Maternal plasma microRNA profiles in twin-twin transfusion syndrome and normal monochorionic twin pregnancies

Abstract

Introduction: Ultrasound-based staging systems for twin-twin transfusion syndrome (TTTS) are limited by radiologic expertise, fetal positioning, and timing of the exam, and may benefit from incorporation of objective biochemical measures for diagnosis and prognostication. microRNA expression is altered in amniotic fluid of TTTS patients, however the invasive nature of amniocentesis has precluded practical incorporation of these biomarkers into current staging systems. Therefore, we sought to assess whether non-invasive maternal plasma microRNAs can distinguish between TTTS and normal monochorionic diamniotic (MCDA) twin pregnancies.

Methods: Maternal blood samples were collected for patients with normal MCDA twin pregnancies (n = 11) or prior to selective fetoscopic laser photocoagulation (SFLP) for patients with TTTS (n = 36). Extracted microRNA from a panel of 24 microRNAs was compared between groups.

Results: miR-26a-5p (P = 0.004), miR-222-3p (P = 0.007), and miR-145-5p (P = 0.047) were downregulated and miR-320a-3p (P = 0.005) was upregulated in the maternal plasma of TTTS patients compared to controls. miR-26a-5p, miR-320a-3p, and miR-222-3p in combination were strong predictors of TTTS on random forest modeling (area under curve = 0.905). After SFLP, all significantly dysregulated microRNAs in TTTS trended toward levels of expression observed in control MCDA twin pregnancies.

Conclusion: Several microRNAs are differentially expressed in maternal plasma and demonstrate strong predictive capacity for identifying twin-twin transfusion syndrome. These plasma microRNAs could provide minimally invasive means to enhance currently established ultrasound diagnostic criteria for twin-twin transfusion syndrome.

Keywords: circulating biomarker; microRNA; monochorionic diamniotic pregnancy; pregnancy; twin-twin transfusion syndrome.

FIGURE 1

Differential expression of miRNAs in controls versus TTTS. (a) Bar graphs depicting relative miRNA expression of the top 5 differentially expressed miRNAs between controls (n = 11) and TTTS (n = 36). Individual patients represented as dots, and bars represent mean ± standard deviation. Asterisk denotes Benjamini–Hochberg adjusted P-value <0.05. (b) Heat map reflecting relative ddCT of respective miRNA in controls versus TTTS patients. Green boxes represent higher ddCT and red boxes represents lower ddCT, and therefore lower and higher relative expression compared to the globally normalized reference, respectively.

FIGURE 2

Random forest analysis of the predictive capacity of miRNAs for TTTS. (a) Mean decrease in accuracy and mean decrease in Gini coefficients reflect the respective predictive capacity of each miRNA for identifying TTTS. The top three miRNAs in both models, hsa-miR-320a-3p, hsa-miR-26a-5p and hsa-miR-222-3p were utilized in the random forest model for their combined predictive capacity for TTTS. (b) ROC curve for the top three miRNAs in the random forest model depicting strong predictive capacity for TTTS as indicated by an area under the curve (AUC) of 0.905. (c) Hierarchical clustering of patients using the model demonstrates two clusters composed entirely of patients with TTTS, and one cluster containing both TTTS and control patients.

FIGURE 3

Comparison of miRNA expression at multiple time points with respect to SFLP response. (a) Characteristics of the patient cohort. At initial presentation, patients referred to the fetal treatment center were either confirmed to have TTTS requiring SFLP (n = 30) or had an MCDA pregnancy without TTTS or with TTTS wherein SFLP was not indicated (n = 17). All patients in the latter group returned for repeat examination and were found to either have evolving TTTS (n = 6) or no further evidence of TTTS (n = 11). A total of 36 patients with TTTS underwent SFLP, and 17 patients underwent maternal blood sample collection on post-SFLP follow-up. (b) Relative miRNA expression of the top 5 differentially expressed miRNAs before and after SFLP for the 17 patients who provided both pre- and post-SFLP blood samples, with controls depicted as a reference. Bars represent mean ± standard deviation. Single asterisk denotes P < 0.05, double asterisk denotes P < 0.001, n.s. denotes P > 0.05. (c) Matched patient samples for patients with TTTS pre- and post-SFLP demonstrating changes in relative miRNA expression for each patient.

FIGURE 4

Network of genes regulated by the top 5 differentially expressed miRNAs in TTTS and putative gene ontogeny pathways. (a). Network analysis depicts the top 5 differentially expressed miRNAs in TTTS (green) and respective genes regulated by three (red), two (yellow), or one (blue) miRNAs. (b). Chord diagram depicting gene ontogeny (GO) pathways enriched in gene:miRNA sets.