A Noninvasive Method to Sample Immune Cells in the Lower Female Genital Tract Using Menstrual Discs

Abstract

T cells in the human female genital tract (FGT) are key mediators of susceptibility to and protection from infection, including HIV and other sexually transmitted infections. There is a critical need for increased understanding of the distribution and activation of T cell populations in the FGT, but current sampling methods require a healthcare provider and are expensive, limiting the ability to study these populations longitudinally. To address these challenges, we have developed a method to sample immune cells from the FGT utilizing disposable menstrual discs which are noninvasive, self-applied, and low in cost. To demonstrate reproducibility, we sampled the cervicovaginal fluid of healthy, reproductive-aged individuals using menstrual discs across 3 sequential days. Cervicovaginal fluid was processed for cervicovaginal cells, and high-parameter flow cytometry was used to characterize immune populations. We identified large numbers of live, CD45⁺ leukocytes, as well as distinct populations of T cells and B cells. Within the T cell compartment, activation and suppression status of T cell subsets were consistent with previous studies of the FGT utilizing current approaches, including identification of both tissue-resident and migratory populations. In addition, the T cell population structure was highly conserved across days within individuals but divergent across individuals. Our approach to sample immune cells in the FGT with menstrual discs will decrease barriers to participation and empower longitudinal sampling in future research studies.

Introduction

In the female genital tract (FGT), the cervix and vagina are the primary sites of exposure to HIV and other sexually transmitted infections (1, 2). Local T cells at these sites are key mediators of both susceptibility to and protection from infection (3). For example, nonspecific inflammation may lead to T cell activation and increased risk of HIV acquisition (2, 4–7). In contrast, robust Th1 responses are needed for clearance of pathogens such as Chlamydia trachomatis (8). In addition, T cells patrolling the FGT must tolerate resident microbes and other foreign Ags such as those from semen (5, 9–11). Prior phenotypic and transcriptional profiling of mucosal T cell populations has demonstrated that they are highly adapted to their site of residence and are distinct from those in circulation, emphasizing the critical need to sample these populations directly from the site of interest (12–14). For example, limited work on the FGT describes an association between cervical T cell populations, menstrual cycle, and microbial community state; however, these studies have been restricted to small sample sizes (15–17). Despite the intriguing associations between the FGT and these biological variables, it has been challenging to comprehensively characterize their complex interplay in diverse populations due to the difficulty in sampling the FGT longitudinally.

Common methods for sampling immune cell populations in the FGT include biopsy, cytobrush, and cervicovaginal lavage (CVL) (18, 19). Tissue collected either from biopsy or discarded surgical tissue is considered the “gold standard” for obtaining large cell yields and allows for precise sampling of anatomical sites (18, 19). This approach has identified two key T cell populations in the FGT: tissue-resident memory (T_RM) cells and migratory cells, which circulate between the tissue, regional lymph nodes, and peripheral blood (17, 20, 21). However, these important studies were restricted to small sample sizes, and most were conducted at a single time point. Cytobrushes have been used to elucidate endocervical CD4⁺ and CD8⁺ T cell populations including the impact of HIV (22–26), but the collection site is anatomically distinct from ectocervical biopsy and may preferentially sample monocytes over lymphocytes (18, 27, 28). In addition, both biopsy and cytobrush may disrupt the mucosal barrier and temporarily increase susceptibility to or transmission of infection, including HIV (29). CVL, in contrast, collects T cells present in the cervical fornixes and vaginal lumen (i.e., “luminal” T cells), and it does not disrupt the mucosal barrier of the FGT. However, CVL yields substantially fewer cells than cytobrush and tissue biopsy sampling (18). Finally, each of these methods requires implementation by a healthcare provider in a clinical setting with a speculum examination, which can be both uncomfortable and expensive, limiting the number of participants and time points for longitudinal sampling.

In this study, we present a noninvasive approach to sample luminal T cells from the FGT, utilizing self-inserted menstrual discs to collect cervicovaginal fluid (CVF) (Fig. 1A, 1B). Menstrual discs are a disposable personal hygiene product designed to collect menstrual fluid. In the research setting, they have been used for the study of cells present in menstrual blood (30, 31). In addition, they can be used to collect CVF in individuals while not menstruating (32–34). For example, menstrual discs have been used previously for the study of cytokines and other soluble markers in CVF (34–36). Given the position of the disc, menstrual discs are expected to collect both cervical secretions and fluid pooled in the vaginal fornixes. We have found that this approach yields high numbers of immune cells including T cells and captures both tissue-resident and migratory populations with reproducible distributions and phenotypic features within individuals. This inexpensive, noninvasive approach does not require a clinical setting and thus reduces barriers associated with sequential sampling and empowers a larger and more diverse population of individuals to participate in future studies of FGT immunity.

FIGURE 1.

Menstrual disc sample collection and processing overview.

(A) Softdisc with firm plastic ring and soft plastic cup used to collect CVF. (B) Placement of the menstrual disc across the cervix and positioned to collect CVF. Images in (A) and (B) are subject to copyright (2023) by The Flex Company (www.flexfits.com) (all rights reserved). Softdisc is a registered trademark of The Flex Company. (C) Overview of method for processing and analyzing CVF for immune cells. The figure was created with BioRender.com.

Materials and Methods

Cohort

All participants were enrolled into the Center for Global Infectious Disease Research Biorepository, approved by the Seattle Children’s Institutional Review Board (STUDY00002048). Participants self-reported as healthy, not pregnant, reproductive age, weighing >110 pounds, and none had an intrauterine device or a known hormonal disorder. To validate the reproducibility of our protocol, five participants were invited to donate CVF samples on 3 sequential days, 7–11 d from the last menstrual period during the follicular phase, as well as one blood sample. One participant (participant 3) did not menstruate due to oral contraceptive use, so the timing was not relative to the last menstrual period. Information on acute and chronic medical conditions and medication usage was also collected.

Sample collection

Whole blood was drawn into EDTA vacutainer tubes (BD Biosciences). Blood was processed for PBMCs as previously described (37). For CVF collection, Softdisc menstrual discs were self-inserted by participants and worn for up to 4 h. After self-retrieval, menstrual discs were placed in a 50-ml conical tube and returned to the study team. CVF was stored at 4°C until processing.

Protocol for processing of CVF from menstrual discs

Discs were initially processed in 50-ml conical tubes returned by participants (Fig. 1C). Discs were submerged in 10 ml of complete RPMI 1640 (RPMI 1640 with l-glutamine, 10% FBS, 100 U/ml penicillin, 100 µg/ml streptomycin) and rinsed five times in the medium using a serological pipette to dislodge CVF. The tube containing the disc was centrifuged at 250 × g for 7 min at 4°C. The menstrual disc was then removed from the tube using sterile forceps and the remaining CVF was transferred from the disc to the tube by pipette. The catch basin of the menstrual disc was rinsed with 1 ml of sterile PBS with 2% FBS (FACS buffer), which was transferred to the tube, and the disc was discarded. Mucins were degraded by incubating CVF with 1 mM DTT for 30 min in a 37°C incubator with 5% CO₂, as described (38). The solution was then filtered through a 100-µm nylon cell strainer (VWR) and centrifuged at 400 × g for 10 min at 4°C and the supernatant was discarded. The resulting cell pellet of cervicovaginal cells (CVCs) was washed twice with 40 ml of sterile PBS with 2% FBS, and cells were enumerated using a C-chip hemocytometer (INCYTO). For cryopreservation, the cell pellet was resuspended in 1 ml of freezing medium (50% FBS, 40% RPMI 1640 with l-glutamine, 10% DMSO [MilliporeSigma]), transferred to a cryovial, placed in a 1°C cryogenic freezing container at −80°C overnight, and then transferred to liquid nitrogen.

Prior to flow cytometric analysis, CVCs and PBMCs were thawed in a 37°C water bath, transferred to prewarmed thaw medium (RPMI 1640 with l-glutamine, 20% FBS), and centrifuged at 400 × g for 5 min. Cell pellets were resuspended in 2 ml of complete RPMI 1640 and enumerated. CVF was enriched for CD45⁺ cells using the EasySep release human CD45 positive selection kit (STEMCELL Technologies) according to the manufacturer’s protocol. The resulting CD45-depleted fraction underwent a second enrichment to recover any remaining CD45⁺ cells, and the two CD45-enriched fractions were combined.

Cell staining and flow cytometry acquisition

All flow cytometry was performed at Fred Hutchison Cancer Center utilizing a 28-color panel focused on T cell phenotyping (Supplemental Table I) (39). CVCs and PBMCs were incubated with Fc-blocking reagent (BioLegend) and fixable UV blue Live/Dead reagent (Thermo Fisher Scientific) in PBS for 15 min at room temperature. Cells were then stained with 50 µl of extracellular Ab master mix in brilliant stain buffer (BD Biosciences) for 20 min at room temperature. Stained cells were washed and resuspended in FACS buffer. For intracellular and intranuclear staining, the cells were fixed with freshly prepared fixation buffer (Thermo Fisher Scientific) for 30 min at room temperature, washed, and then stained with 50 µl of intracellular Ab master mix in FACS buffer for 20 min at room temperature. Cells were then washed and resuspended in FACS buffer and kept at 4°C protected from light until analysis. Single-stained compensation controls were prepared for each experiment using Ab capture beads (BD Biosciences) diluted in FACS buffer and amine reactive beads (ArC amine reactive compensation bead kit, Themo Fisher Scientific). Data were acquired using a FACSymphony A5 (BD Biosciences) and FACSDiva acquisition software (BD Biosciences). All samples were run in the same experiment to eliminate batch effects.

Flow cytometry analysis

Flow cytometry data were analyzed using FlowJo v10.8 software. The predetermined gating scheme was applied uniformly across all samples. Initial panel development included fluorescence minus one to set gates (39). A technical replicate sample that had been validated in prior experiments with the same panel and instrument was included to inform gating. Briefly, paired CVCs and PBMCs were gated on CD45⁺ and viability, then CD3⁺ and CD19⁺ populations (Fig. 2A). After gating out mucosal-associated invariant T and γδ T cells, CD3⁺ T cells were gated into CD4⁺ and CD8⁺ T cell populations. Within the CD4⁺ T cell population, regulatory T cells (Tregs) were designated as CD25^hi/CD127^low, and the remaining cells were designated as conventional CD4⁺ cells. CD8⁺, CD4⁺, and CD4 Treg populations were then gated for memory, tissue residency, activation, and suppression markers. Phenotypic analysis was restricted to samples where the parent population of interest contained ≥50 T cells. Cell count, frequency, and mean fluorescence intensity were determined in FlowJo.

FIGURE 2.

Overview of CVC T cell populations.

Data from five individuals sampled across 3 d. Results for samples with >50 cells in the parent population for each analysis are presented. (A) Representative gates for identification of T cells and other immune subpopulations in CVC (top) and PBMC (bottom) samples. γδ T cells and mucosal-associated invariant T cells were excluded from the T cell population prior to gating into CD4⁺ and CD8⁺ subpopulations. (B) After gating out Tregs (CD25^high, CD127^low), conventional CD4⁺ T cells were further gated by CCR7 and CD45RA to define effector memory (T_EM; CD45RA⁻CCR7⁻), central memory (T_CM; CD45RA⁻CCR7⁺), terminally differentiated effector memory (T_EMRA; CD45RA⁺CCR7⁻), and naive (CD45RA⁺CCR7⁺) populations for PBMC (left) and CVC (right) samples. (C) CD8⁺ T cells were also gated by CCR7 and CD45RA to determine memory subtypes, as well as by CD69 and CD103 to determine tissue residency (CD69⁺CD103⁺). (D) Frequency of conventional CVC CD4⁺ memory subtypes across samples. (E) Frequency of CVC CD8⁺ memory subtypes across samples. (F) Frequency of CVC CD8⁺ T_RM cells across samples. (G) Frequencies of CVC Tregs, conventional CCR7⁺, and conventional CCR7⁻ cells of all CD4⁺ T cells. (H) PD-1 expression on CVC CD4⁺ T cell subsets. (I) CCR5 expression on CVC CD4⁺ T cell subsets.

Integrated high-parameter data analysis was restricted to CVC samples containing ≥100 T cells. Comparison of global T cell population structure across samples was first conducted in R v4.3.2 using principal component (PC) analysis (PCA) computed from mean fluorescence intensity of phenotypic markers relevant to viable conventional T cells (i.e., live/dead, CD45, CD19, MR1-tet, and γδ TCR were excluded). PCA plots were visualized using ggplot2 (40). Initial comparison included both CVCs and PBMCs, where only memory T cell populations were considered (i.e., naive T cells excluded), as well as subsequent analysis considering only CVCs. Comparison of local T cell population structure across all samples was performed using FlowSOM and ConsensusClusterPlus in the integrated CATALYST R package (41). T cells across CVC samples were clustered using median fluorescent intensity of the T cell–directed markers used in PCA with a maximum metacluster number of 20. The number of clusters examined for subsequent analysis (k = 13) was determined by cluster stability measure when the relative change in area under a cumulative distribution function curve approached 0. The aggregated marker expression within T cell clusters was scaled from 0 to 1 and visualized with a heat map; hierarchical clustering is shown with dendrograms representing Euclidean distance and plotted with average linkage. Relative abundances of the clusters in each CVC sample were plotted to demonstrate reproducibility in population structure within individuals across samples. Fluorescence intensity data for each cell was dimensionally reduced and visualized using uniform manifold approximation and projection (UMAP) without the default CATALYST data transformation. The resulting UMAP was colored by FlowSOM-assigned clusters.

Statistical analysis

Cell count data (e.g., number of CD3⁺ T cells recovered) are presented as medians. Frequencies of parent populations (e.g., percentage Tregs of CD4⁺ T cells) are presented as means. To compare frequency of populations between CVCs and PBMCs, a mean frequency of the CVC population across all sampled days included in analysis was computed. The mean frequency of the population in the CVCs was then compared with the frequency in paired PBMCs. Given the small sample size and paired data structure, p values were computed with a Wilcoxon signed rank test. Significance was defined as a p value ≤0.05. Analysis was conducted in Stata 14.2.

Results

Overview of immune cells recovered from CVF

Total number of viable CD45⁺ cells recovered from CVF samples varied both between and within individuals (Table I). The median number of viable CD45⁺ cells recovered from CVF was 6235. Within the CD45⁺ cell population, CD3⁺ T cells were the largest subpopulation (median of 1182 cells), but there was also high recovery of CD19⁺ B cells (median of 1072 cells) (Table I). Of CD3⁺ T cells, frequencies ranged from 37 to 87% for CD4⁺ T cells and from 7 to 55% for CD8⁺ T cells. Most CVC T cells were Ag experienced (mean, 97% of conventional CD4⁺ cells and 91% of CD8⁺ cells), including central memory (CCR7⁺CD45RA⁻), effector memory (CCR7⁻CD45RA⁻), and terminally differentiated effector CD45RA⁺ populations (CCR7⁻CD45RA⁺) (Fig. 2B–E). The CD8⁺ compartment of CVCs contained both T_RM (CD69⁺ CD103⁺) (Fig. 2F) and migratory populations, and CD8⁺ T_RM cells were significantly enriched in CVCs relative to PBMCs (mean, 15 versus 0.03%, p = 0.04). In addition to CCR7⁺ and CCR7⁻ conventional populations, CVC CD4⁺ T cells also contained a Treg population (Fig. 2G), the frequency of which was enriched compared with PBMCs (mean, 14 versus 4%, p = 0.04). Similar to conventional CD4⁺ T cells, 98% of Tregs were Ag experienced. Within CVC CD4 T cells, the expression of PD-1 and CCR5 was highest on Tregs, relative to conventional CD4⁺ T cells, and was consistent across sampling days within individuals (Fig. 2H, 2I).

Table I. Counts of immune cell populations across CVF samples.

Participant:	1			2			3			4			5
Sample:	1	2	3	1	2	3	1	2	3	1	2	3	1	2	3	Median
CD45+	14,772	11,157	57,703	4533	13,401	20,587	1225	2036	11,860	6233	212	31,479	2885	965	2762	6233
CD3+	3,174	1,182	11,554	1052	1,966	3,512	38	392	3,897	1808	12	7,823	521	56	384	1182
CD4+	2,037	938	7,484	518	1,197	1,460	14	223	2,037	1567	5	5,449	350	34	271	938
CD8+	902	157	3,404	191	331	659	21	118	1,387	163	2	968	77	4	31	163
CD19+	2,166	727	3,638	1072	2,070	2,599	51	456	1,220	1548	8	4,690	170	25	201	1072
CD3−/CD19− HLA-DR+	5,539	4,082	12,849	605	3,052	3,726	482	392	3,019	611	32	3,251	775	64	452	775

Activation and suppression phenotype of CVC T cells

We next compared the activation and suppression phenotype of specific FGT T cell populations of interest reported in prior studies, specifically CD4⁺ Tregs and memory CD8⁺ T cells (20, 42). Relative to PBMCs, a higher frequency of CVC CD4⁺ Tregs expressed the activation marker CCR5 (mean, 77 versus 39%, p = 0.04) and the suppressive markers ICOS (mean, 54 versus 21%, p = 0.04), PD-1 (mean, 47 versus 5%, p = 0.04), LAG-3 (mean, 30 versus 4%, p = 0.04), CTLA-4 (mean, 38 versus 7%, p = 0.04), and TIM-3 (mean, 42 versus 4%, p = 0.04) (Fig. 3A, 3B). The Treg compartment also contained populations of PD-1⁺TIM3⁺ cells (10–45% within samples), a highly suppressive phenotype observed in tissue Treg populations and rarely observed in the periphery (43). Within CVC memory CD8⁺ T cells there was a higher frequency of cells expressing CCR5 (mean, 85 versus 25%, p = 0.04), HLA-DR (activation) (mean, 40 versus 11%, p = 0.04), and PD-1 (mean, 34 versus 7%, p = 0.04) relative to PBMCs (Fig. 3C, 3D). Although the frequencies of memory CD8⁺ T cells expressing CD38 (mean, 23 versus 11%, p = 0.08) and granzyme B (mean, 32 versus 25%, p = 0.7) was higher in CVCs versus PBMCs, these differences did not reach significance. Taken together, these observations indicate that menstrual disc sampling yielded key populations of CD4 and CD8 T cells with similar levels of activation and suppression as previously reported from current sampling methods.

FIGURE 3.

Characterization of key CVC T cell populations.

Data from five individuals sampled across 3 d (CVCs) and 1 d (PBMCs). Results for samples with >50 cells in the parent population for each analysis are presented. The mean CVC frequency of each population of interest across sampled days was compared with the PBMC frequency using a Wilcoxon signed rank test. (A) Representative flow plots of activation and suppression marker expression from paired CVC and PBMC CD4⁺ Treg populations. (B) Summary plots of frequencies of activation and suppression markers between paired CVC and PBMC CD4⁺ Treg populations. (C) Representative flow plots of activation and suppression marker expression from paired CVC and PBMC memory CD8⁺ T cells. (D) Summary plots of frequencies of activation and suppression markers between paired CVC and PBMC memory CD8⁺ T cells. *p ≤ 0.05.

Reproducibility of T cell population structure

To quantify the reproducibility of global CVC T cell population structure across sequential time points within individuals, we compared memory T cell phenotypes across consecutive days from the FGT, as well as paired PBMCs from one time point, using PCA. When comparing CVC and PBMC memory T cell populations, samples clustered clearly by sample type (Fig. 4A). Within CVC CD3⁺ and CD4⁺ T cell populations, by comparison, samples clustered by individual (Fig. 4B, 4C). There was no clear relationship between day of sampling and clustering pattern for any analysis. Within CD3⁺ populations, the largest contributors to PC1 as determined by loading scores were CD4, activation markers (CD28 and CD38), and suppressive markers (PD-1, CTLA-4, ICOS, and TIM3); the largest contributors to PC2 were memory markers (CD45RA, CCR7, and CD27) and the proliferation marker Ki67. Within the CD4⁺ population, the largest contributors to PC1 were activation markers (CCR5, CD69, CD38, and CD28) and suppressive markers (PD-1, CTLA-4, and TIM3); the largest contributors to PC2 were also activation (HLA-DR, CD137, and CD25) and suppressive (PD-1 and LAG3) markers. Taken together, these observations emphasize that individuals have unique patterns of memory, activation, and suppression markers on CVC T cells that are highly conserved across days.

FIGURE 4.

CVC T cell populations are reproducible within individuals but vary across individuals.

Data from five individuals sampled across 3 d (CVCs) and 1 d (PBMCs). Results for samples with >100 cells in the parent population for each analysis are presented. (A) PCA utilizing high parameter flow data to characterize paired CVC and PBMC CD3⁺ memory T cell populations. (B) PCA characterizing CVC CD3⁺ memory T cell populations. (C) PCA characterizing CVC CD4⁺ memory T cell populations. (D) Unsupervised hierarchical heat map characterizing the median expression of CVC T cell markers across individuals and samples. (E) Dimensional reduction of all markers relevant to T cells was performed and is represented as a UMAP including all CD3⁺ T cells across CVC samples. (F) Heat map characterizing the median expression of T cell markers in each identified cluster. (G) Proportions of cell clusters composing each sample, arranged by participant and visit.

We next examined local CVC T cell population structure across participants using the CATALYST tool. Consistent with our PCA analysis, initial analysis looking at marker expression demonstrated two groups of participants separated by degree of T cell activation (e.g., CCR5 expression was high in participants 1, 3, and 5 and low in participants 2 and 4) (Fig. 4D). We next clustered T cells across all individuals to understand shared population structure. Across all participants, luminal T cells were comprised of two large populations (CD4⁺ and CD8⁺) and a few minor populations, which could be subdivided into 13 phenotypic clusters (Fig. 4E, 4F). Clusters 1, 5, and 7 had high expression of CD4 and high expression of activation markers including CCR5, CD69, and CD137 and the tissue residency marker CD103. In particular, cluster 1 had exceptionally high activation and was distinct from the other CD4 clusters. Clusters 6, 8, and 9 were also CD4 predominant, but with lower degrees of activation and variable expression of CD45RA. Cluster 10 and 11 had CD4 expression, but showed very low activation, with cluster 10 displaying very high CTLA-4 expression and cluster 11 very high LAG-3 expression. Cluster 2 contained both CD4 and CD8 populations, with moderate activation and no expression of suppression markers. Clusters 3 and 4 were CD8 predominant, and both expressed CD103 consistent with T_RM populations. Clusters 12 and 13 were quite distinct with low expression of both CD4 and CD8. Within individuals, the distribution of clusters was recapitulated across time points (Fig. 4G). Taken together, these observations indicate that luminal T cells from the FGT are comprised of diverse populations that vary by tissue residence, activation, and suppressive status and that the distribution of these populations within individuals is preserved across sequential days.

Discussion

The cervix is a critical site for exposure to and acquisition of HIV and other sexually transmitted infections. It is thus essential to understand the immune cell populations at this site, as well as how these populations change longitudinally and in response to variables such as hormonal fluctuations, microbial communities, hygiene practices, and sexual behaviors. In this study, we describe a noninvasive method for obtaining large numbers of live CD45⁺ immune cells from CVF using menstrual discs. These samples are rich in T cells and an abundant source of B cells and HLA-DR⁺ populations, which remain viable after cryopreservation. The number of T cells we recovered was similar to cytobrush, which recovers ∼500 to 5,000 T cells (18, 26), but lower than tissue biopsy, which recovers ∼5,000 to 10,000 T cells (18). Of note, compared with prior reports of cytobrush and biopsy (18, 44, 45), the observation of significant numbers of CD19⁺ B cells appears unique to our sampling approach.

To compare our FGT sampling method to those reported in the literature, we conducted in-depth phenotyping of the T cell populations. We identified FGT T cell populations that closely resembled those previously identified in biopsy and surgically resected tissue, including both migratory and tissue-resident populations (20, 42). Consistent with prior reports, both conventional CD4 and CD8 populations had high expression of CCR5, indicating activation, and for CD4⁺ cells representing an HIV target cell population (17, 20). In cytobrush samples, prior reports have found that ∼65–75% of all CD4⁺ T cells are CCR5⁺ (26, 46), similar to our observation that 57% of all CD4⁺ T cells expressed CCR5. In addition, Tregs were enriched in CVCs and showed high expression of suppressive markers, consistent with prior reports (42). Similarly, we identified CD8⁺ T_RM cells, although the overall frequency was lower relative to those reported from tissue (15% versus 50–80% of total CD8 T cells) (13, 20). These data indicate that our sampling approach identifies key T cell subsets with phenotypic features consistent with prior approaches.

Importantly, our sampling method collects luminal T cells, similar to prior work utilizing CVL. In addition to representing the underlying tissue, data from lung suggest that luminal T cells may contain unique populations, which have discrete function via their ability to migrate between the lumen and the tissue (47–49). More recently, studies using CVL in nonhuman primates have identified a unique population of memory CD4⁺ T cells that migrate into the lumen of the FGT in a CCR5-dependent manner (17). In the context of the cervix and vagina, luminal T cells may preferentially encounter bacteria present as part of vaginal microbiota versus pathogenic bacteria and viruses, which invade the tissue. In this light, an independent study of luminal T cells, not merely as a representation of tissue, may yield novel insights into the immunology of the cervix and vagina.

Intraindividual reproducibility is critical for any sampling method, so we assessed the reproducibility of our approach across three sequential days. Despite variability in absolute viable leukocyte recovery, the T cell population structure was maintained within each individual. Within individuals, we identified high reproducibility of both the global and local population structure, including the distribution of CVC T cell populations, and specific phenotypic characteristics (e.g., the expression of PD-1 on conventional CD4⁺ T cells). In contrast, we observed significant interindividual variability, wherein each participant had a unique T cell population structure that was reproducible across time. These observations suggests that differences observed across individuals may be driven by biological or behavioral variability. This highlights the need for further study to evaluate the role of specific factors in driving CVC T cell population structure across individuals.

Our study has a number of limitations; for example, the absolute number of leukocytes collected varied across days. This may reflect differences in cervical secretions day to day, variable positioning of the disc, or changes in activity level. Reassuringly, despite variation in total collected leukocytes, the relative population structure was preserved. Future implementation of this approach should consider the need for repeat sampling to ensure sufficient cell recovery for the intended application. Furthermore, we did not collect sufficient T cells to perform both phenotypic and functional work and elected to first perform deep phenotyping. Future studies should address functional capacity as well as Ag specificity of collected T cells. In addition, different sampling methods will sample different anatomical zones and yield different populations. The high CD4/CD8 cell ratio of our CVC T cells is most consistent with cervical origin (18, 21, 27), but it is noteworthy that CVF collected via menstrual disc contains cells that are of less precise anatomical origin than endocervical cytobrush or tissue biopsy. Our flow cytometry panel was primarily targeted at T cell phenotyping, so we were not able to assess the relative lymphocyte-to-monocyte frequency; however, we did identify a large population of CD3⁻CD19⁻HLA-DR⁺ cells that fell within the expected forward and side scatter of monocytes (18). Finally, participants may be unfamiliar with the use of menstrual discs, with a risk of incorrect placement in the vagina rather than surrounding the cervix. To address this in future studies, study team members could demonstrate correct positioning of the disc prior to utilization.

We have established a method for sampling of cervicovaginal immune cells that is low cost, noninvasive, and self-administered. This method can be used in longitudinal studies and does not require study participants to interact with a healthcare provider or a clinical setting, reducing the barrier to participation. The high population reproducibility, low cost, and ability to cryopreserve these samples with sufficient cell recovery further emphasizes the potential application of this approach in future studies, such as in large-scale vaccine trials that may take place at multiple sites.

Disclosures

The authors have no financial conflicts of interest.