Development of novel diagnostic system for pancreatic cancer, including early stages, measuring mRNA of whole blood cells

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is the most life‐threating disease among all digestive system malignancies. We developed a blood mRNA PDAC screening system using real‐time detection PCR to detect the expression of 56 genes, to discriminate PDAC from noncancer subjects. We undertook a clinical study to assess the performance of the developed system. We collected whole blood RNA from 53 PDAC patients, 102 noncancer subjects, 22 patients with chronic pancreatitis, and 23 patients with intraductal papillary mucinous neoplasms in a per protocol analysis. The sensitivity of the system for PDAC diagnosis was 73.6% (95% confidence interval, 59.7%‐84.7%). The specificity for noncancer volunteers, chronic pancreatitis, and patients with intraductal papillary mucinous neoplasms was 64.7% (54.6%‐73.9%), 63.6% (40.7%‐82.8%), and 47.8% (26.8%‐69.4%), respectively. Importantly, the sensitivity of this system for both stage I and stage II PDAC was 78.6% (57.1%‐100%), suggesting that detection of PDAC by the system is not dependent on the stage of PDAC. These results indicated that the screening system, relying on assessment of changes in mRNA expression in blood cells, is a viable alternative screening strategy for PDAC.

Abbreviations

CA

cancer antigen

CEA

carcinoembryonic antigen

CI

confidence interval

Cp

crossing point

IPMN

intraductal papillary mucinous neoplasm

NPV

negative predictive value

PDAC

pancreatic ductal adenocarcinoma

PPV

positive predictive value

RTD‐PCR

real‐time detection PCR

1. INTRODUCTION

Pancreatic ductal adenocarcinoma is among the most difficult malignancies to diagnose and treat, resulting in poor prognosis.1, 2 Approximately 10% of PDAC patients are diagnosed in an early phase (ie, stage I and II), when radical treatment involving surgical removal of the PDAC lesions is available.3, 4 Chemotherapy has only limited efficacy in terms of prolonging the survival of PDAC patients.5 The overall survival after diagnosis of PDAC is less than 5 years, which is the poorest among all digestive system malignancies. Therefore, novel diagnostic tools capable of detecting PDAC in the early stages, as well as sufficiently effective treatment options to substantially improve the prognosis of PDAC, are required.

Previously, we reported that the gene expression profile of peripheral blood showed characteristic alterations in the digestive system of cancer patients.6 The local cancer tissues of PDAC were substantially infiltrated with inflammatory cells, including immune‐suppressing cells as well as lymphocytes expressing the immune checkpoint molecule, programmed cell death 1 (PD‐1). In addition, certain genes expressed in whole peripheral blood, as well as CD4⁺ cells and CD14⁺ cells, were found to be important in the inflammation seen in PDAC.7 Therefore, peripheral blood cells reflect the host inflammatory condition in association with local cancer tissues in PDAC.

We developed a blood mRNA PDAC screening system based on determining changes in the mRNA expression levels of 56 genes in whole blood by RTD‐PCR. We undertook a multi‐institutional clinical study to test the performance of this system. The system performance was assessed in 53 PDAC patients, 102 noncancer subjects, 22 chronic pancreatitis patients, and 23 IPMN patients. The system showed sensitivity of 73.6% (95% CI, 59.7%‐84.7%). The specificity for noncancer volunteers, chronic pancreatitis patients, and IPMN patients was 64.7% (54.6%‐73.9%), 63.6% (40.7%‐82.8%), and 47.8% (26.8%‐69.4%), respectively. Thus, the screening system, based on examination of changes in gene expression in blood, could be useful.

2. MATERIALS AND METHODS

2.1. Complementary DNA microarray analysis to determine gene expression in whole blood

Twenty‐eight PDAC patients and 27 noncancer subjects (Table S1) were enrolled for gene expression analysis of whole blood cells, to select candidate genes for establishing a blood PDAC screening system. For blood collection, we used PAXgene blood RNA tubes (PreAnalytix, Hombrechtikon, Switzerland) in which the RNA of whole blood cells was immediately stabilized after collection and mixed with reagents in the tubes. The study was approved by the Institutional Review Board of Kanazawa University (Kanazawa, Japan) and its related hospitals.

We used a cDNA microarray Whole Human Genome 4× 44K Array kit (G4112F; Agilent Technologies, Santa Clara, CA, USA) for analysis of gene expression of whole blood obtained from 17 PDAC patients and 13 noncancer subjects, as described previously.7 Briefly, RNA was extracted from PAXgene blood RNA tubes using a PAXgene Blood RNA System kit (PreAnalytix) in accordance with the manufacturer's protocol, followed by RNA quantification using a NanoDrop ND‐1000 Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Cy‐3‐labeled cRNA was synthesized from 300 ng total RNA using a Quick Amp Labeling kit, and then hybridized to the cDNA microarray. The hybridized microarray was scanned using a cDNA Microarray Scanner (Model G2505B; Agilent Technologies), and the obtained image was processed using Feature Extraction software version 9.5.3.1 (Agilent Technologies) to obtain the value data. The data, which were normalized using GeneSpring GX software version 11.0.2 (Agilent Technologies), were used to select candidate genes for RTD‐PCR analysis. The cDNA microarray data were deposited in Gene Expression Omnibus (NBCI, Bethesda, MD, USA) with the number GSE125158.

2.2. Real‐time detection PCR

Gene expression analysis by RTD‐PCR for candidate genes selected from cDNA microarray data was undertaken as described previously.7 Briefly, cDNA was synthesized from total RNA extracted from samples in PAXgene Blood RNA tubes using an RT² HT First Strand Kit (Qiagen, Hilden, Germany). Synthesized cDNA was measured using Custom RT² Profiler PCR Arrays (Qiagen) according to the manufacturer's instructions. Quantitative PCR was carried out using a LightCycler 480 Instrument II version 1.5.0.39 (Roche, Basel, Switzerland).

2.3. Selection of genes for blood mRNA screening system

To establish the mRNA blood screening system, we used blood RNA from 28 PDAC patients and 27 noncancer subjects. Using RTD‐PCR, we measured the mRNA expression levels of candidate genes whose expression was assessed by cDNA microarray. For the selection of genes, we used a support vector machine with the linear kernel and sequential variable erasing methods.8 Discrimination between PDAC and noncancer subjects was undertaken using Fisher's linear discriminant with a penalized matrix for gene expression correlation analysis. We applied leave‐one‐out cross‐validation, using the RTD‐PCR data of 1 case as a test dataset for both gene selection and classifier building. Discrimination ability was assessed according to the sensitivity and specificity. The number of genes used for classification was based on cross‐validated sensitivity and specificity. Finally, we developed a final classifier by applying the whole algorithm with the determined number of genes to all 55 cases to be mounted in the blood mRNA screening system.

2.4. Clinical study design to assess performance of blood mRNA screening system

The clinical study to test the system performance was designed in consultation with the Japanese Pharmaceutical and Medical Device Agency, which is a Japanese regulatory agency, working together with the Ministry of Health, Labour, and Welfare to protect public health by assuring the safety, efficacy, and quality of pharmaceuticals and medical devices, and was approved by each participating institutional ethical committee (UMIN000020758). The primary end‐points were sensitivity and specificity of the blood mRNA PDAC screening system. The secondary end‐points were the PPV and NPV for the blood mRNA PDAC screening system, and the sensitivity, specificity, PPV, and NPV for CEA and CA19‐9. The clinical study was undertaken in 50 PDAC patients and 100 noncancer subjects. We also enrolled patients with IPMN and chronic pancreatitis, whose clinical data were not available. The inclusion criteria were as follows: PDAC patients, diagnosis by imaging studies or pathologically; chronic pancreatitis and IPMN patients, diagnosis by imaging studies; and non‐cancer subjects, considered normal based on medical records for the previous year or on current medical check‐up. The subjects were aged between 40 and 90 years. All subjects provided written informed consent prior to participating in the study. Exclusion criteria are shown in Table S2. The participating institutions were as follows (all based in Japan): Department of Gastroenterology, Kanazawa University Hospital; Department of Internal Medicine, Toyama Prefectural Central Hospital; Department of Gastroenterology, Toyama Municipal Hospital; Department of Gastroenterology, Tonami General Hospital; Department of Gastroenterology, Kanazawa Medical Center; Ishikawa Saiseikai Kanazawa Hospital; Department of Gastroenterology, Matto Ishikawa Central Hospital; Fukui Saiseikai Hospital; Department of Gastroenterology, Tsuruga Hospital; and Department of Internal Medicine, Keijyu General Hospital.

Blood was collected using PAXgene blood RNA tubes (PreAnalytiX) for mRNA expression measurement. Sera were also collected for CEA and CA19‐9 measurement by the standard chemiluminescent enzyme immunoassay. Results of CEA greater than 5.0 ng/mL and CA19‐9 greater than 37.0 U/mL were judged to be positive.

2.5. Statistical analysis

For the clinical validation study, we used Wilcoxon's signed‐rank test or Fisher's exact test for group comparisons. In evaluating the system performance, point estimates and 95% CIs of sensitivity and specificity were calculated. Pearson's correlation analysis and t test were applied using IBM SPSS Statistics (IBM, Armonk, NY, USA).

3. RESULTS

3.1. Establishment of blood PDAC screening system

We reported previously that the gene expression profile of whole blood cells from the digestive system of cancer patients was different from that of normal noncancer subjects.6 Based on these findings, we established a blood PDAC screening system using RTD‐PCR for genes, the expression levels of which in whole blood were altered in PDAC patients. We first extracted 2 sets of 384 genes assessing the cDNA microarray data of 17 PDAC patients and 13 noncancer subjects (GSE125158). The 384 genes in 1 set were selected by their expression, which was affected at FC >1.5 and P < 0.00013. The other set of 384 genes was selected by a powerful machine learning algorithms‐random forest method,9 in which 200 bootstraps were used to find the top genes that have more power to distinguish PDAC from noncancer subjects. The 179 genes that were included in both gene sets were finally selected. From the remaining genes in this analysis, we chose 134 genes, considering their biological meanings. We also selected 51 genes, whose flags for expression were positive in 12 of 17 PDAC patients and absent in 10 of 13 noncancer subjects. Thus, we finally selected 364 genes to further construct the equation for the PDAC screening system (Figure S1). For 28 PDAC patients and 27 noncancer subjects, including the above subjects, we carried out RTD‐PCR of these 364 genes and 17 housekeeping genes including GAPDH (Table S3). We used GAPDH as a reference as its expression was substantial and showed no difference between PDAC patients and noncancer subjects. Therefore, we have used delta Cp, which was obtained by subtraction of the GAPDH Cp value from that of each gene, for mRNA expression. To determine the gene number for discrimination of PDAC and noncancer subjects, we used the statistical method applied with the support vector machine and recursive feature elimination8 during the leave‐one‐out cross‐validation among these 55 cases (Figure S2). The cross‐validated sensitivity and specificity to discriminate PDAC from noncancer subjects were sufficiently high (93% and 100%, respectively) based on 56 genes; increasing the number of genes above 56 did not increase discrimination performance. By this analysis, we found that 56 genes were sufficient for the most excellent accuracy.

Therefore, we finally selected 56 genes (Table 1) for expression analysis in whole blood cells in 28 PDAC and 27 noncancer subjects.

Table 1.

Fifty‐six genes, their annotations, and coefficient values for equation of the blood mRNA screening system

Refseq #	Gene symbol	Gene description	Coefficient value	GOTERM_Biological process (BP)	GOTERM_Cellular component (CC)	GOTERM_Molecular function (MF)
NM_138340	SLC44A1	Solute carrier family 44 member 1	18.88705	GO:0006656~phosphatidylcholine biosynthetic process, GO:0015871~choline transport, GO:0042426~choline catabolic process, GO:0055085~transmembrane transport	GO:0005654~nucleoplasm, GO:0005739~mitochondrion, GO:0016020~membrane, GO:0070062~extracellular exosome	GO:0015220~choline transmembrane transporter activity
NM_005470	ABHD3	Abhydrolase domain containing 3	18.25234	GO:0006629~lipid metabolic process, GO:0046470~phosphatidylcholine metabolic process	GO:0005886~plasma membrane, GO:0016021~integral component of membrane	GO:0004623~phospholipase A2 activity, GO:0008970~phosphatidylcholine 1‐acylhydrolase activity, GO:0052739~phosphatidylserine 1‐acylhydrolase activity, GO:0052740~1‐acyl‐2‐lysophosphatidylserine acylhydrolase activity
NM_004457	PTPRC	Protein tyrosine phosphatase, receptor type C	15.18565	GO:0001915~negative regulation of T cell mediated cytotoxicity, GO:0001960~negative regulation of cytokine‐mediated signaling pathway, GO:0002244~hematopoietic progenitor cell differentiation, GO:0002378~immunoglobulin biosynthetic process, GO:0030217~T cell differentiation, GO:0030890~positive regulation of B cell proliferation, GO:0048539~bone marrow development, GO:0048864~stem cell development, GO:0050852~T cell receptor signaling pathway, GO:0050857~positive regulation of antigen receptor‐mediated signaling pathway	GO:0005886~plasma membrane GO:0005925~focal adhesion, GO:0045121~membrane raft, GO:0070062~extracellular exosome	GO:0005001~transmembrane receptor protein tyrosine phosphatase activity, GO:0005515~protein binding, GO:0019901~protein kinase binding
NM_001628	TLE2	Transducin like enhancer of split 2	14.37697	GO:0006351~transcription, DNA‐templated, GO:0007165~signal transduction, GO:0016055~Wnt signaling pathway, GO:0045892~negative regulation of transcription, DNA‐templated, GO:1904837~beta‐catenin‐TCF complex assembly	GO:0005615~extracellular space, GO:0005634~nucleus, GO:0005925~focal adhesion	GO:0003714~transcription corepressor activity, GO:0005515~protein binding
NM_014616	FNDC3B	Fibronectin type III domain containing 3B	14.19903	—	GO:0016021~integral component of membrane	GO:0044822~poly(A) RNA binding
NM_032047	LDHA	Lactate dehydrogenase A	13.84082	GO:0001666~response to hypoxia, GO:0005975~carbohydrate metabolic process, GO:0006089~lactate metabolic process, GO:0006090~pyruvate metabolic process, GO:0006096~glycolytic process,GO:0007584~response to nutrient, GO:0042542~response to hydrogen peroxide, GO:0043065~positive regulation of apoptotic process, GO:0043627~response to estrogen, GO:0051591~response to cAMP, GO:0055114~oxidation‐reduction process, GO:0098609~cell‐cell adhesion	GO:0005634~nucleus, GO:0005737~cytoplasm, GO:0005913~cell‐cell adherens junction, GO:0016020~membrane, GO:0016021~integral component of membrane, GO:0070062~extracellular exosome	GO:0004459~L‐lactate dehydrogenase activity, GO:0016616~oxidoreductase activity, acting on the CH‐OH group of donors, NAD or NADP as acceptor, GO:0019900~kinase binding, GO:0042802~identical protein binding, GO:0051287~NAD binding, GO:0098641~cadherin binding involved in cell‐cell adhesion
NM_001186	RPLP1	Ribosomal protein lateral stalk subunit P1	13.75907	GO:0000184~nuclear‐transcribed mRNA catabolic process, nonsense‐mediated decay, GO:0002181~cytoplasmic translation, GO:0006364~rRNA processing, GO:0045860~positive regulation of protein kinase activity	GO:0005737~cytoplasm, GO:0005840~ribosome, GO:0005925~focal adhesion, GO:0070062~extracellular exosome	GO:0003735~structural constituent of ribosome, GO:0005515~protein binding, GO:0030295~protein kinase activator activity, GO:0070180~large ribosomal subunit rRNA binding
NM_001145054	RAB10	RAB10, member RAS oncogene family	13.27343	GO:0006893~Golgi to plasma membrane transport, GO:0007264~small GTPase mediated signal transduction, GO:0007409~axonogenesis, GO:0016192~vesicle‐mediated transport, GO:0016197~endosomal transport, GO:0019882~antigen processing and presentation, GO:0030859~polarized epithelial cell differentiation, GO:0032869~cellular response to insulin stimulus, GO:0071786~endoplasmic reticulum tubular network organization, GO:0072659~protein localization to plasma membrane, GO:0097051~establishment of protein localization to endoplasmic reticulum membrane, GO:0098609~cell‐cell adhesion	GO:0000139~Golgi membrane, GO:0000145~exocyst, GO:0005622~intracellular, GO:0005768~endosome, GO:0005789~endoplasmic reticulum membrane, GO:0005886~plasma membrane, GO:0005911~cell‐cell junction, GO:0005913~cell‐cell adherens junction, GO:0005925~focal adhesion, GO:0030670~phagocytic vesicle membrane, GO:0032593~insulin‐responsive compartment, GO:0055038~recycling endosome membrane, GO:0070062~extracellular exosome, GO:0071782~endoplasmic reticulum tubular network	GO:0003924~GTPase activity, GO:0005515~protein binding, GO:0005525~GTP binding, GO:0019003~GDP binding, GO:0031489~myosin V binding, GO:0098641~cadherin binding involved in cell‐cell adhesion
NM_152523	HTATIP2	HIV‐1 Tat interactive protein 2	12.98543	GO:0001525~angiogenesis, GO:0006357~regulation of transcription from RNA polymerase II promoter, GO:0006915~apoptotic process, GO:0016032~viral process, GO:0030154~cell differentiation, GO:0051170~nuclear import, GO:0055114~oxidation‐reduction process	GO:0005634~nucleus, GO:0005635~nuclear envelope, GO:0005654~nucleoplasm, GO:0005737~cytoplasm, GO:0016020~membrane	GO:0003713~transcription coactivator activity, GO:0005515~protein binding, GO:0016491~oxidoreductase activity
NM_018455	BACH1	BTB domain and CNC homolog 1	12.79754	GO:0000083~regulation of transcription involved in G1/S transition of mitotic cell cycle, GO:0000117~regulation of transcription involved in G2/M transition of mitotic cell cycle, GO:0006281~DNA repair, GO:0006355~regulation of transcription	GO:0005634~nucleus, GO:0005737~cytoplasm, GO:0005829~cytosol, GO:0031463~Cul3‐RING ubiquitin ligase complex	GO:0001078~transcriptional repressor activity, RNA polymerase II core promoter proximal region sequence‐specific binding, GO:0001205~transcriptional activator activity, RNA polymerase II distal enhancer sequence‐specific binding, GO:0001206~transcriptional repressor activity, RNA polymerase II distal enhancer sequence‐specific binding, GO:0003700~transcription factor activity, sequence‐specific DNA binding, GO:0004842~ubiquitin‐protein transferase activity
NM_021023	NFE2L2	Nuclear factor, erythroid 2 like 2	11.75957	GO:0006351~transcription, GO:0006954~inflammatory response, GO:0010628~positive regulation of gene expression, GO:0016567~protein ubiquitination, GO:0030194~positive regulation of blood coagulation, GO:0030968~endoplasmic reticulum unfolded protein response, GO:0034599~cellular response to oxidative stress, GO:0036003~positive regulation of transcription from RNA polymerase II promoter in response to stress, GO:0036499~PERK‐mediated unfolded protein response, GO:0042149~cellular response to glucose starvation, GO:0045766~positive regulation of angiogenesis, GO:0071356~cellular response to tumor necrosis factor, GO:0071456~cellular response to hypoxia, GO:0071498~cellular response to fluid shear stress, GO:1902037~negative regulation of hematopoietic stem cell differentiation, GO:1903071~positive regulation of ER‐associated ubiquitin‐dependent protein catabolic process, GO:1903788~positive regulation of glutathione biosynthetic process	GO:0000785~chromatin, GO:0005634~nucleus, GO:0005737~cytoplasm, GO:0005813~centrosome, GO:0005886~plasma membrane, GO:0032993~protein‐DNA complex	GO:0000980~RNA polymerase II distal enhancer sequence‐specific DNA binding, GO:0001102~RNA polymerase II activating transcription factor binding, GO:0001205~transcriptional activator activity, RNA polymerase II distal enhancer sequence‐specific binding, GO:0003677~DNA binding, GO:0005515~protein binding
NM_080387	TET2	Tet methylcytosine dioxygenase 2	11.42427	GO:0001822~kidney development, GO:0002318~myeloid progenitor cell differentiation, GO:0006211~5‐methylcytosine catabolic process, GO:0006493~protein O‐linked glycosylation, GO:0007049~cell cycle, GO:0014070~response to organic cyclic compound, GO:0020027~hemoglobin metabolic process, GO:0030099~myeloid cell differentiation, GO:0045944~positive regulation of transcription from RNA polymerase II promoter, GO:0061484~hematopoietic stem cell homeostasis, GO:0070989~oxidative demethylation, GO:0080111~DNA demethylation, GO:0080182~histone H3‐K4 trimethylation	GO:0005634~nucleus	GO:0000907~sulfonate dioxygenase activity,GO:0003677~DNA binding,GO:0005515~protein binding,GO:0008198~ferrous iron binding,GO:0008270~zinc ion binding,GO:0018602~2,4‐dichlorophenoxyacetate alpha‐ketoglutarate dioxygenase activity,GO:0019798~procollagen‐proline dioxygenase activity,GO:0034792~hypophosphite dioxygenase activity,GO:0043734~DNA‐N1‐methyladenine dioxygenase activity,GO:0052634~C‐19 gibberellin 2‐beta‐dioxygenase activity,GO:0052635~C‐20 gibberellin 2‐beta‐dioxygenase activity,GO:0070579~methylcytosine dioxygenase activity
NM_000494	MBTPS2	Membrane bound transcription factor peptidase, site 2	11.32147	GO:0008203~cholesterol metabolic process, GO:0031293~membrane protein intracellular domain proteolysis, GO:0034976~response to endoplasmic reticulum stress, GO:0036500~ATF6‐mediated unfolded protein response, GO:0051091~positive regulation of sequence‐specific DNA binding transcription factor activity, GO:1990440~positive regulation of transcription from RNA polymerase II promoter in response to endoplasmic reticulum stress	GO:0000139~Golgi membrane ,GO:0005737~cytoplasm, GO:0016021~integral component of membrane	GO:0004222~metalloendopeptidase activity, GO:0008237~metallopeptidase activity, GO:0046872~metal ion binding
NM_016230	C2orf81	Chromosome 2 open reading frame 81	8.747677	—	—	—
NM_001195215	PPP1R13L	Protein phosphatase 1 regulatory subunit 13 like	8.60306	GO:0000122~negative regulation of transcription from RNA polymerase II promoter, GO:0006351~transcription, DNA‐templated, GO:0006915~apoptotic process, GO:0045597~positive regulation of cell differentiation, GO:0098609~cell‐cell adhesion, GO:1901796~regulation of signal transduction by p53 class mediator	GO:0005634~nucleus, GO:0005737~cytoplasm, GO:0005913~cell‐cell adherens junction, GO:0030054~cell junction, GO:0045171~intercellular bridge	GO:0003714~transcription corepressor activity, GO:0005515~protein binding, GO:0008134~transcription factor binding, GO:0042802~identical protein binding, GO:0098641~cadherin binding involved in cell‐cell adhesion
NM_024693	AKR1B1	Aldo‐keto reductase family 1 member B	8.005092	GO:0001894~tissue homeostasis, GO:0005975~carbohydrate metabolic process, GO:0006700~C21‐steroid hormone biosynthetic process, GO:0006950~response to stress, GO:0042415~norepinephrine metabolic process, GO:0046427~positive regulation of JAK‐STAT cascade, GO:0048661~positive regulation of smooth muscle cell proliferation, GO:0055114~oxidation‐reduction process, GO:0070301~cellular response to hydrogen peroxide	GO:0005615~extracellular space, GO:0005654~nucleoplasm, GO:0005737~cytoplasm, GO:0005829~cytosol, GO:0033010~paranodal junction, GO:0042629~mast cell granule, GO:0043220~Schmidt‐Lanterman incisure, GO:0048471~perinuclear region of cytoplasm, GO:0070062~extracellular exosome	GO:0004032~alditol:NADP+ 1‐oxidoreductase activity, GO:0004033~aldo‐keto reductase (NADP) activity, GO:0009055~electron carrier activity, GO:0016491~oxidoreductase activity, GO:0043795~glyceraldehyde oxidoreductase activity
NM_001776	COL17A1	Collagen type XVII alpha 1 chain	7.970873	GO:0007160~cell‐matrix adhesion, GO:0008544~epidermis development, GO:0031581~hemidesmosome assembly, GO:0050776~regulation of immune response	GO:0005576~extracellular region, GO:0005578~proteinaceous extracellular matrix, GO:0005581~collagen trimer, GO:0005604~basement membrane, GO:0005788~endoplasmic reticulum lumen, GO:0005886~plasma membrane, GO:0005911~cell‐cell junction GO:0030056~hemidesmosome	GO:0005515~protein binding
NM_016613	MCMBP	Minichromosome maintenance complex binding protein	7.570634	GO:0006261~DNA‐dependent DNA replication, GO:0007062~sister chromatid cohesion, GO:0007067~mitotic nuclear division, GO:0051301~cell division	GO:0000790~nuclear chromatin, GO:0005654~nucleoplasm, GO:0005737~cytoplasm, GO:0005886~plasma membrane, GO:0042555~MCM complex	GO:0003682~chromatin binding, GO:0005515~protein binding
NM_016623	FAR1	Fatty acyl‐CoA reductase 1	7.238576	GO:0006086~acetyl‐CoA biosynthetic process from pyruvate, GO:0008611~ether lipid biosynthetic process, GO:0010025~wax biosynthetic process, GO:0035336~long‐chain fatty‐acyl‐CoA metabolic process, GO:0046474~glycerophospholipid biosynthetic process, GO:0055114~oxidation‐reduction process	GO:0005777~peroxisome, GO:0005778~peroxisomal membrane, GO:0005782~peroxisomal matrix, GO:0016021~integral component of membrane	GO:0016491~oxidoreductase activity, GO:0050062~long‐chain‐fatty‐acyl‐CoA reductase activity, GO:0052814~medium‐chain‐aldehyde dehydrogenase activity, GO:0080019~fatty‐acyl‐CoA reductase (alcohol‐forming) activity
NM_032228	OSBPL8	Oxysterol binding protein like 8	6.956427	GO:0006869~lipid transport, GO:0010891~negative regulation of sequestering of triglyceride, GO:0030336~negative regulation of cell migration, GO:0032148~activation of protein kinase B activity, GO:0045444~fat cell differentiation, GO:0046628~positive regulation of insulin receptor signaling pathway, GO:0090204~protein localization to nuclear pore	GO:0005783~endoplasmic reticulum, GO:0016020~membrane, GO:0031965~nuclear membrane, GO:0032541~cortical endoplasmic reticulum	GO:0001786~phosphatidylserine binding, GO:0005548~phospholipid transporter activity, GO:0015485~cholesterol binding, GO:0070273~phosphatidylinositol‐4‐phosphate binding
NM_006682	ZMPSTE24	Zinc metallopeptidase STE24	6.555173	GO:0006508~proteolysis, GO:0006998~nuclear envelope organization, GO:0030327~prenylated protein catabolic process, GO:0071586~CAAX‐box protein processing	GO:0005637~nuclear inner membrane, GO:0005789~endoplasmic reticulum membrane, GO:0016020~membrane, GO:0070062~extracellular exosome	GO:0004222~metalloendopeptidase activity, GO:0008235~metalloexopeptidase activity, GO:0046872~metal ion binding
NM_022763	CCNYL1	Cyclin Y like 1	6.300126	GO:0000079~regulation of cyclin‐dependent protein serine/threonine kinase activity	GO:0005886~plasma membrane	GO:0005515~protein binding, GO:0019901~protein kinase binding
NM_003838	RRAGD	Ras related GTP binding D	6.236001	GO:0007050~cell cycle arrest, GO:0009267~cellular response to starvation, GO:0010506~regulation of autophagy, GO:0016236~macroautophagy, GO:0032008~positive regulation of TOR signaling, GO:0071230~cellular response to amino acid stimulus	GO:0005634~nucleus, GO:0005737~cytoplasm, GO:0005764~lysosome, GO:0005829~cytosol, GO:0034448~EGO complex, GO:1990131~Gtr1‐Gtr2 GTPase complex	GO:0003924~GTPase activity, GO:0005515~protein binding, GO:0005525~GTP binding, GO:0046982~protein heterodimerization activity
NM_003774	GALNT4	Polypeptide N‐acetylgalactosaminyltransferase 4	6.227974	GO:0005975~carbohydrate metabolic process, GO:0016266~O‐glycan processing	GO:0005794~Golgi apparatus, GO:0016021~integral component of membrane, GO:0048471~perinuclear region of cytoplasm, GO:0070062~extracellular exosome	GO:0004653~polypeptide N‐acetylgalactosaminyltransferase activity, GO:0030246~carbohydrate binding, GO:0046872~metal ion binding
NM_015473	UBE2W	Ubiquitin conjugating enzyme E2 W (putative)	5.881697	GO:0006281~DNA repair, GO:0006513~protein monoubiquitination, GO:0071218~cellular response to misfolded protein	GO:0005634~nucleus, GO:0005737~cytoplasm, GO:0016021~integral component of membrane	GO:0004842~ubiquitin‐protein transferase activity, GO:0005515~protein binding, GO:0005524~ATP binding
NM_006410	HEATR5A	HEAT repeat containing 5A	5.784991	—	—	—
NM_000416	UBE4A	Ubiquitination factor E4A	5.612217	GO:0000209~protein polyubiquitination, GO:0007165~signal transduction, GO:0030433~ER‐associated ubiquitin‐dependent protein catabolic process	GO:0000151~ubiquitin ligase complex, GO:0005634~nucleus, GO:0005737~cytoplasm	GO:0004114~3’,5’‐cyclic‐nucleotide phosphodiesterase activity, GO:0005515~protein binding, GO:0016874~ligase activity, GO:0034450~ubiquitin‐ubiquitin ligase activity
NM_201612	ABI1	Abl interactor 1	5.533139	GO:0006928~movement of cell or subcellular component, GO:0007169~transmembrane receptor protein tyrosine kinase signaling pathway, GO:0008154~actin polymerization or depolymerization, GO:0008285~negative regulation of cell proliferation, GO:0016032~viral process, GO:0018108~peptidyl‐tyrosine phosphorylation, GO:0035855~megakaryocyte development, GO:0038096~Fc‐gamma receptor signaling pathway involved in phagocytosis, GO:0048010~vascular endothelial growth factor receptor signaling pathway, GO:0061098~positive regulation of protein tyrosine kinase activity, GO:0072673~lamellipodium morphogenesis, GO:0098609~cell‐cell adhesion	GO:0005622~intracellular, GO:0005634~nucleus, GO:0005737~cytoplasm, GO:0005783~endoplasmic reticulum, GO:0005856~cytoskeleton, GO:0005913~cell‐cell adherens junction, GO:0030027~lamellipodium, GO:0030175~filopodium, GO:0030426~growth cone, GO:0031209~SCAR complex, GO:0045211~postsynaptic membrane, GO:0070062~extracellular exosome	GO:0005515~protein binding, GO:0008092~cytoskeletal protein binding, GO:0017124~SH3 domain binding, GO:0030296~protein tyrosine kinase activator activity, GO:0032403~protein complex binding, GO:0098641~cadherin binding involved in cell‐cell adhesion
NM_005566	PPIH	Peptidylprolyl isomerase H	5.40687	GO:0000398~mRNA splicing, via spliceosome, GO:0000413~protein peptidyl‐prolyl isomerization, GO:0006457~protein folding, GO:0006461~protein complex assembly, GO:0045070~positive regulation of viral genome replication	GO:0005654~nucleoplasm, GO:0005681~spliceosomal complex, GO:0005737~cytoplasm, GO:0016607~nuclear speck, GO:0046540~U4/U6 x U5 tri‐snRNP complex, GO:0071001~U4/U6 snRNP	GO:0003755~peptidyl‐prolyl cis‐trans isomerase activity, GO:0005515~protein binding, GO:0016018~cyclosporin A binding, GO:0043021~ribonucleoprotein complex binding
NM_005767	CENPN	Centromere protein N	4.732161	GO:0007059~chromosome segregation, GO:0007062~sister chromatid cohesion, GO:0007067~mitotic nuclear division, GO:0034080~CENP‐A containing nucleosome assembly, GO:0034508~centromere complex assembly	GO:0000775~chromosome, centromeric region, GO:0000777~condensed chromosome kinetochore, GO:0005654~nucleoplasm, GO:0005829~cytosol	—
NM_015884	MIER1	MIER1 transcriptional regulator	4.196767	GO:0006351~transcription, DNA‐templated, GO:0007165~signal transduction, GO:0031937~positive regulation of chromatin silencing, GO:0043123~positive regulation of I‐kappaB kinase/NF‐kappaB signaling	GO:0005654~nucleoplasm, GO:0005737~cytoplasm, GO:0017053~transcriptional repressor complex	GO:0003677~DNA binding, GO:0004871~signal transducer activity, GO:0005515~protein binding
NM_024834	IKBIP	IKBKB interacting protein	−4.35974	GO:0010165~response to X‐ray	GO:0005783~endoplasmic reticulum, GO:0016020~membrane	GO:0005515~protein binding
NM_020948	SLC22A15	Solute carrier family 22 member 15	−5.63841	GO:0006811~ion transport, GO:0055085~transmembrane transport	GO:0016021~integral component of membrane	GO:0022857~transmembrane transporter activity, GO:0022891~substrate‐specific transmembrane transporter activity
NM_001031716	IFNGR1	Interferon gamma receptor 1	−6.31219	GO:0007165~signal transduction, GO:0009615~response to virus, GO:0060333~interferon‐gamma‐mediated signaling pathway, GO:0060334~regulation of interferon‐gamma‐mediated signaling pathway	GO:0005783~endoplasmic reticulum, GO:0005886~plasma membrane, GO:0030425~dendrite, GO:0031982~vesicle	GO:0004906~interferon‐gamma receptor activity, GO:0005515~protein binding, GO:0042015~interleukin‐20 binding
NM_006164	CFHR3	Complement factor H related 3	−6.547	—	GO:0005615~extracellular space, GO:0070062~extracellular exosome, GO:0072562~blood microparticle
NM_020841	ENTPD1	Ectonucleoside triphosphate diphosphohydrolase 1	−7.19342	GO:0007155~cell adhesion, GO:0007596~blood coagulation	GO:0005886~plasma membrane, GO:0016020~membrane, GO:0070062~extracellular exosome	GO:0005515~protein binding, GO:0005524~ATP binding, GO:0016787~hydrolase activity
NM_006347	CLEC4D	C‐type lectin domain family 4 member D	−7.20897	GO:0002223~stimulatory C‐type lectin receptor signaling pathway, GO:0002250~adaptive immune response, GO:0002292~T cell differentiation involved in immune response, GO:0038094~Fc‐gamma receptor signaling pathway, GO:0045087~innate immune response	GO:0005886~plasma membrane, GO:0016021~integral component of membrane	GO:0030246~carbohydrate binding, GO:0034987~immunoglobulin receptor binding
NM_006663	USP15	Ubiquitin specific peptidase 15	−7.21872	GO:0006511~ubiquitin‐dependent protein catabolic process, GO:0007179~transforming growth factor beta receptor signaling pathway, GO:0030509~BMP signaling pathway, GO:0035616~histone H2B conserved C‐terminal lysine deubiquitination, GO:0060389~pathway‐restricted SMAD protein phosphorylation	GO:0005634~nucleus, GO:0005737~cytoplasm	GO:0004197~cysteine‐type endopeptidase activity, GO:0004843~thiol‐dependent ubiquitin‐specific protease activity, GO:0005160~transforming growth factor beta receptor binding, GO:0005515~protein binding, GO:0046332~SMAD binding, GO:0061649~ubiquitinated histone binding
NM_018699	SLFN12	Schlafen family member 12	−7.8338	—	—	GO:0005524~ATP binding
NM_002838	FGL2	Fibrinogen like 2	−8.25242	GO:0006508~proteolysis	GO:0005577~fibrinogen complex, GO:0070062~extracellular exosome	GO:0008233~peptidase activity
NM_016131	ACSL3	Acyl‐CoA synthetase long‐chain family member 3	−8.30246	GO:0001676~long‐chain fatty acid metabolic process, GO:0006633~fatty acid biosynthetic process, GO:0008152~metabolic process, GO:0014070~response to organic cyclic compound, GO:0034379~very‐low‐density lipoprotein particle assembly, GO:0042998~positive regulation of Golgi to plasma membrane protein transport, GO:0044539~long‐chain fatty acid import, GO:0051047~positive regulation of secretion, GO:2001247~positive regulation of phosphatidylcholine biosynthetic process	GO:0005741~mitochondrial outer membrane, GO:0005778~peroxisomal membrane, GO:0005783~endoplasmic reticulum, GO:0005789~endoplasmic reticulum membrane, GO:0005794~Golgi apparatus, GO:0005811~lipid particle, GO:0016020~membrane, GO:0048471~perinuclear region of cytoplasm	GO:0003824~catalytic activity, GO:0004467~long‐chain fatty acid‐CoA ligase activity, GO:0005524~ATP binding,GO:0016874~ligase activity, GO:0019901~protein kinase binding, GO:0019904~protein domain specific binding, GO:0102391~decanoate‐CoA ligase activity
NM_001003	ZEB2	Zinc finger E‐box binding homeobox 2	−8.42855	GO:0000122~negative regulation of transcription from RNA polymerase II promoter, GO:0006351~transcription, DNA‐templated, GO:0048066~developmental pigmentation, GO:0050790~regulation of catalytic activity,GO:0097324~melanocyte migration	GO:0005634~nucleus	GO:0001227~transcriptional repressor activity, RNA polymerase II transcription regulatory region sequence‐specific binding, GO:0003676~nucleic acid binding, GO:0005515~protein binding, GO:0019208~phosphatase regulator activity, GO:0046872~metal ion binding
NM_021244	FPGT	Fucose‐1‐phosphate guanylyltransferase	−8.48534	GO:0006004~fucose metabolic process	GO:0005737~cytoplasm GO:0005829~cytosol	GO:0003824~catalytic activity, GO:0005525~GTP binding, GO:0016772~transferase activity, transferring phosphorus‐containing groups, GO:0016779~nucleotidyltransferase activity, GO:0047341~fucose‐1‐phosphate guanylyltransferase activity
NM_018420	ECHDC3	Enoyl‐CoA hydratase domain containing 3	−8.66068	GO:0008152~metabolic process	GO:0005739~mitochondrion	GO:0003824~catalytic activity
NM_080546	PRDM5	PR/SET domain 5	−8.77508	GO:0000122~negative regulation of transcription from RNA polymerase II promoter, GO:0000278~mitotic cell cycle, GO:0006351~transcription, DNA‐templated, GO:0016575~histone deacetylation, GO:0045892~negative regulation of transcription, DNA‐templated, GO:0051567~histone H3‐K9 methylation	GO:0005634~nucleus	GO:0001078~transcriptional repressor activity, RNA polymerase II core promoter proximal region sequence‐specific binding, GO:0003676~nucleic acid binding, GO:0005515~protein binding, GO:0008168~methyltransferase activity, GO:0044212~transcription regulatory region DNA binding, GO:0046872~metal ion binding, GO:0070491~repressing transcription factor binding
NM_018042	FAM49B	Family with sequence similarity 49 member B	−8.78217	—	GO:0005929~cilium, GO:0016020~membrane, GO:0070062~extracellular exosome	GO:0005515~protein binding
NM_018079	VPS51 (C11orf2)	VPS51, GARP complex subunit	−8.85243	GO:0006869~lipid transport, GO:0006914~autophagy, GO:0007041~lysosomal transport, GO:0015031~protein transport, GO:0032456~endocytic recycling, GO:0042147~retrograde transport, endosome to Golgi	GO:0000938~GARP complex, GO:0005794~Golgi apparatus, GO:0005829~cytosol, GO:0016020~membrane, GO:0016021~integral component of membrane, GO:0055037~recycling endosome, GO:1990745~EARP complex	GO:0005515~protein binding
NM_017628	DENND1B	DENN domain containing 1B	−10.4299	GO:0015031~protein transport, GO:0032456~endocytic recycling, GO:0035745~T‐helper 2 cell cytokine production, GO:0043547~positive regulation of GTPase activity, GO:0050776~regulation of immune response, GO:0050852~T cell receptor signaling pathway	GO:0005829~cytosol, GO:0030136~clathrin‐coated vesicle	GO:0017112~Rab guanyl‐nucleotide exchange factor activity, GO:0017137~Rab GTPase binding
NM_003260	NABP1 (OBFC2A)	Nucleic acid binding protein 1	−11.0293	GO:0006281~DNA repair, GO:0006974~cellular response to DNA damage stimulus, GO:0007093~mitotic cell cycle checkpoint, GO:0010212~response to ionizing radiation, GO:0042795~snRNA transcription from RNA polymerase II promoter	GO:0000784~nuclear chromosome, telomeric region, GO:0005737~cytoplasm, GO:0070876~SOSS complex	GO:0003697~single‐stranded DNA binding, GO:0003723~RNA binding, GO:0005515~protein binding
NM_018299	CYB5R4	Cytochrome b5 reductase 4	−11.661	GO:0003032~detection of oxygen, GO:0006091~generation of precursor metabolites and energy, GO:0006801~superoxide metabolic process, GO:0015701~bicarbonate transport, GO:0030073~insulin secretion, GO:0042593~glucose homeostasis, GO:0048468~cell development ,GO:0055114~oxidation‐reduction process	GO:0005783~endoplasmic reticulum, GO:0048471~perinuclear region of cytoplasm	GO:0004128~cytochrome‐b5 reductase activity, acting on NAD(P)H, GO:0016174~NAD(P)H oxidase activity, GO:0016653~oxidoreductase activity, acting on NAD(P)H, heme protein as acceptor, GO:0020037~heme binding, GO:0046872~metal ion binding, GO:0071949~FAD binding
NM_004788	LPAR6	Lysophosphatidic acid receptor 6	−11.9776	GO:0035025~positive regulation of Rho protein signal transduction, GO:0051482~positive regulation of cytosolic calcium ion concentration involved in phospholipase C‐activating G‐protein coupled signaling pathway	GO:0005886~plasma membrane, GO:0005887~integral component of plasma membrane, GO:0016021~integral component of membrane	GO:0004930~G‐protein coupled receptor activity, GO:0070915~lysophosphatidic acid receptor activity
NM_006313	SRBD1	S1 RNA binding domain 1	−12.4726	GO:0006139~nucleobase‐containing compound metabolic process	—	GO:0003723~RNA binding
NM_013265	B3GNT5	UDP‐GlcNAc:betaGal beta‐1,3‐N‐acetylglucosaminyltransferase 5	−12.6434	GO:0006486~protein glycosylation, GO:0007417~central nervous system development, GO:0009247~glycolipid biosynthetic process	GO:0000139~Golgi membrane, GO:0005622~intracellular, GO:0016021~integral component of membrane	GO:0008378~galactosyltransferase activity,GO:0008457~beta‐galactosyl‐N‐acetylglucosaminylgalactosylglucosyl‐ceramide beta‐1,3‐acetylglucosaminyltransferase activity, GO:0008499~UDP‐galactose:beta‐N‐acetylglucosamine beta‐1,3‐galactosyltransferase activity, GO:0047256~lactosylceramide 1,3‐N‐acetyl‐beta‐D‐glucosaminyltransferase activity
NM_015626	FAM198B	Family with sequence similarity 198 member B	−13.2931	—	GO:0000139~Golgi membrane, GO:0005794~Golgi apparatus	—
NM_014795	WSB1	WD repeat and SOCS box containing 1	−13.4219	GO:0000209~protein polyubiquitination, GO:0016567~protein ubiquitination, GO:0035556~intracellular signal transduction	GO:0005622~intracellular, GO:0005829~cytosol	GO:0004842~ubiquitin‐protein transferase activity, GO:0005515~protein binding
NM_005857	ATP11B	ATPase phospholipid transporting 11B (putative)	−16.3131	GO:0006811~ion transport, GO:0015917~aminophospholipid transport, GO:0034220~ion transmembrane transport, GO:0045332~phospholipid translocation	GO:0005637~nuclear inner membrane, GO:0005769~early endosome, GO:0005783~endoplasmic reticulum, GO:0005794~Golgi apparatus, GO:0005886~plasma membrane, GO:0016020~membrane, GO:0055037~recycling endosome	GO:0000287~magnesium ion binding, GO:0004012~phospholipid‐translocating ATPase activity, GO:0005515~protein binding, GO:0005524~ATP binding, GO:0015075~ion transmembrane transporter activity

—, no data available.

As a final classifier, obtained through application of the classification algorithm based on 56 genes to all 55 subjects, we obtained the following equation for PDAC diagnosis using the weighted linear sum of each gene's expression level:

$$\begin{array}{cc}& \text{Sum‐value}=\hfill \\ & \text{intercept}+{\mathrm{a}}_1\ast \text{gene}1+{\mathrm{a}}_2\ast \text{gene}2+\dots \dots +{\mathrm{a}}_{56}\ast \ast \text{gene}56\hfill \end{array}$$

In the formula above, a_n is a coefficient value for each gene (Table 1); “a1*gene 1 + a2*gene 2 + ……… + a56**gene 56” was determined based on Fisher's linear discriminant, and its threshold was 209.018. To simplify the threshold value as zero by the formula, intercept value, −209.018, in the equation, was mounted. Thus, sum‐values 0 or higher and less than 0 were taken as positive and negative for PDAC, respectively.

3.2. Characteristics of subjects registered in the clinical study used to assess the performance of blood mRNA screening system

We designed a clinical study to assess the performance of the blood mRNA screening system. The study cohort consisted of 54 patients with PDAC, 25 with IPMN, 22 with chronic pancreatitis, and 104 noncancer subjects (Figure 1). All subjects were assessed for suitability for the safety analysis cohort and full analysis set. After applying the exclusion criteria, we finally included 53 patients with PDAC, 23 with IPMN, 22 with chronic pancreatitis, and 102 noncancer subjects.

Figure 1.

Outline of the clinical study to assess the performance of the blood mRNA screening system for pancreatic ductular adenocarcinoma (PDAC). The study cohort consisted of 54 patients with PDAC, 22 with chronic pancreatitis, 25 with intraductal papillary mucinous neoplasm (IPMN), and 104 noncancer subjects. No cases were excluded from the full analysis set. In the per protocol set, 1 PDAC patient, 2 IPMN patients, and 2 noncancer volunteers were excluded. Thus, 53 PDAC patients, 22 chronic pancreatitis patients, 23 IPMN patients, and 102 noncancer subjects were included in the per protocol analysis

Some of the clinical characteristics, such as age, gender, and body mass index, were different between PDAC patients and noncancer subjects (Table 2). The frequency of alcohol drinking was not different between PDAC patients and noncancer subjects (Table 2).

Table 2.

Clinical characteristics of study subjects to assess the blood mRNA screening system for pancreatic ductal adenocarcinoma (PDAC)

	PDAC	Chronic pancreatitis	IPMN	Noncancer
Objective number	53	22	23	102
Age (y)
Average	69.6	61.7	67.7	63.4
SD	10.7	10.8	8.2	11.4
Number	53	22	23	102
Wilcoxon signed‐rank test	Ref	P = .008	P = .399	P = .003
	—	Ref	P = .031	P = .484
	—	—	Ref	P = .122
Gender
Male	26 (49.1%)	19 (86.4%)	11 (47.8%)	69 (67.6%)
Female	27 (50.0%)	3 (13.6%)	12 (52.2%)	33 (32.4%)
Fisher's exact test	Ref	P = .004	P = 1.000	P = .036
	—	Ref	P = .011	P = .119
	—	—	Ref	P = .093
Height (cm)
Average	159.63	167.93	159.82	162.99
SD	8.63	5.50	8.08	9.31
N	52	21	22	94
Wilcoxon signed‐rank test	Ref	P < .001	P = .717	P = .027
	—	Ref	P = .002	P = .015
	—	—	Ref	P = .156
Body weight (kg)
Average	54.18	59.11	57.90	61.61
SD	12.26	10.18	9.59	11.49
N	51	21	22	95
Wilcoxon signed‐rank test	Ref	0.132	P = .146	P < .001
	—	Ref	P = .884	P = .287
	—	—	Ref	P = .203
BW loss >5 kg over 6 months
No (N)	29	16	12	75
Yes (N)	10	1	0	1
Fisher's exact test	Ref	P = .144	P = .092	P < .001
	—	Ref	P = 1.000	P = .334
	—	—	Ref	P = 1.000
Body mass index
Average	21.04	20.94	22.55	23.09
SD	3.34	3.45	2.72	3.27
N	51	21	22	93
Wilcoxon signed‐rank test	Ref	P = .536	P = .100	P = .002
	—	Ref	P = .037	P = .004
	—	—	Ref	P = 0.477
Smoking
Nonsmoker	31 (58.5%)	6 (27.3%)	9 (39.1%)	63 (61.8%)
Past smoker	13 (24.5%)	5 (22.7%)	8 (34.8%)	22 (21.6%)
Current smoker	6 (11.3%)	8 (36.4%)	2 (8.7%)	12 (11.8%)
Fisher's exact test	Ref	P = .020	P = .419	P = .933
	—	Ref	P = .119	P = .005
	—	—	Ref	P = .226
Alcohol drinking
Nondrinker	33 (62.3%)	6 (27.3%)	6 (26.1%)	42 (41.2%)
Past drinker	3 (5.7%)	4 (18.2%)	0 (0.0%)	4 (3.9%)
Current drinker	17 (32.1%)	9 (40.9%)	13 (56.5%)	49 (48.0%)
Fisher's exact test	Ref	P = .032	P = .027	P = .055
	—	Ref	P = .144	P = .051
	—	—	Ref	P = .398
Familial history of PDAC
No	24 (45.3%)	14 (63.6%)	14 (60.9%)	67 (65.7%)
Yes	3 (5.7%)	0 (0.0%)	1 (4.3%)	3 (2.9%)
Unknown	26 (49.1%)	8 (36.4%)	8 (34.8%)	32 (31.4%)
Fisher's exact test	Ref	P = .539	P = 1.000	P = .344
	—	Ref	P = 1.000	P = 1.000
	—	—	Ref	P = .547

—, not applicable; BW, body weight; IPMN, intraductal papillary mucinous neoplasm; Ref, reference.

3.3. Performance of blood mRNA screening system

The sensitivity and specificity of the blood mRNA PDAC screening system for 53 PDAC patients and 102 noncancer subjects was 73.6% (95% CI, 59.7%‐84.7%) and 64.7% (54.6%‐73.9%), respectively. The specificity for chronic pancreatitis and IPMN patients was 63.6% (95% CI, 40.7%‐82.8%) and 47.8% (95% CI, 26.8%‐69.4%), respectively (Table 3). The sensitivity for the conventional tumor markers CEA and CA19‐9 was 41.5% (95% CI, 28.1%‐55.9%) and 71.7% (95% CI, 57.7%‐83.2%), respectively (Tables 4 and 5). In noncancer subjects, the specificity for CEA and CA19‐9 was 96.1% (95% CI, 90.3%‐98.9%) and 97.1% (95% CI, 91.6%‐99.4%), respectively (Tables 4 and 5). The PPVs of the mRNA blood screening system, CEA, and CA19‐9 for chronic pancreatitis and IPMN were 76.5%‐83.0%, 71.9%, and 95.1%, respectively (Tables 3, 4, 5); the respective values for noncancer subjects were 50.6%, 82.1%, and 92.9%, respectively. The NPVs of the mRNA blood screening system, CEA, and CA19‐9 for all cases were 86.1%, 81.1%, and 90.4%, respectively (Tables 3, 4, 5). Pearson's correlation analysis was applied between these conventional tumor markers and scores of the mRNA screening system. The correlation coefficients for CEA and CA19‐9 with scores of the mRNA screening system were 0.072 and 0.14, respectively, with P‐values of .308 and .048, respectively. These results indicate no correlation between CEA and the mRNA screening system, and very weak correlation between CA19‐9 and the mRNA screening system (Figure 2).

Table 3.

Senstivity and specificity of blood mRNA screening system for pancreatic ductal adenocarcinoma (PDAC)

	PDAC	Chronic pancreatitis	IPMN	Noncancer
Objective number	53	22	23	102
mRNA system result
Positive	39 (73.6%)	8 (36.4%)	12 (52.2%)	36 (35.3%)
Negative	14 (26.4%)	14 (63.6%)	11 (47.8%)	66 (64.7%)
Sensitivity
Point estimation	73.6%	0 (0.0%)	0 (0.0%)	0 (0.0%)
95% CI	59.7%‐84.7%	—	—	—
Specificity
Point estimation	—	63.6%	47.8%	64.7%
95% CI	—	40.7%‐82.8%	26.8%‐69.4%	54.6%‐73.9%

—, not applicable; CI, confidence interval; IPMN, intraductal papillary mucinous neoplasm.

Table 4.

Senstivity and specificity of carcinoembryonic antigen (CEA) for pancreatic ductal adenocarcinoma (PDAC)

	PDAC	Chronic pancreatitis	IPMN	Noncancer
Objective number	53	22	23	102
Result
Positive	22 (41.5%)	9 (40.9%)	2 (8.7%)	4 (3.9%)
Negative	31 (58.5%)	13 (59.1%)	21 (91.3%)	98 (96.1%)
Sensitivity
Point estimation	41.5%	—	—	—
95% CI	28.1%‐55.9%	—	—	—
Specificity
Point estimation	—	59.1%	91.3%	96.1%
95% CI	—	36.4%‐79.3%	72.0%‐98.9%	90.3%‐98.9%

—, not applicable; CI, confidence interval; IPMN, intraductal papillary mucinous neoplasm.

Table 5.

Senstivity and specificity of cancer antigen (CA)19‐9 for pancreatic ductal adenocarcinoma (PDAC)

	PDAC	Chronic pancreatitis	IPMN	Noncancer
Objective number	53	22	23	102
Result
Positive	38 (71.7%)	2 (9.1%)	2 (8.7%)	3 (2.9%)
Negative	15 (28.3%)	20 (90.9%)	21 (91.3%)	99 (97.1%)
Sensitivity
Point estimation	71.7%	—	—	—
95% CI	57.7%‐83.2%	—	—	—
Specificity
Point estimation	—	90.9%	91.3%	97.1%
95% CI	—	70.8%‐98.9%	72.0%‐98.9%	91.6%‐99.4%

—, not applicable; CI, confidence interval; IPMN, intraductal papillary mucinous neoplasm.

Figure 2.

Scatter plots of carcinoembryonic antigen (CEA), cancer antigen (CA)19‐9, and sum‐values of blood mRNA screening system for pancreatic ductular adenocarcinoma. Serum concentration values of CEA (A) and serum unit values of CA19‐9 (B) with sum‐values of the blood mRNA screening system of 53 patients with pancreatic ductular adenocarcinoma (filled circle), 23 with intraductal papillary mucinous neoplasm (open triangle), 22 with chronic pancreatitis (open diamond), and 102 noncancer subjects (open square). The interval of y‐axis and x‐axis values is expressed in logarithmic scale. Solid line indicates regression lines, dashed line indicates the cut‐off value for tumor markers (A, B). R ² values are 0.021 (A) and 0.08 (B)

Regarding disease stages, the blood mRNA screening system showed a sensitivity of 78.6% for stages I and II, and a similar value of 71.8% for stages III and IV (Table 6).

Table 6.

Sensitivity of mRNA screening system for pancreatic ductal adenocarcinoma (PDAC) staging

	Stage
	I, II	III, IV
Number	14	39
Positive by mRNA screening system	11	28
Sensitivity (%)	78.6	71.8

3.4. Clinical characteristics and blood mRNA screening system

In terms of clinical parameters and sum‐values of the blood mRNA screening system, we applied correlation analyses for continuous data (age, height, and body weight) and t tests for category data (gender, body weight loss of more than 5 kg over 6 months, smoking, alcohol drinking, and familial history of PDAC) with sum‐values of the blood mRNA screening system (Table S4). Although the P value for the age correlation analysis was less than .05, correlation efficiency was weak (0.145). Average values of the sum‐values of the blood mRNA screening system for nonsmoker and current smoker were significantly different (P = .047), but both values were less than zero. Although the sum‐value of past drinker was significantly higher than those of nondrinker and current drinker (P = .045 and P = .043, respectively), the sum‐value for the current drinker was not higher than current drinker or nondrinker.

4. DISCUSSION

In this study, we developed a blood mRNA PDAC screening system measuring the expression of 56 genes in whole blood by RTD‐PCR, followed by weighted linear summation. We undertook a multicenter clinical study to validate the performance of this blood mRNA screening system for PDAC detection, by gene expression analysis using RTD‐PCR. This system showed that sensitivity was 73.6% and the specificity for healthy subjects was 64.7%.

The principle of this system is unique, being based on changes in gene expression in the peripheral blood.6, 7, 10, 11 Previously, we reported that PDAC is related to inflammatory changes in local tumor tissue, as well in peripheral blood.7, 10 Therefore, we assumed that alterations in the gene expression profile in the peripheral blood of PDAC patients would be associated with systemic inflammatory responses of the host. Based on this principle, we selected genes that showed differences in expression between the whole blood from PDAC patients and noncancer subjects.

The prognosis of PDAC is extremely poor due to in part to difficulty of diagnosis in the early stages.12 At present, there is no definitive screening method for PDAC, and various methods are used clinically, including ultrasonography and computed tomography, for imaging‐based diagnosis and determination of tumor markers in blood.13 With regard to blood test tumor markers, CA19‐9,14 Span‐1,15 CA50, CA242, DUPAN‐2, and TAG‐7216, 17, 18 are categorized as glucose‐chain antigens; CEA is an embryonic antigen19 that is typically present in high levels in PDAC, although not in all cases. In addition to these conventional tumor markers, a new approach has been developed using excreted small particles, called exosomes; these are believed to include molecules specific for cancers,20 although the analysis methods are still under development.

As diagnosis of PDAC in the early stages is extremely rare, there are insufficient data regarding the performance of the mRNA screening system for early‐stage PDAC. The PDAC patients in this study were mainly in advanced stages (ie, III or IV); only 14 patients were in stage I or II. The sensitivity of the mRNA screening system for these PDAC patients was 78.6%, illustrating the potential benefit of this system for early detection of PDAC. We previously disclosed that PDAC local tissues were associated with inflammation, even if they were still small.7 Thus, the high sensitivity of the mRNA screening system, capable of detecting PDAC in early stages, might be an important beneficial feature, one that has proven difficult using other conventional tumor marker systems. Based on this previous study, the principle of the mRNA screening system relies on gene expression alteration of peripheral blood cells affected by development of PDAC. This is the feature that distinguishes the system from conventional tumor markers (ie, CEA and CA19‐9), which are produced by cancer tissues. In this context, the novel mRNA system would be a useful screening method as the examination involves a blood test, which can be undertaken without the need for specific medical devices and is minimally invasive.

The establishment of a PDAC screening algorithm requires further development, including the novel blood mRNA screening system described herein, along with other tumor markers and diagnostic imaging tools.21 Although ultrasonography is noninvasive, it cannot be used to determine dilatation of the pancreatic duct or tumorous lesions in the pancreas; therefore, patients also require endoscopy,22 computed tomography, and MRI examinations.23 As discussed above, conventional tumor markers are not sufficient for screening for PDAC in the early stages, which could allow the disease to be cured by surgery.

The genes included in this system were selected biostatistically, based on changes in their expression levels in the whole blood of PDAC patients forming a different cohort from that of the clinical study. The mechanisms by which these genes are affected in PDAC are intriguing; some genes are involved in immunological processes, consistent with the previous finding that PDAC is associated with both local and systemic immune reactions.7 Further investigations to determine the biological significance of these genes will provide insight into the immunopathological features of PDAC, leading to further development of diagnostic tools as well as immunological treatments for the disease.

In conclusion, we developed a novel in vitro diagnostic system based on assessment of the host immune response to PDAC. As it remains unclear how this novel system should best be used as an adjunct to conventional PDAC diagnostic or screening systems, further investigations to improve the PDAC screening algorithm, including our novel screening system, should be undertaken to improve the prognosis of PDAC patients by diagnosis in the early stages of disease progression.

DISCLOSURE STATEMENT

This work was financially supported in part by KUBIX Inc., and was assessed by our institutional review board.

Supporting information

 

 

 

 

 

 

 

Sakai Y, Honda M, Matsui S, et al. Hokuriku Liver Study Group . Development of novel diagnostic system for pancreatic cancer, including early stages, measuring mRNA of whole blood cells. Cancer Sci. 2019;110:1364–1388. 10.1111/cas.13971