Seroepidemiology of Measles, Mumps and Rubella on Bonaire, St. Eustatius and Saba: The First Population-Based Serosurveillance Study in Caribbean Netherlands

The National Immunization Program (NIP) on Bonaire, St. Eustatius and Saba (i.e., Caribbean Netherlands (CN)) includes the measles-mumps-rubella (MMR) vaccine since 1988/89. Seroepidemiological data is an important tool to evaluate the NIP, hence a cross-sectional representative population-based serosurveillance study was conducted for the first time in CN in mid-2017. Participants (n = 1829, aged 0–90 years) donated a blood sample and completed a health-related questionnaire. MMR-specific IgG antibodies were determined using a bead-based multiplex immunoassay and risk factors were analyzed using logistic regression models. Overall seroprevalence was high for measles (94%), but lower for mumps and rubella (both 85%). In NIP eligibles, including women of childbearing age, rubella seroprevalence (88%) exceeded the threshold for protection (85%); however, for measles (89%) this protective level (95%) was not met. MMR seropositivity was lowest in children who became CN resident at 11–17 years of age (especially for measles (72%)), mostly originating from Latin America and other non-Western countries. Interestingly, rubella seroprevalence was lowest in non-NIP eligible adults from Dutch overseas territories and Suriname (75%). Taken together, MMR immunity is generally good in CN, nonetheless some risk groups were identified. Additionally, we found evidence for a unique island epidemiology. In light of recent regional measles outbreaks, disease monitoring remains of utmost importance.


Introduction
Measles, mumps and rubella (MMR) are highly contagious viral diseases. Vaccination with trivalent MMR vaccine is safe and very effective at protecting against disease and severe

Laboratory Analyses
After the fieldwork, blood samples were air-shipped to the laboratory of the National Institute for Public Health and the Environment the Netherlands (RIVM) and stored instantly at −80 • C until analyses. MMR-specific IgG antibodies were determined with a fluorescent bead-based multiplex immunoassay using Luminex technology, as described previously [10]. In short, following standard protocol, a 3.2 mm (1/8 inch) punch was taken from the DBS and incubated in 300 µL phosphate-buffered saline containing 0.1% Tween-20 and 3% bovine serum albumin (i.e., assay buffer) at 4 • C overnight on a shaker to release serum (resulting in a 1:200 dilution) [11,12]. Sera were further diluted to 1:4000 in assay buffer. Controls, blanks and the international standard for rubella (RUBI-1-94), which was calibrated against the international standard for measles and an in-house standard for mumps, were included on each plate. Antibody concentrations were obtained by interpolation of the mean fluorescent intensity in the reference serum curve using a logistic-5PL regression type and expressed in international units per mL (IU/mL) for measles and rubella and RIVM units per mL (RU/mL) for mumps-as no international standard is available. An antibody concentration of ≥0.120 IU/mL for measles [13] and ≥10.0 IU/mL for rubella [14] was considered protective and used as cut-off for seropositivity. For mumps, no correlate of protection is available: An antibody concentration of ≥45.0 RU/mL was used as arbitrary criterion for seroprevalence, upon agreement by the European Sero-Epidemiology Network [15].

Seroprevalence and GMC
Data were analyzed in SAS v.9.4 (SAS Institute Inc., USA) and R v.3.6. Analyses took account of the survey design: To match the population distribution on each island as of January 1, 2017, overall seroprevalence and geometric mean concentrations (GMC) for IgG antibodies were estimated by linear weighting, taking into account sex, age group and country of birth (and neighborhood on Bonaire). Differences in seroprevalence of MMR-specific antibodies between islands and gender were determined by estimating the parameters of the beta distribution for these seroprevalence rates using the methods of moments [16]. Risk ratios, their corresponding 95% confidence intervals (CI) and p values were estimated by Monte Carlo simulations of these seroprevalence estimates. Dissimilarities in GMC between islands and gender were identified by calculating the difference in natural logarithmic (ln) concentrations and tested using a t-test. Age-specific seroprevalence, GMC and 95% CI were calculated for CN and per island. p values of <0.05 were considered statistically significant.

Waning Immunity after MMR Vaccination
Linear regression analyses of MMR ln-antibody concentrations were conducted to study the persistence of antibodies after one and two MMR vaccination(s) received in the Dutch overseas territories. Analyses were restricted to participants who had received MMR-1 between 13-16 months of age (as vaccine response and waning of antibodies is shown to be different in children up to 12 months of age [17]) and MMR-2 between 8-10 years of age, both given at least 2 months before inclusion in the study. Additionally, maximum age at study inclusion was 9 and 30 years for one and two dose(s), respectively. For mumps, those with self-reported mumps symptoms in the preceding year were excluded.

Risk Factors for Seronegativity
Risk factors for MMR seronegativity were identified using separate logistic regression models. A complete case analysis was conducted for both mumps and rubella (n = 1816; Figure S2). Allowing the measles model to converge, those born before introduction of routine vaccination were exclude (n total = 1075), i.e., a period characterized by widespread measles circulation causing nearly all participants to be seropositive (on Bonaire from 42 years of age, and on St. Eustatius and Saba from 36 years). Information on the history of MMR vaccinations on Bonaire before 1988 was derived from neighboring Dutch Leeward Antilles island Curaçao as the same NIP was applied. Studied risk factors included sociodemographics, vaccination history and other health-related factors. Aside from age and sex, variables with p < 0.10 in univariate analyses were included in the multivariate analyses. Backward selection was used to identify independent determinants in which p < 0.05 was considered statistically significant associated. Crude and adjusted odds ratios (OR) and 95% CIs were estimated as well as unadjusted seroprevalence and 95% CIs for all studied factors.

Study Characteristics
In the present study, 1900 participants (response rate 24.5%) were included, of which 1829 donated a blood sample ( Figure S2; 824 (45%) men and 1005 (55%) women; aged 3 months to 90 years), with equal distribution over the islands according to population size (Bonaire: 1129 (62%); St. Eustatius: 477 (26%); Saba: 223 (12%); Table 1). Most participants originated from the Dutch overseas territories (comprising CN, Aruba, Curaçao and St. Maarten) and Suriname (DOT-Sur; n = 1312, 72%), followed by Latin America and other non-Western countries (LA-nonW; n = 281, 16%), and indigenous Dutch and other Western countries (iD-Wes; n = 223, 12%). Almost half of the participants reported to be low educated (n = 883), compared to 26% middle and 18% high (8% unknown). On Saba, relatively more iD-Wes, LA-nonW and those with a high educational level participated-consistent with its population composition [18,19]-as compared to Bonaire and St. Eustatius. Among NIP eligible participants, i.e., those born in the MMR vaccination era, registered vaccination coverage with at least one dose against MMR ranged between 69-76% among the islands, and 8-9% were unvaccinated (and the remainder self-reported to have (partly) participated in the NIP).  Table 2 shows the overall weighted IgG seroprevalence and GMC of MMR in the total CN population, stratified by island and sex, and among NIP eligibles and non-NIP eligible adults. In total, 72.0% (n = 1337) were seropositive for all three pathogens, and 2.5% seronegative (n = 47; of which n = 26 had not reached the NIP eligible age, including all infants between 3-5 months of age (n = 6) for whom protective maternal antibody concentrations could be expected). There was no difference in overall seroprevalence for MMR between islands (all p > 0.05). Weighted MMR seroprevalence and GMC, stratified by age groups, for CN are depicted in Figure 1A-C, and per island in Figure S3A-C. The possible effect of storage and transportation on antibody concentrations of DBS samples in this study was investigated: No significant difference (all p > 0.05, one-way ANOVA) was found between MMR antibody concentrations of samples obtained at the start of the study (having the longest storage period (4 weeks)) compared to samples stored shorter (1, 2 and 3 weeks)-while displaying a similar age distribution.
Seroprevalence in CN was 74.1% at two years of age (after MMR-1) and rose to 93.2% at 5 years, with a corresponding increase in GMC from 37 to 197 RU/mL, respectively, reflecting the response to MMR-2 at 4 years on St. Eustatius and Saba ( Figure 1B and Figure S3B). At 10 years of age (after MMR-2 on Bonaire), seroprevalence in CN was 89.2% and GMC 154 RU/mL. Thereafter, seroprevalence and GMC fluctuated between 80-94% and 106-169 RU/ml, respectively, with age group 18-20 years displaying the highest prevalence (94.4%). All islands showed a similar trend in seroprevalence with age, except for age group 18-34 years in which seroprevalence was considerably higher on Bonaire (90.2%) than the other islands (<70%; Figure S3B). Overall seroprevalence in non-NIP eligible adults was lowest in residents from DOT-Sur (82.2%, and, e.g., in age group 60-90 years: 85.3% vs. 92.8% in iD-Wes and 88.6% in LA-nonW; Figure 2B)).
Seroprevalence of rubella showed a similar age pattern as measles among NIP eligibles ( Figure 1C) and was consistent across the islands ( Figure S3C). After MMR-1, seroprevalence in CN was 94.2% for two-year-olds and fluctuated between 87-100% until 18 years of age. Participants from LA-nonW aged 12-17 years were least seropositive (79.5%; Figure 2C). GMC reached its highest peak at four years of age at 76.0 IU/mL and declined to 33.3 IU/mL at seven years ( Figure 1C). From there it steadily declined towards 40 years of age (ranging between 25-39 IU/mL), reaching its lowest concentration at age group 40-44 years (19.8 IU/mL), but still above the cut-off for protection. Seroprevalence varied between 83-91% in adults aged 18-40 years. Unlike measles, seroprevalence remained indifferent after 40 years of age, varying between 76-87% and GMC between 25-47 IU/mL. Interestingly, seroprevalence and GMC in non-NIP eligible adults from DOT-Sur and LA-nonW were substantially lower than in iD-Wes (e.g., overall seroprevalence was 75.4%, 81.8% and 97.0%, respectively), and this also applied to the different age groups among them ( Figure 2C)).

Risk Factors for Seronegativity
Risk factors for measles were solely studied among NIP eligible participants as non-NIP eligible adults were nearly all seropositive. In multivariate analysis, men (vs. women), infants aged 0-1 years (vs. 2-10), those who have been resident of CN since age group 11-17 years (vs. 0-1) and participants who self-reported to have (partly) followed the NIP and who were unvaccinated (vs. two or more doses) had significantly higher odds of being seronegative (Table 3). For mumps, participants aged 0-1 and 2-10 years (vs. 11-17), those who have been resident of CN since age group 11-17 years (vs. 0-1), individuals who were vaccinated once, those self-reported to have (partly) followed the NIP, those who were unvaccinated and who were not eligible for the NIP (vs. two or more doses) were found to be significant risk factors for seronegativity in multivariate analysis ( Table 4). The multivariate model for rubella revealed that all age groups except 11-17 years (vs. 2-10), those who have been resident of CN since age 0-17 years (vs. 40-59), people who self-reported to have (partly) followed the NIP, those unvaccinated and who were not eligible for the NIP (vs. two or more doses) were significantly associated with seronegativity (Table 5).

Discussion
This cross-sectional population-based serosurveillance study estimated the level of humoral immunity against MMR and risk factors associated with seronegativity in CN. Overall seroprevalence was high for measles (94%), but lower for mumps and rubella (both 85%). In NIP eligibles, including women of childbearing age, rubella seroprevalence (88%) exceeded the threshold for protection (85%); however, for measles (89%) this level (95%) was not met [20,21]. MMR seropositivity was lowest in children who became CN resident at 11-17 years of age (especially for measles (72%)), mostly originating from Latin America and other non-Western countries. MMR vaccinations elicited good antibody responses and receiving two doses of MMR (vs. one) indicated prolonged humoral immunity. Interestingly, rubella seroprevalence was lowest in non-NIP eligible adults from DOT-Sur (75%), illustrative of a specific island epidemiology in the pre-vaccination era.
Overall seroprevalence for measles in CN (94%) was consistent with our previously reported estimate for Bonaire [22]. As reviewed by Dimech and colleagues [23], other large population studies reported measles seroprevalence rates between 54-96% (e.g., Italy: 74%, USA: 93%), of which the Netherlands was among the highest (96%). Seroprevalence for mumps in CN (85%) was rather similar to the USA (88%) [24], but somewhat lower than the Netherlands (91%) [25]. Rubella seroprevalence in CN (85%) was mostly lower than studies performed elsewhere, e.g., Colombia (89%), the Netherlands (95%) and Thailand (98%) [23]. Main drivers coinciding with differences in seroprevalence and antibody responses between and within populations can be attributed to vaccination status (and vaccine effectivity in general) as well as (past) natural exposure to these pathogens.
Children who reside in CN since age 11 years, i.e., after the regular NIP, had a high likelihood of being MMR seronegative. Indeed, lowest seroprevalence was observed in LA-nonW residents aged 12-17 years (e.g., for measles 66%) as they were less vaccinated. Hence, they probably missed vaccination opportunities in their country of birth due to lack of goods or migration-as their beliefs on vaccination (e.g., anti-vaccination) were indifferent from their peers (data not shown)-and did not catch up on missed vaccinations upon arrival to CN (which is regular policy). Based on these findings and in light of recent dissemination of measles across the region and influx of refugees [4], vaccination policy with respect to eligible immigrants aged <18 years was tightened where possible.
Moreover, male sex was an independent determinant for measles seronegativity among NIP eligibles (note: Additional risk factor analyses for rubella and mumps among NIP eligibles revealed a similar-although non-significant-association with sex (data not shown)). This sex difference was most prominent on St. Eustatius, while according to our registry vaccination coverage was even somewhat higher in men. To note, although vaccination status was available from a large proportion of participants, not all records could be retrieved. In that case, we used a self-reported variable on overall NIP attendance as a surrogate, which could not differentiate between (number of) vaccines unfortunately, and recall bias might have played a role too. Hence, we could not exclude that vaccination coverage against MMR on St. Eustatius was slightly higher in women as compared to men. Conversely, given the higher GMC in women too, women might serologically respond better to the vaccine components, as postulated by others [26,27]. Nonetheless, as cellular immunity is assumed to be an essential part of protection [28], higher risk of susceptibility among vaccinated men remains questionable. Future research in outbreak settings might provide clarification on this potential sex difference in vaccinated cases.
We detected significant dissimilarities between NIP eligibles and non-NIP eligible adults. Generally, the latter have (frequently) been naturally exposed to MMR during their life, elucidating high GMCs, indicative of lifelong protection [25,29,30]. This concept is best underlined by measles, a highly infectious agent that is capable of disseminating throughout susceptible populations [31]. Hence, nearly all non-NIP eligible adults in our study were seropositive for measles displaying high antibody concentrations, i.e., were infected, possible boosted regularly and thus protected. However, this was different for the less infectious pathogens mumps and, in particular, rubella. Interestingly, adult participants who were born on the islands or resided there since childhood were more likely to be seronegative. This was confirmed by a lower seroprevalence and GMC in adults from DOT-Sur and LA-nonW descent when compared to iD-Wes who were born in rubella endemic countries (mostly the Netherlands) prior to introduction of MMR vaccination. Principally for rubella, differences in seroprevalence between countries in the pre-vaccination period have been described [32]. Hence, as CN was even more remote and isolated during the pre-globalization/vaccine era, we hypothesize that introduction and transmission of rubella occurred less often due to its lesser infectious character, causing less circulating and exposure, affecting less people. Whilst a proportion of these inhabitants might still be susceptible currently and future cases cannot be ruled-out completely, disease in elderly is expected to be mostly mild, and yet sufficient herd immunity should prevent transmission. Fortunately, seroprotection for rubella was above the threshold for protection in NIP eligibles, including women of childbearing age who are at risk of developing congenital rubella syndrome-resulting in serious birth defects or miscarriage-via infection with rubella during pregnancy [20].
Consistent with literature, waning immunity of measles and rubella specific IgG antibodies after vaccination was present, but much slower after a second dose, staying well above the cut-off for seropositivity [29,30,33]. This indicates long lasting immunological humoral memory-when extended with similar rate of waning. Although this underlines the purpose of booster vaccination-besides preventing primary vaccine failure-we could not draw firm conclusions on the persistence of these antibodies as these data were cross-sectional. Similarly, the non-prospective character of our data was most likely the reason why mumps antibody levels were indifferent eight years after the first dose. Furthermore, seroprevalence rates for mumps should be interpreted with caution as a defined correlate of protection-albeit recent research endeavored [34]-is still lacking. While two doses of MMR (vs. one) indicated a lower risk of mumps seronegativity in our study, outbreaks among twice vaccinated students-with intensive and homogenous contact-have been reported [35]. In fact, in contrast to St. Eustatius and Saba, we observed a high seroprevalence and GMC in young adults on Bonaire and self-reporting of mumps symptoms was highest among this group too. This, together with confirmed cases from nearby Dutch Leeward Antilles island Aruba and recurrent traffic between these islands, could suggest possible exposure to mumps on Bonaire, whereas this likelihood might be lower on St. Eustatius and Saba, which are more isolated.
All infants too young to be vaccinated with MMR were seronegative in our study; even among the infants 3-5 months of age for whom protective maternal antibody concentrations could be expected. This phenomenon is well-known among babies from mothers who have not been naturally infected as antibody concentrations from vaccination are significantly lower and thus reach cutoffs for seropositivity earlier [36]. Timely vaccination and close monitoring remains of great importance, especially in light of recent regional circulation of the measles virus and migration of large populations at risk (e.g., from Venezuela) [3,4]. While considering an optimal age for vaccination, health authorities should take into account several factors, including immunological response, vaccine coverage, herd immunity thresholds and risk of infection [37]. Although no cases of measles have been detected via the surveillance systems in CN recently, the public health department on Bonaire decided to lower the age of MMR-2 from 9 years to 18 months of age (as of January 1, 2019), in order to timely and adequately protect young infants and reduce the risk of viral introduction and transmission.
This study has some additional limitations. Due to the overall response rate of 25%, the possibility of non-response bias cannot be excluded (as described earlier [9]). However, we partly corrected for this by weighting our sample on important sociodemographic characteristics. Further, to overcome logistical hurdles, we used DBS to collect our blood samples. Whilst this is a widely used and validated method for measuring antibodies, we could not exclude that storage and transportation might have had some effect on the antibody levels. We have investigated this and found little overall effect that has not affected our results.

Conclusions
In conclusion, this is the first large-scale serosurveillance study in CN providing evidence on humoral immunity against MMR. The CN population is overall well protected against MMR, albeit some groups were identified that could be at risk of infection. Our data also indicate that infectious disease epidemiology on these islands might have been different in the pre-vaccination era as compared to past MMR endemic countries, such as the Netherlands. Particularly in light of recent outbreaks in the region, it is important to have sensitive disease surveillance in place and to sustain high vaccination coverage in order to meet herd immunity thresholds and, ultimately, reach the WHO measles and rubella elimination goals [20]. Lastly, it is highly recommended to conduct serosurveillance studies in CN on a regular basis in the future in order to monitor the protection against vaccine-preventable diseases and timely detect (additional) gaps in population immunity [8].