The authors have declared that no competing interests exist.
Conceived and designed the experiments: BFA PJL JWP JMC. Performed the experiments: JWP KLH DMM PJL. Analyzed the data: BFA JWP. Wrote the paper: BFA JWP KLH DMM JMC PJL.
Efforts to monitor malaria transmission increasingly use cross-sectional surveys to estimate transmission intensity from seroprevalence data using malarial antibodies. To date, seroconversion rates estimated from cross-sectional surveys have not been compared to rates estimated in prospective cohorts. Our objective was to compare seroconversion rates estimated in a prospective cohort with those from a cross-sectional survey in a low-transmission population.
The analysis included two studies from Haiti: a prospective cohort of 142 children ages ≤11 years followed for up to 9 years, and a concurrent cross-sectional survey of 383 individuals ages 0–90 years old. From all individuals, we analyzed 1,154 blood spot specimens for the malaria antibody MSP-119 using a multiplex bead antigen assay. We classified individuals as positive for malaria using a cutoff derived from the mean plus 3 standard deviations in antibody responses from a negative control set of unexposed individuals. We estimated prospective seroconversion rates from the longitudinal cohort based on 13 incident seroconversions among 646 person-years at risk. We also estimated seroconversion rates from the cross-sectional survey using a reversible catalytic model fit with maximum likelihood. We found the two approaches provided consistent results: the seroconversion rate for ages ≤11 years was 0.020 (0.010, 0.032) estimated prospectively versus 0.023 (0.001, 0.052) in the cross-sectional survey.
The estimation of seroconversion rates using cross-sectional data is a widespread and generalizable problem for many infectious diseases that can be measured using antibody titers. The consistency between these two estimates lends credibility to model-based estimates of malaria seroconversion rates using cross-sectional surveys. This study also demonstrates the utility of including malaria antibody measures in multiplex assays alongside targets for vaccine coverage and other neglected tropical diseases, which together could comprise an integrated, large-scale serological surveillance platform.
Efforts to monitor malaria transmission to inform control strategies increasingly use cross-sectional surveys to estimate transmission intensity from seroprevalence data based on malaria antibodies
Investigators have estimated malaria transmission intensity from cross-sectional prevalence surveys using seroconversion rates estimated with a reversible catalytic model
The objective of this study was to estimate the malaria seroconversion rate using antibody measures against merozoite surface protein-119 (MSP-119) from incident seroconversions measured in a longitudinal cohort of Haitian children ages 0–11 years old, and compare it to the rate estimated with a reversible catalytic model fit to a cross-sectional survey of Haitians aged 0–90 years old. Since the longitudinal data provide a direct measure of the seroconversion rate, a comparison of estimates from the two approaches provides an important check of the model's consistency as currently applied in low-transmission settings.
Study populations were set up initially to monitor transmission of lymphatic filariasis (LF) in a setting of intense LF transmission. Both longitudinal and cross sectional studies were carried out in the coastal plain near Léogâne, Haiti, where up to half of the population was infected with
The protocols for both studies were reviewed and approved by the Centers for Disease Control and Prevention's Institutional Review Board and the Ethical Committee of St. Croix Hospital (Léogâne, Haiti). After explaining the purpose of the study in Creole, individuals were asked to provide verbal consent to participate in the research. The human subjects review boards approved the verbal consent process due to low literacy rates in the study communities. In cases of longitudinal follow-up, the study team documented consent at each study visit. Mothers provided consent for young children, and children 7 years or older provided assent. Consent forms included specific permission to share specimens and to test the samples for other infectious diseases.
A recombinant GST/MSP-119 fusion protein cloned from
We used adult US citizens with no history of foreign travel as an unexposed population for antibody cutoff values to classify individuals as seropositive. The mean +3*SD for the MSP-119 antibody response was calculated from the logged values of the negative control antibody responses. Because samples from the longitudinal study and the cross-sectional study were run with two different bead lots, two different cutoff values were used: a cutoff of 365 MFI-bg units was calculated from 63 negative control sera for the longitudinal study bead set, and a cutoff of 477 MFI-bg units was calculated from 70 negative control samples for the cross-sectional study bead set.
Four children <6 months in the longitudinal cohort and one child age 2 months in the cross-sectional study had evidence of maternal antibodies to multiple antigens (including malaria) in the multiplex panel; we classified them as seronegative in their first year of life for the analysis. For children who were classified as seropositive during follow-up we plotted their antibody responses to identify those who were incident cases versus those who were positive at their first visit.
We estimated age-specific seroprevalence in the longitudinal cohort by combining measurements into one-year to three-year age groups that included enough measurements to estimate each prevalence with reasonable precision. We also estimated the seroconversion and seroreversion rates using incident conversions and reversions divided by the person time at risk over follow-up
With short periods between measurements such as days, the approach to estimating the rates directly is unbiased
With cross-sectional data, direct information about seroconversion and seroreversion is unknown since the same individual is not observed at two points in time. Instead, there is only current status information for an individual at one point in time when they are a particular age. The reversible catalytic model for incidence data can be fit to prevalence data with a simplification where exposure time is measured by age in years
Among the 142 children enrolled in the longitudinal study in Leogane, each child was followed for an average of 5.1 years (range = 0.5, 9.1). The study included 771 total antibody measurements, and the average number of measurements per child was 5.4 (range = 2, 9). The 383 individuals enrolled in the cross-sectional survey in Miton ranged in age from two weeks to 90 years old. Antibody response increased with age in both samples (
Panel A includes antibody responses from the longitudinal study in Léogâne, Haiti, 1991–1999. Panel B includes antibody responses from the cross-sectional survey in Miton, Haiti, 1998.
Age Category (years) | Median Age (years) | N | % | 95% CI |
[0–2) | 1.0 | 127 | 5 | (1, 9) |
[2–3) | 2.5 | 98 | 7 | (2, 14) |
[3–4) | 3.5 | 114 | 10 | (4, 16) |
[4–5) | 4.6 | 103 | 15 | (7, 23) |
[5–6) | 5.4 | 115 | 12 | (6, 20) |
[6–8) | 6.8 | 125 | 17 | (8, 26) |
[8–11.9] | 9.1 | 89 | 12 | (3, 24) |
All ages [0–11.9] | 4.5 | 771 | 11 | (7, 16) |
N is number of measurements.
CI: Confidence Interval.
Age Category (years) | Median Age (years) | N | % | 95% CI |
[0–5) | 2 | 51 | 6 | (1, 16) |
[5–10) | 7 | 67 | 13 | (6, 24) |
[10–15) | 12 | 82 | 34 | (24, 45) |
[15–20) | 16 | 43 | 49 | (33, 65) |
[20–30) | 24 | 51 | 57 | (42, 71) |
[30–40) | 33 | 32 | 53 | (35, 71) |
[40–90] | 50 | 57 | 54 | (41, 68) |
All ages [0–90] | 14 | 383 | 36 | (31, 41) |
Ages [0–11] | 7 | 157 | 14 | (9, 20) |
N is number of individuals.
CI: Confidence Interval.
In the longitudinal study, 25 children were identified as seropositive for malaria during follow-up. Of these, 13 were incident seroconversions and 12 had seroconverted by their first measurement (
Panel A includes individuals with incident seroconversions, and panel B includes those who were seropositive at their first measurement. The dashed line marks the cutoff value (365) used to determine seropositive antibody levels. The light grey lines plot antibody profiles for seronegative children, and the solid black line in each plot is a loess smoother over the seronegative children antibody levels.
Seroprevalence estimates for the age categories in
Study | Estimation approach | Seroconversion rate (95% CI) | Seroreversion rate (95% CI) |
Longitudinal cohort (Léogâne) | Incident cases, ages ≤11 y | 0.020 (0.010, 0.032) | 0.153 (0.073, 0.266) |
RC model, ages ≤11 y |
0.021 (0.001, 0.096) | 0.163 (0.001, 0.729) | |
Cross-sectional survey (Miton) | RC model, ages ≤11 y |
0.023 (0.001, 0.052) | 0.001 (0.001, 0.255) |
RC model, all ages | 0.039 (0.027, 0.052) | 0.024 (0.005, 0.043) |
CI: Confidence Interval; RC: Reversible Catalytic model.
* Lower 95% confidence intervals truncated at the lower bound of possible parameter values (0.001).
In this analysis of two separate studies from Haiti we found that the seroconversion rate for the MSP-119 malaria antibody was very similar when estimated from incident seroconversions in a longitudinal cohort and from a model fit to cross-sectional seroprevalence data. This finding is important because it provides a direct validation of the use of cross-sectional malaria seroprevalence data to estimate seroconversion rates in low-transmission settings. Earlier analyses demonstrated that the seroconversion rate measured from cross-sectional surveys was strongly associated with the entomological inoculation rate
The estimation of seroconversion rates using cross-sectional data is a widespread and generalizable problem for many infectious diseases. Although the model used to estimate malaria seroconversion rates from cross-sectional surveys is an extreme simplification of a complex immunological process, numerous field studies including the present study have found that the model fits the data well
More recently, Bretscher et al.
Our results suggest that the simpler calculation of seroconversion rates directly from incident seroconversions is a viable alternative to a more complex, Hidden Markov Model approach. First, as we have demonstrated there is no need to collapse individual information into age categories and group level seroprevalence to estimate seroconversion rates with longitudinal data – rates can be estimated directly from incident seroconversions
Serum samples from these communities were tested by multiplex as part of a study of risk factors for acquisition of LF; MSP-119 was included in the multiplex, along with antigens from enteric pathogens and vaccine-preventable diseases to better understand the public health context in these communities and the potential interactions between LF and other infections. The current results illustrate the potential of this approach to capture seroincidence data for infections beyond those that were the initial focus of the study. Less clear at this point is the extent to which reversible catalytic models might be successfully applied to the other infections we monitored. These efforts are the focus on ongoing efforts in our labs. Independent of whether or not reversible catalytic models can be applied generally as measures of transmission intensity, multiplex serologic assays represent a powerful tool for capturing useful public health data with simple, dried blood spot surveys.
A limitation of the analysis was that only 25 children were classified as seropositive in the longitudinal study, which meant that it was impossible to estimate age-specific seroconversion rates and it was also impossible to estimate the seroreversion rate with precision. The wide confidence intervals and variation in point estimates for the seroreversion rate in analyses limited to children ≤11 years reflect the lack of information needed to accurately estimate that parameter (
These studies were not designed to follow malaria specifically and we have no parasitologic data that would confirm that children were, in fact, malaria infected. That limitation notwithstanding, not all malaria infections may manifest as patent parasitemia, so serological measures of infection may be a more accurate representation of the true level of transmission in the population – particularly in areas of low transmission
Our finding of close agreement between malaria seroconversion rates estimated in a prospective cohort study with those estimated using a reversible catalytic model fit to cross-sectional prevalence data lends additional credibility to the use of cross-sectional, serological surveys to monitor malaria transmission in low-transmission settings. These results demonstrate the utility of including malaria antibody measures in multiplex assays alongside targets for vaccine coverage and other neglected tropical diseases, which together could comprise an integrated, large-scale surveillance platform.
(PDF)
We express our gratitude to the families and children who participated in these studies and to the members of the filariasis research team, including Mark Eberhard, Thomas Streit, David Addiss, Michael Beach, Jacky Louis Charles, Jean Marc Brissau, and the many trainees who contributed to the follow up of these children.