Pancreatic Adenocarcinoma: Unconventional Approaches for an Unconventional Disease

Pancreatic cancer is the most lethal common tumor in America. The five-year survival is estimated to be 9.3% among all cases and 2.9% among patients with metastatic disease, both lowest among all common tumors (1). With its rising incidence and unabated mortality, pancreatic cancer is the third leading cause of cancer-related death, with a projection that it will the second by 2030 (Fig. 1; refs. 2, 3). In 2019, pancreatic ductal adenocarcinoma (PDAC; 95% of patients) represented 3.2% of all new cancer cases, with an estimated 56,770 new cases and 45,750 deaths (7.5% of all cancer-related deaths; ref. 1).

PDAC is a complex and clinically challenging disease, defined by multiple genetic and environmental factors. The majority of PDAC arises de novo, with hereditary or genetic factors accounting for only 5–10% of cases (4). Risk factors associated with PDAC development include: smoking (RR: 2–3), nonhereditary or chronic pancreatitis (RR: 2–6), chronic diabetes mellitus (RR: 2), obesity and/or sedentary lifestyle (RR: 2), nontype O blood group (RR: 1–2), and age (≥97% of cases occur over the age of 45; refs. 4, 5). A few genetic syndromes and mutations correlate with higher PDAC lifetime risk. Individuals with hereditary pancreatitis, associated with Trypsin-1 (PRSS1) or serine protease inhibitor Kazal-type 1 (SPINK1) mutations, poses a lifetime PDAC risk of 50%, while patients with Peutz–Jegher syndrome and familial atypical multiple mole and melanoma syndrome carry lifetime risks of 30%–40% and 10%–20%, respectively (6). Other syndromes, such as Lynch syndrome (associated with MLH1, MSH2, and MSH6 mutations), hereditary breast and ovarian cancer syndromes (caused by BRCA1/2 or PALB2 mutations), Ataxia-telangiectasia (caused by mutations in the ATM), and Li-Fraumeni Syndrome (caused by germline p53 mutations), contribute to a lesser degree (6, 7). Inherited germline mutations in CDKN2A, MLH1, BRCA1, BRCA2, TP53, and ATM are associated with familial PDAC history, and screening for these mutations is recommended by National Comprehensive Cancer Network guidelines (8, 9).

Treatment improvements in many common tumors, for example, breast and prostate, are, in part, a consequence of advances in disease diagnosis. Unfortunately, there are no reliable or readily available screening tests for PDAC, and the majority of patients with PDAC do not exhibit symptoms until advanced stage. The majority of PDAC tumors (60%–70%) originate at the head of the pancreas, and this location dictates subsequent symptomatology (10, 11). Head tumors typically present with pain, jaundice, pruritus, pale stools, dark urine, and gastric outlet obstruction. Body and tail tumors are largely asymptomatic and present late with distant metastases or local disease with multivisceral and vascular invasion. Both locations are associated with anorexia, weight loss, and generalized abdominal pain (12). A comparative case–control analysis of patients with PDAC (n = 120) to control patients (n = 180) revealed that bile obstruction (OR: 20), pale stool (OR: 31), anorexia (OR: 41), abdominal pain (OR: 30), and unusual bloating/belching (OR: 20 and 17) are the most common general pancreatic cancer symptoms (13). Nonspecific early symptoms hamper early clinical diagnosis of PDAC, supporting research into noninvasive, cost-effective screening methods.

Development of accurate diagnostic tests is limited by the dearth of effective biomarkers. Currently, carbohydrate antigen 19-9 (CA19-9), a sialyated Lewis blood group antigen, and carcinoembryonic antigen (CEA), are used as circulating biomarkers of pancreatic cancer. CA19-9 is not sensitive nor specific, and is elevated in other pancreatic diseases, such as pancreatitis, pancreatic pseudocyst, choledocholithiasis, and cirrhosis (14). Currently, CA19-9 is used to monitor the course of patient disease, including postsurgical recurrence (15). CEA is also neither sensitive nor specific for early PDAC, and is elevated in alcoholic cirrhosis, hepatitis, and biliary disease, and, thus, its utility in screening is limited (13, 16). Efforts to identify clinically relevant biomarkers are ongoing, and there are recent exciting developments in both diagnostic and predictive biomarkers. Given the wealth of new potential biomarkers and potential screening assays, the reader is referred to a recent thorough review by Hasan and colleagues (15).

The poor 5-year survival in patients with PDAC reflects the late diagnosis, limited treatment options, and molecular and biophysical properties of PDAC that contribute to resistance. Surgical resection remains the only current curative intent therapy for pancreatic cancer. However, surgical therapy is limited to 15%–20% of all patients with PDAC, and often nodal metastases and microscopically positive margins are noted following resection (1, 15). Resection typically requires a pancreaticoduodenectomy (Whipple procedure) to remove tumors localized to the pancreatic head and uncinate process (15, 16). Multiple variations of the Whipple procedure (including extended lymphadenectomy, pylorus-sparing, and a variety of anastomotic techniques, as well as minimally invasive) are not associated with improved survival outcomes (5).

Minimally invasive surgical techniques are gaining popularity in oncology as an alternative to open procedures, with the potential to reduce infection, blood loss, postoperative pain, and surgical site infection. Despite the technical challenges of a Whipple procedure, the volume of laparoscopic and robotic assisted procedures has been increasing (17). While meta-analysis comparing both laparoscopic and robotic approaches to open Whipple procedures reported lower blood loss, shorter hospital stay, and decreased morbidity without any difference in major complications, mortality, or reoperation rates, the LEOPARD-2 trial reported a higher mortality in laparoscopic procedures compared with open procedures (18, 19). Robotic surgeries are also safer than laparoscopic procedures, reduce the length of surgery, decrease postoperative morbidity, and decrease the duration of hospital stay, without evidence of increased mortality (17, 20–23). Furthermore, robotic Whipples afford a lower complication rate, margin positivity rate, and wound infection rate compared with open procedures (17).

To increase the rate of R0 (tumor negative margins) versus R1 (residual tumor margins) resections, neoadjuvant chemotherapy protocols are being introduced in surgery (24). Murphy and colleagues report the potential benefit of neoadjuvant fluorouracil, leucovorin, oxaliplatin, and irinotecan (FOLFIRINOX) and losartan followed by surgical resection in a phase II clinical trial. Of the 49 enrolled patients, 42 underwent attempted surgery and an R0 (margin-negative) resection was achieved in 69% (CI, 55%-82%; ref. 25). With neoadjuvant therapy, overall median progression-free survival increases to 17.5 months and overall survival to 31.4 months, although control outcomes were not given (24). Similarly, FOLFIRINOX with gemcitabine transitions 52% patients with borderline resectable disease to R0 outcomes (25). Retrospective data analysis of 30 patients with locally advanced or borderline resectable PDAC, who received neoadjuvant gemcitabine plus nab-paclitaxel, with and without subsequent S-1 therapy reveals that eight were amenable for surgery, and six achieved R0 resection (26). Collectively, this work demonstrates the potential to improve survival outcomes in traditionally nonoperable patients and to expand current curative intent therapy.

Palliative treatment of metastatic PDAC relies on systemic, nonspecific chemotherapeutics (Fig. 1). Current treatment options and several recent trials for metastatic PDAC are summarized in Table 1 (27–47). First-line chemotherapy in metastatic PDAC typically entails combination therapy with gemcitabine and a nanoparticle formulation of paclitaxel (Abraxane), or the combination therapy FOLFIRINOX. Monotherapy with gemcitabine is considered for patients not amenable to more aggressive combination therapy, showing only modest improvement in overall survival (30). First-line therapies, gemcitabine with or without Abraxane and FOLFIRINOX, show marginal improvement in overall survival for patients with metastatic cancer, with overall survival falling short of one year in most trials. Lack of target specificity, dose-limiting toxicity, and poor drug penetrance underlie major limitations in PDAC therapy efficacy in metastatic disease, and highlight the discrepancy in care between PDAC and other solid tumors.

Currently, there is a dearth of targeted therapies in PDAC. Most targeted agents, such as antiangiogenic agents (bevacizumab), tyrosine kinase inhibitors (cetuximab), checkpoint inhibitors (durvalumab), and recently hyaluronic acid therapies (PEGPH20), have been inefficacious in metastatic pancreatic cancer. (Table 1). Several targeted therapies afforded only marginal improvements (≤2 months) in overall and progression-free survival, while others yielded worse outcomes. For example, treatment combining MEK1/2 and AKT inhibitors selumetinib and MK-2206 showed worse overall survival than mFOLFIRINOX alone (46). Similarly, repurposing successful clinical agents (such as anti-EGFR and HER2 agents) have also failed (32–38). Certain targeted therapies do show efficacy in select population groups. Olaparib, a PARP inhibitor, improved response rates in patients with DNA damage repair deficiency (10–20% of patients with PDAC). PFS was 24.7 weeks (3.9–41.1 weeks) with two partial responses and six stable disease responses (48). Similarly, the neurotropic receptor tyrosine kinase inhibitor larotrectinib, received FDA approval for NTRK fusion–mutant tumors (<1% of PDAC), based on a successful phase III clinical trial (49). While larotrectinib may be an efficacious therapy in this subgroup, early data suggests that rapid resistance may be a problem (50).

Several failed therapies have targeted pathways regulated by the RAS family, known to be mutated in PDAC tumors. A summary of RAS mediated signaling is detailed in Supplementary Fig. S1. The RAS family of gene (HRAS, KRAS, and NRAS) encode 21 kDA proteins with GTPase activity, responsible for regulating mitogenic cell signaling pathways (51). In PDAC, KRAS mutations are observed in 95% of tumors, with codons glycine-12, glycin-13, or glutamine-61 residues frequently altered in tumor cells (51). Therapies targeting RAS-regulated pathways have consistently failed in PDAC, due to the high frequency of activating mutations. Initially thought to be undruggable, there is ongoing exciting research in RAS-specific targets, which recognize transient, druggable binding pockets. The first, S-IIP, is the target of RAS inhibitors in the G12C mutant (only 1%–4% of PDAC tumors) with a covalent KRAS(G12C) inhibitor AMG 510, under clinical investigation (52). Additional studies have discovered four total pockets: S-IIP pocket, Kobe site, cyclen site, and P4 site, all of which are of clinical interest (52–57).

Immunotherapy is affording a paradigm shift in human cancer treatment. Immune checkpoint inhibitor (ICI) mAbs that target CTL-associated protein 4 (CTLA-4) or programmed death ligand 1 (PD-L1)/programmed death 1 (PD-1) have generated durable efficacy in many types of human cancers. In 2017, the FDA granted accelerated approval to the anti-PD-1 mAb Keytruda for patients with solid tumors, including pancreatic tumors, which have either mismatch repair deficiency (MMR-D) or high microsatellite instability (MSI). In the first phase I trial evaluating the anti-PD-L1 antibody MS-936559, objective responses were achieved in melanoma, non–small cell lung cancer, renal cell cancer, and ovarian cancer, but not in the 14 patients with pancreatic cancer (58). Similar disappointing results were seen in a 60-patient phase IIA trial with anti-PD-L1 antibody durvalumab (59). However, a robust 62% objective response rate to pembrolizumab was observed in pancreatic cancer patients with mismatch repair deficient (MMR-D; n = 8) (60). Therefore, limited efficacy of anti-PD-L1 therapy is confounded by the low, only 0.8%, prevalence of MMR-D tumors (8). Similar shortcomings were seen in response to anti-CTLA-4 mAb immunotherapy. In phase II and IIa trials, no objective response was achieved with Ipilimumab (61). Combinational immunotherapy of anti-PD-L1 and anti-CTLA-4 mAbs showed increased efficacy, albeit at a low rate with an objective response rate of 3.1% (61).

The underlying mechanism of human pancreatic cancer unresponsiveness to ICI immunotherapy is still elusive. It is thought that pancreatic cancer is a nonimmunogenic “cold” cancer type (62). Human pancreatic carcinoma, except for the MMR-D subtype, exhibits a median mutational load of 4, much lower than more immunogenic melanoma and lung cancer (62, 63). It might be expected that a lower mutational load would decrease neoantigen presentation on tumors, thereby decreasing tumor infiltration by CTLs, and limiting anti-PD-L1 therapy (64, 65). However, CTLs do infiltrate human pancreatic carcinoma, suggesting that CTL tumor infiltration level is unlikely a major factor that underlies human pancreatic cancer nonresponse to ICI immunotherapy (66–69). Furthermore, pancreatic tumor cells highly express PD-L1 and PD-L1 appears to suppress CTL tumor infiltration in the tumor microenvironment (70–73).

Yet, despite abundant PD-L1 expression and infiltrating CTLs, pancreatic cancer does not respond to anti-PD-1/PD-L1 ICI immunotherapy (60, 8). Therefore, it is reasonable to suspect that a PD-L1–independent mechanism may confer immune evasion to pancreatic tumors. This interpretation is supported by data demonstrating that targeting tumor-promoting mediators in the tumor microenvironment sensitizes pancreatic cancer to ICI immunotherapy in preclinical mouse tumor models (7, 73).

Poor survival outcomes in PDAC necessitate novel, innovated approaches towards disease management. Prior therapies failed to capture the biologically relevant pathways driving PDAC tumorigenesis. Future research must focus on addressing the biological and biophysical properties unique to PDAC. As PDAC treatment employs both surgery and chemotherapy, we will discuss new advancements in these fields including: (i) enhanced visualization of tumor margins during surgery to increase the incidence of complete resection; (ii) better targets for biologic therapy and small-molecule inhibitors; and (iii) improvements in drug delivery to overcome biophysical limitations of PDAC treatment such as delivery within the local fibrotic tumor and peritumoral microenvironment as well as improved systemic penetrance.

To achieve complete R0 resection, surgeons must be able to adequately visualize the entire tumor. Conventional brightfield surgery is inherently limited by the human eye, and surgeons rely on histology to evaluate their resection. To augment tumor visualization, several nonspecific near-infrared (NIR) fluorescent probes are used (74–78). These dyes spread lymphatically, and aid in visualizing bulk tumor margins and sentinel lymph nodes. Such surgical visual aids (SVA) are employed in oncologic surgery to allow NIR tumor visualization, and clinical trials are ongoing in many cancer subtypes, including: colon cancer (LUM015, RAPIDO-TRACT, and Fluorescent lectin), lung cancer (EC17 and ICG), and breast cancer (BLZ-100, LYM015, AVP-620, EC17), as well as in PDAC (discussed below; refs. 76–78). While advantageous over bright-field surgery, these dyes are inherently limited by nonspecific diffusion, low specificity, and poor tumor retention. Newer methods of fluorescent agent delivery, enhanced tumor specificity, and improved resolution are under development to improve primary tumor detection and identify early metastatic niches.

Polymeric nanoparticle SVAs offer an effective alternative to current, nonspecific SVAs. Relying on bulk diffusion, nanoparticles accumulate within metabolically active cells and are retained longer than lower molecular weight dyes. Responsive polymeric nanoparticles covalently linked with rhodamine dyes (HRF-eNP), designed to expand and retain dye once internalized within cells and the tumor microenvironment, readily localize to PDAC tumors in a Panc1 rat xenograft model. HRF-eNP enable improved visualization and delineation of tumor margins compared with bright-field imaging (Supplementary Fig. SI2A and S12B; ref. 79). Targeting and retention of the HFR-eNPs in peritoneal tumor sites is highly effective, with specificity, sensitivity, and accuracy of 0.99, 0.92, and 0.95, respectively, in the 202 normal samples and 253 histologically confirmed samples surveyed. Use of HRF-eNPs allows submillimeter tumor detection, a currently unmet need in image resolution. A hyaluronic acid derived nanoparticle containing indocyanine green, NanoICG, provides better tumor visualization (2.30 ± 0.67 AU in tumor vs. 0.41 ± 0.10 AU in normal pancreas) compared with indocyanine green alone (0.77 ± 0.12 AU in tumor vs. 0.35 ± 0.12 AU in normal pancreas) in a KPC orthotopic xenograft PDAC murine model (80). While NanoICGs are effective for delineating tumor from adjacent normal tissue, significant accumulation occurs in the stomach, liver, small intestine, and kidney. Finally, tripolymer fluorescent nanospheres (TFN) possessing a surface coating of peanut agglutinin proteins exhibit high affinity for the tumor-associated antigen Thomsen Friedenreich, and shows high affinity for PDAC, when comparing histologically fixed samples of PDAC and normal tissue (81). TNFs nanoparticles localize primarily to PDAC tumors in an in vivo model of PDAC, using subcutaneously implanted CRL2460 cells. The TNFs nanoparticles enable complete resection of PDAC tumors with minimal removal of normal tissue. Furthermore, signal levels in noncancerous tissue were well below detection threshold, reducing false positives. Relying solely on bulk diffusion, these nanoparticle carriers show promise as a cost-effective, reproducible method for image-guided surgery.

In contrast to nonspecific nanoparticle delivery, targeted fluorophore-conjugated antibody SVAs are being evaluated. Clinical results obtained using anti-CEA (NCT02973672 and NCT0278028), anti CA19-9 (NCT02587230), anti-EGFR (NCT02736578 and NCT03384238), and anti-VEGF (NCT02743975) based SVAs are promising with improved R0 resection rates and better surgical staging of advanced cases (82). Newer antibody conjugate systems, therefore, are of keen interest, including dual-antibody systems to increase the signal, improve specificity, and overcome the heterogeneity present in cell surface expression. For example, Zettlitz and colleagues describe a dual-labeled targeting anti-prostate stem cell antigen (PSCA) conjugated with a near IR dye and a ¹²⁴I for functional PET imaging. PDAC tumors highly express the PSCA surface antigen with high sensitivity and specificity (83). Use of this dual-labeled antibody allows sequential NIRF and PET visualization of subcutaneous PDAC tumors in a murine patient-derived xenograft (PDX) model (84). These bifunctional anti-PSCA antibodies retain high binding specificity, with minimal noncancerous uptake. A dual antibody system that contains both a fluorescent dye, Alexa Fluor 700, and a quencher IRDye, QC-1, fluoresces only after proteolytic cleavage (85). This responsive system repurposes two antibodies, cetuximab and trastuzumab. Both proteolytically activated immunoconjugates provide improved signal resolution of pancreatic AsPC-1 orthotopic xenografts, as well as enhance signal detection and resolution compared with nonspecific probes. Furthermore, dual imaging improves signal resolution, maximizing signal in cancer cells that have higher receptor expression of both targets. With the discovery of more specific PDAC targets, newer SVA can be designed for better tumor detection and enhanced visualization.

Finally, there is ongoing activity to develop theranotics, combined visual agents and adjuvant therapy at the time of surgery. For example, gold nanoparticles functionalized with an anti-glypican-1 antibody, which targets a cell surface proteoglycan, and an NIR dye, for tumor visualization, loaded with oridonin inhibit PDAC migration and EMT (86). These nanoparticles are highly specific for PDAC tissue, and provide excellent resolution of tumor margins. Adequate fluorescent signal remains even after 48 hours in BxPC-3 orthotopic xenografts (87). Furthermore, these SVAs reduce tumor volume in the BxPC-3 model, affirming both a therapeutic efficacy as well as a specificity. Another PDAC gold nanoparticle theranostic targets the cell surface proteoglycan GPC1 (88). These gemcitabine-loaded nanoparticles, labeled with an NIR dye (GPC1-GEM-NP), allow simultaneous tumor visualization and drug delivery. The combined theranostic system shows excellent uptake in PDAC BxPC-3 and Panc-1 cell lines (>99% by flow cytometry), but minimal uptake (<1%) in control 293T cells. In an orthotopic BxPC-3 murine model, the GPC1-GEM-NPs exhibit greater tumor retention (high signal at 48 hours) and accumulation than GEM-NPs (lacking the GPC1 antibody) alone (loss of signal by 48 hours), with off-target NP accumulation in the liver and spleen (88). Theranostics are an exciting surgical tool, providing adjuvant therapy while enhancing surgical visualization.

As discussed above, only 10%–15% of patients with PDAC are currently amenable to surgery. Often many patients progress despite surgery, as the result of incomplete resection or undetected metastatic disease at the time of surgery (89). SVAs improve R0 resection rates and identify micrometastatic disease. In addition, the use of theranostic agents provides adjuvant treatment of micrometastases while guiding further management. It is also reassuring that several recent trials have improved R0 resection rates of previously nonsurgical candidates through better neoadjuvant chemotherapy protocols (vide supra). Through better tumor visualization, improved treatment and staging is a possibility.

Identification of novel targets in pancreatic cancer is essential to improving treatment outcomes. Available therapies for advanced and metastatic PDAC are predominately nonspecific chemotherapeutics. To date few targeted therapies exist, as many have failed because they targeted KRAS-dependent pathways. Efficacious treatment of PDAC requires identification of PDAC-specific targets, which are KRAS-independent, and which target multiple tumor hallmarks to prevent disease resistance. Several new targets have been identified that met these criteria, and preclinical advancements in these targets are exciting (Table 1).

The dual-endothelin-1/VEGF-signal peptide receptor (DEspR) is a single transmembrane receptor, identified as an embryonic-lethal null mutation phenotype characterized by dysregulated vasculogenesis and angiogenesis, and impaired cardiac organogenesis (90, 91). DEspR binds two known ligands, ET-1 and the signaling peptide of VEGF. Both of these pathways strongly correlate with PDAC progression, but currently, there are no effective therapies targeting either one. PDAC tumor cells highly express DEspR, with minimal expression on normal pancreatic tissue. This receptor regulates multiple tumor processes: cancer stem cell (CSC) viability and anoikis resistance, angiogenesis, and cancer cell invasiveness. The exact signaling pathway is unclear, but appears to have RAS-independent signaling, through FAK, SRC, and STAT3-mediated pathways (90). Treatment of a subcutaneous xenograft RNU rat model of Panc1 with a mAb to inhibit DEspR reduces tumor volume compared with gemcitabine (Supplementary Fig. SI3A; ref. 91). Inhibition of DEspR captures several key PDAC pathways and may address the role of ET-1 and VEGF in PDAC. Further research is needed to address its safety and effect on normal cells and to interrogate its role in normal cell signaling.

Several cancer subtypes, including squamous cell carcinomas, lung, cervical, and pancreas highly express GPR87, a recently identified G-coupled protein receptor related to the lysophosphatidic acid (LPA) receptor (92–95). GPR87 regulates multiple pathways, which influence tumor proliferation, angiogenesis, anoikis resistance, and drug resistance. GPR87 signaling involves multiple downstream mediators, with both RAS-dependent and RAS-independent signaling (96, 97). In pancreatic cancer, GPR87 mRNA expression is significantly elevated and correlates with worse overall survival (97). PDAC cell lines, AsPC-1, Capan-1, BxPC-3, MIA PaCa2, Capan-2, and PANC1 overexpress GPR87 compared with normal human pancreatic ductal epithelial cells. Inhibition of GPR87 significantly reduces tumor viability in vivo and decreases subcutaneous tumor growth compared with controls (Supplementary Fig. S3B; ref. 97). GPR87 attractive features include: it regulates multiple tumor hallmarks, correlates with worse prognosis, and is highly expressed in most PDAC. However, GPR87 is expressed on many normal tissues, albeit at lower levels. Like DEspR, GPR87 inhibition is not completely characterized, and it is unclear if GPR87 regulates survival in healthy epithelium.

CD147, a member of the immunoglobulin superfamily, regulates several key cancer hallmarks in PDAC and is highly expressed in PDAC (87%), hepatocellular carcinoma (83%), and glioblastoma multiforme (79%) (98, 99). CD147 expression positively correlates with poor survival and greater metastatic spread (99). CD147 influences multiple downstream signaling pathways, which are both RAS-dependent and independent, to include anoikis resistance, epithelial-to-mesenchymal transition, chemotherapeutic resistance, and CSCs pluripotency and proliferation (100–105). A mAb targeting CD147, HAb18IgG, enhances radiation-induced cell death in CFPAC-1 and MIA PaCa-2 cells, and potentiates the effect of gemcitabine therapy (106). Inhibition of CD147 reduces tumorsphere formation, a surrogate of CSC-inhibition, in BxPC-3 and MIA PaCa-2 PDAC cell lines. A combination therapy of a ⁹⁰Y-l radiolabeled anti-CD147 antibody and gemcitabine significantly improves the efficacy of gemcitabine therapy in a murine MIA PaCa-2 xenograft model (107). Although CD147 is an exciting target, data suggest that CD147 is a promiscuous receptor, and downstream regulation depends on cell surface density (100). Therefore, treatment responses to CD147 therapy may be highly variable.

Microfibril associated protein 5 (MFAP5), an extracellular matrix glycoprotein involved in elastic microfibril assembly, regulates both PDAC cells and cancer-associated fibroblasts (CAF). Elevated MFAP5 expression correlates with worse prognosis and more aggressive disease (108, 109). Murine antibodies targeting MFAP5 reduce PDAC cell viability and cell motility (108). In in vivo PDXs, inhibition of MFAP5 sufficiently arrests cancer cell growth and decreases collagen I deposition, and impairs CAF collagen production and decreases intratumoral microvessel permeability (108). Treatment with MFAP5 improves delivery of paclitaxel by decreasing biophysical barriers, such as high intratumoral pressure (108). Given the shortcomings of hyaluronidase therapy, anti-MFAP5 therapy is an attractive alternative, reducing both tumor cell viability and CAF-induced desmoplasia.

Another target of tumor barrier function is Claudin-4 (CLDN4), a main constituent protein in tight junctions. PDAC cells overexpress claudin-4 compared with normal pancreatic cells, with higher expression carrying a worse prognosis (110, 111). Inhibition of CLDN4 reduces cell adhesion of PDAC cells and enhances intracellular accumulation of chemotherapeutics (111, 112). In vivo administration of anti-CLDN4 antibodies promotes intratumoral accumulation of FOLFIRINOX, and increases efficacy in subcutaneous MIA PaCa2 murine models (111). This effect is likely a result of reducing PDAC cell adhesion and interfering with drug efflux pumps. A recent report suggests that anti-CDLN4 antibodies are safe and relatively nontoxic, but clinical data is lacking (112). Both CLDN4 and MFAP5 are promising targets, which could disrupt natural PDAC barriers to chemotherapeutics and overcome biophysical defenses.

An exciting regulator of PDAC stromal cells is BCL2-associated athanogene 3 (BAG-3), a cochaperone of HSP70. BAG-3 regulates multiple signaling pathways in PDAC, facilitating cross-talk between: tumor-associated macrophages and PDAC cells and PDAC cells and pancreatic stellate cells via an IL6-dependent manner (113, 114). BAG3 expression correlates with fibrosis in PDAC tissue, and plays a key role in tumor collagen barrier regulation (115). Administration of humanized anti-BAG-3 antibodies reduces the growth of subcutaneous MIA PaCa-2 tumors in murine xenografts. The anti-BAG3 antibody are highly tumor specific, and do not distribute into normal parenchyma (115). While data are limited, no toxicity was observed with BAG3 administration in healthy animals. Currently, Intrepida Bio is developing an anti-BAG3 antibody for clinical trials in PDAC.

Improvements in PDAC therapy requires highly selective targets, which regulate multiple, essential cancer pathways. The targets presented here are both novel and clinically relevant, having several advantages over previous, failed biologic targets (e.g., EGFR, IGFR, and VEGF). These include targeting multiple cancer hallmarks, inhibition of tumor cell and stromal function, KRAS-independent mechanisms, and modulation of PDAC CSC function.

Small-molecule inhibitors have not shown success in recent trials in PDAC. Significant resistance has been encountered in these therapies due to RAS-dependence of their targets. Future endeavors must focus on pathways that directly regulate PDAC proliferation, metastasis, and survival. Ongoing research evaluating MEK/ERK inhibitors and autophagy, EGFR inhibitors, and reevaluation of more classical pathways, is clinically important, but will not be discussed here (116–119). Instead we discuss several novel PDAC targets with RAS-independent signaling pathways. These inhibitors impact tumor energy regulation and chemoresistance in addition to PDAC cell progression and metastasis. While not exhaustive, these new studies highlight recent preclinical advances demonstrating in vivo efficacy in highly expressed, biologically relevant targets.

The hexosamine biosynthesis pathway (HBP) is a major regulator of cancer survival. In normal cells, the HBP is a shunt of the glycolysis pathway, acting as a sensor of energy availability in normal cells (120). HBP plays a role in cancer cell signaling, via O-GlcNAcylation/mTOR/adenosine monophosphate (AMP)-activated protein kinase (AMPK) pathways, tumor suppressor kinase 1/2 (LATS1/2)-mediated pathways, and ER stress regulation (121). Not surprisingly, HBP regulates cancer cell metabolism and directs cellular proliferation, transcription, EMT, and self-renewal in CSCs (122). The rate limiting enzyme in HBP is glutamine-fructose amidotransferase 1 (GFAT1), which is overexpressed in 35.7% of the 176 PDAC samples in TCGA. Inhibition of GFAT1 in vitro, using a small-molecule glutamine analogue (6-diazo-5-oxo-l-norleucine, DON), decreases expression of several self-renewal genes: SOX2, OCT4, and KLF4 in MIA-PaCa2 and S2VP10 cell lines, and reduces both tumor cell proliferation and collagen matrix deposition in murine orthotopic KPC and CAF xenografts (123). Reduction of PDAC ECM density also enhances tumor sensitivity to anti-PD1 therapy, increasing CD8+ tumor infiltration (124). GFAT1 is a promising clinical target, as its inhibition decreases tumor desmoplasia while enhancing anti-PD1 therapy. The success of anti-GFAT1 therapy in lung and breast cancer will hopefully translate to PDAC (120).

Another metabolic target of interest is glutamate-oxaloacetate transaminase 1 (GOT1), which regulates amino acid metabolism and the malate-aspartate shuttle. GOT1 overexpression is common in PDAC tumors and correlates with worse disease prognosis In KRAS-mutant PDAC, alterations in GOT1 metabolism fuel biosynthesis, proliferation, and redox balance, leading to dependence on GOT1 for tumor maintenance (125, 126). A novel inhibitor of GOT-1, PF-04859989, forms a covalent adduct with the substrate of GOT1, irreversibly inhibiting enzymatic function (127). Loss of GOT1 function limits PDAC metabolism and inhibits PDAC proliferation in GOT-1 dependent cell lines. A natural inhibitor of GOT1, Aspulvinone O, exhibits similar efficacy (128). Inhibition of GOT1 by Aspulvinone O induces apoptosis in SW1990 cell lines in vitro, and modulates cellular metabolism via shifting mitochondrial respiration and increasing ROS production (128). Also, Aspulvinone O inhibits PDAC proliferation in vivo in murine subcutaneous SW1990 xenografts. These results are promising, as GOT1 inhibition could starve KRAS-mutant PDAC tumors and prevent disease progression.

Increased production of phosphocreatine via creatine kinase-B occurs in pancreatic cancer to meet the high energy demand of growing tumors (129). Importing phosphocreatine via the creatine transporter SLC6A8 allows cells in the metastatic niche to maintain ATP under high metabolic demand (130, 131). Depletion of SLC6A8 in colon and pancreatic cancer reduces the ability of cancer cells to colonize the liver in murine xenografts (132, 133). The orally available small-molecule RGX-202 inhibits SLC6a8 in KRAS wild-type and KRAS-mutant cell lines, as well as human PDX models (134). Inhibition of SLC6A8 correlates with decreased tumor metabolic activity, decreased proliferation, and a dramatic reduction in cell survival. Tumors treated with RGX-202 are also more susceptible to gemcitabine, likely reflecting a change in cellular metabolism and nucleotide synthesis (134, 131). Although the effect of SLC6A8 on normal cellular metabolism remains to be full understood, these results are exciting, and suggest that in phosphocreatine-dependent cells, SLC6A8 inhibition could effectively starve PDAC tumors.

The transcriptional coregulator C-terminal Binding Proteins (CtBP) 1 and 2 regulate key tumor hallmarks: apoptosis, proliferation, and regulation of the pluripotency state (135–137). Overexpression of CtBPs results in oncogenic transformation of human colon epithelium, similar to HRAS mutation, and may define an early activation step in gastrointestinal malignancies (136, 137). PDAC tumors highly overexpress both CtBP1 and 2, and both isoforms are known to regulate PDAC metastasis (138). CtBPs contain a targetable dehydrogenase domain, required for their transcriptional function. The novel anti-CtBP inhibitor [4-chloro-hydroxyimino phenylpyruvate (4-Cl-HIPP)] targets this dehydrogenase domain (131). Loss of CtBP2 decreases metastatic capacity and reduces peritoneal PDAC seeding in a genetically engineered CKP PDAC murine model (131, 138). In primary CKP tumors, treatment with 4-Cl-HIPP reduces tumor volume, equivalent to gemcitabine treatment alone. When dosed together, 4-Cl-HIPP enhances the action of gemcitabine, reducing primary CKP tumors by greater than 50% (138). While early in development, anti-CtBPs are promising adjuvants to traditional chemotherapeutics.

Canalicular multispecific organic anion transporter 2 (ABCC3) is a member of the ATP-binding cassette (ABC family) and is associated with chemoresistance in PDAC. ABCC3 is regulated by p53, and controls PDAC proliferation via STAT3 and HIF1α signaling (139). ABCC3 overexpression correlates with worse prognosis after resection, greater chemoresistance, and is present in early cancerous lesions (140, 141). A nonsteroidal anti-inflammatory drug derivative, MCI-715, inhibits ABCC3 via inhibiting STAT3 and HIF1α downstream signaling, leading to apoptosis in AsPC1, HPAFII, and CFPAC-1 PDAC cell lines. Administration of MCI-715 reduces tumor volume and improves overall survival in a HPAFII subcutaneous xenograft model, a KPC transgenic murine model, and a PDX model (142). Early data is promising that anti-ABCC3 therapy is effective alone and further research into synergistic potential between MCI-715 and conventional chemotherapeutics would be of considerable interest.

The steroid receptor coactivator family (SRC) are nuclear receptors, which act as scaffolds for assembly of coactivator homocomplexes. Of the three SRC members, SRC-3 is more frequently overexpressed in human tumors and SRC-3 levels increase from low-grade PanIN to malignant PDAC (143). Using bufalin, a 14-β-hydroxy steroid, inhibition of SRC3 significantly reduces PDAC cell viability in Panc-1 and Capan-2 cell lines (144). In vivo, treatment with bufalin reduces tumor proliferation and increases animal survival in orthotopic Panc-1-Luc cell line xenografts. Several studies report SRC-3 as a potent oncogene, which may contribute to the genetic complexity of PDAC (145–147).

The small-molecule inhibitors described here possess significant advantages over previously tested, and failed, agents in PDAC. Several of these agents target PDAC cellular metabolism, starving metastatic niches and limiting primary tumor growth. Other targets strip PDAC cells of protective chemoresistance defenses, making tumors vulnerable to conventional chemotherapeutics. These inhibitors target proteins that regulate multiple tumor hallmarks, influencing proliferation, metastasis, and CSC potentiation. All of the agents demonstrate efficacy in vivo and have potential to be translated to clinical trials. Although unlikely to be sufficient by themselves, several show a benefit when combined with conventional chemotherapeutics.

As an alternative to the delivery of antibodies and small-molecule inhibitors, nanoparticle-based strategies deliver high pay-loads of cytotoxic agents to the tumor as a concentrated package (Fig. 1). The targeting or localization of the drug-loaded nanoparticle typically relies on the greater metabolic activity and leakier vasculature of tumors compared with normal tissue, a property termed the enhanced permeability and retention effect (EPR; refs. 148, 149). While the differences in biochemical and biomechanical properties between normal and cancerous tissue can be useful in aiding drug delivery, the EPR effect is often exaggerated in many preclinical models, where murine tumor proliferation rates exceed those of human tumors (150). This higher proliferation rate produces less developed, leakier blood vessels, which are not reflective of most human tumor vessels. Therefore, modifications to nanoparticle materials to improve tumor specificity, such as including tumor specific proteins in the polymeric structure, is likely necessary (151).

Resistance to cancer therapies is both molecular and biophysical in nature. Molecular resistance includes the redundancy of pro-oncogenic pathways, modification or alteration of tumor-specific proteins and receptors, and survival of target negative cells. The challenge of significant tumor heterogeneity is reflected in the common failure of targeted therapies in patients with pancreatic cancer. Biophysical mechanisms of resistance encompass environmental factors within the tumor microenvironment that pose barriers to drug delivery (152). The composition of the tumor extracellular matrix, a dense collagen and elastic network, interspersed with glycosaminoglycans and proteoglycans, produces a hydrophilic gel, thereby increasing local hydrostatic pressure. In PDAC, spare blood vessel density and thick desmoplastic matrix further inhibit drug penetrance (151, 152). In addition, the tumor microenvironment tends to be more acidic, contains numerous proteases, and is a nidus of reactive oxygen species, all of which can deactivate the drug, degrade protein therapies, and alter drug kinetics (150–152). Nanoparticle-based drug delivery vehicles are designed to overcome the above challenges with local accumulation of drug payload (e.g., conventional chemotherapeutics, nucleic acids, or immunomodulatory agents). Various nanoparticle formulations are being investigated, ranging from polymeric, protein, or lipid assembly, and we next discuss recent preclinical advances in nanoparticles for the treatment of PDAC.

Biocompatible polymers, such as poly lactic acid, PLA, and copolymer derivatives, such as poly lactic-co-glycolic acid (PLGA) are commonly employed in nanoparticle development. A paclitaxel-loaded PLGA nanoparticle (NP) with an anti-MUC1 antibody, TAB004, on the surface readily internalizes in MUC1-positive cells, with MUC1-positive cells exhibiting decreased viability when treated with the nanoparticle versus paclitaxel alone (153). In vivo imaging studies of isocyanine green encapsulated TAB004 conjugated NPs shows localization to KCM pancreatic cell line orthotopic tumors following a single IP administration of 50 mg/kg of the nanoparticle. In another study, α-mangostin, a sonic hedgehog inhibitor, encapsulated PLGA NPs reduce growth in CSC-like cells and prevent EMT by upregulating E-cadherin and inhibiting N-cadherin and Slug, SNAI2, NANOG, c-MYC, and Oct4 (Supplementary Fig. SI4A; ref. 154). In the genetically engineered KPC mouse model of PDAC, this NP formulation prevents progression of pancreatic intraneoplasia to adenocarcinoma and liver metastasis. As with in vitro studies, the in vivo efficacy of this nanoparticle stems from inhibition of pluripotency factors, Gli targets, and the Shh pathway, preventing crucial steps in tumor progression. Using a rat model of metastatic PDAC, a paclitaxel-loaded responsive polymeric NP accumulates in pancreatic tumor cells following intraperitoneal injection and specifically delivers drug to these tumors (155). The NPs maintain tight encapsulation of paclitaxel prior to reaching the tumor but then responsively swell and release a high dose of paclitaxel when in the acidic tumor environment. Although treatment with the NPs does not significantly improve survival outcomes compared to free paclitaxel; substantially fewer side-effects are present, suggesting that NP delivery under this method may reduce dose-limiting toxicities and enable completion of treatment regimens.

Polymeric nanoparticles also provide a means to deliver poorly soluble immunomodulatory or tumor-suppressive agents that cannot otherwise be delivered intravenously (156). Curcumin, a suppressor of carcinogenesis in several cancer subtypes: breast, colon, pancreatic, prostate, and ovarian, acts through several pathways, regulating cell-cycle arrest through upregulation of p16, p53, and p14ARF, induction of apoptosis through BAX, BAD, and BCL2 proteins, and modulating NFκB signaling (156). Curcumin inhibits metastasis and angiogenesis in numerous xenograft cancer models, particularly in orthotopic xenograft models of PDAC, suggesting that its benefits extend beyond cancer prevention. Encapsulation sidesteps curcumin's poor aqueous solubility and stability and offers a potentially viable method for delivery at sufficient levels. Curcumin-loaded PLGA-chitosan nanoparticles internalize and retain in Panc1 and MIA PaCa2 cells at a higher level than curcumin itself, with greater in vitro potency and reduced tumor migration via a scratch assay (Supplementary Fig. SI4B; ref. 156). In vivo studies are needed to assess the potential benefit of curcumin.

To improve the local delivery and penetrance of polymeric NPs, which typically rely on EPR and bulk diffusion, modifications to the polymer backbone are being explored. By changing the specific characteristics of the polymer (e.g., charge, hydrophilicity, lipophilicity) or through the addition of surface aptamers, peptides, or even antibodies (which improve selective internalization) better pharmacokinetic profiles are achieved. For example, a sequentially triggered NP relies on an aptamer conjugated to a cell-penetrating peptide-campothecin prodrug (Apt/CPP-CPTD) and amphiphilic PEG copolymer. The NP utilizes a GBI-10 aptamer “camouflage,” relying on GBI-10′s negative charge and tenascin-C's targeting ability to reduce nonspecific uptake of a cell-penetrating peptide nanoparticle (Supplementary Fig. S4C; ref. 157). The NP polymer backbone contains a reductively cleaved disulfide linkage, taking advantage of the greater glutathione concentration within the tumor environment, to release camptothecin following NP internalization. In vitro, the NP performs equivalently to free campothecin in MIA PaCa2 cells, but shows improved drug penetration, as measured by spheroid penetrance. After intravenous administration, the NPs localize primarily in the tumors of a subcutaneous xenograft MIA PaCa2 mouse, and the NP treatment group, containing an equivalent 10 mg/kg of camptothecin, affords a significant decrease in tumor volumes compared with free camptothecin alone, or saline control, demonstrating improved efficacy.

Protein-based nanoparticles offer potentially greater biocompatibility, improved stability, and alternative functionalization capability compared with polymeric nanoparticles (158). Abraxane, a paclitaxel loaded albumin-based nanoparticle, is the prime example and is currently used clinically (27, 159, 160). More recently, the tumor suppressor gene, IL24, has been loaded into a bovine serum albumin (BSA) NPs. Successful gene delivery occurs in Panc1 and BxPC-3 cancer cells, with a marked reduction in cell proliferation and an increase in apoptosis associated with elevation of VEGF expression (161). Treatment of subcutaneous BxPC-3 and Panc-1 xenografts with the NPs significantly decreases tumor volume relative to controls. Encapsulation of Parviflorin D, an anti-proliferative agent derived from the Plectranthus genus, in an albumin-based NP significantly improves drug delivery by increasing drug solubility in vitro, and demonstrates improved potency in BxPc-3 and Panc-1 PDAC cell lines (162). To improve the specificity of albumin-based nanoparticles, Ji and colleagues surface-conjugated the arginine-glycine-aspartic acid (RGD) peptide to their nanoparticles. These RGD modified BSA nanoparticles show greater internalization and retention in BxPC-3 cells compared with BSA nanoparticles alone, and are safe and tolerated in vitro and in vivo (163). The greater specificity of these RGD-albumin nanoparticles improves internalization beyond the EPR effect. Encapsulation of the nanoparticles with gemcitabine significantly improves the efficacy of gemcitabine treatment in BxPC-3 cells, suggesting that this formulation may improve overall success of BSA derived nanoparticles and offer an advancement over Abraxane.

Investigating whether tumor derived exosome-based nanoparticles repress tumor cell growth is another active research area and an exciting alternative to polymer-based approaches (164). Protein characterization of pancreatic cancer cell line, SOJ, exosome-derived NPs reveals a diverse packaging of proteins, including glucose-6-phosphate isomerase, Hsp90, and BSDL (164). Treatment of SOJ cells with the SOJ-derived exosome NPs reduces proliferation in a dose-dependent response, with similar, but variable response in other cell lines (165). Evidence of nanoparticle efficacy includes: a decrease in overall PI3K/Akt/GSK-3β activity, an increase in PTEN dephosphorylation, an elevation of BAX, and a decrease in BCL2 expression. The viability of the control human umbilical vein endothelial cells is not affected by the SOJ NPs.

Currently, two nanoparticle formulations, Abraxane (discussed above) and Onivyde, are FDA approved for treatment of advanced PDAC. Onivyde is a liposomal formulation of irinotecan approved for patients with metastatic PDAC having failed gemcitabine therapy. Onivyde extends survival in combination with fluorouracil and leucovorin, compared with these drugs alone (mOS 6.1 months vs. 4.2 months; ref. 166). A phase I/II trial is evaluating Onivyde with fluorouracil, leucovorin, and oxaliplatin as a first line therapy in PDAC. These clinical successes hold promise for nanoparticle research as part of cancer therapeutic development.

The direct modification of pro-oncogenic genes and reactivation of tumor suppressor genes, through gene therapy, offers a potentially more specific approach to treating pancreatic cancer. Known genetic mutations have been discussed, and represent key targets for drug developers. For example, the loss of crucial tumor-suppressive genes can result in acquired drug resistance in cancer cells, limiting the efficacy of many therapies (167). Although gene delivery, considered a holy grail in cancer therapy, is advancing in the clinic, many successful preclinical studies have failed to show efficacy in clinical trials (167, 168). In this section, we describe several recent nucleic acid treatments in PDAC (Fig. 1).

Delivery of silencing RNAs (siRNA) that target oncogenes offers a therapeutic potential to abrogate tumor progression. siRNAs rapidly degrade before reaching the tumor tissue, therefore, a protective carrier is needed to enable selective delivery of the siRNA to the tumor. For example, programmable DNA nanoparticles (DNP), containing a nucleolin targeting aptamer, deliver BCL2 targeting siRNA and reduce tumor volume in DMS53 subcutaneous xenografts (169). An alternative strategy uses a peptide-based nanoparticle to deliver siRNA for Ras inhibition. This peptide-based siRNA loaded nanoparticle successfully reduces KP1 cell viability and KPCC pancreatic tumors in mice (170). The former approach, through a specific targeting aptamer, offers the advantage of selective targeting to the tumor, and reduces off-targeting effects.

Rather than relying on delivery of siRNA through a nanoparticle carrier, an exciting alternative approach involves reprograming mesenchymal stroma/stem cells (AD-MSC) to release trimeric and multimeric sTRAIL (171). These modified AD-MSCs produce stable multimeric, soluble sTRAIL structures, via lentiviral transduction, capable of inducing apoptosis in BxPC-3, MIA PaCa-2, and primary PDAC cell lines with greater cytotoxicity than rhTrail (Supplementary Fig. SI5A). Modified AD-MSCs transplanted into subcutaneous tumor bearing BxPC-3 mice produce sufficient levels of sTRAIL to decrease tumor volume. However, this technique will be technically challenging and expensive compared with nanoparticle delivery, and may not be amenable to metastatic disease treatment.

miRNA delivery provides an alternative to direct gene modification, relying on pan-suppressive effects. A pentablock copolymer of poly(ethylene glycol) diacrylate (PEG-DA) and miR-345 coblock polymer with encapsulated gemcitabine delivers both miR-345 and gemcitabine to significantly reduce Capan-1 cell viability in vitro and tumor burden in vivo in subcutaneous xenograft models (Supplementary Fig. S5B; ref. 172). Human umbilical cord mesenchymal stromal cell–derived exosomes, containing miR-145-5p, rapidly internalize in human Panc-1 cell lines affording an increase in apoptosis and cell-cycle arrest (173). Overexpression of miR-145-5p significantly reduces tumor volume in a subcutaneous murine xenograft model of Panc1. Cell-derived exosomes offer better potential biocompatibility than polymeric nanoparticles; however, polymeric nanoparticles are better characterized and readily functionalized compared with the native systems.

An alternative approach to nucleic acid therapy is the use of aptamers (single-stranded oligonucleotides) for targeted delivery (174, 175) Aptamers are a practical alternative due to ease of chemical synthesis, chemical modification, and high stability; however, limited success in aptamer-targeted therapies have been seen in cancer (176–179). A novel aptamer–drug conjugate (ApDC) utilizing AS1411 binds selectively to the nucleolin, a cellular protein highly expressed in several PDAC subtypes (180). Capan-1, MIA PaCa2, and AsPC-1 cancer cell lines, but not normal H6c7 cell lines, internalize and retain stable conjugates of AS1411 and gemcitabine (Supplementary Fig. S5C; ref. 181). The AS1411-gemcitabine conjugate shows a slight decrease in potency, but greater target specificity than gemcitabine alone, and demonstrates reduced tumor volumes in a subcutaneous xenograft model of Capan-1 in mice. An alternative approach utilizes a pancreatic cancer–specific aptamer, 2′-Fluoropyrimidine RNA aptamers (2′F-RNA) conjugate designed to deliver a small activating RNA (saRNA) for upregulating C/EBPα, an inhibitor of p21 (181). Downregulation of C/EBPα occurs from loss of KDM6B, leading to increased metastatic drive in pancreatic cancer. C/EBPα-based aptamers are highly specific to cancer cell lines, demonstrating good internalization and retention in Panc1, AsPC-1, MIA PaCa2, and Capan-1 cell lines. Restoration of C/EBPα in Panc-1 cell lines correlates with decreased cell proliferation. Treatment of subcutaneous models of Panc1 with C/EBPα-aptamer affords a modest decrease in tumor size, without significant toxicity, confirming target specificity and relevance of C/EBPα in PDAC progression. Expanding on this concept, researchers are investigating more potent chemotherapeutics, typically reserved for antibody–drug conjugates, for the development of aptamer–drug conjugates. For example, an auristatin-bound EGFR-targeting aptamer is cytotoxic to Panc1, MIA PaCa2, and BxPC-3 cell lines, offering the potential to create highly specific antitumor agents with potency exceeding conventional chemotherapeutics, with a greater therapeutic window (182).

Identification of more selective aptamers along with target validation is an active area of research. For example, cell-SELEX screening of human pancreatic cancer (HPAC)-derived spheroids to target potential cancer stem cell (CSC)-initiating or CSC-related genes yielded seven aptamers exhibiting high affinity binding, and target CD133, CD244, IHH, NANOG, and ALDH1-expressing cells (183). Excitingly, these aptamers bind to circulating tumor cells derived from patients. Further validation of these targets is necessary, but may provide an additional method for aptameric drug delivery.

Advances in nucleic acid delivery have progressed significantly since their inception. With the development of more stable carriers and application of aptamers as drug delivery vectors, nucleic acid therapy shows exciting progress in PDAC therapy. However, relative to other therapies discussed here, nucleic acid therapies are not widely used in the clinic. It is unknown how stable these therapies will be in clinical practice or how well tolerated, although the recent approvals of nucleic acid therapies provide impetus for continued discovery, research, and translation.

Outcomes in PDAC are limited by the dearth of available, successful treatment options. Nonspecific early symptoms, tumor aggressiveness, and diagnosis of late-stage disease result in clinical outcomes that are significantly worse than those of other cancers. For early-stage disease, surgical resection with curative intent is still the preferred option, but the utilization of new imaging modalities may be able to improve surgical outcomes through improved detection of microscopic disease and delineation of tumor margins to assure complete resection. FDA-approved surgical cameras with NIR detection already enhance visualization of the tumor beyond conventional surgical approaches. Coupled with new developments in systemic tumor detection, visual contrast agents, and combined theranostic materials, the cure rates in specific patient subsets will likely be improved.

Current treatment modalities in PDAC are disappointing. To date, there is no targeted therapy that shows superior outcomes compared with current systemic chemotherapy even though these regimens can cause unacceptable toxicity. Developing successful targeted therapies for PDAC is complicated by tumor heterogeneity, biophysical limitations in the tumor microenvironment, and redundancy of pro-oncogenic pathways. Furthermore, while cancer immunotherapy has been a boon to many patients with cancer, pancreatic cancer stands out as one of the few human cancers that does not respond to the current immunotherapy protocols. The failure in implementing therapies used to treat other cancers underscores these unique challenges in PDAC. To address these issues, advances in identifying novel targets and improving the implementation and delivery of therapeutics specifically in PDAC are being employed. These approaches address specific patterns and pathways unique to PDAC with the goal of preventing cancer cells from escaping therapy and host immunosurveillance, as well as avoiding the development of chemoresistance. Through specific targeting of PDAC cancer cells and known biochemical processes, advanced disease can be treated and its progression stopped or significantly delayed.

Several discussed agents augment conventional chemotherapeutics and inhibit pathways that confer resistance. Novel methods of drug delivery are affording both safer and targeted delivery of conventional chemotherapeutics, with the goals of improving their therapeutic window, providing specific delivery to the tumor, or avoiding the systemic toxicity responsible for limiting doses in clinic. In concert with other treatment modalities, improved and multimodal targeting will prevent drug evasion and tumor progression characteristic of late-stage PDAC. Continuing advancements in tumor-omics, as well as development of novel delivery and imaging systems, as described herein, must translate to better patient outcomes in the coming decade. PDAC outcomes are dismal and this is particularly evident in light of the success in significantly reducing cancer-related deaths in breast, prostate, colon, and lung. Failures in PDAC therapy and disparities in outcomes relative to other cancers outline a unique clinical challenge, one that will require a multidisciplinary approach and relies on innovative, evidence-based treatments.

Y. Colson has ownership interest (including patents) in Patent US 7,671,095 B2/US 8,334,324 B2/US 8,338,492 B2 (available for license) and Patent US 8,795,707 B2 (available for license). M.W. Grinstaff has ownership interest (including patents) in patent application by BU (available for licensing). No potential conflicts of interest were disclosed by the other authors.

This work was supported, in part, by NIH (F30 CA220843 to C. Gromisch; R01 CA227433 and R01 CA232056 to M.W. Grinstaff and Y. Colson; CA133085 to K. Liu), Veterans Administration (CX001364 to K. Liu), and Boston University.