Lymphatic filariasis (LF) is a neglected tropical disease (NTD) transmitted by mosquitoes and caused by the roundworms Brugia malayi, Brugia timori, and Wuchereria bancrofti. Microfilariae (Mf), transmitted via the bite of an infected mosquito, go on to develop into adult worms that nest in the lymphatic vessels and can lead to impairment of the lymphatic system [1]. If left untreated, infection can result in severe lymphedema (and, in males, scrotal hydrocele) leading to permanent disability and disfigurement [2]. As of 2021, an estimated 882.5 million people in 44 countries were at risk of LF, with W. bancrofti responsible for 90% of cases [2].

Mass drug administration (MDA) is the World Health Organization’s (WHO) recommended preventive chemotherapy strategy for eliminating LF. MDA regimens reduce the Mf density in the bloodstream and the intensity of infection in communities to a level where transmission is unlikely to be sustained [2,3]. WHO currently recommends the use of point-of-care tests to detect circulating W. bancrofti filarial antigen (Ag) produced by adult worms as the standard serological indicator for LF programmatic surveys [4]. However, there are concerns that Ag testing may have lower sensitivity to identify infections when Ag prevalence is very low, e.g., following effective MDA. This can potentially lead to failure to identify signals of transmission [5–7], particularly among subgroups that are known to have lower Ag prevalence, e.g., young children [8].

Serological assays that test for immunoglobulin G (IgG) antibodies (Ab) to pathogen-specific Ag can provide evidence of an individual’s infection or vaccination status [9]. In general, IgM Ab appear in the early period of an infection and are considered indicative of current/recent infection, whilst IgG Ab usually develop a week or more post-infection and can persist for long periods, sometimes for life [10]. Thus, IgG Ab responses are generally considered markers of middle/late-stage active infection, past infection, or immunity [9,10].

Serological tests that detect the presence of anti-filarial IgG Abs (including Wb123 Ab, Bm14 Ab, and Bm33 Ab) have been found to be more sensitive for detecting LF infection than Ag testing alone [5,6]. Studies have shown that Ab responses may develop before patent infection or antigenemia [11], thus Ab testing may provide an earlier indication of infection or identify asymptomatic individuals. Thus, at the population level, seroprevalence of anti-filarial Abs may provide earlier signals of LF transmission and/or resurgence and enable a timelier response compared to using Ag prevalence alone [5,12]. Further, Wb123 is a larval Ag, meaning that Wb123 Ab-positive results are more likely to be associated with exposure to infected mosquitoes (and therefore an indicator of ongoing transmission) than Bm14 Ab-positive or Bm33 Ab-positive results [13]. A study from American Samoa found that the average Number Needed to Test (NNTest^av) to identify one LF-seropositive individual was lower for Abs than for Ag, reinforcing the potential value of LF Ab serosurveillance [14].

However, the serological profile for LF is complex because of the long lifespan of adult filarial worms; individuals may be infectious (microfilaraemic) for many years, and Abs, Ag, and Mf could be detected concurrently, including in asymptomatic infections. On the other hand, an individual may still be seropositive for Ag and Abs for months to years following a cleared infection [15]. Currently, there are knowledge gaps related to the sensitivity and specificity of Abs to identify active LF infection, and whether Abs can be used to discriminate between active, recently cleared, or past LF infection.

Samoa is a tropical island nation in the Western Pacific Region with a population of 222,382 in 2022 [16]. Despite substantial reductions in LF transmission following multiple national and targeted MDA rounds, most recently in September 2023, Samoa remains endemic for LF [17,18]. A nationwide Transmission Assessment Survey (TAS) was conducted in 2013, in which two of three evaluation units (EU) passed target thresholds, and a second TAS took place in 2017 where all three EUs failed [18]. In 2017, the WHO released guidelines on ‘Alternative MDA Regimens to Eliminate LF’ following several countries reporting suboptimal results after MDA with standard two-drug regimens [19]. In countries endemic for LF but without onchocerciasis or loa loa that were using diethylcarbamazine and albendazole (DA) for MDA, recommendations were to switch to annual triple-drug MDA using ivermectin, diethylcarbamazine, and albendazole (IDA) in settings with (i) implementation units that had not started, or had fewer than four effective DA rounds; (ii) implementation units that had not met epidemiological thresholds in sentinel and spot-check site surveys or in TAS despite meeting drug coverage target; and (iii) communities where post-MDA or post-validation surveillance identified infection suggesting local transmission [19]. In 2018, Samoa was the first country in the world to distribute IDA nationwide.

The operational research project Surveillance and Monitoring to Eliminate LF and Scabies from Samoa (SaMELFS) was established to monitor the impact of nationwide triple-drug MDA on LF transmission in Samoa [18]. The first SaMELFS survey took place in 2018, 1.5-3.5 months following the first national distribution of IDA, and was considered a baseline survey for measuring the impact of triple-drug MDA on LF prevalence. A previously published paper reported Ag prevalence at the population level, as well as by age group (5-9 years vs ≥10 years), and found an adjusted overall Ag prevalence of 4.0% (95% confidence interval [CI]: 3.1-7.0), and a three-fold higher Ag prevalence among participants aged ≥10 years compared to those aged 5-9 years [18].

This study complements our previous paper that reports baseline Ab seroprevalence (Bm14, Wb123, and Bm33 Abs) estimates from the SaMELFS 2018 survey [20]. The current study aimed to investigate the potential utility of anti-filarial Abs to guide and support decision-making for national LF elimination programs. Our objectives were to i) investigate the sensitivity and specificity of individual Abs and combinations of Abs to identify Ag-positives; and ii) estimate the average number needed to test (NNTest^av) to identify LF-seropositives (seropositive for Ag and/or Ab) and iii) compare the efficiency of the different serological indicators by target age group and sampling design.

Overall, 3940 participants aged ≥5 years had DBS collected for MBA. Of these, Ab results were not available for 89 (2.3%) participants: 35 samples had a low bead count (one of which also had an invalid FTS result), no reason was given for 21 samples (of which 14 had no FTS result recorded and 7 were Ag-negative), no consent was given for 17 samples, and 16 duplicate samples were excluded (of which three were Ag-positive and Mf-positive, and two were Ag-positive and Mf-negative). A further 56 participants without an FTS result were excluded (30 had invalid FTS results, 16 did not have enough blood, and 10 had no reason given), giving a total sample of 3795 participants included in the analyses (S1 Fig).

As more countries near LF elimination, sensitive and specific diagnostic tools are required to identify residual LF infections and enable timely intervention to prevent disease resurgence. This study has provided evidence on the potential value of using Abs as an adjunct to Ag for LF surveillance programs, supporting previous studies suggesting that Abs could be more sensitive indicators of LF infection than Ag alone [11]. Our study has quantified the proportion of LF-seropositive individuals that may be missed if Ag were used alone, as well as those missed when individual Abs and different combinations of Ab and Ag are used. We found that Bm14 Ab had the highest ROC of the three Ab seromarkers to classify participants as Ag-positive, which was equal to testing positive for both Wb123 Ab and Bm14 Ab, or testing positive for all three Abs. This information is important for supporting WHO to select the most sensitive and specific diagnostic tools for LF surveillance.

Identifying asymptomatic individuals and residual foci of infection becomes increasingly challenging as LF prevalence decreases following successful MDA. The current target product profile for diagnostic tests for LF surveillance aims for a minimum sensitivity of 85% and specificity of 98.8% [26]. In our study, we used Ag-positivity as a reference for assessing the sensitivity and specificity of Abs because Ag is currently the main indicator used for LF surveillance and programmatic decisions. We found that all Abs and combinations of Abs had a sensitivity >85%, but no individual Ab, or combination of Abs, had a specificity >98.8%. We acknowledge the limitations of using Ag-positivity as a reference, as Ag can persist for months following effective treatment [27] and being Ag-positive does not always indicate active infection. Further, due to the timing of our survey immediately following MDA, we were unable to use Mf data to identify infectious patients in our sample.

Of the three Ab seromarkers, Bm33 Ab had the lowest ROC to classify participants as Ag-positive. The high seroprevalence of Bm33 Ab in previous surveys [12,20] suggests it is very long-lasting post-infection. Thus, the poor discriminatory value of Bm33 Ab limits its use for LF surveillance in programmatic surveys, particularly in endemic areas with previously high infection rates. Contrastingly, we found Bm14 Ab had the highest ROC of the three Abs, but the lowest sensitivity to detect Ag-positives. Though the higher ROC of Bm14 Ab is promising, it may not be sufficiently sensitive to support LF surveillance alone, but could be useful if tested in combination with Ag. The ROC of Wb123 Ab was between that of Bm14 Ab and Bm33 Ab, suggesting Wb123 Ab may be more discriminative than Bm33 Ab to identify true positives and more sensitive than Bm14 Ab. This finding is supported by a study in Papua New Guinea that found Wb123 Ab responses were more closely correlated to persistent Mf infection compared to Ag and Bm14 Ab responses [28]. A further 2015 study of Togolese school children who were born shortly before or after MDA was discontinued found that 4.7% were Wb123 Ab-positive, despite recent surveys finding very few Ag-positive or Mf-positive cases [29].

An investigation of Ab responses in three sequential TAS in American Samoa reported that seropositivity to a combination of Wb123 Ab and Bm14 Ab, or any of Bm14 Ab, Wb123, or Bm33 Ab in TAS-2 were highly sensitive and specific for predicting schools with Ag-positive children in the following survey [12]. Our study found that testing positive to both Wb123 Ab and Bm14 Ab had a sensitivity >85% and specificity >90%, suggesting that testing for both Wb123 Ab and Bm14 Ab, possibly alongside Ag, could be an approach to improve LF surveillance. Indeed, studies have found that Ag-positive and Mf-positive children are generally seropositive for at least two Abs [30].

As countries approach elimination and enter the Stop MDA Phase, WHO recommends conducting TAS to estimate LF Ag prevalence among 6-7-year-old children to determine whether the critical threshold of <1% Ag prevalence (in areas where W. bancrofti is endemic and Aedes is the primary vector [31]) has been reached. However, our study found that i) the probability of testing positive for all seromarkers was significantly higher among ≥10-year-olds compared to 5-9-year-olds; ii) the NNTest^av was significantly lower among ≥10-year-olds compared to 5-9-year-olds; and iii) the correlation between seromarker prevalence at the PSU level was weakest when restricted to 5-9-year-olds compared to ≥10-year-olds. Given our findings, conducting TAS among older aged school children and/or among community members, particularly in suspected hotspots, may increase the likelihood of detecting persistent LF transmission and could prevent the premature cessation of MDA and other surveillance activities.

Interestingly, we found that the ratio of Ab-positivity to Ag-positivity was largest in female participants aged 5-9 years in purposively selected PSUs, and among participants aged ≥10 years in both purposively selected and randomly selected PSUs, suggesting that there might be gender differences in Ab responses, and therefore the utility of Abs for surveillance. Previous studies have found that females elicited Bm14 Ab responses at an earlier age than males [11]. It would be interesting to investigate whether Ab responses are influenced by gender, and if so, how this may impact LF surveillance moving forward.

The serological course and duration of persistence of LF Abs is poorly understood and more research is needed to understand anti-filarial Ab kinetics. Indeed, longitudinal monitoring has shown that individual Abs are detected at different times over the course of infection and their expression can differ by age [11]. Further, population-level Ab prevalence has been seen to drop after MDA [5,32,33], but Abs can persist for a long time following clearance of Ag from the original infection (though the exact duration is uncertain) [14,15]. Thus, Ab-positivity is not always indicative of active infection [15]. This can have cost implications for national programs if Ab-positive individuals are treated unnecessarily. Further research may aid in the interpretation of Ab signals and elucidate whether infection can be classified as active or past based on serological patterns of Ab alone.

Despite these knowledge gaps, there are advantages of Ab serosurveillance. Firstly, there is a long latency period between LF infection and the development of antigenemia (Ag tests may take months to become positive after infection) [29]. By contrast, Ab responses develop before patent infection or antigenemia, and antigenemia is not a requirement for robust Ab responses [11]. Longitudinal monitoring of Abs in Haiti suggest that Bm33 Ab may be a marker of very early LF infection; this study found that Bm33 was the first Ag to elicit an Ab response and was detected on average more than one year before other Ab responses [11]. Bm14 Ab and Wb123 Ab responses developed after Bm33 Ab, but on average six months before the development of antigenemia [11].

Secondly, the high sensitivity of Ab compared to Ag is a particular asset in the context of LF endgame strategies and may enable the identification and management of LF hotspots and residual transmission foci. Spatial analysis using Ag and Ab seroprevalence estimates in American Samoa found that clusters of Ab seropositives were strongly correlated with the presence of multiple Mf-positives, indicating the potential utility of Ab serosurveillance to identify ongoing transmission [34].

Thirdly, there is the potential for monitoring seroprevalence for different seromarkers at different stages of the elimination program. For example, Abs that are rapidly cleared (such as Bm14 Ab and Wb123 Ab) could be monitored in countries in the Stop MDA Phase, whilst less specific but highly sensitive Abs (such as Bm33 Ab) could be used in elimination settings among populations that should be immunologically naïve with no prior LF exposure. This could aid in identifying cut-off measures for more recent exposure or active infection [35,36]. Lastly, technologies such as MBA only require a small volume of blood collected on filter paper to detect Abs, which is logistically and resourcefully appealing. MBA have the added advantage of simultaneously detecting Ab to other diseases of interest, such as vaccine-preventable diseases, arboviruses, and other NTDs, that can enable integrated disease surveillance, management, and treatment.

In conclusion, as more countries approach elimination, further work is urgently needed to determine how Ab serosurveillance can be used to support national programs. Abs are more sensitive than Ag for identifying LF-seropositive participants and may be a useful tool for LF surveillance and elimination programmes. Here, we have established baseline Ab and Ag seroprevalence estimates in Samoa that can be monitored longitudinally to investigate the impact of MDA and further investigate the sensitivity of Ab compared to Ag over time. Further, the low specificity of Abs means that further research is required to fully understand the behaviour of each Ab to determine how Ab serosurveillance can best be used to support LF surveillance programmes.