Abstract

   Endometrial cancer (EC) is a major cause of death among gynecologic
   malignancies. To improve early detection of EC in patients, we carried
   out a large plasma-derived exosomal microRNA (miRNA) studies for
   diagnostic biomarker discovery in EC. Small RNA sequencing was
   performed to identify candidate exosomal miRNAs as diagnostic
   biomarkers in 56 plasma samples from healthy subjects and EC patients.
   These miRNA candidates were further validated in 202 independent plasma
   samples by droplet digital PCR (ddPCR), 32 pairs of endometrial tumors
   and adjacent normal tissues by quantitative real-time PCR (qRT-PCR),
   and matched plasma samples of 12 patients before and after surgery by
   ddPCR. miR-15a-5p, miR-106b-5p, and miR107 were significantly
   upregulated in exomes isolated from plasma samples of EC patients
   compared with healthy subjects. Particularly, miR-15a-5p alone yielded
   an AUC value of 0.813 to distinguish EC patients with stage I from
   healthy subjects. The integration of miR-15a-5p and serum tumor markers
   (CEA and CA125) achieved a higher AUC value of 0.899. There was also a
   close connection between miR-15a-5p and clinical manifestations in EC
   patients. Its exosomal expression was not only associated with the
   depth of muscular infiltration and aggressiveness of EC, but also
   correlated with levels of reproductive hormones such as TTE and DHEAS.
   Collectively, plasma-derived exosomal miR-15a-5p is a promising and
   effective diagnostic biomarker for the early detection of endometrial
   cancer.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12943-021-01352-4.

   Keywords: Endometrial cancer, Liquid biopsy, Plasma-derived exosomal
   miRNA, ddPCR, Cancer diagnosis, Early detection

Main text

   Endometrial cancer (EC) is the second highest incidence of gynecologic
   cancer [[53]1]. Patients have to undergo uterine apoxesis for accurate
   EC diagnosis, since there are no effective biomarkers [[54]2]. Exosomes
   originate from the endosome, and then fuse with the plasma membrane
   under the traction of molecular motors, and are released to the
   extracellular environment [[55]3, [56]4]. Exosomes are detected in body
   fluids such as plasma, urine, and amniotic fluid [[57]5]. Exosomes
   encapsulate biomolecules such as proteins and miRNAs, maintain their
   integrity in the circulation, and transfer them to recipient cells
   [[58]4]. MiRNA is the most abundant type in the RNA cargo of exosomes
   [[59]6, [60]7], and exosomal miRNAs (exomiRs) are usually
   tumor-specific [[61]8]. ExomiRs have received increasing attention in
   precision medicine, due to their non-invasiveness, and high
   accessibility and stability [[62]2, [63]9]. Recent studies have shown
   that exomiRs have the potential to be efficient biomarkers for the
   screening, diagnosis, and monitoring of cancers [[64]10–[65]13].
   However, exomiRs as biomarkers have not yet been reported in EC.

   To improve early detection of EC patients, we carried out a large
   plasma-derived exosomal miRNA study for biomarker discovery in EC
   ([66]Supplementary Methods). Candidates were identified by miRNA
   sequencing in plasma samples from healthy controls (HC) vs. EC
   patients, and were further validated in independent plasma samples and
   endometrial tumor tissues (Table S[67]1 and Figure S[68]1).
   Plasma-derived exosomal miR-15a-5p was identified as a promising
   diagnostic biomarker for early detection of endometrial cancer.

Identification of exomiRs for EC diagnosis

   Exosomes were isolated from plasma of EC patients and age-matched HC
   subjects. The previously reported method to identify the shape and size
   of exosomes [[69]5] was used with CD81, TSG101 and GM130 as positive or
   negative markers, respectively. Indeed, the fraction isolated from
   plasma was enriched in exosomes (Figure S[70]2).

   The miRNA sequencing was then performed in plasma-derived exosomes from
   25 EC and 31 HC subjects. On average, approximately 50 million reads
   were generated in each library, and 384 exomiRs were detected in each
   sample (Table S[71]2). Forty-nine miRNAs were differentially expressed
   between HC and EC groups (p < 0.01) (Fig. [72]1a and Table S[73]3).
   Eighteen of them also differentially expressed between tumor and
   adjacent normal tissues in The Cancer Genome Atlas (TCGA) EC samples
   [[74]14] (Table S[75]4 and Figure S[76]3). Next, a set of exomiRs from
   these 18 miRNAs were selected as a best panel to distinguish EC from HC
   subjects using random forest algorithm (Fig. [77]1b). Clustering
   analysis showed that these samples were largely divided into two
   distinct groups (EC vs HC) by these six exomiRs (Fig. [78]1c). The AUC
   for each exomiR ranged from 0.693 to 0.819 with a mean of 0.757. The
   AUC of the combined six exomiRs achieved 0.983 (Fig. [79]1d).
   miR-106b-5p, miR-107, miR-15a-5p, and miR-3615 were significantly
   up-regulated, while miR-139-3p and miR-574-3p were significantly
   down-regulated in plasma-derived exosomes of EC compared with HC (Fig.
   [80]1e). Consistent trends in expression of these exomiRs were observed
   in tumor tissues from TCGA EC patients (Fig. [81]1f).

Fig. 1.

   Fig. 1
   [82]Open in a new tab

   Dysregulated miRNAs in plasma exosomes of EC patients. a Volcano plot
   displaying differentially expressed miRNAs between plasma exosome of EC
   and HC. Small RNA sequencing was performed in plasma exosomes of 25 EC
   patients and 31 HC controls. There were 49 differentially expressed
   with p-value < 0.01 (i.e., above the dotted line). b Feature vectors
   forming the best panel to discriminate two different groups were
   determined by the random forest algorithm. Six miRNAs (miR-106b-5p,
   miR-107, miR-15a-5p, miR-139-3p, miR-3615, and miR-574-3p) that
   indicated higher variable importance in the random forest tree were
   identified as candidate biomarkers. c Hierarchical clustering analysis
   of 6 candidate miRNAs roughly divided plasma samples into two distinct
   groups (EC vs HC). d ROC curves to evaluate the sensitivity and
   specificity of 6 candidate miRNAs to discriminate EC and HC subjects. e
   The expression levels of 6 candidate miRNAs in plasma exosomes of EC
   (n = 25) and HC (n = 31) subjects. f The expression levels of 6
   candidate miRNAs in EC tumor tissues and matched para-carcinoma tissues
   from TCGA (n = 18). HC: healthy controls; PC, para-carcinoma tissues

Validation of diagnostic exomiRs by ddPCR in independent plasma samples

   Next, we applied droplet digital PCR (ddPCR) to verify these six
   exomiRs in an independent validation cohort including 115 EC and 87 HC
   plasma samples. Two stable high-abundance miRNAs (let-7b-5p and
   miR-26a-5p) were selected as endogenous references, due to their high