Introduction

Pancreatic cancer (PC) is one of the most common malignancies, and the number of PC cases has doubled over the past two decades. The incidence of PC varies significantly across regions and populations, with the highest rates observed in North America, Europe, and Australia [1,2,3]. Recent years have seen a rapid increase in deaths due to PC, which can be attributed to global population growth and age structure changes and is closely linked to social and economic development [4]. According to predictions, PC is expected to become the third leading cause of cancer-related deaths in the European Union [5]. By 2030, it is projected to overtake breast, prostate, and colorectal cancers and become the second leading cause of cancer-related deaths in the United States [6].

Pancreatic ductal adenocarcinoma (PDAC) is a primary histological subtype of PC, accounting for 90% of all cases [7, 8]. Surgical resection is one of the methods of a potential cure, but most PDAC patients are unsuitable for operations when they are diagnosed [9, 10]. Therefore, screening and diagnosis should be conducted as early as possible to ensure a positive outcome. The diagnosis of PDAC relies on endoscopic ultrasonography-guided fine needle aspiration (EUS-FNA), magnetic resonance imaging (MRI), and computed tomography (CT) [11,12,13,14]. However, there are some problems, such as invasiveness, high cost, and exposure of subjects to radiation [15, 16]. In addition, the molecular composition of tumours is complex and dynamic, and repeated endoscopy examinations create a significant burden on patients [17]. Although the potential role of diagnostic biomarkers of cancer is constantly evolving, reliable diagnostic biomarkers for PC are still lacking. For instance, carbohydrate antigen 19-9 (CA19-9) has been extensively studied as a biomarker for detecting PC [18]. However, due to the lack of specificity of CA19-9, it can be expressed in various liver and gallbladder diseases as well as other types of malignant tumours, and elevated levels can also occur in some benign obstructive diseases [19]. Fucosyltransferase 3 (also known as the Lewis gene) is the key enzyme involved in the biosynthesis of CA19-9. Approximately 5–10% of individuals are Lewis antigen-negative, which means they do not secrete or secrete very little CA19-9, and to some extent, this also hinders the diagnosis of PC [20]. Therefore, CA19-9 alone cannot offer a conclusive diagnosis and must be combined with different clinical presentations, imaging tests, and biomarkers [15, 21].

In recent years, liquid biopsy has garnered attention due to its advantages of lower invasiveness and the ability to continuously monitor cancer progression. While blood is considered the most critical biofluid for liquid biopsy (Fig. 1), other clinical samples, such as cerebrospinal fluid, saliva, ascites, pleural effusion, and urine, have also been used [22,23,24,25,26]. Different sample sources have unique characteristics, with the prevailing view suggesting that blood samples carry a richer molecular information profile. Although noninvasive samples such as stool, urine, and saliva may contain less biomarker information than blood, they can provide valuable information about the location of diseases. For instance, certain biomarkers in urine may be associated with kidney or bladder conditions [27], while stool biomarkers may be linked to digestive system disorders [28]. Currently, the potential targets of liquid biopsy are circulating tumour cells (CTCs), circulating tumour DNA (ctDNA), noncoding RNAs (ncRNAs), messenger RNAs (mRNAs), and extracellular vesicles (EVs), which can provide information about tumour genomics, transcriptomics, and proteomics.

Fig. 1
figure 1

Common samples, biomarkers, and clinical applications in liquid biopsy for pancreatic cancer. Blood is typically the most commonly used material in liquid biopsy, in addition to pancreatic juice, saliva, urine, and stool. Circulating tumour cells, circulating tumour DNA, noncoding RNAs, and extracellular vesicles are among the most common biomarkers. Liquid biopsy has a wide range of clinical applications, playing a crucial role in early diagnosis, treatment monitoring, and prognosis evaluation. Created with BioRender.com

Full size image

Liquid biopsy exhibits high utility in the management of PC, with applications spanning early diagnosis, treatment strategies, drug resistance, recurrence monitoring, and prognosis assessment for PC patients. This review provides an overview of the biomarkers and detection methods utilized in liquid biopsy and their applications in the early diagnosis, treatment response, and prognosis evaluation of PC (Table 1). We also discuss the future trends of liquid biopsy and assess its limitations to improve current management strategies for patients.

Table 1 The isolation and detection techniques in liquid biopsy
Full size table

Biomarkers and detection methods

Circulating tumour cells

CTCs, which detach from the primary tumour, can enter the circulatory system and travel through the bloodstream. However, the majority of CTCs die in the peripheral blood within 1 to 2.5 hours due to mechanical forces or immune system attacks. Nevertheless, a small fraction of CTCs can survive and initiate distant metastasis [67, 68]. Numerous metastatic precursors within CTCs increase the risk of tumour metastasis and recurrence [69,70,71]. According to most perspectives, CTCs are believed to exhibit specific differences from primary tumours despite originating from primary tumours. This heterogeneity leads to their detachment from the primary tumour and acquisition of epithelial-mesenchymal transition (EMT) characteristics, facilitating intravascular infiltration and enhancing their potential for metastasis [72, 73].

The analysis process of CTCs mainly involves three stages: enrichment, detection, and characterization. Most enrichment methods are applied based on the surface phenotype or physical properties of CTCs. The CellSearch system, developed using an antibody targeting epithelial cell adhesion molecule (EpCAM), is the sole CTC detection technology approved by the U.S. Food and Drug Administration due to its ability to detect CTCs expressing EpCAM [29, 74]. However, this strategy cannot detect cells with low EpCAM expression due to the potential loss of epithelial antigens during the EMT process [34]. Furthermore, the abundance of CTCs varies across different types of cancer; the CellSearch system is more suitable for tumours with higher CTC abundance [75]. In microfluidic devices, affinity-based separation methods can also be employed. Designing microfluidic devices with varying materials, sizes, and structures to manipulate blood flow patterns creates additional opportunities for interacting CTCs and antibodies [32, 57, 76]. Pahattuge et al. [77] introduced a modular microfluidic system called SMART-Chip. They demonstrated that the SMART-Chip platform could significantly reduce the processing time by more than 50% when handling blood samples obtained from patients with PDAC and colorectal cancer compared to manual sample processing. Furthermore, microfabricated porous membranes can be employed to filtrate and isolate CTCs due to their size, which is larger than that of normal blood cells [78, 79].

CTCs are primarily detected using protein expression, immunocytochemistry, and nucleic acid methods. Flow cytometry allows for the quantitative assessment and characterization of protein expression in CTCs, offering the advantage of evaluating multiple biomarkers to characterize CTCs comprehensively. However, it exhibits lower sensitivity for detecting rare populations of CTCs [80]. Immunohistochemical staining and immunofluorescence are commonly employed techniques for detection and characterization purposes. Immunofluorescence, in particular, enables the visual confirmation of protein expression and localization by fluorescent markers. In the conventional cytofluorimetry approach, isolation is achieved through the utilization of specific antibodies that recognize markers selected on CTCs. This method utilizes monoclonal antibodies specifically targeting antigens expressed by CTCs, which results in the exclusion of CTCs that do not express such antigens but are present in the circulation. Consequently, this presents a challenge in obtaining or develo** novel antibodies against specific targets [81]. The flexibility of immunofluorescence technology makes it a powerful tool for studying the protein expression of tumour cells. For instance, with the use of multimarker immunofluorescence panels, researchers can gain a more comprehensive understanding of the distribution and expression patterns of different CTC subtypes [82]. This not only aids in tumour classification and staging but also provides valuable insights for personalized therapy. In addition, CTCs can also be detected using techniques such as high-resolution image scanning, mutational analysis, and single-cell next-generation sequencing (scNGS). The molecular characteristics of CTCs were initially determined on enriched fractions, which provided limited information about tumour heterogeneity. In recent years, the rapidly advancing single-cell sequencing technology has become the preferred method for isolating individual CTCs and studying tumour heterogeneity. These technologies will facilitate the comprehensive characterization of CTCs at multiple omics and functional levels, enabling effective monitoring of the dynamic changes in tumour heterogeneity in individual cancer patients [83, 84].

Although the precise role of CTCs in tumour development remains elusive, they offer a valuable approach for obtaining comprehensive insights into tumours through liquid biopsy. In PC management, CTCs play a significant and beneficial role in patient diagnosis, prognostic evaluation, recurrence monitoring, and treatment decisions. In this regard, we have summarized the clinical applications of CTCs in various aspects of PC management in recent years (Table 2).

Table 2 Clinical application of CTCs in PC in recent years
Full size table

Circulating tumour DNA

Cell-free DNA (cfDNA) is a crucial genetic component found in the bloodstream; its origin primarily stems from apoptotic, necrotic, and actively secreted fragments originating from healthy, inflamed, and tumour tissue. These fragments are typically approximately 150–180 base pairs in length [97,98,99]. CtDNA represents a distinctive subset of cfDNA released into the blood by CTCs. Compared to cfDNA, ctDNA is present in relatively lower amounts in the bloodstream, constituting only 1% (or even less than 0.01%) of cfDNA [98,99,100]. Most ctDNA fragments have lengths ranging from 160 to 200 base pairs, and they are less influenced by intratumoural heterogeneity compared to tumour tissues [97, 101, 102]. Additionally, ctDNA has a half-life of approximately 15 minutes to 2.5 hours, which means that it serves as a real-time tumour biomarker. In contrast, traditional blood protein biomarkers usually take weeks to manifest, and ctDNA can dynamically reflect the status of a tumour at a specific moment [103, 104]. Furthermore, ctDNA carries tumour-related genomic information, such as gene expression levels, mutations, the methylation status, and microsatellite instability. Compared to traditional biopsy markers, ctDNA is an ideal biomarker, especially for the real-time monitoring of treatment effectiveness and prognosis assessment.

CtDNA detection includes ctDNA preparation, library construction, analysis, and data alignment. One aspect of ctDNA detection focuses on genetic mutations. Single-base mutations have the potential to activate oncogenes, disrupting the balance between oncogenes and tumour suppressor genes, thereby instigating tumorigenesis. Another aspect involves DNA methylation, which plays a role in tumour initiation that is similar to that of DNA mutations [105, 106]. Mutation detection is a vital component of the analysis. Due to the extremely low abundance of ctDNA, employing highly sensitive techniques for detecting tumour mutations is crucial. Conventional approaches rely on polymerase chain reaction (PCR), but recent advancements in PCR and sequencing technologies have paved the way for alternative methods, including quantitative PCR (qPCR), digital PCR (dPCR), droplet digital PCR (ddPCR), and next-generation sequencing (NGS). qPCR allows real-time monitoring of DNA amplification with higher speed, reproducibility, and quantification. NGS platforms offer several advantages, including the ability to screen for unknown mutations and structural and copy-number variations. dPCR and ddPCR involve partitioning DNA samples into thousands or even millions of separate compartments or droplets, effectively reducing background noise associated with traditional methods and enabling the detection of tumour DNA at a variant allele frequency (VAF) below 0.1% [107,108,109]. In recent years, integrated detection strategies combining gene editing techniques, functional enzymes, and nanomaterials have been developed to effectively increase the net content of mutation fragments, thereby facilitating the identification of target gene mutations within ctDNA [106]. There are various methods for DNA methylation detection. Whole-genome bisulfite sequencing (WGBS-seq) is considered the gold standard for DNA methylation analysis. It can identify partially methylated regions in cancer cells. However, the sensitivity of this method may be compromised by DNA degradation [26, 110].

KRAS mutations are the most prevalent genetic alteration in PC. They are present in over 90% of patients and are considered an early driving factor in PDAC [108]. Castells et al. [111] demonstrated that the presence of KRAS mutations in plasma DNA served as a highly specific molecular marker for diagnosis and prognosis in a PDAC cohort of 44 patients. However, it is essential to emphasize that previous cfDNA sequencing results have not only identified mutations known to exist in tumours but have also uncovered a multitude of variations that are absent in tumour tissues [112]. In particular, some patients undergoing chemotherapy may harbour minimal residual lesions composed of drug-resistant cells. In such cases, the mutations detected in cfDNA in the bloodstream do not exclusively originate from tumour cells. CfDNA may also carry mutations from other sources, including those induced by the disease state or treatment. Undoubtedly, KRAS mutations are one of the vital indicators for evaluating PDAC, and their role has received widespread attention as a primary focus in many PC studies (Table 3). However, overall, the application of ctDNA and mutation analysis in PDAC still requires further strategies to thoroughly assess this detection method.

Table 3 Clinical application of cfDNA/ctDNA in PC in recent years
Full size table

Noncoding RNAs

NcRNAs were once perceived to have a limited impact on tumour initiation and progression due to their inability to encode proteins. However, emerging evidence has highlighted the essential regulatory functions of ncRNAs. In addition to their capacity to modulate gene and protein expression, ncRNAs actively participate in diverse tumorigenic processes, including EMT, autophagy, and apoptosis [122,123,124]. NcRNAs can be classified into two main categories based on their lengths: small noncoding RNAs (sncRNAs), with a length of less than 200 nucleotides, and long noncoding RNAs (lncRNAs), with a length exceeding 200 nucleotides [125, 126]. In addition, circRNAs, which are RNA molecules with a circular structure, have been recognized for their significant regulatory roles in gene expression, cell proliferation, cell differentiation, and disease development in recent years. SncRNAs encompass several subtypes, including microRNAs (miRNAs), small nucleolar RNAs, small nuclear RNAs, piwi-interacting RNAs, and tRNA-derived small RNAs [127]. Among them, miRNAs are the most extensively studied factors in cancer research, and liquid biopsy identifies miRNAs actively secreted by CTCs and tumour cells themselves [128, 129]. MiRNAs can influence genes, with thousands of miRNAs regulating approximately 60% of the genes. Their principal function involves binding to recognition sites in the 3' untranslated region, thereby reducing mRNA stability and suppressing gene expression [130, 131].

LncRNAs play a regulatory role in protein and miRNA functions and expression levels and contribute to chromatin remodelling [130, 73, 95, 96]. However, some research suggests that there may not be a significant difference in CTC counts between blood samples collected before and after chemotherapy, possibly due to variations in CTC identification and treatment strategies [29, 93, 197]. In addition to quantitative analyses, the molecular characteristics of CTCs are also frequently used to assess a patient's treatment response [198]. Some CTC measurement techniques enable genetic profiling of CTCs, allowing the detection of key gene mutations, such as those in KRAS, HER2, and TP53 [199,200,201]. Furthermore, programmed death ligand 1 (PD-L1) staining methods can be employed to evaluate the status of CTCs in patients receiving monoclonal antibody therapy, with PD-L1-negative CTC patients often achieving better treatment outcomes [202]. In most cases, CTCs express chemokine receptors, with CXC-motif chemokine receptor 4 (CXCR4) being the most commonly expressed receptor. Continuous monitoring of CXCR4 during treatment serves as a predictive biomarker, providing information to identify which patients are likely to benefit from treatment or develop resistance [203, 204]. Regarding drug sensitivity, Wu et al. [205] conducted a study wherein they collected CTCs from patients diagnosed with PDAC and expanded them ex vivo into organoids. The sensitivity of these organoids to nine drugs (GEM, 5-fluorouracil, erlotinib, irinotecan, olaparib, oxaliplatin, paclitaxel, palbociclib, and trametinib) was examined. A significant correlation was observed between the drug sensitivity of CTCs and clinical outcomes. This indicates that the drug sensitivity of CTCs holds the potential to predict therapeutic outcomes in PDAC, thus enabling the avoidance of ineffective treatments. CECs have been proposed as a potential tool to predict how patients will respond to antiangiogenic cancer therapies. However, it is important to recognize that their diverse phenotypes may exhibit different dynamics during the course of treatment. Given the unique characteristics of CECs and their crucial role in liquid biopsies, this avenue of research holds promise and warrants further exploration. A clinical trial focused on late-stage pancreatic cancer patients monitored CEC levels during neoadjuvant therapy and observed an overall increase in CECs in response to combination therapy that was attributed to chemotherapy-induced vascular damage exacerbating CEC release [206]. Furthermore, research concerning surgery, which is a common treatment method, has indicated that CEC levels typically decrease after tumour resection. This decline may result from the disruption of PDAC-derived growth factor recruitment of endothelial cells after tumour removal, subsequently reducing CEC levels [207].

The longitudinal assessment of ctDNA enables dynamic monitoring of disease trajectory, including treatment monitoring and the detection of minimal residual disease, and serves as an alternative biomarker for overall disease burden [208]. Tao et al. [209] conducted a study to examine the role of ctDNA in monitoring treatment response in a cohort of 17 PDAC patients who were treated with the FOLFIRINOX regimen (fluorouracil, irinotecan, and oxaliplatin). Among the 12 patients who responded to chemotherapy, 11 exhibited a reduction in the mutant allele fraction (MAF) of cfDNA. In contrast, the remaining 5 patients who developed chemotherapy resistance showed an increase in the ctDNA MAF during disease progression. These findings suggest that the levels of ctDNA partly reflect the tumour burden. In another study, Groot et al. [210] identified a substantial decrease in the probability of detecting ctDNA in the bloodstream of patients who underwent neoadjuvant chemotherapy compared to those who did not receive any preoperative chemotherapy (21% vs. 69%; p < 0.001). Although the practice of longitudinal ctDNA monitoring in PDAC cases remains limited, these studies underscore the potential of ctDNA as a crucial monitoring biomarker during the therapeutic course.

The emergence of chemoresistance presents formidable obstacles for nonsurgical candidates, thereby exacerbating their clinical predicament. Several miRNAs are considered key regulatory elements involved in acquiring chemoresistance in PDAC. Lu et al. [163] demonstrated that the expression of plasma miR-20a-5p in PDAC patients who exhibited resistance to GEM was markedly diminished compared to that in nonresistant patients (p < 0.01). The authors proposed that miR-20a-5p potentially regulates the expression of the RRM2 protein, thereby exerting an influence on the sensitivity of tumour cells to GEM. MiRNA levels have the potential to serve as informative indicators regarding disease progression, whether assessed before treatment initiation or during the treatment course. In a study by van der Sijde et al. [211], the elevated expression levels of serum miR-373-3p before FOLFIRINOX therapy was identified as a predictive factor for disease progression. Correspondingly, the reduced expression levels of miR-194-5p following a single cycle of FOLFIRINOX treatment indicated disease deterioration. LncRNA holds considerable importance in guiding the therapeutic approach for PC. Zhang et al. [215,216,217]. Although the CTC count is commonly used as the determining criterion in studies, distinguishing CTC subgroups can also reflect the tumour status to some extent. Certain CTC subgroups with specific phenotypes, for example, indicate the tendency of the tumour for metastasis [218, 219]. Semaan et al. [33] identified and characterized 4 CTC subpopulations that can be used for the clinical stratification of PC, providing a valuable perspective for applying liquid biopsy technologies in prognostic prediction. Some studies suggest that CTC distribution shows spatial heterogeneity and that portal venous blood may be a better option for assessing PDAC prognosis than peripheral venous blood [206].

Research on ctDNA primarily focuses on three KRAS mutations (G12D, G12V, and G12R) [226]; however, G12R has a lower detection rate than the other two mutations [227, 228]. Ako et al. [227] studied postoperative recurrence and overall survival in PDAC by analysing two KRAS mutations (G12D and G12V); they discovered that patients with both KRAS mutations had significantly lower disease-free survival. Interestingly, in a separate study, Guo et al. [216] specifically examined the influence of the G12D mutation on the prognosis of PDAC. Among 26 patients with PDAC, those with the KRAS G12D mutation had notably reduced overall survival (12.1 vs. 24.9 months, p < 0.001) and recurrence-free survival (6.3 vs. 17.4 months, p < 0.001) compared to those without the mutation. Notably, patients with the KRAS G12D mutation exhibited a distinct early recurrence trend and poorer clinical outcomes.

In previous studies, several reports have confirmed the predictive role of ncRNAs in PC. Most of these studies constructed multifactor prognostic risk models and performed survival analyses with these models [189, 229,

Conclusion

In conclusion, the application of liquid biopsy in the clinical management of PC aligns with the concept of precision medicine. Biological samples obtained through noninvasive procedures can provide detailed information about various aspects of the tumour, which aids in monitoring tumour development and evaluating treatment responses. Moreover, this information assists clinical physicians in understanding the molecular mechanisms of tumour occurrence and development and providing more accurate and personalized treatment decisions for each patient. There are also some limitations, including low-sensitivity detection techniques, nonstandardized analysis workflows, and small sample sizes; these limitations are significant barriers of liquid biopsy. It cannot be denied that with the continuous advancement of technological methods and large-scale clinical trials, many biomarkers have begun to demonstrate their value, indicating broad prospects for their application. Liquid biopsy will become an indispensable technology for tumour diagnosis and treatment.