Preexisting yellow fever virus and West Nile virus immunity and pregnancy outcomes in a Nigerian cohort with endemic orthoflavivirus exposure

ABSTRACT

Yellow fever virus (YFV) and West Nile virus (WNV) co-circulate with other arboviruses, including Zika (ZIKV), dengue (DENV), and chikungunya virus (CHIKV), in sub-Saharan Africa. Associations between preexisting YFV and WNV immunity with symptoms and adverse infant outcomes among pregnant women exposed to orthoflaviviruses are unknown. We retrospectively studied a prospective cohort of pregnant women enrolled between 2019 and 2022 in Jos, Nigeria. Rapid tests identified ZIKV, DENV, and CHIKV IgM/IgG reactivity for enrolment; 216 women underwent Western blot for YFV and WNV IgG. Logistic regression evaluated associations between arboviral seropositivity and maternal symptoms or adverse infant outcomes. Sequential serology of mother-infant pairs estimated the persistence of passively transferred maternal YFV antibodies. YFV IgG was detected in 50.5% (109/216) and WNV IgG in 5.1% (11/216) of maternal samples. YFV and WNV seropositivity was significantly associated with maternal symptoms (OR = 2.02, 95% CI: 1.35–3.02, P = 0.001), as was YFV seropositivity alone (OR = 1.77, 95% CI: 1.21–2.61, P < 0.004). CHIKV IgM reactivity was significantly associated with abnormal infant outcomes (OR = 2.38, 95% CI: 1.43–4.02, p = 0.001), but not ZIKV and DENV IgM reactivity. Passive maternal YFV IgG waned in infants at a median of 3.1 months (IQR: 1.65-5.35 months) after birth. YFV and WNV seropositivity was associated with maternal symptoms but not with adverse infant outcomes. Rapid waning of maternal YFV IgG highlights infant vulnerability and supports enhanced surveillance and maternal immunization strategies.

Introduction

Arthropod-borne viruses (arboviruses) such as yellow fever virus (YFV), West Nile virus (WNV), Zika virus (ZIKV), dengue virus (DENV), and chikungunya virus (CHIKV), pose persistent public health threats to maternal and child health globally, particularly in sub-Saharan Africa and tropical regions [1,2]. Infections with ZIKV, DENV, and CHIKV during pregnancy have been linked to adverse pregnancy outcomes, including congenital abnormalities, stillbirth, and preterm birth [3-5]. However, the roles of WNV and YFV infections in pregnancy are limited despite their known ability to cause viremia and neuroinvasive disease [6,7]. Prior studies have focused primarily on acute infection [8] with limited data on the impact of prior YFV and WNV exposure in pregnant women [9].

In orthoflavivirus-endemic regions, sequential or concurrent exposures may trigger complex immunologic interactions. Prior DENV or ZIKV immunity has been associated with both protection and antibody-dependent enhancement (ADE) [10], whereby preexisting antibodies enhance infection severity in humans [11,12]. These dynamics are less well understood for YFV and WNV, especially during pregnancy. Prior studies in general populations have explored interactions between YFV and DENV, including findings that YFV vaccination does not increase DENV severity and that prior DENV exposure may reduce YFV disease severity [13]. Whether YFV and/or WNV immunity, acquired through natural infection or vaccination, modulates the course of subsequent arboviral infections such as ZIKV, DENV, or CHIKV remains largely unknown, especially in Africa. This knowledge gap is particularly relevant for pregnancy, where immunologic priming could influence symptom severity or vertical transmission risk.

Arboviruses, including ZIKV, DENV, CHIKV, YFV, and WNV, co-circulate in Nigeria, where annual transmission and overlapping mosquito vectors create an environment of recurring outbreaks and co-infections [14]. Our research and surveillance efforts in this region have emphasized cross-reactive immunity in DENV and ZIKV infection and pregnancy outcomes associated with ZIKV, DENV, and CHIKV, which are known for causing abnormal infant outcomes and maternal symptoms [4,15]. In contrast, YFV has received comparatively less attention, likely due to assumptions of widespread vaccine-derived immunity and high efficacy of the YFV vaccine [16]. In response to repeated yellow fever epidemics, Nigeria incorporated YFV vaccination into its national immunization schedule in 2004, targeting infants at 9 months of age [17]. While this has improved early-life protection, maternal exposure remains understudied, and the population-level impact of YFV and WNV immunity in pregnancy is poorly defined [18].

In endemic settings with significant baseline immunity in pregnant women, passive immunity may provide critical neonatal protection against orthoflavivirus infections [19]. Maternal WNV and YFV IgG antibodies can cross the placenta to provide transitory immunity to newborns [8]. However, the duration of this protection remains uncertain, and antibody waning occurs within the first few months of life for other orthoflaviviruses [20]. Studies on DENV and ZIKV suggest that passively acquired maternal antibodies may decay leaving infants vulnerable to infections before their immune systems fully mature [21]. In the mouse model, sub-neutralizing levels of maternally transferred ZIKV antibodies after waning can increase disease severity in secondary DENV infection and cause antibody-dependent enhancement (ADE) in pups [22]. Therefore, understanding the kinetics of maternal antibody persistence is essential to determine whether maternal immunity influences susceptibility to (or protection from) orthoflavivirus infections in neonates.

Our study determined the seroprevalence of YFV and WNV in our prospective cohort of pregnant women in Nigeria [23], adding to our surveillance of ZIKV, DENV, and CHIKV [4,15]. Using this new data, we re-examined associations between maternal arbovirus serostatus with maternal symptoms and adverse infant outcomes, enabling a more complete analysis of circulating arboviruses. Finally, we measured the rate of antibody passive transfer among YFV seropositive mothers and the time to antibody waning among infants. Understanding the potential impact of WNV and YFV immunity during arbovirus infection in pregnancy has crucial implications for maternal and infant health policies, vaccination strategies, and vector control efforts in orthoflavivirus-endemic regions.

Methods

Study design and participants

This study retrospectively analysed archived serum samples from a prospective cohort of pregnant women recruited from antenatal clinics at Jos University Teaching Hospital (JUTH) and Our Lady of Apostles (OLA) Hospital in Jos, Nigeria between April 1, 2019 and January 31, 2022, as previously described [4,15].

A total of 1,006 pregnant women were screened, with 787 symptomatic and 219 asymptomatic participants based on six arboviral symptoms (fever ≥37.5 °C, rash, headache, arthralgia, conjunctivitis, or myalgia). Rapid diagnostic testing for ZIKV, DENV, and CHIKV IgM/IgG using the ChemBio DPP assay identified 312 women with positive IgM/IgG reactivity to any of the three viruses (200 IgM ± IgG and 112 IgG only). Of these, 240 women consented to participate in the prospective study. Twenty-four samples were unavailable, resulting in 216 maternal serum samples included in the final analyses for YFV and WNV IgG testing (Supplementary Figure 1).

Maternal arboviral symptom associations were assessed in all 216 women. Symptoms reported within 7 days were recorded corresponding to the maternal serum sample. Infant outcome data were available for 149 mother-infant pairs. Infants were followed for examination and sample collection to coincide with routine follow-up visits: 6, 10, 14 weeks, and 6 months. Microcephaly and low birth weight were defined as head circumference or weight, respectively, with Z-score less than or equal to -2 standard deviations (SD). Z-score was calculated as the difference between the observed value and the median reference value divided by the SD, derived from sex- and age-specific reference populations used in the WHO Multicentre Growth Reference Study [24]. Infants were preterm if delivered earlier than 37 weeks gestational age; Fenton growth charts were used to calculate Z-scores for preterm infants [25]. A subset of 96 mother-infant pairs had infant samples available for testing; 36 infants born to YFV-seropositive mothers were followed longitudinally to estimate the median time to loss of maternal antibodies. Ethical approval for the study was granted by the institutional review boards of Jos University Teaching Hospital (JUTH/DCS/ADM/127/XXVIII/1338) and the Harvard T.H. Chan School of Public Health (IRB18-1258).

Western blot analysis

Western blot analysis detected IgG antibodies against YFV and WNV in mothers previously identified as IgM-positive for DENV, ZIKV, or CHIKV. Given the potential for cross-reactivity among orthoflaviviruses, previous studies have shown that detecting envelope (E) and nonstructural protein 1 (NS1) may not reliably distinguish past infections [26]. To improve specificity, we determined antibodies against pre-membrane (prM) protein, which can distinguish four orthoflavivirus infections (DENV, ZIKV, WNV, and YFV), as diagnostic for the seroprevalence determination [26].

Vero cells were infected with YFV or WNV at a multiplicity of infection (MOI) of 1 and incubated at 37 °C until a 50% cytopathic effect was observed. Cells were lysed in NP-40 buffer containing protease and phosphatase inhibitors, incubated on ice for 30 min, and clarified by centrifugation (15,200 × g, 10 min, 4 °C).

Protein lysates (100 µL) were resolved on 4–15% SDS-PAGE gels (Bio-Rad, 5671082) and transferred to nitrocellulose membranes. Membranes were blocked in 4% non-fat milk in PBST, cut into strips, and probed overnight at 4 °C with maternal serum (10 µL/sample, diluted in 2% milk/PBST). Blots were washed, incubated with HRP-conjugated anti-human IgG (1:2000, Abcam, Ab97175) for 1 h at room temperature, and developed with DAB substrate (Thermo-Fisher, 34065). Molecular weight markers (Bio-Rad, 161-0374) were used to identify bands of interest, including E, NS1, and prM proteins. For equivocal or co-reactive samples, membranes were further tested using a half-membrane format containing lysates from six orthoflaviviruses (DENV1, DENV2, DENV4, WNV, ZIKV, and YFV), as described previously [26], to assess cross-reactivity across antigens.

Data visualization and statistical analysis

Data were managed in FileMaker Pro 18 (Cupertino, CA) and analysed using STATA/MP 18 (College Station, TX). Descriptive statistics summarized socio-demographic variables and arbovirus seroprevalence. Bivariate associations between maternal arbovirus reactivity and select characteristics (maternal age, trimester, hospital site, year of screening) with clinical outcomes were tested using two-tailed Fisher’s exact tests. Because of low WNV reactivity, WNV and YFV WB reactivity were assessed together, and YFV reactivity was additionally assessed alone for associations. After bivariate modelling, variables with p-values <0.25 in Chi-square or Fisher’s exact tests or with known clinical relevance to outcomes were included in multivariable logistic regression to estimate adjusted odds ratios (ORs) and 95% confidence intervals (CIs). Variables with p-values <0.05 in the adjusted logistic regression were considered significant.

Forest plots were generated to display adjusted associations with maternal symptoms and infant outcomes. UpSet plots illustrated overlapping IgG reactivities to ZIKV, DENV, YFV, WNV, and CHIKV. Kaplan-Meier survival analysis estimated the time to waning of maternal YFV IgG in infants, with median decay times calculated from the survival curve.

Role of the funding source

The funders of this study had no role in study design, data collection, data analysis, interpretation, writing, or decision to submit this report.

Results

Arboviral seroreactivity profiles among pregnant women in Nigeria

Among the 216 women in the final analytical cohort, selected based on IgM and /or IgG reactivity, IgM reactivity was most common for CHIKV (95/216, 44.0%), followed by DENV (83/216, 38.4%), and then ZIKV (34/216, 15.7%). Notably, among IgM-positive people, 3.7% (8/216) showed triple IgM positivity, indicating widespread co-infection. Additionally, 26.9% (58/216) were IgG-positive for at least one of the three viruses despite being IgM-negative, suggesting prior arbovirus exposure. The western blot analysis confirmed that 109 of 216 individuals (50.5%) were IgG-positive for YFV, while 11 individuals (5.1%) tested positive for WNV. YFV IgG was present across all groups with previous ZIKV, DENV, or CHIKV IgM reactivity, further confirming high levels of orthoflavivirus exposure. Seroreactivity to ZIKV, DENV, CHIKV, WNV, and YFV overlapped significantly based on the IgM and/or IgG results (Figure 1).

Figure 1.

Seroreactivity of Arboviruses Among Pregnant Women in Nigeria. (A) UpSet plot displaying the intersections of seroreactivity among ZIKV, CHIKV, DENV, YFV, and WNV. The bar graph represents the number of mothers with seroreactivity of arboviruses, while connected dots indicate virus combinations. (B) Representative Western blot images for YFV or WNV IgG positivity with other orthoflaviviruses in serological testing. The blot includes bands for envelope (E), nonstructural protein 1 (NS1), and pre-membrane (prM) proteins, with the prM band used to assess the orthoflavivirus exposure.

Maternal symptoms and YFV/WNV serostatus

Among 216 women, maternal arboviral symptoms were associated with several serological and demographic variables. YFV/WNV seropositive mothers had significantly higher odds of experiencing symptoms during a subsequent arbovirus infection compared to seronegative mothers (OR = 2.02, 95% CI: 1.35–3.02, P = 0.001) (Table 1; Figure 2A). When YFV reactivity was considered alone, the association remained statistically significant (OR = 1.77, 95% CI: 1.21–2.61, P < 0.004) (Supplementary Table 1). In contrast, DENV IgM reactivity was inversely associated with symptoms (OR = 0.515, 95% CI: 0.289–0.916, P = 0.024). Neither ZIKV IgM (P = 0.484) nor CHIKV IgM (P = 0.168) reactivity was significantly associated with symptoms.

Table 1.

Associations between yellow fever virus and/or West Nile virus seroreactivity and maternal symptoms upon subsequent arbovirus infection among pregnant women in Jos, Nigeria.

	Symptomatic		Fisher’s exact p-value	Logistic regression
	No, #(%)	Yes, #(%)		Odds ratio (95% CI)	p-value
YFV/WNV WB reactivity			0·03*
No	42 (41·2)	60 (58·8)		Ref	–
Yes	31 (27·2)	83 (72·8)		2·02 (1·35–3·02)	0·00*
ZIKV IgM reactivity			0·43
No	64 (35·2)	118 (64·8)		Ref	–
Yes	9 (26·5)	25 (73·5)		1·44 (0·52–3·98)	0·48
DENV IgM reactivity			0·08
No	39 (29·1)	95 (70·9)		Ref	–
Yes	34 (41·5)	48 (58·5)		0·52 (0·29–0·92)	0·02*
CHIKV IgM reactivity			0·31
No	37 (30·6)	84 (69·4)		Ref	–
Yes	36 (37·9)	59 (62·1)		0·67 (0·38–1·19)	0·17
Site			<0·0001*
JUTH	14 (14·6)	82 (85·4)
OLA	59 (49·2)	61 (50·8)
Year screened			0·22
2019	26 (28·9)	64 (71·1)		Ref	–
2020	28 (41·8)	39 (58·2)		0·58 (0·43–0·78)	<0·0001*
2021–22	19 (32·2)	40 (67·8)		0·82 (0·78–0·85)	<0·0001*
Age (years) a	30 [26–34]			1·00 (0·93–1·06)	0·88
Trimester			0·47
1st & 2nd (1-27 weeks).	42 (36·2)	74 (63·8)
3rd (≥ 28 weeks)	31 (31·3)	68 (68·7)

*Asterisks denote p < 0·05. Statistical analyses include Fisher’s exact test and logistic regression. ^a Median [interquartile range] is shown for Age (years).

Abbreviations: YFV, yellow fever virus; WB, Western blot; ZIKV, Zika virus; DENV, dengue virus; CHIKV, chikungunya virus; JUTH, Jos University Teaching Hospital; OLA, Our Lady of Apostles Hospital.

Figure 2.

Forest Plot of Arbovirus Seroreactivity and Associations with Maternal Symptoms and Infant Outcomes in Nigerian Cohort. (A) Odds ratios (ORs) and 95% confidence intervals (CIs) for the association between arbovirus reactivity and maternal symptoms. (B) Odds ratios (ORs) and 95% confidence intervals (CIs) for the association between arbovirus reactivity and abnormal infant outcomes.

Symptom prevalence also varied by study site, with a higher proportion of symptomatic women at JUTH compared to OLA (85.4% vs. 50.8%, P < 0.001). Screening year also impacted symptom odds, with women enrolled in 2020 (OR = 0.579, 95% CI: 0.430–0.779, P < 0.001) and 2021–2022 (OR = 0.817, 95% CI: 0.784–0.852, P < 0.001) less likely to report symptoms compared to those screened in 2019, indicating environmental and vector exposure differences over time. Neither maternal age nor trimester at screening was significantly associated with symptoms.

Infant outcomes and maternal arbovirus serostatus

Among 149 mothers with infant outcome data, abnormal infant outcomes were not significantly associated with maternal YFV/WNV IgG serostatus in neither Fisher’s exact test nor multivariable logistic regression (OR = 1.28, P = 0.503) (Table 2; Figure 2B). Similarly, YFV IgG seropositivity alone did not predict infant abnormalities (Supplementary Table 2). However, CHIKV IgM reactivity emerged as a significant predictor. Mothers positive for CHIKV IgM had over two times higher odds of having infants with abnormal outcomes (OR = 2.51, 95% CI: 1.48–4.20, P = 0.001). Maternal age emerged as a significant predictor (OR = 1.07, 95% CI: 1.02–1.11, P = 0.004), suggesting increased risk with advancing age. Trimester and screening year did not significantly affect abnormal infant outcomes.

Table 2.

Associations between yellow fever virus and/or West Nile virus seroreactivity and abnormal infant outcomes upon subsequent arbovirus infection among pregnant women in Jos, Nigeria.

	Abnormal infant outcome		Fisher’s exact p-value	Logistic regression
	No, # (%)	Yes, # (%)		Odds ratio (95% CI)
YFV/WNV WB reactivity			0·36
No	46 (75·4)	15 (24·6)		Ref	–
Yes	60 (68·2)	28 (31·8)		1·28 (0·62–2·64)	0·50
ZIKV IgM reactivity			1·00
No	86 (71·1)	35 (28·9)		Ref	–
Yes	20 (71·4)	8 (28·6)		1·01 (0·58–1·75)	0·98
DENV IgM reactivity			0·86
No	61 (79·1)	26 (29·9)		Ref	–
Yes	45 (72·6)	17 (27·4)		1·37 (0·73–2·58)	0·32
CHIKV IgM reactivity			0·03*
No	61 (79·2)	16 (20·8)		Ref	–
Yes	45 (62·5)	27 (37·5)		2·51 (1·45–4·35)	0·00*
Site			0·57
JUTH	32 (68·1)	15 (31·9)
OLA	74 (72·6)	28 (27·4)
Year screened			0·33
2019	51 (66·2)	26 (33·8)
2020	26 (72·2)	10 (27·8)
2021–22	29 (80·6)	7 (19·4)
Age (years) a	30 [26–33]			1·07 (1·02–1·11)	0·00*
Trimester			0·47
1st and 2nd (1-27 weeks).	60 (74·1)	21 (25·9)
3rd (≥28 weeks).	46 (67·7)	22 (32·3)

*Asterisks denote p < 0·05. Statistical analyses include Fisher’s exact test and logistic regression. ^a Median [interquartile range] is shown for Age (years).

Abbreviations: YFV, yellow fever virus; WB, Western blot; ZIKV, Zika virus; DENV, dengue virus; CHIKV, chikungunya virus; JUTH, Jos University Teaching Hospital; OLA, Our Lady of Apostles Hospital.

Comprehensive confirmatory arbovirus testing was outside the scope of this study, but we did perform microneutralization for ZIKV, DENV and CHIKV on 82 mothers and 6 infants and nucleic acid testing on 58 mothers and 1 infant [15]. Our microneutralization data did provide confirmation for ZIKV, DENV and CHIKV and enabled documentation of CHIKV perinatal transmission (Supplementary Table 3) [4].

Maternal immunity passive transfer

Among 96 infants with available follow-up samples, 51 were born to YFV IgG-seropositive mothers. Of these, 36 infants (70.6%) tested YFV IgG-positive at birth, indicating transplacental transfer of maternal antibodies. The longitudinal follow-up revealed a progressive decline in passively acquired IgG over time, with variable rates of antibody waning across individuals (Supplementary Figure 3).

By six months of age, 22 infants (22/26, 84.6%) had waning antibodies. Kaplan-Meier survival analysis estimated a median time to IgG waning of 3.1 months (IQR: 1.65-5.35 months) (Figure 3). The results demonstrate that over 50% of infants become YFV seronegative within the first four months, while most maternal antibodies begin to wane over 6–12 months [27]. These findings demonstrate the transience of maternally derived YFV immunity in infants and highlight the vulnerability of YFV exposure in early infancy.

Figure 3.

Persistence of Maternal YFV IgG in Infants in North. North-Central Nigeria. Kaplan–Meier survival curve showing the proportion of infants with detectable maternal YFV IgG over time. The blue line represents the survival estimate, and the shaded area represents the 95% confidence interval. The median time to IgG waning is 3.1 months (IQR: 1.65–5.35 months), as indicated by the red dashed lines.

Discussion

In this retrospective study of pregnant women and their infants, we observed high seroreactivity to YFV (109/216, 50.5%) and lower seroreactivity to WNV (11/216, 5.1%), based on IgM and/or IgG results (Figure 1A). YFV and WNV IgG seropositivity were significantly associated with increased risk of maternal symptoms during a subsequent arbovirus infection but not with risk of adverse infant outcomes. In a subset analysis of 36 infants, maternal YFV IgG waned with a median duration of 3.1 months (IQR: 1.65-5.35 months) and indicated limited protection of infants after waning since the YFV vaccine is not given until 9 months [28]. These findings raise important considerations about the sufficiency of maternally derived immunity in these highly endemic regions.

High IgM reactivity to CHIKV and DENV reflects intense natural transmission, whereas broad YFV IgG seropositivity likely reflects a combination of natural infection and widespread vaccination coverage through Nigeria’s national childhood immunization programme, established in 2004 [29]. The observed 50.5% YFV seropositivity in our cohort is notably higher than the predicted YFV-17D vaccine coverage (30-40%) in Nigeria up to 2016 [30], likely reflecting the recent surge in epidemics along with vaccine campaigns. In contrast, WNV IgG was detected in only a small subset of individuals (6.2%). Its low prevalence limited the statistical power to assess independent associations with maternal or infant outcomes. Most observed effects were, therefore, likely driven by YFV seroreactivity. These findings highlight a high degree of orthoflavivirus exposure from natural infections and vaccination, particularly for YFV and DENV. Given the endemic nature of these viruses, our results highlight the importance of continued serological surveillance to mitigate the burden of orthoflavivirus-related morbidity among pregnant women in high-risk regions.

The seropositivity to multiple viruses reflected the diagnostic complexity in endemic regions. This is particularly relevant for orthoflaviviruses, which share considerable sequence homology across structural (E, prM) and non-structural (NS1) proteins [31]. The live-attenuated YFV vaccine expresses both structural and non-structural proteins [32]. As a result, the YFV vaccine-induced broad immune response may be indistinguishable from natural infection using standard serological assays. This cross-reactive immune landscape complicates efforts to distinguish YFV vaccine-derived immunity from natural YFV infection, as well as to interpret recent orthoflavivirus infections in individuals with prior YFV vaccination. More specific diagnostics tools, such as virus-specific neutralization assays and multiplexed viral antigens or nucleic acid-based assays, are needed to improve orthoflavivirus surveillance and attribution of clinical diseases.

YFV and WNV IgG positivity were associated with 2.02 times the odds of maternal symptoms, even after adjusting for confounders. YFV reactivity alone did not reach significance in a univariate model but emerged as significant in multivariable regression (Supplementary Table 1). The multivariable regression models suggest possible synergistic effects of co-infection or immune priming from previous orthoflavivirus exposures. These results align with other studies from endemic settings, where past orthoflavivirus immunity exacerbates the severity of clinical symptoms, potentially through weakly neutralizing antibodies or in lower concentrations through Fc-mediated functions [33]. The observation raises the possibility that prior YFV or WNV exposure may influence inflammatory or symptomatic responses during pregnancy with other arbovirus coinfections. This is consistent with the prior finding where Japanese encephalitis virus (JEV) neutralizing antibodies were associated with increased symptomatic DENV illness in a paediatric cohort in Thailand [34].

We further examined whether YFV/WNV immunity modifies the effect of ZIKV and DENV IgM seroreactivity on maternal symptoms using a multiple logistic regression model with interaction terms between YFV/WNV IgG and individual IgM groupings (ZIKV only, DENV only, or both). While the interaction term was not significant, predictive margins revealed that among ZIKV/DENV IgM-positive mothers, those with YFV/WNV IgG had consistently higher predicted probabilities of symptoms compared to those without YFV/WNV IgG. The predicted probability of symptoms was 0.75 among mothers with both ZIKV and DENV IgM and YFV/WNV IgG compared to 0.45 in those with both IgM and no YFV/WNV IgG. Similarly, probabilities were higher among those IgM-positive for ZIKV only or DENV only with YFV/WNV IgG. However, interpretation of certain subgroups, such as IgM without YFV/WNV IgG, is limited by small sample size and model non-convergence. These findings suggest a potential effect modification by preexisting YFV/WNV immunity on symptom expression, which warrants further investigation in larger studies.

Despite the burden of maternal symptoms, neither YFV/WNV seroreactivity nor IgM reactivity with ZIKV or DENV was associated with adverse infant outcomes in this study. Instead, CHIKV IgM reactivity was robustly associated with abnormal infant outcomes (OR = 2.51). The association brings concerns about the neurotropic potential of CHIKV and its potential role in pregnancy complications [35]. However, the association between YFV and WNV with maternal symptoms but not with infant outcomes suggests the limitation of orthoflaviviruses’ vertical transmission in the presence of YFV or WNV prior exposures. Additionally, the rapid waning of maternal YFV IgG in infants raises concerns about a window of vulnerability before vaccine-induced immunity is established. Infants may be unprotected during outbreaks occurring in the first months of life. While considerably higher than WNV, YFV IgG seroprevalence was still only 50.5% indicating that half of mothers have no immunity to transfer to their infants. This suggests the revision of maternal vaccination strategies for women of childbearing age and education on safe mosquito protection and environmental control interventions such as protective clothing, mosquito nets, and safe repellents, at least until 9 months, the age of YFV vaccination, for newborns and new mothers in endemic regions.

Our study has limitations related to the lack of temporal resolution for exposure due to retrospective design. The variability in follow-up timing and incomplete infant outcome data may have reduced the power of associations. While it was beyond the scope of our study and assay to distinguish YFV infection from live-attenuated vaccine, we believe that our findings of an association between YFV immunity (whether natural or vaccine-derived) and symptoms in acute orthoflavivirus infection are important to report and indicate the need for further research with larger sample sizes and tests that can distinguish between natural and vaccine-derived YFV immunity. Finally, our study population included only pregnant women positive for ZIKV, DENV, and CHIKV IgM and IgG and their infants, which enabled us to look at the role of YFV and WNV immunity in infection with those viruses, but we did not have a negative control group. Future studies will need prospective longitudinal cohort designs with better-defined infection timing, maternal-fetal transmission dynamics, and durability of passive immunity.

Nonetheless, this study provides novel seroprevalence data on YFV and WNV immunity among pregnant women in Nigeria and on the persistence of passively transferred maternal YFV IgG antibodies in neonates. The increased odds of arboviral symptoms among those with YFV and/or WNV antibodies during pregnancy suggest the possibility of ADE in secondary orthoflavivirus infections, which indicates the need for further studies. Understanding how immunity from prior vaccination or infection affects pregnancy outcomes and neonatal vulnerability remains essential for optimizing vaccination policies and clinical care in endemic settings.

Funding Statement

Motsepe Presidential Research Accelerator Fund and NIH (R21AI137840, R01AI149502).National Institute of Allergy and Infectious Diseases

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The deidentified data underlying the results of this study will be stored along with the data dictionary in the Harvard Dataverse repository (https://dataverse.harvard.edu) at the time of publication. Researchers can request access to the participant-level dataset through the Harvard Dataverse for review and approval by the study investigators.

Supplemental Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/22221751.2025.2544720.