For a long time, the pancreatic duct has been considered as the cellular origin of PDAC due to its similarity to duct-like morphology (33, 34). However, this hypothesis was challenged when genetic mouse models confirmed that acinar cells may also be a cellular source of PDAC (35–37). A fusion of the cytokeratin 19 promoter (specifically active in pancreatic ductal cells) with mutant KRAS failed to induce PanIN or PDAC, which negates the possibility of a ductal origin of PanIN (38). Nevertheless, recent analysis of somatic variants in human PDAC and precursor lesions has confirmed that PanIN is a disease that can spread through the entire ductal system, suggesting that PanIN may still originate from ductal cells. In patients, multiple discrete PanIN lesions often represent a single neoplasm that has spread along the ductal system (32). Therefore, the hypothesis of ductal cells as the origin of PanIN has been proposed once again. KRAS mutation in pancreatic ducts accelerates PDAC’s progression in a mouse model of chronic obstructive pancreatitis (39). A recent study utilizing tamoxifen-dependent CreERT2 mediated recombination mice (Hnf1b: CreER^T2; Kras^G12V ) suggests that ectopic expression and elevated levels of oncogenic mutant KRAS in pancreatic ducts generate early and late PanIN as well as PDAC (40). However, this doesn’t necessarily mean that only ductal cells are the original cells of PanIN, as more evidence suggests the diversification of its origin cells.

Acinar cells are a crucial type of pancreatic constituent cells that have been found to be the predominant origin cells for KRAS^G12D-induced PanINs in various studies. Acinar cells are the main type of pancreatic constituent cells. When oncogenic Kras is activated in combination with chronic inflammation, a high-fat diet, or mutations in tumor suppressor genes (SMAD4/DPC4 and P53), adult acinar cells lead to the transformation of PanIN lesions into PDAC (41–43). During embryonic development in mouse models, the pancreatic lineage expresses endogenous KRAS oncogene, which can replicate human PanIN and PDAC in a faithful manner. When endogenous KRAS^G12V oncogene is expressed in embryonic acinar/centroacinar cells in mice, they produce PanIN and PDAC. These findings suggest that PDAC may originate from acinar/centroacinar cells or their precursors that differentiate into duct-like cells (30).

Studies have shown that changes in the transcriptional network of acinar cells can affect early events induced by KRAS^G12D (35, 37, 44). In mice expressing KRAS^G12D but lacking MIST1, an acinar-restricted transcription factor, severe exocrine pancreatic defects occur, and the initiation and progression of PanIN are greatly accelerated (44). Since Mist1 is not expressed in centroacinar cells or duct cells, these findings suggest that altering the acinar Mist1 transcriptional network has a profound effect on the development of PanIN. Kopp et al. conducted a comparison of the propensity of ductal/centroacinar cells and acinar cells to form PanIN in response to oncogenic KRAS. To induce recombination of the KRAS^G12D allele in ductal/centroacinar cells or acinar cells, they used Sox9^CreER or Ptf1a^CreER mice, respectively. Their findings showed that all Ptf1a^CreER;Kras^G12D mice displayed abundant PanIN, in contrast, only 57% of Sox9^CreER;Kras^G12D mice displayed a small amount of PanIN (45).

Guerra et al. created a PanIN model of LSL-Kras^G12V;Elas-tTA/tetO-Cre mice that allows controlled temporal expression of KRAS^G12V oncogene in acinar and central acinar cells. Between 1 and 3 months of age in this mice, acinar-ductal metaplasia was observed. The metaplastic acinar structures comprised highly proliferative cells that included both acinar cells and duct-like cells. The Notch, TGF-β, and EGF signaling pathways are typically maintained at high levels in PDAC, providing cells with growth and survival advantages. Metaplastic acinar structures expressed Notch target genes, and exhibited mosaic expression patterns for ErbB2, and pERK. This expression pattern was similar to that of PanIN, which means that both follow similar molecular pathways. Thus, it is possible that KRAS^G12D-induced PanIN may originate from acinar cells undergoing acinar-ductal metaplasia (46). TGF-a has been found to induce transdifferentiation of acinar cells into ductal cells, further supporting the notion that acinar cells can develop ductal properties and potentially be the origin of PanIN (47). The ductal origin of PanIN is also possible, although this is more unusual (48).

The ductal phenotype of PanIN and other PDAC precursor lesions such as IPMN and MCN, suggests that pancreatic ducts are the origin of pancreatic cancer (49). However, recent evidence emerging from mouse models and lineage tracing studies challenges the notion that the pancreatic duct is the exclusive origin of PDAC. A significant amount of evidence suggests that PDAC may also originate from centroacinar-acinar cells through the ADM process or the expansion of centroacinar cells (28, 30, 37). Centroacinar cells, located at the interface of acinar and terminal duct cells, are considered candidates for pancreatic progenitor cells. Centroacinar cells are much like duct cells, with only subtle differences in the characteristics of the apical membrane. Centroacinar cells are a ductal cell type located at the center of acini, with active Notch signaling, and expression of the endocrine differentiation regulator Sox9 (50). SOX9 is a transcription factor required for the maintenance of the pancreatic progenitor pool and for determining pancreatic endocrine and exocrine cell fates (51–53). Wang et al. discovered that SOX9 was expressed in central acinar cells, ductal epithelial cells, PDAC and its precursor lesions, but was rarely expressed in other pancreatic tumors (54). When co-expressed with oncogenic KRAS, SOX9 mutant mice accelerated the formation of PDAC precursor lesions (45). Pten is expressed in pancreatic ducts, central acinar cells, and pancreatic islets. Stanger et al. discovered that mice with pancreas-specific deletion of Pten exhibit extensive ductal metaplasia with developed PanIN lesions and malignant transformation. They further found that centroacinar cells are highly proliferative before the onset of metaplasia, and this hyperproliferative state exists at an even greater level after metaplastic changes. Ductal metaplasia in Pten mutant mice does not arise from acinar cells, concluding that centroacinar cells are instrumental in initiating the process of metaplasia (28). However, more research is necessary to support the evidence that PanIN can originate from central acinar cells. A subset of central acinar cells exhibit several markers, such as SOX9, HES1, and ALDH1B1, which are known to play significant roles in the development of PDAC and are also associated with cancer stem cells. This has led to the hypothesis that a subset of central acinar cells may function as cancer stem cells.

In human PDAC primary xenografts and pancreatic cancer cell lines, a subpopulation of cells with potent tumor initiation or cancer stem cells (CSCs) has been identified by scholars (55–58). CSCs contribute to PDAC initiation and metastasis, are pluripotent, self-renewing, and tumor-forming, and responsible for resistance to chemotherapy and radiotherapy. CSCs account for less than 1% of all pancreatic cancer cells, and their origin remains uncertain. However, it is currently believed that they may emerge from transformed stem/progenitor cells or terminal cell dedifferentiation (59). The reactivation of embryonic programs and the expression of genes related to self-renewal are common to both stem cells and cancer (60–62). Terminally differentiated cells in the adult pancreas also display a high degree of plasticity. Efforts have been made to transform acinar cells into the endocrine lineage, particularly β cells. Pancreatic acinar cells are prone to dedifferentiate into duct-like cell phenotypes in response to stresses such as injury or inflammation. This evidence well demonstrates the ability of terminal cells to dedifferentiate and retrograde (63–66).

Currently, there is limited research on PDAC CSCs. CD133, CD24, CD44, EPCAM, and ESA are currently considered the most comprehensive biomarkers for pancreatic CSCs (67, 68). The CD133 pancreatic CSCs co-express the CXCR4 receptor play a crucial role in tumor metastasis (58). Animal experiments have confirmed a highly tumorigenic subpopulation of CD44 co-express CD24, EpCAM, and CD133 pancreatic cancer cells that exhibit the biological properties of CSCs, making them often resistant to chemo and radiotherapy (55). In addition to cell surface markers, the cellular molecule ALDH1, which catalyzes the oxidation of intracellular aldehydes and transforms retinol to retinoic acid, has also been identified as a pancreatic ductal adenocarcinoma stem cell marker (67). The ABCG autofluorescent vesicular cell subset in PDAC has potent CSCs properties with distinct tumorigenic potential (69). C-Met is also critical in the biology of pancreatic CSCs (55). The double corticoid-like kinase DCLK1 and the G protein-coupled receptor LGR5 containing leucine-rich repeats are also markers of pancreatic CSCs (68). Additionally, 26S proteasome activity, CD90, and side group (SP) are also included in biomarkers of CSCs (70–73).

Although the majority of studies suggest that PDAC arises from PanIN, relatively few investigations have focused on PanIN CSCs. A subpopulation of cells containing high levels of DCLK1 and acetylated αTubulin (AcTub) are present in mouse ADM and PanIN, these cells display unique sphere-forming capacities similar to CSCs. AcTub+/DCLK1+ cell subpopulation in ADM and PanIN epithelium is acinar cell-derived, and these cells have been shown to enhance clonogenic capacity in vitro and have high metastatic potential in vivo, both factors are crucial for the development and progression of PanIN (74, 75). Additionally, oncogenic KRAS effectively initiates PanIN development in the DCLK1-expressing cell subset, and deletion of DCLK1 greatly reduces PanIN formation (75). However, it remains to be further clarified and characterized whether these are subpopulations of acinar cells specifically produced by precancerous lesions and PDAC.

According to studies, multiple genetic alterations found in PDAC are also present in PanIN, and their frequency increases with PanIN grade (84). Recently, genomic DNA samples obtained by microdissection from PDAC and related PanIN 2 and PanIN 3 lesions in ten pancreatic cancer patients were subjected to exome sequencing analysis. The results showed that all adjacent lesions originated predominantly from a common ancestral source, with most somatic mutations present at the PanIN 2 stage or earlier (85). Laser microdissection technology combined with pyrosequencing was also used by Michael Goggins et al. to perform more detailed genetic testing of the early development of PanIN. They found that almost all early development of PanIN requires somatic mutation, with KRAS mutation initiating PanIN formation in genetically engineered mouse models (86). More than 99% of PanIN 1 lesions contain KRAS mutation, while early sporadic p16/CDKN2A, GNAS or BRAF mutations may further promote PanIN development based on KRAS mutation (86). Based on the view that PanIN 2 is the first true precancerous stage of PDAC, some scholars have screened 30 potential early diagnostic genes in this stage, including HNF3, S100P, TFF1, CDC37, AGR2, PI15, CST3, LGALS4, and PCOLN3, which have received attention. However, there is much more to PanIN than early genetic and molecular changes that remain to be explored in depth (87).

The use of genetically engineered mouse models has significantly advanced the study of PanIN and PDAC. By continuously monitoring the signaling pathways involved in pancreatic carcinogenesis using these models, researchers have been able to gain a better understanding of the mechanisms driving PDAC precursor progression. Activation of numerous cell fate regulation signaling pathways occurs during PanIN and PDAC and plays a crucial role in PDAC development ( Figure 3 ).

There is a hypothesis that PanIN could potentially arise from CSCs. While the origin of these cells is not entirely clear, there is a shared characteristic between stem cells and CSCs in terms of their expression of genes associated with self-renewal (60, 61). Therefore, alterations in genes related to stemness may signify an early genetic change in the development of PanIN.

In the early stages, PanIN’s molecular characteristics are relatively simple, and the likelihood of developing invasive PDAC is low. However, high-grade PanIN (PanIN 3) is more uncertain and complex due to severe nuclear atypia, lumen necrosis, and marked epithelial sprouting, which indicates the development of PDAC (123). At present, high-grade PanIN is not only the main precursor lesion of PDAC but also an ideal target for early detection (124). Unfortunately, clinical and imaging methods cannot detect high-grade PanIN, and thus molecular detection methods are critical (125, 126).

Maitra et al. utilized immunohistochemistry to examine the protein expression of genes related to PDAC progression in 55 PanIN lesion tissue microarrays. Their findings revealed that molecular abnormalities in PanIN were not random and were typically classified as “early” changes (such as prostate stem antigen and MUC5 expression, or P16 loss), “intermediate” changes (such as cyclin D1 expression), and “late” changes (such as P53, proliferating antigen, and MUC1 expression, or Smad4/Dpc4 loss) (127). However, further experiments involving sequencing and pedigree tracking have raised questions about these results. Using tumor sequencing, adjacent PanIN lesions, and normal tissues from 10 PDAC patients, Murphy et al. discovered that the tumors were genetically identical to the adjacent PanIN lesions for over 50% of the genetic mutations. While there were some variations in molecular mutations in each sample, the overall results indicated that the adjacent PanIN and tumor originated from a common ancestor and that most of the somatic mutations in the tumor occurred in PanIN 2 or earlier. Furthermore, PanIN 2 contains as many mutations as PanIN 3 and tumor tissue, indicating that PanIN 2 can potentially progress into a tumor even in the absence of PanIN 3. In addition, precancerous lesions may require epigenetic modifications, aneuploidy, or changes based on the expression of certain proteins to develop into invasive tumors (85). For instance, the proteins gal-1, annexins A4 and A5, ANXA4, vimentin, and laminin, which exhibit differential expression in PanIN 3, are also dysregulated in PDAC (102, 128). C-myc was identified as an important regulatory protein in the dysregulated protein network in PanIN 3 tissue, which supports the pathological and genomic progression model of PanIN to PDAC from a proteomic perspective (129–131).

Hosoda et al. conducted an analysis of 17 high-grade PanIN lesions in 15 patients and 16 low-grade PanIN lesions in 10 patients, revealing a presence of KRAS oncogenic mutations in 94% of both high and low-grade PanIN lesions. Meanwhile, in high-grade PanIN lesions, 29% exhibited RNF43 mutations, 18% displayed CDKN2A mutations, and 12% had mutations in GNAS and TP53. Additionally, one high-grade PanIN lesion was found to have mutations in PIK3CA, TGFBR2, and ARID1A (124). Though mutations in RNF43 and GNAS were also observed in low-grade PanIN lesions, no mutations in CDKN2A or TP53 were detected.

In the development of non-invasive pancreatic duct lesions, the inactivation of the P16 gene has been found to be a contributing factor (132). Analysis via immunohistochemistry of lesional tissues from 70 patients with pancreatic diseases demonstrated that loss of P16 expression occurred in pancreatic tissue with pancreatic duct dysplasia, preceding P53 and DPC4 inactivation. P16 expression inversely governs the expression of P21 and P53, both of which play an important role in cell cycle regulation and the onset and development of PanIN. As such, assessments based on immunohistochemistry can enhance the efficiency of PanIN diagnoses (133–135). While the early identification of high-grade PanIN may facilitate timely treatment, studying PanIN 2 may provide greater clinical diagnostic value and impede the progression of PanIN 2 into high-grade PanIN and aggressive PDAC.

Histopathology is considered the most accurate and reliable gold standard among all cancer diagnostic methods. However, obtaining sufficient material for biopsy can be challenging, and the invasive nature of tissue collection can affect diagnostic accuracy. The sensitivity of CT and MRI is limited by the size of the tumor. Spiral CT has a sensitivity of only 72% in detecting small pancreatic masses, and MRI performs similarly (142, 143).

Endoscopic ultrasonography (EUS) has been found to have higher diagnostic efficiency than CT and MRI, and EUS-guided fine needle aspiration (EUS-FNA) is particularly superior to histopathological diagnosis. However, EUS is an invasive procedure with potential complications like bleeding and pancreatitis. Endoscopic retrograde cholangiopancreatography (ERCP) is another option to diagnose PanIN, but it carries a postoperative risk of inducing pancreatitis and requires strict candidate selection (144).

Recently, serial pancreatic-juice aspiration cytologic examination (SPACE) has emerged as an advanced method for malignancy diagnosis (145–149). Nakahodo et al. retrospectively compared various imaging methods in patients with high-grade PanIN undergoing surgery and found that focal pancreatic parenchymal atrophy on CT/MRI and a hypoechoic stenotic pattern on EUS are important for the early diagnosis of PDAC, especially in the diagnosis of high-grade PanIN (150, 151). A comprehensive approach should be utilized to improve the diagnosis of PanIN, given the limitations of current imaging techniques and diagnostic methods.

Currently, early detection strategies for PDAC are primarily focused on pancreatic cystic fluid, pancreatic fluid, and plasma samples, but development of universal methods for early detection of PDAC is still necessary. The sensitivity, specificity, and accuracy of tumor marker CA19-9 in the diagnosis of PDAC are 83.1%, 73%, and 75%, respectively, but it is not suitable for early detection of PDAC due to its low value in predicting early stages of PDAC (152, 153). Genetic testing can provide a more sensitive diagnostic approach than conventional methods. Detection of KRAS mutations in the patient’s pancreatic fluid or plasma can offer stronger evidence for the diagnosis of PDAC, although its presence alone is not sufficient for such diagnosis (154). In addition, the protein encoded by GPC in cancer exosomes can be used as a potential marker for non-invasive diagnosis and screening tools by ctDNA biopsy diagnosis (141).

At present, there are few specific biomarkers can distinguish between low-grade PanIN, high-grade PanIN, and PDAC directly. MAbDas-1, a monoclonal antibody against the colonic epithelial antigen, has shown promise as a highly specific marker for precancerous lesions located in the upper gastrointestinal tract. Moreover, MAbDas-1 is capable of distinguishing between low- and high-risk IPMN lesions (155). Through the study of cystic fluid from 169 patients with pancreatic cystic lesions, Das et al. found that the Das-1 ELISA exhibited a sensitivity of 88%, specificity of 99%, and accuracy of 95% for high-risk lesions (156). Expression of Das-1 was not identified in 56 normal pancreatic ducts or 95 low-grade PanIN lesions; however, it was markedly raised in high-grade PanIN and PDAC. As such, mAbDas-1 proves highly specific for advanced PanIN and PDAC and could aid in preoperative diagnoses, as well as clinical risk stratification (157).

Using a PDAC genetically engineered mouse model, Dieter et al. employed cathepsin-activatable near-infrared probes combined with flexible confocal fluorescence laser microscopy to differentiate normal pancreatic tissue, low-grade PanIN, high-grade PanIN, and early PDAC (158). This method offers a promising approach to early diagnosis of PDAC with greater accuracy compared to traditional diagnostic methods.

Metabolomics techniques offer a promising approach for the detection and identification of cancer by facilitating the comprehensive analysis of trace metabolites in biological fluids and tissues (159–161). Notably, the metabolomic profiles of PDAC patients are distinct from that of healthy controls (161–163). Potential metabolic signatures can be identified by using nuclear magnetic resonance (NMR) spectroscopy combined with multivariate and univariate statistical analysis and screening for differential metabolites including glucose, amino acids, carboxylic acids, and coenzymes (164, 165). However, due to its relatively low sensitivity, NMR should be combined with other techniques like GC-MS and LC-MS to improve biomarker identification for early clinical detection and diagnosis of PDAC (165, 166). By effectively combining these techniques, researchers can maximize the potential of metabolomics as a diagnostic tool for PDAC.

MicroRNAs (miRNAs) are a class of small non-coding RNAs that are closely associated with cancer progression (167). Abnormal expression of miRNAs is believed to be an early event in pancreatic carcinogenesis (168). MiRNAs are implicated not only in PDAC progression but also in the process from PanIN to cancer. MiR-21 activates KRAS signaling via the transcription factor ELK1, promoting tumor cell proliferation, migration, and invasion (169). MiR-224 overexpression activates normal fibroblasts into tumor-associated fibroblasts, thereby increasing drug resistance in tumor cells (170). These studies demonstrate the significant impact of miRNA on promoting disease progression. Moreover, miRNAs are abundant and stable in serum, making them a promising marker for clinical diagnosis of PDAC (171–173). Several studies have identified specific miRNAs in PanIN, including miR-10, miR-16, miR-21, miR-100, and miR-155, which are not found in normal pancreatic ducts (174). Additionally, miR-196b has been identified as a potential biomarker for differentiating PanIN 3 from low-grade PanIN, thus facilitating screening of high-risk cancerous lesions and assisting in clinical treatment selection.