Sage Journals: Discover world-class research

Abstract

Early detection of cancers and their precise subtyping are essential to patient stratification and effective cancer management. Data-driven identification of expression biomarkers coupled with microfluidics-based detection shows promise to revolutionize cancer diagnosis and prognosis. MicroRNAs play key roles in cancers and afford detection in tissue and liquid biopsies. In this review, we focus on the microfluidics-based detection of miRNA biomarkers in AI-based models for early-stage cancer subtyping and prognosis. We describe various subclasses of miRNA biomarkers that could be useful in machine-based predictive modeling of cancer staging and progression. Strategies for optimizing the feature space of miRNA biomarkers are necessary to obtain a robust signature panel. This is followed by a discussion of the issues in model construction and validation towards producing Software-as-Medical-Devices (SaMDs). Microfluidic devices could facilitate the multiplexed detection of miRNA biomarker panels, and an overview of the different strategies for designing such microfluidic systems is presented here, with an outline of the detection principles used and the corresponding performance measures. Microfluidics-based profiling of miRNAs coupled with SaMD represent high-performance point-of-care solutions that would aid clinical decision-making and pave the way for accessible precision personalized medicine.

Keywords

cancer hallmarks early detection circulating miRNA software-as-medical-device biomarker prioritization prognostic model multiplexed detection microfluidics liquid biopsy precision medicine

Introduction

MicroRNAs (miRNAs) are small non-coding regulatory RNAs that bind to specific target gene transcripts perturbing their correct translation.¹ Given this epigenetic function, the dysregulation of miRNAs is a significant mechanism promoting the development and progression of various cancers. Several factors contribute to miRNA dysregulation, including errors in miRNA biogenesis machinery,^2,3 abnormal epigenetic changes,^4,5 copy number changes in genes coding miRNAs,^6,7 and aberrant transcription factor control of miRNA genes.⁸

The dysregulated miRNA drives the progression of cancer,^9,10 and based on the association with the specific hallmarks of cancer,¹¹ it could be classified; namely:

OncomiR: Oncogenic miRNA whose target genes are tumor suppressors. Their overexpression downregulates the protein expression of these target genes, thereby favoring the growth and metastasis of cancer cells. For example, miR-21 is frequently overexpressed, downregulating target tumor-suppressor genes PDCD4 in cancers of the colon/rectum and breast.^12,13 MiR-155 is an oncomiR whose high expression downregulates the protein expression of tumor suppressor genes SHIP1, VHL, and WEE1 in lymphoma and solid tumors.^14,15

TS-miR: Tumor Suppressor miRNA whose target genes are oncogenes. Low expression of TS-miR has the effect of increasing the protein expression of oncogenes, thereby promoting the malignant progression of cancer. Downregulation of miR-15 and miR-16 increases the expression of anti-apoptotic molecules BCL2, CDC2, and ETS1, thereby enhancing the progression of cancer.^6,16

MetastamiR: Metastasis-related miRNAs that modulate (either promote or suppress) the progression to metastasis by targeting genes participating in the processes of the metastatic cascade, like migration, invasion, colonization, cancer stemness, and the epithelial-to-mesenchymal transition (EMT).¹⁷ For eg, miR-10b is a metastamiR that negatively regulates the T-box transcription factor 5 and homeobox D10 in cell invasion, colonization, and metastasis in multiple cancers.¹⁸

AngiomiR: miRNAs that maintain vascular integrity and regulate angiogenic processes, for eg, miR-126 in breast cancer.¹⁹ Some angiomiRs could be detected in exosomes; for eg, exomiR-1229 secreted by colorectal cancers that modulate the HIPK2/VEGF pathway.²⁰

Immuno-miR: A class of miRNAs that have a profound effect on T-cell biology regulating their development, maturation, differentiation, activation, apoptosis, and aging. They are implicated in enabling the cancer tissue to evade immune surveillance.²¹ MiR-155 is a key modulator of the inflammatory response in cancers of epithelial origin including those of the colon, rectum, and breast, as well as in liquid cancers like B-cell lymphoma.²²

Significant altered expression of miRNAs is frequently reported in the literature for various cancers; for eg, miR-221, miR-301, miR-376a, and miR-21 in pancreatic cancers,^23,24 and miR-155 and let-7a-2 in lung cancer.²⁵ MiRNAs are present in body fluids such as plasma, urine, cerebrospinal fluid, pleural effusions, and colostrum, circulating tumor cells (CTCs), exosomes, and stool^26,27; for eg, miR-21 in the serum samples of B cell lymphoma patients.²⁸ Furthermore miRNAs are stable,^29,30 and circulating miRNAs with signal aberrant expression constitute an important class of biomarker candidates for screening, diagnosis, subtyping, and prognosis of cancer. In a recent study, liquid biopsy has been used to derive early-stage miRNA biomarkers for diagnosing esophageal squamous cell carcinoma.³¹ Liquid biopsy is a radical non-invasive method for tumor sampling in clinical oncology, and facilitates the simultaneous profiling of multiple biomarkers, paving the way for precise targeted therapy and optimal patient outcomes.³² Figure 1 shows the different dysregulation classes of miRNAs that could be detected using a liquid biopsy. In the first part of this review, we summarize data-driven approaches for establishing an optimal panel of miRNA biomarkers for the desired objectives.

Figure 1.

MiRNAs and cancer hallmarks. Aberrations in miRNA functions are crucial to tumorigenesis and the development of cancer hallmarks. Such miRNAs could serve as early-stage biomarkers detectable in a liquid biopsy.

Though circulating miRNAs make attractive early-stage biomarker candidates, they pose substantial detection challenges due to their very low prevalence (∼0.01% of total RNA in human body), small size (∼20-30 nucleotides), low copy number (∼10 000 copies per cell), and poor sequence differentiation (miRNAs in a given family differing in just one base).³³ In addition, miRNA concentrations between plasma and serum are different, and thus the collection source of circulating miRNAs exerts a critical role.³⁴ These constraints require the development of detection strategies with single-nucleotide specificity to a minute target RNA population against a background of abundant non-target RNA molecules. Quantitative Reverse Transcriptase Polymerase Chain Reaction (qRT-PCR), next-generation sequencing methods like miRNA-Seq, and Northern blotting constitute standard methods for profiling miRNA transcript abundance.^35,36 Emerging techniques for detecting miRNAs include using synthetic DNA and RNA nanostructures with better performance profile.^33,37,38 However, a viable alternative technology to miRNA detection and quantification in body fluids would be microfluidics-based strategies³⁹ that afford real-time point-of-care solutions with high specificity, high sensitivity, low detection limits, quicker response, and less sample volume. Other features of microfluidics devices include advantages in cost, time, and user-friendliness, which help in fostering the adoption of such devices in the development of biomarker-based analytical diagnostics.^40,41 In the latter section of this review, we discuss different options for incorporating microfluidic strategies towards profiling an identified miRNA biomarker panel.

AI Assisted miRNA Biomarker Panel Discovery

miRNA Biomarker Discovery

Genome-wide –omics big data have marked a paradigm shift in the process of biomarker discovery from the traditional hypothesis-driven knowledge-based approach to a hypothesis-agnostic data-driven modern approach. The data-driven approach is systematic, ideally unbiased (up to assumptions in experimental design), and cost-efficient with respect to the identification of biomarkers optimal for a phenotype of interest. It must be noted that bias in experimental design will induce an implicit hypothesis in the collected data. Well-known public repositories of big omics data include the Cancer Genome Atlas (TCGA),⁴² the Gene Expression Omnibus (GEO),⁴³ and the International Cancer Genome Consortium (ICGC).⁴⁴ TCGA houses multiple omics big data (including miRNA-Seq) with > 20 000 cancer and matched normal samples for over 33 cancer types. GEO is a global repository supported by the National Center of Biotechnology Information (NCBI) that archives high-throughput gene expression and genomics data. ICGC is an international cancer research collaboration for sharing –omics data pertaining to nearly 86 cancer types. Dedicated databases of miRNA expression profiles include miRmine,⁴⁵ DIANA-miTED,⁴⁶ TissueAtlas,⁴⁷ and Chinese Glioma Genome Atlas.⁴⁸ Other databases supporting the investigation of miRNA roles in cancers include dbDEMC,⁴⁹ miRCancer,⁵⁰ miR2Trait,⁵¹ and CancerMIRNome.⁵²

MiRNA-Seq –omics data could be obtained using a variation of RNA-Seq, where larger RNA fragments (mRNAs and rRNAs) are initially excluded by size selection, read counts obtained, and transformed to an expression matrix. Differentially expressed miRNAs could be found from the miRNA-Seq –omics expression matrix using a log fold change between the cancer samples and the controls, along with a significance test. More complex workflows could yield specific DE miRNAs pertaining to some clinical metadata of interest like clinical or pathological stage, survival status, response to treatment, etc. The clinical annotation of the samples may be used to design regression models that could yield specific biomarkers containing predictive utility with reference to the classes in the clinical annotation.

Classes of miRNA Biomarkers

A prime interest in cancer research is the identification of early-stage biomarkers. Below, we illustrate the use of clinical annotation in the form of AJCC staging to identify early-stage biomarkers. Our two-tiered protocol has been described earlier^53,54 in summary:

(i) Tier-I: The miRNA features specific to a clinical variable can be identified by using a two tier contrast that was developed in our lab. A contrast between the controls on the one hand and each class in the clinical variable of interest on the other could identify class-specific miRNA biomarkers. In the case of AJCC staging data, this contrast is performed using the following design matrix:

y = α + β_{1} x_{1} + \dots β_{i} x_{i . . .} + β_{n} x_{n} \dots

(1)

where the independent variables indicate the class, and n is the number of classes of the clinical variable of interest, here Stage I, Stage II, Stage III, and Stage IV. The intercept α is the baseline expression estimated from the controls, and β_i represents the log fold change (lfc) in the mean expression of Stage-i with respect to the controls.

(ii) Tier-II: Contrast between the various classes of the clinical variable of interest could identify class-salient biomarkers. This between-class analysis is performed for all pairs of classes, using the results from the following design matrix:

y = β_{0} x_{0} + \dots β_{i} x_{i \dots} + β_{n} x_{n} \dots . .

(2)

where the controls themselves constitute one of the indicator variables (x_i), and the β_i are coefficients estimated only from the corresponding samples.

The design matrix of contrasts is then applied to find the log fold change and associated P-value for each contrast in both tier-I and tier-II. This could be done by supplying the contrast matrix as an argument to the contrastsFit function in limma. The contrast against the control demands a stronger significance (adj. P < .001) than the pairwise contrasts (P < .05). Using an |lfc| threshold ∼ 2 for each contrast, tier-I constraints yield significant upregulated or downregulated DE miRNAs. Application of tier-II constraints to the stage-specific miRNAs (from tier-I) yields stage-salient miRNAs.

In addition, an ordinal model could be used to model miRNA expression, treating the cancer stage as a numeric variable. This is given by the following equation:

y = a X + b \dots . .

(3)

where X denotes stage annotation, with possible numeric values of 0, 1, 2, 3, and 4 corresponding to control, Stage I, Stage II, Stage III, and Stage IV respectively. Monotonicity in expression could be gauged from the following inequalities for mean expression:

Control < I < II < III < IV \dots . .

(4)

denoting monotonic upregulation; and

Control > I > II > III > IV \dots . .

(5)

denoting monotonic downregulation. Significant DE genes that exhibit either pattern are flagged as monotonically expressed (ME) genes. Table 1 presents a representative set of biomarker classes that could be useful in the development of AI devices for the overarching goal of effective cancer management.

Table 1.

Classes of miRNA Biomarkers That Could Be Useful in Building AI-Based Cancer Diagnosis and Prognosis Models.

S.No	Biomarker class	Definition
1	Differentially Expressed (DE)	\|lfc\| > 2 for some stage w.r.to controls
2	Significant DE	DE in all stages with tier-I P-value < .001.
3	Stage-specific	Significant DE with maximum \|lfc\| in some stage
4	Stage-salient	Stage-specific that passes all tier-II significance filters against the other stages
5	Monotonically Expressed (ME)	Significant DE with monotonic mean expression across the stages
6	Progression-significant	Stage-salient that is also ME; typically Stage-IV salient
7	Ordinal-model significant	Significant from model with the stage as numeric (adj. P < .05) [Eqn. 3]
8	Stage-agnostic	ME that is ordinal-model significant (but not stage-salient)
9	Hub driver	Significant DE that is also network central
10	Prognostic	ME or Stage-salient that is significantly associated with overall survival (OS)

Construction of ML Model

The identified biomarkers could be used in the construction of an ML model for screening, early-stage diagnosis, and subtyping purposes. Machine learning models are generally characterized by the {T,P,E} formalism, where they learn from experience E of –omics expression data with respect to the class of tasks T of interest, improving on some performance measure P.^55,56 The generalized workflow for building such models for oncology applications is presented in Figure 2. The response variable of interest is extracted from clinical metadata, and appended to the miRNA expression data matrix. In a pre-processing step, miRNAs with minimal variance in expression (expression σ < 1) and samples with missing values in the clinical variable of interest are removed. The dataset is log₂-transformed through voom, which stabilizes the variance in expression in the remaining genes.⁵⁷ The dataset is split into train dataset and test dataset, and biomarker identification is performed with the train dataset to prevent leakage into the test dataset. Potential biomarkers could belong to one of the classes enumerated in Table 1, especially stage-salient and ME miRNAs. Feature selection is then performed to optimize the feature space up to 16 biomarkers (usually three or four) for detection with multiplexed microfluidics. This may be done using techniques like Recursive Feature Elimination (RFE),⁵⁸ which uses backward variable selection, and Boruta wrapper algorithm,⁵⁹ which uses top-down elimination of features irrelevant to the learning problem. The consensus between the feature selection techniques could be used as the final feature space for ML model development. A wide range of algorithms that work with tabular data could be used for model construction, including SVM,⁶⁰ neural networks (shallow and deep),⁶¹ and ensemble methods such as Random Forest⁶² and XGBoost.⁶³ Model hyperparameters may be optimized with ten-fold cross-validation on the train dataset using libraries such as caret⁶⁴ and KerasTuner.⁶⁵ The optimized models are evaluated on the unseen test set and the best model then determined. The performance of the best model may be assessed on an external out-of-domain dataset. Comprehensive information on developing machine intelligence in oncology could be found in contemporary reviews of the state of the art.^66–71 Successful models represent a Software-as-Medical-Device (SaMD) endpoint⁷² that could then be refined with real-time clinical data for continuous deployment and integration. Examples of such models built using miRNA biomarkers have been recently reported.^73–75

Figure 2.

A general workflow for the development of an ML model with miRNA biomarker features for varied diagnostic applications. The model is created with miRNA –omics data and could lead to the production of SaMD.

Construction of Prognostic Model

The development of prognostic models offers another application for miRNA biomarkers to improve the practice of medicine. Prognostic models are combinations of predictor variables that score the risk of encountering a well-defined endpoint, like death or relapse of disease, in a specified period for individuals with a given state of health.^76–78 Parametric techniques to model time-oriented data include logistic regression, Kaplan–Meier curves, and Cox proportional hazards models. Neural networks represent an emerging class of nonlinear nonparametric prognostic models.⁷⁹ Prognostic models could be calibrated and evaluated for discrimination between patients in two groups in a given period of time using the concordance index.^80,81 A generalized statistical survival analysis for developing a prognostic model based on miRNA biomarkers would involve the following steps (Figure 3):

constructing univariate Cox regression models for each biomarker of interest, including stage-salient and ME miRNAs, and identifying the prognosis-significant miRNAs (P < .05)

applying robust feature selection methods such as LASSO cox regression or SVM-RFE to reduce the dimensionality

constructing a multivariate Cox regression with backward variable selection to arrive at an optimal prognostic signature of three to four miRNAs. This step yields an expression for the risk score in terms of the optimal biomarker panel:

\begin{aligned} Risk_score & = β_{1} \times miRN A_{1} + \dots \\ + β_{i} \times miRN A_{i} + \dots + β_{n} \times miRN A_{n} \dots . . \end{aligned}

(6)

where n is the size of the prognostic signature, miRNA _i denotes the expression level of the i^th miRNA, and β_i denotes the effect size (or weight) of the i^th miRNA. The risk score could be used to stratify patients into high-risk and low-risk prognoses. Such miRNA-based prognostic models could be augmented with clinical factors from imaging and laboratory procedures to improve patient stratification for better treatment outcomes,⁸² and yield composite models such as the palliative prognostic score, palliative prognostic index, and Glasgow prognostic score.⁸³ Some alternatives to survival analysis for prognostication include the use of non-parametric models, and tests such as Gehan's (Gehan-Breslow-Wilcoxon), Prentice, and Taron-Ware.⁸⁴

Microfluidic Systems for Profiling miRNA Signatures

The miRNA biomarkers comprising the final diagnostic or prognostic final models could be detected and quantified using microfluidic-based multiplexed technologies; Figure 4 summarizes a scheme for translating biomarker discovery into a microfluidics-based SaMD. Microfluidics platforms are combinations of fluidic operating units including channels (1-1000 mm in size), chambers, and nanoscale structures like pillars, wires, and tubes. They serve to integrate, automate, multiplex, and thereby streamline biochemical processes, offering significant advantages in performance and cost. Microfluidics have been used to study the unique “fluidic signatures” of cancer, including growth, metastases,^85,86 induction of angiogenesis,^87,88 and the tumor microenvironment.^87,88 In particular, they have been used to study CTCs, which carry essential cancer biomarkers but possess short half-life, posing a detection challenge that is overcome with the use of probes such as aptamers, peptides, immune-affinity proteins, and cytokines.^35,37

Figure 3.

A schematic for the development of a prognostic model with miRNA biomarker features. The model is created with miRNA –omics data and could lead to the production of SaMD.

Figure 4.

Microfluidic detection of miRNA biomarkers. A scheme for developing and deploying the microfluidics-based SaMD is shown.

Microfluidics-based devices are perfectly suited for converting miRNA biomarker signatures of cancers into clinical outcomes such as screening, diagnosis, and prognosis. These devices could be characterized in terms of the design of the microfluidic platform as well as the detection principle used. Platform designs could be based on nanomaterials, pico-droplets, lab-on-a-chip, and paper-based analytical devices (μPADs). The detection methods could be varied and encompass optical, electrochemical, and conventional PCR readouts. Table 2 summarizes various microfluidics-based devices that have been used for detecting miRNA cancer biomarkers, along with their respective detection principles.

Table 2.

A Classification of Microfluidics Strategies by Their Detection Principle. The Specific Biomarkers Identified and the Performance (if Available) of the Respective Strategies are Noted, Along with the Primary Reference Source. Performance is Given in Terms of Limit of Detection (LOD), Analysis or Detection Time, Sample Volume, Sensitivity, Specificity, and/or Linear Detection Range

S.No	Microfluidic Device	Detection method	Biomarker	Cancer	Performance	Ref.
1	Nanoparticle-based	Fluorescence	miR-125, miR-126, miR-191, miR-155, and miR-21	Breast	LOD of 0.146 aM; linear detection range between 1 aM and 10 nM; detection time 35 min; sample requirement 2 µL	Chu et al, 2021 ⁸⁹
2	Nanoparticle-based	Colorimetry	let-7a, miR-145, and miR-21	Lung	LOD of 260 fM	Lee et al, 2020⁹⁰
3	Self-priming Polydimethylsiloxane (PDMS)	Fluorescence	miR-100, miR-155, and Let-7a	Breast	LOD of 42.635 pM	Zou et al, 2021 ⁹¹
4	Self assembled PolyL-lysine (PLL)	Fluorescence	miR-4732, miR-k12-5, miR-3646, and miR-4484	Breast	LOD of 1 pM; linear detection range between 100 nM and 1 pM; detection time of 30 min	Gao et al, 2020 ⁹²
5	PolyMethyl Methacrylate (PMMA)	Fluorescence	miR-21	Breast	-	Salima et al, 2017⁹³
6	Magnetic bead coated	Fluorescence	miR-21	Ovarian	LOD of 1.4 aM	Sung et al, 2021⁹⁴
7	Laminar flow-assisted dendritic amplification (LFDA)	Fluorescence	miR-196a and miR-331	Cancer biomarker	LOD of 0.045 pM for miR-196a and 0.45 pM for miR-331	Kim et al, 2019 ⁹⁵
8	Droplet-based Quantum dots	Fluorescence	miRNA-20, miRNA-21, miRNA-155, and miRNA-221	Cancer biomarker	LOD range of 35-39 pM	Ye et al, 2020 ⁹⁶
9	μPADs	Fluorescence	miRNA-21 and miRNA-31	A549 and HeLa cancer cell lines	LOD of 0.20 fM for miR-21 and 0.50 fM for miR-31	Cai et al, 2020 ⁹⁷
10	Au Nanoflare Probe	Fluorescence	miR-1246	Breast	Showed a sensitivity of 100% and a specificity of 92.9% Assay takes about 4 h and requires a plasma sample of 40 µL	Zhai et al, 2018 ⁹⁸
11	Tethered Cationic Lipoplex Nanoparticle (tCLN) biochip	Total Internal Reflection Fluorescence	miR-21, miR-25, miR-155, miR-210, and miR-486	Lung	Sensitivity 96.9%, specificity 93.3%, assay takes ∼ 2.5 h, and sample volume ∼ 60 µL serum	Liu et al, 2020 ⁹⁹
12	Triggerable Mutually Amplified Signal (TMAS) Chip	SERS	miR-21	Breast	LOD of 2.33 fM and assay takes ∼ 30 min	Wang et al, 2019 ¹⁰⁰
13	Surface acoustic wave exosome lysis and an ion exchange nanomembrane	Electrochemical	miR-550	Pancreatic	LOD of 2 pm with total analysis time ∼1.5 h	Taller et al, 2015 ¹⁰¹
14	CRISPR/Cas13a	Electrochemical	miRNA-19b and miRNA-20a	Brain	LOD of 10 pM; sample volume < 0.6 µL; readout time 9 min and overall assay time < 4 h	Bruch et al, 2019 ¹⁰²
15	Superparamagnetic core–shell nanoparticles	Electrochemical	miRNA-122	Liver cells in hepatitis, alcoholism, and obesity	LOD of 2-18 pM	Horny et al, 2021 ¹⁰³
16	µPADs	Electrochemical	Synthetic miR-21	-	LOD of 0.434 fM	Sun et al, 2017¹⁰⁴
17	3D printed	Droplet-based PCR	miR21	MCF-10A and MDA-MB-231 cell lines	-	Jiao et al, 2019 ¹⁰⁵
18	Fluidigm^Ⓡ high throughput platform	qRT-PCR	miR-30c, miR-26b, miR-451, miR-223, miR-24, miR-874, miR-1274a, miR-1207-5p, miR-93, miR-106a, and miR-20b	Prostate	-	Moltzahn et al, 2011 ¹⁰⁶
19	Horseshoe-shaped orifice micromixer (HOMM) chip	PCR	miR-9, miR-16, miR-21, and miR-429	Breast	Sensitivity 96.8% and specificity of 80%	Kim et al, 2021 ¹⁰⁷
20	Polyurethane-based nanoporous anion exchange membrane	Ion conductivity	miR-146a	Oral	LOD of 1 pM and assay time ∼ 20 min	Senapati et al, 2014¹⁰⁸

The optical detection methods include Surface Enhanced Raman Spectroscopy (SERS),¹⁰⁹ Surface Plasmon Resonance (SPR),¹¹⁰ upconversion nanoparticles,¹¹¹ fluorimetry, Förster resonance energy transfer (FRET),¹¹² and colorimetry.¹¹³ All these methods share very high sensitivity relative to other detection principles. SERS achieves enhancement of the regular Raman spectroscopy signal of analytes (including single-molecule miRNAs) by a factor of ∼ 10¹⁰-10¹¹, either directly via adsorption onto metal nanostructured surfaces or indirectly via a reporter mechanism, and with a spectral profile tailored for multiplexing and stable quantitative readouts.¹¹⁴ SPR refers to the collective coherent oscillations of excited delocalized electrons on a thin metal surface due to light incident at a particular angle to the surface. The specific angle of incidence is a function of the characteristic target such as miRNAs, leading to differential detectable signals.¹¹⁵ SPR and SERS display the lowest detection limits (< 1 aM miRNAs); however, these techniques pose challenges in scaling and reproducibility. Upconversion nanoparticles absorb low-energy photons and emit high-energy fluorescence post target capture, with an ability to detect 0.1 nM, but remain limited by conversion efficiency. Fluorescence-based detection remains the most sought-after readout method, due to the ease of detection and sensitivity. Laser-induced fluorescence (LIF) is a special type of fluorescence detection method with an ultra-sensitive detector, and works with samples of different dimensions and states (solid, liquid, gas). Further, the LIF instrument is small, cheap, and portable making it the ideal detector for microfluidic paper-based analytical devices (μPADs).¹⁰⁹ μPADs are microfluidics-based systems that predominantly make use of fluorescence and colorimetric detection principles. μPADs are advantageous for numerous reasons: one, in terms of microfluidics, fluid flow could be easily controlled by capillary forces on them, ruling out the requirement of any external driving forces; two, the cellulose paper substrate remains compatible to carry out various chemical and biochemical reactions and furnishes passive liquid transfer owing to its hydrophilicity, porosity, and homogeneous structure; three, PADs are highly disposable due to their low sample consumption, cheap cost, and low weight; four, paper allows for preserving various biomolecular reagents like antibodies, enzymes, nucleic acids, and cells by simple freeze drying of the storing areas.^36,112 In the work of 96, LIF was used for the detection of miR-21 and miR-31.⁹⁷ Nanomaterial-based designs utilize colorimetric properties of gold nanoparticles (AuNPs). Two-dimensional nanomaterials like graphene oxide, MoS_2,¹¹⁰ and WS₂¹¹¹ have been employed for miRNA capture, leading to a readout based on FRET. FRET refers to the non-radiative energy transfer within a light-sensitive donor–acceptor molecular couple, and is extremely sensitive to extremely small changes in donor–acceptor distance, their relative orientation, and the lifetime of the donor excited state. Complementary techniques for miRNA capture could be used: (i) the target miRNA binds to a complementary RNA sequence, which is coupled to a FRET pair and immobilized in a micro-channel, thus giving rise to a fluorescence signal⁸⁹; (ii) the target miRNA acts to separate the donor and acceptor fluorophores in a FRET pair, resulting in fluorescence quenching; and (iii) two different oligonucleotide probes, with donor and acceptor fluorophore dyes, hybridize to adjacent regions of the target miRNA, yielding the FRET signal of proximity of fluorophore dyes.¹¹⁶ Despite the gains in sensitivity and reliability, the clinical utility of colorimetric and FRET-based microfluidic devices could be circumscribed by the requirement that each target miRNA have its own detection element.

Electrochemical detection provides highly sensitive detection with quick response times.¹¹³ The use of microelectrodes relaxes the necessary analyte volume, and the electrochemical system directly transduces the biological process into an output signal. Thus, by bringing the best of both the worlds of microfluidic systems and electrochemical readout systems, diverse strategies for biomarker detection have been invented (Table 2). For instance, Sun et al coupled μPADs with electrochemical readout, designing a microfluidic device with (i) gold (Au) nanorods for improving the conductivity of the cellulose paper substrate; and (ii) cerium dioxide—Au conjugated with glucose oxidase enzyme for serving as the electrochemical probe.¹⁰⁴ The synergistic composite of cerium dioxide—Au conjugated with glucose oxidase enzyme helps in improving the device's sensitivity and specificity.

In addition to the above, microfluidics-based strategies offer refinements to the conventional PCR-based detection method, overcoming inefficiencies in heat transfer and reaction times. Microfluidics afford controllable thermal cycling, which could be used in droplet-based PCR systems for detecting miRNA biomarkers.¹⁰⁸ Droplet-based microfluidic systems have been used for nucleic acid amplification-free single-molecule RNA detection, via Cas13a assays with fluorescence readout.¹¹⁷ Such devices could be used to count cell-free miRNA biomarkers from serum samples, which could be used for cancer screening applications (for eg,[³⁴]). New directions for microfluidics-based strategies include the demonstration of loop mediated isothermal amplification (LAMP) followed by fluorescence measurement for multiplexed detection of multiple strains of pathogen DNA.¹¹⁸ DNA markers have been shown effective in multicancer early detection tests based on liquid biopsies,¹¹⁹ a result that invites further research into mining the omics biomarkers driving cancer hallmarks.

In summary, the key drivers for adopting microfluidic systems for miRNA detection include: (i) modular nature, with the freedom to mix and match the detection platform and readout physics; (ii) requirement of low sample volume; (iii) affordance of spatiotemporal control; and (iv) multiplexed sub-pM sensitivity, all of which help in the design of efficient portable high-performance systems for the automated detection of miRNA signatures of cancer. On the other hand, some major limitations include: (i) charge-based or affinity-based non-specific binding of detection probes (eg, to device surfaces), and fouling of microchannels, especially in the case of nanoparticle/polymer-based microfluidic detection systems; (ii) lack of unified protocols and standardized benchmarks, raising quality control issues that frustrate clinical validation; (iii) high costs of development and clinical trials, especially in the case of diagnostics in oncology; and (iv) regulatory compliance and market adoption issues that tend to plague novel maturing technologies, especially pronounced in the case of diagnostic medical devices.¹²⁰ Most of these challenges are common to emerging high-throughput technologies, and microfluidics-based SaMD-coupled miRNA profiling could be no exception. Potential drawbacks notwithstanding, continued research of microfluidics-based devices would overcome the obstacles, and deliver increased optimization and adoption of such quantitative diagnostics.

In conclusion, miRNAs are an effective and compelling class of biomarkers for the early detection of cancers towards clinical management of these deadly killers. AI-based methods enable the discovery of early-stage miRNA signatures for cancer screening, diagnosis, and prognosis, yielding SaMDs that could be deployed in conjunction with microfluidic sensing platforms for delivering real-time diagnostic and prognostic information. The power and precision of these systems is likely to drive their increasing currency in cancer diagnosis especially as a point-of-care solution in resource-constrained settings.

Footnotes

Abbreviations

Acknowledgments

We would like to thank Profs. K.S. Rajan and R. John Bosco Balaguru for helpful discussions. We are grateful to the management of SASTRA Deemed University for infrastructure and support. We would like to thank reviewers for the careful reading of our manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Ashok Palaniappan

References

Bartel

. Metazoan MicroRNAs. Cell. 2018;173(1):20-51. doi:10.1016/j.cell.2018.03.006

Iliou

da Silva-Diz

Carmona

, et al. Impaired DICER1 function promotes stemness and metastasis in colon cancer. Oncogene. 2014;33(30):4003-4015.

Völler

Reinders

Meister

Bosserhoff

. Strong reduction of AGO2 expression in melanoma and cellular consequences. Br J Cancer. 2013;109:3116-3124. doi:10.1038/bjc.2013.646

Tsuruta

Kozaki

Uesugi

, et al. miR-152 is a tumor suppressor microRNA that is silenced by DNA hypermethylation in endometrial cancer. Cancer Res. 2011;71(20):6450-6462.

Furuta

Kozaki

Tanaka

Arii

Imoto

Inazawa

. miR-124 and miR-203 are epigenetically silenced tumor-suppressive microRNAs in hepatocellular carcinoma. Carcinogenesis. 2010;31(5):766-776.

Calin

Dumitru

Shimizu

, et al. Frequent deletions and down- regulation of micro-RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proc Natl Acad Sci USA. 2002;99(24):15524-9.

Tagawa

Seto

. A microRNA cluster as a target of genomic amplification in malignant lymphoma. Leukemia. 2005;19(11):2013-2016.

Chang

Lee

, et al. Widespread microRNA repression by Myc contributes to tumorigenesis. Nat Genet. 2008;40(1):43-50.

Kasinski

Slack

. MicroRNAs en route to the clinic: Progress in validating and targeting microRNAs for cancer therapy. Nat Rev Cancer. 2011;11(12):849-864.

10.

Stahlhut

Slack

. MicroRNAs and the cancer phenotype: Profiling, signatures and clinical implications. Genome Med. 2013;5(12):111.

11.

Hanahan

Weinberg

. Hallmarks of cancer: The next generation. Cell. 2011;144(5):646-674.

12.

Asangani

Rasheed

Nikolova

, et al. MicroRNA-21 (miR-21) post- transcriptionally downregulates tumor suppressor Pdcd4 and stimulates invasion, intravasation and metastasis in colorectal cancer. Oncogene. 2008;27(15):2128-2136.

13.

Frankel

Christoffersen

Jacobsen

Lindow

Krogh

Lund

. Programmed cell death 4 (PDCD4) is an important functional target of the microRNA miR-21 in breast cancer cells. J Biol Chem. 2008;283(2):1026-1033.

14.

Gironella

Seux

Xie

, et al. Tumor protein 53-induced nuclear protein 1 expression is repressed by miR-155, and its restoration inhibits pancreatic tumor development. Proc Natl Acad Sci USA. 2007;104(41):16170-5.

15.

Tili

Croce

Michaille

. miR-155: On the crosstalk between inflammation and cancer. Int Rev Immunol. 2009;28(5):264-284.

16.

Cimmino

Calin

Fabbri

, et al. miR-15 and miR-16 induce apoptosis by targeting BCL2. Proc Natl Acad Sci USA. 2005;102(39):13944-9.

17.

Hurst

Edmonds

Welch

. Metastamir: The field of metastasis-regulatory microRNA is spreading. Cancer Res. 2009;69(19):7495-7498.

18.

Teruya-Feldstein

Weinberg

. Tumour invasion and metastasis initiated by microRNA-10b in breast cancer. Nature. 2007;449(7210):682-688.

19.

Wang

Olson

. Angiomirs–key regulators of angiogenesis. Curr Opin Genet Dev. 2009;19(3):205-211.

20.

Zhang

, et al. Exosomal miR-1229 derived from colorectal cancer cells promotes angiogenesis by targeting HIPK2. Int J Biol Macromol. 2019;1(132):470-477.

21.

Kroesen

Teteloshvili

Smigielska-Czepiel

, et al. Immuno-miRs: Critical regulators of T-cell development, function and ageing. Immunology. 2015;144(1):1-10.

22.

Eis

Tam

Sun

, et al. Accumulation of miR-155 and BIC RNA in human B cell lymphomas. Proc Natl Acad Sci USA. 2005;102(10):3627-3632.

23.

Stahlhut

Slack

. MicroRNAs and the cancer phenotype: Profiling, signatures and clinical implications. Genome Med. 2013;5:111. doi:10.1186/gm516

24.

Lee

Gusev

Jiang

, et al. Expression profiling identifies microRNA signature in pancreatic cancer. Int J Cancer. 2007;120(5):1046-1054.

25.

Roldo

Missiaglia

Hagan

, et al. MicroRNA expression abnormalities in pancreatic endocrine and acinar tumors are associated with distinctive pathologic features and clinical behavior. J Clin Oncol. 2006;24(29):4677-4684.

26.

Yanaihara

Caplen

Bowman

, et al. Unique microRNA molecular profiles in lung cancer diagnosis and prognosis. Cancer Cell. 2006;9(3):189-198.

27.

Preethi

Selvakumar

Ross

, et al. Liquid biopsy: Exosomal microRNAs as novel diagnostic and prognostic biomarkers in cancer. Mol Cancer. 2022;21(1):54.

28.

Weber

Baxter

Zhang

, et al. The microRNA spectrum in 12 body fluids. Clin Chem. 2010;56(11):1733-1741.

29.

Lawrie

Gal

Dunlop

, et al. Detection of elevated levels of tumour-associated microRNAs in serum of patients with diffuse large B-cell lymphoma. Br J Haematol. 2008;141(5):672-675.

30.

Park

Zhou

Elashoff

, et al. Salivary microRNA: Discovery, characterization, and clinical utility for oral cancer detection. Clin Cancer Res. 2009;15(17):5473-5477.

31.

Miyoshi

Zhu

Luo

, et al. A microRNA-based liquid biopsy signature for the early detection of esophageal squamous cell carcinoma: A retrospective, prospective and multicenter study. Mol Cancer. 2022;21(1):44.

32.

Lone

Nisar

Masoodi

, et al. Liquid biopsy: A step closer to transform diagnosis, prognosis and future of cancer treatments. Mol Cancer. 2022;21:79. https://doi.org/10.1186/s12943-022-01543-7

33.

Mlcochova

Hezova

Stanik

Slaby

. Urine microRNAs as potential noninvasive biomarkers in urologic cancers. Urol Oncol: Semin Orig Invest. 2014;32(1):41.e1-41.e9. doi:10.1016/j.urolonc.2013.04.011

34.

Tiberio

Callari

Angeloni

Daidone

Appierto

. Challenges in using circulating miRNAs as cancer biomarkers. BioMed Res Int. 2015;2015:731479. doi: 10.1155/2015/731479

35.

Chandrasekaran

MacIsaac

Dey

, et al. Cellular microRNA detection with miRacles: MicroRNA-activated conditional looping of engineered switches. Sci Adv. 2019;5(3):1-8.

36.

Mousavi

Mahdian

SMA

Ebrahimi

, et al. Microfluidics for detection of exosomes and microRNAs in cancer: State of the art. Molecular Therapy - Nucleic Acids. 2022;28:758-791. doi:10.1016/j.omtn.2022.04.011

37.

Pardee

Green

Ferrante

, et al. Paper-based synthetic gene networks. Cell. 2014;159(4):940-954. doi:10.1016/j.cell.2014.10.004

38.

Baabu

PRS

Srinivasan

Nagarajan

, et al. End-to-end computational approach to the design of RNA biosensors for detecting miRNA biomarkers of cervical cancer. Synth Syst Biotechnol. 2022;7(2):802-814. doi:10.1016/j.synbio.2022.03.008

39.

Sanjay

Dou

, et al. Biomarker detection for disease diagnosis using cost-effective microfluidic platforms. Analyst. 2015;140(21):7062-7081.

40.

Soum

Park

Brilian

Kwon

Shin

. Programmable paper-based microfluidic devices for biomarker detections. Micromachines (Basel). 2019;10(8):516.

41.

Campbell

Balhoff

Landwehr

Rahman

Vaithiyanathan

Melvin

. Microfluidic and paper-based devices for disease detection and diagnostic research. Int J Mol Sci. 2018;19(9):2731.

42.

Cancer Genome Atlas Research Network, Weinstein

Collisson

, et al. The cancer genome atlas pan-cancer analysis project. Nat Genet. 2013;45(10):1113-1120.

43.

Edgar

Domrachev

Lash

. Gene expression omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 2002;30(1):207-210.

44.

Zhang

Baran

Cros

, et al. International cancer genome consortium data portal-a one-stop shop for cancer genomics data. Database. 2011;bar026. doi:10.1093/database/bar026

45.

Panwar

Omenn

Guan

. Mirmine: A database of human miRNA expression profiles. Bioinformatics. 2019;33(10):1554-1560.

46.

Kavakiotis

Alexiou

Tastsoglou

Vlachos

Hatzigeorgiou

. DIANA-miTED: A microRNA tissue expression database. Nucleic Acids Res. 2022;50(D1):D1055-D1061.

47.

Ludwig

Leidinger

Becker

, et al. Distribution of miRNA expression across human tissues. Nucleic Acids Res. 2016;44(8):3865-3877.

48.

Zhao

Zhang

Wang

, et al. Chinese Glioma genome atlas (CGGA): A comprehensive resource with functional genomic data from Chinese glioma patients. Genomics Proteomics Bioinformatics. 2021;19(1):1-2. doi:10.1016/j.gpb.2020.10.005

49.

Wang

Ling

, et al. dbDEMC 3.0: Functional exploration of differentially expressed miRNAs in cancers of human and model organisms. Genomics Proteomics Bioinformatics. 2022;20(3):446-454. doi:10.1016/j.gpb.2022.04.006

50.

Xie

Ding

Han

. miRCancer: A microRNA–cancer association database constructed by text mining on literature. Bioinformatics. 2013;29(5):638-644.

51.

Babu

Palaniappan

. Mir2trait: An integrated resource for investigating miRNA-disease associations. PeerJ. 2022;10:e14146. doi:10.7717/peerj.14146

52.

Wang

, et al. CancerMIRNome: An interactive analysis and visualization database for miRNome profiles of human cancer. Nucleic Acids Res. 2022;50(D1):D1139-D1146.

53.

Sarathi

Palaniappan

. Novel significant stage-specific differentially expressed genes in hepatocellular carcinoma. BMC Cancer. 2019;19(1):663.

54.

Muthamilselvan

Raghavendran

Palaniappan

. Stage-differentiated ensemble modeling of DNA methylation landscapes uncovers salient biomarkers and prognostic signatures in colorectal cancer progression. PloS one. 2022;17(2):e0249151. doi:10.1371/journal.pone.0249151

55.

Mitchell

. Machine learning. McGraw-Hill Education; 1997.

56.

. Applications of machine learning in cancer prediction and prognosis. Cancer Inform. 2007;11(2):59-77.

57.

Ritchie

Phipson

, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47.

58.

Kuhn

. Building predictive models in R using the caret package. J Stat Softw. 2008;28:1-26. doi:10.18637/jss.v028.i05

59.

Kursa

Rudnicki

. Feature selection with the boruta package. J Stat Softw. 2010;36(11):1-13. http://www.jstatsoft.org/v36/i11/

60.

Dimitriadou

Hornik

Leisch

Meyer

Weingessel

. Misc functions of the department of statistics (e1071), TU Wien. R Package. 2008;1:5-24.

61.

Chollet

. Deep learning with Python. 2nd ed. Simon and Schuster; 2021.

62.

Liaw

Wiener

. Classification and regression by randomForest. R News. 2002;2(3):18-22.

63.

Chen

Guestrin

. XGBoost: A Scalable Tree Boosting System. KDD ‘16: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016; 785–794. https://doi.org/10.1145/2939672.2939785

64.

Kuhn

. Caret: classification and regression training. Astrophysics Source Code Library. 2015;ascl-1505.

65.

O'Malley, Bursztein

Long

. 2019. KerasTuner [Online]. Available: https://github.com/keras-team/keras-tuner. [Accessed: 05-Feb-2022].

66.

Beam

Kohane

. Artificial intelligence in healthcare. Nat Biomed Eng. 2018;2:719-731. https://doi.org/10.1038/s41551-018-0305-z

67.

Bertsimas

Wiberg

. Machine learning in oncology: Methods, applications, and challenges. JCO Clin Cancer Inform. 2020;4:885-894. doi: 10.1200/CCI.20.00072

68.

Kourou

Exarchos

Papaloukas

Sakaloglou

Exarchos

Fotiadis

. Applied machine learning in cancer research: A systematic review for patient diagnosis, classification and prognosis. Comput Struct Biotechnol J. 2021;6(19):5546-5555. doi: 10.1016/j.csbj.2021.10.006

69.

Tran

Kondrashova

Bradley

, et al. Deep learning in cancer diagnosis, prognosis and treatment selection. Genome Med. 2021;13:152. doi:10.1186/s13073-021-00968-x

70.

Luchini

Pea

Scarpa

. Artificial intelligence in oncology: Current applications and future perspectives. Br J Cancer. 2022;126:4-9. doi:10.1038/s41416-021-01633-1

71.

Steyaert

Pizurica

Nagaraj

, et al. Multimodal data fusion for cancer biomarker discovery with deep learning. Nat Mach Intell. 2023;5:351-362. doi:10.1038/s42256-023-00633-5

72.

Gordon

Stern

. Challenges and opportunities in software-driven medical devices. Nat Biomed Eng. 2019;3(7):493-497.

73.

Alarcón-Zendejas

Scavuzzo

Jiménez-Ríos

, et al. The promising role of new molecular biomarkers in prostate cancer: From coding and non-coding genes to artificial intelligence approaches. Prostate Cancer Prostatic Dis. 2022;25(3):431-443. doi: 10.1038/s41391-022-00537-2

74.

Ling

Aldoghachi

Chong

, et al. Addressing the clinical feasibility of adopting circulating miRNA for breast cancer detection, monitoring and management with artificial intelligence and machine learning platforms. Int J Mol Sci. 2022 Dec 6;23(23):15382. doi: 10.3390/ijms232315382

75.

Sullivan

Zhu

Hicks

. Putting the “mi” in omics: Discovering miRNA biomarkers for pediatric precision care. Pediatr Res. 2023 Jan;93(2):316-323. doi: 10.1038/s41390-022-02206-5

76.

Steyerberg

Moons

van der Windt

, et al. Prognosis research strategy (PROGRESS) 3: Prognostic model research. PLoS Med. 2013;10(2):e1001381. doi: 10.1371/journal.pmed.1001381

77.

Patrick

Thomas

. Survival analysis and interpretation of time-to-event data: The tortoise and the hare. Anesth Analg. 2018;127(3):792-798. doi: 10.1213/ANE.0000000000003653

78.

Rai

Mishra

Ghoshal

. Survival analysis: A primer for the clinician scientists. Indian J Gastroenterol. 2021;40:541-549. doi:10.1007/s12664-021-01232-1

79.

Ohno-Machado

. Modeling medical prognosis: Survival analysis techniques. J Biomed Inform. 2001;34(6):428-439.

80.

Tripepi

Jager

Dekker

, et al. Statistical methods for the assessment of prognostic biomarkers (part II): Calibration and re-classification. Nephrol Dial Transplant. 2010;25(5):1402-1405.

81.

Tripepi

Jager

Dekker

, et al. Statistical methods for the assessment of prognostic biomarkers (part I): Discrimination. Nephrol Dial Transplant. 2010;25(5):1399-1401.

82.

Steyerberg

Mushkudiani

Perel

, et al. Predicting outcome after traumatic brain injury: Development and international validation of prognostic scores based on admission characteristics. PLoS. 2008;Med 5:e165. doi:10.1371/journal.pmed.0050165

83.

Hui

. Prognostication of survival in patients with advanced cancer: Predicting the unpredictable? Cancer Control. 2015;22(4):489-497. doi: 10.1177/107327481502200415

84.

Etikan

Bukirova

Yuvali

. Choosing statistical tests for survival analysis. Biom Biostat Int J. 2018;7(5):477-481.

85.

Portillo-Lara

Annabi

. Microengineered cancer-on-a-chip platforms to study the metastatic microenvironment. Lab Chip. 2016;16(21):4063-4081.

86.

Zhang

Chua

Khoo

. Worm-based microfluidic biosensor for real-time assessment of the metastatic status. Cancers (Basel). 2021;13(4):1-14.

87.

Cho

Choi

Kim

Shin

Choi

. Microfluidic system to analyze the effects of interleukin 6 on lymphatic breast cancer metastasis. Front Bioeng Biotechnol. 2021;8(February):1-9.

88.

Liu

Xie

Zhang

, et al. Biomimetic tumor-induced angiogenesis and anti-angiogenic therapy in a microfluidic model. RSC Adv. 2016;6(42):35248-35256.

89.

Chu

Gao

Tang

, et al. Attomolar-level ultrasensitive and multiplex microRNA detection enabled by a nanomaterial locally assembled microfluidic biochip for cancer diagnosis. Anal Chem. 2021;93(12):5129-5136.

90.

Lee

Doyle

. Hydrogel-based colorimetric assay for multiplexed microRNA detection in a microfluidic device. Anal Chem. 2020 Mar 24;92(8):5750-5755.

91.

Zou

Liu

Xia

Yin

. A multiplex and fast detection platform for microRNAs based on a self-priming microfluidic chip and duplex-specific nuclease. Analyst. 2021;146(2):628-635.

92.

Gao

Qiang

Chu

Han

Zhang

Han

. Microfluidic chip for multiple detection of miRNA biomarkers in breast cancer based on three-segment hybridization. AIP Adv. 2020;10(4):045022.

93.

Salim

Athira

Kandaswamy

Vijayakumar

Saravanan

Sairam

. Microfluidic device for novel breast cancer screening by blood test using miRNA beacon probe. Biomed Microdevices. 2017;19(4):1-1.

94.

Sung

Huang

Chen

Lee

. Extraction and quantification of microrna biomarkers for diagnosis of ovarian cancer on an integrated microfluidic platform. 2021 IEEE 34th International Conference on Micro Electro Mechanical Systems (MEMS); 2021;366-369. doi:10.1109/MEMS51782.2021.9375425

95.

Kim

Hosokawa

Maeda

. Sensitivity enhancement of MicroRNA detection using a power-free microfluidic chip. Analytical Science. 2019;35(11):1227-1236.

96.

Wei

Zhang

Yang

. Multiplexed detection of micro-RNAs based on microfluidic multi-color fluorescence droplets. Anal Bioanal Chem. 2020;412(3):647-655.

97.

Cai

Zhang

Wang

. A microfluidic paper-based laser-induced fluorescence sensor based on duplex-specific nuclease amplification for selective and sensitive detection of miRNAs in cancer cells. Talanta. 2020;216:120996. doi:10.1016/j.talanta.2020.120996

98.

Zhai

Pan

, et al. In situ detection of plasma exosomal microRNA-1246 for breast cancer diagnostics by a au nanoflare probe. ACS Appl Mater Interfaces. 2018;10(46):39478-39486.

99.

Liu

Kannisto

, et al. Non-invasive detection of exosomal MicroRNAs via tethered cationic lipoplex nanoparticles (tCLN) biochip for lung cancer early detection. Front Genet. 2020;11:258.

100.

Wang

Zhang

Liu

Wang

. Triggerable mutually amplified signal probe based SERS-microfluidics platform for the efficient enrichment and quantitative detection of miRNA. Anal Chem. 2019;91(8):5043-5050.

101.

Taller

Richards

Slouka

, et al. On-chip surface acoustic wave lysis and ion-exchange nanomembrane detection of exosomal RNA for pancreatic cancer study and diagnosis. Lab Chip. 2015;15(7):1656-1666.

102.

Bruch

Baaske

Chatelle

, et al. CRISPR/Cas13a-powered electrochemical microfluidic biosensor for nucleic acid amplification-free miRNA diagnostics. Adv Mater. 2019;31(51):1905311.

103.

Horny

Dupuis

Siaugue

Gamby

. Release and detection of microrna by combining magnetic hyperthermia and electrochemistry modules on a microfluidic chip. Sensors. 2021;21(1):1-18.

104.

Sun

Wang

Jian

, et al. Ultrasensitive microfluidic paper-based electrochemical/visual biosensor based on spherical-like cerium dioxide catalyst for miR-21 detection. Biosens Bioelectron. 2018;105:218-225. doi:10.1016/j.bios.2018.01.025

105.

Jiao

Zhao

Tang

Shi

Wang

. Droplet-based PCR in a 3D-printed microfluidic chip for miRNA-21 detection. Anal Methods. 2019;11(26):3386-3393.

106.

Moltzahn

Olshen

Baehner

, et al. Microfluidic-based multiplex qRT-PCR identifies diagnostic and prognostic microRNA signatures in the Sera of prostate cancer patients serum miRNAs in prostate cancer. Cancer Res. 2011;71(2):550-560.

107.

Kim

Park

Lee

, et al. Multi-miRNA panel of tumor-derived extracellular vesicles as promising diagnostic biomarkers of early-stage breast cancer. Cancer Sci. 2021;112(12):5078-5087.

108.

Senapati

Slouka

Shah

, et al. An ion-exchange nanomembrane sensor for detection of nucleic acids using a surface charge inversion phenomenon. Biosens Bioelectron. 2014 Oct 15;60:92-100.

109.

Zhao

Liu

Chu

Chen

. Paper-based laser induced fluorescence immunodevice combining with CdTe embedded silica nanoparticles signal enhancement strategy. Sensors Actuators, B Chem. 2017;242:87-94. doi:10.1016/j.snb.2016.11.004

110.

Xiao

Chandrasekaran

, et al. Affinity-modulated molecular beacons on MoS2 nanosheets for MicroRNA detection. ACS Appl Mater Interfaces. 2018;10(42):35794-35800.

111.

Zhou

Kan

, et al. Highly sensitive and selective strategy for microRNA detection based on WS2 nanosheet mediated fluorescence quenching and duplex-specific nuclease signal amplification. Anal Chem. 2014;86(3):1361-1365.

112.

Tabeling

. Recent progress in the physics of microfluidics and related biotechnological applications. Curr Opin Biotechnol. 2014;25:129-134. doi:10.1016/j.copbio.2013.11.009

113.

Bruch

Johnston

Kling

, et al. CRISPR-powered electrochemical microfluidic multiplexed biosensor for target amplification-free miRNA diagnostics. Biosens Bioelectron. 2021;177:112887.

114.

Oliveira

Dalot

Fortunato

, et al. Microfluidic SERS devices: Brightening the future of bioanalysis. Discov Mater. 2022;2:12. doi:10.1007/s43939-022-00033-3

115.

Wang

Lin

Cao

Liang

Zhang

. Surface plasmon resonance sensors on Raman and fluorescence spectroscopy. Sensors (Basel). 2017;17(12):2719. doi: 10.3390/s17122719

116.

Jin

Geißler

Qiu

Wegner

Hildebrandt

. A rapid, amplification-free, and sensitive diagnostic assay for single-step multiplexed fluorescence detection of MicroRNA. Angew Chem Int Ed Engl. 2015;54(34):10024-9. doi: 10.1002/anie.201504887

117.

Tian

Shu

Jiang

, et al. An ultralocalized Cas13a assay enables universal and nucleic acid amplification-free single-molecule RNA diagnostics. ACS Nano. 2021;15(1):1167-1178.

118.

Zhang

Tian

Liu

, et al. Hand-powered centrifugal microfluidic platform inspired by the spinning top for sample-to-answer diagnostics of nucleic acids. Lab Chip. 2018;18(4):610-619.

119.

Klein

Richards

Cohn

, et al. Clinical validation of a targeted methylation-based multi-cancer early detection test using an independent validation set. Ann Oncol. 2021;32(9):1167-1177.

120.

Nahavandi

Baratchi

Soffe

, et al. Microfluidic platforms for biomarker analysis. Lab Chip. 2014;14(9):1496-1514.

Microfluidics for Profiling miRNA Biomarker Panels in AI-Assisted Cancer Diagnosis and Prognosis

Abstract

Keywords

Introduction

AI Assisted miRNA Biomarker Panel Discovery

miRNA Biomarker Discovery

Classes of miRNA Biomarkers

Construction of ML Model

Construction of Prognostic Model

Microfluidic Systems for Profiling miRNA Signatures

Footnotes

Abbreviations

Acknowledgments

Declaration of Conflicting Interests

Funding

ORCID iD

References