The Investigation of the Correlation between Urine Biomarkers and

Pancreatic Ductal Adenocarcinoma

Jiahui Shen

Institute of Problem Solving, Beijing Normal University Hong Kong Baptist University United International College,

Zhuhai, Guangdong, China

Keywords: Pancreatic Ductal Aadenocarcinoma, Urine Biomarkers, Multinomial Logistic Regression Model, Prediction

Of Cancer.

Abstract: Pancreatic ductal adenocarcinoma (PDAC)'s low survival rate has long been a world unsolved problem. Many

past studies in recent decades proved the possibility of detecting such disease in its early stages, using a new

screening panel with several urinary biomarkers. However, limited studies truly focused on the statistical

correlation between the fluctuation of urinary biomarker concentrations and PDAC diagnosis status. Our study

sought to demonstrate a possible correlation between biomarker concentration values in urine samples and

confirmed cases of PDAC that could be used for the early diagnosis of PDAC patients. Based on the

correlation of the different biomarker measurements with our investigation, we obtained data from Kaggle

originally from an open access paper. We estimated odds ratios (ORs) and 95% CIs in a multinomial logistic

regression model. From the analysis of p-value, LYVE1, REG1B, and TFF1 are all possible biomarkers to

indicate a patient's PDAC status. Multinomial logistic regression was made to show the correlation between

selected biomarkers and diagnosis. Our study suggested that a possible real correlation exists between urinary

biomarkers' concentration and PDAC diagnosis status. Our model could be used to detect patients in their

early disease stages to some degree.

1 INTRODUCTION

Pancreatic ductal adenocarcinoma, also known as

PDAC, is the most common malignant tumor of the

pancreas. It arises from cells in the ducts or ducts of

the pancreas, hence its name. PDAC has a low

survival rate of about 9% at 5 years and is one of the

deadliest cancers in the world. Initially, the tumor

may not show any signs or symptoms. However, over

time, it may cause abdominal pain, nausea, and

vomiting, and lead to weight loss and, in most cases,

complications that eventually lead to a person's death.

In terms of today's medical research developments,

complete surgical removal of the tumor is the only

chance to cure PDAC. If the disease is detected early,

the 5-year survival rate can be increased to 70% when

the tumor is still small and resectable. However,

because pancreatic ductal adenocarcinoma is difficult

to detect at an early stage, many patients are already

at an advanced stage of cancer when diagnosed and

https://orcid.org/0000-0003-0924-1786

the disease is already difficult to cure. Therefore,

finding a detection method for early PDAC is an

important clinical need, which may greatly improve

the survival chances of patients.

Since the first risk prediction model for coronary

heart disease was introduced in 1976, prediction

models for various diseases, including cancer, have in

the intervening decades, several tests for Pancreatic

ductal adenocarcinoma have emerged, and the

methods can be broadly classified into two

categories. The first type is based on the use of

imaging, where patients can be distinguished from

pancreatic ductal adenocarcinoma by the radiomics

score (rad-score) using multidetector computed

tomography (MDCT), which distinguishes focal-type

autoimmune pancreatitis (fAIP) from pancreatic

ductal adenocarcinoma (MDCT). adenocarcinoma

(PDAC) (Li et al. 2021). The second category is the

use of various biomarkers to discriminate PDAC

from benign pancreatic disease and healthy

individuals. the source of most biomarkers is blood,

318

Shen, J.

The Investigation of the Correlation between Urine Biomarkers and Pancreatic Ductal Adenocarcinoma.

DOI: 10.5220/0011368200003444

In Proceedings of the 2nd Conference on Artiﬁcial Intelligence and Healthcare (CAIH 2021), pages 318-325

ISBN: 978-989-758-594-4

for example, CA 19-9 and CEA in serum can be used

as detection markers (Poruk et al. 2013). Mayerle, J.

& Kalthoff, H. et al. generated metabolomic

profiles of plasma and serum samples by gas

chromatography-mass spectrometry and liquid

chromatography-tandem mass spectrometry

identification of 477 metabolites and selected nine

plasma metabolites, all of which could be identified

to distinguish pancreatic cancer from chronic

pancreatitis. Serum mRNA for LGALS9 (Galectin-9)

was detected to be highly expressed in PDAC patients

compared to the normal pancreas and could be used

as a new assay biomarker in addition to prognostic for

stage IV patients (Seifert et al. 2020). With medical

developments in the assay new blood testing

techniques with nanoparticles are added, which are

expected to distinguish PDAC patients from healthy

individuals (Caputo & Caracciolo 2020). For the

detection of the resectable phase of PDAC, the exlr

signal of PDAC can be detected by analyzing plasma

extracellular vesicle long RNA, reported (Yu et al.

2020). In addition to biomarkers in blood, there are

volatile organic compounds (VOCs) in alveolar air

(Princivalle et al., 2018) and MicroRNA (MiR) in

pancreatic fluid (Nakamura et al. 2019) both with

good sensitivity and specificity for pancreatic tumors.

One more source of biomarkers in urine, and the urine

biomarker set, including LYVE1, REG1A, and TFF1,

has been shown to be effective in the early detection

of PDAC in studies as early as 2015 (Radon et al.,

2015). In a recent study, Sahni, S. and Pandya, A. R.

et al. used a non-targeted urine metabolic panel to

identify novel metabolite biomarker profiles for

PDAC diagnosis, and six metabolites were screened

and showed very high potential in the detection and

diagnosis of PDAC in both early (stages I and II) and

late (stages III and IV) patients.

In contrast to previously published articles on the

detection of Pancreatic ductal adenocarcinoma, this

study changes the detection of PDAC from traditional

imaging methods (e.g., CT) or blood markers to urine

biomarkers. While imaging methods are expensive

and require training of dedicated personnel for

testing, while many biomarkers in the blood (e.g.,

CA19-9 serum test) can be used to diagnose PDAC,

urine instead of blood allows for completely non-

invasive sampling, high volume collection, and easily

repeatable measurements, with a smaller dynamic

range and less complex proteome than blood. In

addition, continuous ultrafiltration of blood is

expected to result in the accumulation of at least some

biomarkers in the urine, leading to higher

concentrations. Therefore, sensitivity, specificity,

positive and negative predictive values are superior to

conventional methods. Second, previous studies have

used urine biomarkers to detect PDAC, and they used

REG1A in urine as one of the biomarkers. expression

of REG1A increased with the progression of PanINs

to cancer, but REG1B was highly expressed in the

earliest PanINs, showing a better difference. Despite

their similar performance, this study confirmed that

REG1B was superior to REG1a in comparisons

between control samples and stage I-IIA PDAC

samples. therefore, all subsequent experiments in this

study used REG1B as a component of the biomarker

set.

In this paper, A secondary data analysis is

conducted to study the statistical correlation between

four urine bio-markers and Pancreatic ductal

adenocarcinoma, and to predict whether this patient

has pancreatic problems and determine whether he

has pancreatic cancer at an early stage. First, the type

of variable was determined after obtaining the data

and corresponding to obtaining the p-value.

Continuous variables were assessed using ANOVA,

while categorical variables were tested with the χ

test. Then a polynomial logistic regression model was

built to come and correlations were assessed by the

above.

2 DATA SOURCE

The data was obtained from Kaggle. It is a

community that allows users to find and publish

datasets, explore and build models in a web-based

data science environment. It is also a community

where one can collaborate with other data scientists

and machine learning engineers as well as participate

in competitions to solve data science challenges. The

data was uploaded to Kaggle by John Davis and the

data was initially derived from an open-access paper

by Silvana Debernardi and a colleague in PLoS

medicine published on December 10, 2020.We

selected this secondary data based on the relevance of

the different biomarker measurements to our study

and the number of reported NAs. These clinical

specimens come from multiple centers, such as Barts

Pancreas Tissue Bank（BPTB), University College

London (UCL), University of Liverpool (LIV),

Spanish National Cancer Research Center (ESP), the

University of Cambridge Hospital, and the University

of Belgrade. All samples were collected prior to

surgery or chemotherapy treatment and were

potentially age and sex matched. The data including

a total of 590 biomarker panels tested on urine

samples, 332 of which were collected in 2013 by

Vanessa W and colleagues to study the association

The Investigation of the Correlation between Urine Biomarkers and Pancreatic Ductal Adenocarcinoma

319

between pancreatic ductal adenocarcinoma and

urinary metabolic features, and the latter 258 samples

were collected by Debernardi and colleagues

collected additional samples at the time of the study.

The majority of samples were from BPTB and LIV

with 409 and 132 samples, respectively, with the

remaining 8% of samples originating from other

centers. The first category was 183 control samples,

with no known pancreatic disease or malignancy

confirmed. The second category was benign disease

samples, which included 119 cases of chronic

pancreatitis, 54 cases of gallbladder disease, 20 cystic

lesions of the pancreas, and 15 cases with abdominal

pain and gastrointestinal symptoms, for a total of 208

samples. The remaining 199 were all PDAC patients.

The male to female ratio remained essentially 1:1 at

291 and 299 respectively, and the mean age of the

sample is 59.1 years old.

3 RESEARCH VARIABLES

The pancreas plays a vital role in exocrine function,

helping digest protein, cholesterol, and fat. Pancreatic

cancer can severely impair the normal function of the

pancreas. Therefore, we selected four biomarkers

from urine biomarkers that are closely related to

pancreas and used their values as independent

variables in this study. The following are the main

characteristics of the four urine biomarkers and they

are all continuous variables. Variable，creatinine, as

a urinary biomarker of kidney function, is a protein

often used as an indicator of kidney function. In

patients with PDAC, decreased protein digestion may

lead to increased urinary creatinine production.

Variable, lymphatic vessel endothelial hyaluronan

receptor 1 (LYVE1) is a protein that may play a role

in tumor metastasis, growing tumors may produce

large amounts of YVLE1 for cell metastasis.

Variable, REG1B stand for regenerating family

member 1 beta is a Protein Coding gene. It may be

associated with pancreas regeneration; damaged

pancreas tissue may release large amounts of REG1B

during regeneration. TFF1 is trefoil factor 1, serves

as a variable that may be associated with regeneration

and repair of the urinary tract. Increasing TFF1 in the

gastrointestinal mucosa can help repair the damaged

digestive tract. In addition, age and gender were also

independent variables in this study. Age was a

continuous variable and gender was used as a

categorical variable, with M for male and F for

female.

The dependent variable in the study is diagnosis,

which is a categorical variable. There are three

diagnostic classifications, 1 represents control (no

pancreatic disease), 2 stands for benign hepatobiliary

disease (119 of which are chronic pancreatitis), and 3

for pancreatic ductal adenocarcinoma, i.e., pancreatic

cancer. Through the correlation between these four

urine biomarkers and the patient's diagnosis and data

analysis, the early diagnosis of PDAC will be more

accurate.

Table 1: Summary of Data Collected, including the type of independent variable, the number of independent variables in the

dependent variable category, the mean, standard deviation and median, and the relative p-value of each independent variables.

Type no pancreatic

disease

(N=183)

benign

hepatobiliary

disease

(N=208)

Pancreatic ductal

adenocarcinoma

(N=199)

P-value

Age (years)

Mean (SD)

Median [Min, Max]

Continuous

56.3 (12.2)

57.0 [26.0, 89.0]

54.7 (13.3)

54.0 [26.0, 82.0]

66.2 (10.5)

67.0 [29.0, 88.0]

<0.001

Gender

Female

Male

Categorical

115 (62.8%)

68 (37.2%)

101 (48.6%)

107 (51.4%)

83 (41.7%)

116 (58.3%)

<0.001

Creatinine

Mean (SD)

Median [Min, Max]

Continuous

0.798 (0.559)

0.713 [0.0679,

3.45]

0.848 (0.616)

0.746 [0.0566,

3.34]

0.916 (0.724)

0.724 [0.0792,

4.12]

0.189

Lymphatic vessel

endothelial hyaluronan

receptor 1 (LYVE1)

Mean (SD)

Median [Min, Max]

Continuous

1.21 (1.92)

0.146 [0.000129,

8.32]

2.08 (2.37)

1.21 [0.000226,

11.0]

5.79 (3.78)

5.62 [0.00127,

23.9]

<0.001

CAIH 2021 - Conference on Artiﬁcial Intelligence and Healthcare

320

Regenerating family

member 1 beta (REG1B)

Mean (SD)

Median [Min, Max]

Continuous

41.3 (61.9)

17.6 [0.00110,

544]

64.2 (116)

20.2 [0.00280,

864]

226 (277)

23 [1.65,

1400]

<0.001

Trefoil Factor 1 (TFF1)

Mean (SD)

Median [Min, Max]

Continuous

169 (278)

59.8 [0.00529,

1880]

448 (646)

210 [0.0132,

4460]

1150 (1430)

723 [0.0212,

13300]

<0.001

4 STATISTICAL ANALYSIS

We performed a preliminary processing of the data.

As shown in Table 1, continuous variables are

expressed as mean (SD), while categorical variables

are reported as absolute numbers and percentages.

Firstly, the correlation between the number of

pancreatic disease-healthy, benign hepatobiliary

disease-noncancerous and PDAC—pancreatic

cancer) is evaluated by ANOVA test while

categorical variables like age and sex are done by χ

test. The level of significance, α was set at 0.05.

Next, we produced density plots of biomarkers

and then we performed logarithmic transformation on

the four urine biomarkers. At last, we established a

multinomial logistic regression model to predict

outcomes as we have more than 2 categorical

outcomes (healthy, benign, PDAC) that cannot be put

into meaningful orders.

Multinomial Logistic regression analysis was

used to estimate the odds ratio (OR) and 95%

confidence interval (95% CIs) are calculated. A 2-

sided P-value less than 0.05 was considered

significant. Data management and statistical analyses

were performed using R, version 4.1.1.

Figure 1: The process of data analysis.

The Investigation of the Correlation between Urine Biomarkers and Pancreatic Ductal Adenocarcinoma

321

5 RESULTS

According to table-1, P-value for creatinine is 0.189,

which is higher than the level of significance. This

indicates that there is no significant statistical

correlation between levels of creatinine and diagnosis

(cancerous, non-cancerous condition, healthy).

Therefore, we will remove creatinine in further

studies and continue to investigate the correlation

between the remaining 5 independent variables and

diagnosis. The p-value for age, sex, LYVE1, REG1B,

and TFF1 all have p-value way smaller than the alpha

level, indicating there is a correlation between these

biomarkers and confounders and diagnosis.

LYVE1's mean and median are slightly different

while the other 2 biomarkers having a relatively large

mean median difference and SD, indicating that

LYVE1 is less likely to be affected by individual

differences. The large difference will be discussed

further in the discussion session.

We conclude that LYVE1 may be impacted more

by PDAC than the other 2, but since all 3 biomarkers

show statistical correlation with the diagnosis, we

should use 3 together when making predictions on a

patient's health conditions. Through the density plot

of 3 urine biomarkers, it was found that the

distributions of the four urine biomarkers are all

lognormal distributions.

Figure 2: Diagrams of Density of 3 Urine Biomarkers.

CAIH 2021 - Conference on Artiﬁcial Intelligence and Healthcare

322

There are also a noticeable number of outliers for

samples under all 3 diagnoses, as can be seen among

3 box plots. Because we're investigating the possible

correlation between biomarkers and PDAC thus no

patients' samples, regardless of their underlying

health conditions, should be ignored.

Figure 3: Mean level of different urine biomarkers displayed by different sexes under 3 diagnoses.

The Investigation of the Correlation between Urine Biomarkers and Pancreatic Ductal Adenocarcinoma

323

As represented by TFF1 and REG1B in the box

plot (Figure 3), gender differences were shown under

the same diagnosis: male patients showed higher

levels of the 2 biomarkers than females, while both

sexes showed an increasing trend in the levels of the

2 biomarkers at the diagnosis of PDAC.

We also observed an increasing trend for LYVE1

when patients were diagnosed with PDAC. However,

unlike the other 2 biomarkers, the number of

extremes for this marker decreased when diagnosed

with PDAC.

We also noticed that for REG1B, dispersion of

data increased for both females and males when

diagnosed with PDAC while the dispersion for

creatinine remain similarly across diagnosis and

gender. Therefore, we assume that these data may not

follow a normal distribution and need to use a

different approach.

LYVE1 is also a significant indicator for health

(OR: 1.202, 95% CI: 1.099-1.315) and PDAC (OR:

2.418, 95% CI: 1.874-3.118). REG1B was identified

as a significant risk factor for health (OR: 0.834, 95%

CI: 0.716–0.973) and PDAC (OR: 1.247, 95% CI:

1.004–1.550). TFF1 was identified as a significant

risk factor for health (OR: 1.187, 95% CI: 1.080–

1.305) and PDAC (OR: 1.166, 95% CI: 1.013–1.343).

Formula 1: Summary of Multinomial Regression Model

Used.

6 DISCUSSION

Even though we concluded from our analysis on data

that LYVE 1 impacted most by PDAC, we should

still consider using all 3 biomarkers when analyzing

urine samples collected from clinics. The differences

of mean and standard deviation for REG1B and TFF1

among different diagnoses, according to several past

studies that also used urine biomarkers, were a likely

outcome of patients' other health conditions. Also, as

we discovered that the data obtained followed a log-

normal distribution, thus the extreme values we

thought to be outliers are normal under such

contribution.

LYVE1 itself was discovered to be a protein that

played a role in the autocrine regulation of cell

growth and tumor metastasis; in the meantime, the

other 2 biomarkers are more associated with other

organs and tissues. These urine biomarkers are not

unique to PDAC, thus could be affected by cancer or

disease. For instance, TFF1 is an indicator of urinary

canal's self-repair, but also present in normal breast

tissues; thus, situations like a male patient with

prostate carcinoma, a prostate cancer, near the male

urinary canal, may have a significantly higher level

of TFF1 than other patients. In the meantime, breast

cancer in women can also significantly increase the

TFF1 levels shown by a 2017 Japanese study. Serum

TFF1 and TFF3 but not TFF2 is higher in women

with breast cancer than in women without breast

cancer. Thus, the great differences in levels of TFF1

among individuals under the same pancreatic

diagnosis may not be due to PDAC we invested in but

other cancers not indicated in the study when samples

were collected. Therefore, the 3 biomarkers (L, R,

and T) should be analyzed together when testing and

predicting the diagnosis of PDAC for susceptible

patients in the clinic. Future studies could focus on

distinguishing between biomarkers or find

biomarkers that are uniquely correlated to PDAC

when making predictions in the clinics.

7 CONCLUSIONS

In summary, we performed secondary data analysis

with data obtained on Kaggle, including categorical

variables, chi-square test, and estimated dominance

ratio (OR) using multinomial logistic regression

analysis and calculated 95% confidence intervals

(95% CI). Ultimately, a correlation between the

number of biomarkers in urine (continuous variable)

and different diagnoses (no pancreatic disease health,

benign hepatobiliary disease non-cancer, and PDAC

pancreatic cancer) was successfully demonstrated

and urine biomarkers (LYVE1, REG1B, and TFF1)

could be used to screen for no pancreatic disease,

benign The urine biomarkers (LYVE1, REG1B, and

TFF1) can be used to screen for no pancreatic disease,

benign hepatobiliary disease, and pancreatic ductal

adenocarcinoma, cancer, providing a completely non-

invasive and convenient method for detecting PDAC.

REFERENCES

Abdou, A. G., Aiad, H. A., & Sultan, S. M. (2008). pS2

(TFF1) expression in prostate carcinoma: correlation

with steroid receptor status. APMIS: acta pathologica,

microbiologica, et immunologica Scandinavica,

CAIH 2021 - Conference on Artiﬁcial Intelligence and Healthcare

324

116(11), 961–971. https://doi.org/10.1111/j.1600-

0463.2008.01009.x

Caputo, D., & Caracciolo, G. (2020). Nanoparticle-enabled

blood tests for early detection of pancreatic ductal

adenocarcinoma. Cancer letters, 470, 191–196.

https://doi.org/10.1016/j.canlet.2019.11.030

Davis, V. W., Schiller, D. E., Eurich, D., Bathe, O. F., &

Sawyer, M. B. (2013). Pancreatic ductal

adenocarcinoma is associated with a distinct urinary

metabolomic signature. Annals of surgical oncology,

20(3), S415–S423. https://doi.org/10.1245/s10434-

012-2686-7

Debernardi, S., O'Brien, H., Algahmdi, A. S., Malats, N.,

Stewart, G. D., Plješa-Ercegovac, M., Costello, E.,

Greenhalf, W., Saad, A., Roberts, R., Ney, A., Pereira,

S. P., Kocher, H. M., Duffy, S., Blyuss, O., &

Crnogorac-Jurcevic, T. (2020). A combination of

urinary biomarker panel and PancRISK score for earlier

detection of pancreatic cancer: A case-control study.

PLoS medicine, 17(12), e1003489.

https://doi.org/10.1371/journal.pmed.1003489

Ishibashi, Y., Ohtsu, H., Ikemura, M., Kikuchi, Y., Niwa,

T., Nishioka, K., Uchida, Y., Miura, H., Aikou, S.,

Gunji, T., Matsuhashi, N., Ohmoto, Y., Sasaki, T., Seto,

Y., Ogawa, T., Tada, K., & Nomura, S. (2017). Serum

TFF1 and TFF3 but not TFF2 are higher in women with

breast cancer than in women without breast cancer.

Scientific reports, 7(1), 4846.

https://doi.org/10.1038/s41598-017-05129-y

Li, J., Liu, F., Fang, X., Cao, K., Meng, Y., Zhang, H., Yu,

J., Feng, X., Li, Q., Liu, Y., Wang, L., Jiang, H., Shao,

C., Lu, J., & Bian, Y. (2021). CT Radiomics Features

in Differentiation of Focal-Type Autoimmune

Pancreatitis from Pancreatic Ductal Adenocarcinoma:

A Propensity Score Analysis. Academic radiology,

S1076-6332(21)00209-9. Advance online publication.

https://doi.org/10.1016/j.acra.2021.04.014

Mayerle, J., Kalthoff, H., Reszka, R., Kamlage, B., Peter,

E., Schniewind, B., González Maldonado, S., Pilarsky,

C., Heidecke, C. D., Schatz, P., Distler, M., Scheiber, J.

A., Mahajan, U. M., Weiss, F. U., Grützmann, R., &

Lerch, M. M. (2018). Metabolic biomarker signature to

differentiate pancreatic ductal adenocarcinoma from

chronic pancreatitis. Gut, 67(1), 128–137.

https://doi.org/10.1136/gutjnl-2016-312432

Nakamura, S., Sadakari, Y., Ohtsuka, T., Okayama, T.,

Nakashima, Y., Gotoh, Y., Saeki, K., Mori, Y., Nakata,

K., Miyasaka, Y., Onishi, H., Oda, Y., Goggins, M., &

Nakamura, M. (2019). Pancreatic Juice Exosomal

MicroRNAs as Biomarkers for Detection of Pancreatic

Ductal Adenocarcinoma. Annals of surgical oncology,

26(7), 2104–2111. https://doi.org/10.1245/s10434-

019-07269-z

Poruk, K. E., Gay, D. Z., Brown, K., Mulvihill, J. D.,

Boucher, K. M., Scaife, C. L., Firpo, M. A., &

Mulvihill, S. J. (2013). The clinical utility of CA 19-9

in pancreatic adenocarcinoma: diagnostic and

prognostic updates. Current molecular medicine, 13(3),

340–351.

https://doi.org/10.2174/1566524011313030003

Princivalle, A., Monasta, L., Butturini, G., Bassi, C., &

Perbellini, L. (2018). Pancreatic ductal

adenocarcinoma can be detected by analysis of volatile

organic compounds (VOCs) in alveolar air. BMC

cancer, 18(1), 529. https://doi.org/10.1186/s12885-

018-4452-0

Radon, T. P., Massat, N. J., Jones, R., Alrawashdeh, W.,

Dumartin, L., Ennis, D., Duffy, S. W., Kocher, H. M.,

Pereira, S. P., Guarner posthumous, L., Murta-

Nascimento, C., Real, F. X., Malats, N., Neoptolemos,

J., Costello, E., Greenhalf, W., Lemoine, N. R., &

Crnogorac-Jurcevic, T. (2015). Identification of a

Three-Biomarker Panel in Urine for Early Detection of

Pancreatic Adenocarcinoma. Clinical cancer research:

an official journal of the American Association for

Cancer Research, 21(15), 3512–3521.

https://doi.org/10.1158/1078-0432.CCR-14-2467

Sahni, S., Pandya, A. R., Hadden, W. J., Nahm, C. B.,

Maloney, S., Cook, V., Toft, J. A., Wilkinson-White,

L., Gill, A. J., Samra, J. S., Dona, A., & Mittal, A.

(2021). A unique urinary metabolomic signature for the

detection of pancreatic ductal adenocarcinoma.

International Journal of Cancer, 148(6), 1508.

https://doi.org/10.1002/ijc.33368

Seifert, A. M., Reiche, C., Heiduk, M., Tannert, A.,

Meinecke, A. C., Baier, S., von Renesse, J., Kahlert, C.,

Distler, M., Welsch, T., Reissfelder, C., Aust, D. E.,

Miller, G., Weitz, J., & Seifert, L. (2020). Detection of

pancreatic ductal adenocarcinoma with galectin-9

serum levels. Oncogene, 39(15), 3102–3113.

https://doi.org/10.1038/s41388-020-1186-7

Yu, S., Li, Y., Liao, Z., Wang, Z., Wang, Z., Li, Y., Qian,

L., Zhao, J., Zong, H., Kang, B., Zou, W. B., Chen, K.,

He, X., Meng, Z., Chen, Z., Huang, S., & Wang, P.

(2020). Plasma extracellular vesicle long RNA

profiling identifies a diagnostic signature for the

detection of pancreatic ductal adenocarcinoma. Gut,

69(3), 540–550. https://doi.org/10.1136/gutjnl-2019-

318860

The Investigation of the Correlation between Urine Biomarkers and Pancreatic Ductal Adenocarcinoma

325