC) curves.
Ethics. This study was performed according to the Helsinki Declaration as described previously 30  Participants were informed about the study-objectives, the samples and their right to refuse to take part or withdraw from the study without consequences for their treatment. Written informed consent was obtained before enrolment. All patients received treatment according to national guidelines. ROC curves. Graphpad Prism version 6.02 for Windows (GraphPad Software, San Diego CA, USA) was used to plot ROC curves and calculate the area under curve (AUC); for IP-10, IL-10, and CCL4 the concentrations (pg/ml) were applied, whereas for anti-PGL-I IgM the OD 450 corrected for background (ELISA) and ratio value (UCP-LFAs) was used.
Statistical analysis. Differences in cytokine or antibody levels between test groups, as determined with UCP-LFA, were analysed with the One-way ANOVA for non-parametric distribution (Kruskall-Wallis) and Dunn's correction for multiple testing using GraphPad Prism. For IP-10, IL-10 and CCL4 the concentrations (pg/ml) and for anti-PGL-I IgM the UCP-LFA ratio values were utilized. The statistical significance level used was p ≤ 0.05.