Improving Replication in Endometrial Omics: Understanding the Influence of the Menstrual Cycle

Abstract

The dynamic nature of human endometrial tissue presents unique challenges in analysis. Despite extensive research into endometrial disorders such as endometriosis and infertility, recent systematic reviews have highlighted concerning issues with the reproducibility of omics studies attempting to identify biomarkers. This review examines factors contributing to poor reproducibility in endometrial omics research. Hormonal fluctuations in the menstrual cycle lead to widespread molecular changes in the endometrium, most notably in gene expression profiles. In this review, we examine the variability in omics data due to the menstrual cycle and highlight the importance of accurate menstrual cycle dating for effective statistical modelling. The current standards of endometrial dating lack precision and we make the case for using molecular-based modelling methods to estimate menstrual cycle time for endometrium tissue samples. Additionally, we discuss statistical considerations such as confounding and interaction effects, as well as the importance of recording the detailed and accurate clinical information of patients. By addressing these methodological challenges, we aim to establish more robust and reproducible research practises, increasing the reliability of endometrial omics research and biomarker discovery.

Keywords: endometrium; experimental design; gene expression; omics.

Figure 4

PCA plots from nine RNA-seq studies from the GEO database. Count data were obtained from the GREIN (GEO RNA-seq Experiments Interactive Navigator) platform [54], and molecular model time was obtained using the endest R package [27]. Samples were plotted using the study-provided cycle time (Figure 4ai,bi,ci,di,ei,fi,gi) and molecular time (Figure 4aii,bii,cii,dii,eii,fii,gii). Two studies had no provided cycle time in their sample metadata and were plotted using molecular time only (Figure 4h,i). The PCA plots reveal that the variance explained by the first two principal components generally has a stronger concordance with the molecular model timing compared to histological or LH+ dating.

Figure 1

Examples of genes that significantly change in the menstrual cycle. Gene expression from RNA-seq with samples histologically dated from the menstrual (M), early proliferative (EP), mid-proliferative (MP), late proliferative (LP), early secretory (ES), mid-secretory (MS), or late secretory (LS) phase. Each data point represents an endometrial sample taken from a unique patient. Data are from GEO Series GSE234354 [27].

Figure 2

A PCA plot of RNA-seq gene expression data from endometrial tissue samples from across the menstrual cycle. The variance observed in the first two principal components primarily reflects menstrual cycle-related changes in gene expression. Data are from GEO Series GSE234354 [27].

Figure 3

Examples of genes that significantly change expression in the menstrual cycle identified using molecular methods for estimating menstrual cycle time. Each data point represents an endometrial sample taken from a unique patient. Time 0 corresponds to the beginning of menstruation and time 100 corresponds to the conclusion of the secretory phase. Gene expression from RNA-seq with cycle time was estimated using the endest R package [27]. Data are from GEO Series GSE234354 [27].

Figure 5

An illustration of the effects of cycle time being confounded with experimental groups. All samples are histologically dated as mid-secretory and are all control samples. The samples are separated into two groups to perform differential gene expression and three different combinations of comparisons are shown. p-value histograms are useful for diagnosing issues [59]. Under the null, we expect a flat distribution, as seen in hypothetical experiment 2. The composition of different samples for each experimental group can determine the distribution of p-values, e.g., conservative, flat, and anti-conservative distributions. Severe imbalances such as in hypothetical experiment 1 can result in an anti-conservative p-value distribution which can result in identifying false positive effects that are due to cycle effects. In a real experiment, it would not be possible to distinguish if the effect is due to the confounding variable or the effect of interest.