Evaluation of Immunodiagnostic Tests for Leprosy in Brazil, China and Ethiopia

Leprosy remains persistently endemic in several low- or middle income countries. Transmission is still ongoing as indicated by the unabated rate of leprosy new case detection, illustrating the insufficiency of current prevention methods. Therefore, low-complexity tools suitable for large scale screening efforts to specifically detect M. leprae infection and diagnose disease are required. Previously, we showed that combined detection of cellular and humoral markers, using field-friendly lateral flow assays (LFAs), increased diagnostic potential for detecting leprosy in Bangladesh compared to antibody serology alone. In the current study we assessed the diagnostic performance of similar LFAs in three other geographical settings in Asia, Africa and South-America with different leprosy endemicity. Levels of anti-PGL-I IgM antibody (humoral immunity), IP-10, CCL4 and CRP (cellular immunity) were measured in blood collected from leprosy patients, household contacts and healthy controls from each area. Combined detection of these biomarkers significantly improved the diagnostic potential, particularly for paucibacillary leprosy in all three regions, in line with data obtained in Bangladesh. These data hold promise for the use of low-complexity, multibiomarker LFAs as universal tools for more accurate detection of M. leprae infection and different phenotypes of clinical leprosy.

In 2016, 59% of the new cases worldwide were diagnosed with MB leprosy with ratios of MB and PB patients varying per endemic region 11 . Since PB cases are generally not detected using serological tools for anti M. leprae antibody detection, additional biomarkers are needed to identify the remaining 41% of PB leprosy patients. Moreover, as not all PGL-I seropositive individuals will develop disease, new diagnostic tests should be based on disease-and infection specific biomarkers allowing the distinction between individuals requiring therapeutic or prophylactic therapy, respectively. Tests based on signatures combining humoral-and cellular biomarkers may help to guide administration of postexposure prophylaxis (PEP), a currently introduced strategy aimed at reduction of transmission by M. leprae infected individuals without clinical symptoms of leprosy 12,13 .
In Bangladesh we previously demonstrated, using a field-friendly lateral flow assay (LFA) 7,14,15 , that combined detection of a humoral immune-marker (M. leprae PGL-I specific IgM antibodies) with additional cellular immune-markers (IP-10, CCL4 and IL-10) significantly improved distinction between M. leprae infected and non-infected individuals 7 . In this setting the BI of most leprosy patients was less than 1 which generally corresponds with the absence of anti-PGL-I antibodies 6,16 . The detection of additional cellular markers increased the sensitivity of the assay for these individuals with 39% compared to the LFA based on antibody detection alone 7 .
In the current study, whole blood samples of leprosy patients, their household contacts (HHC) and endemic controls (EC) were collected in Asia, Africa and South-America to evaluate the diagnostic potential of the previously used LFAs in Bangladesh applying detection of IP-10, CCL4 and PGL-I specific antibodies 7 . Additionally, a new UCP-LFA for detection of C-reactive protein (CRP) was developed and evaluated in these cohorts as CRP, an acute phase protein produced by the liver in response to inflammation, is elevated in LL/BL leprosy patients 17 and active tuberculosis (TB) 18,19 .
The Asian cohort originated from Guizhou, the province with the second highest leprosy prevalence in China, a country with overall low leprosy endemicity 20 . Patients originated from the Qianxinan and the Guiyang prefecture, with a prevalence of 0.085/10,000 and 0.011/10,000, respectively and mostly MB patients (MB/PB ratio: 8.2). The South American cohort was recruited in the state of Pará, Brazil, a region hyperendemic for leprosy with an annual new case detection rate of 35.34 per 10,000 with an MB to PB patient ratio of 1.932. Active transmission is ongoing in this area, evidenced by the high number of children amongst new cases (6.4% in 2013) 21,22 . The African cohort was collected in Kokosa Woreda (Oromia region) in Ethiopia with a prevalence of 0.32/10,000 and 5.9-fold more MB than PB patients 23 . Thus, this study describes the evaluation of multiple UCP-LFAs for leprosy in low (China), medium (Ethiopia) and hyperendemic (Brazil) settings.

Results
Patient cohorts. The extent of humoral and cellular immune responses against M. leprae differs within the leprosy spectrum 10 , ranging from predominantly humoral in MB patients to largely cellular immune responses in PB patients. The reported ratio of MB/PB patients as well as the level of endemicity differ between the three regions where the study cohorts were recruited (Table 1).
LL/BL, on the MB side of the leprosy spectrum, was the major form of leprosy observed in China and Ethiopia (respectively 76% and 71%), whereas in the Brazilian cohort the different forms of leprosy were more equally divided (Table 1). In China and Ethiopia the majority of the patients were male, while in Brazil a more even distribution of males and females was observed (Table 1).
Diagnostic value of single UCP-LFAs. Supernatants of WBAs (both M. leprae antigen-stimulated, WCS and Mlep, and non-stimulated (Nil)) were used to assess the significance of 4 single UCP-LFAs. The diagnostic potential of each marker (anti-PGL-I IgM, CRP Nil , IP-10 Nil , IP-10 Mlep , IP-10 WCS , CCL4 Nil , CCL4 Mlep , CCL4 WCS ) was evaluated based on Ridley-Jopling classifications by assessing the ability of the markers to discriminate LL/BL and BT/TT patients from HHC and EC (Supplementary Table S1). The diagnostic value of each marker was assessed by computing ROC curves, and the associated AUCs (Fig. 1). In the hyperendemic Brazilian cohort ( Supplementary Fig. S1), anti-PGL-I IgM, CRP and IP-10 (Nil, WCS and Mlep) levels differed significantly between LL/BL patients versus HHC and EC as indicated by the significant AUC values (Fig. 1). Similarly, CCL4 Nil levels for LL/BL patients and EC in Brazil were significantly different ( Fig. 1), while a good distinction between BT/TT patients and EC was provided by IP-10 Nil and IP-10 WCS.
In the Chinese cohort (supplementary Fig. S2), anti-PGL-I IgM and IP-10 (Nil, WCS and Mlep) levels significantly differed in LL/BL patients from HHC and EC (Fig. 1). Additionally, CRP levels distinguished LL/BL patients from HHC. BT/TT patients significantly differed from HHC and EC in anti-PGL-I IgM levels and IP-10 Nil . Moreover, IP-10 WCS , IP-10 Mlep, CCL4 Nil and CCL4 WCS differentiated BT/TT patients from EC.
In the Ethiopian cohort ( Supplementary Fig. S3), anti-PGL-I IgM levels distinguished LL/BL patients from HHC and EC, whereas CRP levels reached significance between these patients and EC. CCL4 WCS and CCL4 Mlep discriminated BT/TT patients from HHC and EC.
Overall, the data show that IP-10 was the most significant cellular marker to identify both LL/BL and BT/TT leprosy patients in low-and high endemic populations, anti-PGL-I IgM and CRP are relevant for diagnosis of LL/ BL patients and CCL4 contributes to the detection of BT/TT patients.

Identification of multi-biomarker signatures.
To identify a biomarker signature specific for leprosy disease in general, we included in this signature besides anti-PGL-I IgM also cellular markers based on the AUCs (Fig. 1).
In the Chinese cohort both IP-10 and CCL4 enabled the distinction between BT/TT patients and EC, with the highest AUC for a single analyte for IP-10 Nil and CCL4 Nil (Fig. 1). In Brazil, IP-10 Nil and IP-10 WCS discriminated BT/TT patients from EC, whereas in Ethiopia CCL4 WCS and CCL4 Mlep showed diagnostic value for these patients. Optimal cut-offs of the selected biomarkers were determined based on the Youden's index 24 . All individuals positive for both selected cellular markers were designated positive ( Supplementary Fig. S4). The use of multi-biomarker signatures consisting of cellular markers and humoral anti-PGL-I IgM seropositivity resulted in four possible outcomes depicted in Fig. 2. With a sensitivity for LL/BL patients of 91%, 97% and 75% in China, Brazil and Ethiopia respectively, the majority of LL/BL patients was identified by the PGL-I UCP-LFA with little added value of the cellular markers identifying 2%, 3% or 5% additional patients respectively in line with the immune responses within the leprosy spectrum.  On the other hand, for BT/TT patients the combination of cellular and humoral markers increased the test sensitivity with 50% to 54%, resulting in an overall sensitivity for BT/TT leprosy of 80% (China), 71% (Brazil) and 75% (Ethiopia). Similar analysis was applied to previously described data from a cohort in Bangladesh 7 , additionally detecting 39% of BT/TT patients resulting in an overall sensitivity of 63%. Importantly, specificity was not relevantly affected by the inclusion of cellular biomarkers in China and Ethiopia and was only moderately decreased in Brazil and Bangladesh. Of the HHCs, 10%, 13%, 15% and 49% were positive for cellular markers in the Chinese, Ethiopian, Bangladeshi and Brazilian cohorts, respectively. These data show that biomarker profiles based on humoral and cellular markers can identify patients at both ends of the leprosy spectrum, irrespective of geographical region.
Decision tree as a field-tool to assess leprosy risk profiles. For detection of leprosy patients and/ or individuals at risk of developing leprosy in field situations, a decision tree based on low-complexity diagnostic tests may facilitate decision making by local health workers. As biomarker based diagnostic tests should be globally applicable, we constructed a decision tree irrespective of cohort applying anti-PGL-I IgM and IP-10 as general biomarkers for leprosy patients. As anti-PGL-I IgM levels measured by UCP-LFA were highly specific for M. leprae infection and can be considered a correlate of risk for developing leprosy, this was designated the first step of the decision tree. This step identified 132 individuals (based on the cut-offs described in Fig. 2), among whom 104 leprosy patients (57% of total patients in the 3 cohorts) and 28 individuals without leprosy (5%) (Fig. 3).
In the next step, IP-10 Nil levels were used for anti-PGL-I IgM seronegative individuals, which detected an additional 59 patients thereby identifying 89% of all leprosy cases. Of the individuals without clinical leprosy 230 (43%) were IP-10 Nil positive. Subsequent assessment of the IP-10 levels in response to M. leprae WCS in the IP-10 Nil positive individuals increased specificity; while 81% of the patients were identified this step reduced the number of individuals without leprosy that were identified as being at risk to 137 (26%; Supplementary Fig. S5A). Stratifying these data by cohort showed that based on this three-step decision model of the individuals without leprosy 38% in Brazil (n = 131; 46 HHC; 74 EC), 22% (n = 11; 7 HHC; 4 EC) in Ethiopia and 16% (n = 23; 16 HHC; 7 EC) in China are considered at risk of developing leprosy or transmitting M. leprae (Supplementary Fig. S5B). Thus, use of a decision tree based on multiple, different types of markers measured in low-complexity assays can therefore guide decision making on who needs (prophylactic) treatment in large-scale screening efforts in field settings.

Discussion
Diagnostic tests for M. leprae infection will provide a useful asset in large scale screening efforts to identify individuals who need prophylactic treatment. In a previous study we demonstrated that the combination of field-friendly UCP-LFAs for leprosy detecting cellular and humoral biomarkers as compared to humoral markers alone, increased the sensitivity for detection of patients in Bangladesh by 32% for MB and 36% for PB cases 7 . The current study investigates the use of similar UCP-LFAs in three other cohorts with different leprosy endemicity in order to evaluate their worldwide applicability as field-friendly, point-of-care tests for leprosy based on combined multiplex detection of biomarker profiles.
In line with previous studies 4,6-8 , anti-PGL-I IgM levels identified the majority of LL/BL patients with high bacillary loads. The detection of cellular markers (IP-10 and CCL4) further improved the sensitivity of BT/TT leprosy patients up to 54%. Elevated levels of CRP in LL/BL (MB) leprosy were in agreement with other reported studies 17 . Cellular markers thus increased sensitivity of leprosy diagnosis and in particular showed added value for BT/TT patients in the three cohorts compared to assessment of anti-PGL-I IgM seropositivity alone. BT/TT patients at the PB side of the leprosy spectrum worldwide comprise 41% of the leprosy patients, signifying the relevance of cellular markers in leprosy diagnostics 11 . Especially with respect to detection of M. leprae infection in the absence of clinical symptoms, in particular relevant for the HHC group, cellular markers also diagnosed M. leprae infection more often than anti-PGL-I IgM testing.
The low rate (6%) of anti-PGL-I IgM positivity in the Brazilian HHC contrasted earlier observations from the same region 22 that indicated a remarkably high 77.6% seropositivity in a comparable student population. This difference is thought to be a consequence of improved assay-specificity of the UCP-LFA used in the current study compared to the earlier applied ELISA 6 : the UCP-LFA format is a virtually background-free reporter technology 25 thereby detecting M. leprae infection with higher specificity.
The rate of positivity for cellular markers in HHC correlated with the level of endemicity ranging from 10% in China (low endemic) to 53% in Brazil (high endemic), which is in agreement with previous findings 26 . In Brazil intense transmission continues in the area of this study as revealed by particularly high rates of leprosy cases amongst children 27 . Moreover, it has been reported that in the majority (19 out of 27) of Brazilian states 50% of the individuals is exposed to high or hyperendemic rates of infection 28 . Thus, the Brazilian EC group tested is therefore not unaffected by M. leprae, as these school children are likely to have been in contact with M. leprae infected individuals 22 . However, using quantitative signals as measured by UCP-LFA in field-settings, cut-off values are adjustable. This facilitates a stepwise approach that can be accommodated for various diagnostic questions (postexposure prophylaxis, monitoring, classification) each with different sensitivity/specificity requirements.
Factors to be considered for the appropriate applicability of UCP-LFAs based on combined biomarker profiles thus are regional differences in the MB/PB ratio and the level of endemicity. Cellular markers clearly represent valuable diagnostic tools in countries with high percentages of PB patients (i.e. Bangladesh) 11 . The level of cellular markers is more frequently elevated in HHC in regions with high leprosy endemicity as the rate of M. leprae infection corresponds with high new case detection 26 . Consequently, HHC resemble BT/TT patients with respect to positivity for certain cellular biomarkers. In this regard, the use of a multi-step decision approach, using initial categorization based on anti-PGL-I IgM seropositivity, followed by additional steps based on a cellular biomarkers can identify more sensitively and specifically those at risk of developing leprosy or transmitting bacteria.
The UCP-LFA format applied in this study facilitates rapid testing based on the presence of selected biomarkers as it is compatible with the use of finger stick blood (FSB). UCP-LFAs could thus serve as a rapid FSB screening test for M. leprae infection applicable as triage in large scale screening of HHCs aiming to provide PEP to infected individuals, to reduce transmission by infected but non-symptomatic individuals 12,13 . For leprosy diagnosis, on the other hand, a subsequent test using overnight stimulation with M. leprae antigens similar to the Quantiferon TB test 29 can be applied to increase test specificity for leprosy avoiding unnecessary use of antibiotics.
Longitudinal studies sampling contacts of leprosy patients during yearly follow-up are ongoing. This approach will include intra-individual biomarker comparison of individuals before and at diagnosis of clinical leprosy aimed at identification of biomarker signatures specific for early disease in individuals yet lacking symptoms 7,30 . The biomarker profile investigated in the current study indicates a high level of similarity of the immunological response of BT/TT and HHC based on 4 biomarkers. Longitudinal studies will provide biomarker signatures that can be used as correlates of disease in infected individuals before clinical manifestation of leprosy.
In summary, despite minor differences in biomarker specificity due to levels of leprosy endemicity, this study demonstrates that UCP-LFA rapid tests are well suited for diagnosis of leprosy patients and M. leprae infected individuals irrespective of geographical region. Multiplex UCP-LFAs will enable the assessment of biomarker signatures in leprosy endemic areas, which can facilitate guidance of prophylactic treatment within large-scale screening efforts to reduce transmission and disease while limiting administration of antibiotics 13 .

Materials and Methods
Study cohorts. Leprosy patients were diagnosed based on histological findings and clinical observations determined by experienced leprologists and a leprosy pathologist as previously described 4,31 . Patients were categorized according to WHO classification (MB/PB) and Ridley-Jopling classification and bacterial indices (BI) were determined. All leprosy patient whole blood was collected at initial diagnosis prior to multidrug therapy (MDT). . 50 ul of diluted sample was added to microtiter plate wells and mixed with 250 ng of target-specific UCP conjugate (IP-10, CCL4 and CRP) before target-specific LF strips were placed in the corresponding wells. Immunochromatography was allowed to continue for at least 30 min until dry. LF strip analysis. LF strips were scanned locally using portable LF strip readers adapted for measurement of the UCP label (ESEQuant LFR reader with 980 nm excitation and 550 nm emission; QIAGEN Lake Constance GmbH, Stockach Germany). LF strips were shipped to the LUMC and re-analysed using a UCP dedicated benchtop reader (UPCON; Labrox, Finland). Results are displayed as the ratio value between Test and Flow-Control signal based on relative fluorescence units (RFUs) measured at the respective lines.
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. Informed consent was provided by parents/guardians on behalf of all child participants. All patients who were diagnosed with leprosy received free multidrug treatment (MDT) according to national guidelines. Data analysis. Graphpad Prism version 7.02 for Windows (GraphPad Software, San Diego CA, USA) was used to perform Mann-Whitney U tests, Kruskal-Wallis with Dunn's correction for multiple testing, plot receiver operating characteristic (ROC) curves and calculate the area under curve (AUC). The cut-off with the optimal sensitivity and specificity was determined using the Youden's index 24 .