Liquid biopsy and biomarkers in pancreatic ductal adenocarcinoma: from concept to clinical translation—a narrative review

Introduction

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal solid malignancies, with a 5-year survival rate of approximately 9–11% despite advances in surgical, medical, and supportive care (1,2). More than 80% of patients are diagnosed at an advanced or metastatic stage when curative treatment is no longer feasible, reflecting non-specific early symptoms and the absence of effective early detection strategies (1-3). Current diagnostic approaches, including cross-sectional imaging, endoscopic ultrasound, and CA19-9, play central roles in clinical management but lack sufficient sensitivity and specificity for reliable early-stage detection or consistent identification of minimal residual disease (MRD) (4-6).

This persistent diagnostic gap has intensified interest in liquid biopsy and circulating biomarkers as minimally invasive tools for earlier detection, molecular characterization, and longitudinal monitoring of PDAC (7,8). Circulating tumor DNA (ctDNA), circulating tumor cells (CTCs), extracellular vesicles (EVs), circulating RNA species, and proteomic or metabolomic signatures each capture distinct aspects of tumor and host biology (7-9). Over the past decade, these modalities have evolved from exploratory assays to increasingly sophisticated platforms capable of integrating multiple analytes from a single blood draw (9,10). This integrative concept has been termed the “circulome”, defined as the comprehensive multidimensional molecular profile derived from all circulating analytes—including ctDNA, EVs, CTCs, RNAs, proteins, and metabolites (9). The major component of this framework is illustrated in Figure 1.

Figure 1 The pancreatic cancer circulome: multi-analyte liquid biopsy integration framework. This schematic illustrates the major circulating biomarker classes released by PDAC and the complementary biological information each contributes to liquid biopsy analysis. CTCs represent cellular dissemination and allow characterization of metastatic potential. ctDNA reflects tumor-specific mutations (such as KRAS and TP53) and methylation changes. EVs, including exosomes, carry miRNAs (e.g., miR-21, miR-155, miR-196a, miR-122), mRNAs, and proteins that mirror tumor-stroma signaling. Additional circulating analytes include serum proteins (CA19-9, THBS2, REG4), noncoding RNAs (miRNAs, lncRNAs, circRNAs), and small-molecule metabolites (amino acids and lipids), which collectively represent systemic metabolic and microenvironmental alterations. Together, these modalities form a multidimensional “circulome” that can be captured through liquid biopsy to support diagnosis, risk stratification, MRD detection, and treatment monitoring in PDAC. Created with BioRender.com. CA19-9, carbohydrate antigen 19-9; circRNAs, circular RNAs; CTCs, circulating tumor cells; ctDNA, circulating tumor DNA; EV, extracellular vesicle; lncRNAs, long non-coding RNAs; miRNAs, microRNA; MRD, minimal residual disease; mRNAs, messenger RNAs; PDAC, pancreatic ductal adenocarcinoma; REG4, regenerating family member 4; THBS2, thrombospondin-2.

Numerous recent reviews have summarized individual liquid biopsy modalities in PDAC (7,8,10). The present narrative review differs in three important respects. First, it focuses specifically on the period from 2018 onward, which has seen rapid maturation of multi-analyte and multi-omics technologies and initial extensive validation studies in defined clinical contexts (10-12). Second, rather than cataloguing biomarkers by modality alone, we emphasize their integration into a unified circulome framework and evaluate performance across discrete clinical scenarios including high-risk surveillance, symptomatic diagnosis, post-resection MRD detection, and treatment monitoring (9-13). Third, we explicitly incorporate discussions of practical implementation barriers, evolving regulatory landscapes, and equity and access considerations—dimensions critical for translation beyond research settings but often underrepresented in earlier syntheses (14,15).

At the same time, it is increasingly evident that impressive analytical performance in controlled research settings does not automatically translate into clinical utility. Many reported diagnostic metrics derive from retrospective or enriched case-control cohorts and may be inflated by spectrum bias, overfitting, and limited external validation (10,16-18). Moreover, performance requirements for screening or surveillance in low-prevalence populations differ fundamentally from those for treatment monitoring or MRD detection (2,14). These considerations underscore the need for cautious interpretation of high area under the curve (AUC) values and for critical appraisal of study design, validation status, and intended clinical use (19,20).

In this context, we present a structured narrative synthesis of contemporary evidence on circulating biomarkers in PDAC. We review the biological rationale, typical performance characteristics, and principal limitations of primary liquid biopsy modalities, and discuss their potential roles across the PDAC care continuum (7,8,19-21). Particular attention is given to sources of bias, challenges in standardization and validation, and the regulatory and economic factors that will shape clinical adoption (14,19,20). Our goal is not to propose practice-ready algorithms, but to provide a balanced, clinically grounded framework for understanding where liquid biopsy in PDAC currently stands and what is required for responsible translation into routine care (1,19,21). We present this article in accordance with the Narrative Review reporting checklist (available at https://apc.amegroups.com/article/view/10.21037/apc-25-41/rc).

Methods

Review design

This article was designed as a structured narrative review to provide a critical, clinically oriented synthesis of recent evidence on liquid biopsy and circulating biomarkers in PDAC. The objective was to integrate findings across diverse biomarker modalities and clinical contexts rather than to perform a formal quantitative synthesis. Accordingly, this review does not follow a systematic review or meta-analysis framework; no PRISMA flow diagram was generated, and no formal risk-of-bias assessment tool [e.g., Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) or Prediction model Risk Of Bias ASsessment Tool (PROBAST)] was applied. The narrative approach was chosen to allow flexible, concept-driven integration of heterogeneous evidence spanning molecular technologies, clinical applications, and implementation considerations (19,20).

Search strategy

We conducted a structured search of PubMed/MEDLINE and Embase for English-language human studies published between January 2018 and January 2026. Search terms combined PDAC descriptors (“pancreatic cancer”, “pancreatic ductal adenocarcinoma”, “PDAC”) with liquid biopsy modalities (“liquid biopsy”, “ctDNA”, “cfDNA”, “circulating tumor cells”, “extracellular vesicles”, “exosomes”, “microRNA”, “non-coding RNA”, “proteomics”, “metabolomics”, “multi-analyte”) using database-appropriate Boolean operators. We also performed manual screening of reference lists from key reviews and guideline papers to identify additional eligible primary studies. A summary of the search strategy and selection process is provided in Table 1. Because this was a narrative review, we did not register a protocol and did not attempt exhaustive retrieval of all eligible studies.

Eligibility and study selection

Eligible studies were clinical investigations in humans evaluating blood- or urine-based biomarkers for PDAC in one or more of the following contexts: symptomatic diagnosis, surveillance of high-risk individuals, detection of MRD, recurrence monitoring, prognostication, or treatment response assessment (1,14,15,19). Biomarker modalities of interest included classical serum markers and panels, ctDNA/cell-free DNA (cfDNA), CTCs, EVs and exosomal RNAs, circulating non-coding RNAs, and proteomic or metabolomic signatures (7-10,20,21).

Both prospective and retrospective studies were considered if they reported quantitative performance metrics such as sensitivity, specificity, AUC, predictive values, hazard ratios (HRs), or lead time over imaging (15,16). Systematic reviews, meta-analyses, narrative reviews, and clinical guidelines were used for contextual background and to help identify primary studies, but were not treated as primary sources of evidence (1,7,8,21). Animal-only studies, single-patient case reports, purely mechanistic in vitro work, conference abstracts without full manuscripts, and studies lacking primary biomarker performance data were excluded. Title/abstract screening and full-text assessment were performed by the authors, with disagreements resolved by discussion.

Data extraction and synthesis

From each eligible study, we extracted the study design, population characteristics, sample type (blood or urine), biomarker modality and assay platform, clinical context (e.g., symptomatic diagnosis, high-risk surveillance, MRD detection, recurrence monitoring, treatment monitoring), comparator groups, and reported performance metrics [e.g., sensitivity, specificity, AUC, positive predictive value (PPV)/negative predictive value (NPV), HRs, or lead time]. We synthesized findings narratively and organized the evidence by biomarker modality and by clinical use-case across the PDAC care continuum. Because of heterogeneity in study designs, cohorts, platforms, and outcome definitions, results are presented as representative ranges rather than pooled estimates.

Classical serum markers and the current baseline

Carbohydrate antigen 19-9 (CA19-9) remains the most widely used serum biomarker in routine management of PDAC, with established roles in supporting diagnosis, estimating prognosis, and monitoring treatment response (1,4-6,22). However, its limitations for early detection are well recognized. Across multiple cohorts, the sensitivity of CA19-9 for early-stage PDAC is modest, and its specificity is reduced by elevation in benign hepatobiliary conditions such as cholestasis, pancreatitis, and other inflammatory disorders (4,5,22,23). In addition, approximately 5–10% of individuals who are Lewis antigen-negative cannot synthesize CA19-9, resulting in false-negative results even in the presence of advanced disease (4,5,22).

Despite these shortcomings, CA19-9 remains the clinical reference standard due to its availability, low cost, and extensive clinical experience (1,5,22). In contemporary practice, its principal value lies in monitoring known disease rather than in screening or early diagnosis (1,22). This historical and ongoing role makes CA19-9 an important benchmark against which any novel biomarker or biomarker panel must be compared, not only in terms of analytical performance but also in terms of reproducibility, cost, and implementation feasibility. Key classical and emerging serum biomarkers, along with representative performance ranges, are summarized in Table 2 (6,24-26).

To overcome the intrinsic limitations of single-analyte testing, numerous studies have evaluated multi-marker serum panels that combine CA19-9 with additional proteins such as thrombospondin-2 (THBS2), regenerating family member 4 (REG4), carcinoembryonic antigen (CEA), carbohydrate antigen 125 (CA125), or prothrombin induced by vitamin K absence-II (PIVKA-II) (6,24-26). In retrospective and enriched case-control cohorts, several of these combinations have demonstrated improved discrimination between PDAC and benign pancreatic disease compared with CA19-9 alone (6,24-26). Similarly, glycoproteomic approaches and exosome-associated protein markers have been proposed as complementary strategies that may partially circumvent the limitations imposed by Lewis antigen status (24).

However, it is important to emphasize that most reported performance gains for serum biomarker panels derive from single-center, retrospective, or case-control studies, often comparing clinically overt PDAC with healthy or highly selected control populations (3,20,21). Such designs are prone to spectrum bias and may overestimate real-world diagnostic accuracy, particularly in clinically challenging scenarios such as differentiating PDAC from chronic pancreatitis or in evaluating asymptomatic high-risk individuals (14,22). Consequently, while multi-analyte serum panels represent a rational and promising evolution beyond CA19-9, none have yet demonstrated sufficient, prospectively validated performance to support population-level screening or to replace CA19-9 in routine clinical workflows (1,14,21).

In this context, CA19-9 should be viewed not as an obsolete marker, but as a pragmatic clinical anchor: it remains useful for longitudinal monitoring of established disease and as a comparator within more complex multi-analyte frameworks (1,5,22). Any new serum-based biomarker strategy must therefore demonstrate not only superior analytical performance in controlled studies, but also reproducibility, clinical utility in realistic patient populations, and acceptable cost-effectiveness before widespread adoption can be justified (1,14,21).

ctDNA and cfDNA

ctDNA, a tumor-derived fraction of circulating cfDNA, is among the most intensively studied liquid biopsy modalities in PDAC, offering minimally invasive access to both mutational and epigenetic tumor features (7,8,27-29). Most mutation-based assays focus on recurrent alterations in genes such as KRAS, TP53, and GNAS (28-32). In retrospective and enriched case-control cohorts, these approaches typically achieve sensitivities of approximately 30–50% for early-stage disease, with performance improving in advanced or metastatic PDAC, where sensitivities often approach or exceed 70–80% in selected series (16,31-34). These differences reflect the strong dependence of ctDNA detectability on tumor burden, vascularization, and biological shedding characteristics (32,34-36).

Despite the near-universal presence of KRAS mutations in PDAC, the specificity of mutation-based ctDNA assays is not absolute. Clonal hematopoiesis of indeterminate potential and inflammatory conditions can release somatic mutations into the circulation that mimic tumor-derived variants, potentially leading to false-positive results (18,29,34). Sensitivity also varies according to tumor stage, volume, anatomical location, and the extent of desmoplastic stroma, which can limit DNA release into the bloodstream (32,34,36).

To address the limited sensitivity of mutation-based assays in early disease, methylation-based ctDNA approaches have been developed to capture epigenetic alterations that often precede overt structural mutations (33,34). In several retrospective and case-control studies, such assays have achieved AUC values above 0.90 for distinguishing PDAC from benign pancreatic disease (33,36-38). In enriched cohorts, early-stage sensitivities exceeding 70% have been reported, representing a substantial improvement over mutation-only strategies. However, these estimates may be inflated by spectrum bias and require confirmation in prospective, clinically realistic settings (33,36,38).

Beyond primary diagnosis, serial ctDNA measurements have shown particular promise in the post-resection and advanced-disease settings. In multiple studies, ctDNA positivity after surgery or during follow-up has preceded radiologic evidence of recurrence by several months, supporting a potential role in MRD detection and early relapse monitoring (15,35,39). Similarly, in advanced disease, dynamic changes in ctDNA levels and mutational profiles can reflect treatment response and the emergence of resistance mechanisms in near real time (18,29,38).

Important technical and clinical challenges continue to limit widespread implementation. Early-stage PDAC typically sheds ctDNA at very low allele fractions, necessitating highly sensitive techniques such as digital polymerase chain reaction (PCR) or deep sequencing (29,34,36). Pre-analytical variability in blood collection, processing, and storage can further compromise reproducibility, while assay cost and technical complexity raise concerns about scalability in routine clinical practice (19,20,34). As a result, the current evidence suggests a stage-dependent performance profile: modest detection rates in early disease that may improve when ctDNA is combined with protein or other biomarker classes, and greater robustness in advanced disease and MRD settings, where ctDNA more reliably reflects tumor dynamics (37,38).

Several sources of bias must be considered when interpreting reported ctDNA performance. Lead-time bias is particularly relevant in MRD studies, as earlier molecular detection of recurrence does not necessarily translate into improved survival without effective intervention (15,35). Spectrum bias is also common, as many diagnostic studies compare clinically overt PDAC with healthy controls rather than with clinically realistic comparator groups such as chronic pancreatitis or new-onset diabetes, inflating apparent accuracy (3,14,21). Finally, reported performance is often derived from retrospective or enriched cohorts, underscoring the need for prospective, multi-center validation to define the actual clinical utility of ctDNA across different PDAC care pathways (1,19,21).

Taken together, ctDNA currently appears closest to routine clinical application in the post-resection MRD and advanced-disease monitoring settings, where its ability to track molecular disease dynamics may inform therapeutic decision-making (15,35,39). In contrast, limited sensitivity in early-stage disease remains the principal barrier to its broader use in screening or as a first-line diagnostic test (1,14,21). A comparative overview of liquid biopsy modalities is provided in Table 3.

CTCs

CTCs represent intact malignant cells that have detached from the primary tumor or metastatic deposits and entered the bloodstream, providing a potential window into tumor dissemination, metastatic competence, and therapeutic resistance (28,40,41). In PDAC, however, CTCs are considerably less abundant than in several other solid tumors, reflecting the dense desmoplastic stroma, low vascular permeability, and anatomical features of the pancreas that limit tumor cell intravasation (40,41).

Across most published studies, CTC detection rates in PDAC are highly variable and strongly dependent on disease stage, assay platform, and enrichment strategy (40,42-44). In retrospective cohorts and case-control designs, CTCs are detected in only a minority of patients with localized disease, with higher detection rates in advanced or metastatic PDAC (42,43,45). As a result, reported sensitivities for diagnostic purposes are generally inferior to those achieved with ctDNA or multi-analyte approaches, particularly in early-stage disease (28,29,40). For this reason, CTC-based assays have not demonstrated sufficient performance to support roles in screening or first-line diagnosis (19,20,40).

Where CTCs may offer particular value is in prognostication and biological characterization rather than in primary detection. Multiple studies have shown that the presence or persistence of CTCs is associated with poorer survival, higher metastatic burden, and increased risk of recurrence after surgery or systemic therapy (42,43,45). Because CTCs are intact cells, they can, in principle, provide information on protein expression, cellular phenotype, epithelial-mesenchymal transition, and, in selected settings, enable single-cell genomic or transcriptomic analyses that are not accessible through cell-free analytes (46,47). Advanced ex vivo models, such as bioengineered three-dimensional (3D) co-cultures using tunable peptide amphiphile extracellular matrices, have been developed to more faithfully recapitulate the in vivo PDAC microenvironment, including cancer stem cell niches and stromal interactions, thereby improving the biological relevance of biomarker discovery and drug-testing platforms (48).

Nevertheless, significant technical and biological challenges limit the clinical utility of CTCs in PDAC. The extreme rarity of these cells in circulation necessitates complex and expensive enrichment platforms, and no single capture strategy reliably recovers the full spectrum of phenotypically diverse tumor cells (41,44,49). Epitope-based methods may miss cells that have undergone epithelial-mesenchymal transition, whereas size- or deformability-based approaches are prone to contamination by non-malignant cells (41,44,49). In addition, pre-analytical factors such as blood volume, processing time, and operator-dependent variability further compromise reproducibility (40,44,49).

Several sources of bias also affect the interpretation of CTC studies in PDAC. Many reports are based on small, single-center, retrospective cohorts and compare advanced PDAC with healthy controls rather than with clinically realistic comparator groups (19,20,40). This design inflates apparent discriminatory performance and limits generalizability. Moreover, the biological meaning of CTC detection is context-dependent. While the presence of CTCs may correlate with aggressive disease biology, it does not necessarily translate into actionable clinical decisions in the absence of validated intervention strategies (50).

In summary, although CTCs offer unique opportunities to study tumor biology and metastatic processes in PDAC, their current clinical utility is primarily confined to research settings and exploratory prognostic applications (19,20,40). Compared with ctDNA and multi-analyte liquid biopsy approaches, CTCs are unlikely to play a significant role in early detection or routine disease monitoring in the near term. However, they may contribute complementary biological information within integrated multi-modal frameworks (47,51).

EVs and exosomes

EVs, including exosomes, are membrane-bound nanoparticles actively released by both tumor and stromal cells and present in most biological fluids. They carry proteins, lipids, and nucleic acids that reflect the molecular and functional state of their cell of origin (52-54). In PDAC, EVs have attracted particular interest as liquid biopsy substrates because their lipid bilayer protects enclosed cargo—such as microRNAs (miRNAs), messenger RNAs (mRNAs), and proteins—from enzymatic degradation, and because they capture aspects of tumor-stroma and tumor-host signaling that are not fully represented by ctDNA alone (52-54).

Early EV-based diagnostic studies in PDAC reported strikingly high performance, driven in part by the identification of tumor-associated exosomal surface proteins and RNA cargos (52,53). However, subsequent work has demonstrated that EV biomarker performance is susceptible to pre-analytical variables, isolation methods, and analytical pipelines, leading to substantial inter-study heterogeneity and, in some cases, inconsistent results (52-54). These methodological issues prompted the development and iterative refinement of the Minimal Information for Studies of Extracellular Vesicles (MISEV) guidelines, which aim to standardize EV isolation, characterization, and reporting and to improve reproducibility across laboratories (52,53,55).

More recently, multi-marker EV-based approaches have shown more robust and reproducible performance than single-analyte strategies. Panels combining exosome-associated proteins or RNAs—such as EPCAM, EGFR, MUC1, ALPPL2, and selected miRNAs—have demonstrated improved discrimination between PDAC and benign pancreatic disease compared with CA19-9 alone in several retrospective and case-control cohorts (11,24,27,52,55-57). In optimized research settings, some of these EV-based assays have reported AUC values exceeding 0.90 for distinguishing PDAC from chronic pancreatitis or healthy controls (27,52,55-57).

However, it is essential to interpret these results cautiously. The majority of published EV biomarker studies in PDAC are single-center, retrospective, or based on enriched case-control populations, which are particularly susceptible to spectrum bias and overestimation of diagnostic accuracy (21,32,50). Few EV-based assays have been evaluated in prospective, unselected, or surveillance-embedded cohorts, and independent external validation remains limited (21,32,50).

Beyond diagnosis, EVs are biologically attractive substrates for longitudinal disease monitoring because of their abundance and stability in circulation. Changes in exosomal RNA or protein cargo have been associated with tumor burden, treatment response, and disease progression in exploratory studies, suggesting potential roles in treatment monitoring and MRD assessment (55-57). However, most such applications remain investigational, and there is currently no definitive evidence that EV-guided monitoring improves clinical outcomes (21,32,50).

Several technical and translational barriers continue to limit clinical implementation. EV isolation remains labor-intensive and method-dependent, with significant contamination from lipoproteins and non-tumor vesicles in many workflows (52-54). Quantification and normalization strategies are not yet standardized, and inter-platform variability remains substantial (52,55). These factors complicate assay harmonization, regulatory approval, and large-scale deployment in routine clinical practice (21,32,50).

In summary, EVs and exosomes represent a biologically rich and promising component of the liquid biopsy landscape in PDAC. At present, however, their most significant potential lies in combination with other biomarker classes within multi-analyte or multi-omics platforms rather than as standalone diagnostic tools. Robust prospective validation, methodological standardization, and demonstration of clinical utility will be essential before EV-based assays can be responsibly integrated into routine PDAC care (10,21,32,50).

Circulating RNA biomarkers

Circulating RNA species represent a diverse and information-rich class of liquid biopsy analytes with potential applications in PDAC (32,58,59). Among these, miRNAs are the most extensively studied because of their relative stability in circulation—conferred by encapsulation within EVs or association with carrier proteins—and their close links to oncogenic signaling pathways (27,52,53,58). In retrospective and optimized case-control cohorts, multi-miRNA panels have often outperformed CA19-9 in distinguishing PDAC, including early-stage disease, from healthy or benign controls, with several studies reporting AUC values around or exceeding 0.90 in these selected settings (12,27,55,58).

However, as with other circulating biomarkers, it is essential to interpret these results in light of the study design. Most high-performing miRNA signatures have been developed and tested in small, single-center, retrospective cohorts using highly selected control groups. This setting is prone to spectrum bias and overestimation of diagnostic accuracy (21,27,55). When evaluated in more heterogeneous or clinically realistic populations, performance often attenuates (21,32). Incorporating miRNAs into multivariate models alongside CA19-9 and other clinical variables has consistently improved discriminatory performance in several studies, suggesting that RNA markers may be most useful as components of composite panels rather than as standalone tests (12,32,59).

Beyond miRNAs, long non-coding RNAs (lncRNAs) and circular RNAs (circRNAs) provide additional layers of biological and diagnostic information (58,60). Several circulating lncRNAs correlate with tumor stage, metastatic burden, or prognosis, potentially offering greater tissue specificity in selected contexts (21,58). CircRNAs, owing to their covalently closed structure, exhibit exceptional stability and are often enriched in EVs; specific circRNA species have been associated with advanced disease features and poorer survival in exploratory studies (58,60). Nevertheless, the clinical evidence supporting these RNA classes remains preliminary and confined mainly to discovery-phase investigations (21,58,60).

Reproducibility represents the principal barrier to clinical translation of circulating RNA biomarkers. Substantial variability arises from pre-analytical factors such as sample collection, processing, and storage, as well as from differences in RNA extraction methods, normalization strategies, and analytical pipelines (27,34,55). Small cohort sizes, multiple hypothesis testing, and statistical overfitting further contribute to inconsistent findings across studies (21,32,58). Although standardized and automated RNA extraction and quantification workflows could mitigate some of these issues, they are not yet uniformly adopted (34,55).

Several sources of bias warrant explicit consideration. Many RNA biomarker studies lack external validation and have not been tested against clinically realistic comparator groups such as chronic pancreatitis or new-onset diabetes (21,27,32). In addition, impressive performance metrics are often reported without a clear distinction between training and validation cohorts, raising concerns about optimistic bias (21,32,58). These limitations underscore the need for larger, multi-center, prospectively designed studies with predefined analytical and clinical endpoints (21,32,59).

In current clinical practice, circulating RNA biomarkers—particularly miRNA-based panels—are best viewed as adjunctive tools that may complement established markers and imaging in selected diagnostic dilemmas, rather than as standalone diagnostic tests (1,21,32). Their greatest potential likely lies in integration into multi-analyte or multi-omics platforms, where they can contribute biologically complementary information to more robust composite signatures (10,12,59,60).

Protein and metabolomic signatures

Proteomic and metabolomic biomarkers aim to capture downstream functional consequences of tumor biology and tumor-host interactions that may not be fully reflected by nucleic-acid-based assays alone (32,34,50). In PDAC, numerous studies have explored circulating protein panels, glycoproteomic signatures, and metabolite profiles as tools for diagnosis, risk stratification, and disease monitoring (24,61-63). Because these analytes are closer to phenotype than genomic or epigenomic alterations, they offer a biologically attractive complement to ctDNA, EVs, and circulating RNA markers (32,34,50).

Several multi-protein panels, often incorporating CA19-9 together with markers such as THBS2, REG4, and other inflammatory or tumor-associated proteins, have demonstrated improved discrimination between PDAC and benign pancreatic disease in retrospective and enriched cohorts (6,24-26,61,64-66). Similarly, glycoproteomic approaches that interrogate cancer-associated changes in protein glycosylation patterns, as well as metabolomic signatures reflecting altered amino acid and lipid metabolism, have yielded promising results in exploratory studies (61-63). In optimized case-control settings, some of these platforms report AUC values in ranges typically considered good to excellent for diagnostic tests, although these estimates largely reflect discovery-phase studies in selected populations (61-64).

However, as with other liquid biopsy modalities, most reported performance metrics derive from single-center, retrospective, or highly selected study populations (21,32,34,50). Proteomic and metabolomic measurements are susceptible to pre-analytical and analytical variability, including differences in sample collection, storage, batch effects, and platform-specific processing pipelines (34,62,63). These factors complicate cross-study comparisons and contribute to limited reproducibility. Moreover, many reported biomarker signatures are developed using high-dimensional data from relatively small cohorts, increasing the risk of statistical overfitting and optimistic bias (21,32,34).

From a clinical perspective, protein and metabolite biomarkers are unlikely to replace established imaging and histopathological diagnostics as standalone tests (1,21,50). Current data therefore support the use of serum biomarker strategies for early PDAC detection mainly as adjuncts to imaging and structured risk assessment, rather than as standalone screening tools (1,21,70). Their most significant potential lies in combination strategies, where they can complement nucleic-acid-based markers by capturing systemic responses, tumor-stroma interactions, and metabolic reprogramming that are not directly accessible through ctDNA or RNA analysis (10,12,32,34). Indeed, several multi-analyte platforms that integrate protein or metabolite features with genomic or epigenomic data have demonstrated superior performance compared with single-modality approaches in exploratory validation studies including recent mRNA-based liquid biopsy assays evaluated in resectable PDAC (10-12,64-68).

Several sources of bias and limitations must be acknowledged. Many proteomic and metabolomic studies lack independent external validation and are tested against non-representative control groups, inflating apparent diagnostic accuracy (21,32,50). In addition, the biological specificity of some proteins and metabolic changes is limited, as similar alterations can occur in inflammatory or non-malignant conditions affecting the pancreas or hepatobiliary system (21,22,50). These factors underscore the need for larger, multi-center, prospectively designed studies with standardized analytical workflows and predefined clinical use-cases (21,32,34).

In summary, proteomic and metabolomic biomarkers represent an important and biologically informative component of the liquid biopsy landscape in PDAC (32,34,50). At present, however, their role is best viewed as complementary within multi-analyte frameworks rather than as independent, practice-ready diagnostic tools (10,21,32,34).

Multi-analyte and multi-omics platforms and the circulome

Multi-analyte and multi-omics platforms aim to overcome the limitations of single-modality liquid biopsy by integrating complementary signals from ctDNA (mutation and methylation), EV proteins and RNAs, circulating RNAs, and serum protein or metabolite features into unified classifiers. This integration is often described within a “circulome” framework, in which multiple circulating analyte classes are jointly interpreted to better reflect tumor biology and tumor-host interactions from a single blood draw (9,10).

Across enriched and case-control cohorts, multi-analyte approaches frequently outperform CA19-9 and single-modality assays, particularly for discriminating PDAC from benign pancreatic disease and for detecting earlier-stage disease (10-12,37,64-67). Notably, several platforms combine protein markers (e.g., CA19-9, THBS2, REG4) with nucleic-acid features (ctDNA or circulating RNA signatures), yielding higher AUC values than either component alone in retrospective or single-center validations (10-12,37,64-66).

However, the apparent performance of multi-omics models is highly sensitive to study design and validation rigor. Many studies develop high-dimensional models using relatively small datasets, increasing the risk of overfitting, spectrum bias, and optimistic estimates if external validation is limited or absent (21,32,34,50). In addition, differences in pre-analytical handling (collection tubes, processing delays), platform-specific batch effects, and normalization strategies can reduce reproducibility across laboratories and populations (29,34,50).

For clinical translation, the key requirement is not only high discrimination but also demonstrated clinical utility in defined use-cases—for example, high-risk surveillance adjuncts, symptomatic diagnosis in indeterminate imaging, or post-resection MRD strategies—tested prospectively in realistic cohorts with prespecified thresholds and decision-impact endpoints (15,17,21,32,35). Until such evidence is established, multi-analyte and multi-omics assays should be interpreted as promising but investigational tools whose ultimate value will depend on standardization, external validation, cost-effectiveness, and equitable implementation (1,21,32,34,50).

Clinical contexts and diagnostic performance

The clinical value of any liquid biopsy biomarker in PDAC depends not only on its statistical discriminatory performance, but also on the specific clinical context in which it is applied (1-3,10,21,32,34,50). Although many studies emphasize AUC as a summary metric, AUC alone does not define clinical utility (21,32,34,50). In low-incidence settings such as screening or high-risk surveillance, even tests with excellent AUC values may generate unacceptable numbers of false positives unless specificity is exceptionally high (2,3,14,21). Nonetheless, prospective surveillance cohorts of high-risk individuals have reported modest but clinically meaningful yields of PDAC and suggest a shift toward more resectable disease and improved survival when cancers are detected under surveillance (2,3,14,69,71). Conversely, in post-resection surveillance or MRD detection, sensitivity becomes the dominant requirement, as missed recurrences carry significant clinical consequences (15,35,50). For this reason, clinically meaningful implementation requires context-specific performance thresholds rather than relying solely on AUC (21,32,34,50).

High-risk surveillance

High-risk surveillance is among the most frequently proposed applications of liquid biopsy, particularly for individuals with familial risk, pathogenic germline variants (e.g., BRCA, PALB2, ATM), or syndromic predisposition such as Peutz-Jeghers syndrome (1-3,14). In this setting, disease prevalence remains low on an annual basis, meaning that acceptable screening tools must achieve extremely high specificity—often approaching or exceeding 99%—to avoid excessive false-positive findings, unnecessary invasive procedures, and psychological harm (2,3,14,21).

Several ctDNA methylation assays and EV-derived miRNA or multi-analyte panels have shown encouraging performance in enriched cohorts when combined with magnetic resonance imaging (MRI) or endoscopic ultrasound (11-13,37,64,65). However, most supporting studies are retrospective or case-control in design, and true prospective, surveillance-embedded validation remains limited (14,21,32). As a result, while these biomarkers may eventually contribute to risk-adapted surveillance strategies, they cannot yet be considered substitutes for imaging-based programs and should be viewed as investigational adjuncts (1,14,21).

Symptomatic diagnosis and indeterminate imaging

In patients with symptoms or indeterminate pancreatic lesions on imaging, liquid biopsy biomarkers may have a different, more immediately plausible role: helping clarify diagnostic uncertainty or prioritize further testing (1-3,21,32,50). In this context, acceptable specificity remains high, but modest reductions in specificity may be tolerable if sensitivity improves and if the test reduces diagnostic delay or unnecessary procedures (21,32,50).

Protein panels, EV-derived miRNA signatures, and combined multi-analyte approaches have shown improved discrimination between PDAC and chronic pancreatitis in retrospective cohorts (6,24-27,37,52,64-67). Nonetheless, many reported performance metrics are derived from comparisons against healthy controls rather than clinically realistic differential diagnoses, which likely inflates apparent accuracy (2,3,21,32,50). Prospective studies embedded in real diagnostic pathways are needed before such tests can be confidently incorporated into routine practice (1,21,32,34,50).

MRD detection after resection

Detection of MRD after curative-intent surgery is currently the most clinically mature and actionable application of ctDNA-based assays in PDAC (15,35,50). Multiple studies demonstrate that ctDNA positivity after resection or during follow-up often precedes radiologic recurrence by several months, supporting its role as an early molecular indicator of relapse (15,35). In this context, the dominant requirement is near-perfect specificity—since false positives could lead to unnecessary treatment escalation—combined with high sensitivity for low-volume disease (15,35,50).

While early results are promising, it remains essential to recognize that lead-time bias may exaggerate perceived benefit unless earlier molecular detection is shown to translate into improved survival or quality of life (21,32,50). Ongoing and recent studies evaluating ctDNA-guided adjuvant or salvage therapy strategies will be critical to establishing the clinical utility of these approaches (15,35).

Treatment monitoring and response assessment

In advanced disease, serial ctDNA or circulating RNA measurements may provide dynamic information on treatment response, emerging resistance, and clonal evolution (16,18,29,32,34,35). Here, the clinical goal is not early detection but timely identification of non-responders or resistance mechanisms. In this setting, moderate reductions in specificity may be acceptable if earlier treatment adaptation improves outcomes (29,32,34,50).

Although several studies suggest that changes in circulating biomarkers can precede radiologic progression, most evidence remains observational, and integration into routine treatment decision-making remains investigational (18,29,32,34,35).

Summary perspective

Taken together, these use cases illustrate why reported performance metrics must always be interpreted in light of clinical context, study design, and disease prevalence (21,32,34,50). A biomarker that performs well in one scenario may be clinically unusable in another (21,32,50). Table 4 summarizes representative studies across these major clinical contexts, but it is important to emphasize that most current evidence supports adjunctive or investigational use rather than guideline-endorsed implementation (1,21,32,50).

Figure 2 presents a conceptual and purely illustrative framework for how investigational liquid biopsy modalities might be integrated across the PDAC care continuum in different clinical scenarios (1,21,32,34,50). This schematic is not intended to represent a clinical algorithm or practice guideline, but rather to summarize potential use-cases discussed in the literature and to highlight where future prospective validation and decision-impact studies would be required (21,32,34,50).

Figure 2 Conceptual framework for investigational liquid biopsy integration across the PDAC care continuum. This figure presents a conceptual and illustrative framework showing how circulating biomarkers might be investigated across different stages of PDAC care. The pathways shown are hypothesis-generating and not guideline-endorsed clinical recommendations. In high-risk individuals, investigational multi-analyte panels could potentially be explored as adjuncts to imaging-based surveillance to inform surveillance intensity. In patients with indeterminate imaging findings, combinations of CA19-9, protein panels, and EV-miRNA or ctDNA assays may be studied as adjunctive tools to help refine diagnostic assessment. In the post-resection setting, serial ctDNA testing may allow investigational detection of MRD and molecular relapse prior to radiologic recurrence, potentially supporting risk stratification, closer surveillance, or consideration of clinical trials or multidisciplinary review. During systemic therapy, longitudinal ctDNA or circulating RNA measurements may provide exploratory insights into treatment response or emerging resistance. All proposed applications remain investigational and require prospective validation, demonstration of clinical utility, and regulatory approval before any routine clinical implementation. CA19-9, carbohydrate antigen 19-9; ctDNA, circulating tumor DNA; EV-miRNA, extracellular vesicle microRNA; MDT, multidisciplinary team; MRD, minimal residual disease; PDAC, pancreatic ductal adenocarcinoma.

Practical implementation challenges

Despite the rapid scientific maturation of liquid biopsy technologies, multiple practical barriers continue to limit their widespread clinical implementation in PDAC (21,29,32,34,50). One of the most important challenges is pre-analytical variability. Differences in blood collection tubes, processing delays, storage conditions, and isolation protocols for ctDNA, EVs, or circulating RNAs can produce substantial and sometimes systematic differences in measured biomarker levels (29,34,50,52,55). These effects are especially pronounced for low-abundance analytes and contribute significantly to inter-study inconsistency and limited reproducibility (21,34,50,55).

Analytical heterogeneity further complicates translation. Platforms for sequencing, PCR, proteomics, metabolomics, and vesicle characterization differ widely in sensitivity, specificity, and quantitative stability (29,34,50,51,62). Cutoff definitions, normalization strategies, and batch effects remain insufficiently standardized, and many published studies incompletely report pre-analytical and analytical workflows, making independent replication difficult (21,29,50,55). These methodological issues are a significant reason why promising signatures often fail to generalize beyond the discovery setting (21,32,34,50).

Economic and infrastructural considerations represent an equally important barrier. A single-analyte test, such as CA19-9, typically costs on the order of tens to a few hundred US dollars. In contrast, comprehensive multi-omics assays that combine ctDNA sequencing, EV profiling, and proteomic or metabolomic mass spectrometry can cost approximately 1,000–3,000 USD per test, depending on the platform and region (21,29,34,50). In practice, this cost and infrastructure gap means that patients treated in large academic or well-resourced health systems may gain earlier access to multi-omics testing, while those in lower-resource or rural settings remain limited to CA19-9 and imaging, potentially widening existing survival disparities in PDAC (1,21,32,50). In addition, such assays require specialized laboratory infrastructure, bioinformatics pipelines, and trained personnel, and are currently available mainly in large academic centers or specialized reference laboratories (29,34,50).

Without deliberate attention to cost control and implementation strategy, the adoption of advanced liquid biopsy platforms risks widening existing disparities in pancreatic cancer outcomes (1,21,32,50). Patients in low-resource settings, rural areas, or health systems without advanced molecular diagnostics infrastructure may have limited or no access to these technologies (21,32,50). Thus, innovations intended to improve early detection and personalized care could paradoxically exacerbate inequities if deployed without parallel efforts to ensure affordability, scalability, and geographic accessibility (21,32,50).

Regulatory pathways constitute an additional bottleneck. Only a small number of liquid biopsy assays relevant to PDAC are currently cleared or approved by major regulatory agencies, and most remain available only as laboratory-developed tests or research-use-only platforms (1,21,29,32,50). For complex multi-analyte assays, demonstrating analytical validity, clinical validity, and clinical utility is particularly challenging, especially in the absence of standardized reagents, workflows, and performance benchmarks (21,29,34,50).

Finally, workflow integration and turnaround time must be considered. Tests that are expensive, slow, or difficult to interpret are unlikely to be adopted in routine clinical pathways, particularly in time-sensitive diagnostic or treatment-monitoring settings (21,29,32,50). For liquid biopsy to achieve meaningful clinical impact in PDAC, future development must focus not only on improving analytical performance but also on simplifying workflows, reducing costs, and ensuring equitable access (21,32,34,50).

Regulatory landscape and guidelines

Regulatory oversight of liquid biopsy diagnostics continues to evolve, particularly as assays become increasingly complex and incorporate multi-analyte and algorithmic components (1,21,29,32,34,50). In most jurisdictions, these tests are regulated as in vitro diagnostics and must demonstrate analytical validity, clinical validity, and, for broad clinical adoption, evidence of clinical utility (1,21,29,34,50). For assays intended to serve as companion diagnostics, additional regulatory requirements ensure appropriate alignment with specific therapeutic indications (1,21,29).

At present, only a limited number of liquid biopsy assays have received regulatory clearance or approval in oncology, and most are focused on advanced-disease genotyping rather than early detection or surveillance (1,21,29,32,34). In PDAC, the vast majority of ctDNA, EV, circulating RNA, and multi-analyte platforms remain investigational. They are available primarily as laboratory-developed tests or research-use-only assays (1,21,29,32,50). This regulatory status reflects not only technical complexity but also the absence of definitive prospective evidence demonstrating improved patient outcomes (21,32,34,50).

For EV-based assays, the International Society for EVs has played an important role in promoting standardization through the MISEV guidelines, which define minimal requirements for vesicle isolation, characterization, and reporting (52,53). The updated MISEV2023 recommendations further emphasize the need for rigorous methodological transparency, acknowledging that inconsistent practices have contributed substantially to irreproducibility in the field (52,53,55). Many PDAC-focused EV biomarker studies still fall short of full guideline adherence, which remains a significant barrier to regulatory acceptance (21,50,52,55).

Current clinical guidelines for pancreatic cancer management generally regard liquid biopsy approaches as investigational for screening, early detection, or routine surveillance (1,17,21,50). Their use is typically limited to research settings or selected circumstances in advanced disease when tissue biopsy is infeasible (1,17,29,50). This conservative stance appropriately reflects the current evidence base, which is dominated by retrospective and exploratory studies rather than prospective, decision-impact trials (21,32-34,50).

Guideline adoption will likely occur in a stepwise and context-dependent manner (1,17,21,32). For example, ctDNA-based MRD detection after resection may be considered clinically earlier than multi-omics screening platforms, provided that ongoing trials demonstrate improved outcomes with biomarker-guided interventions (15,17,18,35). More complex multi-analyte and algorithm-driven tests will require robust evidence not only of diagnostic accuracy, but also of clinical benefit, cost-effectiveness, and reproducibility across diverse healthcare settings (21,32,34,50).

Overall, regulatory and guideline frameworks remain a critical gatekeeper for the responsible clinical translation of liquid biopsy technologies in PDAC (1,21,32,34,50). Continued progress will depend on methodological standardization, transparent reporting, and prospective trials designed to demonstrate not only earlier detection but also meaningful improvements in patient outcomes (21,32,34,50).

Strengths and limitations

This narrative review has several strengths. We performed a structured search across multiple major databases using predefined eligibility criteria, supplemented by manual reference screening, to ensure broad coverage of the recent literature on liquid biopsy biomarkers in PDAC (7,8,10,21,29,32,34). The synthesis integrates evidence across diverse biomarker modalities—including ctDNA, CTCs, EVs, circulating RNAs, and proteomic and metabolomic signatures—and explicitly frames their performance within clinically relevant contexts such as high-risk surveillance, symptomatic diagnosis, MRD detection, and treatment monitoring (1,10,17,21,32,50). In addition, we extend beyond technical performance to discuss regulatory, economic, and equity considerations, which are critical for responsible clinical translation (1,21,32,34,50).

However, important limitations must be acknowledged. Although the literature search was structured, this work is intentionally a narrative rather than a systematic review (7,8,10,21). No formal PRISMA workflow or quantitative meta-analysis was performed, and no standardized risk-of-bias or quality assessment tool was applied to individual studies (21,32,34). As a result, the review does not provide a comprehensive, quantitatively weighted assessment of the entire evidence base, and the conclusions should be interpreted as qualitative and contextual rather than definitive (21,32,34).

The included literature is heterogeneous with respect to study design, patient populations, comparator groups, assay platforms, and outcome definitions, which complicates direct comparison across modalities and likely inflates apparent performance in some settings (10,21,32,34,50). Many cited studies are single-center, retrospective, or case-control in design, and publication bias toward positive findings is likely (10,21,32,34). Furthermore, the review focuses primarily on studies published from 2018 onward and may therefore underrepresent earlier foundational work, which we cite selectively for historical context (4,5,23,37,63).

Finally, most available studies emphasize diagnostic or surrogate endpoints rather than hard clinical outcomes such as survival or quality of life, and robust cost-effectiveness data are scarce (1,17,21,32,34,50). These limitations underscore the need for cautious interpretation and for prospective, multi-center trials to define the actual clinical value of liquid biopsy approaches in PDAC (15,17,18,21,32,34,35).

Future perspectives

The future trajectory of liquid biopsy in PDAC is likely to be shaped less by the discovery of additional candidate biomarkers and more by rigorous validation, standardization, and implementation science (1,21,32,34,50). Although the number of proposed circulating biomarkers continues to grow, relatively few have progressed beyond exploratory or single-center evaluation (10,21,32,34). Harmonization of pre-analytical workflows—including standardized blood collection tubes, processing intervals, storage conditions, and vesicle isolation methodologies—remains a fundamental prerequisite for reproducibility and for meaningful comparison across studies (19,21,52-55). Without such standardization, even analytically impressive biomarker signatures are unlikely to translate into routine clinical practice (21,32,34,50).

Multi-analyte and multi-omics integration will underpin the next significant advances in the field (9,10,34,50,59). Individual modalities such as ctDNA, EVs, or circulating RNAs have not, on their own, demonstrated sufficiently robust and consistent performance for broad clinical use (19-21,32,34). In contrast, combined panels interpreted through machine-learning approaches have repeatedly shown improved discriminatory ability in enriched cohorts (10-12,37,62,64-67). The concept of a circulome-based assay—where layered molecular information is extracted from a single blood sample—offers a compelling biological and technological framework (9,10,34,59). However, future development must prioritize transparent model construction, avoidance of overfitting, external validation, and prospective testing in clinically realistic populations (21,32,34,50), following reporting and evaluation frameworks such as TRIPOD and PROBAST.

Equally important is the question of scalability and equity. If advanced liquid biopsy platforms remain expensive, technically complex, and confined to specialized centers, their adoption risks widening existing disparities in PDAC outcomes (1,21,32,34,50). Future innovation should therefore emphasize simplified, cost-conscious assay designs, miniaturized or automated platforms, and workflows that can be deployed beyond major academic institutions (21,32,34). Point-of-care or near-patient testing strategies, although still aspirational, represent an important long-term goal for broadening access (34,50,59).

The most decisive step forward will be the conduct of pragmatic, prospective trials embedded in real clinical pathways (15,17,18,21,32,35). Such studies must evaluate not only diagnostic accuracy, but also downstream effects on clinical decision-making, patient outcomes, quality of life, and cost-effectiveness (1,17,21,32,34,50). In high-risk surveillance and post-resection monitoring in particular, it will be essential to demonstrate that earlier molecular detection leads to interventions that meaningfully improve survival or other patient-centered outcomes (15,17,18,35).

Finally, regulatory and guideline integration will depend on the accumulation of this higher-level evidence (1,17,21,50). Stepwise adoption is likely, with narrower, more clearly defined use cases—such as MRD detection—preceding broader applications such as screening or early diagnosis (15,17,18,21,35). Taken together, the future of liquid biopsy in PDAC lies not in identifying a single “ideal” biomarker, but in the careful, evidence-driven integration of complementary biological signals, standardized across laboratories, prospectively validated, and implemented in a manner that ensures both clinical impact and equitable access (1,21,32,34,50,59).

Conclusions

Liquid biopsy has evolved from a largely exploratory concept into a credible translational strategy in PDAC. Across multiple biomarker classes—including ctDNA, CTCs, EVs, circulating RNA species, and proteomic or metabolomic signatures—converging evidence indicates that blood-based biomarkers can complement established tools such as imaging and CA19-9 in selected clinical contexts. However, no single marker currently achieves the performance required for population-level screening or definitive early diagnosis.

The strongest and most clinically mature evidence at present supports the use of ctDNA-based assays for MRD detection and postoperative surveillance, where earlier molecular detection of recurrence may enable more timely intervention. In other settings, including high-risk surveillance, symptomatic diagnosis, and treatment monitoring, most liquid biopsy approaches should still be regarded as investigational adjuncts rather than practice-ready tools.

Meaningful translation into routine care will depend on several interrelated factors: rigorous standardization of pre-analytical and analytical workflows; prospective, multi-institutional validation in clinically realistic populations; and demonstration that biomarker-guided strategies improve patient outcomes rather than merely test performance metrics. Economic feasibility, scalability, and equitable access must also be addressed to ensure that technological advances do not exacerbate existing disparities in pancreatic cancer care.

If these challenges can be met, integrated multi-analyte and multi-omics liquid biopsy strategies have the potential to shift PDAC management toward earlier, more precise, and more individualized intervention. Until then, careful, evidence-driven, and context-specific implementation remains essential.

Acknowledgments

The authors thank Dr. Chuanhui Peng from the Department of Hepatobiliary Surgery at The First Affiliated Hospital of Zhejiang University for serving as a mentor and providing invaluable guidance throughout the research process. The authors also thank the clinical and academic staff of The First Affiliated Hospital of Zhejiang University and Zhejiang University School of Medicine for their institutional support and for fostering an environment that encourages research and critical inquiry in pancreatic cancer. No individuals or organizations provided dedicated funding, editorial assistance, or technical writing support for the preparation of this manuscript. The authors used artificial intelligence-based tools only for grammar refinement, plagiarism checking, and improving clarity and active voice usage in this manuscript; all scientific content and interpretations were generated by the authors, who take full responsibility for the final version.

Footnote

Reporting Checklist: The authors have completed the Narrative Review reporting checklist. Available at https://apc.amegroups.com/article/view/10.21037/apc-25-41/rc

Peer Review File: Available at https://apc.amegroups.com/article/view/10.21037/apc-25-41/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://apc.amegroups.com/article/view/10.21037/apc-25-41/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Conroy T, Pfeiffer P, Vilgrain V, et al. Pancreatic cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol 2023;34:987-1002. [Crossref] [PubMed]
2. Shen MJ, Jamali A, Katona BW. Challenges of early detection of pancreatic cancer. J Clin Invest 2025;135:e191942. [Crossref] [PubMed]
3. Bengtsson A, Andersson R, Linders J, et al. Biomarkers for Early Detection of Pancreatic Cancer. Visc Med 2025; Epub ahead of print. [Crossref]
4. Ballehaninna UK, Chamberlain RS. The clinical utility of serum CA 19-9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: An evidence based appraisal. J Gastrointest Oncol 2012;3:105-19. [Crossref] [PubMed]
5. Lee T, Teng TZJ, Shelat VG. Carbohydrate antigen 19-9 - tumor marker: Past, present, and future. World J Gastrointest Surg 2020;12:468-90. [Crossref] [PubMed]
6. Mehta S, Bhimani N, Gill AJ, et al. Serum Biomarker Panel for Diagnosis and Prognosis of Pancreatic Ductal Adenocarcinomas. Front Oncol 2021;11:708963. [Crossref] [PubMed]
7. Dubrovsky G, Ross A, Jalali P, et al. Liquid Biopsy in Pancreatic Ductal Adenocarcinoma: A Review of Methods and Applications. Int J Mol Sci 2024;25:11013. [Crossref] [PubMed]
8. Zhao Y, Tang J, Jiang K, et al. Liquid biopsy in pancreatic cancer - Current perspective and future outlook. Biochim Biophys Acta Rev Cancer 2023;1878:188868. [Crossref] [PubMed]
9. Wu J, Hu S, Zhang L, et al. Tumor circulome in the liquid biopsies for cancer diagnosis and prognosis. Theranostics 2020;10:4544-56. [Crossref] [PubMed]
10. Reese KL, Pantel K, Smit DJ. Multibiomarker panels in liquid biopsy for early detection of pancreatic cancer - a comprehensive review. J Exp Clin Cancer Res 2024;43:250. [Crossref] [PubMed]
11. Yang Z, LaRiviere MJ, Ko J, et al. A Multianalyte Panel Consisting of Extracellular Vesicle miRNAs and mRNAs, cfDNA, and CA19-9 Shows Utility for Diagnosis and Staging of Pancreatic Ductal Adenocarcinoma. Clin Cancer Res 2020;26:3248-58. [Crossref] [PubMed]
12. Kawai M, Fukuda A, Otomo R, et al. Early detection of pancreatic cancer by comprehensive serum miRNA sequencing with automated machine learning. Br J Cancer 2024;131:1158-68. [Crossref] [PubMed]
13. Abdel-Razeq R, Mansour A, Barbar M, et al. Enhancing Early Detection of Pancreatic Cancer in Genetically Predisposed Individuals: Integrating Advanced Imaging Modalities with Emerging Biomarkers and Liquid Biopsy. Biologics 2025;19:511-23. [Crossref] [PubMed]
14. Bogdanski AM, Acedo P, Wallace MB, et al. Recommendations, evidence and sustainability of screening for pancreatic cancer in high-risk individuals. Best Pract Res Clin Gastroenterol 2025;74:101974. [Crossref] [PubMed]
15. Li PCW, Winters HD, Geng X, et al. Real-world microscopic residual disease (MRD) monitoring with circulating tumor DNA (ctDNA) in pancreatic adenocarcinoma (PDAC). J Clin Oncol 2025;43:696.
16. Toledano-Fonseca M, Cano MT, Inga E, et al. Circulating Cell-Free DNA-Based Liquid Biopsy Markers for the Non-Invasive Prognosis and Monitoring of Metastatic Pancreatic Cancer. Cancers (Basel) 2020;12:1754. [Crossref] [PubMed]
17. Tarabay A, Swales L, Smolenschi C, et al. Precision medicine strategy in pancreatic ductal adenocarcinoma. ESMO Open 2025;10:105899. [Crossref] [PubMed]
18. Mahadevia H, Majeed U, Patel J, et al. Circulating Tumor DNA and Tissue Testing for Pancreatobiliary Tumors. JAMA Netw Open 2025;8:e2531373. [Crossref] [PubMed]
19. Heredia-Soto V, Rodríguez-Salas N, Feliu J. Liquid Biopsy in Pancreatic Cancer: Are We Ready to Apply It in the Clinical Practice? Cancers (Basel) 2021;13:1986. [Crossref] [PubMed]
20. Stosic K, Senar OA, Tarfouss J, et al. A Comprehensive Review of the Potential Role of Liquid Biopsy as a Diagnostic, Prognostic, and Predictive Biomarker in Pancreatic Ductal Adenocarcinoma. Cells 2023;13:3. [Crossref] [PubMed]
21. Toledano-Fonseca M, Brozos-Vázquez E, García-Ortiz MV, et al. Unveilling the potential of liquid biopsy in pancreatic cancer for molecular diagnosis and treatment guidance. Crit Rev Oncol Hematol 2025;212:104807. [Crossref] [PubMed]
22. Luo G, Jin K, Deng S, et al. Roles of CA19-9 in pancreatic cancer: Biomarker, predictor and promoter. Biochim Biophys Acta Rev Cancer 2021;1875:188409. [Crossref] [PubMed]
23. Ge L, Pan B, Song F, et al. Comparing the diagnostic accuracy of five common tumour biomarkers and CA19-9 for pancreatic cancer: a protocol for a network meta-analysis of diagnostic test accuracy. BMJ Open 2017;7:e018175. [Crossref] [PubMed]
24. Halder K, Borazanci EH, Jameson GS, et al. Exosomal ALPPL2 and THBS2 as biomarkers for early detection and disease monitoring of pancreatic ductal adenocarcinoma. Br J Cancer 2025;133:1335-43. [Crossref] [PubMed]
25. Ahmadipour M, Bhattacharya A, Sarafbidabad M, et al. CA19-9 and CEA biosensors in pancreatic cancer. Clin Chim Acta 2024;554:117788. [Crossref] [PubMed]
26. Wang M, Bu H, Luo W, et al. CA19-9, CEA and PIVKA-II as a novel panel of serum markers for diagnosis of pancreatic cancer. Clin Biochem 2025;137:110902. [Crossref] [PubMed]
27. Patel RD, Patel B, Crnogorac-Jurcevic T. Extracellular Vesicle-Derived miRNAs as Diagnostic Biomarkers for Pancreatic Ductal Adenocarcinoma: A Systematic Review of Methodological Rigour and Clinical Applicability. Biomark Insights 2025;20:11772719251381960. [Crossref] [PubMed]
28. Buscail E, Maulat C, Muscari F, et al. Liquid Biopsy Approach for Pancreatic Ductal Adenocarcinoma. Cancers (Basel) 2019;11:852. [Crossref] [PubMed]
29. Lee JS, Park SS, Lee YK, et al. Liquid biopsy in pancreatic ductal adenocarcinoma: current status of circulating tumor cells and circulating tumor DNA. Mol Oncol 2019;13:1623-50. [Crossref] [PubMed]
30. Raufi AG, May MS, Hadfield MJ, et al. Advances in Liquid Biopsy Technology and Implications for Pancreatic Cancer. Int J Mol Sci 2023;24:4238. [Crossref] [PubMed]
31. Yousef M, Yousef A, Hurd MW, et al. KRAS mutation detection by liquid biopsy for pancreatic ductal adenocarcinoma. J Hematol Oncol 2025;18:44. [Crossref] [PubMed]
32. Juthani R, Manne A. Blood-based biomarkers in pancreatic ductal adenocarcinoma: developments over the last decade and what holds for the future- a review. Front Oncol 2025;15:1555963. [Crossref] [PubMed]
33. Furukawa T, Fukada I, Hayashi N, et al. Clinical predictors of KRAS mutation detection in liquid biopsies for pancreatic ductal adenocarcinoma. Pancreatology 2025;25:736-42. [Crossref] [PubMed]
34. Lin Z, Adeniran EA, Cai Y, et al. Early Detection of Pancreatic Cancer: Current Advances and Future Opportunities. Biomedicines 2025;13:1733. [Crossref] [PubMed]
35. Ma L, Guo H, Zhao Y, et al. Liquid biopsy in cancer current: status, challenges and future prospects. Signal Transduct Target Ther 2024;9:336. [Crossref] [PubMed]
36. Botta GP, Abdelrahim M, Drengler RL, et al. Association of personalized and tumor-informed ctDNA with patient survival outcomes in pancreatic adenocarcinoma. Oncologist 2024;29:859-69. [Crossref] [PubMed]
37. Chen X, Hu X, Liu T. Development of liquid biopsy in detection and screening of pancreatic cancer. Front Oncol 2024;14:1415260. [Crossref] [PubMed]
38. Cohen JD, Javed AA, Thoburn C, et al. Combined circulating tumor DNA and protein biomarker-based liquid biopsy for the earlier detection of pancreatic cancers. Proc Natl Acad Sci U S A 2017;114:10202-7. [Crossref] [PubMed]
39. Cox M, Vitello D, Chawla A. Translating the multifaceted use of liquid biopsy to management of early disease in pancreatic adenocarcinoma. Front Oncol 2025;15:1520717. [Crossref] [PubMed]
40. Martini V, Timme-Bronsert S, Fichtner-Feigl S, et al. Circulating Tumor Cells in Pancreatic Cancer: Current Perspectives. Cancers (Basel) 2019;11:1659. [Crossref] [PubMed]
41. Luo K, Wang X, Zhang X, et al. The Value of Circulating Tumor Cells in the Prognosis and Treatment of Pancreatic Cancer. Front Oncol 2022;12:933645. [Crossref] [PubMed]
42. Effenberger KE, Schroeder C, Hanssen A, et al. Improved Risk Stratification by Circulating Tumor Cell Counts in Pancreatic Cancer. Clin Cancer Res 2018;24:2844-50. [Crossref] [PubMed]
43. Bobek V, Gurlich R, Eliasova P, et al. Circulating tumor cells in pancreatic cancer patients: enrichment and cultivation. World J Gastroenterol 2014;20:17163-70. [Crossref] [PubMed]
44. Brychta N, Drosch M, Driemel C, et al. Isolation of circulating tumor cells from pancreatic cancer by automated filtration. Oncotarget 2017;8:86143-56. [Crossref] [PubMed]
45. Stoecklein NH, Fluegen G, Guglielmi R, et al. Ultra-sensitive CTC-based liquid biopsy for pancreatic cancer enabled by large blood volume analysis. Mol Cancer 2023;22:181. [Crossref] [PubMed]
46. Song BG, Kwon W, Kim H, et al. Detection of Circulating Tumor Cells in Resectable Pancreatic Ductal Adenocarcinoma: A Prospective Evaluation as a Prognostic Marker. Front Oncol 2021;10:616440.
47. Zhao X, Ma Y, Dong X, et al. Molecular characterization of circulating tumor cells in pancreatic ductal adenocarcinoma: potential diagnostic and prognostic significance in clinical practice. Hepatobiliary Surg Nutr 2021;10:796-810. [Crossref] [PubMed]
48. Osuna de la Peña D, Trabulo SMD, Collin E, et al. Bioengineered 3D models of human pancreatic cancer recapitulate in vivo tumour biology. Nat Commun 2021;12:5623. [Crossref] [PubMed]
49. Sturm N, Ettrich TJ, Perkhofer L. The Impact of Biomarkers in Pancreatic Ductal Adenocarcinoma on Diagnosis, Surveillance and Therapy. Cancers (Basel) 2022;14:217. [Crossref] [PubMed]
50. Wang Q, Šabanović B, Awada A, et al. Single-cell omics: a new perspective for early detection of pancreatic cancer? Eur J Cancer 2023;190:112940. [Crossref] [PubMed]
51. Götze J, Meißner K, Pereira-Veiga T, et al. Identification and characterization of tumor and stromal derived liquid biopsy analytes in pancreatic ductal adenocarcinoma. J Exp Clin Cancer Res 2025;44:14. [Crossref] [PubMed]
52. Verel-Yilmaz Y, Fernández JP, Schäfer A, et al. Extracellular Vesicle-Based Detection of Pancreatic Cancer. Front Cell Dev Biol 2021;9:697939. [Crossref] [PubMed]
53. Hsu SK, Jadhao M, Liao WT, et al. The Role of Exosomes in Pancreatic Ductal Adenocarcinoma Progression and Their Potential as Biomarkers. Cancers (Basel) 2023;15:1776. [Crossref] [PubMed]
54. Luan X, Wang X, Bian G, et al. Exosome applications for the diagnosis and treatment of pancreatic ductal adenocarcinoma: An update Oncol Rep 2025;53:13. (Review). [Crossref] [PubMed]
55. Tiyuri A, Hatami H, Mobarezi Z, et al. Exosomal RNA biomarkers in pancreatic ductal adenocarcinoma: Systematic review and meta-analysis. Clin Chim Acta 2026;578:120532. [Crossref] [PubMed]
56. Reese M, Flammang I, Yang Z, et al. Potential of Exosomal microRNA-200b as Liquid Biopsy Marker in Pancreatic Ductal Adenocarcinoma. Cancers (Basel) 2020;12:197. [Crossref] [PubMed]
57. Song P, Dong J, Yu T, et al. Accurate Diagnosis of Pancreatic Ductal Adenocarcinoma by Detection of miRNA-196a Biomarker in Exosome Using Solution-Gated Graphene Transistor with Antifouling Design. Adv Healthc Mater 2025;14:e2404572. [Crossref] [PubMed]
58. Buscail E, Alix-Panabières C, Quincy P, et al. High Clinical Value of Liquid Biopsy to Detect Circulating Tumor Cells and Tumor Exosomes in Pancreatic Ductal Adenocarcinoma Patients Eligible for Up-Front Surgery. Cancers (Basel) 2019;11:1656. [Crossref] [PubMed]
59. Volkov H, Shlayem R, Shomron N. Deciphering the Diagnostic Potential of Small Non-Coding RNAs for the Detection of Pancreatic Ductal Adenocarcinoma Through Liquid Biopsies. Int J Mol Sci 2025;26:8108. [Crossref] [PubMed]
60. Munnings R, Gibbs P, Lee B. Evolution of Liquid Biopsies for Detecting Pancreatic Cancer. Cancers (Basel) 2024;16:3335. [Crossref] [PubMed]
61. Hou J, Li X, Xie KP. Coupled liquid biopsy and bioinformatics for pancreatic cancer early detection and precision prognostication. Mol Cancer 2021;20:34. [Crossref] [PubMed]
62. Sogawa K, Yonekubo R, Shimizu M, et al. Serum fucosylated receptor expression-enhancing protein 5 as a biomarker for early stage pancreatic cancer. Oncol Lett 2025;30:556. [Crossref] [PubMed]
63. Athanasiou A, Kureshi N, Wittig A, et al. Biomarker Discovery for Early Detection of Pancreatic Ductal Adenocarcinoma (PDAC) Using Multiplex Proteomics Technology. J Proteome Res 2025;24:315-22. [Crossref] [PubMed]
64. Fahrmann JF, Capello M, Hanash S. Liquid biopsy-based biomarkers for early detection of pancreatic cancer. Precis Cancer Med 2018;1:19.
65. Lucas AL, Simeone DM, Katona BW, et al. Validation of a Serum-Based Biomarker Signature for Detection of Early-Stage Pancreatic Ductal Adenocarcinoma. Gastroenterology 2025;S0016-5085(25)05976-1.
66. Palma NA, Lucas AL, Katona BW, et al. A High Performing Biomarker Signature for Detecting Early-Stage Pancreatic Ductal Adenocarcinoma in High-Risk Individuals. Cancers (Basel) 2025;17:1866. [Crossref] [PubMed]
67. Shin DW, Cho JY, Cho S, et al. Development of a serum protein biomarker panel for the diagnosis of pancreatic ductal adenocarcinoma using a machine learning approach. Sci Rep 2025;15:35659. [Crossref] [PubMed]
68. Lee JC, Kang SW, Sim EJ, et al. Novel mRNA biomarker-based liquid biopsy for the detection of resectable pancreatic cancer. BMC Cancer 2025;25:762. [Crossref] [PubMed]
69. Overbeek KA, Levink IJM, Koopmann BDM, et al. Long-term yield of pancreatic cancer surveillance in high-risk individuals. Gut 2022;71:1152-60. [Crossref] [PubMed]
70. Zhang C. Examining the Use of Serum Biomarkers to Guide Early Detection of Pancreatic Cancer. Clin J Oncol Nurs 2025;29:151-6. [Crossref] [PubMed]
71. Blackford AL, Canto MI, Dbouk M, et al. Pancreatic Cancer Surveillance and Survival of High-Risk Individuals. JAMA Oncol 2024;10:1087-96. [Crossref] [PubMed]
72. Nagai M, Sho M, Akahori T, et al. Application of liquid biopsy for surgical management of pancreatic cancer. Ann Gastroenterol Surg 2020;4:216-23. [Crossref] [PubMed]