Pancreatic cancer is listed as one of the most lethal cancers, with a 5-year overall survival rate of <9%. In 2019, in the US, 56,770 new pancreatic cancer cases were confirmed, and its estimated deaths accounted for 7%–8% of all cancer-related deaths [1]. In addition, pancreatic cancer is predicted to become the second leading cause of cancer-related death in the next decade [2]. While advances in the treatment of other cancer types have dramatically improved the overall outcomes of patients, the incidence and mortality rates of pancreatic cancer have only decreased slightly over the past 30 years [1, 3,4,5]. The high mortality rate might be a consequence of the combination of a late diagnosis, resistance to therapies and insufficiency of the effective treatment modality.

To date, surgical resection remains the only potential curative treatment. However, only 15% of patients have resectable tumors. The majority of patients are diagnosed at an advanced stage and are mainly be treated with chemotherapy regimens. Although novel chemotherapy regimens have been established recently, such as FOLFIRINOX and gemcitabine plus nab-paclitaxel, their overall efficacy remains limited. The overall survival duration of patients with metastatic pancreatic cancer is still <1 year, regardless of treatment with FOLFIRINOX or nab-paclitaxel plus gemcitabine [6, 7]. In addition, long-lasting debates regarding the survival benefits of chemoradiation therapy exist. With research progress in molecularly targeted therapy, tumor therapy has undergone revolutionary changes. Some cancers, such as lung cancer, have entered the era of molecularly targeted therapy. However, in patients with PDAC, only the combination of gemcitabine plus erlotinib is associated with a statistically significant increase in survival compared to gemcitabine alone. However, the actual benefit is small, suggesting that only a small subset of patients might intrinsically benefit from this treatment. Nonetheless, molecularly targeted therapy remains the only hope for patients with PDAC.

Accordingly, many published studies have been conducted using the latest next-generation high-throughput sequencing technology, and it is widely accepted that pancreatic cancer is a disease of genetic alterations. The most commonly mutated genes are generated from The Cancer Genome Atlas (TCGA) data and are presented in Fig. 1. However, the number of druggable targets in individual patients is low. Thus, the classic progression model of PDAC must be re-examined, and possible solutions should be identified. PDAC originates from a series of precursor lesions, such as pancreatic intraepithelial neoplasia (PanIN), intraductal papillary mucinous neoplasm (IPMN), and mucinous cystic neoplasm (MCN) [8]. The dysregulation of several core pathways and a myriad of genomic alterations drive pancreatic tumorigenesis [9]. Whole-genome sequencing has revealed the main driver genes in pancreatic cancers, including KRAS, CDKN2A, TP53, and SMAD4/DPC4 [10,11,12,13]. These genes are mutated in different stages of precursor lesions, and their dysregulation promotes the differentiation and proliferation of pancreatic cancers (Fig. 2) [8]. In general, KRAS mutations emerge from stage 1 lesions (PanIN-1) to promote the initiation process, while CDKN2A mutations occur in PanIN-2 to facilitate further progression. In addition, mutations in TP53 and SMAD4 are frequently detected in PanIN-3 and invasive tumors, driving the proliferation and expansion of pancreatic cancers [14, 15]. Therefore, studies investigating the roles of these driver genes in pancreatic cancers are of paramount importance to provide additional strategies for use in clinical practice. Here, we review the biological functions of these driver genes and summarize their clinical implications reported to date.

Fig. 1: Mutation profile of pancreatic cancer in the TCGA dataset.
figure 1

Mutation information on 178 pancreatic cancers in the TCGA dataset was analyzed. KRAS/CDKN2A/TP53/SMAD4 are the most commonly mutated genes in pancreatic cancer, with mutation rates of 77%, 63%, 22%, and 16%, respectively. In addition, missense mutations and nonsense mutations are the main alteration types.

Fig. 2: Classical progression model of pancreatic cancer.
figure 2

Pancreatic cancer is considered a disease of multiple genetic alterations, and mutations in KRAS/CDKN2A/TP53/SMAD4 promote the initiation and progression of precursor lesions. KRAS mutations occur in the early stage of PanIN-1; the loss of cdkn2a occurs in PanIN-2; and the loss of p53 and smad4 occurs in the later stage of precursor lesions. A series of other mutations cooperate to promote the tumorigenesis and metastasis of PDAC.


General introduction

As the most common genetic driver in pancreatic cancers, the KRAS gene is mutated in ~93% of pancreatic cancers [16,17,18,19]. The KRAS protein is a small GTPase responsible for interacting with cell membrane growth factor receptors and controlling the switch of multiple signaling pathways and cellular processes. The proteins in the human RAS family usually consist of two functional domains: the G domain and the membrane-targeting domain. Different isoforms of the RAS family have a similar G domain but vary in the membrane-targeting domain. The G domain spans residues 1-164 and functions as a molecular switch for the downstream signaling pathway by binding to GTP/GDP. GTP-bound RAS prompts the membrane-targeting domain to interact with effector proteins and activate downstream signaling pathways, while GDP-bound RAS inactivates the process, and GTP/GDP conversion is catalyzed by the SOS1 protein. The most frequent mutation of KRAS in pancreatic cancer occurs in codon G12 of exon 2(e.g., G12D (40%) and G12C (33%)). Approximately 10% of mutations occur on codons 13, 61, 117, and 146 [20]. Point mutations in codon 12 of the KRAS oncogene prevent the conversion from GTP to GDP, resulting in the constitutive activation of downstream signaling pathways and markedly promoting carcinogenesis in various cancer types [20, 21].

KRAS mutations in pancreatic cancer

Similar to other cancers, pancreatic cancers are derived from a series of precursor lesions, among which pancreatic intraepithelial neoplasia (PanIN) is the most common lesion [22, 23]. Oncogenic KRAS mutations have been identified in 95% of PDAC tissues [9, 16, 17]. According to many studies, oncogenic KRAS mutations drive the initiation and progression of different stages of PanINs, and this process involves changes in the gene mutation rate, from 50% in PanINs to 95% in PDAC [17, 19]. In addition, factors that promote and activate KRAS, such as inflammation, oxidants and TGF-β, all contribute to the initiation and development of pancreatic tumors. In contrast, NSAIDs and antioxidants decrease KRAS activity and prevent tumor progression [19]. Nonetheless, KRAS also functions as a fundamental factor in the progression from IPMNs to PDAC [23].

Decades of research have discovered and clarified the complex picture of KRAS-regulated biological processes, including cell metabolism, tumor cell signaling, the tumor microenvironment, micropinocytosis, apoptosis, and redox homeostasis [24,25,26,27,28]. According to Sunil et al., the endogenous expression of KRASG12D induced all three stages of PanINs in all genetically engineered mouse models, and a small percentage of these animals developed metastatic PDAC [29]. In addition, KRAS mutations play an important role in the maintenance and proliferation of PDAC. Collins et al. constructed models with the reversible expression of oncogenic KRAS, and a reduction in KRAS expression led to rapid tumor relapse [30]. However, a KRAS mutation alone is insufficient to drive carcinogenesis, and hundreds of changes in gene expression, especially in tumor suppressor genes, cooperate to drive the formation of the final invasive PDAC [19, 31]. Moreover, due to the consecutive activation of oncogenic KRAS, the RAF/MEK/ERK pathway, PI3K/AKT/mTOR pathway, RalA/B pathway and NF-κB pathway are all activated to promote the proliferation of PDAC (Fig. 3) [32]. Our previous studies also revealed a novel KRAS/ERK/FBW7/cMyc pathway in PDAC cell lines, and all these effectors and pathways represent potential drug targets for further study [33,34,35].

Fig. 3: Pathways of key driver genes and therapeutic targets in pancreatic cancer.
figure 3

Oncogenic mutations in KRAS activate downstream signaling pathways, such as the PI3K/Akt/mTOR, KRAS/Ral, and KRAS/Raf/MEK pathways. Therapeutic methods include directly targeting KRAS, targeting upstream EGFR, or targeting downstream effectors such as PI3K, Akt, mTOR, Raf and MEK. Loss-of-function mutations in CDKN2A/TP53/SMAD4 attenuate the tumor suppressive functions of downstream signaling pathways. Therapeutic targets for tumor suppressor genes include restoring the function of wild-type p53, HSP90 inhibitors, vaccine therapy targeting mut-p53, Wee-1 kinase inhibitors (not shown), CDK4/6 inhibitors and TGF-β inhibitors.

KRAS mutations for early diagnosis

The high mortality rate of pancreatic cancer is closely related to the fact that only a small percentage of patients are diagnosed at the early stage [36]. Considering the crucial role of oncogenic KRAS in pancreatic cancer, scientists have tried to determine its potential efficacy in diagnosis and medical treatment since its discovery. Over 30 years ago, oncogenic KRAS mutations were also detected in duodenal fluid [37]. Examinations of KRAS mutations involve endoscopic ultrasonography-guided fine needle aspiration (EUS) and circulating tumor DNA (ctDNA) analyses. EUS-FNA is the main well-established tool for collecting cytological and histological samples from locally advanced PDAC [38]. The combination of a KRAS mutation assay and cytopathology dramatically increases the sensitivity, accuracy, and negative predictive value of pancreatic cancers compared to cytopathology alone [39]. Notably, the KRAS mutation assay not only facilitates a differential diagnosis of PDAC and pseudotumorous chronic pancreatitis but also helps distinguish these diseases from autoimmune pancreatitis [40].

Liquid biopsy has rapidly emerged in recent years as a promising tool for early diagnosis, monitoring the effect of treatment and predicting prognosis. In the study performed by Cohen et al., tumor-specific KRAS mutations were detected in 30% of plasma samples from 221 patients with pancreatic cancer, and the combination of KRAS mutations and elevated CA19-9 levels was more sensitive in detecting pancreatic cancer than CA19-9 levels alone [41]. In addition, the KRAS mutation detection rates in ctDNA are notably connected with the stages of tumor progression, namely, 53% in metastatic disease and 34% in localized disease [42]. Negative detection of KRAS mutations before treatment is closely associated with a good prognosis and therapeutic response, regardless of tumor resection. In addition, the emergence of KRAS-mutated ctDNA within 1 year after surgery may predict poor overall survival [43, 44]. Moreover, KRAS mutation in exosome-derived DNA (exoDNA) is also an important factor for predicting tumor resectability and the overall survival of patients with pancreatic cancer [45, 46]. However, the detection of KRAS mutations alone in both ctRNA and exoDNA inevitably leads to false-negative results, and thus, multiple biomarkers and mutations must be evaluated simultaneously to improve the accuracy, sensitivity, and specificity of pancreatic cancer detection. In addition, liquid biopsy is not very sensitive for the early detection of tumors <10 mm in size [47].

Targeting KRAS for treatment

Multiple methods targeting KRAS mutations have been proposed in the past three decades. Phase II/III studies of strategies targeting KRAS or KRAS-related pathways in pancreatic cancer are summarized in Table 1. Drugs that directly target KRAS are the most obvious option, but the lack of a potential binding pocket apparently renders this mutation “undruggable” [9, 48]. Recently, in parallel with a better understanding of KRAS biochemistry and new methods to identify potential chemicals that target KRAS, several strategies have been pursued for the direct targeting of KRAS, such as targeting nucleotide exchange and RAS-effector interactions [49]. A group of small molecules has been identified to be able to bind to KRAS and inhibit SOS-mediated nucleotide exchange, thus preventing the activation of KRAS [50, 51]. Ongoing clinical trials investigating treatments that directly target KRAS are listed in Table 2. Strategies targeting downstream effectors also represent a promising method, but the administration of single MEK, RAF, or PI3K inhibitors has been unsatisfactory in KRAS-mutated cancer [21, 52]. Indirect targeting of KRAS by inhibiting membrane localization is also a theoretically plausible option, but its clinical efficacy was extremely disappointing [53, 54]. In addition, because reovirus can induce oncolysis in the context of activated RAS signaling, clinical trials have examined the role of reovirus in PDAC [55, 56]. Moreover, the use of RNA interference (RNAi) to suppress the KRAS protein has been reported to inhibit pancreatic cancer growth in vivo, and its role is being explored in a clinical trial (NCT01676259) [57, 58]. However, despite 35 years of research on KRAS-mutated pancreatic cancer, very few effective drugs have been produced for clinical use. Moreover, most drugs that enter phase III clinical trials are closely related to KRAS pathways, indicating that strategies targeting KRAS still hold promise to conquer this disease in the future.

Table 1 Summary of phase II or III clinical trials investigating KRAS-related therapy in patients with pancreatic cancer.
Table 2 Ongoing clinical trials targeting mutations in KRAS.

Compared with wild-type KRAS, mutated KRAS is highly relevant to the poor survival rate of patients with PDAC [59,60,61]. According to the National Comprehensive Cancer Network (NCCN) guidelines for pancreatic cancer (version 1.2020), erlotinib is the only molecular therapy that has produced significant outcomes in combination with gemcitabine. Interestingly, a survival advantage was observed in patients with tumors expressing wild-type KRAS when erlotinib was prescribed with gemcitabine/capecitabine (median survival rates of 7.9 months and 5.7 months for patients with wild-type KRAS and mutated KRAS, respectively, P = 0.005) [62]. In addition, in a phase IIb study of gemcitabine/nimotuzumab, patients with wild-type KRAS had a better overall survival rate than patients with mutated KRAS [63]. However, the role of the KRAS status in predicting the efficacy of treatment with erlotinib remains elusive, and the minimal benefit of erlotinib has prevented its use as a clinical treatment [64,65,66].


General introduction

CDKN2A, cyclin-dependent kinase inhibitor 2A, which is located on chromosome 9p21, is one of the most important tumor suppressor genes, with a crucial role in the regulation of the cell cycle by directly or indirectly targeting CDK4/6-cyclins. Proteins encoded by CDKN2A, p14ARF and p16INK4A share exons 2 and 3 but differ in exon 1, resulting in the translation of two unrelated proteins that function through different pathways [67]. Notably, p16INK4A, one of the INK family inhibitors, binds to CDK4/6 and inhibits the activation of D-cyclins, further preventing the phosphorylation of retinoblastoma (Rb) and limiting cell cycle entry [68]. The p14ARFprotein also promotes cell cycle arrest by binding to and inactivating mouse double minute 2 homologue (MDM2), an E3 ubiquitin ligase that mediates the degradation of p53 (Fig. 3).

Biological and oncogenic functions

Numerous studies have confirmed important roles for CDKN2A in cancer development, aging and type 2 diabetes. Importantly, p16INK4A is expressed at high levels during islet regeneration, and the overexpression of p16INK4A inhibits beta-cell proliferation [69]. The expression of p16INK4Aalso increases with aging in both rodent and human islets, indicating that p16INK4A might be responsible for impaired beta-cell proliferation in aging mice [69, 70]. Increased expression of p16INK4A has also been observed in almost all rodents with aging, and this change occurs in parallel with decreased proliferative and regenerative capacity [71, 72]. In addition to its role in aging, CDKN2A is also one of the most frequently mutated tumor suppressor genes, and genetic alterations in CDKN2A have been detected in many types of cancers (30%–50% of pancreatic cancer cases) [73,74,75,76]. The inactivation of CDKN2A cooperates with KRAS mutations and drives the malignant transformation of the pancreas [77]. Loss-of-function p16INK4A mutations stimulate pancreatic neoplastic development in the intermediate or late stages, and the dysregulation of p14ARF advances tumor development and metastasis [73, 78]. Regarding PDAC, the loss of either p14ARF or p16INK4A facilitates malignant progression and differentiation [74]. In families with melanoma, genetic alterations in both p14ARF and p16INK4A dramatically increase the risk of pancreatic cancers [75]. Moreover, the deletion of CDKN2A in a mouse model promoted the tumorigenesis of pancreatic neuroendocrine tumors (PanNETs), the second most common pancreatic malignancy, and reduced survival [79]. Genetic alterations in CDKN2A in pancreatic cancer mainly include deletions, mutations, and promoter hypermethylation [80]. Approximately 40% of PanNETs harbor aberrant hypermethylation of CDKN2A, and a high level of methylation may predict the malignant behavior of PanNETs [81].

Prognostic role of CDKN2A mutations

In contrast to KRAS mutations, the prognostic role of CDKN2A mutations remains controversial among different studies of PDAC (Table 3). In 88 patients with pancreatic cancer treated with preoperative chemoradiation, the polymorphic genotypes of p16INK4A were significantly associated with a shorter median time to tumor progression, namely, 10.8 months for patients with polymorphic genotypes and 16.2 months for patients with the wild-type genotype [82]. The deletion of CDKN2A also results in poor outcomes for patients who have undergone partial pancreatoduodenectomy and radical lymphadenectomy [83]. Oshima et al. investigated the genetic status of 106 patients with PDAC undergoing radical surgery and discovered that the loss of CDKN2A was significantly correlated with lymphatic invasion and postoperative metastases [84].

Table 3 Prognostic and predictive roles of mutant driver genes in clinical trials.

Targeting CDKN2A function for treatment

CDKN2A mainly functions by inhibiting CDK4/6. Specific drugs targeting CDKN2A have not yet been reported, but a series of therapies targeting CDK4/6 has been implicated in patients with a loss of CDKN2A. Palbociclib is an oral, small-molecule inhibitor of CDK4/6 that has been approved by the FDA as a treatment for metastatic breast cancer [85]. By inducing apoptosis and cell cycle arrest, palbociclib enhances the therapeutic efficacy of gemcitabine and inhibits the invasiveness of PDAC cells [86]. Notably, these functions are mainly effective in Rb-positive PDAC cells, suggesting that the Rb protein, which is downstream of CDKN2A, might also serve as a predictive biomarker for palbociclib. The clinical efficacy of palbociclib in PDAC is still under investigation. Ongoing clinical trials targeting CDKN2A/TP53/SMAD4 for the treatment or diagnosis of pancreatic cancers are presented in Table 4. A combination of therapeutic and predictive biomarkers might be needed in future clinical trials.

Table 4 Ongoing clinical trials targeting P53/SMAD4/CDKN2A pathways for the treatment and diagnosis of pancreatic cancer.


General introduction

As the most frequently mutated tumor suppressor gene in all cancers, the TP53 gene is estimated to be mutated in 60%-70% of pancreatic cancers [17]. The TP53 gene encodes the p53 protein, which binds to specific sequences through its DNA binding domain and regulates the transcription of downstream molecules to exert its functions in various biological processes, including the cell cycle, mitochondrial respiration, cell metabolism, autophagy and stem cell maintenance and development [87]. TP53 is commonly activated by oncogenic mutations or cellular stress, such as DNA damage and oxidative stress, preventing p53 from interacting with MDM2/4 and therefore stabilizing p53. As its level increases, p53 increases the transcription of downstream genes, such as P21 and Bcl-2, thus driving cell cycle arrest and repairing or eliminating damaged cells to inhibit the accumulation of oncogenic mutations [88]. Interestingly, p53 both represses and induces the expression of different genes in a context-dependent manner [89]. Consistent with its function, mutations in TP53 usually occur in the DNA binding domain, and most of the mutations are missense mutations, providing a great opportunity for cancer cells to proliferate and survive in a mild stress environment [90]. However, mut-p53 does not simply lose its original function but rather gains other abilities to promote cancer development through different mechanisms, including remodeling the tumor microenvironment and enhancing cell metabolism [91, 92]. Once mutated, mut-p53 binds specifically to the Hsp90 chaperone machinery, a system that senses cellular stress, such as protein misfolding and oncogenic signaling. Hsp90 specifically blocks the activity of MDM2 and CHIP to prevent degradation of the p53 protein, resulting in the accumulation of dysfunctional p53 proteins in cells [91].

TP53 mutations in pancreatic cancer

Genetic alterations in p53 without loss of heterozygosity have been detected in early PanIN, and homozygous mutations in p53 have been observed in PanIN-3, indicating the potential for p53 to drive the carcinogenesis of PDAC [93]. Morton et al. constructed a mouse model with mut-p53 (Trp53R172H) and found that mut-p53 facilitated malignant transformation from premalignant lesions to PDAC [94]. Furthermore, the same study also showed that mut-p53 enhanced the invasion of pancreatic tumor cells and promoted lymph node metastasis. In contrast, supplementary expression from a retroviral p53 vector substantially inhibited the growth of primary pancreatic tumors [95]. In addition, when TP53 was mutated together with KRAS in the mouse pancreas, metastatic PDAC was widely distributed, and the tumor tissue exhibited a high degree of genomic instability [96].

Prognostic role of TP53 mutations

Despite the importance of TP53 mutations in tumorigenesis and progression, many studies have investigated the predictive and prognostic roles of p53 expression and reported seemingly controversial results (Table 3). Although mutations in TP53 increase the stability and accumulation of the p53 protein, immunohistochemical staining for p53 is not related to the overall survival of patients with PDAC who undergo complete pancreatic resection [59, 60, 97]. However, in patients with metastatic pancreatic cancer treated with FOLFIRINOX, high levels of p53 expression in the tumor are significantly correlated with a poor overall survival rate but not a poor PFS or response rate [98]. In patients receiving adjuvant gemcitabine treatment, p53 expression is inversely related to disease-free survival and overall survival, consistent with the high tolerance to gemcitabine cytotoxicity exhibited by a mut-p53 cell line [99, 100]. While mutated and wild-type p53 do not result in a difference in patient survival, an analysis of cancer genomic data showed that specific mutation types, such as mutations at Arg248 and Arg282, result in a notably poor outcome in several tumors [101]. Moreover, the abnormal expression of p53 predicts a high risk of locoregional recurrence (P = 0.020) [84]. In summary, genetic alterations in TP53 are capable of predicting advanced tumor progression, but their prognostic status requires further study.

Treatments targeting TP53 mutations

Approximately half of all human cancers harbor mutations in TP53, and multiple strategies targeting TP53 have been proposed. Gain-of-function mutations in TP53 markedly increase the proliferation and metastasis of tumor cells, consistent with the decreased function of wild-type p53. Therefore, the identification of compounds or therapies that restore wild-type p53 activity or delete the mut-p53 protein shows promise for treating TP53-mutated cancers [102]. Several compounds have been reported to restore the transcriptional activity of the mut-p53 protein, such as PRIMA-1 and APR-246, which bind to thiols in the core domain of mut-p53 [103]. PRIMA-1 rescues the function of mut-p53 and induces apoptosis and cell cycle arrest in pancreatic cancer cells [104]. In addition, PRIMA-1 increases the sensitivity of p53-mutated PDAC cells to gemcitabine and erlotinib. An inhibitor of histone deacetylase 6 (HDAC6), a positive regulator of HSP90, also blocks the oncogenic activity of mut-p53 by increasing the degradation of mut-p53 without altering wild-type p53 [105]. Furthermore, HSP90 inhibitors exhibited significant efficacy in reducing the growth and angiogenesis of pancreatic cancer both in vitro and in vivo but failed in a phase II study of patients with advanced pancreatic cancer [106,107,108].

Furthermore, the restoration of wild-type p53 by delivering nanoparticles carrying plasmid DNA induces apoptosis in pancreatic cancer cells [109]. The combination of wt-p53-expressing plasmid DNA and gemcitabine significantly inhibited tumor proliferation compared with gemcitabine alone, with 77.3% and 61.7% reductions in tumor growth, respectively [109]. A phase II study is ongoing to evaluate the efficacy of the combination of targeted p53 gene therapy plus gemcitabine/nab-paclitaxel in patients with metastatic pancreatic cancer (NCT02340117). Moreover, Chung et al. immunized patients with solid tumors with a p53-expressing modified vaccinia Ankara virus (p53MVA), which activates peripheral T cells to kill cells overexpressing mut-p53, and a clinical response was observed in 3/11 patients [110].

Because cancer cells with mut-p53 lose control over the G1 checkpoint and rely heavily on the G2 checkpoint to repair cellular DNA damage, treatments that inhibit the key regulator of the G2 checkpoint, Wee1 kinase, are predicted to abrogate the DNA repair process and induce synthetic lethality in TP53-mutated cancer cells [111]. A phase II clinical trial evaluated the efficacy of adavosertib (AZD1775), a Wee1 kinase inhibitor, in combination with gemcitabine and radiotherapy in patients with locally advanced pancreatic cancer and reported an extremely encouraging result: a median OS duration of 21.7 months and a median PFS duration of 9.7 months [112]. Although the median OS duration was 15.2 months for patients receiving chemoradiotherapy in LAP07 trials, Wee1 inhibitors dramatically improved clinical outcomes and hence provided opportunities for future treatment [113]. Last but not least, downstream molecules of mut-p53 and the ability to block the formation of mut-p53 complexes and other proteins represent targets for anticancer drugs [102, 114].


General introduction

SMAD4 (Sma (Caenorhabditis elegans) mothers against decapentaplegia homologue 4), also known as DPC4 (deleted in pancreatic cancer, locus 4), is a tumor suppressor gene that is mutated in a wide range of diseases and cancers, particularly in pancreatic cancers, with a mutation rate of ~20%–50% [11, 115]. The SMAD family consists of 8 proteins and plays a crucial role in mediating TGF-β signaling. Although SMAD4 is not obligatory for the activation of TGF-β signaling pathways, it is indispensable for producing a strong signaling response [116]. SMAD4 shuttles between the nucleus and cytoplasm and forms a heterodimeric complex with SMAD2/SMAD3, which is phosphorylated by activated TGF-β receptors. The SMAD complex subsequently enters the nucleus and interacts with downstream proteins to regulate the transcription of target genes. The E3 ubiquitin ligase ectodermin targets SMAD4 for degradation in the nucleus and antagonizes TGF-β signaling, thus blocking downstream pathways regulating cell differentiation and proliferation [117]. In contrast, the deubiquitinase USP9x reverses the ubiquitination caused by ectodermin and restores the function of SMAD complexes [118].

SMAD4 mutations in pancreatic cancer

The TGF-β/SMAD4 signaling pathway mediates the growth of cancer cells by promoting cell cycle arrest, apoptosis and DNA damage repair, while genetic alterations in SMAD4 attenuate the tumor suppressor function of the TGF-β pathway [119, 120]. In contrast, enhancement of the epithelial-mesenchymal transition (EMT) process in a SMAD4-dependent manner is commonly presumed to increase the invasion and metastasis of cancer cells [121]. As one of the driver genes, SMAD4 is mutated in half of PDAC cases, with homozygous deletions occurring in 30% of cases and chromosome allelic loss existing in 20% of cases, and the mutations remarkably decrease immunohistochemical staining for the SMAD4 protein [122, 123]. In addition, loss of SMAD4 expression has been detected in high-grade precursor lesions rather than in low-grade lesions, suggesting that the inactivation of SMAD4 promotes progression to a later stage of tumorigenesis [124]. Similarly, SMAD4 deletion alone in a transgenic mouse model was insufficient to initiate the development of PDAC, while SMAD4 inactivation substantially enhanced the progression of KRASG12D-initiated neoplasms [125]. In addition, in mouse models of PDAC carrying mutations in both KRAS and TRP53, the inactivation of SMAD4 increased metastasis, but the expression of wild-type SMAD4 decreased metastasis and increased proliferation [126]. Our previous studies have also demonstrated a disparity in progression and migration that might result from TGF-β-induced autophagy and PGK1-mediated metabolic reprogramming, depending on the SMAD4 status [127, 128]. Alterations in SMAD4 also regulate the differentiation of PDAC: SMAD4 insufficiency is beneficial to retain epithelial features, while wild-type SMAD4 promotes the EMT process [125]. Surprisingly, the EMT process is dispensable for pancreatic cancer metastasis but promotes the proapoptotic function of the TGF-β signaling pathway [129, 130]. SMAD4 deletion remarkably increases the resistance of both PDAC cell lines and mouse models to radiotherapy, and this decrease in radiosensitivity is correlated with the induction of ROS production and autophagy [131].

Prognostic and predictive roles of SMAD4 mutations

The relationship between SMAD4 mutations and clinical outcomes has been extensively investigated in numerous studies but remains elusive [132,133,134]. Andrew et al. investigated the relationship between SMAD4 expression and overall survival in 119 patients with PDAC and showed that the loss of SMAD4 expression was remarkably associated with an improved median survival in patients who underwent pancreatic resection in the univariate analysis (13.6 vs 6.4 months, P = 0.0257) [132]. However, the majority of other studies reported shorter overall survival in patients with SMAD4-inactivated PDAC (Table 4). The loss of SMAD4 also predicts a significant benefit from postoperative adjuvant chemotherapy (P = 0.002) [135]. The loss of SMAD4 is also significantly associated with an increased metastatic burden, and wild-type SMAD4 is related to local recurrence, indicating that systematic chemotherapy might achieve satisfactory outcomes in patients with SMAD4 mutations [136, 137]. However, different studies have reported contradictory findings regarding the role of the SMAD4 status in predicting the recurrence pattern of patients with resected PDAC [135, 138].

Treatments targeting SMAD4

Although several anticancer agents targeting SMAD4-deficient cells have been discovered, no results from animal models or human trials have been published to date [139, 140]. Since the TGF-β signaling pathway promotes the progression and metastasis of PDAC in the absence of SMAD4, strategies targeting TGF-β might provide new methods for clinical treatments. Vactosertib (TEW-7197), an inhibitor of TGF-β signaling, combined with nanoliposomal irinotecan and 5-FU dramatically improved the survival outcome of an animal model of pancreatic cancer and suppressed the migration and invasion of pancreatic cancer cells [141]. Furthermore, a phase Ib clinical trial is ongoing to evaluate the efficacy of vactosertib with FOLFOX in patients with metastatic pancreatic cancers (NCT03666832). Synthetic lethality is another potential option available for targeting SMAD4 mutations, as the loss of SMAD4 is commonly accompanied by a passenger deletion of mitochondrial malic enzymes 2 (ME2), a housekeeping gene that functions with ME3 to sustain NADPH synthesis in mitochondria. Inhibition of ME3 substantially slows the growth and proliferation of ME2-null pancreatic cells, suggesting that compounds targeting ME3 are promising treatments for patients with SMAD4 mutations [142]. Moreover, due to the potential ability of wild-type SMAD4 to predict locally advanced pancreatic cancer, patients with intact SMAD4 expression might benefit more from intense local therapy than systematic chemotherapy. However, no ongoing registered clinical trial is evaluating the role of the SMAD4 status in radiotherapy. Further clinical trials examining the SMAD4 status might be able to improve the efficacy of radiotherapy in the treatment of local pancreatic cancers.

Other genes


BRCA1/2 are the most common genes mutated in familial pancreatic cancers, and mutations in BRCA1/2 increase the risk of pancreatic cancer susceptibility [143]. Germline BRCA1/2 mutations occur in 4%–7% of all pancreatic cancers [144]. As the key factors involved in DNA damage repair, BRCA1/2 cooperate to mediate recombination between homologous DNA sequences to repair double-strand DNA breaks (DSBs), while mutations in BRCA1/2 lead to inappropriate DSB repair during the cell cycle and gross chromosomal rearrangements [145]. In addition, as one of the genes involved in Fanconi’s anemia (FA) pathways, BRCA2 and other FA proteins, together with BRCA1, are required for the repair of DNA interstrand cross-links [146]. In BRCA1/2-deficient cells, the accumulation of DSBs and genomic instability drive malignant transformation and progression [147]. Likewise, poly (ADP-ribose) polymerase 1 (PARP1) is responsible for the repair of single-strand DNA breaks (SSBs), and the simultaneous dysregulation of PARP1 and BRCA1/2 results in genomic instability and cell death (Fig. 4). This synthetic lethality provides evidence for the ability of PARP1 to serve as a potential target for the treatment of BRCA1/2-mutated cancers.

Fig. 4: DNA damage repair pathway and the mechanism of PARP inhibitors.
figure 4

DNA damage repair mainly includes the repair of DSBs and SSBs. PARP and ATR/CHK1 are responsible for SSB repair, while ATM, BRCA1/2 and other BRCAness-related genes are necessary for the homologous recombination repair of DSBs. PARP inhibitors block the repair of SSBs and increase DSBs. Mutations in germline BRCA1/2 or other BRCAness-related genes impair the homologous recombination repair of DSBs, leading to the accumulation of DSBs. The dysfunction of two pathways causes synthetic lethality, genomic instability and cell death.

In parallel with the roles of BRCA1/2 in DNA damage repair, advanced pancreatic cancers with BRCA 1/2 mutations also respond well to chemotherapies containing DNA-interacting regimens, such as platinum compounds [148, 149]. PARP inhibitors are effective in patients carrying germline BRCA mutations with several types of advanced cancers, including breast cancer, prostate cancer and ovarian cancer [150,151,152]. In a phase III clinical trial of 154 patients with germline BRCA-mutated metastatic pancreatic cancers, the PARP inhibitor olaparib dramatically prolonged progression-free survival (PFS) compared with the placebo (7.4 months vs 3.8 months, respectively), while no statistically significant difference in overall survival was observed between the olaparib and placebo groups. According to the NCCN guidelines version 1.2020, olaparib has been highlighted as a maintenance therapy for patients with metastatic PDAC carrying germline BRCA1/2 mutations who have not experienced disease progression. Furthermore, ongoing clinical trials are evaluating the efficacy of PARP inhibitors against both germline and somatic mutations in BRCA1/2 (NCT03601923). In addition, genetic alterations in the PALB2, CHK2, ATM and RAD51 genes result in defects in homologous recombination and DNA repair in the absence of BRCA1/2 mutations, and this phenocopy of BRCA1/2 mutations is defined as BRCAness [153]. Concerning mutations in other genes involved in BRCAness, their response to PARP inhibitors is also being investigated in patients with advanced PDAC (NCT03601923 and NCT04171700).


ATM, ataxia telangiectasia mutated, is an indispensable gene that senses and repairs DNA damage [154]. As a serine/threonine kinase, ATM reacts to cellular DSBs by phosphorylating downstream proteins to activate multiple cellular processes, including cell cycle arrest, apoptosis and DNA repair [154]. Hu et al. conducted a large case-control study of 3030 patients with pancreatic cancer and found that germline mutations in ATM occurred in 2.3% of patients compared with 0.37% in the normal population (OR, 5.71; 95% CI, 4.38–7.33) [155]. In addition, next-generation sequencing of two pedigrees of familial pancreatic cancers revealed that loss-of-function mutations are the main type of ATM mutation and are highly correlated with the predisposition to familial pancreatic cancer [156]. Based on these studies, ATM mutations promote the tumorigenesis and malignancy of pancreatic cancers.

In contrast to BRCA1/2 mutations, the loss of ATM is an independent prognostic factor for poor overall survival in patients with resectable pancreatic cancers [157,158,159]. Lukas et al. constructed a transgenic mouse model and cell lines with ATM deletion and showed that the loss of ATM enhanced the malignant features of pancreatic cancer cells, such as genomic instability and migratory properties, and promoted proliferation under metabolic stress [160]. In addition, in the same study, both in vitro and in vivo experiments confirmed that ATM deficiency increases radiosensitivity and induces synthetic lethality in combination with PARP inhibitors. A phase II proof-of-concept trial is being conducted to investigate the role of the PARP inhibitor niraparib in treating advanced PDAC patients carrying mutations in ATM and other BRCAness-related genes (NCT03601923).

In addition to PARP1, the ATR (Ataxia telangiectasia and Rad3-related) protein also mediates SSB repair, and its inhibitors are capable of inducing synthetic lethality in the context of ATM deletion [161]. The PARP inhibitor olaparib or the ATR inhibitor VE-822 dramatically inhibited the growth of ATM-deficient pancreatic cancer in vitro and in vivo [160]. Moreover, the combination of olaparib, an ATR inhibitor, and cisplatin has been tested in a clinical trial for refractory solid tumors (NCT02723864). Two distinct kinase signaling cascades account for the response to cellular DNA damage, as checkpoint kinase 1 (CHK1) participates in ATR-mediated SSB repair, while checkpoint kinase 2 (CHK2) is involved in ATM-mediated DSB repair [162]. Inhibitors of CHK1 or CHK2 are capable of eliciting the cytotoxicity associated with chemotherapy in pancreatic cancer cells, and multiple studies have been performed to determine their efficacy in clinical practice [163,164,165]. Homologous recombination-related gene mutations are estimated to occur in 15.4% of PDACs, indicating that a considerable percentage of patients with pancreatic cancer might benefit from inhibitors targeting the DNA repair process [166]. Moreover, although numerous potential targets in the DNA repair process have been identified, PARP inhibitors are the only drugs available for personalized therapy in patients with pancreatic cancers.


PALB2, partner and localizer of breast cancer 2 (BRCA2), interacts with BRCA1/2 and modulates the localization of BRCA2 to facilitate the homologous recombination process during DNA damage repair [167]. Similar to BRCA2, PALB2 is also one of the genes involved in Fanconi’s anemia pathways, and mutations in BRCA2 and PALB2 have been confirmed to increase the susceptibility to breast cancer [146]. In addition, PALB2 is mutated in 0.6%–3% of hereditary pancreatic cancers, while the mutation rate varies in different populations [168,169,170].

While PALB2 mutations are estimated to increase the risk of breast cancers 5–9-fold in different age groups, the relation between PALB2 mutations and the predisposition to pancreatic cancer remains controversial [155, 166, 171, 172]. In a single case report of metastatic and gemcitabine-resistant pancreatic cancer, Maria et al. found that mitomycin C and cisplatin significantly prolonged symptom-free survival to at least 3 years, and exome sequencing revealed the connection of PALB2 mutations with high sensitivity to DNA damaging agents [173]. Moreover, interstrand crosslinking agents dramatically inhibited pancreatic cancer growth in mouse models with deletions of PALB2 and BRCA1/2 [174]. In a phase I study of the PARP inhibitor talazoparib, a clinical benefit was observed in a patient with a PALB2 mutation, suggesting an expanded range of potential targets of the PARP inhibitor [175]. Due to its close relationship with BRCA1/2, several ongoing clinical trials are investigating the role of the PARP inhibitor in treating PALB2-mutated pancreatic cancers (NCT04300114 and NCT03337087).


The BRAF protein is a member of the RAF family of serine/threonine protein kinases. With the wide prevalence of KRAS mutations in pancreatic cancer, downstream signaling pathways, such as RAF/MEK/ERK, PI3K/AKT/mTOR, and RalA/B, all enhance cancer progression, proliferation and differentiation [176]. As one of the most common mutations in melanoma, BRAF mutations, mainly the BRAF V600E point mutation, have been detected in 1.4%-3% of pancreatic cancers and are mutually exclusive with KRAS mutations [177,178,179]. According to Eric et al.,the BRAF V600E mutation alone was sufficient to drive PanIN lesions in the mouse pancreas, while the combination of BRAF and TP53 mutations led to the formation of PDAC [180]. In the same study, MEK1/2 inhibitors induced a profound survival benefit in mice with PDAC. Therefore, BRAF mutations and RAF/MEK/ERK signaling play a pivotal role in the initiation and progression of PDAC expressing wild-type KRAS.

BRAF inhibitors, including vemurafenib and dabrafenib, have been used in clinical practice and dramatically changed the treatment of melanoma expressing the BRAF V600 mutant[181]. In addition, the combination of dabrafenib and the MEK inhibitor trametinib significantly improved the PFS outcome of patients with metastatic melanoma carrying the BRAF V600 mutation compared with dabrafenib alone [182]. The efficacy of dabrafenib and trametinib has also been verified during adjuvant therapy for patients with stage III melanoma carrying BRAF V600 mutations [183]. Kazimierz et al. described a patient with advanced PDAC carrying BRAF mutations, and dabrafenib dramatically improved the patient’s clinical condition for 6 months [184]. However, no clinical trials have been conducted to examine the efficacy of BRAF inhibitors in patients with metastatic pancreatic cancers.

Challenges and perspectives

What factors might be responsible for the failure of various molecularly targeted therapies: from the laboratory to the clinic

For the past several decades, a milieu of genetic targets have been tested for potential efficacy, with many of them succeeding in the preclinical stage but failing in clinical trials. Many mechanisms have been identified to explain this frustrating condition, including the complex biological features and microenvironment of PDAC [185, 186]. The high density of stromal cells and interaction between tumor cells and the microenvironment might lead to an unsatisfactory drug response and chemotherapy resistance [187]. As a result, reliable preclinical tools that imitate real tumor biology in the human body are difficult to establish. Different cell lines and genetically engineered mouse models have been widely used to examine the functions of genetic drivers. However, the cell lines only recapitulate a small group of patients with PDAC, and the function of the extracellular matrix is easily neglected in cell-based experiments [188]. Genetically engineered mouse models represent state-of-the-art methods to investigate the functions of genetic alterations and responses to new drugs, but species disparity and other underlying mechanisms, such as the effect of microbes, limit the ability of this model to assess new chemotherapies and detailed pancreatic tumor biology [176]. As only a small percentage of targeted drugs have entered phase III trials or clinical practice in the past several decades, we must reflect on our research strategies and obtain additional insights into the biology of PDAC.

Recently, organoids have emerged as new techniques and reliable models of human organs and diseases in vitro. Compared to the low cellularity of the primary tumor tissue, organoids are derived from cancer cells and possess high neoplastic purity, which may assist researchers in identifying more actionable genetic alterations. In addition, organoids avoid the differences between human tumors and mouse models because they are directly constructed from the tumor tissue rather than from injecting tumor cells into mice [189]. Tiriac and colleagues successfully generated pancreatic cancer organoids from human samples and found that the organoids exhibit high similarity to the primary tumor specimens in terms of the genetic hallmarks. They also analyzed the therapeutic profile of organoids in response to different treatments, and this pharmacotranscriptomic signature showed high concordance with chemotherapy sensitivities [190]. Moreover, alterations in the KRAS or TP53 gene in organoids promote the initiation and progression of PDAC [191]. Without the highly dense stroma, the organoid potentially represents a more efficient preclinical tool to test the effects of molecularly targeted therapies on pancreatic cancer cells. Notably, coculture of cancer cells, the stroma and other peritumoral components in an organoid model can provide an environment that is similar to that of pancreatic cancer in humans. Therefore, new tools and therapeutic methods to identify and examine more molecular medicines are needed in future studies.

What is the role of genetic testing in the treatment of pancreatic cancer?

A precise therapy based on genomic data might represent a new era in cancer treatment. A mixture of genetic mutations endow pancreatic cancer with different properties in different patients, and personalized therapy based on the mutation types might provide unprecedented clinical benefits. An analysis of molecular profiles revealed that pancreatic cancer is not a single disease, and this heterogeneous disease has been divided into several subgroups with various responses to chemotherapy [192]. In particular, patients with a defect in homologous recombination exhibit a satisfactory clinical response to platinum-based chemotherapy. Olaparib, the only orphan drug used to treat pancreatic cancer, has potential therapeutic efficacy in patients not only with germline BRCA1/2 mutations but also with other BRCAness-related gene mutations. The anti-PD-1 receptor antibody pembrolizumab dramatically reduced tumor progression in patients with solid tumors presenting with high microsatellite instability or mismatch repair deficiency, which accounts for <2% of patients with PDAC [193]. Moreover, the FDA approved the TRK inhibitors larotrectinib and entrectinib as treatments for solid tumors with NTRK gene infusions, and TRK inhibitors were tolerated and promoted relatively prolonged survival [194]. Therefore, the identification of genetic alterations provides opportunities to administer precise chemotherapy, particularly in patients in whom first-line therapy has failed. According to the NCCN guidelines version 1.2020, actionable targets include fusions of ALK, NRG1, NTRK and ROS1, mutations in BRAF, BRCA1/2, HER2, KRAS and PALB2, and mismatch repair deficiency. In addition, an examination of the response of xenograft tumors can predict the chemotherapy scheme resulting in the greatest sensitivity to increase the overall survival rate [173].

However, many limitations in genetic testing still exist, since only a small percentage of patients with genomic results ultimately receive genome-guided therapy [195, 196]. First, at least 20 days are needed to determine the genomic profile, which is a relatively long period to wait in some cases. Second, even if the genomic profile is available, only 10% of patients have actionable genetic alterations that have been verified in clinical trials, and the clinical benefit of precision medicine in a single individual remains unclear. Moreover, the high cost of genetic sequencing and the related processes may also limit its clinical applications.

Personalized therapy and combinatorial therapy in pancreatic cancer

The increasing prevalence of next-generation sequencing has made it possible to identify druggable genetic alterations. As discussed before, patients with germline BRCA mutations, ALK fusions, mismatch repair deficiency and other aberrations have benefitted from targeted therapies. However, only a small percentage of patients harbor these aberrations. Moreover, the feasibility of identifying and utilizing actionable aberrations as a routine clinical practice still needs to be confirmed and normalized in additional trials [197]. Notably, the aforementioned targeted therapies are related more to cancer treatment than the characteristics of pancreatic cancer. Transcriptional profile analysis has uncovered subtypes of PDAC and their different responses to chemotherapies. Our previous research also demonstrated that patients with a high strain ratio in EUS respond well to gemcitabine plus nab-paclitaxel, which indicates that a high strain ratio could provide guidance for the utility of stroma-disrupting agents [198]. Therefore, integration of the clinical and molecular information of pancreatic cancers is of significance to identify subgroups to offer more strategies for personalized therapies.

Due to the high malignancy of pancreatic cancer, only a few drugs can be applied to systemic therapy (Table 5). Since the end of the last century, gemcitabine-based therapies have been widely tested in clinical trials, but only a few combinations have been proven to significantly extend overall survival with systematic therapy. The combined use of targeted drugs has also been reported in several clinical trials, but most failed to provide better efficacy. In the clinical trial SWOG S1115, the combination of AKT and MEK inhibitors did not result in improved overall survival [199]. In addition, the high toxicity of combinatorial therapy remains an important issue, and some studies reported that only patients with a good performance benefitted from gemcitabine-based combinatorial therapy [200]. However, the success of FOLFIRINOX and nab-paclitaxel plus gemcitabine in the past decade suggests that combinatorial therapies of different chemotherapeutic agents could still be inevitable in further trials. Appropriate combinations and subgroup analyses might also be necessary in further trials.

Table 5 Summary of drugs showing favorable result in phase III clinical trials of systematic therapy.


KRAS, CDKN2A, TP53, and SMAD4 have been confirmed to be mutated in a wide range of pancreatic cancers and play a crucial role in driving tumorigenesis and metastasis through different mechanisms. Here, we conducted a general review of the biological functions and clinical implications of these four driver genes in pancreatic cancer. Despite tremendous efforts to investigate the functions of driver gene mutations, with the majority of studies focused on KRAS, the detailed mechanism remains elusive, and clinical trials have seldom reported significant benefits on overall outcomes. Considering the sophisticated crosstalk among these altered genes, monotherapy inhibiting a single target is unlikely to produce a remarkable clinical benefit, and a combination of multiple targeted drugs might provide further opportunities for treatment.