Plasma Kallikrein-Activated TGF-β Is Prognostic for Poor Overall Survival in Patients with Pancreatic Ductal Adenocarcinoma and Associates with Increased Fibrogenesis

Pancreatic ductal adenocarcinoma (PDAC) is a hard-to-treat cancer due to the collagen-rich (fibrotic) and immune-suppressed microenvironment. A major driver of this phenomenon is transforming growth factor beta (TGF-β). TGF-β is produced in an inactive complex with a latency-associated protein (LAP) that can be cleaved by plasma kallikrein (PLK), hereby releasing active TGF-β. The aim of this study was to evaluate LAP cleaved by PLK as a non-invasive biomarker for PDAC and tumor fibrosis. An ELISA was developed for the quantification of PLK-cleaved LAP-TGF-β in the serum of 34 patients with PDAC (stage 1–4) and 20 healthy individuals. Biomarker levels were correlated with overall survival (OS) and compared to serum type III collagen (PRO-C3) and type VI collagen (PRO-C6) pro-peptides. PLK-cleaved LAP-TGF-β was higher in patients with PDAC compared to healthy individuals (p < 0.0001). High levels (>median) of PLK-cleaved LAP-TGF-β were associated with poor OS in patients with PDAC independent of age and stage (HR 2.57, 95% CI: 1.22–5.44, p = 0.0135). High levels of PLK-cleaved LAP-TGF-β were associated with high PRO-C3 and PRO-C6, indicating a relationship between the PLK-cleaved LAP-TGF-β fragment, TGF-β activity, and tumor fibrosis. If these preliminary results are validated, circulating PLK-cleaved LAP-TGF-β may be a biomarker for future clinical trials.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) has a 5-year survival rate of less than 11%, and is expected to be one of the leading causes of cancer death in the next decade [1,2]. PDAC is a very stroma-rich tumor, as the stroma may account for more than 80% of the total tumor mass [3]. The dense stroma is primarily a consequence of tumor fibrosis, also known as desmoplasia, and consists of extracellular matrix (ECM) components, including various collagens, immune cells, endothelial cells, and cancer-associated fibroblasts (CAFs) [3][4][5]. CAFs play a major role in the formation and remodeling of the ECM, including an increase in type III and VI collagen synthesis, which have been shown to be related to short overall survival (OS) in patients with PDAC [6,7]. Under normal conditions, fibroblasts are in a quiescent state, but they can be activated and differentiated into CAFs by factors such Figure 1. Illustration of plasma kallikrein (PLK)-mediated TGF-β activation: TGF-β in its latent state in complex with latency-associated peptide (LAP) (LAP-TGF-β) is bound to the extracellular matrix (ECM) via latent TGF-β binding protein (LTBP). PLK cleaves LAP between 58 and 59 , resulting in the release of active TGF-β and the PLK-cleaved LAP-TGF-β fragment. The active TGF-β will induce multiple biological functions, including tumor progression, metastasis, and collagen formation, leading to tumor fibrosis. The PLK-cleaved LAP-TGF-β fragment is released into circulation, and as a result, becomes detectable in serum and plasma samples through antibody binding. (ECM) via latent TGF-β binding protein (LTBP). PLK cleaves LAP between R 58 and L 59 , resulting in the release of active TGF-β and the PLK-cleaved LAP-TGF-β fragment. The active TGF-β will induce multiple biological functions, including tumor progression, metastasis, and collagen formation, leading to tumor fibrosis. The PLK-cleaved LAP-TGF-β fragment is released into circulation, and as a result, becomes detectable in serum and plasma samples through antibody binding.
In this study, we developed and validated a competitive ELISA for the quantification of PLK-cleaved LAP-TGF-β in circulation ( Figure 1). We evaluated the diagnostic and prognostic value for this biomarker in patients with PDAC, and investigated its association with circulating biomarkers of fibrosis.

Target Identification and Antibody Specificity
PLK-mediated cleavage of LAP-TGF-β at R'58 leads to a fragment of LAP being released into circulation [24]. The peptide 59 LASPPSQGEV 68 from the C-terminal side of the PLK cleavage site was used as a biomarker target for the LAP fragment, reflecting the PLK cleavage of TGF-β ( Figure 1). BLAST was used on the amino acid sequence for homology to other human proteins, using NPS@: Network Protein Sequence Analysis with UniprotKB/Swiss-prot database, and no potential extracellular off-targets were found [32]. A conjugate of the peptide and Keyhole Limpet Hemocyanin (KLH), connected by a cysteine-linker (LASPPSQGEV-GGC-KLH), was used for the immunization of mice to produce monoclonal antibodies in the same manner as described in Nissen et al. [33,34].

Assay Development and Validation
To evaluate the technical performance of the PLK-cleaved LAP-TGF-β ELISA, it underwent the following validation tests: determination of the measuring range, inter-and intra-assay variation, dilutional linearity, spiking accuracy, analyte stability, and interference tolerance of biotin, lipids, and hemoglobin. The measurement range (lower limit of measurement range (LLMR) and upper limit of measurement range (ULMR)) and the inter-and intra-variation were estimated based

Assay Development and Validation
To evaluate the technical performance of the PLK-cleaved LAP-TGF-β ELISA, it underwent the following validation tests: determination of the measuring range, inter-and intra-assay variation, dilutional linearity, spiking accuracy, analyte stability, and interference tolerance of biotin, lipids, and hemoglobin. The measurement range (lower limit of measurement range (LLMR) and upper limit of measurement range (ULMR)) and the inter-and intra-variation were estimated based on 10 independent runs of the assay using quality control samples comprising five human serum samples that span the linear part of the standard curve, three additional human serum samples, and two samples of known peptide concentration spiked into the assay buffer. LLMR and ULMR mark the boundaries of the linear range of the standard curve. Intra-assay variation was summarized as the mean coefficient of variance (CV%) between double-determinations of sample measurements of each assay run for the 10 quality control samples across the 10 independent runs. Inter-assay variation was summarized as the mean CV% between assay runs for the 10 quality controls samples across the 10 independent runs. Dilutional linearity was tested by serially diluting four human serum samples 2-fold and calculating the percentage recovery relative to the undiluted samples.
Spiking accuracy was assessed by spiking a serum sample of high PLK-cleaved LAP-TGF-β concentrations into three human serum samples with low PLK-cleaved LAP-TGF-β concentrations and calculating the percentage recovery relative to the expected concentration in the spiked serum sample.
To determine the analyte stability, PLK-cleaved LAP-TGF-β was measured in three human serum samples after 24 h and 48 h of storage at 4 • C and 20 • C, with storage at −20 • C as a reference. Freeze/thaw stability was determined using PLK-cleaved LAP-TGFβ measurements of human serum samples after one to four freeze/thaw cycles with the samples after one cycle as a reference to calculate the percentage recovery.
As hemoglobin, biotin, and lipids are commonly interfering substances, the influence of these on the assay was investigated by spiking human serum samples with low or high concentrations of the substances (hemoglobin: low = 2.5 mg/mL, high = 5 mg/mL; biotin: low = 3 ng/mL, high = 9 ng/mL; and lipids: low = 1.5 mg/mL, high = 5 mg/mL). The interference was calculated as the percentage recovery of the spiked samples, with non-spiked samples as a reference.

PLK-Cleaved LAP-TGF-β Assay Protocol
The PLK-cleaved LAP-TGF-β assay went through optimization regarding the antibody/coater peptide ratio, assay buffer, incubation time, and temperature, as well as the conjugation of horseradish peroxidase (HRP) to the antibody. For the final protocol, 96-well streptavidin-coated plates were coated with 100 µL of 1.5 ng/mL biotinylated selection peptide (LASPPSQGEV-K-(biotin)) dissolved in assay buffer (25 mM TBS-BTB 2 g NaCl/L, pH 8.0), and incubated at 20 • C in darkness with 300 revolutions per minute (rpm), shaking for 30 min. The plates were washed five times in washing buffer (20 mmol/L TRIS, 50 mmol/L NaCl, pH 7.2), followed by the addition of 20 µL of a 2-fold serial dilution of selection peptide (LASPPSQGEV) starting at 38.75 ng/mL, 1:2 diluted serum sample or 1:4 diluted EDTA plasma sample to the wells, followed by the addition of 100 µL of 33.3 ng/mL HRP-conjugated PLK-cleaved LAP-TGF-β targeting antibody dissolved in assay buffer and incubated at 4 • C in darkness with 300 rpm shaking for 20 h (±1 h). After another five washes in washing buffer, 100 µL of tetramethylbenzidine (TMB) (Kem-En-Tec Diagnostics (Cat. No. 4380)) was added to each well, and the plates were incubated at 20 • C in darkness with 300 rpm shaking for 15 min, followed by the addition of 100 µL of 1% sulfuric acid to stop the reaction. Plates were analyzed using a VersaMax ELISA microplate reader (Molecular Devices, San Jose, CA, USA) at 450 nm, with 650 nm as reference. A standard curve was generated using a four-parametric mathematical fit, and the data were analyzed using GraphPad Prism (version 9).

Assessment of Type III and Type VI Collagen Formation
Formation of type III and type VI collagen was assessed by measuring the biomarkers PRO-C3 (cat. nr. 1700AF06) and PRO-C6 (cat. nr. 4000AF02), respectively, in human serum samples via ELISA [6,35,36]. The biomarkers were measured according to the manufacturer's instructions (Nordic Bioscience A/S, Herlev, Denmark).

Subjects
The PLK-cleaved LAP-TGF-β biomarker was measured in pre-treatment serum samples from patients with PDAC (n = 34) and gender-matched healthy individuals (n = 20). Subject demographics are shown in Table 1, and include: age, gender, number of metastatic sites, body mass index (BMI), stage (American Joint Commission on cancer, 8th edition), and performance status [37,38]. In accordance with the Declaration of Helsinki, version 8, all subjects gave written informed consent. Healthy control samples (including the matched serum and plasma samples) were obtained from Valley BioMedical (Winchester, VA, USA), a commercial vendor with an appropriate institutional review board/independent ethical committee-approved sample collection. For the matched samples, both serum and plasma were obtained from six healthy individuals, also from Valley BioMedical (Winchester, VA, USA). All patients with PDAC were included in the Danish BIOPAC study "BIOmarkers in patients with PAncreatic Cancer (BIOPAC)-can they provide new information of the disease and improve diagnosis and prognosis of the patients?" (ClinicalTrials.gov ID: NCT03311776;). The BIOPAC study is a prospective multicenter open cohort study with ongoing enrollment [38]. from patients prospectively. The serum samples were measured blinded to the clinical information. The patients were followed until May 2022 or death, whichever came first.

Statistics
Biomarker levels in healthy individuals and patients with PDAC or subgroups of patients were compared using the Mann-Whitney test. Wilson/Brown Receiver operating characteristic (ROC) curve analysis was used to further investigate the diagnostic potential of the biomarker. Correlations between PLK-cleaved LAP-TGF-β levels in matched serum and plasma samples were calculated using Pearson correlation. Kaplan-Meier survival analysis was used to assess differences in OS between high (>median) and low (<median) PLK-cleaved LAP-TGF-β biomarker levels in patients with PDAC. The prognostic value of the PLK-cleaved LAP-TGF-β biomarker was further evaluated using univariate and multivariate Cox proportional-hazards regression models, including PLK-cleaved LAP-TGF-β levels (dichotomized at the median), age (continuous scale), and stage (stage 1-3 versus stage 4).
Statistical analyses and graphic designs were made using GraphPad Prism (version 9) and MedCalc (version 19.3) software.

Technical Evaluation of the PLK-Cleaved LAP-TGF-β ELISA
The measurement range (LLMR to ULMR) of the PLK-cleaved LAP-TGF-β assay was determined to be 0.05 to 2.2 ng/mL, with an IC50 of 0.34 ng/mL. The intra-and inter-assay variations were 3% and 11%, respectively. The mean dilution recovery for human serum was 115%, 114%, and 101%, observed from undiluted to a 1:2, 1:4, and 1:8 dilution, respectively, and the mean spiking recovery was 94%. After five freeze/thaw cycles, the analyte recovery in serum was 113%. Analyte recovery from human serum after storage at 4 • C for 48 h was from 98%, demonstrating a much higher stability of the PLK-cleaved LAP-TGF-β fragment than previously shown [31]. For storage at 20 • C for 24 h, the recovery was 88%. Analyte recoveries for the interference of hemoglobin, biotin, and lipids were 101%, 102%, and 98% for low concentrations, and 100%, 105%, and 100% for high concentrations, respectively. Altogether, this demonstrates that the assay is technically stable, robust, and unaffected by the most commonly interfering substances from serum. The technical evaluation results are summaries in Table 2.

Specificity of the PLK-Cleaved LAP-TGF-β ELISA
To determine the antibody specificity of the PLK-cleaved LAP-TGF-β ELISA, the signal inhibition of the selection peptide (LASPPSQGEV), elongated peptide (RLASPPSQGEV), truncated peptide (ASPPSQGEV), non-sense standard peptide (PNASPLLGS), and a nonsense coater peptide (YPNASPLLGS-K-(Biotin)) were used, as shown in Figure 2a. The selection peptide led to a dose-dependent signal inhibition, while the truncated, elongated, and non-sense peptide showed no signal inhibition, and no signal was detected with the non-sense coating peptide. This demonstrates a high specificity of the assay towards the PLK-cleaved LAP-TGF-β neoepitope (LASPPSQGEV).
To further support the specificity and to confirm the ability to target PLK-cleaved LAP-TGF-β, a cleavage experiment was performed. The LAP-TGF-β antibody only showed reactivity towards kallikrein-cleaved LAP-TGF-β, and not intact LAP-TGF-β (Figure 2b).

PLK-Cleaved LAP-TGF-β Levels Were Correlated in Matched Serum and Plasma Samples
Platelet derived TGF-β could result in higher LAP-TGF-β levels in serum compared to plasma [26,27,29,39]. To evaluate this interference, PLK-cleaved LAP-TGF-β was measured in matched serum and citrate-plasma samples. Citrate tubes were used, as these have been shown to result in lower TGF-β levels than EDTA tubes [26].
We did not observe a notable difference in PLK-cleaved LAP-TGF-β levels between serum and plasma samples from the same donors (Figure 3a). A significant correlation between PLK-cleaved LAP-TGF-β measured in serum and plasma was found using Pearson correlation (Pearson's r = 0.90 and p = 0.014) (Figure 3b).

PLK-Cleaved LAP-TGF-β is Elevated in Serum from Patients with PDAC, and Sho Diagnostic Potential
To evaluate the clinical and biological relevance of the PLK-cleaved LAP-TGF-β omarker, it was measured in pretreatment serum samples from 34 patients with PD (stage 1 (n = 1), stage 2 (n = 7), stage 3 (n = 8), and stage 4 (n = 18)), and healthy individu (n = 20). The clinical characteristics are shown in Table 1. The concentration of PL cleaved LAP-TGF-β was significantly elevated (p < 0.0001) in patients with PDAC (med = 1.92 ng/mL, range 0.11-7.48) compared to healthy individuals (median = 0.30 ng/m range: 0.05-1.52) (Figure 4a). In this preliminary study, the biomarker was able to sign cantly discriminate between healthy individuals and patients with PDAC, with an a under the receiver operating characteristic (AUROC) curve of 0.92, p < 0.0001 (Figure 4 further supporting the diagnostic potential of PLK-cleaved LAP-TGF-β.

Overall Survival in Patients with PDAC Is Associated with PLK-Cleaved LAP-TGF-β
The prognostic value of PLK-cleaved LAP-TGF-β was evaluated in the patients with PDAC after dividing the patients into two groups based on PLK-cleaved LAP-TGF-β levels above and below the median, and examining using Kaplan-Meier analysis and uni-and multivariate Cox proportional hazards models for an association with OS. The median OS was 5.2 months (95% CI: 1.5-60.2) in patients with high levels of PLK-cleaved LAP-TGF-β (above median), compared to 19.8 months (95% CI: 5.6-57.1) for patients with low levels of PLK-cleaved LAP-TGF-β (below median). Likewise, when evaluated through univariate analysis, patients with high levels of PLK-cleaved LAP-TGF-β had a significantly decreased OS compared to patients with low levels (log-rank p = 0.013) ( Figure 5).

PLK-Cleaved LAP-TGF-β is Elevated in Serum from Patients with PDAC, an Diagnostic Potential
To evaluate the clinical and biological relevance of the PLK-cleaved LAP-TG omarker, it was measured in pretreatment serum samples from 34 patients with (stage 1 (n = 1), stage 2 (n = 7), stage 3 (n = 8), and stage 4 (n = 18)), and healthy ind (n = 20). The clinical characteristics are shown in Table 1. The concentration cleaved LAP-TGF-β was significantly elevated (p < 0.0001) in patients with PDAC = 1.92 ng/mL, range 0.11-7.48) compared to healthy individuals (median = 0.30 range: 0.05-1.52) (Figure 4a). In this preliminary study, the biomarker was able to cantly discriminate between healthy individuals and patients with PDAC, with under the receiver operating characteristic (AUROC) curve of 0.92, p < 0.0001 (

Overall Survival in Patients with PDAC is Associated with PLK-Cleaved LAP-TGF-β
The prognostic value of PLK-cleaved LAP-TGF-β was evaluated in the patients with PDAC after dividing the patients into two groups based on PLK-cleaved LAP-TGF-β levels above and below the median, and examining using Kaplan-Meier analysis and uniand multivariate Cox proportional hazards models for an association with OS. The median OS was 5.2 months (95% CI: 1.5-60.2) in patients with high levels of PLK-cleaved LAP-TGF-β (above median), compared to 19.8 months (95% CI: 5.6-57.1) for patients with low levels of PLK-cleaved LAP-TGF-β (below median). Likewise, when evaluated through univariate analysis, patients with high levels of PLK-cleaved LAP-TGF-β had a significantly decreased OS compared to patients with low levels (log-rank p = 0.013) (Figure 5).  The association between high levels of PLK-cleaved LAP-TGF-β and poor OS was statistically significant after adjusting for age and the subdivision of metastatic and nonmetastatic tumors (stage 1-3 versus stage 4): (HR: 2.57, 95% CI: 1.22-5.44, p = 0.014).

High PLK-Cleaved LAP-TGF-β Levels Associate with Both Collagen Type III and Type VI Formation in Patients with PDAC
To evaluate the association between collagen formation (fibrosis) and PLK-cleaved LAP-TGF-β, pre-treatment serum levels of type III collagen (PRO-C3) and type VI collagen (PRO-C6), pro-peptides were measured in patients with PDAC. Both PRO-C3 and PRO-C6 were significantly higher in patients with high levels of PLK-cleaved LAP-TGF-β (>median), compared to patients with low levels of PLK-cleaved LAP-TGF-β (<median) (Figure 6). This indicates that PLK-activated TGF-β is associated with increased collagen type III and VI formation in patients with PDAC.

Discussion
In this study, we developed and validated an ELISA assay for measuring PLK-activated TGF-β by targeting the cleaved LAP-TGF-β fragment in serum. Analyte stability and ex vivo activated platelet derived TGF-β are two major issues when attempting to measure TGF-β in serum directly. Large amounts of TGF-β can be released ex vivo from platelets, resulting in false-positive measurements of TGF-β [17,26,39,40]. To prevent the ex vivo activation of platelet derived TGF-β, TGF-β is often measured in plasma instead of serum [26,27]. With the PLK-cleaved LAP-TGF-β assay, we found a similarity and correlation between the PLK-cleaved LAP-TGF-β levels in matched serum and plasma samples, which indicated that the analyte was not influenced by ex vivo platelet derived activation in serum.
We demonstrated the diagnostic and prognostic potential of PLK-cleaved LAP-TGFβ in a preliminary study, with serum samples from 34 patients with PDAC and 20 healthy individuals. We showed that PLK-cleaved LAP-TGF-β was significantly higher in serum from patients with PDAC, compared to healthy individuals. To our knowledge, previously published studies only investigated this fragment in plasma from patients with liver diseases, where it was found to be a potential marker for monitoring the clinical course of chronic liver disease [25,28]. In the present study, high levels of PLK-cleaved LAP-TGF-β (> median) were significantly associated with short OS in patients with PDAC, and independent of age and stage. We also found that high levels of PLK-cleaved LAP-TGF-β were associated with type III and type VI collagen formation, suggesting an association with desmoplasia in PDAC. Both type III and type VI collagen formation have been shown to be associated with other common markers of fibrosis, including α-smooth muscle actin (α-SMA), and they are known to be increased and predictive of poor OS in patients with PDAC [6,37,[41][42][43].
TGF-β has earlier been shown to be elevated in serum from patients with pancreatic

Discussion
In this study, we developed and validated an ELISA assay for measuring PLK-activated TGF-β by targeting the cleaved LAP-TGF-β fragment in serum. Analyte stability and ex vivo activated platelet derived TGF-β are two major issues when attempting to measure TGF-β in serum directly. Large amounts of TGF-β can be released ex vivo from platelets, resulting in false-positive measurements of TGF-β [17,26,39,40]. To prevent the ex vivo activation of platelet derived TGF-β, TGF-β is often measured in plasma instead of serum [26,27]. With the PLK-cleaved LAP-TGF-β assay, we found a similarity and correlation between the PLK-cleaved LAP-TGF-β levels in matched serum and plasma samples, which indicated that the analyte was not influenced by ex vivo platelet derived activation in serum.
We demonstrated the diagnostic and prognostic potential of PLK-cleaved LAP-TGF-β in a preliminary study, with serum samples from 34 patients with PDAC and 20 healthy individuals. We showed that PLK-cleaved LAP-TGF-β was significantly higher in serum from patients with PDAC, compared to healthy individuals. To our knowledge, previously published studies only investigated this fragment in plasma from patients with liver diseases, where it was found to be a potential marker for monitoring the clinical course of chronic liver disease [25,28]. In the present study, high levels of PLK-cleaved LAP-TGF-β (>median) were significantly associated with short OS in patients with PDAC, and independent of age and stage. We also found that high levels of PLK-cleaved LAP-TGF-β were associated with type III and type VI collagen formation, suggesting an association with desmoplasia in PDAC. Both type III and type VI collagen formation have been shown to be associated with other common markers of fibrosis, including α-smooth muscle actin (α-SMA), and they are known to be increased and predictive of poor OS in patients with PDAC [6,37,[41][42][43].
TGF-β has earlier been shown to be elevated in serum from patients with pancreatic cancer, as well as being predictive of poor OS [44,45]. Elevated circulating TGF-β levels have also been shown to predict poor OS in other types of cancers, including pulmonary cancer, hepatic cancer, and leukemia [46][47][48]. These studies measured TGF-β in serum using an ELISA with an antibody that targets either activated TGF-β or the LAP-TGF-β complex directly. These assays will either have the previously discussed issues of analyte stability or will not be able to separate active TGF-β from latent TGF-β which could influence the results, depending on which form of TGF-β the antibody is targeting.
There are and have been many ongoing clinical trials investigating the utility of various anti-TGF-β drugs for use in cancer treatment; however, despite promising preclinical studies, many of these trials fail due to lack of consistency or being unable to recapitulate the data shown in preclinical studies [49][50][51][52]. This demonstrates an ongoing need for better translational TGF-β biomarkers. The PLK-cleaved LAP-TGF-β ELISA cross-reacts to mice, and could be used in pre-clinical studies as well. Thus, the PLK-cleaved LAP-TGF-β biomarker could potentially be used for identifying patients that would benefit the most, and for monitoring the effect of anti-TGF-β drugs in clinical as well as pre-clinical studies.
The present study has several limitations. Since the number of patients with PDAC and healthy individuals are low, future validation in larger cohorts are needed to determine the utility of this biomarker in patients with PDAC. In addition, the group of healthy individuals were slightly younger than the patient group, which can affect the diagnostic potential shown in this study. Furthermore, the PDAC patients included in the BIOPAC study have good performance status, and are more likely to receive anticancer therapy, which may provide selection bias [37]. Another limitation is that the assay only measures the activation of TGF-β that is activated through the PLK-mediated cleavage of LAP. Although we demonstrated the applicability of PLK activated TGF-β in PDAC, TGF-β can also be released from the LAP complex, due to other factors including LAP cleavage mediated by other proteases and integrin interactions [19,[21][22][23].

Conclusions
In summary, we developed a competitive ELISA for the measurement of PLK-activated TGF-β with a high analyte stability and indications of being unaffected by ex vivo platelet derived TGF-β activation. We demonstrated the diagnostic and prognostic value of this biomarker in patients with PDAC. Furthermore, we showed an association between high levels of PLK activated TGF-β, and type III and type VI collagen formation in patients with PDAC, suggesting the involvement of TGF-β in the induction of tumor fibrosis. These are preliminary results, but if validated, circulating PLK-cleaved LAP-TGF-β may be a biomarker for future clinical trials in relation to anti-TGF beta drugs, as well as tumor fibrosis modulation.