2.1 Pathogenesis of PDAC

Despite the aggressive nature of PDAC, it is preceded by the development of noninvasive precancerous lesions years prior to the appearance of malignant cells.¹⁸ Several categories of precancerous lesions exist based on their size and involvement with the pancreatic ductal system.¹⁹ The most common premalignant lesion of PDAC is pancreatic intraepithelial neoplasia (PanIN), which are neoplasms of the pancreatic duct that are less than 5 mm.^18-21 Next, less than 10% of PDACs arise from intraductal papillary mucinous neoplasms (IPMNs), which are finger-like, microcystic lesions that protrude into the pancreatic duct.¹⁸ Finally, mucinous cystic neoplasms are the least common precancerous lesions.¹⁹ They are characterized by distinctive ovarian-type stroma that do not involve the pancreatic ductal system.^{18, 19} While the development and progression of precancerous lesions have been well studied, the origins of these lesions remain a controversy: studies have reported these lesions originating from both ductal and acinar cells.²²

Classification of the precancerous lesions hold clinical significance as well. PanINs are much smaller compared with IPMNs, making them less likely to be detected incidentally.^{18, 23} Precancerous lesions are also categorized as low-grade versus high-grade morphologically: low-grade lesions are characterized by basally oriented nuclei with mild to moderate cytological atypia, while high-grade lesions are marked by loss of nuclear polarity, significant architectural alterations and cytological atypia.²⁴ Low-grade lesions are more commonly found compared with high-grade lesions, but high-grade lesions have been associated with higher risk of progression to carcinoma, and studies have shown the possible direct link between high-grade IPMN and PDAC.¹⁸

2.2 Genetic and molecular alterations in PDAC

PDAC is a heterogeneous disease marked by four common genetic alterations: KRAS oncogenic activation and CDKN2A, TP53, SMAD4 tumor suppressor gene inactivation.²⁵ Activating KRAS mutations are the earliest and most common genetic alterations to occur in PDAC progression; they are present in greater than 90% of low-grade PanINs and 90–95% of PDAC.^26-28 KRAS encodes a membrane-bound GTPase that regulates a range of critical intracellular processes such as cell growth, differentiation, and survival by moving between its GDP-bound inactive and GTP-bound active form.^{25, 29} KRAS is usually present in quiescent cells in its inactive form; upon growth stimulation, guanine nucleotide exchange factors (GEFs) catalyze the GDP–GTP exchange to change KRAS into its active state.^{30, 31} Most activating KRAS mutations result in impairment of the protein's GTPase, causing KRAS to be locked in its GTP-bound active form and continued activation of major downstream pathways.³²

While activating KRAS mutations are often one of the earliest mutations and promote PDAC initiation, research has suggested that KRAS mutations alone are incapable of promoting PDAC progression.^{25, 33} Located on chromosome 9, CDKN2A encodes the p16^INK4A protein, which negatively regulates cell cycle progression from G1-phase to S-phase by disrupting complex formation between CDK4/6 and cyclin D.^{34, 35} Loss of CDKN2A is key in the progression of PDAC as it allows cells with activating KRAS mutations to escape cell senescence.³⁶ Another one of the four commonly mutated driver genes, TP53 encodes transcription factor p53, which binds onto DNA to promote transcription of genes (e.g., CDKN1A, BAX, etc.) that trigger cell cycle arrest or apoptosis upon DNA damage.^{37, 38} TP53 mutations that result in functional loss of the transcriptional factor are observed in 50–75% of PDAC patients and typically occur following loss of CDKN2A.³⁹

Finally, SMAD4 is a crucial mediator in the transforming growth factor-ß (TGF-ß) family signal transduction, which has been found to have context-dependent and often contradictory roles in PDAC pathogenesis.⁴⁰ In nonmalignant cells, TGF-ß signaling directly promotes tumor suppression by inhibiting expression of cyclin-dependent kinases (CDKs) and enhancing expression of CDK inhibitors (e.g., p15 and p21).⁴¹ On the other hand, TGF-ß signaling can act as a tumor promoter in malignant cells by inducing epithelial-to-mesenchymal transition (EMT) and creating an immunosuppressive environment.⁴¹ Loss of SMAD4 function inhibits the antiproliferative effect of TGF-ß and enhances tumor aggressiveness.⁴²

In addition to these four driver genes, a wide range of genes are also mutated in different subsets of PDAC, including BRCA1/2, PALB2, ATM, ARID1A, KMT2D, BRAF, FGFR1, albeit at a lower frequency (≤10%).^43-47 These alterations converge to disrupt several core cellular signaling pathways and processes, including KRAS signaling, TGF-ß signaling, DNA damage control, apoptosis, regulation of G1/S phase transition, and Hedgehog signaling.^{25, 48}

2.3 Signaling pathways in PDAC tumorigenesis and metastasis

A number of signaling pathways have been identified to play critical roles in the tumorigenesis, prognosis, metastasis, and response to treatment of PDAC, including RAS and its downstream pathways, JAK/STAT pathway, Wnt/β-Catenin pathway, and Notch pathway.^{49, 50}

2.4 EMT in PDAC

EMT is a dynamic and biologically critical process that plays a role in mediating PDAC's highly aggressive, immunosuppressive, and resistant nature.^{93, 94} In different tissues of the body, epithelial cadherin molecules known as cell surface E-cadherin help epithelial sheets maintain their structural integrity.^{95, 96} During EMT, cancer cells lose their epithelial features and markers, including E-cadherin, occludin, claudin, and begin expressing mesenchymal markers such as N-cadherin, fibronectin, and vimentin.^97-99 Morphologically, cells undergoing EMT change from polygonal shapes to spindle-shaped while also leading to the disruption of normal cell–cell and cell–matrix adhesion, remodeling of cytoskeleton, and loss of cellular polarity.¹⁰⁰ Once cancer cells undergo EMT, they acquire migratory and invasive capabilities that help facilitate growth and metastasis while becoming more resistant to chemotherapy and other cytotoxic therapies.^101-104 EMT is modulated by a complex network of epigenetic modifications, transcription regulation, and multiple signaling pathways regulated by major EMT-inducing transcription factors (EMT-TFs) Snai1 (Snail), Snai2 (Slug), Zeb1, Zeb2, and Twist (Twist basic helix-hoop-helix transcription factor 1).^105-108

A major contributor to EMT is cancer-associated fibroblasts (CAFs).¹⁰⁹ CAFs secrete a myriad of cytokines that promote EMT of PDAC cells, including TGFβ, IL-1, and IL-6.^{110, 111} Under normal TGFβ and SMAD signaling, the pathway is growth suppressive: binding of TGFβ with TGFβ receptor I/II leads to phosphorylation of SMAD2 and SMAD3, which heterotrimerize with SMAD4 and translocate to the nucleus to act as a transcription factor that downregulates c-Myc and upregulates CDK inhibitors p15 and p27, ultimately resulting in cell cycle arrest.^112-114 However, TGFβ signaling induces the engagement of EMT in addition to driving fibrosis and immunosuppression in the pancreatic cancer TME.^{115, 116} Activated SMAD2/3/4 promotes transcription of multiple aforementioned EMT-TFs.^117-119 Furthermore, IL-1α/β mediates a sustained inflammatory TME, which is critical driving mechanism of EMT, through the NF-κB pathway.^120-123 Expression of TNF-α or constitutively active IKK2 are also capable of inducing a EMT phenotype by upregulating the expression of mesenchymal marker vimentin and EMT transcription factor ZEB1 while downregulating epithelial marker E-cadherin.¹²⁴

Within the TME, the extracellular matrix (ECM), which is primarily produced by CAFs, also contributes to promoting EMT in addition to hindering the infiltration of treatment and cytotoxic T cells.¹²⁵ Collagen is the major component of the ECM, accounting for over 80% of ECM mass in PDAC.¹²⁶ Ligation of ECM proteins and integrins causes clustering of integrins and the formation of focal adhesions on the cell membrane where the integrins connect to actin cytoskeleton via linker proteins.^{127, 128} This interaction triggers the activation of EMT-promoting pathways, including the FAK/SFK, Rac/Cdc42, and Fyn/YES pathways.¹²⁹ Particularly, the FAK pathway can promote EMT by upregulating N-cadherin expression through the JNK signaling pathway.^130-132 FAK can also activate Yes-associated protein (YAP) and Transcriptional Coactivator with PDZ-Binding Motif (TAZ), forming the YAP–TAZ complex that translocates to the nucleus to transactivate genes that code for EMT-TFs.^{133, 134} The dense ECM also limits vascularization and perfusion, creating a hypoxic TME that heightens chemoresistance and tumor aggressiveness.^135-137 Hypoxia-inducible factors (HIFs) are activated to help cells adapt to the hypoxic environment through transcriptional regulation; of note, HIF-1α also binds to the promoter region of EMT-TFs.^138-140 Despite the heavy emphasis on CAFs role in promoting EMT, other cells such as tumor-associated macrophages (TAMs) play a role in inducing a mesenchymal phenotype in PDAC cells.

2.5 TME of PDAC

PDAC is characterized by a unique TME with an extensive desmoplastic stroma that can make up to 90% of the tumor volume.¹⁴¹ Cellular crosstalk between neoplastic and non-neoplastic cells, including fibroblasts, endothelial, immune cells, create an immunosuppressive or “cold” TME in PDAC (Figure 1).¹⁴² The dense desmoplastic stroma in PDAC is made up of ECM constituents such as collagen secreted by CAFs upon signaling from neoplastic cells.¹⁴³ Interestingly, research has demonstrated that CAFs demonstrate both tumor-suppressive and tumor promoting properties.¹⁴⁴ This could be attributed to the recent identification of different subtypes of CAFs (inflammatory, myofibroblastic, and antigen presenting) that exhibit distinctive gene expression profiles and function.¹⁴⁵ Regardless, the resulting desmoplasia promotes tumor progression by significantly limiting drug penetration and immune cell infiltration. CAFs are further capable of creating an immunosuppressive TME by secreting suppressive factors that recruit immunosuppressive T regulatory cells (Tregs) and myeloid cells and upregulating immune checkpoints on effector T cells.¹⁴⁶

Subsets of myeloid cells, most notably TAMs and myeloid-derived suppressor cells (MDSCs), also contribute to the immunosuppressive TME of PDAC.¹⁴³ TAMs can be placed on a spectrum from antitumoral M1 macrophages to protumoral M2 macrophages for simplification.^{147, 148} M2 macrophages support an immunosuppressive TME by secreting IL-6 and TGF-ß, promoting the differentiation of CD4⁺ T cells into Tregs, and expressing high levels of T cell suppressive ligands such as PD-L1, which can trigger T cell exhaustion.^{143, 146} Distinct from TAMs, MDSCs are immature myeloid cells characterized by the expression of signature markers CD11b, CD33, and CD14 or CD15/66b.¹⁴³ They can be classified into two main populations: polymorphonuclear (PMN)-MDSCs/granulocytic-MDSCs and mononuclear MSDCs (M-MDSCs).¹⁴⁹ MDSCs, based on the subpopulation they belong in, promote an immunosuppressive TME through a range of different mechanisms.¹⁵⁰ Multiple key enzymes, including arginase-1 and indoleamine 2,3-dioxygenase, are activated, depleting available l-arginine and l-tryptophan and causing T cell inhibition.¹⁵¹ Similar to TAMs, MDSCs also express inhibitory receptors (e.g., PD-L1 and CTLA-4) and recruit Tregs through CD40 engagement to induce T cell exhaustion and tolerance.^{151, 152} The high production of reactive oxygen species and nitric oxide by PMN-MDSCs and M-MDSCs respectively inhibit the fitness, proliferation, and movement of T cells within the TME.^{151, 153}

4.1 Locally advanced pancreatic cancer

Staging of the tumor plays a critical role in determining whether the patient should receive NAT or not. NAT has shown to be beneficial for patients with LAPC or BRPC. A study of 415 LAPC patients from 2013 to 2019 showed that surgical resection after NAT is achievable in a highly selected group of patients.¹⁷¹ Eighty-four (20%) of the 415 patients underwent surgical resection after NAT (median duration of 5 months) and had significantly higher median OS (35.3 vs. 16.3 months) and smaller tumor size compared with patients who did not undergo surgical resection.¹⁷¹ NAT not only provides better outcomes for patients with LAPC, but it also facilitates more precise selection of patients who will benefit from and should undergo surgical resection.^{158, 172} Patients who experience disease progression or no response to neoadjuvant chemotherapy are unlikely to respond favorably to adjuvant chemotherapy as well, thus increasing the risk of recurrence after surgical resection and indicating a more aggressive malignancy. Consequently, these patients who are precluded from surgical resection are most likely spared from an unnecessary and complicated surgical procedure. Several other studies have also reported favorable results after surgical resection following NAT even though imaging post-NAT shows continued irresectability.^{158, 173, 174} Tee et al.¹⁷⁵ found that the 2-year OS rate of LAPC patients who underwent pancreatectomy with arterial resection following NAT was significantly higher than patients who underwent upfront surgery (62.3 vs. 25.8%, p = 0.038). Overall, NAT should be administered for patients with LAPC to not only reduce tumor size but also to help select candidates for resection who do not have disease progression, which is an indicator of a biologically aggressive disease.¹⁵⁸

4.2 Borderline resectable pancreatic cancer

Approximately 30% of PDAC patients have borderline resectable tumors, and current guidelines suggest the use of NAT following the completion of several RCTs.¹⁶⁴ Jang et al.¹⁷⁶ reported the first prospective RCT for NAT on BRPC patients. Patients in the experimental arm were treated with four cycles of neoadjuvant gemcitabine-based chemoradiotherapy followed by surgical resection, while patients in the control arm received upfront surgery followed by four cycles of adjuvant gemcitabine-based chemoradiotherapy.¹⁷⁶ Both groups received another four cycles of gemcitabine maintenance chemotherapy. The study found that the neoadjuvant group had significantly better 2-year survival rate (840.7% vs. 26.1%, hazard ratio HR 1.495, p = 0.028) and R0 resection rate (51.8 vs. 26.1%, p = 0.004).¹⁷⁶ The study was terminated due to the statistically significant benefits of neoadjuvant chemoradiation.¹⁷⁶

In the recently completed phase III PREOPANC-1 trial, 246 patients with resectable pancreatic cancer or BRPC were randomized to either upfront surgery with adjuvant gemcitabine or neoadjuvant gemcitabine-based chemoradiotherapy and adjuvant gemcitabine.^{177, 178} Long-term follow-up showed a significant increase in median OS for patients that underwent NAT compared with upfront surgery (15.7 vs. 14.3 mo, HR 0.73, p = 0.025).¹⁷⁸ Furthermore, R0 resection, which is a prognostic predictor for PDAC, was achieved in 49 out of 68 (72%) of patients who underwent neoadjuvant chemoradiotherapy versus 35 out of 82 (43%) of patients who underwent upfront surgery (p < 0.001).¹⁷⁸ Interestingly, interim analysis of the clinical trial did not yield any significant results, highlighting the need for future RCTs to follow-up long term to obtain meaningful results. Previous RCTs have also reported the safety and the benefits of NAT against BRPC in terms of R0 resection rate and median OS.^{176, 179-181}

4.3 Resectable pancreatic cancer

While the use of NAT for BRPC is supported by other RCTs, the evidence is not as clear for patients with resectable tumors at presentation. Some studies support the use of NAT against resectable tumors.¹⁶ The phase II-III Preop-2/JSAP-05 trial in Japan compared the effects of NAT with gemcitabine and S-1 chemotherapy versus upfront surgery, both of which were followed by adjuvant S-1 chemotherapy on resectable patients.¹⁸² The RCT demonstrated superior survival, with a median OS of 37 versus 27 months, for the group who underwent NAT (HR 0.72, 95% confidence interval CI 0.55–0.94, p = 0.015).¹⁸² However, administering NAT and delaying surgery may result in losing the opportunity for surgical resection, particularly for patients with resectable tumors at presentation, as a result of disease progression during NAT.¹⁷⁷ The hazard ratio of 0.73 in favor of NAT in the PREOPANC-1 trial was primarily driven by patients with BRPC (HR 0.67, p = 0.045) rather than with resectable tumors (HR 0.79, p = 0.23).^{177, 178} Furthermore, results of the SWOG-S1505 trial showed that patients with resectable disease who underwent NAT (either mFOLFIRNOX or gemcitabine plus nab-paclitaxel) did not show superior results compared with patients with similar disease progression who underwent surgery first historically.^{177, 183} A meta-analysis of 7 RCTs that studied NAT on resectable pancreatic cancer or BRPC concluded that more evidence is needed to demonstrate whether patients with resectable tumors should be treated with NAT as opposed to upfront surgery.¹⁸⁴ The current standard of care based on the 2022 NCCN guidelines for resectable PDAC is upfront surgery and adjuvant treatment.¹⁸⁵ The guidelines advise the use of NAT when high risk features such as significantly elevated carbohydrate antigen (CA) 19-9, large primary tumor or lymph nodes, suspicions of advanced or metastatic disease are observed.¹⁸⁵ A list of completed RCTs on the use of NAT for patients with resectable pancreatic cancer or BRPC are summarized in Table 1.

4.4 Late-stage PDAC

Patients with late-stage PDAC, which refers to T_2-4N₀M₀ and nodal or distal metastatic PDAC, are inoperable, and palliative therapy had been the only available treatment option historically.^{188, 189} Current guidelines recommend (m)FOLFIRINOX or gemcitabine with nab-paclitaxel (GnP) for patients with better performance status (ECOG 0–1) and single agent gemcitabine with or without a second agent for patients with worse performance status (ECOG 2–3).¹⁹⁰ Recent developments in novel treatments like targeted therapy and immunotherapy have offered many of these inoperable patients another avenue of treatment and a chance to participate in ongoing clinical trials for new agents. Furthermore, germline and somatic testing, suggested by current NCCN guidelines, provide the opportunity for more personalized medicine and a wider range of options for these patients.^{190, 191} Figure 2 summarizes in a flow chart the current guidelines on the use of NAT depending on resectability.

6.1 Imaging response

Once the decision to administer NAT has been made, it is critical to be able to accurately assess patient response to NAT as it offers meaningful prognostic value and can guide future course of treatment. However, methods for assessment have not been standardized. Staging and restaging of PDAC have been traditionally done based on a radiographic response, most commonly with multidetector computed tomography (CT), positron emission tomography (PET) with CT, magnetic resonance imaging, and endosonography.¹⁹⁶ The 2022 NCCN guidelines recommend PET/CT scan for detection of extra pancreatic metastases in the case of high risk patients as PET offers a more in depth look at the morphological and metabolic activity of the lesions.¹⁹⁷ However, PET/CT should only be used in conjunction with high-quality contrast-enhanced CT, and not to replace it.¹⁹⁸ Having said that, assessing tumor response to NAT and determining resectability with only imaging is insufficient.¹⁹⁶ A recent systemic review on BRPC and LAPC reported that most patients had stable disease after NAT and did not experience change in tumor size, volume, or density; only a fraction of patients were observed to have a partial response with tumor size reduction and a decrease in standardized uptake value on the PET scan.¹⁹⁹ The imaging criteria were found to be conflicting with the histopathological tumor regression grading, underlining the inadequacy of measuring NAT response with imaging only.¹⁹⁹ Furthermore, determining tumor response and resectability following NAT using imaging may be tricky as inflammation caused by NAT can mimic a solid tumor.²⁰⁰

6.2 CA 19-9

Biomarkers have also been used to supplement imaging assessment of NAT response. Currently, CA 19-9 is the most widely used and only Food and Drug Administration-approved biomarker to manage PDAC.²⁰¹ Multiple studies have demonstrated the prognostic value of measuring CA 19-9 before and throughout treatment for PDAC.^202-204 An elevated baseline or preoperative CA 19-9 level has been associated with a poor prognosis for PDAC with a shorter median survival; similarly, an increase in CA 19-9 levels throughout treatment is also a sign of disease progression and predictor of worse outcomes. A study with 93 and 41 PDAC patients in the training and validation set respectively reported that CA 19-9 elevation is an early and reliable sign for PDAC recurrence: 2.45 times elevated CA 19-9 values showed recurrence with 90% positive predictive rate.²⁰⁵ In terms of NAT, CA 19-9 levels post-NAT have been found to be predictive of tumor size reduction and survival. In a study of 250 patients, a CA 19-9 response ≥85% was a strong independent predictor of tumor size reduction ≥25% (HR 2.40, p = 0.007) and improved survival (HR 0.47, p = 0.007).²⁰⁶ Results from Heger et al.’s²⁰⁷ retrospective study showed that CA 19-9 levels after neoadjuvant FOLFIRINOX more accurately predicts resectability and survival than dynamic values. Combined, this evidence suggests the need to incorporate CA 19-9 response as an endpoint for NAT and decision making in resectability after NAT.^{206, 207} Having said that, CA 19-9 levels can be difficult to interpret when the patient has a biliary obstruction or presents with a normal CA 19-9 level initially.¹⁹⁶ Other biomarkers such as carcinoembryonic antigen and CA 125 have been found over the past decade to complement serum CA 19-9 levels for diagnosis and prognosis.²⁰⁸

6.3 Circulating tumor cells and circulating tumor DNA

Recently, research has focused on the role of circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA) from liquid biopsies as biomarkers for the screening and monitoring of PDAC patients.^209-213 Effenberger et al.²¹⁴ showed that CTC status affects the outcome of PDAC patients independent of other risk factors and can stratify patients based on OS and progress-free survival (PFS). CTCs were identified in 78% of patients and were correlated with increasing stage (p < 0.001) in a study by Court et al.²¹⁵ Of the 53 patients who underwent surgical resection in Court et al.’s²¹⁵ study, those who eventually experienced occult disease has significantly higher number of CTCs (p < 0.0001). Furthermore, Court et al. found that CTCs were a multivariate predictor of OS (HR 1.38, 95% CI 1.01–1.88, p = 0.040) and a univariate predictor of recurrence-free survival postsurgery (HR 2.36, 95% CI 1.17–4.78, p = 0.017).²¹⁵ Similarly, another study with 112 patients found that higher levels of percent-ctDNA were a multivariate, independent prognostic factor for OS (HR 4.35, 95% CI 1.84–10.24, p = 0.001) in pancreatic cancer.²¹⁶ All this evidence show promise for the use of CTC and ctDNA from liquid biomarkers to assess patient status.

However, there exists several issues with using CTC and ctDNA for screening, diagnostic, and prognostic purposes that have yet to be overcome.²¹⁷ First, a meta-analysis on ctDNA in pancreatic cancer reported a low pooled sensitivity of only 0.64, which could be attributed to the inadequate amount of ctDNA released in the plasma as a result of low tumor necrosis in early PDAC.²¹⁸ This low sensitivity makes it difficult for the use of ctDNA in the screening of PDAC. A more recent meta-analysis calculated a pooled specificity of ctDNA of 0.92 in PDAC, which is only marginally better than the specificity of CA 19-9.^{217, 218} While the theoretical benefits of using ctDNA as a new marker have yet to make a significant clinical impact due to technological limitations, several ongoing studies are currently exploring their use in a clinical setting. An ongoing multicenter study (NCT04246203) conducted by the University of Munich is monitoring and comparing disease progression between patients with and without the presence of detectable ctDNA in a liquid biopsy. CAPS5 (NCT02000089), the current CAPS program, may have also incorporated surveillance of circulating epithelial cells from pancreatic duct fluid as a primary outcome.²¹⁹ Overall, the use of CTC and ctDNA for diagnosing and monitoring PDAC is extremely promising as technologies become more developed.

6.4 Histopathologic response

Another method to assess patient tumor response to NAT is through a histopathologic approach. Multiple grading systems of tumor regression/response in posttreatment pancreatectomy exist for PDAC, including the Evans, White, Ishikawa, and College of American Pathologists (CAP) system.^220-222 The CAP system, one of the more commonly used systems, has four scoring tiers: score 0, complete response; score 1, near complete response; score 2, partial response; score 3, poor or no response.²²³ While the CAP system has been criticized for the lack of difference in survival between score 2 and 3 patients, Maeda et al.⁶ reported that scores 2 and 3 have different prognostic meanings for patients treated with neoadjuvant chemotherapy versus neoadjuvant chemoradiotherapy. Despite scores 2 and 3 not having a significantly different OS for neoadjuvant chemoradiotherapy, a CAP score of 2 was found to be in an independent prognostic factor for improved OS and DFS for neoadjuvant chemotherapy.⁶

A pathological complete response (pCR) is defined as the absence of any observable cancer cells in the pancreas or lymph node on final pathology or a CAP score of 0.²²⁴ Achieving pCR after NAT and resection is a prognostic factor for improved outcomes, while long-term survival is less common if patients have a poor histopathologic response following NAT and pancreatectomy.^{196, 225, 226} In a retrospective study of 186 BRPC/LAPC patients who underwent NAT and pancreatectomy at Johns Hopkins Hospital, 19 (10%) achieved pCR.²²⁶ Achieving pCR was found to be an independent prognostic factor for DFS (HR 0.45, 95% CI 0.22–0.93, p = 0.030) and OS (HR 0.41, 95% CI 0.17–9.97, p = 0.044).²²⁶ However, pCR is rarely achieved in PDAC, with a reported observance of 4–15%.^226-229 A follow-up study at Johns Hopkins Hospital found that only 30 out of 331 patients achieved pCR after NAT and pancreatectomy.²²⁴ Interestingly, even in the rare occurrence of achieving pCR, a portion of these patients develop recurrence despite the lack of any observable cancer cells.²²⁴ Out of the 29 eligible pCR patients, 14 (48%) had recurrence with nine out of the 14 dead at the last follow-up of the study.²²⁴ No clinicopathological factor, including the NAT regimen, radiation modality, and location of recurrence, was significantly associated with OS, DFS, or PFS, highlighting the potential limits of a histopathological approach.²²⁴

Given the variable outcomes despite achieving pCR, Yin et al.²⁰⁰ incorporated liquid biopsy with next-generation sequencing (NGS) of resected specimen to introduce the new concept molecular complete response (mCR). mCR is defined as the absence of tumor-related mutations in both resected specimen and plasma (CTC and ctDNA).²⁰⁰ Thirty-six patients who achieved pCR were selected from a pool of 479 patients who underwent NAT.²⁰⁰ Six patients were identified as mCR, while 15 patients who were positive for either tumor-related mutations in resected samples or plasma (CTC or ctDNA) were classified as non-mCR.²⁰⁰ Although no statistical significance on patient outcome could be achieved due to the small sample size, this study was the first to report the presence of CTC and ctDNA in patients who achieved pCR after NAT and highlights the value of liquid biopsies in prognostic assessment and outcome prediction.²⁰⁰ As future research increases the sensitivity of these biomarkers from liquid biopsies, they can be incorporated into patient assessment prior to and after NAT treatment to determine resectability and guide future adjuvant therapeutic choices.

Ultimately, we recommend a combination of established methods—imaging, CA 19-9, and histopathological—to assess patient response to NAT, reevaluate resectability, and to eventually determine a plan of care for patients going forward. Other more novel methods such as biomarkers and molecular analysis can be incorporated, but more research is required to validate these findings. Most studies suggest that surgical resection be performed 4–8 weeks after completion of NAT if the patient is a candidate for surgery.¹⁷⁰

7.1 Retrospective studies comparing NAT regimens

A number of retrospective studies and meta-analyses have been completed to determine the optimal NAT regimen.^236-238 Tang et al.²³⁸ conducted a meta-analysis that included eight retrospective studies to compare head-to-head neoadjuvant FOLFIRINOX versus GnP for localized pancreatic cancer. Three out of the eight included studies reported a significantly prolonged OS for FOLFIRINOX, while the remaining five demonstrated a similar efficacy in terms of OS.²³⁸ Furthermore, a sub-group analysis adjusted for age also showed significantly longer OS for FOLFIRINOX when compared with GnP.²³⁸ Yet despite the 1–3-year survival rates being higher for patients treated with FOLFIRINOX, the 5-year survival rates were comparable between patients who were treated with FOLFIRINOX versus GnP.²³⁸ In terms of perioperative parameters, the meta-analysis found that patients with either treatment had statistically similar resection rates and R0 resection rates, but patients who were treated with FOLFIRINOX had a significantly lower perineual invasion rate, which may contribute to its longer OS.²³⁸ Ultimately, Tang et al.²³⁸ concluded that FOLFIRINOX is noninferior compared with GnP for patients who are capable of receiving FOLFIRINOX. A more recent single institution retrospective study comparing FOLFIRINOX and GnP for LAPC demonstrated that NAT with FOLFIRINOX is associated with more favorable results versus GnP as well.²³⁷ A better OS was observed in the FOFLIRINOX group (85.1 vs. 54.3 weeks); the authors suggested that this may be a result of the higher rates of conversion/radical surgery for patients treated with FOLFIRINOX (41.8 vs. 22.2%).²³⁷ However, FOLFIRINOX has been associated with higher risk of adverse events, particularly in the elderly, so patient condition must also be taken into account in addition to the efficacy of the chemotherapy agents when deciding the treatment course.^{11, 239}

7.2 Completed and ongoing RCTs comparing NAT regimens

Despite the results from these retrospective studies, RCTs that directly compare different neoadjuvant chemotherapeutic regimens for PDAC are required to provide conclusive evidence. Table 2 shows four RCTs that compare different neoadjuvant chemotherapeutic and novel immunotherapeutic agents. Notably, the SWOG S1505 trial randomized patients with resectable PDAC to either perioperative (three cycles neoadjuvant and three cycles adjuvant) FOLFIRINOX or GnP. The researchers found no significant difference between the 2-year OS rate, the primary endpoint, between the two groups (41.6 vs. 48.8%), nor statistically significant improvement compared with the a priori 2-year OS rate of 40%.²⁴⁰ And while the SWOG S1505 trial demonstrated safety and relatively high resectability rates of both neoadjuvant chemotherapeutic regimens, it found no significantly improved median OS (22.4 vs 23.6 months) of either regimen compared with historical standards.²⁴⁰

Several other RCTs that are investigating the optimal neoadjuvant chemotherapeutic and immunotherapeutic regimen are summarized in Table 3. In the ongoing multicenter phase III PREOPANC-2 RCT, patients with resectable pancreatic cancer or BPRC are randomized into two groups. The first group receives eight cycles of neoadjuvant FOLFIRINOX followed by surgical resection without adjuvant therapy; the second group receives three cycles of gemcitabine with hypofractionated radiotherapy (36 Gy in 15 cycles) during the second cycle followed by surgical resection and four cycles of adjuvant gemcitabine. The 2019 ASCO guidelines suggests a total of 6 months of chemotherapy, taking into consideration both neoadjuvant and adjuvant chemotherapy.²⁴¹

7.3 Patient performance status and NAT regimen

In addition to the molecular basis of disease and the efficacy of chemotherapy regimens, the optimal NAT regimen is also dependent on the performance status of patients. FOLFIRINOX and mFOLFIRINOX should be limited to patients with ECOG 0–1, while GnP is acceptable for patients with ECOG 0–2 according to the 2022 NCCN guidelines.¹⁹⁷ 5-Fu + leucovorin + liposomal irinotecan is reasonable for patients with ECOG 0–2 if patients have not been previously treated with irinotecan.¹⁹⁷ Furthermore, comorbidities should be incorporated into the decision making concerning the neoadjuvant regimen and the decision to offer surgery and radiotherapy on a case-by-case basis, particularly for older adults.²⁵³ Gemcitabine or fluorouracil should be considered for patients with an ECOG of 2 or comorbidities that preclude them from more aggressive regimens.²⁵⁴

7.4 Neoadjuvant radiotherapy

The use of radiotherapy neoadjuvant style also remains a controversial issue. RCTs such as PREOPANC-1 and Jang et al.¹⁷⁸ have successfully showed the advantages of neoadjuvant chemoradiotherapy compared with upfront surgery for patients with BRPC and LAPC. This conclusion is supported by other retrospective studies.²⁵⁵ Yet, due to the initial results of the ESPAC-1 trial showing a significant survival benefit of adjuvant chemotherapy but negative effects of adjuvant radiotherapy, neoadjuvant radiotherapy use has been minimized in Europe despite still being commonly used in the United States.¹⁵⁸ Katz et al.²³⁵ also reported in an abstract report of the ALLIANCE A021501 trial that no improvement in OS or R0 resection rate were found with the addition of stereotactic body radiation therapy. Furthermore, ESPAC-5F trial demonstrated that neoadjuvant mFOLFIRINOX or gemcitabine plus capecitabine resulted in similar OS durations, and significantly better than neoadjuvant 50.4-Gy capecitabine-based chemoradiotherapy.²⁵⁶ These clinical trials along with other studies question whether the addition of radiotherapy in NAT is beneficial or may actually be detrimental.^{4, 177, 257} Consequently, given the lack of conclusive evidence on the effects of radiotherapy in NAT, we suggest that the use of radiotherapy in NAT should be limited to RCTs only.

9.1 Targeting germline mutations

Several studies have combined to show that cancer predisposing germline mutations, which are directly passed down from parents to children, are present in 3.8–9.7% of PDAC patients.^{273, 274} These mutations predispose individuals for several hereditary cancers, such as ovarian and breast cancer (BRCA1, BRCA2, ATM, PALB2), atypical multiple mole melanoma (CDKN2A), Peutz-Jeghers (STK11), and so on.^274-277 The NCCN recommends that all patients diagnosed with PDAC to undergo germline mutation testing regardless of stage and family history.¹⁹⁷ Additionally, genetic testing for at-risk individuals with family history of cancers can help enhance screening measures and reach asymptomatic individuals earlier on in the disease.⁴⁷

Mutations that negatively affect DNA damage repair genes, including BRCA1, BRCA2, ATM, and PALB2, are one of the better studied group of mutations.^{278, 279} These genes encode for key proteins in homology-directed repair (HDR), which is an error-free repair system for DNA double strand breaks (DSB).^{280, 281} When loss of function mutations occur in one or more of these genes, HDR is impaired, and cells become more reliant on nonhomologous end joining and other DSB repair systems, which are more error prone.²⁸² Previous studies have reported that HDR-mutated ovarian and breast cancers are more susceptible to drugs that induce DSB, such as platinum-based chemotherapy.^283-285 In 2018, Blair et al.²⁸⁶ reported that within a group of BRCA1/2-mutated PDAC patients, platinum-based adjuvant chemotherapy was associated with better outcomes compared with nonplatinum-based adjuvant chemotherapy and no adjuvant chemotherapy (31.0 vs. 17.8 vs. 9.3 mo, p < 0.001). Furthermore, in a retrospective study of 50 HDR-mutated patients out of 262 total patients, Park et al.²⁸⁷ demonstrated patients with HDR mutations had better PFS compared with non-HDR-mutated patients when treated with first-line platinum therapy, but not when the HDR-mutated patients with nonfirst line platinum therapy. Finally, a recently completed phase II study in patients with demonstrated that gemcitabine + cisplatin treatment is an effective treatment against BRCA1/2- and PALB2-mutated PDACs.²⁸⁸ Based on these evidence, platinum-based chemotherapy, such as gemcitabine + cisplatin or FOLFIRINOX, should be considered as first-line treatments for patients with HDR mutations.

Poly (ADP-ribose) polymerase (PARP) inhibitors (PARPi) are another group of drugs that have shown great promise against HDR-mutated PDAC.^{289, 290} PARP enzymes play a role in repairing DNA single strand breaks (SSB) through base excision repair. PARPi prevents repair of SSB by trapping PARP enzymes on DNA, creating a stall in the replication fork and converting the SSB into a DSB.^{282, 289} While normal cells without HDR mutations are able to repair the induced DSB via HDR, HDR-deficient cells are unable to repair the DSB in an error-free way, leading to accumulation of DNA errors and eventual tumor cell death.²⁹¹ Several different PARPi (Olaparib, Velaparib, and Rucaparib) have been developed and tested over the past few years with promising results.²⁹² In the phase III POLO trial, patients with germline BRCA1/2 mutations and metastatic PDAC that had not progressed during first-line platinum-based chemotherapy were randomly assigned to receive either maintenance Olaparib or placebo.²⁹³ Median PFS was found to be significantly higher in the Olaparib group compared with the placebo group (7.4 vs. 3.9 mo, HR 0.53, 95% CI 0.35–0.82, p = 0.004).²⁹³ Given the success of the POLO trials, other clinical trials have begun to test the use of Olaparib alone or with in combination with other treatments. A phase II Johns Hopkins study (NCT04753879) is currently recruiting to assess the efficacy of maintenance Olaparib and Pembrolizumab after low dose gemcitabine, nab-paclitaxel, capecitabine, cisplatin, and irinotecan (GAX-CI) in patients with untreated metastatic pancreatic cancer.²⁹⁴ Similarly, the National Cancer Institute is also conducting a phase II trial (NCT04548752) comparing the use of Olaparib alone versus Olaparib and Pembrolizumab in metastatic germline BRCA1/2-mutated pancreatic cancer.²⁹⁵

Other PARPi are currently undergoing similar clinical trials as well. In a phase II trial (NCT03140670), Reiss et al.²⁹⁶ found that Rucaparib is safe and effective for patients with metastatic BRCA1/2- or PALB2-mutated pancreatic cancer with a median PFS of 13.1 months (95% CI 4.4–21.8 months) and median OS of 23.5 months (95% CI 20–27 months). The use of Niraparib and dostarlimab (anti-PD-1) on BRCA1/2 and PALB2-mutated metastatic pancreatic cancer is currently being tested in a phase II trial (NCT04493060).²⁹⁷

9.2 Targeting somatic mutations

Testing for somatic genetic alterations with available targeted therapy is recommended for patients with LAPC or metastatic PDAC suitable for therapy.²⁹⁸ While KRAS may be the most common mutation in PDAC, present in 90–95% of tumors, developing an inhibitor targeting KRAS mutations have been difficult due to lack of a high affinity binding sites for drugs.^{299, 300} In 2018, Janes et al.³⁰¹ found a small molecular compound, ARS-1620, that was capable of binding to a previously discovered allosteric binding pocket in a specific KRAS mutation (KRAS^G12C) and sequester it in its GDP-bound inactive form. ARS-1620 was shown to have antitumor effects via inhibition of phosphor-ERK, phosphor-S6, and phospho-AKT in KRAS^G12C preclinical models, leading to the development of other KRAS^G12C inhibitors (e.g., Sotosarib, Agadrasib).³⁰¹ And while in 2021, Sotorasib (AMG510) and Adagrasib (MRTX849) received FDA approval and FDA breakthrough therapy designation respectively, the impact of these inhibitors will be limited as KRAS^G12C mutation is only present in 1% in PDAC cases.^302-305 However, multiple inhibitors targeting other KRAS mutations, pan-KRAS inhibitors, SHP2 inhibitors, and therapeutic vaccines targeting KRAS are under development and testing.³⁰⁶ MRTX1133, a promising, novel KRAS^G12D inhibitor, is awaiting clinical trial after having shown preclinical efficacy—eight out of 11 KRAS^G12D-mutated cell line xenografts and patient-derived xenografts showed tumor regression after MRTX1133 treatment.^{47, 307} BI 1701963, which is currently under phase I trial, is a pan-KRAS inhibitor that disrupts the binding of KRAS to GEFs SOS1.³⁰⁸ Additionally, after promising initial safety and tolerability were reported for SHP2 inhibitor TNO155, multiple clinical trials are being conducted the drug in combination with other therapies.³⁰⁹ Table 4 below provides a list of ongoing clinical trials targeting KRAS and its upstream and downstream pathways.

9.3 Microsatellite instability

Microsatellite instability (MSI) represents another subset of PDAC characterized by deficits in the DNA mismatch repair, which can occur as a result of genetic disorders like Lynch syndrome.³¹⁷ Studies have found that compared with microsatellite stable tumors, MSI-high tumors are more responsive to FOLFIRINOX and comparatively less responsive to 5-Fu and gemcitabine.^{18, 318} Furthermore, it has been reported MSI-high tumors are more responsive to immune checkpoint inhibitor therapy, and Pembrolizumab (anti-PD-1) is now FDA approved for MSI-high cancers.³¹⁹ In the phase II KEYNOTE-158 trial, researchers explored the efficacy of Pembrolizumab in noncolorectal MSI-high or mismatch repair deficient cancers: out of the 22 patients with MSI-pancreatic cancer, there was an overall response rate of 18.2% (one complete response and three partial responses), suggesting a meaningful response despite it being lower than the overall response rate of all noncolorectal cancers in the study (34.3%).^{320, 321}

9.4 Molecular subtypes of PDAC

Using analysis of primary PDAC RNA expression profiles, several studies have defined different molecular subtypes of PDAC and their associated transcriptional networks.³²² Several classification methods have been developed, but integration of these methods reveal two basic subtypes of PDAC: classical and basal.^323-325 Classical subtypes show higher expression of epithelial and pancreatic lineage markers and are correlated with a better prognosis; on the other hand, basal subtypes are enriched for cell cycle progression, TGF-ß signaling, and EMT and are correlated with a poorer prognosis.³²⁶ Studies have begun exploring the clinical significance of these molecular subtypes in terms of treatment regimen.³²⁶

To understand the clinical significance of different molecular subtypes, the COMPASS trial recruited patients with metastatic pancreatic cancer prior to first-line therapy.³²⁷ Patient tumors were (RNA) sequenced, and tumors were classified into classical (80%) and basal (20%).³²⁷ Worse prognosis for basal tumors were confirmed with a 60% radiographic progression rate for basal tumors compared with the 15% of classical tumors.³²⁷ GATA6 expression in the tumor was identified as a suitable surrogate biomarker for differentiating between molecular subtypes and predict chemosensitivity.³²⁷ Building on this, the Pancreatic Adenocarcinoma Signature Stratification for Treatment (PASS-01) trial is currently recruiting and randomizing metastatic PDAC patients to either mFFX or GnP.³²⁸ This trial will explore the use of molecular profiling, GATA6 expression, and chemosensitivity signatures to predict response to therapy.

Furthermore, in the ongoing phase III ESPAC-6 trial (NCT05314998), resected PDAC patients are randomized to either oxaliplatin- or gemcitabine-based adjuvant chemotherapy by either standard clinical criteria (control arm) or transcriptomic treatment specific signature (experimental arm).³²⁹ The trial seeks to determine whether allocating patients based on transcriptomic stratification is superior to using the standard clinical criteria by comparing DFS and OS. Ultimately, identifying transcriptional subtypes could offer patients more selective chemotherapy treatments, but further understanding on the mechanisms and clinical significance of these subtypes is required.

9.5 Targeted therapy in NAT

Overall, there is a lack of research on the use of targeted therapy neoadjuvant style; most research is still currently focused on the efficacy of the targeted therapies as a second- or third-line therapy and/or in conjunction with chemotherapy in an adjuvant manner. However, results of NGS can certainly help determine the NAT regimen. For example, in the case of BRCA1/2 mutations, FOLFIRINOX and other platinum-based chemotherapy should be chosen as the NAT regimen if patient performance status allows. Ultimately, there is an urgent need to not only identify possible therapeutic targets and their associated treatments, but to also test their efficacy in both adjuvant and neoadjuvant settings to overcome the immunosuppressive and chemoresistant nature of PDAC.