Inflammatory mechanisms of menstrual cycle changes following COVID-19 vaccination in adolescents.

Participant recruitment and enrollment has been described in detail elsewhere [ 16 ]. Briefly, participants were recruited from social media advertisements, a posting on the institution’s webpage, through other ongoing studies, and through word-of-mouth referrals. The inclusion criteria were: 1) females aged 13–20 years; 2) self-reported menstrual cycle averaging 22–35 days; 3) menarche at least 12 months prior to participation and having regular cycles for at least 6 months; and 4) previous receipt of initial COVID-19 vaccine. Exclusion criteria were: 1) the use of oral contraceptives or any exogenous hormones in the 3 months prior to participation; 2) presence of factors indicative of secondary dysmenorrhea (e.g., self-reported diagnosis of endometriosis, presence of persistent pelvic pain throughout the month); 3) diagnosis of chronic pain condition; 4) currently pregnant or breastfeeding; 5) history of pelvic inflammatory disease or sexually transmitted disease; 6) developmental delay, diagnosis of autism, or significant cognitive impairment that may preclude understanding of study procedures. See Fig. 1 for the participant flow diagram. The study was approved by the Mass General Brigham Institutional Review Board (protocol 2021P002543). Potential participants were screened for eligibility by telephone; eligible participants (and a parent of a minor) were sent an email with the Information Sheet to review and a Research Electronic Data Capture (REDCap; [ 25 , 26 ]) survey link to the online consent/assent/permission page. After completing the enrollment survey link, participants were mailed a packet including printed instructions, the saliva sample collection supplies (2.0 mL cryovials and Saliva Collection Aids, Salimetrics, LLC, Carlsbad, CA), and a separate plastic bag for freezer storage of completed samples. A video call for the first sample collection (T1; see Fig. 2 ) was arranged to occur 0–24 h prior to receiving the booster. T1 video calls began with the participant rinsing their mouth with water. During the 10 min between mouth rinse and sample collection, participants were instructed how to set up the mobile app (see below) and told which days over the following 3–4 weeks they would be collecting samples. Participants were provided a 1-page “study calendar” where they wrote down sample collection dates and determined best times for text message reminders. Study calendars were customized depending on the anticipated vaccine brand (either three or four weeks for Pfizer-BioNTech and Moderna vaccines, respectively; see below). Research team members verbally described the study procedures and sent the questionnaire link by email to complete the questionnaire immediately after freezing the saliva sample. Vaccine brand was confirmed the day after vaccination, and T5 collection dates were modified as needed. After collecting T5, participants returned the frozen saliva samples and were compensated $20 for each saliva sample collected (i.e., up to $100). Participants used a mobile app, OnTimePoint ™ (OTP) developed by Salimetrics, LLC, to log when they completed each saliva sample. Participants logged into OTP using a study ID code and participant code that was given to them during the T1 video call. No identifying information was entered into OTP, either by the participant or by the researcher using the OTP web portal. OTP logs the date and start/stop time of each sample, displays a visual picture of a vial to show participants how much saliva to collect, and reminds them to put the sample in the freezer when they’ve completed each collection. Participants who did not consent to receive text messages received the reminders via email. The pre-booster sample (T1) was collected as described above (within the 24 h prior to vaccination). T2 was scheduled for the calendar day after receiving the booster (approximately 24 h post-vaccination), and T3 was scheduled for the following day (i.e., approximately 48 h post-vaccination). T4 was collected 14 days after the booster, and T5 was collected either 21 days (Pfizer-BioNTech) or 28 days (Moderna) after vaccination (see Fig. 2 ). Ovarian hormones estradiol (E2) and progesterone were assayed at all five time points. Pro-inflammatory cytokines were assessed at T1, T2, and T3, and included IFNγ, IL-1β, and IL-6. Antibodies of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2, the virus that causes COVID-19), including anti-nucleocapsid immunoglobulin G (IgG) [anti-nucleocapsid IgG], anti-S1 receptor-binding domain (RBD), and anti-spike IgG, were assayed at T1, T4, and T5. The Salimetrics ELISA immunoassay is specifically designed and validated for the qualitative measurement of human IgG specific to the SARS-CoV-2 anti-nucleocapsid IgG. All samples were tested for cytokines and anti-SARS-CoV-2 antibodies using commercial chemiluminescent immunoassay kits (Meso Scale Diagnostics, Rockville, MD) according to manufacturer’s instructions, using minimum required sample dilutions specific for saliva and adjustments made to the standards to accommodate lower analyte levels in saliva. All assays were conducted twice on each sample, and results of the two measurements were averaged to yield the final value. For quality assurance, samples were repeat-tested if the percent coefficient of variation (CV%) of each duplicate test was above 15%. Concentrations were determined by interpolation from standard curves that met the criteria of an R 2 equal to or above 0.99. Controls also were required to read within established ranges for each assay. Demographic variables (e.g., date of birth, race, ethnicity, etc.) and menstrual cycle characteristics (e.g., age at menarche, first day of bleeding for each menstrual cycle, and severity of menstrual symptoms; [ 27 ]) were assessed using an instrument designed for this study. As shown in Fig. 3 , menstrual cycle length was determined by calculating the number of days between the first day of bleeding for the three menstrual cycles prior to vaccination (Cycles B-3, B-2, B-1), as well as for the cycle during which the vaccine was received (Cycle B), and three cycles later (Cycles B + 1, B + 2, and B + 3). For additional details on collection of menstrual cycle dates pre- and post-vaccination, see [ 16 ]. In our initial analysis, we observed high rates of missing data at time points T3 and T4 (mean missing rate more than 40%). Due to the extent of missingness and its potential impact on the reliability of longitudinal comparisons, we decided to exclude data from these time points and focus our analyses on earlier assessments where data completeness was substantially higher (T1, T2, and T5). We conducted three sets of analyses on participants without recent SARS-CoV-2 infection at baseline, defined as having a baseline (T1) anti-nucleocapsid IgG measure below 2.2 AU/mL. Individuals with anti-nucleocapsid IgG levels greater than 2.2 ( n = 14) were excluded from the analyses due to likelihood of having a recent SARS-CoV-2 infection. This threshold was determined by previous Salimetrics, LLC testing of pre-COVID saliva samples and represents three standard deviations above the mean for each antibody detected. Spiked serum from confirmed SARS-CoV-2 individuals and tested serum confirmed seropositive individuals to validate the assay. The assay is catalog number K15383U ( https://www.mesoscale.com/products/sars-cov-2-panel-2-igg-k15383u/ ) and the assay is validated for saliva by MSD. The three analyses were as follows: We used multiple linear regression to examine the relationship between post-booster changes in menstrual cycle length and baseline (T1) SARS-CoV-2 antibodies and cytokine measures. The outcome was the difference between a specific post-booster cycle length (each B, B + 1, B + 2, and B + 3) and the average cycle length over the three pre-booster cycles (average of B-3, B-2, and B-1). We also considered the difference between the average cycle length across all four post-booster cycles (B through B + 3) and the pre-booster average cycle length (B-3 to B-1). Using multiple linear regression, we identified predictors of two biomarkers—levels of anti-spike IgG at T5 and IL-1β at T2—indicative of immune response to and inflammatory burden of booster shots, respectively. Covariates included the baseline levels of these two biomarkers, E2, progesterone, and the interactions of E2 and progesterone with the baseline levels of anti-spike IgG and IL-1β. To identify distinct immune response profiles, we conducted a clustering analysis based on anti-spike IgG levels at T5, which reflects the immune response to the booster vaccination. We used the K-medoids approach and selected the optimal number of clusters by maximizing the average silhouette width. We visualized the distributions of anti-spike IgG at T1 and T5 for each cluster. Univariate logistic regression was then performed to assess differences between clusters in clinical and demographic variables, which include age, race, age of menarche, menstrual symptom severity, post-booster menstrual cycle length changes, E2, progesterone, and cytokines and SARS-CoV-2 antibodies. There were two individuals who were missing saliva samples or data at T5 and were thus excluded from these analyses. For analyses 1 and 2, we handled missing covariates using multiple imputation with chained equations (R package mice; [ 28 ]). We set the number of imputations at five and for each imputation, 10 iterations were used. We applied Rubin’s rule [ 29 ] to combine results across five imputations. For all analyses, we did not correct for multiple comparisons due to the small sample size of our study.

Baseline characteristics of the study cohort are presented in Table 1 . The study sample included 47 participants who received a booster vaccination during the study period. At the time of the booster vaccination, the average time since receipt of the initial vaccine series was 7.63 months ( SD = 1.17; range = 5.31–11.54). Based on the results in Table 2 , we found that the only significant association was between the IL-1β level at T2 and the changes in cycle length of the cycle in which the booster was received (Cycle B). The association was estimated to be negative, which indicates a higher T2 level of IL-1β was associated with shorter post-booster cycle length. This significant result was almost preserved two cycles later (B + 2, p -value = 0.056), but not for the rest of the post-booster cycles. Table 3 summarizes the results from the regression analysis. None of the predictors were statistically significantly associated with the two outcome measures of interest, nor their respective baseline measures. The lack of associations with the baseline measures might be due, in part, to the heterogeneity of the participants’ response profiles (see below). Specifically, a group of participants in our study exhibited little to no immune response after booster vaccination. In Fig. 4(a) , we plotted levels of anti-spike IgG at T1 and T5 for participants with both measures available. Two distinct patterns emerged: one where the level remained mostly unchanged, and one with a large increase from T1 to T5. The K-medoids analysis confirmed the existence of two well-separated clusters based on the levels of anti-spike IgG at T5 (the low and high clusters), and the high cluster with a higher average T5 anti-spike IgG levels are also the group of participants with a large increase in anti-spike IgG level post-booster ( Fig. 4a ). The distribution of the baseline anti-spike IgG levels of these two clusters revealed that the high cluster tended to have lower T1 anti-spike IgG levels than the low cluster ( Fig. 4b ). The regression analysis, however, did not find any significant contrasts between the two clusters in terms of their T1 and T2 profiles ( Table 4 ). The strongest difference we observed was for the menstrual symptoms at T1 with a p -value = 0.084. We also considered whether the different response profiles may have been due to vaccination type (brand), so we conducted a Fisher’s exact test to assess the association between vaccine type and cluster identity. The result was not statistically significant ( p = 1), indicating that cluster identity is not simply a reflection of vaccine type.

This study investigated the role of inflammatory responses to the COVID-19 vaccine in menstrual irregularities among adolescents. Although it is well-known that vaccines provoke a controlled immune response [ 17 ], including the activation of pro-inflammatory cytokines, the implications of this response on reproductive health are understudied, particularly in adolescent populations. Our findings revealed that IL-1β plays a significant role in explaining the shortening of the menstrual cycle following COVID-19 vaccination in adolescent girls, primarily impacting the cycle in which the vaccine was received and possibly two cycles later. This provides evidence for a link between vaccine-induced inflammation and menstrual cycle changes. This finding is particularly relevant given the complex interplay between the immune and reproductive systems. The menstrual cycle is regulated by a delicate balance of hormones, and inflammatory cytokines such as IL-1β can influence this balance by interacting with components of the HPO axis [ 30 ]. When the COVID-19 vaccine activates the immune system, there is a strong production of pro-inflammatory cytokines, including IL-1β [ 31 ]. IL-1β is known to inhibit the release of gonadotropin-releasing hormone (GnRH) from the hypothalamus, which disrupts its usual pulsatile signaling [ 32 , 33 ]. GnRH pulses most rapidly during the follicular phase to promote follicle development [ 34 , 35 ]. This signaling is crucial for stimulating the pituitary gland to produce luteinizing hormone (LH) and follicle-stimulating hormone (FSH), which regulate ovarian function [ 32 , 35 ]. As such, the suppression of GnRH can disrupt the menstrual cycle through alterations of the hormonal rhythms necessary for proper follicle development and ovulation. This disruption can manifest in several ways. Inflammation during the follicular phase has been linked with anovulation [ 23 ], and insufficient LH and FSH production appear to reduce ovarian steroidogenesis, leading to lower levels of estrogen and progesterone [ 34 ]. The lack of a progesterone surge can lead to premature reductions in additional hormones, shortening the luteal phase and facilitating early menstruation [ 36 ]. This is known clinically as “short luteal phase” and “luteal phase defect.” Alternatively, inflammation can also lead to delayed ovulation, which is thought to be due to either slowed or delayed follicle development, leading to a longer follicular phase and typical-length luteal phase [ 37 ]. The observation of shortened menstrual cycles in adolescent girls differs from the frequently-documented lengthening seen in adult women (e.g., [ 7 , 8 , 38 ]). We speculate inflammation may affect menstrual cycles differently in adults compared to adolescents due to the maturity of their HPO axes. In adults, whose HPO axes are fully developed, inflammation-induced suppression of LH and FSH appears to be more likely to prolong the menstrual cycle (for a review, see [ 8 ]). Low FSH can delay follicular development and ovulation, while insufficient LH can impair corpus luteum function and progesterone production, extending the follicular phase [ 34 ]. The mature and complex feedback mechanisms in adults could contribute to hormonal disruptions resulting in cycle lengthening rather than shortening. During adolescence, however, the HPO axis is still developing [ 39 ], which likely contributes to variations in menstrual cycle length. Indeed, among adolescents, 21- to 45-day menstrual cycles are considered typical as their bodies establish regular menstrual rhythms [ 40 ]. Longitudinal research by Sun and colleagues [ 41 ] found that, among 23 healthy, early postmenarchal adolescent girls, the hormonal composition of the menstrual cycle differed from that which is typically seen in adults. For example, girls were found to have lower levels of estradiol, progesterone, and gonadotropins during the luteal phase than adults. The authors also found that many of the girls’ menstrual cycles were characterized by luteal insufficiency, which they defined as a short luteal phase and/or low levels of progesterone. Given this lower baseline of hormone levels and tendency towards shorter luteal phases, when inflammatory cytokines disrupt the menstrual cycle, it is possible that the adolescent body may preferably shorten, as opposed to lengthen, the menstrual cycle. These distinctions underscore how age-related differences in HPO axis regulation influence menstrual responses to inflammation. It is also possible that individual immune responses to vaccination could influence menstrual side effects. Indeed, it is well-documented that greater immune activation following vaccination is associated with greater side effects (e.g., [ 42 – 45 ]). In the present study, two distinct clusters of vaccine response as measured by anti-spike IgG were found: those with high and low immune responses. There was a modest, positive association between menstrual symptoms and anti-spike IgG, though this finding did not reach statistical significance so must be considered with caution. It is possible that a larger sample size could have yielded results more consistent with the literature (i.e., that greater side effects follow greater immune response). Alternatively, this finding may suggest that downstream inflammatory signaling, as seen in IL-1β elevation, plays a more direct role on the menstrual cycle than do overall anti-spike IgG levels. Interpretation of baseline (T1) anti-spike IgG in the cluster analyses warrants caution. Most T1 values were low and compressed near the assay floor, which limits between-person ranking. Consequently, the apparent tendency for the “high” follow-up cluster to start lower at T1 than the “low” follow-up cluster ( Fig. 4b ) should not be over-interpreted as evidence that the “low” cluster failed to rise because of an already-established immune response. Typical secondary (memory) responses to mRNA boosters are rapid and robust across serostatus strata; fold-changes are often smaller when pre-booster titers are higher (ceiling effects), not larger, and booster responses are generally seen even in those with minimal pre-booster signal. This pattern is well documented across cohorts and vaccine platforms, including in Frontiers in Immunology reports [ 46 ]. Cluster differences should be interpreted descriptively rather than as causal reflections of pre-existing immunity. In the future, earlier post-booster timepoints (e.g., days 7–14) may be sampled when secondary responses typically peak, to better resolve kinetics and reduce interpretive ambiguity. Despite these immune differences, we did not observe significant associations between changes to menstrual cycle length and baseline ovarian hormone levels, despite our previous paper which found that participants experienced shorter cycles when the booster was received during the luteal phase compared to the follicular phase [ 16 ]. This may suggest that the effect of inflammation on menstrual cycle length is not associated with co-occurring shifts in progesterone and estradiol. It is also possible that the timing of salivary hormone sampling did not align with peak points of hormonal variation, limiting our ability to detect relevant changes or that absolute ovarian hormone levels may not reflect an accurate identification of cycle phase. More frequent sampling or different methods of sampling (e.g., serum) may be necessary to capture a more comprehensive picture of the effect of vaccine-related inflammation on ovarian hormones. There are a number of limitations that are worth mentioning. First, the current study included a sample of individuals who had previously received the initial vaccine series. Even though participants were required to wait at least six months between the initial series and the booster, the exposure to previous vaccination may have impacted results. An additional limitation is the non-harmonized T5 sampling (Pfizer 21 days, Moderna 28 days post-booster), inherited from earlier brand schedules. This 7-day offset may bias absolute titers and complicate cross-brand comparisons at T5; accordingly, we reported T5 results descriptively and avoid brand-to-brand inference. Our primary finding, the association between 24-h IL-1β and booster-cycle shortening, does not depend on T5 and is unaffected. Future work may standardize post-dose windows or model kinetics directly. Third, inflammatory markers were assessed in saliva samples, which may capture oral inflammation and may be different from inflammatory markers in serum samples. We also did not assess other recent vaccinations, non-COVID viral infections, or other health events that may impact inflammation, immune, or menstrual cycle functioning. Additionally, these findings may not generalize to other populations, including those with chronic health or inflammatory conditions or those using exogenous hormones. Finally, the sample size was relatively small and did not include a control group, so findings must be interpreted with caution. Future research should further explore the role of immune responses in menstrual regulation across different age and demographic groups. Replicating these findings in larger, more diverse populations would help support the generalizability of these results. Furthermore, in light of our finding that there were different levels of immune response to the vaccine, future research is needed to better understand whether responsiveness to vaccination is associated with changes in menstrual characteristics, including pain. Lastly, integrating menstrual health into routine vaccine safety studies could provide more comprehensive data and enhance public and scientific understanding of vaccine-related side effects. This study sheds light on the direct relationship between immune system activation and altered menstrual functioning in adolescent girls. It was found that COVID-19 vaccination led to shortened menstrual cycle lengths in girls, and IL-1β mediated this effect. These findings enhance our understanding of the biological mechanisms underlying vaccine-related menstrual changes and align with prior research indicating that such disruptions are typically short-lived. As the relationship between immune responses and reproductive health continues to be explored, it is important to incorporate menstrual health into vaccine safety monitoring and research.

Following the widespread distribution of Coronavirus Disease (COVID-19) vaccines, global reports emerged regarding alterations to menstrual cycles following vaccination (e.g., [ 1 ]). Early media coverage highlighted cases of longer and shorter cycles and heavier and more painful menstrual periods post-vaccination [ 2 , 3 ], alongside worries about potential implications for fertility [ 4 ]. These concerns sparked scientific inquiry into the relationship between COVID-19 vaccination and women’s menstrual and reproductive health outcomes (e.g., [ 5 , 6 ]), leading to global research efforts that identified changes in adult women’s menstrual cycles (e.g., [ 7 ]). Data from adult women have consistently identified several common menstrual differences following vaccination, including increased cycle length, heavier bleeding, and greater menstrual pain, compared to unvaccinated cohorts (e.g., [ 7 , 8 ]). On average, women experience a slight lengthening of their menstrual cycles after vaccination - typically about one day - with some studies indicating increases of up to eight days [ 9 – 11 ]. These changes are more pronounced when the vaccine is administered during the follicular phase of the menstrual cycle [ 10 , 12 ], suggesting that the timing of vaccination relative to the menstrual cycle and circulating sex hormones may play a key role in these changes. These findings also indicate that menstrual disturbances appear to be transient, with no evidence suggesting long-term effects on fertility [ 13 , 14 ]. Much less research has focused on adolescent menstrual cycle changes, although one study found that 25.1% of vaccinated adolescents reported at least one menstrual change following vaccination, such as heavier bleeding, shorter cycles, longer cycles, and increased pain [ 15 ]. However, 22.6% of participants also reported irregularities prior to vaccination, suggesting that these changes may partly reflect the natural variability of menstrual cycles during adolescence. Payne and colleagues [ 16 ] reported pre- and post-vaccination menstrual cycle data in a sample of 65 adolescent girls, comparing those who received a COVID-19 booster vaccine to a control group who did not receive a booster vaccine during the study timeframe. Among the booster group, post-booster cycles were, on average, five days shorter than their pre-booster cycles. This change was observed only in those who were vaccinated during the luteal phase of their menstrual cycles and was driven by a finding of shorter cycles in the second post-booster cycle. These findings highlight the need to better understand how adolescent girls may be uniquely affected by COVID-19 vaccination, as well as potential causes of these observed alterations. The mechanisms involved in menstrual cycle alterations following COVID-19 vaccination are unknown, but it is hypothesized that the changes are due to interactions between inflammatory and immune processes triggered by the vaccine. Vaccination induces a multi-system inflammatory response (for a discussion, see [ 17 ]), including the release of pro-inflammatory cytokines, such as interferon gamma (IFNγ), interleukin-1 beta (IL-1β), interleukin-6 (IL-6), interleukin-8 (IL-8), and tumor necrosis factor-alpha (TNF-α), which play central roles in regulating immune and systemic functions [ 18 ]. While the precise effects of these cytokines on the hypothalamic-pituitary-ovarian (HPO) axis remain unclear, it is well-established that inflammation can disrupt neuroendocrine signaling [ 19 ]. As the HPO axis is predominantly regulated by neuroendocrine signaling [ 20 , 21 ], such disruptions could influence reproductive function. Supporting this hypothesis, a study investigating the effects of COVID-19 vaccination on menstrual cycles and serum anti-Müllerian hormone (AMH) levels reported temporary menstrual irregularities in some women following vaccination [ 22 ]. This study also reported declines in AMH, a marker of ovarian reserve, during the first six months post-vaccination, with levels returning to baseline by nine months. These findings indicate that COVID-19 vaccination can temporarily alter the menstrual cycle and HPO axis at the neuroendocrine level. However, this study did not explore the role of specific pro-inflammatory cytokines or the timing of vaccination within the menstrual cycle, leaving the mechanisms of this effect unknown. Emerging research suggests that inflammation occurring during the follicular phase may disrupt the HPO axis [ 23 ], potentially altering cycle length, flow, and pain. This phase, which precedes ovulation, appears to be particularly sensitive to hormonal and inflammatory fluctuations. Disruptions during this window may therefore have a greater effect on menstrual patterns. This aligns with research indicating that menstrual cycle changes are most pronounced in women who received the COVID-19 vaccination during the follicular phase [ 10 , 12 , 16 ]. Although inflammation has frequently been hypothesized as an explanation for menstrual cycle changes, this link has yet to be explicitly studied as a mechanism involved in menstrual cycle alterations following COVID-19 vaccination, particularly in adolescents who are, by definition, experiencing many changes in HPO axis functioning during this developmental transition [ 24 ]. As such, the aim of the present study is to assess the role of inflammatory processes prior to and in the 24 h post-COVID-19 booster vaccination and determine the relationship of inflammation to observed menstrual cycle changes post-vaccination among adolescents. We also aimed to determine whether ovarian hormones or menstrual cycle characteristics measured at baseline are associated with inflammatory and immune responses to booster vaccination. Specifically, we hypothesized that inflammatory markers would be associated with changes in menstrual cycle length following COVID-19 booster vaccination and that ovarian hormones would be associated with both inflammatory and immune response to the booster vaccine.