Molecular alterations and targeted therapy in pancreatic ductal adenocarcinoma

Advanced technologies facilitate the diagnosis of PDAC and the detection of tumour mutations. In addition to tumour biopsies, NGS has been performed using multiple types of specimens, such as pancreatic cyst fluid [5], secretin-stimulated juice [6], and cell-free DNA collected from the blood [7]. The use of more easily acquired specimens not only facilitates PDAC screening [8] but also obviates complications and costs.

Whole-genome sequencing reveals the mutational landscape of PDAC, and PDAC has been divided into four subtypes according to the variations in chromosomal structure: stable, locally rearranged, scattered, and unstable, each of which has its own distinctive mutational signatures [9, 10]. Researchers have also attempted to combine transcriptomic and genomic analysis to define PDAC subtypes because the mutational and transcriptional profiles do not overlap and an integrated genomic and transcriptomic analysis may reveal PDAC heterogeneity more thoroughly [11, 12].

The categorization of PDAC into various subtypes has potential clinical applications, as the basis of precision oncology is differentiating patients who may respond to a certain treatment from others and recognizing promising therapeutic targets [13]. Inspiringly, The Know Your Tumour programme revealed that 26% of the PDAC profiles harboured actionable molecular alterations, and molecularly matched precise therapy for patients with PDAC substantially improved their overall survival (OS) (hazard ratio (HR) = 0.42, P value = 0.0004) [14].

Tumorigenesis mainly results from genetic aberrations [15, 16]. As the amount of information about the genetic events involved in PDAC increases, the identification of ideal therapeutic targets is becoming possible. The aberrant genetic events in PDAC are generally divided into oncogene activation and tumour suppressor inactivation, and the four major genetic mutations observed in PDAC occur in KRAS, TP53, CDKN2A and SMAD4. These four commonly mutated major genes have been used to characterize PDAC and provided a pleiotropic roadmap for identifying ideal targets that may benefit most patients [17]. PDAC develops through a stepwise progression, and the progression from preneoplastic lesions to PDAC is a process characterized by the accumulation of genetic mutations. Early-stage precancerous lesions already appear to harbour mutations that are required for PDAC progression [18, 19]. For example, the most common KRAS and TP53 mutations are detected in early-stage intraepithelial neoplasia [20], suggesting that they play an important role in tumour onset.

In addition to the four major canonical genes involved in PDAC, genes involved in stabilizing chromatin, remodelling chromatin or editing point mutations in cancer cells, e.g. BRCA, APOBEC and KDM6A, also warrant investigation. Their low mutation frequencies in PDAC raise doubt about their clinical importance. Nonetheless, the poor prognosis of patients with PDAC suggests that any target, even if few people benefit from a treatment targeting that gene, is encouraging and merits investigation. Based on the aforementioned genetic events, researchers have attempted to therapeutically target these genetic variants and the altered pathways. In general, targeted treatment has been implemented using three approaches: inhibiting the dysregulated activation of oncogenes, interfering with the inactivation of tumour suppressors and exploiting the biological functional deficiency of certain genes, such as BRCA. Recent genetic-based explorations of precise targets in PDAC are shown in Table 1.

The most well-known oncogene involved in PDAC is RAS. RAS plays important roles in the signalling pathways regulating cell growth and differentiation to promote cell proliferation and differentiation and inhibit apoptosis. RAS switches between the inactive GDP-bound state and the active GTP-bound state, and recruited RAS guanine nucleotide exchange factors [21] and GTPase-activating proteins [22] are responsible for managing the transient activation of RAS.

KRAS mutations are the most common mutations identified in human solid tumours, and approximately 90% of patients with PDAC harbour the G12 mutation in KRAS [23‐26]. The most frequent point mutations at G12, G13 and Q61 [22] inhibit the intrinsic GTPase activity of RAS, thus sustaining the GTP-bound state of the RAS protein, which is established to be oncogenic [27, 28] (Fig. 1). Constitutively activated KRAS subsequently upregulates the endogenous expression of the upstream protein epidermal growth factor receptor (EGFR) and induces its hyperactivation [29, 30], and increased RAS levels and EGFR activity induce robust increases MEK/ERK activity, leading to intraepithelial neoplasia [31]. Furthermore, the overexpressed CA19-9 modifies fibulin-3 and enhances its interaction with EGFR, suggesting that CA19-9 and EGFR play intricate roles in PDAC tumorigenesis [32].

None of the direct KRAS inhibitors have reached clinical application, despite more than three decades of intensive effort; hence, KRAS was once considered an undruggable therapeutic target [33]. This frustrating fact is partially due to the multiple alternative signalling pathways of KRAS [34‐36]. Aberrantly activated RAS triggers downstream signalling by the RAF/MEK/ERK pathway, the PI3K/PDK1/AKT/mTOR pathway, RALGDS, TIAM1, and RIN1 [21]. These molecules further translocate to the nucleus and function as transcriptional modulators.

KRAS G12C provides a specific cysteine for drugs to bind, and thus small molecules have been designed to irreversibly bind this specific mutant target. By screening cysteine-reactive compounds, two fragments (6H05 and 2E07) were chosen as KRAS G12C-specific inhibitors [37, 38]. ARS853 was efficacious in KRAS G12C mutant cancer cells through the trapping mechanism [39], and ongoing phase I/II trials (NCT03785249 and NCT04330664) are assessing the efficacy of MRTX849, a small molecule that selectively modifies the mutant cysteine residue in KRAS G12C [40]. The relative frequency of the KRAS G12C mutation in PDAC is approximately 3% [25], suggesting that a certain subgroup of patients with PDAC may benefit from this type of treatment. In addition to small molecule inhibitors, RNA interference has been applied to target KRAS directly. Advances in endoscopic ultrasonography have assisted with the accurate placement of RNA interference molecules, such as siG12D-LODER™, into the parenchyma of patients with PDAC, and phase I/IIa trials have confirmed that this therapeutic strategy is well tolerated [41]. Engineered exosomes facilitate RNA interference efficiency as well [42] and may be applied as treatments for KRAS-mutant PDAC.

First-generation EGFR inhibitors, such as gefitinib and erlotinib, show little efficacy (median disease-free survival of patients treated with erlotinib: HR = 0.94, 95% confidence interval (CI) 0.76–1.15, P value = 0.26) [3, 4], partly due to the resistance caused by the non-EGFR members of the ERBB family [43, 44]. Irreversible tyrosine kinase inhibitors, such as afatinib and neratinib, have been developed to prevent the activation of the entire ERBB family. According to the results of previous clinical trials, afatinib is a more promising choice when selecting treatment for patients with KRAS-mutant lung cancer compared with gefitinib [45] or erlotinib [46], and a clinical trial of the efficacy of afatinib in patients with PDAC is ongoing (NCT02451553). Another EGFR inhibitor, nimotuzumab, improved the OS of patients with locally advanced or metastatic pancreatic cancer in a phase II trial (the median OS was 8.6 months vs 6.0 months, HR = 0.69, P value = 0.03), and patients with KRAS wild-type PDAC appear to benefit more from nimotuzumab than patients with KRAS mutant PDAC (the median OS was 11.6 months vs 5.6 months, P value = 0.03) [47]. In contrast, vandetanib failed to show efficacy (the median OS was 8.83 months vs 8.95 months, HR = 1.21, P value = 0.303) [48]. Another clinical trial indicated no benefit of cetuximab in the recruited patients either (the median OS was 6.3 months vs 5.9 months, HR = 1.06, P value = 0.23) [49]. These unsatisfactory outcomes suggest the presence of other potential resistance mechanisms that probably exist in PDAC to circumvent the inhibition of EGFR and imply that an alternative treatment strategy, i.e. the combination of EGFR inhibitors with other pharmaceuticals, may be more effective. For example, the combined inhibition of EGFR and C-RAF led to complete tumour regression in murine PDAC models and human patient-derived xenografts [50]. A phase II trial (NCT01222689) revealed modest antitumour activity following the application of erlotinib plus selumetinib to patients with locally advanced or metastatic PDAC (the median OS was 7.3 months, 95% CI 5.2–8.0 months) [51]. IGF-1R exhibits crosstalk with EGFR and mediates tumour resistance to EGFR inhibitors, and a phase II clinical trial (NCT00769483) showed that MK-0646, an IGF-1R antagonist, synergistically improved OS when applied with gemcitabine (10.4 months vs 5.7 months, P value = 0.02) [52]. In addition, nanoparticles (C18-EEG-GE11) have been developed to target EGFR and precisely deliver drugs to PDAC cells [53].

Proteins downstream of KRAS, such as the RAF/MEK/ERK pathway or the PI3K/PDK1/AKT/mTOR pathway, have also attracted increasing interest [54, 55]. MEK is required for the viability and proliferation of tumours [23]; thus, diverse MEK inhibitors have been developed.

No significant difference was observed in the clinical trials performed to verify the efficacy of MEK inhibitors applied as a monotherapy, i.e. selumetinib and trametinib, in patients with advanced PDAC (selumetinib HR = 1.03, 80% CI 0.68–1.57, P value = 0.92; trametinib HR = 0.98, 95% CI 0.67–1.44, P value = 0.453) [56, 57]. The failures of trametinib and selumetinib appear to be due to the activation of receptor tyrosine kinases (RTKs) [58]. Accordingly, multidrug combinations of MEK inhibitors are being tested in clinical trials. High-throughput screening revealed the highest relative efficacy of AZD6244 (selumetinib) in PDAC cell lines. When applied together with AZD6244, BKM120, a PI3K inhibitor, leads to robust apoptosis in PDAC-derived organotypic models or murine models, resulting in a longer median survival (131.5 vs 71 days) [59] and indicating that the combined inhibition of MEK and PI3K may have clinical value. AKT inhibitors also produce potent synergistic effects with MEK inhibitors on PDAC [54]. Ulixertinib, an ERK inhibitor, exerts an inhibitory effect on solid tumour xenograft models [60] and appears to prevent tumour growth to a greater extent when combined with MEK inhibitors [61]. In summary, interventions that simultaneous target the two major downstream pathways of KRAS, i.e. RAF/MEK/ERK and PI3K/PDK1/AKT, represent a direction for future exploration in KRAS-mutant PDAC treatment, and clinical trials have been performed to verify the effectiveness of this strategy [62].

In addition to the simultaneous inhibition of multiple pathways, many other adjuncts to MEK inhibitors with various mechanisms have been developed. ABT-263 relieves the inhibition of BCL-XL to BIM; hence, the MEK inhibitor-induced expression of the pro-apoptotic protein BIM increases cell apoptosis and reduces the tumour volume in KRAS mutant cancer models [63]. Multiple members of the RTK/RAS/MAPK pathway have a synthetic lethal interaction with MEK, as they induce tumour resistance to MEK inhibitors by triggering an adaptive reactivation of the MAPK pathway. Therefore, the simultaneous blockade of MEK and its synthetic lethal interactors may be another strategy for KRAS mutant PDAC [58, 64‐66]. SHP2 inhibition (by SHP099) and SHOC2 suppression (by gene knockout) were performed to confirm the effectiveness of this strategy in murine models. The combined application of trametinib and SHP099 or trametinib and SHOC2 knockout resulted in tumour stasis [67, 68]. In addition to the direct cytostatic effect on the tumour, MEK inhibitors also exert an inhibitory effect on several immunosuppressive immune cells, indicating potential synergy with immunotherapy. The application of GDC-0623 (cobimetinib), a MEK inhibitor, with an anti-CD40 antibody in murine models produced striking synergistic effects [69]. A strategy targeting both MEK and CDK4/6 not only delays tumour progression but also increases T-cell infiltration and tumour sensitivity to immune checkpoint inhibitors in xenograft models [70]. Interestingly, in breast cancer, the combined application of trametinib and rosiglitazone transforms cancer cells into adipocytes. This combination exploits the plasticity of cancer cells and destroys the resistance of cancer cells to conventional chemotherapy [71]. Further clinical trials assessing the efficacy of these combination therapies in PDAC will be worthwhile.

Rigosertib, an inhibitor of PI3K and PLK1, failed to improve the prognosis of patients with metastatic PDAC (OS HR = 1.24, 95% CI 0.85–1.81) [72]. In addition, paradoxically, activated AKT was observed after the inhibition of PI3K. Everolimus, an mTOR inhibitor [73], failed equally against metastatic PDAC (the median progression-free survival (PFS) was 1.8 months and the median OS was 4.5 months) [74]. Recent studies also aimed to combine PI3K inhibitors with other targeted treatments, such as MK-2206 plus selumetinib (the OS was shorter in the experimental arm, HR = 1.37, P value = 0.15) [75], and GDC-0941 plus ulixertinib (synergistic inhibitory activity in PDAC cell lines) [76].

Most patients with PDAC harbour KRAS mutations, as described above. In the small group of patients with KRAS wild-type PDAC, other mutations, such as NTRK and NRG1, initiate PDAC tumorigenesis and represent actionable targets.

Gene fusion is rare but oncogenic in KRAS wild-type cell lines [77]. The frequency of NTRK fusion and NRG1 fusion is 0.3% and 0.5%, respectively [78]. Chromosomal rearrangement of the NTRK gene family promotes the expression of tropomyosin receptor kinases with chimeric rearrangements, which are characterized by ligand-independent constitutive activation [77]. These chimeric proteins signal via the same MAPK and PI3K-AKT pathway as normal TRK proteins, and they participate in possible crosstalk with tyrosine kinases [79].

In solid tumours with NTRK gene fusions, TRK inhibitors such as larotrectinib showed significant and lasting antitumour activity, regardless of the tumour types (the overall response rate was 75%, 95% CI 61–85%) [80]. Hyperactivated chimeric TRK proteins also represent potential targets in NTRK fusion-positive PDAC. A pooled analysis of clinical trials (NCT02122913, NCT02637687, NCT02576431, NCT02097810, NCT02568267, EudraCT, and 2012-000148-88) revealed that the selective TRK inhibitors larotrectinib and entrectinib are effective against solid tumours that harbour NTRK gene fusions, including PDAC (the larotrectinib response rate was 79%, 95% CI 72–85%; and the entrectinib response rate was 57%, 95% CI 43.2–70.8%), and larotrectinib and entrectinib have received the FDA breakthrough designation of targeting NTRK fusion-positive solid tumours [81, 82]. Next-generation TRK inhibitors, such as selitrectinib and repotrectinib, are being developed to address on-target resistance [83].

NRG1 is a direct ligand of ERBB3 and ERBB4 receptors; accordingly, various NRG1 fusions initiate PDAC via the overactivation of ERBB receptor signalling pathway [84].

The ectopic ERBB signalling pathway, including constitutive activation of MEK, ERK, and PI3K, represents a potentially promising target in NRG1 fusion-initiated KRAS wild-type PDAC [85]. The anti-ERBB3 antibody GSK2849330 and pan-ERBB inhibitors afatinib and neratinib impaired cell proliferation in multiple cancer cell lines with NRG1 rearrangements. An anti-ERBB3 antibody led to tumour regression in an ovarian cancer-derived xenograft model, suggesting that the selective inhibition of ERBB3 may exert more potent antitumour effects than pan-ERBB inhibitors [86‐88]. MCLA-128 (zenocutuzumab) docks on ERBB2 and blocks the binding of an NRG1 fusion protein to ERBB3. The effectiveness of MCLA-128 has been confirmed in patients with PDAC harbouring an NRG1 fusion [89]. Moreover, a phase II clinical trial of MCLA-128 in patients with solid tumours expressing an NRG1 fusion has been launched (NCT02912949).

In contrast to the direct stimulation of oncogenes, tumour suppressors were originally designed to restrain tumorigenesis. Notably, p53 is a transcription factor that regulates the expression of several genes, and its biological functions include the inhibition of cell proliferation by inducing p21 expression, promoting the apoptosis of tumour cells by stimulating Bax expression, maintaining genetic stability, and inhibiting tumour vascularity [90, 91]. TP53 is the most commonly inactivated tumour suppressor in PDAC. Approximately 70% of patients with PDAC harbour alterations in the TP53 gene [23, 26].

TP53 reactivators include cys-targeting agents such as CP-31398 and APR-246, Zn²⁺ chelators such as COTI-2, and other proteins that potentially stabilize p53, help p53 refold, or inhibit the aggregation of aberrant p53 [92]. APR-246 (PRIMA-1MET) performed well in blocking the growth of haematological malignancies, prostate cancers and oesophageal adenocarcinomas [93, 94]. COTI-2 also exhibited potency in TP53-mutant squamous cell carcinoma [95]. Further studies are needed to verify whether these reactivators improve the prognosis of patients with TP53 mutant PDAC, and a clinical trial of COTI-2 is ongoing (NCT02433626). In addition to reactivation, the inhibition of murine double minute 2 (MDM2) is another emerging tactic for targeting TP53-mutant tumours. The p62-NRF2-MDM2 axis is involved in tumour progression and programming [96], and MDM2 antagonizes p53 through direct interaction or ubiquitin-dependent degradation [97]; therefore, the inhibition of MDM2 may increase the activity of p53 and restrain p53 mutant cancers [98]. Recent studies have confirmed the efficacy of MDM2 inhibitors, such as Nutlin, MA242, SP141 and MI-319, in vitro and in vivo [99‐102]. However, clinical trials of MDM2 inhibitors in patients with PDAC are currently lacking.

CDKN2A is a multifunctional gene that produces p16 and p19 to arrest the cell cycle at the G1/S checkpoint through a CKD4/6-regulated mechanism [103], and the proteins bind to MDM2 to block the reduction in p53 levels [16]. Approximately 60% of patients with PDAC harbour CDKN2A mutations [23, 26], with an odds ratio of 12.33, indicating that germline mutations in CDKN2A are associated with a high risk of developing PDAC [104].

CDK4/6 is a potential target in CDKN2A-deficient tumours [105], [106]. Ribociclib and palbociclib have already shown efficacy and safety in metastatic breast cancer and liposarcoma [107, 108]. The efficacy of CDK4 inhibitors has also been confirmed in PDAC preclinical models [10‐111], and related clinical trials (NCT02501902) are underway. Researchers have postulated that CDK4/6 inhibitors, which exert a limited antitumour effect as a monotherapy, show greater promise when combined with other targeted agents [112]. For instance, CDK4/6 inhibitors block the DNA repair machinery, increasing the sensitivity of PDAC cells to PARP inhibitors [113]. In addition, the combined inhibition of CDK4/6 and MEK modulates the PDAC microenvironment, increasing the sensitivity of PDAC cells to immune checkpoint blockade [70]. The application of abemaciclib and YAP1 or HuR inhibitors also exerts a synergistic inhibitory effect on PDAC cell lines [114].

Approximately 40% of patients with PDAC harbour SMAD4 mutations [16, 23, 26]. SMAD4 mediates the pleiotropic signalling network downstream of the transforming growth factor-β (TGF-β) pathway and exerts paradoxical effects on tumorigenesis. SMAD4 prevents the tumour-promoting activity of proinflammatory cytokines and induces cell cycle arrest and apoptosis in precancerous cells. In PDAC, however, SMAD4 mutations interfere with the trimeric assembly of its C-terminal domain, which is important for its transduction activity [115], therefore preventing the normal transduction of TGF-β signals. Thus, its role switches from a suppressor to a promoter in precancerous cells [116]; moreover, TGF-β activity in mast cells induces cancer resistance to gemcitabine [117], and TGF-β suppresses the activity of normal immune cells, helping cancer cells escape from the immune system [118].

The TGF-β SMAD4 signalling pathway mediates the tumour-stroma interaction. PDAC has two distinct epithelial-mesenchymal transformation (EMT) subtypes, the complete EMT and partial EMT, and the latter is speculated to result in an increased metastasis rate via the formation of clusters of circulating tumour cells [119]. Cancer-associated fibroblasts secreting TGF-β may induce the partial EMT and switch PDAC proliferation phenotypes, contributing to PDAC heterogeneity [120]. PDAC with an impaired TGF-β-SMAD4 signalling pathway per se may modulate the fibrotic response and mechanophenotype [121], indicating that molecular alterations in tumours not only control PDAC progression but also reprogram the metabolic phenotypes of cells in the TME. Heterozygous mutation of SMAD4 attenuates the metastatic potential of PDAC cells while increasing their proliferation. Reportedly, SMAD4 is also correlated with glucose transporter expression and matricellular fibrosis. Clinical studies have confirmed that SMAD4 inactivation is associated with a poor prognosis [122, 123].

Because of the dual roles of SMAD4 in cancer cells, agents have been designed to inhibit rather than activate TGF-β in SMAD4-deficient tumours [124, 125]. Galunisertib, a TGF-β inhibitor, showed efficacy in a preclinical investigation [126]. Phase I/II trials showed that the combined application of galunisertib and gemcitabine prolonged OS (estimated HR = 0.796) [127, 128].

The RUNX3 expression level is strongly correlated with the SMAD4 status. Accordingly, RUNX3 also functions as both a tumour suppressor and promoter in PDAC and regulates the balance between cancer cell proliferation and dissemination. RUNX3 combined with DPC4 helps distinguish PDAC subtypes and enables more precise clinical decisions [129]. In SMAD4-negative PDAC, PGK1 is selected as the decisive gene to determine the PDAC metabolic phenotype and balance metastasis and proliferation. Nuclear PGK1 determines the metastatic potential of PDAC cells, thus helping to predict metastatic patterns of PDAC cells and providing guidance for precise therapy [130].

CAR-T is a hotspot of immunotherapy. The autologous T cells of patients are isolated and reprogrammed to precisely target tumour-associated antigens [163]. CAR-T has already proven to be effective against haematological neoplasms [164], and the FDA has approved Kymriah and Yescarta, two CAR-T drugs targeting CD19-expressing cancer cells, for clinical application [165, 166]. In addition to CD19, other characteristic surface biomarkers of solid tumours also have the potential to be designed as CAR-T therapeutic targets, as shown in Table 2. For example, the diverse tumour-specific glycosylated antigens provide a roadmap for CAR-T targets [167, 168]. CAR-T targeting the abnormal O-glycosylation site, i.e. the Tn and STn antigens on MUC1, has already been shown to inhibit the growth of PDAC cell lines [169] and control PDAC xenograft growth in murine models [170]. The combination of CEA-CAR-T with rhIL-12 exerted significant antitumour effects in vitro and in vivo [171], and a phase II/III trial (NCT04037241) to evaluate the efficacy of CEA-CAR-T is recruiting patients. CD133 is a marker of cancer stem cells and is related to tumour metastasis and recurrence; a phase I trial (NCT02541370) confirmed the safety of CAR-T-133 in patients with advanced metastatic malignancies [172]. Mesothelin (MSLN) is implicated in tumour invasion and is widely overexpressed in solid tumours, including PDAC [173]. The targeting of mesothelin by CAR-T controls the metabolic active volume in murine models [174], and a phase I trial (NCT02159716) suggested that MSLN CAR-T is safe in patients with solid tumours, including PDAC [175]. Moreover, dual-receptor CAR-modified T cells that simultaneously recognize CEA and MSLN were designed to attenuate the “on-target, off-tumour” toxicity [176]. The appealing KRAS protein is also involved in the exploration of CAR-T; experiments using CAR-T targeting mutant KRAS G12D suggested that the loss of heterozygosity at the HLA may reduce the efficacy of immunotherapy, and a phase II trial (NCT01174121) of this CAR-T is ongoing [177]. HER2/ERBB2 is a transmembrane protein that induces tumour initiation and progression; therefore, HER2 potentially represents an ideal target, and the safety of CAR-T-HER2 has been confirmed in a phase I trial (NCT01935843) [178]. In addition, a study used switchable CAR-T targeting HER2 to increase its efficacy and reduce its toxicity [179]. Programmed cell death protein-1 (PD-1) is a famous immune checkpoint receptor that is involved in tumour immune evasion. In addition to small molecule inhibitors, chPD1 T cells have been designed to target PD1 precisely, and a preclinical study observed protective antitumour responses of chPD1 T cells in multiple models of solid tumours [180]. B7-H3 overexpressed on the PDAC cell surface is another attractive target, xenograft PDAC models certified the effectiveness of CAR-T targeting B7-H3, and 4-1BB co-stimulation enhanced this antitumour activity [181].

In addition to CAR-T, antibody-drug conjugates (ADC) and bispecific T-cell engagers (BiTE) are also designed to confer selective toxicity to PDAC cells. ADC combine antibodies against tumour-specific antigens with cytotoxic agents; hence, cell toxins are able to precisely target cancer cells.

The most common cell toxins are microtubule-disrupting agents. For example, DMUC5754A conjugates an anti-MUC16 antibody to monomethyl auristatin E (MMAE); however, it was ineffective at treating patients with PDAC in phase I trial [182]. MLN0624 conjugates anti-guanylyl cyclase C to MMAE, and it is reported to have a limited benefit for patients with PDAC [183]. A glypican-1 antibody has been conjugated to monomethyl auristatin F (MMAF) and significantly inhibits the growth of xenografts derived from patients with PDAC [184]. Anetumab ravtansine conjugates an anti-mesothelin antibody to the tubulin inhibitor DM4, and it exhibited great tolerance in a phase I trial and warrants future investigation [185].

In addition to cytoskeleton-disrupting agents, other drugs have also been conjugated to antibodies, such as DS-8201a, which conjugates a topoisomerase I inhibitor with HER-2 antibodies. A phase I trial supported the use of DS-8201a as a potentially promising treatment [186].

BiTEs simultaneously bind tumour-associated antigens and the CD3 epitope on the T cell surface, forming an immune synapse and resulting in the targeted lysis of tumour cells [187]. For example, MT110 (solitomab) links EpCAM with CD3 and redirects T cells to selectively kill PDAC cells [188]. However, a phase I trial revealed adverse events of solitomab and prevented dose escalation to therapeutic levels [189].

Immune checkpoint inhibitors, such as ipilimumab and nivolumab, also show potential in antagonising tumours [190]. An increasing number of trials have been designed to combine PD-1 or programmed cell death 1 ligand 1 (PD-L1) inhibitors with other treatments [191]. However, only a subgroup of tumours are sensitive to immune checkpoint blockade; thus, indicators are required to guide the treatment more efficiently [192]. The tumour mutational burden exhibits a strong linear correlation with the objective response rate to PD-1 inhibition. PDAC with a low number of genomic mutations is more resistant to PD-1 inhibitors than PDAC with a high number of genomic mutations [193]. A high degree of microsatellite instability (MSI-H) results in a high tumour mutational burden [194]. Therefore, mismatch repair deficiency (dMMR) and subsequent MSI-H are good predictors of the efficacy of PD-1 or PD-L1 inhibitors [195]. The latest phase II KEYNOTE-158 trial revealed a benefit of PD-L1 inhibitors in combination with pembrolizumab in patients with MSI-H/dMMR cancers (the objective response rate in the pancreatic cancer subgroup was 18.2%, 95% CI 5.2–40.3%) [196]. Approximately 1% of patients with PDAC exhibit dMMR/MSI-H; therefore, the clinical value of applying PD-1 or PD-L1 antibodies in PDAC is limited.