A RATIO BETWEEN PLGF AND SFLT1 IS PREDICTIVE FOR NEURAL INVASION IN PATIENTS SUFFERING FROM PANCREATIC CANCER

Abstract

The present invention is directed to a method for identifying neural invasion in a pancreatic cancer patient, the method comprising the steps of: determining in a sample of a pancreatic cancer patient the expression level of PlGF and sFlt1; and calculating a ratio of the expression levels of PlGF and sFlt1, wherein the expression level of PlGF forms part either of the numerator or of the denominator of the ratio; characterized in that, a deviation of the ratio of the patient sample from a reference sample or a predetermined threshold value is indicative for presence of neural invasion of the pancreatic cancer in said patient. Preferably said deviation, when the ratio is expressed using the expression level of PlGF in the numerator, is an increase in the ratio of the patient sample compared to a reference sample or a predetermined threshold value and/or said deviation, when the ratio is expressed using the expression level of PlGF in the denominator, is a decrease in the ratio of the patient sample compared to a reference sample or a predetermined threshold value.

Claims

1. A method for identifying neural invasion in a pancreatic cancer patient, the method comprising the steps of: determining in a sample of a pancreatic cancer patient the expression level of P1GF and sFlt1; and calculating a ratio of the expression levels of P1GF and sFlt1, wherein the expression level of P1GF forms part either of the numerator or of the denominator of the ratio; characterized in that, a deviation of the ratio of the patient sample from a reference sample or a predetermined threshold value is indicative for presence of neural invasion of the pancreatic cancer in said patient.

2. The method according to claim 1, wherein the deviation, when the ratio is expressed using the expression level of P1GF in the numerator, is an increase in the ratio of the patient sample compared to a reference sample or a predetermined threshold value and/or wherein the deviation, when the ratio is expressed using the expression level of P1GF in the denominator, is a decrease in the ratio of the patient sample compared to a reference sample or a predetermined threshold value.

3. The method according to claim 1, wherein the ratio between the expression levels of P1GF and sFlt1 is calculated using the metric [P1GF/sFlt1] or the metric [sFlt1/P1GF].

4. The method according to claim 1, wherein, when the ratio is expressed using the expression level of P1GF in the numerator, the predetermined threshold value is selected from the range of 0.15 to 0.35, preferably from the range of 0.2 to 0.3, more preferably from the range of 0.22 to 0.26.

5. The method according to claim 1, wherein, when the ratio is expressed using the expression level of P1GF in the denominator, the predetermined threshold value is selected from the range of 6.67 to 2.86, preferably from the range of 5 to 3.33, more preferably from the range of 4.55 to 3.85.

6. The method according to claim 1, wherein the sample of the pancreatic cancer patient is or is derived from a tissue sample or a body fluid, preferably is or is derived from a body fluid, more preferably is or is derived from serum, plasma and/or urine.

7. The method according to claim 1, wherein the expression levels of P1GF and sFlt1 are determined using PCR or an immunological method, preferably the immunological method is an ELISA.

8. The method according to claim 1, wherein the pancreatic cancer patient has not yet undergone partial or full resection of said pancreatic cancer.

9. A use of a ratio of the expression levels of P1GF and sFlt1 in a sample of a patient suffering from pancreatic cancer for identification or diagnosis of neural invasion in said patient.

10. The use of the method according to claim 1 or a ratio of the expression levels of P1GF and sFlt1 in a sample of a patient suffering from pancreatic cancer for assigning a pancreatic cancer patient to a specific therapy, preferably the specific therapy is selected from partial or full tumour resection, neoadjuvant chemo and/or radiation therapy, adjuvant chemo and/or radiation therapy and/or palliative therapy including palliative pain therapy.

11. The use of a method according to claim 1 or a ratio of the expression levels of P1GF and sFlt1 in a sample of a patient suffering from pancreatic cancer for disease prognosis for a pancreatic cancer patient, wherein, when the ratio is expressed using the expression level of P1GF in the numerator, the higher the ratio between expression levels of P1GF and sFlt1 for the pancreatic cancer patient is above the reference sample or threshold value, the less favourable the disease prognosis is for this patient, or, when the ratio is expressed using the expression level of P1GF in the denominator, the lower the ratio between expression levels of P1GF and sFlt1 for the pancreatic cancer patient is below the reference sample or threshold value, the less favourable the disease prognosis is for this patient.

12. A method of predicting responsiveness of a pancreatic cancer patient to an anti-P1GF therapy, the method comprising the steps of a method according to claim 1, wherein, when the ratio is expressed using the expression level of P1GF in the numerator, an increase in the ratio between expression levels of P1GF and sFlt1 above the reference sample or threshold value is indicative of an increase in likeliness that this patient will benefit from anti-P1GF therapy, or, when the ratio is expressed using the expression level of P1GF in the denominator, a decrease in the ratio between expression levels of P1GF and sFlt1 below the reference sample or threshold value is indicative of an increase in likeliness that this patient will benefit from anti-P1GF therapy.

13. A method of predicting whether a pancreatic cancer patient will benefit from partial or full tumour resection, the method comprising the steps of a method according to claim 1, wherein, when the ratio is expressed using the expression level of P1GF in the numerator, the higher the ratio between expression levels of P1GF and sFlt1 for the pancreatic cancer patient is above the reference sample or threshold value, the less likely this patient will benefit from partial or full tumour resection, or, when the ratio is expressed using the expression level of P1GF in the denominator, the lower the ratio between expression levels of P1GF and sFlt1 for the pancreatic cancer patient is below the reference sample or threshold value, the less likely this patient will benefit from partial or full tumour resection.

14. A kit for performing a method according to claim 1, wherein the kit comprises means for determining the expression level of P1GF and/or sFlt1 as well as instructions for using said kit to calculate a ratio of the expression levels of P1GF and sFlt1.

15. An anti-P1GF compound for use in the treatment of a pancreatic cancer patient, wherein, when the ratio of the expression levels of P1GF and sFlt1 in a sample of said patient suffering from pancreatic cancer is expressed using the expression level of P1GF in the numerator, said pancreatic cancer patient has an increase in the ratio between the expression levels of P1GF and sFlt1 relative to a reference sample or a predetermined threshold value, preferably to a threshold value of 0.22, more preferably of 0.23, even more preferably of 0.24 or 0.25.

16. The anti-P1GF compound for use according to claim 15, wherein the anti-P1GF compound comprises an anti-P1GF antibody, preferably an anti-P1GF antibody comprising the CDR regions corresponding to the sequences of SEQ ID NO: 1 to 6.

Description

FIGURES

[0098] FIG. 1: High circulating PIGF/sFlt1 serum ratio is associated with neural invasion and reduced survival in patients with PDAC following curative surgery. A, Elevated circulating PIGF/sFlt1 serum ratio (referred to as PIGF/sFlt1.sup.circ) in the overall cohort of PDAC patients (n=80) as compared to healthy controls (ctr; n=80). Shown is the scatter dot plot with the median and interquartile range, p<0.0001, Mann-Whitney test. B, PIGF/sFlt1.sup.circ serum ratios do not differ between the subgroup of patients with non-resectable, i.e. locally advanced or metastatic tumors (M1, palliative, n=45) and the subgroup of patients with resectable tumors prior to curative-intent surgery (R0; n=35). Shown is the scatter dot plot with the median and interquartile range, p=0.869, Mann-Whitney test. C, Median PIGF/sFlt1.sup.circ serum ratios are elevated to 0.32 in patients with neural invasion (Pn1; n=25) as compared to 0.19 in patients without neural invasion (Pn0; n=20). Shown is the scatter dot plot with the median and interquartile range; p<0.0001, Mann-Whitney test. D, In contrast to neural invasion, PIGF/sFlt1.sup.circ serum ratios are not associated with the incidence of lymph node metastasis, which is in line with the notion that PIGF specifically functions as modifier at the tumor-nerve interface. Shown is the scatter dot plot with the median and interquartile range; n=11 for N0 (lymph node metastasis absent) and n=29 for N1 (lymph node metastasis present); p=0.55, Mann-Whitney test. E-F, Elevated PIGF/sFlt1.sup.circ serum ratios are associated with shorter overall survival in patients with PDAC who underwent curative-intent surgery (E), but not in patients with non-resectable (locally advanced or metastastic) PDAC undergoing palliative therapy (F). Shown is Kaplan-Meier survival in patients with PIGF/sFlt1.sup.circ<median (n=15) or PIGF/sFlt1.sup.circ>median (n=16), respectively, undergoing curative-intent surgery for resectable PDAC (HR: 4.106; 95% confidence interval: 1.131 to 14.91; Log-rank p=0.0128; E), and Kaplan-Meier survival in patients with palliative (non-resectable) disease allocated to groups with PIGF/sFlt1.sup.circ<median (n=18) and PIGF/sFlt1.sup.circ >median (n=19); HR: 0.845; 95% confidence interval: 0.3264 to 2.187; Log-rank p=0.153 (H). G, Elevated PIGF/sFlt1.sup.circ is associated with the severity of tumor-related pain. Pain was quantified using visual analogue scales (VAS 0-10) and grouped into mild (VAS 1-3), moderate (VAS 4-6), and strong pain (VAS 7-10). Shown are means±SEM of PIGF/sFlt1.sup.circ serum ratios in patients allocated to groups with mild, moderate or strong pain, respectively (*p<0.01; Mann-Whitney test).

[0099] FIG. 2: Optimal cut point and ROC curve analyses of PIGF/sFlt1.sup.circ serum ratios were performed using CRAN cutpointr software. A, Panel depicts optimal cut point and distribution of PIGF/sFlt1.sup.circ by class. Analysis displayed 92.0% sensitivity and 85.0% specificity for the discrimination between patients without neural invasion (no) and patients with neural invasion (yes) at a cut-off PIGF/sFlt1.sup.circ serum ratio of 0.2404 (95% confidence interval: 0.23-0.31). B, Panel depicts corresponding ROC curve of PIGF/sFlt1.sup.circ (AUC=0.935; 95% confidence interval: 0.855-0.992).

[0100] FIG. 3: PIGF and its receptors are expressed at the tumor-nerve interface. A, PIGF mRNA expression in tissues of human PDAC (n=24) and corresponding healthy pancreas (control; ctr) was determined using qPCR and normalized to GAPDH. Shown is the bar graph with mean±SEM; p=0.002. (B) Human DANG pancreatic tumor cells were grown orthotopically in the pancreas of NMRI.sup.nu/nu mice, and human (representing tumor cell derived) and murine (representing host derived) PIGF proteins in orthotopic tumor (PDAC) and paired healthy pancreatic tissues (control; ctr) determined using species-specific ELISA. Host-derived murine PIGF (indicated as mouse) is rarely expressed in the pancreas of mice (control; ctr) but induced in the stroma of PDAC xenografts. Tumor cell derived PIGF (indicated as human) is expressed in xenografted tumors. *p<0.01. C, ELISA based quantification of PIGF protein in supernatants of various human PDAC cell line cultures (n=3). D, Primary neurons (PN) and Schwann cells (SC) were isolated from newborn rats and cultivated with control media (ctr) or conditioned media (CM) derived from MiaPaca tumor cell cultures. PIGF mRNA expression was determined using qPCR, normalized to GAPDH and shown as mean±SEM (n=3; p=0.0072). PIGF expression is induced in Schwann cells upon cultivation with tumor cell conditioned supernatants. E-F, Representative IHC for the PIGF receptor Nrp1 in tissues of PDAC (E) and healthy pancreas (F). Intrapancreatic (in F) and intratumoral (in E) nerves are depicted by asterisks. Arrows point to Nrp1 expression in ductal epithelial cancer cells (arrowheads) and nerves (arrows). G, Nrp1 mRNA and protein as well as VEGFR1 mRNA expression levels were determined in human PDAC cell lines and in an immortalized human pancreatic ductal epithelial cell line (HPDE) using qPCR and normalized to GAPDH or immunoblotting, respectively (n=3). H, Expression of Nrp1 and VEGFR1 mRNA transcripts in dorsal root ganglia (DRGs), primary neurons (PN) and Schwan cells (SC) was determined using qPCR, normalized to GAPDH and shown as mean±SEM.

[0101] FIG. 4: The extent of neural invasion of extratumoral nerves predicts early disease recurrence. A-B, Morphometric analysis of neural invasion (NI) of nerves in PDAC specimens (referred to as intratumoral nerves) and in paired adjacent healthy pancreatic tissues (referred to as extratumoral nerves); n=24. Shown are (A) incidence of neural invasion in intra- and extratumoral nerves with absence (NI-) or presence (NI+) of neural invasion (p=0.19, Fisher's exact test) and (B) nerve fraction as determined by quantifying the percentage of tumor-invaded nerves per total nerves (p=0.002, Mann Whitney test). C-D, Within extratumoral nerves, a high neural invasion score (C) and a high nerve fraction (D; as determined by the fraction of tumor-invaded nerves per total nerves) are associated with shorter disease-free survival (DFS) in patients following curative-intent surgery. Shown is Kaplan-Meier disease-free survival in patients allocated to groups with extended invasion of extratumoral nerves (NI score 2) versus absent or focal neural invasion (NI score 0 and 1, respectively; HR: 4.46; 95% confidence interval: 1.069 to 18.60; Log-rank p=0.0042 in C), and in patients allocated to groups with extratumoral nerve fraction <median and nerve fraction >median (HR: 3.22; 95% confidence interval: 0.94 to 11.11; Log-rank p=0.0228 in D). E-F, Intraneural invasion (i.e. tumor cell invasion of the intraneural space) was quantified in extratumoral nerves (for a scheme displaying the respective morphometric analysis please refer to FIG. 6A). Kaplan-Meier depicts disease-free survival in the subgroups of patients with or without intraneural invasion of extratumoral nerves (HR: 3.63; 95% confidence interval: 0.90 to 14.70; Log-rank p=0.0037), and in patients allocated to groups with intraneural nerve fraction of extratumoral nerves<median and nerve fraction>median (HR: 3.78; 95% confidence interval: 0.82 to 17.38; Log-rank p=0.0139.

[0102] FIG. 5: The extent of neural invasion of extratumoral nerves is associated with reduced overall survival. A-B, Neural invasion was quantified in extratumoral nerves. Shown is Kaplan-Meier overall survival in patients allocated to groups with extended extratumoral NI (NI score 2) versus absent or focal NI (NI score 0 and 1, respectively; HR: 4.46; 95% confidence interval: 1.069 to 18.60; Log-rank p=0.0042 in A), and in patients allocated to groups with extratumoral nerve fraction <median and nerve fraction >median (HR: 3.22; 95% confidence interval: 0.94 to 11.11; Log-rank p=0.0228 in B). C-D, Intraneural invasion was quantified in extratumoral nerves. Kaplan-Meier depicts overall survival in the subgroups of patients with or without intraneural nerve invasion of extratumoral nerves (HR: 3.63; 95% confidence interval: 0.90 to 14.70; Log-rank p=0.0037), and in patients allocated to groups with intraneural nerve fraction of extratumoral nerves <median and nerve fraction >median (HR: 3.78; 95% confidence interval: 0.82 to 17.38; Log-rank p=0.0139.

[0103] FIG. 6: Expression of PIGF mRNA transcripts correlates with the extent of neural invasion of extratumoral nerves and increased neural plasticity. A, Morphometric analysis of neural invasion of nerves in PDAC. Scheme illustrates invasion of a nerve (indicated by asterisk) by tumor cell colonies (a, b, c, and e). Morphometric analysis depicts tumor invasion of perineural sheath (referred to as perineural invasion; a-c), the circumferential range of perineural invasion (d=a+b+c) and the invasion of the intraneural nerve space (referred to as intraneural invasion; e) of nerves in tumor tissues (intratumoral) and adjacent healthy pancreas (extratumoral). B, PIGF mRNA transcript levels in tumors without (NI-) and with neural invasion (NI+) of extratumoral nerves are shown as means±SEM, n=24, p=0.0043, Mann-Whitney test. C, Incidence of extratumoral neural invasion in tumors with low (<median) or high (>median) PIGF mRNA transcript levels (n=24, p=0.0325; Fisher's exact test). D, PIGF mRNA transcript levels in tumors without (PNI absent) or with perineural invasion (PNI present) of extratumoral nerves shown as means±SEM (n=24, p=0.0139; Mann-Whitney test). E, Circumferential range of perineural invasion of extratumoral nerves was morphometrically determined and scored as 0 (absent), 1 (¼ circumference), 2 (½ circumference), 3 (¾ circumference), and 4 (whole circumference). Shown are PNI scores as means±SEM of affected extratumoral nerves in tumors allocated to groups with PIGF transcript levels <median or >median (n=24, p=0.0331; Mann-Whitney test). F, Intraneural nerve fraction is determined as the percentage of extratumoral nerves exhibiting intraneural tumor invasion per total nerves analyzed. Bar graph depicts intraneural nerve fraction as mean±SEM in tumors allocated to groups with PIGF transcript levels <median and >median (n=24, p=0.0133; Mann-Whitney test). G, Neural plasticity is increased in tumors with high PIGF mRNA expression levels. Nerve area was morphometrically determined in PDAC tissues and shown as mean±SEM in tumors allocated to groups with PIGF transcript levels <median or >median (n=24, p=0.046; unpaired t-test). H-I, In contrast to neural invasion and neural plasticity, PIGF mRNA expression does not correlate with lymphatic metastasis. PIGF mRNA transcript levels in tumors without (N0) and with lymphatic metastasis (N1). Shown are mean±SEM, n=24, p=0.55, Mann-Whitney test (H). Lymph node fraction as determined by the percentage of tumor-infiltrated lymph nodes per total lymph nodes analyzed (I; p=0.65; Mann-Whitney test) in tumors with low (<median) or high (>median) PIGF mRNA transcript levels.

[0104] FIG. 7: PIGF mediates mutual chemoattraction between tumor cells and Schwann cells and stimulates neurite outgrowth. A, Schwann cells were seeded in the upper chamber of transwell inserts and allowed to migrate towards conditioned media from DANG cell monolayers (seeded in the lower chamber) or control media for 16 h. Neutralizing PIGF using anti-PIGF antibodies inhibits directed migration of Schwann cells as compared to IgG1 control. Shown are means±SEM; n=5; p<0.0001, Mann-Whitney t-test. B-C, Anti-PIGF inhibits directed migration of Panc1 (B) and Capan-2 (C) cells towards chemoattractant stimuli from conditioned Schwann-cell supernatants placed in the lower chamber (n=3; p<0.05). D, DRGs were co-cultured in 3D-matrigel with explanted PDAC xenograft tumor tissue and neurite outgrowth monitored by in vivo phase contrast microscopy. Neurite outgrowth from DRGs towards gradients released by PDAC tissues was determined by measuring the radial distance of the longest neurite extensions from the center of the DRG (proximal); distal: extensions of outgrowing neurites at the opposite site of the DRG as control (n=8; p<0.001). E-I, Whole primary DRG cell cultures (containing neurons and Schwann cells) were isolated from newborn rat DRGs and incubated with supernatants from various PDAC cells (E) or medium containing recombinant nerve growth factor (NGF), glial-derived nerve growth factor (GDNF) and PIGF (F). Overall neurite length was determined (n=3-5, p<0.05) using NeuroQuant® software based on selective immunofluorescent staining of primary neurons for neuron-specific Tuj-1. PIGF stimulates neurite outgrowth. Anti-PIGF was used to neutralize PIGF secreted by DANG cells, resulting in marked inhibition of neurite outgrowth. Representative images of primary neurons cultured with control media (G), DANG supernatant (H) and DANG supernatant with anti-PIGF (I) following selective IF-staining for neuron-specific Tuj-1.

[0105] FIG. 8: PIGF stimulates clonal growth of PDAC cell lines via Nrp1. A-B, Human PDAC (A) and pancreatic neuroendocrine tumor cell lines (B) were subjected to HTCA assays and the effects of PIGF (100 ng/ml), anti-PIGF antibodies and Nrp1-receptor knock-down using shRNA on clonal growth determined (A). Loss of responsiveness to PIGF in QGP cells with siRNA-mediated knockdown of Nrp1 as compared to scrambled controls (B; scr). All data are mean±SEM (n=3-5; *p<0.05). Chemotherapy induces PIGF expression within the neural compartment of PDAC. C, Treatment of DANG tumor bearing mice with gemcitabine induces PIGF expression released by tumor epithelial cells (human) and stromal cells (mouse) as determined using species-specific ELISA. D, Conditioned tumor cell supernatant (CM) and Gemcitabine induce PIGF expression in Schwann cells. Blocking PIGF enhances the efficacy of chemotherapy in vivo. D, DANG tumor-bearing mice received anti-PIGF, gemcitabine (125 mg/kg, twice/week) or the combination of both, and tumor growth was monitored.

[0106] FIG. 9: VEGF is not associated with neural invasion and neuropathic pain and does not predict disease prognosis in patients with PDAC following curative-intent surgery. A, Elevated VEGF serum levels in the overall cohort of PDAC patients (n=61) as compared to healthy controls (ctr; n=40). Shown is the scatter dot plot with the median and interquartile range, p=0.75, Mann-Whitney test. B, Shown is Kaplan-Meier overall survival in patients with serum levels of VEGF<median (n=47) or VEGF>median (n=47), respectively, undergoing curative-intent surgery for resectable PDAC (HR: 1,44; 95% confidence interval: 0.432 to 4.7, Log-rank p=0.55). C, VEGF serum levels in patients with neural invasion (NI+; n=15) and in patients without neural invasion (NI-; n=10); p=0.53, Mann-Whitney test. D, Pain was quantified using visual analogue scales (VAS 0-10) and grouped into mild (VAS 1-3), moderate (VAS 4-6), and strong pain (VAS 7-10). Shown are mean±SEM of circulating VEGF levels in patients allocated to groups with mild (n=9), moderate (n=19) or strong pain (n=9), respectively; p=0.677.

[0107] FIG. 10: PIGF mRNA transcript levels correlate with desmoplasia and inflammation within the tumor stroma. A, PIGF mRNA expression in orthotopic PDAC xenografts grown from human MiaPaca cells in NMRI.sup.nu/nu mice, isolated tumor-associated macrophages (TAMs) and MiaPaca cell cultures was determined using qPCR, normalized to GAPDH and shown as mean±SEM (n=3). B, In vitro expression and secretion of PIGF by mesenchymal embryonic fibroblasts (MEFs) derived from wildtype mice but not from PIGF.sup.-/- mice co-cultured with PDAC tumor cells. Shown is PIGF protein expression in supernatants as quantified by ELISA (n=3). C-D, Semiquantitative morphometric analysis of inflammation and desmoplasia in human PDAC specimens. Shown are inflammation (C) and desmoplasia scores (D) as mean±SEM in tumors allocated to groups with PIGF transcript levels <median or >median (n=24, *p=0.048 and #p=0.049; Mann-Whitney t-test).

[0108] FIG. 11: PDAC stimulates neural remodelling and plasticity. A-C, Non-transformed HPDE (A) and HupT3 PDAC cells (B) were cultured in separate patches divided by a 500 μm gap using IBIDI® inserts, and nascent neurites were visualized by immunofluorescent staining for Tuj-1 following 48 h. Neurite outgrowth was quantified by determining the neurite length and shown as mean±SEM (n=3; p<0.05). D, Conditioned media from PDAC cell lines DANG and HupT3 induce expression of the growth-associated-protein (GAP)-43 in primary neurons. Shown are mean±SEM GAP43 mRNA transcript expression in primary neurons. E, Blocking PIGF in conditioned media from DANG cell cultures abrogated the directed migration of F11 neurons towards chemoattractant released from DANG cell cultures. Shown are mean±SEM (n=3; p<0.05).

[0109] FIG. 12: shows treatment scheme of patients having resectable PDAC and good clinical performance status, wherein the ratio of PIGF/sFlt1 is used for stratification of patients into different treatment groups.

[0110] FIG. 13: shows treatment scheme of patients having resectable PDAC and bad clinical performance status and/or not eligible for chemotherapy, wherein the ratio of PIGF/sFlt1 is used for stratification of patients into different treatment groups.

[0111] FIG. 14: shows treatment scheme of patients having borderline resectable PDAC and good clinical performance status, wherein the ratio of PIGF/sFlt1 is used for stratification of patients into different treatment groups.

EXPERIMENTAL SECTION

Material and Methods

Materials

[0112] Antibodies were from ReliaTech (Braunschweig, Germany; to PIGF), Thermo Scientific (Fremont, CA; to β-actin), R&D Systems (Minneapolis, MN; to Nrp1 and VEGFR-1), and BD Pharmingen (Heidelberg, Germany; to CD31). Secondary antibodies were from Dianova (Hamburg, Germany). Recombinant human PIGF was from R&D Systems (Minneapolis, MN). Neutralizing antibodies to murine PIGF (5D11D4) and to human PIGF (16D3), as well as the IgG1 control antibody (1C8011) for use in vivo were supplied by ThromboGenics (Leuven, Belgium). Quantikine® ELISA kits for human and mouse PIGF and human VEGF were from determine PIGF and VEGF were from R&D Systems (Minneapolis, MN).

Patients and Samples

[0113] Sera samples were obtained from individuals with PDAC treated at Charité-Universitätsmedizin Berlin, Department of Gastroenterology, from 1998-2018. Tumor staging with TNM classification using CT or MRT scans was performed at the time of blood sampling. In patients with locally or metastatic disease who were not applicable for surgery, histopathological diagnosis was obtained by biopsies of primary tumors or metastases. In patients who underwent curative-intent surgery histopathological diagnosis and characteristics were obtained from pathology reports and included TNM as well as grading, neural invasion, lymph node metastasis and (lymph)angioinvasion. Clinical parameters were obtained from systematic review of the medical records. Follow-up visits with imaging studies were performed at least every 3 months. Tumor progression was determined based on multi-phasic computed tomography (CT), or magnetic resonance imaging (MRI). Quantification of pain was determined using visual analogue scale (VAS 0-10), and patients were allocated to groups with no pain (VAS 0), mild pain (VAS 1-2), moderate pain (VAS 3-6) and strong pain (VAS 7-10). Healthy controls were blood donors without medical history of malignant disease and consisted of 58 women and 29 men, with a median age of 45 years (range 41-56 years).

Pancreatic Cancer Specimens

[0114] Tissue samples were from individuals undergoing surgery due to pancreatic cancer at Charité-Universitätsmedizin from 12010-2016 with histologically confirmed pancreatic adenocarcinoma.

Determination of PIGF and sFlt1 Levels in Serum and Culture Supernatants

[0115] Concentrations of PIGF and sFlt1 were determined from serum samples using Elecsys® PIGF and sFlt1 immunoassays. PIGF levels from cell culture supernatants were measured using Quantikine® ELISA kits (R&D Systems) according to the manufacturer.

Quantitative RT-PCR

[0116] RNA from forty 20 μm slices/tissue was purified using the RNeasy-mini kit (Qiagen; Hilden, Germany), and RNA concentration and quality were determined on Agilent's 2100 Bioanalyzer using the RNA-6000-Nano Kit (Agilent; Santa Clara, USA). qRT-PCR was carried out in triplicate 10 μl-reactions using PIGF and sFlt1 TaqMan primer/probes from Applied-Biosystems (Foster City, USA) with the One-Step RT-PCR Kit (Invitrogen) on a CFX96 thermo-cycler (Bio-Rad; Hercules, USA). RNA isolation and qRT-PCR of cultured cells were performed as described. Relative quantification was calculated by the Livak-method.

Immunohistochemical Analyses

[0117] Cryostat-sections were fixed in 4% PFA. Immunoperoxidase-staining was performed using Vectastain Elite ABC-kit (Vector Laboratories; Wertheim-Bettingen, Germany) and AEC as substrate chromogen (DAKO; Hamburg, Germany). The antibody to Npn1 was diluted 1:100 and was omitted in negative controls. For a semi-quantitative evaluation of immunoreactivity, the immunoreactive area was determined relative to the total field measured using AxioVision®.

Cell Lines and Cultures

[0118] For all cell lines used throughout the study, frozen stocks were generated shortly after receipt of cell lines from the repositories, or after authentication by short tandem repeat DNA-Typing. MiaPaCa, ASPC1, and Panc1 cells were from ATCC (Manassas, USA); Capan-1, Capan-2 and DANG cells were from DSMZ (Braunschweig, Germany) and maintained as described.

Preparation of Cell Extracts and Immunoblotting

[0119] 5×10.sup.6 cells were treated with doxycycline for 48 h, lysed in 100 μl RIPA buffer and immunoblotted as described. TCA-precipitates from culture supernatants were prepared as described.

Growth, Migration and Invasion Assays

[0120] 10.sup.5 cells were plated in 24-well-dishes, and cell numbers counted using a hemocytometer. For migration assays, 2×10.sup.5 cells/insert (doxycycline-pretreated, if applicable) were placed in serum-free medium in the upper chamber of 8 μm-Transwells (Corning) and allowed to migrate for 8-12 h towards 1% FCS added to the lower chamber. Tumor cell migration towards gradients from Schwann-cells, cultured in the lower chamber for 16 h in DMEM (0.1% BSA), and vice-versa, was determined. Migrated cells were stained with crystal-violet or DAPI and quantified by counting 5 standardized fields at 100× magnification. For 3D-Matrigel-Invasion assays, 8 μm-Transwells were coated with 1 mg/ml growth-factor-reduced Matrigel® (BD Pharmingen). Individual experiments were performed in triplicates.

Assays to Study Neural Invasion

[0121] Using transwell assays, the effect of tumor derived PIGF on the directed migration of Schwann cells (placed in the upper transwell chamber) towards chemoattractant and neurotrophic gradients released by tumor cells cultured in the lower chamber, was evaluated. Vice versa, the effect of PIGF on the directed migration of tumor cells (placed in the upper transwell chamber) towards chemoattractant and neurotrophic gradients released by Schwann cell cultures in the lower chamber, was evaluated. Antibodies to PIGF were used to neutralize PIGF. The abundance of migrated cells will be quantified as described.

Assays to Study Neural Plasticity

[0122] DRGs from newborn rats were freshly isolated and placed in 12-well plates in a growth-factor-reduced Matrigel-drop, thus providing a planar surface that enables outgrowing neurites to project and align in a 2D manner. Explanted PDAC xenograft tissues in a separate Matrigel-drop placed at 1 mm distance were connected to the DRG with a Matrigel-bridge. The matrigel layer allows gradients of chemokines and axon guidance cues to establish, and neurites to elongate towards the tumor. A matrigel drop without xenograft tumor tissue at the opposite site serves as control for random neurite outgrowth. Time-lapse imaging (Leica DM16000-B) conducted between day 1-5 following co-culture initiation was evaluated using automated acquisition software (Leica LAS AF6000). Images were taken in a chamber with closed environment at 37° C. and 5% CO.sub.2 to follow neurite outgrowth. Radial distance of neurite extension from the center of the DRG were determined and calculated using ImageJ® and IBIDI® chemotaxis software.

[0123] Chemoattraction of neurons by PDAC derived neurotrophic factors was interrogated via co-culture experiments in transwell assays. Whole DRG primary cell cultures or enriched primary neurons were seeded in the upper chamber of Fluor® Blok™ transwell inserts (Corning®). Filter with 1 μm pores selectively allow outgrowing neurites but not neuron cell bodies to migrate and project through the transwell membranes towards chemoattractant gradients established by tumor cells or conditioned tumor cell media, cultured or placed in the lower chamber, respectively. Of note, Fluor® Blok™ inserts are provided with a light-tight polyethylene terephthalate membrane, which efficiently blocks transmission of light from the top chamber of the insert and hence allow fluorescence detection of growing neurites (CellTracker™ labeled) exclusively at the bottom side of the filter membrane. The dynamic process of neurite outgrowth and elongation in response to tumor cell gradients was captured using in vivo time-lapse microscopy imaging (every 10 minutes over 24 hours), and quantified by determining the overall number and the mean length and density of neurites at defined time points. Alternatively, primary neurons were selectively visualized by immunofluorecent detection of β3-tubulin and subjected to morphometric analysis using NeuriteQuant® to determine overall tumor length and branching. Antibodies to PIGF were used to neutralize the effect of PIGF.

[0124] In order to track locomotion of individual neurons, primary neurons were co-cultured with tumor cells in two separate patches divided by a 500 μm gap using IBIDI® inserts. GFP or cell tracking dyes (CellTracker™) were used to selectively visualize neurons and tumor cells. The dynamic process by which neurons migrate towards gradients from tumor cells were evaluated using time-lapse imaging in a Leica DMI6000B live cell microscopy unit (closed environment at 37° C. and 5% CO.sub.2). Using automated acquisition software (Leica LAS AF6000), images were taken every 15 min for up to 24 h to follow cell locomotion. Cell trajectory, distance from origin, velocity, and forward migration index of neurons were calculated as described.

Tumor Models

[0125] Local authorities approved animal experiments. Female NMRI.sup.nu/nu mice (20-24g) were from Charles River Laboratories (Sulzfeld, Germany). For orthotopic MiaPaCa tumors, 10.sup.6 cells were injected into the pancreatic head. After 7 weeks, mice were sacrificed, primary tumors harvested, and enlarged lymph nodes collected. Metastatic nodules in the mesentery were counted as described.

Statistics

[0126] Data are presented as mean±SEM, circulating levels of PIGF and sFlt1 as median with interquartile ranges. Statistical significance was determined by t-test, Fisher's exact test and Mann-Whitney test using SPSS® (v18.0; Chicago, IL) and GraphPad® Prism (v5.0; San-Diego, CA). Tumor-related survival and time-to-progression were calculated based on the date of blood sampling and analyzed using the Kaplan-Meier method and Log-rank test. Cox proportional hazards regression model was used for multivariate analysis. (*) P<0.05 values were considered significant; all tests were two-sided.

Results

Elevated Circulating PIGF/sFlt1 Serum Ratio is Associated with Neural Invasion and Predicts Disease Prognosis in Patients with PDAC

[0127] PIGF is a prognostic parameter in a variety of solid cancers, as PIGF levels in blood and tumor tissues of patients reflect the risk of tumor progression and metastasis. Given that PIGF is released into systemic circulation, initial experiments assessed the expression of circulating PIGF in patients with PDAC. Since binding of PIGF to its soluble receptor Flt1 (sFlt1; sVEGFR-1) may diminish its biological activity, we also determined sFlt1 levels. The ratio PIGF/sFlt1 represents unbound and hence bioactive circulating PIGF. Compared to healthy controls, median circulating PIGF/sFlt1 ratio (from heron referred to as PIGF/sFlt1.sup.circ) was found elevated in the overall cohort of PDAC patients (FIG. 1A), but strikingly did not differ among subgroups of patients with advanced, and thus non-resectable tumors (FIG. 1B; M1) versus locally resectable tumors prior to curative-intent surgery (FIG. 1B; R0). In sharp contrast, PIGF/sFlt1.sup.circ serum ratios were not elevated in patients with chronic pancreatitis (not shown).

[0128] As high rates of disease recurrence following curative-intent surgery constitute an unsolved medical need even in the absence of systemic dissemination, we analyzed the relation of PIGF/sFlt1.sup.circ levels to the known independent risk factors for tumor recurrence, i.e. the presence and extent of lymph node metastasis and the presence of neural invasion. Notably, PIGF/sFlt1.sup.circ ratios were associated with the presence of neural invasion (FIG. 10), which represents a distinct and independent route of metastasis that is characteristic for PDAC. Indeed, median PIGF/sFlt1.sup.circ serum ratios are elevated to 0.32 in patients with neural invasion as compared to 0.19 in patients without neural invasion (FIG. C). In contrast, PIGF/sFlt1.sup.circ ratios did not correlate to the incidence (FIG. 1D) nor the extent of lymph node metastasis (not shown), which is in line with the notion that PIGF exerts a unique function as modifier at the tumor-nerve interface in PDAC.

[0129] The clinicopathological association of PIGF/sFlt1.sup.circ ratios with neural invasion raised the question, whether serum ratios of PIGF/sFlt1.sup.circ also reflected the prognosis. Accordingly, patients were grouped by high (>50.sup.th percentile) or low (≤50.sup.th percentile) levels of PIGF/sFlt1.sup.circ ratios, respectively, and Kaplan-Meier curves based on survival were generated. Indeed, PIGF/sFlt1.sup.circ levels >50.sup.th percentile (>median) were associated with poor prognosis and predicted shorter overall survival in patients with PDAC undergoing curative-intent surgery (FIG. 1E). Notably, in patients with unresectable advanced or metastatic PDAC (representing palliative disease) PIGF/sFlt1.sup.circ levels were not associated with overall survival (FIG. 1F), but with neuropathic pain (FIG. 1G), which is closely associated with the extent and severity of neural invasion in PDAC. In sharp contrast, circulating levels of VEGF did not differ in patients with PDAC as compared to healthy controls, nor were they associated with lymph node metastasis, neural invasion, overall survival, or pain (Suppl. FIG. 1).

[0130] In an attempt to stratify patients to prognostic subgroups, optimal cut point and ROC analyses were performed and displayed 92.0% sensitivity and 85.0% specificity for the discrimination between patients with neural invasion and patients without neural invasion at a cut-off PIGF/sFlt1.sup.circ serum ratio of 0.2404 (FIG. 2).

PIGF and its Receptor Nrp1 are Expressed at the Tumor-Nerve Interface in PDAC

[0131] Consistent with elevated PIGF/sFlt1.sup.circ serum ratios in PDAC patients, PIGF mRNA expression was induced in tissues of human PDAC (FIG. 3A) and mouse orthotopic xenograft tumors (FIG. 3B, and own previous data) as compared to healthy pancreas adjacent to the tumor. Moreover, differential quantification of mouse and human PIGF proteins in xenograft tumor tissues (FIG. 3B) and human PDAC cell lines (FIG. 3C) using species-specific ELISA identified both, the epithelial tumor cells as well as the stromal compartment as source for PIGF production.

[0132] Indeed, while PIGF expression and secretion was found either deficient or variably present in a series of human PDAC cell lines in vitro (FIG. 3C), stromal tumor associated macrophages (TAMs), which were isolated from MiaPaCa xenograft tumor tissues, constituted a major source of PIGF mRNA production in xenografted mouse tumors (Suppl. FIG. 2A).

[0133] In addition to TAMs, the stromal tumor micromilieu consists of a variety of different cell types, of which fibroblasts, Schwann cells and neurons have particularly been characterized to play a pivotal role in PDAC progression and neural metastasis. Notably, in in vitro co-culture experiments, PDAC cells induced the expression of PIGF in mesenchymal embryonic fibroblasts (MEFs) derived from wildtype mice but not from PIGF.sup.-/- mice (Suppl. FIG. 2B). These finding are consistent with our observation that high levels of PIGF mRNA transcripts in human PDAC specimens correlate with increased inflammation and desmoplasia of the tumor stroma (Suppl. FIG. 2C-D) and hence fit the pleiotropic role of PIGF as modulator of the tumor microenvironment. Moreover, conditioned supernatants from PDAC cell cultures were capable to induce PIGF expression in primary Schwan cells, but not in primary neurons (FIG. 3D).

[0134] On the receptor side, we focused on Nrp1, which acts as independent receptor for PIGF and conveys signals different from established VEGFR1-mediated signaling pathways. Nrp1 immunoreactivity was localized to epithelial tumor cells, parenchymal nerves and on vascular endothelial cells in human PDAC tissues (FIG. 3E), but weakly expressed in healthy pancreas (FIG. 3F). Similarly, non-transformed immortalized human pancreatic ductal epithelial cells (HPDE) lack Nrp1 mRNA and protein expression, which contrasts with abundant Nrp1 mRNA transcripts and protein levels in most PDAC cell lines (FIG. 3G). Notably, Nrp1 mRNA expression was also present in dorsal root ganglia, primary neurons and Schwann cells, freshly isolated from newborn mice or rats (FIG. 3H), as well as in permanent F11 neuron cell cultures (not shown).

[0135] In contrast to Nrp1, expression of VEGFR-1 appeared restricted to the stromal compartment of PDAC, and in particular localized to vascular endothelial cells, inflammatory cells and parenchymal nerves (own previous results and not shown). In line with this observation, VEGFR-1 transcripts were rarely expressed among human PDAC cell lines (FIG. 3G), but present in DRGs, primary and F11 neurons as well as Schwann cells (FIG. 3H; and not shown). VEGFR-1 mRNA transcripts were furthermore induced in primary neurons upon cultivation in conditioned media from PDAC cell lines (not shown), consistent with previous findings that expression of VEGFR-1 in nerves was induced in PDAC patients.

[0136] Hence, both epithelial tumor cells and the neuronal compartment of the tumor stroma have the capacity to respond to secreted PIGF ligand.

Neural Invasion of Extratumoral Nerves Predicts Early Disease Recurrence in PDAC

[0137] In routine pathology reports, presence of neural invasion is described as a solely qualitative feature in up to 90% of tumors. To obtain a quantitative description of neural invasion (referred to as NI) and neural plasticity, we implemented a histomorphometric analysis which allows recording a detailed set of histomorphological parameters.

[0138] These parameters are based on evaluation of 30 mm.sup.2 areas within the tumor (referred to as intratumoral) and in healthy tissue adjacent to the tumor margin (referred to as extratumoral), and include (i) nerve number (density) and diameters (hypertrophy), (ii) presence, quality (perineural vs. intraneural) and area of tumor cell invasion, (iii) presence of intraneural and tumoral inflammation and presence of desmoplasia.

[0139] As expected, thorough morphometric evaluation of PDAC specimens revealed that intratumoral NI is present in almost all PDAC patients (FIG. 4A), while the extent of intratumoral NI as quantified by determining the fraction of tumor-invaded nerves per total nerves largely varied among PDAC patients (FIG. 4B), as did the extent of neural plasticity (not shown).

[0140] Mechanistically, migration of tumor cells along nerves represents a process which may ultimately result in the continuous spread of tumor cells beyond the boundaries of the tumor. Hence, NI can occur in nerves within normal pancreatic regions distant to the tumor by means of continuous spread of tumor cells along parenchymal nerves. In addition to determining NI in tumor tissues, we thus also quantified NI in respective healthy pancreatic tissues adjacent to the tumor margin (referred to as extratumoral NI). Although extratumoral NI was significantly less frequent as compared to intratumoral NI, when paired tumor and healthy tissues were analyzed, its incidence remained high (FIG. 4A-B).

[0141] Clinically, the extent of extratumoral NI correlated with an unfavorable course of the disease and early disease recurrence in PDAC patients. Indeed, when patients who underwent curative-intent surgery for PDAC were grouped by a high or low neural invasion score, and Kaplan-Meier curves based on disease-free survival (DFS) and overall survival were generated, a high neural invasion score (score 2) was associated with shorter disease-free survival (FIG. 4C) and overall survival (FIG. 5A) as compared to a low neural invasion score (score 0/1). More specifically, a thorough quantitative analysis of extratumoral nerves linked a high fraction of tumor-invaded nerves per total nerves to shorter disease-free survival (FIG. 4D) and overall survival (FIG. 5B).

[0142] Tumor cell invasion of the intraneural space (referred to as Intraneural invasion) is considered a feature of advanced neural invasion in PDAC. Accordingly, both the presence of intraneural invasion and a high fraction of intraneural invasion of extratumoral nerves predicted early disease recurrence and shorter overall survival in our patients' cohort (FIG. 4E-F and FIG. 5C-D).

[0143] Thus, we are able to quantitatively score the extent of NI of extratumoral nerves and accordingly stratify patients with PDAC into prognostic subgroups.

Expression of PIGF Transcripts Correlates with the Extent of Neural Invasion of Extratumoral Nerves and with Neural Plasticity

[0144] Given that expression of PIGF and its receptors Nrp1 and VEGFR1 are present and/or increased in cellular compartments of the tumor-nerve interface in PDAC, we aimed to determine, whether PIGF transcript levels correlate with neural invasion. Since incidence of neural invasion in intratumoral nerves was high among PDAC patients (FIG. 4A-B), morphometric analysis was mainly focused on neural invasion of extratumoral nerves (FIG. 6A), which is considered an advanced stage of neural spread of tumor cells, and which we found closely linked to disease-free and overall survival in PDAC patients (FIG. 4C-F and FIG. 5A-D). Indeed, tumors with extratumoral NI exhibited higher mean PIGF transcript levels as compared to tumors, in which extratumoral NI was absent (FIG. 6B). Moreover, PIGF transcript levels correlated with the incidence of NI of extratumoral nerves, and all tumors with PIGF transcript levels >median exclusively displayed presence of NI within extratumoral nerves (FIG. 6C). A more detailed morphometric analysis addressed the correlation of PIGF transcript levels with features of advanced and more severe NI. Such features are an extended circumferential range of perineural nerve invasion and the presence of tumor cell invasion within the intraneural space, referred to as intraneural nerve invasion (FIG. 6A). Thus, tumors with perineural invasion of extratumoral nerves pe se exhibited higher mean PIGF transcript levels as compared to tumors without perineural invasion (FIG. 6D), whilst tumors with a high perineural invasion score, based on determining the extent of circumferential perineural nerve invasion, exhibited PIGF transcript levels >median (FIG. 6E). Finally, PIGF transcript levels >median were linked to a higher fraction of intraneural invasion of extratumoral nerves (FIG. 6F). Accordingly, intraneural nerve fraction of intratumoral nerves was also higher in tumors with PIGF mRNA transcript expression >median (n=24, p=0.0133; Mann-Whitney test) as compared to tumors with PIGF mRNA <median. Furthermore, the overall nerve area in tumors with PIGF mRNA transcript levels >median was larger as compared to tumors with PIGF mRNA transcript levels <median, delineating a role of PIGF as modifier of neural plasticity in PDAC (FIG. 6G). In contrast, PIGF mRNA transcript levels did not correlate with the incidence (FIG. 6H) nor the extent of lymph node metastasis (FIG. 61) in these patients.

[0145] PIGF Mediates Mutual Chemoattraction Between Tumor Cells and Schwann Cells

[0146] Schwann cells act as conduits that facilitate the subsequent axonal outgrowth in neuronal regeneration following injury and in neural invasion. More recent reports in PDAC revealed, that Schwann cells have the capacity to temporarily disengage from the perineural sheath and migrate towards tumor cell colonies, and reciprocally chemoattract tumor cells into the direction to the nerve, altogether promoting NI.

[0147] To determine the consequences of PIGF for mutual chemoattraction of Schwann cells and tumor cells within the stromal microenvironment, we initially used monolayer cultures and conditioned supernatants of DANG cells with endogenous expression of PIGF as chemoattractant. Conditioned media from DANG cell monolayers enhanced the directed migration of Schwann cells, while inhibition of PIGF by using anti-PIGF, but not control IgG1 antibodies abrogated this effect (FIG. 7A). Vice versa, conditioned supernatant from Schwann cell cultures stimulated the directed migration of Panc1 (FIG. 7B) and Capan-2 tumor cells (FIG. 7C), while blocking PIGF using anti-PIGF antibodies abolished the directed migration of PDAC cells towards chemoattractant released by Schwann cells (FIG. 7B-C). Thus, PIGF affects the chemotactic profile of PDAC cells and Schwann cells, suggesting that PIGF modifies NI in models of PDAC.

PDAC Xenograft Tissues and Gradients From PDAC Cells Stimulate Neurite Outgrowth

[0148] To assess neural plasticity in PDAC, we performed ex vivo co-culture assays using xenograft PDAC tissue or cell lines, respectively, which were co-cultured with whole DRGs, as well as single cell cultures of primary neurons and Schwann cells, both freshly isolated from newborn rat and mouse DRGs, and permanent cultures of F11 hybridoma neurons.

[0149] When DRGs are cultured in 3D collagen matrices together with (matrix embedded) explants of PDAC xenograft tissues, neurite outgrowth is skewed towards the tumor tissue, as reflected by differences in the radial distance of neurite extension from the center of the DRG (FIG. 7D). Thus, neurites facing the tumor were more elongated than neurites at the opposite site of the DRG, suggesting that tumor gradients stimulate and/or attract neurite outgrowth and elongation.

[0150] To study the directed chemoattraction of nascent neurites towards PDAC gradients, neurons were co-cultured with either human pancreatic ductal epithelial cells (HPDE) or PDAC cells in separate patches divided by a 500 μm gap using IBIDI® inserts (Suppl. FIG. 3A-C), and nascent neurites were visualized by immunofluorescent staining for Tuj-1 following 48 h. Notably, neurite outgrowth as quantified by determining the neurite length was increased over time when neurons were co-cultured with HupT3 tumor cells as compared to non-transformed HPDE cells. Moreover, conditioned media from PDAC cell lines induce expression of the growth-associated-protein (GAP)-43 in primary neurons (Suppl. FIG. 3D), which stimulates growth cone formation and neurite outgrowth. Thus, PDAC derived tissue and growth factors have the capacity to induce neural remodeling in ex vivo models of the tumor-nerve interface.

Tumor Derived PIGF Regulates Neural Plasticity

[0151] To gain a more complex depiction of tumor induced changes of neuronal plasticity and to interrogate chemoattraction of neurons by PDAC derived neurotrophic factors, whole DRG primary cell cultures were incubated with supernatants from PDAC cell lines. Since evidence is emerging that nerve plasticity is controlled by mutual interaction of neurons with their ensheathing Schwann cells, we used whole DRG primary cell cultures (encompassing Schwann cells in addition to primary neurons) instead of using pure (highly enriched) cultures of primary DRG neurons. Primary neurons are selectively visualized by immunofluorecent detection of β3-tubulin and subjected to morphometric analysis using NeuriteQuant®.

[0152] PDAC cell supernatants variably stimulated overall neurite length (FIG. 7E), as did recombinant PIGF and the two major known neurotrophic factors, nerve growth factor (NGF) or glial-derived neurotrophic factor (GDNF; FIG. 7F).

[0153] Importantly, neutralizing endogenous PIGF in conditioned media from DANG cell cultures with anti-PIGF antibodies reduced neurite length (FIG. 7F-I). Moreover, blocking PIGF in conditioned media from DANG cell cultures also abrogated the directed migration of F11 neurons towards chemoattractant released from DANG cell cultures (Suppl. FIG. 3E). Together, these data suggest, that tumor derived PIGF has the capacity to modify cancer-mediated neural plasticity and neural remodeling.

PIGF Stimulates Colony Formation of Tumor Cells Via Nrp1

[0154] Pharmacological studies using anti-PIGF have previously revealed a functional role of PIGF for tumor growth and metastases in syngeneic orthotopic PDAC mouse models. More recently, Nrp1 has emerged as independent receptor for PIGF and conveys signals different from established VEGFR1-mediated signaling pathways. However, the role of Nrp1 for PIGF mediated effects in PDAC epithelial cells has not been assessed yet. Therefore, we experimentally studied the consequences of exogenous PIGF and antibody-based pharmacological inhibition of PIGF, respectively, on colony formation of PDAC cells. Notably, anti-PIGF antibodies prevent biding of PIGF to both, VEGFR-1 and Nrp1, as previously reported.

[0155] In vitro, regulation of clonal growth of PDAC and pancreatic neuroendocrine tumor cells by PIGF was critically dependent on Nrp1. Thus, exogenous PIGF stimulated colony formation of Nrp1-expressing Capan-1 and QGP cells irrespective of VEGFR1 deficiency of QGP cells, but not of MiaPaCa cells, lacking both VEGFR1 and Nrp1 (FIG. 8A). Conversely, pharmacological disruption of autocrine PIGF stimulation using neutralizing anti-PIGF antibodies in VEGFR1-deficient Panc1 and BXPC3 cells reduced clonal growth (FIG. 8A and FIG. 3G). Moreover, knockdown of Nrp1 in VEGFR1-deficient QGP-1 cells abrogated the response to PIGF, while neutralizing endogenous PIGF in DANG cells reduced clonal growth equal to knockdown of the Nrp1 receptor (FIG. 8A-B). Given the overall rare expression of VEGFR-1 in PDAC cell lines (see FIG. 3G), these data indicate that interaction of PIGF with Nrp1-receptor may be required and sufficient to mediate effects PIGF on clonal growth of PDAC.

Blocking PIGF Enhances the Efficacy of Chemotherapy In Vivo

[0156] Treatment of mice with DANG xenograft tumors with the cytostatic agent gemcitabine increased PIGF expression in tumor epithelial cells and the stromal compartment of PDAC xenografts (FIG. 8C). Moreover, exposure of primary neurons and Schwann cells with conditioned tumor cell media or gemcitabine dose-dependently increased PIGF expression in Schann cells, but not in primary neurons (FIG. 8D), altogether providing a rationale to test whether blocking PIGF would enhance chemotherapeutic treatment. Indeed, combination therapy with anti-PIGF and gemcitabine inhibited growth of DANG xenograft tumors more than monotherapies alone (FIG. 8E). Thus, blocking PIGF may enhance the efficacy of chemotherapy.

Discussion

[0157] Understanding the molecular basis of the almost inevitable, therapy-refractory progression of PDAC constitutes a central task for translational research. Our work applies the paradigm of the neurovascular link to understand how tumor cells exploit the shared evolutionary origin of axon-/vessel guidance factors in specialized interactions with nerves (and vessels) in the tumor environment and explores PIGF as a member of the axon and vessel guidance family for such specialized interactions in PDAC.

[0158] Our current data provide evidence, that elevated circulating PIGF levels are associated with neural invasion (NI) and shorter survival in patients with PDAC who underwent surgery, and delineate the function of PIGF for neural invasion, which represent the clinically relevant and characteristic route of metastatic spread in PDAC. Experimentally, we characterize PIGF in its unique role as an axon guidance factor, which supports neural plasticity on one hand, and attracts tumor cells towards nerves on the other hand. Specifically, we provided evidence that PIGF contributes to induction of directional and dynamic changes in outgrowth of primary DRG neurons upon exposure to PDAC derived guidance cues and growth factors and supports mutual chemoattraction of tumor cells with neuronal cells and Schwan cells, respectively.

[0159] Ultimately, findings from these experimental models translated to the clinical observation that circulating PIGF/sFlt1.sup.circ serum ratios are associated with NI. More specifically, induction of tumor PIGF mRNA transcript levels correlated to higher incidence and more specifically to an increased extent of extratumoral NI, which predicted unfavorable prognosis, early disease recurrence and reduced survival in patients with PDAC following curative-intent surgery.

[0160] The neurovascular link offers a unique concept to recognize specialized features in the characteristic way of cancer cell spread along nerves in PDAC. Physiologically, nerves and vessels exhibit comparable structural patterning and share similar ligand-receptor cues to navigate to their target. So called axon guidance factors have originally been characterized in neurogenesis to function as molecular cues, which control growth, navigation and positioning of neurons in the developing brain. These ancient axon guidance signals from the evolutionary older nervous system were co-opted for navigational control by the emerging younger blood vessels. In turn, VEGF-A, which was originally discovered as the key angiogenic growth factor, has recently been characterized as neurotrophic factor in brain development and motoneuron diseases.

[0161] Own previous work provided precedent evidence for a role of axon guidance factor Slit2 and its receptor Robot in NI of PDAC, hence creating a functional counterpart to the report of frequent genomic mutations in the Slit2-Robo pathway. Indeed, reduced expression of components of this pathway correlated to shorter survival in PDAC. Restoring Slit2 function in PDAC cell lines inhibited bidirectional chemoattraction of tumor cells, neurons and Schwann cells, and impaired accelerated directional PDAC cell navigation along outgrowing neurites.

[0162] VEGF-A binds to and signals via two receptor tyrosine kinases, VEGFR-1 and VEGFR-2, which are highly homologous in their structure, but convey distinct and independent biological functions. In contrast, PIGF, a homologue member of the VEGF growth factor family, exclusively binds to VEGFR-1. A recent report established a role of VEGFR1 in cancer related functional and structural remodeling of nerves under pathophysiological conditions of neuropathic pain: Indeed, VEGF and its homologue PIGF induced nociceptive sensitization and augmented pain sensitivity through selective activation of VEGFR-1 (but not VEGFR-2), expressed in sensory neurons in human cancer and mouse models. Moreover, VERGR1 expression and intensity were upregulated in sensory nerves of human PDAC specimens and correlated to perceived pain intensity in PDAC patients. Conversely, genetic loss or pharmacological inhibition of VEGFR1 signaling or VEGF ligands, respectively, impaired tumor associated sprouting and hypertrophy of nerve fibers and attenuated cancer associated pain in animal models.

[0163] In perfect line with these observations, our current data correlate circulating serum levels of PIGF with perceived neuropathic pain in PDAC patients and more specifically characterize a functional role of PIGF in cancer related structural remodeling of neural plasticity. Thus, PIGF transcript levels correlated with the overall nerve area in PDAC, and blocking PIGF by anti-PIGF antibodies abrogated cancer mediated neurite outgrowth from primary DRG neurons in ex vivo co-culture models, and hence reduced neural remodeling and plasticity.

[0164] Furthermore, our experimental and clinical data assign to PIGF a novel function in the control of neural invasion by modifying bilateral chemoattraction between PDAC cells and neuronal cells, in particular Schwann cells, which recently gained increasing attention as key players in NI of PDAC.

[0165] A recent systemic review and meta-analysis advocates NI as independent prognostic factor for tumor recurrence (disease free survival) and overall survival in patients with PDAC in multivariate analysis. Median survival ranged from 7 to 28.5 months for patients with presence of NI versus 8 to 56.1 months for patients, in whom NI was absent. Fittingly, a retrospective analysis in our department characterized NI as an unfavorable prognostic factor with a reduced median survival of 31 months in patients with NI as compared to 79.3 months in patients without NI.

[0166] Despite the utmost clinical impact of NI, neither quantification of NI nor the assessment of extratumoral (intrapancreatic) neural invasion has been implemented as standardized procedures in histopathological reports, taking into account that NI frequently occurs in nerves within normal pancreatic regions distant to the tumor (referred to as extratumoral NI).

[0167] So far, only few studies provide sufficient information on extratumoral (intrapancreatic) NI, with incidences ranging from 76.2% to 97.8%. Accordingly, our current data support further evidence that the severity of extratumoral NI (as determined by the fraction of invaded nerves per total nerves) is associated with reduced survival in PDAC patients and are thus in perfect agreement with published data, which characterized extratumoral neural invasion as an independent, unfavorable prognostic parameter for tumor recurrence and survival in PDAC patients.

[0168] A thorough morphometric and quantitative analysis of NI and neural plasticity now enabled us to characterize PIGF in its role as axon guidance cue by correlating PIGF transcript levels in PDAC tissues with the extent of intra- and extratumoral neural invasion. Thus, induced expression of PIGF transcripts in PDAC specimens correlated with a higher extent of extratumoral NI. Apart from signaling via VEGFR-1, the axon guidance receptor, neuropilin-1 was found to acts as independent receptor for PIGF and conveys signals different from established VEGFR1-mediated signaling pathways. As both receptors are expressed in the neuronal compartment of PDAC and moreover high expression of Nrp1 was associated with poor overall survival in patients following resection of PDAC, it is tempting to speculate that both, VEGFR-1 and Nrp1 mediate observed effects of PIGF on neural plasticity and neural invasion.

[0169] Taken together, our experimental and clinical data assign to PIGF a novel function in the control of neural invasion by modifying bilateral chemoattraction between PDAC cells and neuronal cells at the tumor-nerve interface. More specifically, our data suggest, that blocking PIGF may represents a promising therapeutic strategy to impair metastatic spread along outgrowing neurites and hence to reduce tumor recurrence rates and improve survival following curative-intent surgery in the subgroup of PDAC patients with elevated PIGF/sFlt1.sup.circ serum ratios.

[0170] Upfront tumor resection with following adjuvant chemotherapy has long been the standard of care of patients with resectable tumors. However, despite major improvement of surgical quality, nearly half of patients fail to complete or never receive adjuvant chemotherapy after pancreatectomy, due to postoperative complications, clinical deterioration with poor clinical performance status and limited compliance, or early progressive disease. Especially those patients who are not able to receive adjuvant chemotherapy have a poor prognosis.

[0171] This has led to an ongoing paradigm shift and resulted in an increased emphasis on the use of upfront neoadjuvant chemotherapy for all patients without evidence of metastatic disease. Neoadjuvant chemotherapy, given upfront to surgery, has emerged as standard therapy in a variety of cancer entities including breast cancer, esophagogastric cancer and colorectal cancer, and has several potential benefits in pancreatic cancer as well: such as early treatment of occult (distant or nodal) micro-metastases, improved compliance with chemotherapy and early delivery of systemic chemotherapy to all patients, the potential to downstage tumor size and increase margin-negative (R0) resection rates, as well as select patients who respond to neoadjuvant treatment and are thus likely to benefit from surgery, whilst preventing vain surgery in patients with rapidly progressive disease. Given these advantages, there is an increasing clinical focus on evaluating the use of neoadjuvant therapy in the upfront resectable pancreatic cancer population. The neoadjuvant approach has recently been supported by three randomized controlled trials, which showed a favorable outcome of neoadjuvant gemcitabine or gemcitabine-based chemoradiotherapy versus upfront surgery. Indeed, neoadjuvant therapy decreases nodal positivity, increases margin-negative resection rates and increases overall and disease-free survival. Since FOLFIRINOX has proven superior compared to gemcitabine in both the metastatic and adjuvant setting, it is likely to be a more effective neoadjuvant regime as well. Several ongoing randomized clinical trials hence currently evaluate the efficacy of neoadjuvant FOLFIRINOX given either alone or combined with radiotherapy in comparison to gemcitabine-based regimens.

[0172] It is widely appreciated that neural invasion (NI) limits the clinical benefit of surgery followed by adjuvant therapy. So far however, neural invasion has not been systematically evaluated yet in the neoadjuvant therapeutic concept. Given the ongoing paradigm switch to neoadjuvant therapy, understanding the mechanisms underlying neural invasion as a unique 25 and characteristic route of local tumor cell spread in pancreatic cancer has therefore become a major task to understand limitations of neoadjuvant therapeutic regimens and consecutively better direct treatment stratification of patients with pancreatic cancer.

[0173] Own data provide evidence that incidence and extent of NI of extratumoral nerves strongly correlate to disease prognosis, in particular to time-to-disease-recurrence. However, pathologic analysis to determine NI inevitably requires histology on resected tumor tissue and cannot be addressed on biopsy material only. Thus, neoadjuvant therapy upfront to surgery precludes from determining NI on therapy-naïve tissue and consequently lacks the access to essential prognostic tissue parameters (i.e. neural invasion status).

[0174] Circulating PIGF and sFlt1, and hence PIGF/sFlt1.sup.circ serum ratios, are easily accessible biomarkers which can be repeatedly quantified in the serum of PDAC patients in routine laboratory tests. Since PIGF/sFlt1.sup.circ in the serum of PDAC patients correlate with NI in pancreatic cancer, they can be used as serum biomarkers to predict the risk of NI and to allow identifying those patients with high risk of NI, who will benefit from neoadjuvant therapy and/or anti-PIGF targeted therapy by reducing NI upfront to surgery. Current optimal cut point and ROC analyses bases on our PDAC patients' cohort now determined a cut-off PIGF/sFlt1.sup.circ serum ratio of 0.2404, which allows for discrimination between patients without neural invasion and patients with neural invasion at 92.0% sensitivity and 85% specificity and therefore enables to allocate PDAC patients to prognostic and therapeutic subgroups.

[0175] It is a matter of ongoing debate to what extent neoadjuvant chemotherapy reduced NI, but recent clinical evidence suggests that this effect might be limited. Indeed, neoadjuvant chemotherapy prolongs DFS primarily by increasing time to recurrence of distant metastasis while effects on local recurrence rate and time to local disease recurrence are minor.

[0176] Own data provide evidence that chemotherapy in turn induces expression of PIGF in Schwann cells, and PIGF supports survival and clonal growth of tumor cells (even under conditions of chemotherapeutic treatment), the tumor-nerve interface might be considered as a microenvironmental niche within the tumor stroma, which protects tumor cells from chemotherapeutic treatment and therefore facilitates further metastatic spread of PDAC. Blocking PIGF might therefore constitute a beneficial adjunct to neoadjuvant or adjuvant treatment modalities which (i) prevents PIGF-mediated escape from chemotherapy, (ii) increases efficacy of neoadjuvant chemotherapy and (iii) thereby reduces NI within the tumor cell-neural niche, (vi) ultimately resulting in lower local tumor recurrence rates and improving the hitherto devastating prognosis of pancreatic cancer.