This study was performed at the Regional Institute of Gastroenterology and Hepatology “O. Fodor” in Cluj-Napoca, Romania. This study was prospectively performed and was approved by the Ethics Committee of the hospital and was registered at clinicaltrials.gov (NCT03042442).

Subjects of the study group were at least 18-years-old, with no previous history of any other cancer in the last 5 years. Written consent was given prior to entry into the study. Patients with pancreatic ductal adenocarcinoma, based on the results of endoscopic ultrasonography (EUS), biopsy, or surgery, were enrolled at the time of diagnosis, before any therapeutic intervention had been given, from January 2015 to September 2017.

The exclusion criteria were obvious malabsorption, major depression, artificial nutrition, hyperthyroidism, and other causes of malnutrition. The final diagnosis was based on the histologic results from endoscopic ultrasonography fine needle aspiration (EUS-FNA) or surgery.

The subjects of the control groups were healthy people who were at least 18-years-old, with no previous history of any cancer and other chronic diseases. For the most part, controls were matched to cases for sex and age (plus/minus 5 years). Age, sex, tumor stage, tumor differentiation, body-mass index (BMI), smoking, and the presence of diabetes were noted.

Current body weight and height were measured at the time of inclusion. Diabetes was diagnosed if fasting glucose values met the ADA criteria[23] and the duration since diabetes onset was recorded.

Cachexia was defined as an involuntary weight loss of more than 5% or a weight loss of more than 2% in individuals with a BMI of less than 20 kg/m² over the past 6 mo[24].

Blood samples were collected at the time of diagnosis. Peripheral venous blood was drawn into a tube containing ethylenediaminetetraacetic acid and was prepared by centrifugation at 5000 × g for 5 min. The plasma samples were stored at −80 °C until use.

Selected proteins were quantified in the plasma using Western blot analyses.

The following antibodies were obtained from Abcam (Cambridge, United Kingdom): rabbit polyclonal antibody to ACV Receptor Type IIB antibody (cat. no. ab128544), rabbit polyclonal antibody to GAPDH (glyceraldehyde 3-phosphate dehydrogenase) (cat. no. ab37168), and horseradish peroxidase (HRP) conjugated goat anti-rabbit IgG H.L antibody (cat. no. ab97051). The antibodies against midkine, mouse monoclonal IgG1 MK antibody (sc-46701, Santa Cruz Biotechnology, Santa Cruz, CA, United States) and goat anti-mouse IgG-HRP for MK (sc-2005; Santa Cruz Biotechnology, Santa Cruz, CA, United States) were obtained from Santa Cruz.

Protein concentration was determined using a protein assay kit, Quick Start Bradford Protein Assay (BioRad Laboratories, Inc.). A total of 50 μg of total protein from each plasma sample was loaded per lane onto a 12% polyacrylamide gel. Electrophoresis was performed at 120 mV, and the protein fractions were electrotransferred onto a nitrocellulose membrane at 100 mV for 1 h. The membranes were blocked for 3 h with nonfat dry milk powder (BioRad Laboratories, Inc.) in Trisbuffered saline containing 0.1% Tween20 (TBST) under constant agitation. Subsequently, the membranes were incubated overnight at 4 ºC with a rabbit polyclonal antibody to ACV Receptor Type IIB antibody (cat. no. ab128544AbCam) diluted 1: 1000 in TBST, with nonfat dry milk powder and a mouse monoclonal antibody to IgG1 MK antibody (sc-46701; Santa Cruz Biotechnology, Santa Cruz, CA, United States) diluted 1: 100 in TBST with nonfat dry milk powder. For the loading control, a rabbit polyclonal antibody to GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (cat. no. ab37168 AbCam) was used at a concentration of 0.8 µg/mL in nonfat dry milk powder in TBST. Membranes were washed with TBST and incubated at room temperature for 1 h with a horseradish peroxidase (HRP) conjugated goat anti-rabbit IgG H.L antibody (cat. no. ab97051, Abcam) diluted 10000fold in TBST and a goat anti-mouse IgG-HRP antibody for MK (sc-2005; Santa Cruz Biotechnology, Santa Cruz, CA, United States) diluted 5000fold in TBST, with an additional washing step performed prior to detection. For GAPDH determination, an HRPconjugated goat IgG antirabbit IgG antibody (cat. no. ab97051, Abcam) diluted 10000fold in TBST was used. Total protein expression was normalized by dividing each of the protein units by those for GAPDH.

Proteins were detected by using Clarity™ Western ECL (BioRad Laboratories, Inc.) and the membranes were imaged in a Chemidoc Imaging System (BioRad Laboratories, Inc.) and analyzed using ImageLab Software version 5.2.1 for Windows (BioRad Laboratories, Inc.).

Measurement and confirmation of the ACV and MK protein levels is often performed with normalization against “housekeeping proteins”, such as glyceraldehyde-3-phosphate dehydrogenase, to correct for protein loading and other factors, such as transfer efficiency.

For protein expression, after densitometry, the integrated density value (IDV) for each protein band (ACV and MK) was determined, and the normalized levels of ACV and MK were calculated by dividing the IDV of a protein band by the IDV of the GAPDH band (arbitrarily assigned a value of 100) in the same sample, thus quantifying the expression of the proteins as high or low expression.

Tumor tissue samples from endoscopic ultrasound-fine needle aspiration and surgery were fixed with 10% formalin for pathology studies.

In addition, there was a supplementary group of 14 samples containing normal pancreatic tissues from patients who received partial pancreatectomy for benign tumors that were used as normal controls for the immunohistochemical interpretation.

The perineural invasion was assessed only on surgical tissues. The associations of perineural invasion with characteristics of PDAC were assessed on slides stained with hematoxylin-eosin. The presence of perineural invasion was defined by the infiltration of cancer cells into the perineurium or neural fascicles[25]. The evaluation of perineural invasion was assessed mainly in tissue obtained from surgery after curative treatment. The degree of perineural invasion was defined as follows: 0- less than one occurrence per slide; 1 -two to four occurrences per slide; and 2 - more than four occurrences per slide or intraneural invasion.

For each case a paraffin-embedded block was selected. 3-micron-thick sections were obtained for immunohistochemistry investigation. The primary antibodies used were the following: ACV Receptor Type IIB antibody (cat. no. ab128544) diluted 1: 150 and mouse monoclonal IgG1 MK antibody (sc-46701, Santa Cruz Biotechnology, Santa Cruz, CA, United States) diluted 1: 200. The BOND‐III staining instrument (Leica Biosystems) and Bond Polymer Refine Detection Kit (Leica Biosystems) were used for all antibodies. This analysis was performed with pancreatic tumor and normal tissue samples.

All slides were scored by a pathologist (I.R.) who was blinded to all clinical data. Finally, the tissues were evaluated under a microscope. The intensity of the staining was scored as negative, weak, moderate, or strong (scores of 0, 1, 2, or 3, respectively).

Chi square test or Fisher’s exact test were used for categorical data. Comparisons between two groups of continuous data were performed with a t-test for independent samples for data with a normal distribution or a Wilcoxon rank-sum test otherwise. Univariate and, multivariate Cox proportional hazard models with each protein expression variable adjusted for age ≥ 50 years, gender, stage (T4 vs T1-T3), metastasis, tumor size ≥ 3 cm, and diabetes were built. The Cox proportional hazard assumption and multicollinearity assumptions were checked. Similarly, univariate and multivariate logistic regression models were built to predict metastasis. We checked the models for multicollinearity, misspecification and the goodness of fit.

For all statistical tests, a two-tailed P value was used, along with a 0.05 significance level. All analyses were performed in the R environment for statistical computing and graphics, version 3.4.4.

A total of 114 patients with PDAC and 125 controls with no tumor were enrolled in this study (Table 1).

The mean age of the population was 62.41 years (SD 11.56, range: 27-88 years). There were more males than females (61% vs 39%).

A total of 49 (42.98%) patients with PDAC reported a history of diabetes, including new-onset diabetes for 24 (21.05%) patients.

The most common localization of PDAC was in the pancreatic head (66 patients-49%), followed by a localization in the body (33 patients-24%), the isthmus (17 patients-12%), the pancreatic tail (13 patients-10%) and the uncinated process (7 patients-12%).

Cachexia was more common in women (P = 0.033) and in patients with metastasis (P = 0.011). The presence of cachexia was unrelated to age (P = 0.389), cancer site (P = 0.611), tumor staging (P = 0.148) and perineural invasion (P = 0.12).

MK was expressed in plasma from 54 (47.37%) PDAC patients compared to 20 (16%) controls (P < 0.001). Immunohistochemical staining in PDAC tissue was positive in 56 (49.12%) patients, including weak in 34 (29.82%), moderate in 16 (14.04%), and strong in 6 (5.26%), compared to none of the tissue from the supplementary controls (P = 0.012) (Figure 1).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Immunohistochemical staining of Midkine in pancreatic adenocarcinomaand normal pancreatic tissue. A: Negative Midkine staining in normal pancreatic tissue; B: Negative Midkine staining in pancreatic tumor tissue; C: Weak Midkine expression in pancreatic tumor tissue; D: Moderate Midkine expression in pancreatic tumor tissue; E: Strong Midkine expression in pancreatic tumor tissue.

ACV was expressed in plasma from 72 (63.16%) PDAC patients compared to 39 (31.2%) controls (P < 0.001). Immunohistochemical staining in PDAC tissue was positive in 59 patients (51.8%), including weak in 53 (46.49%) and moderate in 6 (5.26%) compared to none of the tissue from the control group (P = 0.001) (Figure 2).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Immunohistochemical staining of Activin in pancreatic adenocarcinoma and normal pancreatic tissue. A: Negative Activin staining in normal pancreatic tissue; B: Weak Activin expression in pancreatic tumor tissue; C: Moderate Activin expression in pancreatic tumor tissue; D: Strong Activin expression in pancreatic tumor tissue.

Both proteins were detected in the tumor tissue by immunohistochemistry, and the tissue expression was significantly correlated with the plasma expression (P = 0.002 for Activin; and P = 0.046 for Midkine).

Table 2 summarizes the associations of MK and ACV expression with clinicopathological features in pancreatic cancer.

MK was highly expressed in the plasma of patients with an advanced T tumor stage (P = 0.006), metastases (P = 0.046), long-term diabetes (P = 0.002) and nonsmokers (P = 0.032). Nevertheless, there was no relationship between clinicopathological factors and high ACV expression. The association between ACV expression and cachexia was not statistically significant (P = 0.5).

The presence of perineural invasion was observed in only 61 PDAC specimens. The MK expression levels in PDAC were significantly higher in patients with perineural invasion than in those without perineural invasion (P = 0.033). The ACV expression level was not significantly associated with perineural invasion (P = 0.5).

To predict the presence of metastasis in patients with adenocarcinoma, we performed uni- and multivariate analysis by creating regression models, where each expression variable (MK and ACV high expression vs low expression) in the model was adjusted by age > 50 years, sex, diabetes, stage T4 vs T1-3, stage N1 vs N0 (Table 3). There were no statistically significant factors.

Using data from the 114 patients with PDAC, we further evaluated whether the overexpression of ACV and MK was correlated with patient survival.

The univariate and multivariate Cox proportional survival analysis showed that after adjusting for age, gender, adenopathy, tumor size ≥ 3 cm, and metastases, tumor size > 3 cm, metastasis and ACV expression (Table 4) were independent predictors of poor survival (Figure 3). The survival of patients with high or low levels of MK expression was similar (Figure 4).

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Overall survival comparison between high and low-expressed Activin stratified by tumoral stage T1-3 vs T4.

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Overall survival comparison between high and low-expressed Midkine stratified by tumoral stage T1-3 vs T4.

Pancreatic cancer has a high mortality rate, due to late diagnosis. Cachexia and perineural invasion have an increased incidence in pancreatic cancer, leading to decreased quality of life. Combating pain and cachexia through optimal treatment options can lead to increased quality of life and improved survival.

We wanted a better understanding of the involvement of cachexia and pain in pancreatic cancer and their relationship with different clinico-pathological factors, constituting a basis for discovering new therapeutic targets.

Defining the profile of cachexia in pancreatic cancer; establishing the degree of perineural invasion in pancreatic cancer; Highlighting the interrelationship between cachexia and perineural invasion in pancreatic cancer.

We conducted a prospective study in 114 patients with pancreatic cancer and 125 healthy people as controls. Blood samples were used and pancreatic tissue were collected through EUS-FNA or surgery. The method of determining the biomarkers of cachexia (Activin) and perineural invasion (Midkine) in plasma was western blot, respectively immunohistochemistry in pancreatic tissue.

The analysis of the data showed an Activin (ACV) and Midkine (MK) proteins overexpression in plasma of patients with pancreatic cancer vs control, results that were correlated with the expression of proteins in the pancreatic cancer tissue. MK was also significantly correlated with advanced T stage, metastasis, diabetes and perineural invasion. ACV was significantly correlated with survival.

MK can be considered a biomarker for perineural invasion, and ACV a prognostic factor for patients with pancreatic cancer.

This research would open new perspectives in choosing the treatment that involves activin antagonists in order to prolong survival in patients with pancreatic cancer.