The EV, circulating tumor cell (CTC) and circulating tumor DNA (ctDNA) are ideal biosources in blood-based liquid biopsies. They are used to detect, diagnose, and monitor cancer, and they could enable the early diagnosis of cancer, low costs by avoiding complex invasive procedures, tailoring molecular targeted treatments, improving convenience for cancer patients, and ultimately supplementing clinical oncological decision-making. However, the low detection rates of CTC and ctDNA and the suboptimal sensitivity for localized cancer patients hamper their implementation for early BC detection. Here, we report that molecular interrogation of blood exosomal long RNA can offer valuable diagnostics information for BC patients. In our study, we obtained exLR-seq expression profiles from 172 human plasma EV samples. Differences in the exLR levels were then compared between patients with BC, patients with breast benign disease and healthy participants, and a diagnostic signature for BC was finally established. In addition, the exLRs can provide a strong indication of the tumor subtype and can be employable as a predictive biosource upon response to NACT for LABC.
Blood-based liquid biopsy is an ideal method for cancer diagnosis. Tumor protein biomarkers, such as alpha-feto protein (AFP), prostate specific antigen (PSA), CEA, and CA 19 − 9, have been utilized for decades to detect early stage cancer, disease recurrence, and progression after therapy[19–21]. Although the utility of these tumor markers has been well established in clinical practice, they are not entirely specific and remain elevated in non-malignant conditions. In addition, they do not reflect the details of tumor biology, nor do they provide any predictive information on the response to therapy. However, the latest blood-based biomarkers include CTCs, ctDNAs, circulating miRNAs and tumor-derived EVs, and they have the potential of assessing real-time tumor responses to therapy as well as identifying dynamically resistant clones. EVs contain many bioactive molecules, such as proteins and nucleic acids, from parent cells, thereby playing vital roles in intercellular communication [22–25]. Recent studies have suggested that EVs represent an appealing source of diagnostic biomarkers[23, 25]. The analysis of EVs encompasses several advantages over CTCs and ctDNA due to their higher abundancies and stability in the bloodstream, as well as their functionality in supporting tumor-host cross talk or tumorigenesis.
Until Valadi et al. demonstrated that variable RNAs can be transported between cells by EVs in 2007, EVs have begun to attract the attention of scientists[26]. Recent studies have suggested that EVs contain large numbers of long RNAs[11]. Certain functional exLRs were involved in intercellular communication and were valuable for the prediction of treatment responses[27–29]. Moreover, some exLRs have been shown to be differentially expressed between cancer and healthy controls and could have potential for cancer diagnosis[30, 31]. To determine the differences in the plasma exLR profiles among BC, breast benign disease and healthy, we performed exLR profiling of plasma samples from all 172 participants using an optimized exLR-seq strategy we recently developed[11]. We identified 1552 exLRs that were differentially expressed in the BC samples compared with controls (benign + healthy) by the Mann-Whitney U test. We then used the xCell tool and found that these two groups had different cell type proportions. All of these previous studies suggested that exLRs were of great valuable and could serve as a non-invasive diagnostic biomarker.
To establish a plasma exLR d-signature for BC detection, we integrated the random forest algorithm and the LASSO method to shrink the number of variables. Finally, eleven exLR markers were selected and used to construct a BC classifier. For the 5 mRNAs that were used in the final d-signature, all were upregulated in both BC tissues and plasma EV when compared with normal breast tissues or plasma EV from healthy or benign controls, respectively. BEX2, KRT19, CTSG, and HOXB7 have been previously shown to be upregulated and to possess a tumor-promoting function in BC[32–35]. CBR3 is a member of the short-chain reductase family, and it has close correlation with radiotherapy sensitivity and chemotherapy sensitivity in esophageal cancer and BC[36, 37]. Moreover, Wu et al. demonstrated that CBR3 could play an important role in the development and malignancy of BC[38]. Little is known about the 5 pseudogene markers (AC104843.1, AL136981.2, NPM1P25, AL691447.3 and RNA5SP141) in previous cancer studies. The candidate circRNA (chr13: 42953948–42970670 -) is derived from the protein coding gene EPSTI1. EPSTI1 is a recently identified stromal fibroblast-induced gene that is highly overexpressed and critical for BC invasion and metastasis[39]. Previous study has demonstrated that circEPSTI1 (chr13: 43528083–43544806) sponges miR-4753 and miR-6809 to regulate BCL11A expression and affects cell proliferation and apoptosis in triple-negative BC[40]. Collectively, these results suggest that plasma EVs contain a considerable number of exLRs that could serve as potential biomarkers for BC detection.
The d-signature can distinguish BC from controls with high diagnostic accuracy in both training and validation cohorts (92.5% and 92.3%, respectively). Moreover, it is not influenced by the tumor burden and can act as an effective classifier to identify early stage (I/II) BC from controls (AUC = 0.94). The diagnosis of early stage BC is critical in clinical practice, which could enable immediate surgery, thereby improving the prognosis of BC.
BI-RADS based on mammography or ultrasound is routinely used for BC screening. However, there are disadvantages associated with these two forms of screening, including a high rate of false positive tests, frequent false negative results, and considerable implications for public health spending. For example, clinical doctors routinely choose invasive methods, either core needle biopsy or excision biopsy, to further examine patients who have a BI-RADS 4a or higher finding. If all patients with BI-RADS 4a findings receive invasive procedures, more than 90% of the lesions will be benign, which indicates unnecessary overtreatment for these patients. When integrating the diagnosis exLRs signature with BI-RADS 4a or higher findings, the diagnosis accuracy was approximately 91.9%. If we considered only the patients with BI-RADS 4a or 4b, the integrated predictive value was approximately 91.3%. Therefore, by integrating the exLRs signature and film findings, the possibility of false positive and the overtreatment of patients would be greatly decreased.
To further explore the role of exLRs in predicting the neoadjuvant chemotherapy treatment response in BC, we divided 48 patients who received NACT into pathological complete response (pCR) (n = 14) and non-pCR (n = 34) groups based on the post-surgical pathology. Different exLR profiles were found between these two groups. Through GSEA analysis, the steroid biosynthesis pathway is one of the most upregulated pathways in the non-pCR group. Many studies have implicated the function of the sterol synthesis pathway in tumor growth and response to treatment. For example, the sterol composition of the membrane has been shown to regulate EGFR signaling[41] and the sensitivity of head and neck cancer cells to apoptosis[42]. ExMSMO1 enriched in the non-pCR group and could distinguish non-pCR from pCR with an AUC of 0.79 (95% CI: 0.62 to 0.97, SD: 0.09) (Fig. 7D, E). MSMO1 is an intermediate enzyme in the cholesterol biosynthetic pathway. Previous study has indicated that the inactivation of MSMO1 markedly sensitized tumor cells to therapeutic anti-EGFR antibody via increased EGF receptor degradation[43]. In this study, we demonstrated that MSMO1 was overexpressed in BC and correlated with poor survival. Moreover, MSMO1 can decrease the sensitivity of BC cells to PAX and DOX by regulating the mTORC1 signaling pathway. In other words, these findings demonstrate that exMSMO1 can act as a predictive biomarker for neoadjuvant treatment efficacy of BC.
Several limitations in this study warrant mention. First, as a single-center study, it is uncertain whether this diagnosis signature is applicable to other populations composed of different cultures and ethnic groups. Future work with larger cohorts from multiple centers is still needed to externally validate our results. Second, the roles of exLRs in BC subtype distinguishing and treatment efficient predictions are still not fully evaluated in this study. Further work must be done on this issue.