Maternal plasma microRNAs as potential biomarkers for triaging pregnancies of unknown location and ectopic pregnancy diagnosis.

ClinicalTrials.gov NCT04738370 , NCT04176549 .

Christopher Kyriacou: Writing – review & editing, Writing – original draft, Validation, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Sung Hye Kim: Writing – review & editing, Validation, Methodology, Formal analysis, Conceptualization. Maria Arianoglou: Writing – review & editing, Resources, Methodology. Shabnam Bobdiwala: Writing – review & editing, Data curation. Margaret Pikovsky: Writing – review & editing, Data curation. Nina Parker: Writing – review & editing, Data curation. Jennifer Barcroft: Writing – review & editing, Data curation. Maya Al-Memar: Writing – review & editing, Data curation. Phillip R. Bennett: Writing – review & editing, Visualization, Supervision. David A. MacIntyre: Writing – review & editing, Visualization. Tom Bourne: Writing – review & editing, Visualization, Supervision, Funding acquisition, Conceptualization. Vasso Terzidou: Writing – review & editing, Visualization, Supervision, Project administration, Methodology, Investigation, Formal analysis, Conceptualization.

Imperial Health Charity; March of Dimes; National Institute for Health Research Imperial Biomedical Research Centre.

Women alone were included in this work given the nature and relevance of the conditions under investigation. Women with a positive urine pregnancy test attending the early pregnancy assessment units (EPAU) at Imperial College Healthcare NHS Trust in London, UK, between July 2018 and December 2020 were recruited following classification of PUL or diagnosis of EP, mostly between four to eight weeks' gestation. TVUS ± PUL triage was performed as per local clinical guidelines until an outcome was confirmed, and ectopic pregnancies, miscarriages (NVIUP) and viable intrauterine pregnancies were included in the analysis. VIUP were confirmed at the pregnancy dating scan, usually at 12–14 weeks'gestation. FPUL and PPUL were excluded from analysis due to their heterogeneity (FPUL and PPUL cohorts are likely to contain both failing or persistent intrauterine or ectopic pregnancies that cannot be discerned using TVUS). Whole blood samples were collected in EDTA tubes (K 2 EDTA BD Vacutainer®), and plasma was isolated by centrifugation at 1962× g for 10 min at 4 °C (Megafuge 40R, Thermo Scientific, Waltham, USA). Isolated plasma was stored in 500 μL aliquots in RNase-free microtubes at −80 °C. The samples used in this project were collected following the approval of two studies by the Health Research Authority, the Health and Care Research Wales Research Ethics Committee, and either the North of Scotland Research Ethics Committee, or the London-Dulwich Research Ethics Committee. Their references are 14/NS/1078 (Assessment of Biomarkers in Pregnancy of Unknown Location and Ectopic pregnancy – ABPEP: ClinicalTrials.gov NCT04738370 ) and 19/LO/1154 (Assessment of Biomarkers in Ectopic Pregnancy – AMBER: ClinicalTrials.gov NCT04176549 ). Written informed consent was received prior to participation. Findings for this observational cohort study have been reported in accordance with the STROBE guidelines. Raw count data from the Nanostring nCounter miRNA profiling assay are available in Supplementary Table S1 . RNA extraction was performed using Quick-cfRNA Serum & Plasma Kit (Zymo Research, Orange, USA – RRID: SCR_008968 ). Plasma samples were thawed on ice prior to centrifugation at 12,000× g for 15 min to remove any cell debris or precipitates as per manufacturer's instructions. The upper 200 μL of plasma was used for the RNA extraction to avoid cellular and/or platelet contamination. Samples were evaluated for haemolysis macroscopically, and via the use of spectrophotometric measurement of haemoglobin at 414 nm using Nanodrop 1000 (Thermo Scientific, Waltham, USA – RRID: SCR_016517 ). Samples with readings above 0.2 were deemed at risk of haemolysis. In accordance with the manufacturer's protocol, the spike-in control cel-miR-254 (Qiagen, Hilden, Germany – RRID: SCR_008539 ) was added to each sample following lysis and denaturation of plasma with Quick-cfRNA Digestion Buffer to allow normalisation of any technical variation that may have occurred during RNA extraction. The RNA samples for the discovery cohort were evaluated using nCounter cell-free miRNA profiling (Nanostring, Seattle, USA – RRID: SCR_021712 ), allowing direct fluorescence-based molecular barcoding of 798 individual miRNAs per sample without amplification. The nCounter Human v3 miRNA panel was used for each sample and it accounts for more than 95 % of all observed sequencing reads in miRbase version 21 ( https://www.mirbase.org/ ), including all miRNAs denoted as ‘high confidence’ [ 62 ]. Counts of miRNAs were analysed using nSolver version 4.0 (Nanostring, Seattle, USA – RRID: SCR_003420 ) software, with normalisation to the 100 most expressed targets. Normalised miRNA counts were then compared between the outcome groups, EP, NVIUP, and VIUP. The miRNAs expressed below background level in more than 50 % of at least one outcome group were excluded from analysis (background level defined as two standard deviations above mean negative control counts). The raw count data is available in Supplementary Table S1 . Reverse transcription and RT-qPCR was used for technical validation of the discovery work, as well as for the biological validation study which utilised a second, separate validation cohort. Following a literature review, three additional miRNA markers were also added to the biological validation work. Reverse transcription of the extracted RNA samples was achieved using the miRCURY LNA Universal RT miRNA cDNA synthesis kits following the manufacturer's instructions (Qiagen). As part of the protocol, the spike-in control, UniSp6, was added (Qiagen) to allow normalisation of any variation during reverse transcription. Endogenous miRNA controls were identified using NormFinder (Microsoft, Redmond, USA – RRID: SCR_003387 ) software. The most stable combination of miRNAs was chosen for the normalisation of RT-qPCR data [ 63 ]. The stability values created by NormFinder using miRNA counts profiled using nCounter and analysed using nSolver (Nanosting) reflect the intra- and inter-group variation in the expression of each individual and combination of miRNAs. The combination of two miRNAs with greatest stability were used. The NormFinder analysis is also available in Supplementary Table S1 . RT-qPCR was performed as per manufacturer's instructions using custom PCR panels with embedded LNA™ primers (Qiagen, Hilden, Germany – RRID: SCR_008539 ), and miRCURY LNA SYBR® Green master mix (Qiagen, Hilden, Germany – RRID: SCR_008539 . RT-qPCR was performed in duplicate for the technical validation samples. RT-qPCR reactions were carried out on the StepOnePlus™ Real-Time PCR system (Applied Biosystems, Waltham, USA – RRID: SCR_015805 ). The cycle conditions included initial PCR heat activation at 95 °C for 2 min, followed by 45 cycles of denaturation at 95 °C for 10 s and annealing/extension at 56 °C for 60 s. Melt curve analyses from 60 to 95 °C confirmed a single PCR product. Raw fluorescence data were collected using the StepOne Software version 2.3 (Applied Biosystems, Waltham, USA – RRID: SCR_014281 ) for evaluation of Cq values and primer efficiencies using the LinRegPCR program version 2017.1 [ 64 ]. Following this, Cq data was processed by normalising to the inter-plate calibrator (UniSp3), UniSp6 reverse transcription control, cel-miR-254 RNA extraction control and the endogenous controls. Univariate analyses were then performed to compare miRNA fold changes between the different outcome groups. Any participant samples or miRNAs that did not consistently amplify were excluded. For the remainder, any normalised Cq values above 45 or below one were given a value of 45 (the maximum number of cycles in PCR reaction). Statistical software used included nSolver version 4.0 (Nanostring) and PRISM version 8.0 (GraphPad, San Diego, USA – RRID: SCR_002798 ). Demographics (age, ethnicity, gestational age to initial TVUS, and Body Mass Index – BMI) for all patient groups were tested for normality using the Shapiro-Wilk test with mean (standard deviation) and median (interquartile range) appropriately presented. One-way ANOVA/Kruskal-Wallis analyses were selected depending on data distribution, with categorical data analyses using Fisher's. nSolver (Nanostring) statistical analysis allowed comparison of log-transformed geometric means of expression counts assuming unequal variance between different outcome groups in the discovery work using an unpaired two-tailed t -test. MiRNAs with p  < 0.05 and a false discovery rate (FDR) < 0.05 were considered to be differentially expressed between outcomes. Count data were plotted on the logarithmic scale using geometric mean and geometric standard deviation (SD), with statistically significant p values presented. For the technical and biological validation work, univariate analyses of the relative expression fold changes for each miRNA were compared between outcome groups. Data from all miRNA candidates in each outcome group were tested for normality using the Shapiro-Wilk test, and unpaired two-tailed t -test/Mann-Whitney U analysis was selected depending on data distribution. Fold change data were plotted on the logarithmic scale using geometric mean and geometric SD, with statistically significant p values presented. For ROC curves, area under the ROC curve (AUC) with 95 % confidence intervals (95 % CI) were also included. Significance was again determined by p value, with a level <0.05 deemed significant. Whilst no formal power calculation was performed due to the discovery nature of the study, the study population numbers were comparable to similar published work exploring miRNAs in fetal growth restriction and preterm birth [ 19 , 21 ].

A total of 290 women were recruited from EPAU and 120 women (41.4 %) with outcomes of either ectopic pregnancy, miscarriage (NVIUP) or viable intrauterine pregnancy were used for further analysis ( Fig. 1 ). 170 women (58.6 %) were excluded from this study due to their outcome: 141 failed PUL, 9 persistent PUL, 8 IUP of uncertain viability and 12 lost to follow up (LTFU). For the discovery cohort and technical validation using RT-qPCR, 22 women (18.3 %) were selected for analysis, where 7 had EP, 7 NVIUP, and 8 VIUP. One ectopic pregnancy sample did not consistently amplify and was excluded from the technical validation. For the validation cohort, 98 women (81.7 %) were selected where 32 were EP, 16 NVIUP and 50 VIUP. One miscarriage sample subsequently did not consistently amplify and was excluded. Fig. 1 Flowchart of patient recruitment and breakdown of data available for discovery, technical and biological validation studies Abbreviations: FPUL: Failed PUL; LTFU: Lost to follow up; PPUL: Persistent PUL; IPUV: Intrauterine pregnancy of unknown viability; PUL: Pregnancy of unknown location; EP: Ectopic pregnancy; NVIUP: Non-viable intrauterine pregnancy; VIUP: Viable intrauterine pregnancy. Fig. 1 Flowchart of patient recruitment and breakdown of data available for discovery, technical and biological validation studies Abbreviations: FPUL: Failed PUL; LTFU: Lost to follow up; PPUL: Persistent PUL; IPUV: Intrauterine pregnancy of unknown viability; PUL: Pregnancy of unknown location; EP: Ectopic pregnancy; NVIUP: Non-viable intrauterine pregnancy; VIUP: Viable intrauterine pregnancy. The demographics for all three outcome groups for the separate discovery and biological validation cohorts were similar, except for the participants age in the biological validation cohort when comparing NVIUP with VIUP ( p  = 0.02, Kruskal-Wallis analysis), and when comparing ‘Black’ and ‘Other’ ethnicities within viable intrauterine and ectopic pregnancy cohorts ( p  = 0.04, Fisher's exact test) ( Table 1 , Table 2 ). The presence of ectopic pregnancy risk factors within the outcome groups of both the discovery and biological validation cohorts are described in Supplementary Tables S2 and S3 , with the demographics and characteristics of each outcome group further divided by the ethnicity groups of Table 1 , Table 2 in Supplementary Tables S4 and S5 . Table 1 Demographics of discovery and technical validation cohort (N = 22) for microRNA analysis, divided by outcome. Table 1 Demographics VIUP (N = 8) EP (N = 7) NVIUP (N = 7) p -value Mean age - years (SD) 32 (5.9) 35.3 (3.9) 34.4 (6.4) 0.60; 0.77 Individual p -values 0.60, 0.65 0.60, 0.77 0.65, 0.77 Ethnicity N (%) Caucasian 5 (62.5) 4 (57.1) 3 (42.9) 0.44; 0.99 Asian 3 (37.5) 2 (28.6) 3 (42.9) Black 0 (0.0) 1 (14.3) 0 (0.0) Other 0 (0.0) 0 (0.0) 1 (14.3) Individual p -value ranges 0.44–0.99 0.99 0.44–0.99 Mean gestational age - days (SD) 32.3 (4.8) 37.3 (12.9) 44.6 (11.8) 0.092; 0.36 Individual p -values 0.092, 0.36 0.36 0.092, 0.36 Median BMI - kg/m 2 (IQR) 24.8 (21.5–31.3) 24.4 (21.9–31.2) 25.8 (22.2–32.0) a 0.99 Individual p -values 0.99 0.99 0.99 p-value corresponds to one-way ANOVA (parametric, continuous), Kruskal-Wallis (non-parametric, continuous), or Fisher's (categorical) for the difference in study participants' demographics between outcome groups. For Fisher's, each outcome and ethnicity group were individually tested against one another. Individual p -values are specific to each outcome group under comparison, with overall p -value ranges listed in the end column. Abbreviations: SD = Standard deviation; IQR = Interquartile range; VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy; NS = Not significant; BMI = Body mass index. a BMI not available for 1 NVIUP. Table 2 Demographics of biological validation cohort (N = 97) for microRNA analysis, divided by outcome. Table 2 Demographics VIUP (N = 50) EP (N = 32) NVIUP (N = 15) p -value Median age - years (IQR) 31 (25.8–35) 32 (27–37.5) 34 (31–42) a 0.018 a ; 0.68 Individual p -values 0.018 a , 0.68 0.26, 0.68 0.018 a , 0.26 Ethnicity N (%) Caucasian 28 (56.0) 16 (50.0) 7 (46.7) 0.041 b ; 0.99 Asian 7 (14.0) 6 (18.8) 3 (20.0) Black 8 (16.0) b 1 (3.1) b 3 (20.0) Other 7 (14.0) b 9 (28.1) b 2 (13.3) Individual p -value ranges 0.041 b –0.99 0.041 b –0.99 0.077–0.99 Median gestational age - days (IQR) 33 (31.0–38.5) c 42.5 (27.5–48.8) d 40 (34.0–44.0) 0.18; 0.99 Individual p -values 0.18, 0.47 0.18, 0.99 0.47, 0.99 Median BMI - kg/m 2 (IQR) 26.4 (22.8–28.8) e 23.7 (20.6–26.7) f 24 (20.3–28.3) 0.46; 0.99 Individual p -values 0.46, 0.60 0.46, 0.99 0.60, 0.99 p-value corresponds to one-way ANOVA (parametric, continuous), Kruskal-Wallis (non-parametric, continuous), or Fisher's (categorical) for the difference in study participants' demographics between outcome groups. For Fisher's, each outcome and ethnicity group were individually tested against one another. Individual p -values are specific to each outcome group under comparison, with overall p -value ranges listed in the end column. Abbreviations: IQR = Interquartile range; VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy; NS = Not significant; BMI = Body mass index. a Significant difference comparing NVIUP with VIUP. b Significant difference comparing ‘Black’ and ‘Other’ ethnicities within VIUP and EP cohorts. c Gestational age not available for 4 VIUP. d Gestational age not available for 4 EP. e BMI not available for 1 VIUP. f BMI not available for 1 EP. Demographics of discovery and technical validation cohort (N = 22) for microRNA analysis, divided by outcome. p-value corresponds to one-way ANOVA (parametric, continuous), Kruskal-Wallis (non-parametric, continuous), or Fisher's (categorical) for the difference in study participants' demographics between outcome groups. For Fisher's, each outcome and ethnicity group were individually tested against one another. Individual p -values are specific to each outcome group under comparison, with overall p -value ranges listed in the end column. Abbreviations: SD = Standard deviation; IQR = Interquartile range; VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy; NS = Not significant; BMI = Body mass index. BMI not available for 1 NVIUP. Demographics of biological validation cohort (N = 97) for microRNA analysis, divided by outcome. p-value corresponds to one-way ANOVA (parametric, continuous), Kruskal-Wallis (non-parametric, continuous), or Fisher's (categorical) for the difference in study participants' demographics between outcome groups. For Fisher's, each outcome and ethnicity group were individually tested against one another. Individual p -values are specific to each outcome group under comparison, with overall p -value ranges listed in the end column. Abbreviations: IQR = Interquartile range; VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy; NS = Not significant; BMI = Body mass index. Significant difference comparing NVIUP with VIUP. Significant difference comparing ‘Black’ and ‘Other’ ethnicities within VIUP and EP cohorts. Gestational age not available for 4 VIUP. Gestational age not available for 4 EP. BMI not available for 1 VIUP. BMI not available for 1 EP. A total of 62 miRNAs were found to be expressed above the background in more than 50 % of samples in any one of the outcome groups ( Table S6 ). The expression of these miRNAs were compared between outcome groups using nSolver (Nanostring) with 19 miRNAs identified to be significantly different between control (VIUP) and ectopic pregnancy ± miscarriage (NVIUP) outcome groups ( Table 3 , Table 4 , Table 5 ). There were eight miRNAs expressed significantly higher only in EP compared to VIUP ( Table 3 , Fig. 2 ), seven only in NVIUP when compared to VIUP ( Table 4 , Fig. 3 ), and four miRNAs which were increased in both EP and NVIUP when compared to VIUP ( Table 5 , Fig. 4 ). Table 3 Discovery microRNA markers meeting significance when comparing viable intrauterine pregnancy with ectopic pregnancy (N = 15) using nCounter. Table 3 MicroRNA VIUP (N = 8) vs EP (N = 7) Ratio (95 % CI) Relative change p -value hsa-miR-222-3p 4.5∗ (1.7–12.1) EP > VIUP 0.0060 hsa-miR-20a-5p + hsa-miR-20b-5p a 6.0∗ (1.4–25.0) EP > VIUP 0.019 hsa-miR-21-5p 5.4∗ (1.3–23.0) EP > VIUP 0.027 hsa-miR-185-5p 4.8∗ (1.2–18.8) EP > VIUP 0.028 hsa-miR-148a-3p 5.1∗ (1.1–23.2) EP > VIUP 0.036 hsa-miR-19b-3p 4.5∗ (1.1–18.8) EP > VIUP 0.038 hsa-miR-1285-5p 4.4∗ (1.1–18.1) EP > VIUP 0.038 hsa-miR-22-3p 4.6∗ (1.0–20.5) EP > VIUP 0.044 p-value corresponds to unpaired two-tailed t -test, for the difference between outcome groups. Ordered by p -value; ∗Designates ratio significance. Abbreviations: 95 % CI = 95 % confidence interval; VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy. a Nanostring assay ran both these targets together. Table 4 Discovery microRNA markers meeting significance when comparing viable intrauterine pregnancy with non-viable intrauterine pregnancy (N = 15) using nCounter. Table 4 MicroRNA VIUP (N = 8) vs NVIUP (N = 7) Ratio (95 % CI) Relative change p -value has-miR-107 5.3∗ (1.8–15.2) NVIUP > VIUP 0.0047 hsa-miR-223-3p 2.0∗ (1.1–3.7) NVIUP > VIUP 0.028 hsa-let-7d-5p 3.9∗ (1.1–13.9) NVIUP > VIUP 0.035 hsa-miR-191-5p 2.5∗ (1.1–5.8) NVIUP > VIUP 0.037 hsa-miR-29b-3p 4.5∗ (1.0–19.9) NVIUP > VIUP 0.045 hsa-miR-374a-5p 1.6∗ (1.0–2.5) NVIUP > VIUP 0.046 hsa-miR-199a-3p + hsa-miR-199b-3p a 3.7∗ (1.0–13.2) NVIUP > VIUP 0.047 p-value corresponds to unpaired two-tailed t -test, for the difference between outcome groups. Ordered by p -value; ∗Designates ratio significance. Abbreviations: 95 % CI = 95 % confidence interval; VIUP = Viable intrauterine pregnancy; NVIUP = Non-viable intrauterine pregnancy. a Nanostring assay ran both these targets together. Table 5 Discovery microRNA markers meeting significance when comparing viable intrauterine pregnancy with both ectopic pregnancy and non-viable intrauterine pregnancy (N = 22) using nCounter. Table 5 MicroRNA VIUP (N = 8) vs EP (N = 7) Ratio (95 % CI) VIUP (N = 8) vs NVIUP (N = 7) Ratio (95 % CI) Relative change p -value hsa-miR-1246 13.9∗ (2.8–70.3) 6.0∗ (1.2–29.0) EP > VIUP 0.0062 NVIUP > VIUP 0.030 hsa-miR-130a-3p 5.7∗ (1.3–26.0) 8.2∗ (1.9–35.2) EP > VIUP 0.026 NVIUP > VIUP 0.0090 hsa-miR-320e 5.3∗ (1.4–20.6) 3.9∗ (1.3–11.2) EP > VIUP 0.020 NVIUP > VIUP 0.017 hsa-miR-30d-5p 3.0∗ (1.2–7.3) 3.3∗ (1.2–9.1) EP > VIUP 0.022 NVIUP > VIUP 0.023 p -value corresponds to unpaired two-tailed t -test, for the difference between outcome groups. Ordered by p -value of EP/NVIUP vs VIUP for each microRNA; ∗Designates ratio significance. Abbreviations: 95 % CI = 95 % confidence interval; VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy. Fig. 2 Eight discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP (N = 15) in discovery study. nCounter probe counts for A: hsa-miR-222-3p, B: hsa-miR-20a-5p + hsa-miR-20b-5p, C: hsa-miR-21-5p, D: hsa-miR-185-5p, E: hsa-miR-148a-3p, F: hsa-miR-19b-3p, G: hsa-miR-1285-5p, and H: hsa-miR-22-3p detected in human plasma in women in early pregnancy who subsequently had a confirmed VIUP (N = 8, green), EP (N = 7, red), or NVIUP (N = 7, purple) ∗Statistical significance of differences between groups was tested using Nanostring nSolver platform. Graphs show geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) counts on y axis. Only statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy. Fig. 2 Fig. 3 Seven discriminatory miRNAs in early pregnancy plasma when comparing VIUP with NVIUP (N = 15) in discovery study. nCounter probe counts for A: hsa-miR-107, B: hsa-miR-223-3p, C: hsa-let-7d-5p, D: hsa-miR-191-5p, E: hsa-miR-29b-3p, F: hsa-miR-374a-5p, and G: hsa-miR-199a-3p + hsa-miR-199b-3p detected in human plasma in women in early pregnancy who subsequently had a confirmed VIUP (N = 8, green), EP (N = 7, red), or NVIUP (N = 7, purple) ∗Statistical significance of differences between groups was tested using Nanostring nSolver platform. Graphs show geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) counts on y axis. Only statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy. Fig. 3 Fig. 4 Four discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP and NVIUP (N = 22) in discovery study. nCounter probe counts for A: hsa-miR-1246, B: hsa-miR-130a-3p, C: hsa-miR-320e, and D: hsa-miR-30d-5p detected in human plasma in women in early pregnancy who subsequently had a confirmed VIUP (N = 8, green), EP (N = 7, red), or NVIUP (N = 7, purple) ∗Statistical significance of differences between groups was tested using Nanostring nSolver platform. Graphs show geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) counts on y axis. Statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy. Fig. 4 Discovery microRNA markers meeting significance when comparing viable intrauterine pregnancy with ectopic pregnancy (N = 15) using nCounter. p-value corresponds to unpaired two-tailed t -test, for the difference between outcome groups. Ordered by p -value; ∗Designates ratio significance. Abbreviations: 95 % CI = 95 % confidence interval; VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy. Nanostring assay ran both these targets together. Discovery microRNA markers meeting significance when comparing viable intrauterine pregnancy with non-viable intrauterine pregnancy (N = 15) using nCounter. p-value corresponds to unpaired two-tailed t -test, for the difference between outcome groups. Ordered by p -value; ∗Designates ratio significance. Abbreviations: 95 % CI = 95 % confidence interval; VIUP = Viable intrauterine pregnancy; NVIUP = Non-viable intrauterine pregnancy. Nanostring assay ran both these targets together. Discovery microRNA markers meeting significance when comparing viable intrauterine pregnancy with both ectopic pregnancy and non-viable intrauterine pregnancy (N = 22) using nCounter. p -value corresponds to unpaired two-tailed t -test, for the difference between outcome groups. Ordered by p -value of EP/NVIUP vs VIUP for each microRNA; ∗Designates ratio significance. Abbreviations: 95 % CI = 95 % confidence interval; VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy. Eight discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP (N = 15) in discovery study. nCounter probe counts for A: hsa-miR-222-3p, B: hsa-miR-20a-5p + hsa-miR-20b-5p, C: hsa-miR-21-5p, D: hsa-miR-185-5p, E: hsa-miR-148a-3p, F: hsa-miR-19b-3p, G: hsa-miR-1285-5p, and H: hsa-miR-22-3p detected in human plasma in women in early pregnancy who subsequently had a confirmed VIUP (N = 8, green), EP (N = 7, red), or NVIUP (N = 7, purple) ∗Statistical significance of differences between groups was tested using Nanostring nSolver platform. Graphs show geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) counts on y axis. Only statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy. Seven discriminatory miRNAs in early pregnancy plasma when comparing VIUP with NVIUP (N = 15) in discovery study. nCounter probe counts for A: hsa-miR-107, B: hsa-miR-223-3p, C: hsa-let-7d-5p, D: hsa-miR-191-5p, E: hsa-miR-29b-3p, F: hsa-miR-374a-5p, and G: hsa-miR-199a-3p + hsa-miR-199b-3p detected in human plasma in women in early pregnancy who subsequently had a confirmed VIUP (N = 8, green), EP (N = 7, red), or NVIUP (N = 7, purple) ∗Statistical significance of differences between groups was tested using Nanostring nSolver platform. Graphs show geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) counts on y axis. Only statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy. Four discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP and NVIUP (N = 22) in discovery study. nCounter probe counts for A: hsa-miR-1246, B: hsa-miR-130a-3p, C: hsa-miR-320e, and D: hsa-miR-30d-5p detected in human plasma in women in early pregnancy who subsequently had a confirmed VIUP (N = 8, green), EP (N = 7, red), or NVIUP (N = 7, purple) ∗Statistical significance of differences between groups was tested using Nanostring nSolver platform. Graphs show geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) counts on y axis. Statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy. We performed technical validation of the 19 miRNAs identified using nCounter (Nanostring). Two miRNAs, hsa-miR-126-3p and hsa-let-7g-5p were identified to have the highest inter- and intra-group stability using NormFinder and were used as endogenous controls [ 63 ]. Univariate analysis of the DeltaCq values for each miRNA candidate are shown in Table 3 . From the 19 miRNAs examined, one miRNA (hsa-miR-21-5p) showed significant increase in EP compared to VIUP ( Table 6 ), and two miRNAs (hsa-miR-130a-3p and hsa-miR-374a-5p) were found to be significantly increased in NVIUP compared to VIUP ( Table 7 , Table 8 ). Table 6 Technical validation of microRNA markers comparing viable intrauterine pregnancy with ectopic pregnancy (N = 14) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Table 6 MicroRNA VIUP (N = 8) EP (N = 6) VIUP vs EP Fold change Relative change p -value has-miR-21-5p 1.0 (0.28) 3.2 (2.75) 3.2∗ EP > VIUP 0.0013 hsa-miR-222-3p 1.0 (0.79) 1.8 (1.01) 1.8 EP > VIUP 0.12 hsa-miR-148a-3p 1.0 (0.47) 2.2 (2.05) 2.2 EP > VIUP 0.13 hsa-miR-185-5p 1.0 (0.28) 1.2 (0.57) 1.2 EP > VIUP 0.45 hsa-miR-20a-5p 1.0 (0.43) 1.2 (0.52) 1.2 EP > VIUP 0.52 hsa-miR-19b-3p 1.0 (0.48) 1.2 (0.81) 1.2 EP > VIUP 0.54 hsa-miR-1285-5p 1.0 (0.37) 0.9 (0.45) 0.9 EP  VIUP 0.79 hsa-miR-22-3p 1.0 (0.45) 1.2 (0.89) 1.2 EP > VIUP 0.85 Mean DeltaCq (SD) presented in outcome columns. p -value corresponds to unpaired two-tailed t -test (parametric, continuous), or Mann-Whitney U (non-parametric, continuous), for the difference in fold change between outcome groups. Ordered by p -value; ∗Designates relative expression fold change significance. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; SD = Standard deviation. Table 7 Technical validation of microRNA markers comparing viable intrauterine pregnancy with non-viable intrauterine pregnancy (N = 15) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Table 7 MicroRNA VIUP (N = 8) NVIUP (N = 7) VIUP vs NVIUP Fold change Relative change p -value hsa-miR-374a-5p 1.0 (0.14) 2.0 (1.20) 2.0∗ NVIUP > VIUP 0.0022 hsa-miR-199a-3p 1.0 (0.87) 2.0 (1.34) 2.0 NVIUP > VIUP 0.054 hsa-miR-107 1.0 (0.32) 1.5 (0.52) 1.5 NVIUP > VIUP 0.058 hsa-miR-223-3p 1.0 (0.77) 1.6 (1.09) 1.6 NVIUP > VIUP 0.072 hsa-miR-29b-3p 1.0 (0.35) 1.2 (0.22) 1.2 NVIUP > VIUP 0.19 hsa-miR-191-5p 1.0 (0.25) 1.3 (0.62) 1.3 NVIUP > VIUP 0.28 hsa-let-7d-5p 1.0 (0.60) 0.9 (0.36) 0.9 NVIUP < VIUP 0.70 Mean DeltaCq (SD) presented in outcome columns. p -value corresponds to unpaired two-tailed t -test (parametric, continuous), or Mann-Whitney U (non-parametric, continuous), for the difference in fold change between outcome groups. Ordered by p -value; ∗Designates relative expression fold change significance. Abbreviations: VIUP = Viable intrauterine pregnancy; NVIUP = Non-viable intrauterine pregnancy; SD = Standard deviation. Table 8 Technical validation of microRNA markers comparing viable intrauterine pregnancy with both ectopic pregnancy and non-viable intrauterine pregnancy (N = 21) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Table 8 MicroRNA VIUP (N = 8) EP (N = 6) NVIUP (N = 7) VIUP vs EP Fold change VIUP vs NVIUP Fold change Relative change p -value hsa-miR-130a-3p 1.0 (0.61) 1.3 (0.57) 2.2 (1.61) 1.3 2.2∗ EP > VIUP 0.14 NVIUP > VIUP 0.014 hsa-miR-30d-5p 1.0 (0.38) 1.7 (1.01) 1.5 (0.68) 1.7 1.5 EP > VIUP 0.11 NVIUP > VIUP 0.094 hsa-miR-1246 1.0 (0.63) 1.3 (1.36) 0.5 (0.29) 1.3 0.5 EP > VIUP 0.95 NVIUP  VIUP 0.41 NVIUP > VIUP 0.96 Mean DeltaCq (SD) presented in outcome columns. p -value corresponds to unpaired two-tailed t -test (parametric, continuous), or Mann-Whitney U (non-parametric, continuous), for the difference in fold change between outcome groups. Ordered by p -value of EP/NVIUP vs VIUP for each microRNA; ∗Designates relative expression fold change significance. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy; SD = Standard deviation. Technical validation of microRNA markers comparing viable intrauterine pregnancy with ectopic pregnancy (N = 14) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Mean DeltaCq (SD) presented in outcome columns. p -value corresponds to unpaired two-tailed t -test (parametric, continuous), or Mann-Whitney U (non-parametric, continuous), for the difference in fold change between outcome groups. Ordered by p -value; ∗Designates relative expression fold change significance. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; SD = Standard deviation. Technical validation of microRNA markers comparing viable intrauterine pregnancy with non-viable intrauterine pregnancy (N = 15) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Mean DeltaCq (SD) presented in outcome columns. p -value corresponds to unpaired two-tailed t -test (parametric, continuous), or Mann-Whitney U (non-parametric, continuous), for the difference in fold change between outcome groups. Ordered by p -value; ∗Designates relative expression fold change significance. Abbreviations: VIUP = Viable intrauterine pregnancy; NVIUP = Non-viable intrauterine pregnancy; SD = Standard deviation. Technical validation of microRNA markers comparing viable intrauterine pregnancy with both ectopic pregnancy and non-viable intrauterine pregnancy (N = 21) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Mean DeltaCq (SD) presented in outcome columns. p -value corresponds to unpaired two-tailed t -test (parametric, continuous), or Mann-Whitney U (non-parametric, continuous), for the difference in fold change between outcome groups. Ordered by p -value of EP/NVIUP vs VIUP for each microRNA; ∗Designates relative expression fold change significance. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; NVIUP = Non-viable intrauterine pregnancy; SD = Standard deviation. Biological validation was performed for the three technically validated miRNAs (hsa-miR-21-5p, hsa-miR-130a-3p and hsa-miR-374a-5p) in a separate patient population utilised as a second validation cohort ( Fig. 1 , n = 98). Three additional miRNA markers were also added following literature review: hsa-miR-411-5p, hsa-miR-324-3p and hsa-miR-873-3p [ 52 , 54 , 57 , 59 ]. These three additional miRNAs were reported to be expressed at lower levels in ectopic pregnancy when compared to viable intrauterine pregnancy (or termination of pregnancy patients). Whilst this trend was confirmed by the findings of the nCounter assay ( Table S1 ), where the differences in Nanostring counts between EP and VIUP were either significant (hsa-miR-411-5p, p  = 0.040) or borderline (hsa-miR-324-3p, p  = 0.090; hsa-miR-873-3, p  = 0.069), they were excluded from further discovery cohort analysis due to their low levels of expression (below background level in more than 50 % of at least one outcome group). However, in this separate biological validation, we aimed to exploit the inverse expression relationship between these miRNAs and the increasing miRNAs identified in the discovery cohort, to improve the discrimination of ectopic pregnancy. Two of the six miRNAs examined (hsa-miR-21-5p and hsa-miR-411-5p) were biologically validated ( Table 9 ). Hsa-miR-21-5p was significantly increased in EP compared to VIUP ( p  = 0.035) with a mean fold change of 2.8 (SD 5.46), and hsa-miR-411-5p was decreased in EP ( p  = 0.016) with a mean fold change of 0.2 (SD 0.20) ( Table 9 , Fig. 5 ). Hsa-miR-130a-3p, hsa-miR-374a-5p and hsa-miR-324-3p did not show any significant differences between outcome groups ( Table 9 , Table 10 , Fig. S1 ), and hsa-miR-873-3p failed to amplify during RT-qPCR. Six of the 485 normalised Cq values obtained from the five miRNAs that did amplify were given a value of 45 during the analysis (1.2 %). Table 9 Biological validation of microRNA markers comparing viable intrauterine pregnancy with ectopic pregnancy (N = 82) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Table 9 MicroRNA VIUP (N = 50) EP (N = 32) VIUP vs EP Fold change Relative change p -value hsa-miR-411-5p 1.0 (3.65) 0.2 (0.20) 0.2∗ EP  VIUP 0.035 hsa-miR-324-3p 1.0 (0.83) 1.0 (0.74) 1.0 EP = VIUP 0.62 Mean DeltaCq (SD) presented in outcome columns. p -value corresponds to Mann-Whitney U (non-parametric, continuous). For the difference in fold change between outcome groups. Ordered by p -value; ∗Designates relative expression fold change significance. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; SD = Standard deviation. Fig. 5 Two discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP (N = 82) in biological validation study. Fold changes for A: hsa-miR-21-5p, and B: hsa-miR-411-5p in women in early pregnancy who subsequently had a confirmed VIUP (N = 50, green), or EP (N = 32, red) ∗Statistical significance of differences between groups was tested using Mann-Whitney (non-parametric, continuous). Graphs show geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) fold change on y axis. Statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy. Fig. 5 Table 10 Biological validation of microRNA markers comparing viable intrauterine pregnancy with non-viable intrauterine pregnancy (N = 65) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Table 10 MicroRNA VIUP (N = 50) NVIUP (N = 15) VIUP vs NVIUP Fold change Relative change p -value hsa-miR-374a-5p 1.0 (0.42) 0.9 (0.51) 0.9 NVIUP < VIUP 0.26 hsa-miR-130a-3p 1.0 (0.60) 0.9 (0.39) 0.9 NVIUP < VIUP 0.68 Mean DeltaCq (SD) presented in outcome columns. P -value corresponds to unpaired two-tailed t -test (parametric, continuous), or Mann-Whitney U (non-parametric, continuous), for the difference in fold change between outcome groups. Ordered by p -value; Relative expression fold change presented. Abbreviations: VIUP = Viable intrauterine pregnancy; NVIUP = Non-viable intrauterine pregnancy; SD = Standard deviation. Biological validation of microRNA markers comparing viable intrauterine pregnancy with ectopic pregnancy (N = 82) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Mean DeltaCq (SD) presented in outcome columns. p -value corresponds to Mann-Whitney U (non-parametric, continuous). For the difference in fold change between outcome groups. Ordered by p -value; ∗Designates relative expression fold change significance. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; SD = Standard deviation. Two discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP (N = 82) in biological validation study. Fold changes for A: hsa-miR-21-5p, and B: hsa-miR-411-5p in women in early pregnancy who subsequently had a confirmed VIUP (N = 50, green), or EP (N = 32, red) ∗Statistical significance of differences between groups was tested using Mann-Whitney (non-parametric, continuous). Graphs show geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) fold change on y axis. Statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy. Biological validation of microRNA markers comparing viable intrauterine pregnancy with non-viable intrauterine pregnancy (N = 65) using reverse transcription quantitative polymerase chain reaction (RT-qPCR). Mean DeltaCq (SD) presented in outcome columns. P -value corresponds to unpaired two-tailed t -test (parametric, continuous), or Mann-Whitney U (non-parametric, continuous), for the difference in fold change between outcome groups. Ordered by p -value; Relative expression fold change presented. Abbreviations: VIUP = Viable intrauterine pregnancy; NVIUP = Non-viable intrauterine pregnancy; SD = Standard deviation. The predictive ability of the two biologically validated miRNA markers for ectopic pregnancy were assessed using ROC curve analyses ( Fig. 6 ). As individual markers, hsa-miR-21-5p and hsa-miR-411-5p showed ROC AUC of 0.64 ( p  = 0.035) and 0.66 ( p  = 0.017), respectively. To explore the two markers in combination for the prediction of EP, the inverse correlation of hsa-miR-21-5p and hsa-miR-411-5p expression was used to create a ratio ( Fig. 7 ). The ratio of hsa-miR-21-5p/hsa-miR-411-5p showed significant difference between EP and viable intrauterine pregnancies ( p  = 0.0003), and the ratio revealed improved predictive ability with ROC AUC of 0.74 ( p  = 0.0004). Fig. 6 Two discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP (N = 82) in biological validation study. Area under the Receiver operating characteristic (ROC) curve (AUC) for A: hsa-miR-21-5p, and B: hsa-miR-411-5p, in women in early pregnancy who subsequently had a confirmed VIUP (N = 50), or EP (N = 32) ROC curve with area, 95 % confidence interval and p -value presented. Statistically significant differences are presented. Abbreviations: ROC = Receiver operating characteristic. Fig. 6 Fig. 7 Two discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP (N = 82) in biological validation study. A: Fold changes, and B: area under the Receiver operating characteristic (ROC) curve (AUC) for hsa-miR-21-5p/hsa-miR-411-5p ratio, in women in early pregnancy who subsequently had a confirmed VIUP (N = 50, green), or EP (N = 32, red) ∗Statistical significance of differences between groups was tested using Mann-Whitney (non-parametric, continuous). Graph shows geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) fold change on y axis. ROC curve with area, 95 % confidence interval and p -value presented. Statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; ROC = Receiver operating characteristic. Fig. 7 Two discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP (N = 82) in biological validation study. Area under the Receiver operating characteristic (ROC) curve (AUC) for A: hsa-miR-21-5p, and B: hsa-miR-411-5p, in women in early pregnancy who subsequently had a confirmed VIUP (N = 50), or EP (N = 32) ROC curve with area, 95 % confidence interval and p -value presented. Statistically significant differences are presented. Abbreviations: ROC = Receiver operating characteristic. Two discriminatory miRNAs in early pregnancy plasma when comparing VIUP with EP (N = 82) in biological validation study. A: Fold changes, and B: area under the Receiver operating characteristic (ROC) curve (AUC) for hsa-miR-21-5p/hsa-miR-411-5p ratio, in women in early pregnancy who subsequently had a confirmed VIUP (N = 50, green), or EP (N = 32, red) ∗Statistical significance of differences between groups was tested using Mann-Whitney (non-parametric, continuous). Graph shows geometric mean with geometric standard deviation and p -values, with outcome on x axis, and log(10) fold change on y axis. ROC curve with area, 95 % confidence interval and p -value presented. Statistically significant differences are presented. Abbreviations: VIUP = Viable intrauterine pregnancy; EP = Ectopic pregnancy; ROC = Receiver operating characteristic. An in-silico analysis of the genes and pathways regulated by hsa-miR-21-5p and hsa-miR-411-5p was undertaken to further understand their potential role in ectopic pregnancy pathophysiology. The miRDIP miRNA data integration portal (University Health Network, Toronto, Canada – RRID: SCR_016770 ) identified 413 genes with binding sites for these miRNA markers [ 20 , [65] , [66] , [67] ]. Pathway analysis was then performed using the Reactome (Reactome project, Ontario/New York/Hinxton/Oregon, Canada/USA/UK – RRID: SCR_003485 ) platform [ [68] , [69] , [70] , [71] , [72] , [73] ]. Of the 1129 pathways identified using the Reactome platform, 162 met significance (defined as p  ≤ 0.05) and are presented in Table S7 . These include signal transduction pathways via growth factor receptors, transcription regulation pathways, oestrogen dependent gene expression, and cytokine signalling, particularly interleukins (IL-4, IL-6, IL-12, and IL-13). IL-6, a proinflammatory interleukin, has been associated with ovarian hormonal stimulation, follicle genesis, and implantation, with higher levels identified in the fallopian tubes of women with EP [ 53 ]. Pilot work previously presented suggested a difference in both IL-6 and oestradiol expression in EP [ 74 , 75 ].

There is an ongoing clinical need to improve pregnancy of unknown location triage to reduce potential life-threatening consequences of missed or delayed ectopic pregnancy diagnoses. In this study, we have explored miRNAs as potential novel markers that may play a role in improving PUL triage pending diagnosis using transvaginal ultrasonography. We report the identification of 19 miRNAs differentially expressed in the three outcome groups (VIUP, NVIUP and EP) through maternal plasma miRNA profiling in our discovery work. Three of these were technically validated using RT-qPCR, and two miRNA markers, hsa-miR-21-5p and hsa-miR-411-5p, were further validated in a second, separate validation cohort. We found both validated miRNAs were able to differentiate viable intrauterine pregnancies from ectopic pregnancies, with hsa-miR-21-5p increasing and the hsa-miR-411-5p decreasing significantly in EP. The combination of their relative expression as a ratio demonstrated good predictive ability for EP with ROC AUC of 0.74, indicating their potential use in EP prediction as early as four weeks’ gestation. Several putative miRNAs biomarkers of ectopic pregnancy have been previously reported ( Table S8 ) [ 52 , 54 , [56] , [57] , [58] , [59] , 61 , 76 , 77 ]. Dysregulation of these miRNAs have been hypothesized to affect endometrial receptivity, production of reproductive regulators such as kisspeptin, influence implantation site (with associated tissue trauma and inflammation), vascularity and trophoblast formation [ 53 , 54 , 59 , 78 , 79 ]. Whilst there were no peer-reviewed articles that linked hsa-miR-21-5p to EP, there was one paper that discussed hsa-miR-411-5p in this manner [ 54 ]. There are key differences between the literature available and our work: a) none of the studies had focussed on a PUL population, b) many used tissue, or serum samples, c) most had low study numbers, and d) the majority of EP included had later gestational ages. We have examined a unique population in this context with early miRNA detection suggesting the ability of prompt EP prediction and detection (with consequent early intervention), which could reduce morbidity. Plasma collection from women with PUL is a non-invasive (unlike tissue), easily tolerated and well accepted blood test, which unlike serum is not affected by white blood cells or platelets where miRNAs are released during clotting. Yi Feng et al. has previously examined tissue samples from women undergoing termination of pregnancy and compared with fallopian tube samples (invasive) from women undergoing salpingectomy for tubal EP (not PUL) [ 54 ]. Amongst the miRNAs identified, lower expression of hsa-miR-411-5p was noted in 20 EP with an average gestational age of nine weeks. Following our in-silico analyses, a separate literature review of hsa-miR-21-5p confirmed its involvement in the pathophysiology of multiple pregnancy complications. These include dysregulation of endometrial receptivity, proliferation, apoptosis and migration via the PDCD4/AKT, PI3K/AKY (FOXO3) and JAK/STAT3 pathways, hypoxia-related changes, increased postpartum cardiovascular risk in women previously affected by gestational diabetes, success of in-vitro fertilisation and ovarian reserve, placentation, gestational hypertension, pre-eclampsia and intrahepatic cholestasis of pregnancy, preterm premature rupture of membranes and pre term birth, placental and fetal growth [ [80] , [81] , [82] , [83] , [84] , [85] , [86] , [87] , [88] , [89] , [90] , [91] , [92] , [93] , [94] ]. PTEN (phosphates and tensin homologue deleted on chromosome ten) is quoted as one of the most significant pathways within the in-silico analysis. However, PTEN is also linked to hsa-miR-411-5p expression [ 95 ]. Whilst PTEN overexpression negatively affects cell survival, hsa-miR-411-5p causes downregulation, allowing trophoblast proliferation, migration, and invasion [ 95 ]. Nuclear paraspeckle assembly transcript 1 (NEAT1) inhibits hsa-miR-411-5p expression, upregulating PTEN which in turn promotes trophoblast apoptosis and pre-eclampsia [ 95 ]. Whilst these findings support both hsa-miR-21-5p and hsa-miR-411-5p potentially contributing to ectopic pregnancy aetiology in a biologically plausible manner, gaining a functional understanding may prove beneficial when considering their use in combination with other epigenetic markers, proteins or hormones not yet identified to further improve 10.13039/501100000784 EP prediction, diagnosis, and subsequent management. Given the complexity of early pregnancy processes and pathophysiology, opportunities to use various statistical or algorithmic approaches beyond the scope of this paper may also subsequently arise. Limitations of the study pertain to the overall numbers, the specific demographics, and the subjective assessment of ultrasound, which all could potentially confound results as well as limit findings. However, whilst the proportions of each outcome group were not reflective of population prevalence, and patients were recruited following either PUL classification or EP diagnosis, the enriched EP population, including ectopic pregnancies at diagnosis, allowed this relatively uncommon early pregnancy outcome to be evaluated with greater statistical power, with total population numbers comparable to similar published work, and matched demographically as closely as the population allowed [ 19 , 21 ]. The results, from a Health Policy perspective, lay important groundwork for ongoing research into how miRNA can contribute to improving day to day clinical workflow and patient safety in early pregnancy, both of which would likely also carry economic benefits. Unfortunately, from a total of 290 women, 170 were excluded due to their outcome diagnosis of failed or persistent PUL, both heterogenic groups where pregnancy location could not be confidently differentiated, as well as those lost to follow up or with an intrauterine pregnancy of uncertain viability. It is also likely that miRNAs differentiating VIUP with miscarriages (NVIUP) did not validate due to the heterogeneity of the NVIUP population following a PUL classification. For example, intrauterine pregnancies of unknown viability (IPUV) may have miscarried in the first trimester with or without the development of fetal cardiac activity. Validation in a larger population, the inclusion of all pregnancy of unknown location outcomes, and a sensitivity analysis are the next steps in rigorously evaluating the potential clinical usefulness of hsa-miR-21-5p and hsa-miR-411-5p, both alone and as part of ratio. To facilitate such a large deployment of these miRNAs will require the digital signal of these candidate biomarkers to be merged with the digital signals of two or more independent modalities. This fusion of technology would increase generalisability and access to the markers without compromising workflow efficiency and precision, allowing outpatient early pregnancy use in a manner that is rapid but accurate. We have demonstrated the potential of circulating miRNA markers for PUL triage by predicting EP in early pregnancy, which can be measured via the gold-standard RT-qPCR method that is an efficient, cost-effective, and readily available technology for translational use in clinical practice. With better diagnostics, the potential for further advancements in ectopic pregnancy therapeutics is also high, using proposed vehicles such as nanoparticles for targeted interventions in the management of pregnancy complications [ 96 ].

Pregnancy of unknown location (PUL) is classified in early pregnancy when a pregnancy is not visualised on transvaginal ultrasonography (TVUS) following a positive urine pregnancy test (UPT) [ 1 ]. PUL rates have been reported to vary between five to 42 % of all pregnancies [ [2] , [3] , [4] , [5] ]. Management of PUL is based on the triage of these pregnancies to high versus low risk of clinical complications to decide community-based follow-up arrangements pending final diagnosis [ 1 , 6 ]. High-risk outcomes include ectopic pregnancies (EP – pregnancies partially or completely external to the endometrial cavity) or persistent PUL (PPUL – PUL with three or more serial beta-human chorionic gonadotrophin (hCG) values that vary by <15 %), and low-risk outcomes comprise intrauterine pregnancies (viable – VIUP, or non-viable – NVIUP/miscarriage) or failed PUL (FPUL – PUL with a negative UPT two weeks from initial assessment when presenting with a positive UPT) [ 1 , 6 ]. Ectopic pregnancies continue to carry significant risk of maternal morbidity and mortality if diagnosis is delayed or missed [ 7 ]. Whilst ectopic pregnancies, viable intrauterine pregnancies and miscarriages are diagnosed using TVUS, persistent PUL and failed PUL are more heterogeneous and require TVUS with the aid of serial hCG values (PPUL) or a negative urine pregnancy test (FPUL) [ 1 ]. It is likely that both PPUL and FPUL are very early intrauterine or ectopic pregnancies that are too small to confidently identify on TVUS [ 6 ]. Biomarkers are used for PUL triage to facilitate screening of cases as efficiently and as accurately as possible pending a final diagnosis. Biomarkers currently available in clinical practice include hCG (a marker of trophoblast function), and progesterone (a marker of ovarian corpus luteal function) [ 1 , 8 , 9 ]. These are used in isolation, as a ratio (utilising serial measurements over time), or within multinomial logistic regression models [ 1 , [8] , [9] , [10] , [11] , [12] , [13] , [14] , [15] ]. The best available model (the M6) discriminates those at high-risk and low-risk of clinical complications well (area under the curve, AUC between 0.86–0.89), however, it has limited population generalisability [ 9 ]. Although novel marker candidates have been investigated, many are not appropriate or suitable for clinical use. Biomarkers used in isolation have poor sensitivity or specificity, with much of the evidence based on studies with small numbers, that have not been externally validated [ 9 , [16] , [17] , [18] ]. There is thus a clinical need to improve our ability to triage PUL pending diagnosis. MicroRNAs (miRNAs) are single stranded, non-coding, 19–25 nucleotide molecules which regulate messenger RNA stability and transcription [ 19 , 20 ]. A proportion of miRNAs are shed into the circulation in a stable and quantifiable form [ 19 , 21 , 22 ]. They are important gene regulators, and their expression profiles change with pathophysiology [ 19 , 21 ]. They have been implicated in cardiac, pulmonary, neurological, endocrinological, haematological, rheumatological and oncological processes or pathologies [ [23] , [24] , [25] , [26] , [27] , [28] , [29] , [30] , [31] , [32] ]. MiRNAs have been investigated in reproductive health and pregnancy outcomes. Outside of pregnancy, they are implicated in reproductive tract and fibroid development, as well as endometriosis [ [32] , [33] , [34] , [35] , [36] , [37] ]. MiRNAs appear to also play a role in abnormal trophoblast invasion and choriocarcinoma [ [38] , [39] , [40] ]. Antenatally and intrapartum, miRNAs have been linked to mesenchymal hamartomas, neural tube defects, gestational diabetes, pre-eclampsia, fetal growth restriction, pre-term birth and uterine contractility [ 19 , 21 , [41] , [42] , [43] , [44] , [45] , [46] , [47] , [48] , [49] , [50] ]. In early pregnancy, variable miRNA expression has been associated with blastocyst development, site and success of implantation, and thus miscarriage as well as EP [ 20 , 22 , 31 , [51] , [52] , [53] , [54] , [55] , [56] , [57] , [58] , [59] , [60] , [61] ]. MiRNAs have been suggested as novel biomarkers that could improve PUL outcome prediction beyond current clinical capabilities [ 53 ]. The objective of this study was to evaluate the potential of circulating miRNA markers in pregnancy of unknown location triage and ectopic pregnancy diagnosis.

The findings of this study have been filed and published under the remit of two patent applications: 11,075 (microRNA discovery work – GB 2113344·2) and 10,439 (microRNA biological validation work – PCT/GB2020/050,324). The authors have declared that no other conflict of interest exists.