Hybridization Protection Reaction for Sensitive and Robust Gene Expression Profiling of Clinical Formalin-Fixed Paraffin-Embedded Samples

Abstract

Background

RNA profiling of formalin-fixed paraffin-embedded (FFPE) tumor tissues for the molecular diagnostics of disease prognosis or treatment response is often irreproducible and limited to a handful of biomarkers. This has led to an unmet need for robust multiplexed assays that can profile several RNA biomarkers of interest using a limited amount of specimen. Here, we describe hybridization protection reaction (HPR), which is a novel RNA profiling approach with high reproducibility.

Methods

HPR assays were designed for multiple genes, including 10 radiosensitivity-associated genes, and compared with TaqMan assays. Performance was tested with synthetic RNA fragments, and the ability to analyze RNA was investigated in FPPE samples from 20 normal lung tissues, 40 lung cancer, and 30 esophageal cancer biopsies.

Results

Experiments performed on 3 synthetic RNA fragments demonstrated a linear dynamic range of over 1000-fold with a replicate correlation coefficient of 0.99 and high analytical sensitivity between 3.2 to 10 000 pM. Comparison of HPR with standard quantitative reverse transcription polymerase chain reaction on FFPE specimens shows nonsignificant differences with > 99% confidence interval between 2 assays in transcript profiling of 91.7% of test transcripts. In addition, HPR was effectively applied to quantify transcript levels of 10 radiosensitivity-associated genes.

Conclusions

Overall, HPR is an alternative approach for RNA profiling with high sensitivity, reproducibility, robustness, and capability for molecular diagnostics in FFPE tumor biopsy specimens of lung and esophageal cancer.

Introduction

A limited amount of tissue specimen is a significant barrier for translational medicine, particularly in clinical research based on RNA profiling. RNA profiling involves the comprehensive analysis of the expression levels of RNA molecules of interest. Its strengths in sensitivity, reproducibility, quantitation, dynamic range, and multianalyte capability position this technique as a promising molecular diagnostic tool. Tissue specimens are usually archived in formalin-fixed paraffin-embedded (FFPE) blocks, in which the RNA has relatively low integrity (1, 2). Current methods of RNA profiling have inherent limitations for molecular diagnostics using FFPE specimens (3, 4). Therefore, numerous upfront optimizations are necessary for reverse-transcription quantitative polymerase chain reaction (RT-qPCR) assays before being introduced to clinical applications. The challenges of FFPE specimens and assay validation/optimization restrict the RT-qPCR assay to the investigation of limited biomarkers simultaneously (5–7). While next-generation sequencing–based RNA profiling can analyze thousands of transcripts simultaneously, it is more costly and time-consuming (8, 9). Moreover, the FFPE tissue-preserving process results in RNA degradation and modification, which leads to low reproducibility using the current methods (10, 11).

With the advances in genomics and data analytics, precision medicine represents an innovative approach to medicine in tailoring prevention, diagnosis, and treatment based on an individual's unique genetic factors. It aims to optimize therapeutic effectiveness and minimize adverse reactions. Despite its promise, challenges like cost and data accuracy for molecular diagnostics remain. A novel approach that is capable of a multiplexed analysis of RNA biomarkers from a limited amount of low-integrity specimen RNA and achieves high sensitivity and reproducibility will be valuable for translational research in this era of precision medicine.

In the present research, we developed a technique, designated hybridization protection reaction (HPR), that combines the advantages of the nuclease protection assay with real-time RT-qPCR to enable multiplexed quantitative profiling of RNA biomarkers from FFPE samples with consistent quality (12, 13). This probe design enables the quantification of target transcripts without extensive upfront optimization that must be performed prior to RT-qPCR. We evaluated the technical performance of HPR by performing a series of experiments to assess a range of technical variables, including specificity, linearity, limit of detection, and reproducibility, with synthetic RNA molecules and commercial FFPE samples. In addition, we investigated whether HPR can be used for practical clinical applications. Clinical specimens from a prospective cohort of non-small cell lung carcinoma patients receiving definitive high-dose radiotherapy and esophageal squamous cell carcinoma patients undergoing neoadjuvant concurrent chemoradiotherapy were used to test the performance of the HPR technique for analyzing radiosensitivity-associated RNA biomarkers (14–16). HPR has the potential to be an alternative approach for RNA profiling, as it has high sensitivity, reproducibility, robustness, and capability for molecular diagnostics. In addition to RNA profiling from FFPE tissue, HPR can be used for other nucleic acid biomarkers from any sample type.

Materials and Methods

Experimental Design

A series of experiments were conducted to test the technical performance and clinical applicability of HPR. The synthetic RNA oligonucleotides containing 40 to 50-mer sequences from ACTB, IRF1, and HDAC1 transcripts were adopted as standards for evaluating the specificity, linearity, limit of detection, and reproducibility of in-house designed HPR assays. Additional experiments were performed using 20 commercially available lung FFPE tissue samples (purchased from Biochain) and clinical specimens, including 40 lung cancer and 30 esophageal cancer patients, enrolled from a prospective cohort to further demonstrate the reproducibility and clinical utility of HPR as a technical platform for molecular diagnostics based on RNA biomarkers.

HPR

For standard HPR, hybridization was carried out in a reaction mixture containing single or multiple probes (0.5–10 pM each), synthetic RNA fragments (Integrated DNA Technologies), and yeast transfer RNA (Sigma-Aldrich) as carriers. The mixture was first heated to 65°C for denaturation, after which 40 mM sodium acetate (pH 4.5), 300 mM sodium chloride, and 2 mM zinc sulfate were added. The mixture was further incubated at 55°C for 0.5 to 3 hours. S1 nuclease (Thermo Fisher Scientific) was then added to digest single-stranded nucleic acids by incubating the mixture at 50°C for 30 mins. The subsequent digestion of RNase A was performed after the addition of EDTA to the mixture. After dual nuclease digestion, the reaction mixture was diluted and subjected to preamplification using Q5 DNA polymerase (New England Biolabs) with universal primers. Eventually, the preamplification product would be used as the template for the TaqMan assay using a universal primer and a TaqMan probe specific to the target barcode.

Specificity, Linearity, Limit of Detection, and Reproducibility

Specificity was evaluated by using synthetic RNA fragments designed from target or homologous sequences as positive and nonspecific controls, which were introduced separately to the HPR at a high concentration of 1000 pM. The results were calculated as a percentage of undigested probes (100%) and digested probes (0%). The limit of detection, linearity, and reproducibility were assessed by mixing the target-specific probe with various amounts (10 000 to 0.1 pM) of the synthetic oligonucleotide as positive control and measuring the protected probe using a TaqMan assay specific for the assigned barcode.

RNA Isolation and RT-qPCR

Five to 10 mm tissue thickness of FFPE slides were subjected to RNA isolation using Quick-RNA FFPE Miniprep kit (Zymo Research) according to the manufacturer's protocol. Purified RNA was first treated with DNase I to remove DNA contamination. Then, complementary DNA synthesis was carried out using SuperScript IV reverse transcription kit (Thermo Fisher Scientific) with random hexamers, followed by TaqMan gene expression assays in 96-well plates, performed with the QuantStudio 3 real-time qPCR instrument (Thermo Fisher Scientific). The expression level of each gene was analyzed in triplicate and normalized relative to the GAPDH gene.

Comparison Between HPR And RT-qPCR

Forty-eight target genes (Supplemental Table 1) were evaluated for each FFPE. Hybridization reactions were performed in triplicate, with total RNA samples isolated from 20 FFPE lung tissue slides. The same samples were also analyzed with RT-qPCR using TaqMan gene expression assays. The sensitivities of HPR and RT-qPCR were determined by examining the number of gene transcripts detected in each method. The concordance of HPR and RT-qPCR was further demonstrated by 12 target genes, which expressed with a mean threshold cycle (Ct) of 25 to 40 in RT-qPCR for assessing the differences between HPR and RT-qPCR. For both methods, the expression levels were normalized to GAPDH as a control gene. The level of degradation of RNA samples from 20 commercially available FFPEs was obtained by RNA integrity number analysis using Agilent 2100 bioanalyzer. The reproducibility of HPR on high degradation RNA was determined by calculating the coefficient of variation based on the mean value of triplicates.

Clinical Subject Enrollment and Specimen Collection

From 2016 to 2020, 86 patients who were ages ≥ 20 years old and who had pathological or cytological proof of non-small cell lung carcinoma receiving curative or radical high-dose radiotherapy (biological equivalent dose ≥ 66 Gy₁₀) or esophageal squamous cell carcinoma undergoing neoadjuvant concurrent chemoradiotherapy were enrolled. Of these, 70 had preserved pathology specimens available and were included in the translational study. All participants signed the informed consent for an institutional research ethics committee-approved protocol.

Radiosensitivity-Associated Biomarkers Analysis

A previously published 10-gene expression assay (15) was conducted using clinical specimens. We designed a multiplex HPR assay for assessing the expression levels of 10 radiosensitivity-associated transcripts in routine lung cancer and esophageal cancer FFPE samples obtained from needle or forceps biopsy. Total RNA was isolated from 40 lung needle biopsy and 30 esophageal mucosal forceps biopsy FFPE samples and introduced to the HPR hybridization mixture. Each reaction contained the probe mixture for 10 transcripts and extra carrier RNA to enhance the overall hybridization efficiency. The results of gene expression levels from the HPR were applied to calculate the radiosensitivity index. Specifically, the expression levels of 10 messenger RNAs (AR, JUN, STAT1, PRKCB, RELA, ABL, SUMO1, PAK2, HDAC1, and IRF1) were ranked from 10 to 1 according to the quantification results and then subjected to the calculation algorithm as reported by Scott et al. (15).

Statistical Analysis

The linear correlation was used to determine the performance of the detection range of each assay. The Student t-test was used to determine the number of assays detected by HRP and RT-qPCR methods and the detection limit of each assay. The ANOVA with correction for multiple comparisons by the Bonferroni test was used to compare the concordance and performance between the HRP and RT-qPCR methods. The GraphPad Prism version 9 (Graphpad Software Inc.) was used for the statistical analysis.

Results

The Design of the HPR

To quantitatively assay HPR, a unique DNA probe was designed for each gene of interest. Each probe comprised 4 independent regions, including the target transcript hybridization region, target-specific assigned barcode, 5′ and 3′ universal primer sites, and upstream and downstream flanking regions. The target transcript hybridization region contained 30 to 40 base sequences and was designed to be completely complementary with the transcript of the target gene to prevent mismatch cleavage by single-stranded specific nucleases and coupled to an assigned target-specific barcode that provided the binding site for the TaqMan probe. The target-specific assigned barcode was a 25-base sequence, designed with a low nonspecific hybridization potential from a set of 240 000 orthogonal DNA barcode probes (17). 5′ and 3′ universal primer sites were common sequences among all probes for enrichment and amplification in the RT-qPCR step. The 5′ and 3′ flanking regions were designed to complement universal primer sites and target-specific assigned barcodes. The formation of the 5′ and 3′ self-complementary hairpin structures protects against undesired cleavage by single-stranded specific nucleases (Fig. 1A).

During the HPR process, the probes for the genes of interest were pooled into a library and used as a single reagent for hybridization. The hybridization was then accomplished by combining the RNA sample with the probe library, leading to the formation of DNA-RNA hybrid complexes. Unhybridized probes and nontarget RNA were removed in a 2-step nuclease treatment using S1 nuclease and RNase A. This dual digestion allowed the protected HPR probes to be intact and serve as ideal DNA templates for further preamplification without any interference from a large excess of probes or sample RNA (Fig. 1B). After digestion, the solution was be diluted and introduced to a preamplification step using universal primers to generate sufficient copies of protected probes for subsequent quantification (18, 19). The expression levels of targets were measured by quantification of amplified HPR probes using universal primers and TaqMan probes specific for target barcodes (Fig. 1C).

Technical Performance of the HPR

The limits of detection, linearity, and reproducibility of the HPR assays were assessed using synthetic RNA fragments for ACTB, IRF1, and HDAC1 as positive controls. The Ct values of positive controls at different concentrations (10 000 to 0.1 pM) after the HPR were measured with respect to distilled water as blanks, with repeats in duplicates. The overlapping points in Fig. 2 indicate that the Ct values of each replicate were highly reproducible at all concentrations. Linear correlations within the detection range between 3.2 and 10 000 pM across all positive controls were extremely high (correlation coefficient, R² ≥ 0.99, Fig. 2). The limit of detection of each assay was determined by comparing mean Ct values between 3 positive replicates at the lowest concentration to the blank using the Student t-test. The limits of detection of the control assays ranged from 2.5 to 6 pM in a hybridization solution containing 500 ng of yeast transfer RNA. Typically, each mammalian cell contains about 20 pg total RNA. Thus, the data suggest that the detection limit of the HPR technique could be as low as 1200 to 2880 molecules per cell.

The specificity of HPR was assessed by performing assays for ACTB, IRF1, and HDAC1 using synthetic RNA fragments at concentrations of 1000 pM. Each assay was evaluated using one positive control designed from the target sequence and 2 nonspecific controls designed from homologous transcripts with a sequence identity of 90% to 92%. By using an undigested probe as the quantification standard, positive RNA fragments showed low Ct values upon application of the HPR technique, providing more than 80% of protection (Fig. 3). In contrast, nonspecific controls generated low protection (2%–10%). Notably, an increase in protection to above 10% was observed when using mismatch RNA as a nonspecific control for the HDAC1 assay (Fig. 3). We noticed that the HDAC1 assay was designed with a relatively high G/C content region, which might have led to decreased accessibility of S1 nuclease to the mismatch site. S1 nuclease has been reported to show weaker activity toward extremely G/C-rich mismatch regions, and the optimization of buffer compositions may improve the enzyme activity (20). Overall, these results demonstrate that HPR achieved high specificity in RNA detection.

Comparison Between HPR And RT-qPCR

One of the advantages of HPR is that this approach depends on the target RNA hybridization in a specific region containing 30 to 40 bases, thus avoiding analysis bias from degraded RNA. This implies that HPR may be an ideal approach for the evaluation of low-integrity RNA samples extracted from FFPE tissue, which is examined for clinicopathological diagnosis.

The performance of HPR was compared to standard RT-qPCR in the quantitative analysis of endogenous transcript expression. Of the 48 expression assays, there were 3 genes that could not be analyzed by either HPR or RT-qPCR. The remaining 45 genes were used to determine how many genes could be detected by each method. With HPR, nearly all transcripts could be analyzed in 20 FFPE samples. In contrast, the number of detected transcripts by RT-qPCR was fewer for genes with low expression levels (defined as mean Ct ≥ 37 by RT-qPCR) in FFPE samples (RT-qPCR, 17/20 vs HRP, 20/20).

The concordance of HPR and RT-qPCR was further demonstrated by the selected 12 genes. As shown in Fig. 4, the expression pattern for each transcript was similar between HPR and RT-qPCR, suggesting that HPR can provide comparable information to data derived with RT-qPCR. A 2-way ANOVA test was executed to evaluate the quantitative differences between HPR and RT-qPCR, and it revealed no significant difference between the 2 methods in nearly all 12 test genes (Supplemental Table 2, Supplemental Data 2). The only exception was for SOBP expression. The HPR assay for SOBP messenger RNA seemed to be less sensitive than its RT-qPCR counterpart, which led to a significant difference between HPR and RT-qPCR (P = 0.032). This finding might be attributed to the potential formation of a stem-loop structure in the SOBP-specific HPR probe, which is likely to impede hybridization. Overall, these results confirm that HPR can provide precise quantification data on FFPE samples to enable RNA profiling in clinical use.

We further investigated the reproducibility of HPR for highly degraded low-integrity RNA. The mean, median, and range of the RNA integrity number values were 2.1, 2.3, and 1.1 to 2.5, respectively, while the average percentage of RNA fragments with > 300 bases was 56%. The performance of HRP and RT-qPCR was evaluated by assessing the Ct and CV for commonly used housekeeping genes including ACTB, GAPDH, B2M, and GUSB. The results are shown in Fig. 5 and Supplemental Table 3 and show that the housekeeping genes require significantly fewer Ct numbers (P < 0.001) and have smaller CV values (P < 0.01) with the HRP technique. Additionally, for all of the 45 tested genes, the mean value of CV assayed by the HRP and RT-qPCR techniques were 2.3 and 10.1, respectively. These findings demonstrated that the results of the HRP assay were highly reproducible in low-integrity RNA samples. Overall, compared to the RT-qPCR assay, the HPR technique had the ability to profile targeted transcripts from highly degraded FFPE samples with superior reproducibility.

Clinical Performance of HPR In Needle/Forceps Biopsies

The potential of HPR in clinical practice was further evaluated. The clinical samples obtained from routine needle or forceps biopsy have been described as challenging samples for molecular diagnostics, since the majority of the tissue has already been exhausted for standard pathological analysis (e.g., immunohistochemistry stains and oncogenic driver mutation analysis). Using HPR, all 10 transcripts could be detected in all test samples within 45 cycles (Fig. 6), suggesting that HPR could be utilized for the analysis of gene expression profiles in lung needle biopsy and esophageal mucosal forceps biopsy FFPE specimens.

Discussion

In this study, we reported a novel approach for the quantification of multiple RNA transcripts from low-integrity RNA isolated from a limited amount of sample. This approach combined hybridization and the TaqMan assay system. The developed system was highly sensitive, reproducible, and robust for characterizing specific transcripts; and it was easy to use. Our head-to-head comparison using the same test set demonstrated that HPR is capable of quantifying genes of interest with low expression levels at greater reproducibility than conventional standard RT-qPCR. In addition, our results indicate that HPR can be applied to characterize radiosensitivity-associated transcripts from an extremely limited amount of tissue in biopsy samples from non-small cell lung carcinoma and esophageal squamous cell carcinoma cancer patients (14–16).

We demonstrated several advantages of HPR over current gene expression methods, such as RT-qPCR. First, the target transcript was measured through specific hybridization and nuclease protection. The presence of low-integrity RNA target transcripts did not affect the analysis by HRP, and the expression level of each gene of interest within a total RNA sample could be easily measured by quantifying the protected probes and calculating concentrations with standard curves generated from stepwise diluted reference transcript (21–23). In the highly degraded test FFPE samples, HRP showed significantly fewer Ct numbers and smaller CV values in both housekeeping genes and the 12 selected genes. In contrast, RT-qPCR requires extensive upfront optimization for each assay before being introduced to the assay for analysis of low-integrity RNA samples (24). Second, all target-specific probes and RNA samples are combined and hybridized in a single solution, rather than being bound to the surface of a solid material. This kinetic advantage allows for greater sensitivity in detecting transcripts with low expression levels (25, 26). Indeed, in the 20 tested genes with low expression levels assayed by RT-qPCR, HRP successfully assayed all transcripts, but RT-qPCR only worked for 85% of FFPE samples. Third, HPR provides a convenient way for multiplex preamplification to increase overall detection sensitivity using universal primers. In other existing methods, preamplification of complimentary DNA is carried out with multiplex primer sets for multiple transcripts of interest. The process of multiplex primer design is a critical issue requiring complex considerations, such as the formation of secondary structures, cross-reactivity between primers, and generation of nontarget amplicons (27).

Another advantage of HPR is that it can be easily modified for various applications and analysis. For example, HPR can be applied to the analysis of micro RNA (miRNA) signatures, as this technique is able to discriminate among highly homologous miRNAs. In contrast, the current approach for miRNA analysis is based on reverse transcription using stem-looped primers to generate amenable templates for the TaqMan assay (28, 29). This use of stem-loop RT primers may be tricky and poorly reproducible when profiling highly homologous miRNAs. Another potential application of the HPR technology is measuring biomarkers of oncogenic driver mutations from cell-free DNA. Studies have shown the existence of tumor-derived small fragment DNA in the peripheral blood of cancer patients, which suggests the potential of liquid biopsy for early detection at a curable stage, surveillance of disease progression, and prediction of drug responses (30–32). Since single-stranded specific nuclease can cleavage single base pair mismatch in heteroduplex DNA, HPR holds great potential for multiplex analysis of oncogenic driver mutations in the presence of high background levels of wild-type DNA.

In HRP, the target-specific probe is an antisense probe and a template for the subsequent analysis. After nuclease digestion, the amount of remaining probe is measured by using the TaqMan probe complemented with an assigned target barcode. This approach shows high specificity and sensitivity and a limit of detection of 1200 molecules per cell with a high linear correlation coefficient of the dynamic range (R² > 0.99). Furthermore, all target-specific probes can be pooled as a single reagent to detect multiple targets in a single hybridization to achieve a multiplexed assay without extensive upfront optimization. Compared to conventional RT-qPCR techniques, HPR is more suitable for the detection of low-integrity RNA with low levels of expression in clinical samples with scarce tissue. It is in the best interest of patients and clinicians to profile multiple RNA biomarkers in various diseases to improve disease diagnosis (e.g., predicting therapeutic response) and prognostication. Of note, patients who have received definitive radiotherapy usually only have tissue samples from biopsy specimens, which contain significantly fewer tumor cells than surgically excised specimens. Furthermore, advanced histology subtyping and molecular characterization also consume significant amounts of tissue in current routine clinical practice. These factors limit the available tissue sample for research. Notably, a recent peer-reviewed study on the prediction of radiotherapy benefits using the radiosensitivity index suggested a new paradigm for an individualized radiation dose (16). The tool provides a cost-effective solution to characterize the radiosensitivity index using clinical samples, which is timely in the era of personalized radiation oncology (33). Of note, our preliminary analysis revealed the radiosensitivity index assayed by HRP was associated with the clinical outcomes for lung and esophageal cancer patients receiving radiotherapy (data not shown). Thus, HPR is more useful for radiosensitivity tests from clinical samples.

The analysis of gene expression profiles in low-quality RNA presents a significant challenge in molecular diagnostics. Our study enrolled patients with lung cancer undergoing definitive radiotherapy and esophageal cancer undergoing neoadjuvant chemoradiotherapy. Patients undergoing definitive or neoadjuvant radiotherapy usually had limited tissue specimens available for molecular profiling, as often only small biopsy samples are taken rather than larger surgical samples. If only a minimal amount of specimen is available, it can be challenging to obtain a comprehensive molecular profile. This limitation can hinder the quality of RNA profiling, impacting the implantation of precision medicine. Various technologies have demonstrated their capacity as holistic analysis platforms for FFPE samples (34). A notable example is TempO-Seq, which utilizes templated oligonucleotides to hybridize with target RNA and generate amplifiable templates for next-generation sequencing analysis (35). It represents a dependable high-throughput solution for examining expression levels derived from low-quality RNA. A comparison of common technologies in assaying RNA profiles is summarized in SupplementalTable 4. Among these methodologies, HPR exhibits the ability to analyze low-integrity RNA samples with a broad dynamic range. It may rival closely traditional qRT-PCR in terms of sample throughput and the simultaneous assay of multiple targets. However, when it comes to the screening of potential biomarkers for the development of specialized panels in molecular diagnostics, HPR falls short in its competitiveness compared to other novel technologies. On the other hand, HRP may provide a platform for assaying specific expression panels aimed at molecular diagnostics with its strengths lying in standardization, cost-effectiveness, accuracy, and reproducibility, particularly for low-quality FFPE samples.

Conclusion

This study proposed a novel approach that combined the advantages of nuclease protection and real-time RT-qPCR to analyze clinical specimens of limited tissue with low-integrity RNA. This robust approach measured the target transcripts using a designed DNA probe and barcode-specific real-time RT-qPCR and had high sensitivity and reproducibility. This novel technique may empower the discovery of more RNA biomarkers for disease diagnosis and management, as well as the development of practical molecular diagnostics.

Supplemental Material

Supplemental material is available at Clinical Chemistry online.

Nonstandard Abbreviations

FFPE, formalin-fixed paraffin-embedded; RT-qPCR, reverse-transcription quantitative polymerase chain reaction; HPR, hybridization protection reaction; Ct, threshold cycle; miRNA, micro RNA.

Human Genes

ACTB, actin beta; B2M, beta-2-microglobulin; HDAC1, histone deacetylase 1; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GUSB, glucuronidase beta; IRF1, interferon regulatory factor 1; SOBP, sine oculis binding protein homolog.

Author Contributions

The corresponding author takes full responsibility that all authors on this publication have met the following required criteria of eligibility for authorship: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. Nobody who qualifies for authorship has been omitted from the list.

Feng-Ming Hsu (Investigation-Equal, Project administration-Equal, Writing—original draft-Lead), Yih-Leong Chang (Formal analysis-Equal, Resources-Lead), Chung-Yung Chen (Formal analysis-Equal, Methodology-Equal), Shu-Rung Lin (Data curation-Equal, Investigation-Equal, Methodology-Equal), and Jason Chia-Hsien Cheng (Conceptualization-Equal, Funding acquisition-Lead, Writing—review & editing-Lead).

Authors’ Disclosures or Potential Conflicts of Interest

Upon manuscript submission, all authors completed the author disclosure form.

Research Funding

This work was supported by the Ministry of Science and Technology, Taiwan, R.O.C. [grant number: MOST-109-2622-E-003-015-CC1; MOST-110-2622-E-002-012-CC1; MOST-111-2622-E-002-027-CC1; MOST-110-2622-E-002-020; MOST-111-2622-E-002-039]; the Ministry of Health and Welfare, Taiwan, R.O.C. [grant number: MOHW111-TDU-B-221-114006]; the Medein Science Corporation, Taipei, Taiwan; and the BRAXX Biotech Corporation, Taipei, Taiwan.

Disclosures

None declared.

Role of Sponsor

The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, preparation of manuscript, or final approval of manuscript.

Acknowledgments

We thank the staff of the National Taiwan University College of Medicine, the Eighth Core Lab, Department of Medical Research, and National Taiwan University Hospital for technical assistance during this work. We thank Uni-edit (www.uni-edit.net) for editing and proofreading this manuscript.

References