Combinatorial Blood Platelets-Derived circRNA and mRNA Signature for Early-Stage Lung Cancer Detection

Abstract

Despite the diversity of liquid biopsy transcriptomic repertoire, numerous studies often exploit only a single RNA type signature for diagnostic biomarker potential. This frequently results in insufficient sensitivity and specificity necessary to reach diagnostic utility. Combinatorial biomarker approaches may offer a more reliable diagnosis. Here, we investigated the synergistic contributions of circRNA and mRNA signatures derived from blood platelets as biomarkers for lung cancer detection. We developed a comprehensive bioinformatics pipeline permitting an analysis of platelet-circRNA and mRNA derived from non-cancer individuals and lung cancer patients. An optimal selected signature is then used to generate the predictive classification model using machine learning algorithm. Using an individual signature of 21 circRNA and 28 mRNA, the predictive models reached an area under the curve (AUC) of 0.88 and 0.81, respectively. Importantly, combinatorial analysis including both types of RNAs resulted in an 8-target signature (6 mRNA and 2 circRNA), enhancing the differentiation of lung cancer from controls (AUC of 0.92). Additionally, we identified five biomarkers potentially specific for early-stage detection of lung cancer. Our proof-of-concept study presents the first multi-analyte-based approach for the analysis of platelets-derived biomarkers, providing a potential combinatorial diagnostic signature for lung cancer detection.

Keywords: biomarkers; cancer diagnosis; circular RNA; liquid biopsy; lung cancer; messenger RNA; platelets.

Figure 1

Titration experiment using total RNA derived from blood platelets of cancer patient and asymptomatic individual. (a) Six different inputs (1 ng, 3 ng, 6 ng, 12 ng, 24 ng and 48 ng) of total RNA derived from platelets of a non-cancer individual (Control) were tested using human immunology v2 panel with nCounter platform. Total number of counts detected (after negative background removal) follow a linear regression model (R² = 0.99, p-value < 0.0001). (b) A similar experiment was performed using total RNA derived from lung cancer platelet sample, and in this case, the total number of counts after background removal follow a linear regression model (R² = 0.99, p-value < 0.0001). (c) Principal Component Analysis (PCA) assessing RNA profile of Control sample using six different initial inputs of total RNA. (d) PCA assessing RNA profile of Cancer sample using six different initial inputs of total RNA.

Figure 2

Wet and dry lab workflow for the study of mRNA and circRNA derived from blood platelets using nCounter technology. Total RNA is extracted from human platelets and directly hybridized with nCounter probes. Hybridization and purification are performed on the nCounter prep-station and counting of the hybridized barcode is performed on the nCounter digital analyzer. Bioinformatic workflow consists of quality control (QC) and data filtering, normalization of counts, differential expression (DE) analysis, and the generation of a prediction model through the use of ML algorithms.

Figure 3

Platelet-mRNA and circRNA detection using human immunology v2 panel and 78-circRNA panel. (a) Venn diagram showing mRNAs identified in lung cancer and control samples using human immunology v2 panel. (b) Venn diagram showing circRNAs identified in lung cancer and control samples using 78-circRNA custom panel. (c) Number of transcripts detected in blood platelets derived from cancer patients and non-cancer controls using human immunology v2 panel (Mann–Whitney U test, p-value > 0.05). (d) Number of circRNAs detected in blood platelets derived from cancer patients and non-cancer controls using 78-circRNA custom panel (Mann–Whitney U test, p-value > 0.05).

Figure 4

Normalization and differential expression analysis of mRNA and circRNA. (a) RLE plot of the normalized mRNA data generated using DESeq2. (b) Volcano plot of differentially expressed mRNAs. The negative log of the adjusted p-value (base 10) is plotted on the Y-axis, and the log of the FC (base 2) is plotted on the X-axis. Red dots indicate significantly downregulated mRNA and green dots represent significantly upregulated mRNA (adjusted p-value < 0.05). (c) RLE plot of the normalized circRNA data generated using DESeq2. (d) Volcano plot of differentially expressed circRNAs. Green dot represents the significantly upregulated circRNA (adjusted p-value < 0.05).

Figure 5

ML analysis using single signature of mRNA and circRNA. (a) AUC ROC curve of the 28-mRNA signature using RF classifier for group classification. (b) Violin plot of the classification score of samples generated using the 28-mRNA predictive model (*** indicates p-value < 0.001 in a two-tailed Mann–Whitney U test). (c) AUC ROC curve of the 21-circRNA signature using RF classifier for group classification. (d) Violin plot of the classification score of samples generated using the 21-circRNA predictive model (*** indicates p-value < 0.001 in a two-tailed Mann–Whitney U test).

Figure 6

ML analysis using combinatorial signature of mRNA and circRNA. (a) AUC ROC curve of the 8-mRNA-circRNA signature using RF classifier for group classification. (b) Confusion matrix indicating the number of the correctly classified and misclassified samples based on the previous prediction model. (c) Violin plot of the classification score per samples generated using the 8-mRNA-circRNA predictive model (*** indicates p-value < 0.001 in a two-tailored Mann–Whitney U test).