KRAS wild-type pancreatic ductal adenocarcinoma: molecular pathology and therapeutic opportunities

The MAPK pathway is the most important core signaling pathway in PDAC [3, 8, 9, 11, 26], in which the activation of RAS is a crucial step. Such activation is mediated by a guanine nucleotide exchange factor [30]. Active RAS is self-inactivated by intrinsic GTPase activity, based on a GTPase-activating protein [31]. KRAS mutations in PDAC often involve codons 12, 13 and 61; most of them are G12D or G12R substitutions, which decrease the intrinsic GTPase activity, thus resulting in prolonged activation of RAS [31].

Notably, the activation of MAPK pathway is possible also in the absence of KRAS mutations [30]. A recent study by Singhi et al. definitively demonstrated that the MAPK signaling is activated in about one third of KRAS wild-type PDAC [11]. In this cohort, BRAF alterations were the most prevalent, and included activating missense mutations, amplification and kinase fusions. BRAF mutations were mutually exclusive with those of KRAS; furthermore, kinase fusions in BRAF were not present in KRAS mutant cases [11]. BRAF encodes for the B-RAF proto-oncogene, a serine/threonine kinase belonging to a family of MAPK kinases [30]. The most common mutations in BRAF usually involve codon 600 and often results in a V600E substitution, leading to constitutive activation of its kinase function, which is the results also of the other aforementioned alterations [32, 33]. The mutual exclusivity of KRAS and BRAF mutations in PDAC suggests that the activating mutations of these genes can compensate for each other in PDAC oncogenesis via activation of the MAPK pathway [30]. Recent data also suggest that other molecular alterations may modulate cancer cell (including PDAC) susceptibility to MAPK inhibition: indeed, PTEN loss was shown to portend intrinsic resistance to MEK inhibitors and synergistic activity of combined MAPK/phosphatidylinositol 3 kinase (PI3K) pathway inhibition [34]; however, combined MAPK/PI3K inhibition did not prove effective in either cell line models of PDAC [35] or PDAC patients [36], presumably owing to the lack of PTEN alterations.

B-RAF is immediately downstream of RAS and triggers MAP2K1/MEK, which activates MAPK1/ERK2, important mediators of the MAPK pathway. Then, the activating signal is passed on by a chain of kinase reactions, overall establishing MAPK signaling pathway [30]. Consequently, activating mutations of KRAS and BRAF produce at last the activation of this pathway, which is crucial for pancreatic cancer (Fig. 2). Interestingly, MAPK kinases are classified into three classes based on their distinctive effectors: extracellular signal-regulated kinases (ERK), Jun amino-terminal kinases (JNK) and p38-mitogen-activated protein kinases (p38MAPK) [37]. Each class regulates distinct functions: ERK is involved in proliferation, JNK in apoptosis and differentiation and p38MAPK in stress responses [37]. To enact specific cell responses, activated MAPK kinases translocate into the nucleus, where, through phosphorylation of different transcription factors, they can directly modulate the expression of different genes [30, 38]. Additional mechanisms that may replace KRAS/BRAF function for RTK/RAS/MAPK activation included mutations or amplifications of GNAS, EGFR, ERBB2, MET, ERBB3 and FGFR2 [9].

Regarding therapeutic approaches to MAPK pathway targeting, in addition to the above described MAPK inhibitors, some other interesting data come from a large, multicenter, non-randomized trial of 581 PDAC patients, as part of the so-called “Know Your Tumor” initiative [39]. In this cohort, wild-type KRAS tumors accounted for 8% of the whole PDAC cohort, and a significant proportion (24%) of these had alterations in other MAPK pathway effectors, including BRAF activating alterations. In this setting, patients who received molecular target therapy, including MAPK inhibitors and BRAF-targeted drugs, achieved an improved progression-free survival and median overall survival [38]. Other data regarding direct BRAF inhibitors come from a recent study, which reported a 6-months remission of a BRAF- mutant PDAC patient treated with the inhibitor dabrafenib [40]. Moreover, preclinical data indicate that even wt-BRAF might constitute a suitable therapeutic target in PDAC, taking advantage of the therapeutic synergism of “vertical” combination strategies simultaneously targeting BRAF and MEK along the same pathway, regardless of KRAS mutational status [41].

From the point of view of molecular pathology, an important consideration regards the specific BRAF mutation to be encountered in PDAC. Indeed, mutations that are the most common in other tumor types, including melanoma and thyroid carcinoma, are not frequently present in PDAC. Singhi and colleagues [11] reported typical V600E and V600K BRAF mutations in only about 1/5 of PDAC with BRAF alterations. This point requires the adoption of next-generation sequencing (NGS) technologies in clinical practice to guarantee in-depth analysis of the coding regions of BRAF, coupled with the possibility to investigate other BRAF alterations, such as the presence of fusion genes.

Another genetic alteration enriched in KRAS wild-type PDAC is microsatellite instability (MSI)/defective mismatch repair (dMMR) [11, 42‐45]. A recent study has definitively demonstrated that MSI/dMMR PDAC harbor KRAS mutations less frequently than conventional PDAC, reaching statistically significant values [42]. However, it should be acknowledged that about one third of MSI/dMMR PDAC can present KRAS mutations [43, 44]. Microsatellites are short, repetitive DNA sequences of 1–6 base pairs present throughout the genome, mostly in noncoding regions [35]. Their repetitive nature renders them very sensitive to DNA mismatching errors, which can occur during DNA replication or iatrogenic damage [45]. Cancers harboring a dMMR are very often hypermutated and typically accumulate mutations in microsatellites: this condition is termed MSI [45].

The MSI phenotype was first described in the familial cancer condition known as Lynch syndrome (LS), where the MMR genes MLH1, MSH2, MSH6 or PMS2 harbor germline mutations and portend marked susceptibility to develop several types of cancer, including PDAC [43]. MSI can be tested using immunohistochemistry (IHC) and molecular tests, including classic polymerase chain reaction (PCR) and next-generation sequencing (NGS) [45‐47]. Immunohistochemical analysis is more reliable for cancers belonging to the spectrum of LS [43], where the loss of expression of the heterodimers MLH1-PMS2 and/or MSH2-MSH6 represent a very reliable surrogate of dMMR [44, 45]. For other cancer types, PCR or NGS approaches appear more robust for MSI/dMMR assessment and should be used instead of IHC. In PDAC, the use of IHC appears a reliable method (Fig. 3), although this cancer type does not represent one of the most common malignancies in LS patients [44, 45].

MSI in PDAC has been described with variable frequencies, with a prevalence ranging approximatively from 0.1 to 7% in the most recent investigations on large cohorts of patients [30, 48‐51]. However, a recent comprehensive seminal paper has established its prevalence in PDAC at around 1–2% [44]. Highest values of MSI prevalence have been reached in specific PDAC subtypes, such as medullary variant, mucinous/colloid variant and IPMN-derived carcinomas [40, 44, 51]. Among conventional PDAC, Humphris et al., who interrogated 385 pancreatic cancer genomes for mutational signatures inferring defects in DNA repair, identified MSI in 1% of tumors [48]. Along this line, Hu et al., using an NGS assay designed to perform targeted deep sequencing of cancer-associated genes, found a 0.8% MSI frequency in a cohort of 833 PDAC [50]. Although it is a rare condition, the new frontiers of immunotherapy have opened new important therapeutic opportunities for tumors with this molecular alteration. Indeed, because of the defective mismatch repair machinery, tumors with MSI are hypermutated neoplasms; this genetic alteration leads to the synthesis of several (up to 50 times more than microsatellite stable tumors) aberrant and potentially immunogenic neo-antigens by the tumor cells [52]. An important self-response to the presence of these neo-antigens is a diffuse infiltration of the tumor area by cytotoxic T-cell lymphocytes (CTLs) [52]. Recent studies have highlighted the concomitant expression of multiple active immune checkpoint markers, such as the Programmed Cell Death Protein 1 (PD-1) and its ligand PD-L1. Their interaction cause T cells functional exhaustion and unresponsiveness [53]. Based on this discovery, Le et al. evaluated in 2015 the clinical activity of pembrolizumab, an anti-PD-1 immune checkpoint inhibitor, in a cohort of metastatic colorectal and non-colorectal carcinoma patients with or without MSI [54]. However, no patient with PDAC was included in this study. The results of this phase 2 trial clearly showed that MSI status was able to predict clinical benefits from immune checkpoint blockade therapy with pembrolizumab [55, 56]. In 2017, the same group published the “KEYNOTE-158” trial, investigating PD-1 blockade efficacy in patients with advanced dMMR cancers across different tumor types, including also PDAC [57]. If initially the objective response rate was similar between colorectal versus other tumor subtypes, an update of the trial including a total of 22 MSI/dMMR PDAC patients, showed only 4 of 22 patients with objective responses, which represented the lowest objective response among the different cancers investigated [58]. These findings highlight the complex biology of PDAC and call for further investigations in this field.

Of interest, tumor mutational burden (TMB) is another important variable in this setting, since it has been strictly correlated to the response to immunotherapy. TMB is a value that represents the rate of mutations in a tumor, and is expressed as the number of mutations per megabase. In PDAC, a high TMB is a rare finding and very often occurs simultaneously with MSI [11, 47, 50]. New NGS approaches can provide simultaneously data on MSI and TMB, with potentially immediate consequences on therapeutic opportunities for these specific PDAC molecular subgroups.

From the point of view of molecular pathology, it is important to acknowledge that IHC, performed using the antibodies for all 4 MMR proteins (MLH1, PMS2, MSH2 and MSH6) represents a very reliable methodology to assess MSI in PDAC. However, in cases of doubtful IHC results, MSI/dMMR should be assessed with PCR-based MSI or NGS. In cases with limited tissue (e.g.: endoscopic/ultrasound guided fine needle biopsy), NGS should be adopted as the first choice, as highlighted by a recent study on this topic [44]. Due to the potential importance associated with a diagnosis of MSI/dMMR in the context of PDAC, including eligibility for immunotherapy trials, it is of primary importance that all methodologies for MSI/dMMR assessment follow standardized protocols (including all pre-analytical phases) and up to date recommendations [45].

The most important original report on this topic is a recent massive sequencing-based paper with a cohort of 3594 PDAC, where Singhi et al. showed that kinase fusion genes are one of the most frequent putative driver alterations in KRAS wild-type PDACs [11]. They found specific kinase fusions in FGFR2 (12 cases), RAF (7 cases), ALK (5 cases), RET (4 cases), MET (2 cases), NTRK1 (2 cases), ERBB4 (1 case) and FGFR3 (1 case), representing about 7.6% of genetic alterations of all KRAS wild-type PDAC. In the majority of fusion genes, a serine/threonine kinase or tyrosine kinase catalytic domain was fused to an oligomerization domain, which may represent a mutual mechanism of activation [11]. All these kinase fusions were mutually exclusive and, similar to BRAF, they were not present in PDACs with KRAS alterations [11].

Regarding the therapeutic approach to PDAC patients with kinase fusions, there are still limited data in the literature. Interestingly, a previous research identified 4 PDAC patients with ALK fusions; 3 of them showed stable disease with normalization of serum CA19-9 levels after treatment with an ALK inhibitor [59]. Notably, Pishvaian et al. recently described 2 PDAC patients with NTRK1 fusions; such patients were already metastatic at the time of diagnosis but showed partial responses to the specific targeted drug entrectinib [60]. Regarding BRAF and RAF fusions, these molecular alterations have not been described in PDAC but in another exocrine pancreatic cancer, which is acinar cell carcinoma [61]. Intragenic/in-frame deletions in BRAF represent other potential driver alterations in KRAS wild-type PDAC. Although also in this case there are limited data, Aguirre et al. reported a single PDAC patient with oncogenic BRAF deletion with response to the pathway inhibitor trametinib [62].

In the last few years, several studies have detected neuregulin 1 (NRG1) fusion genes across different cancer types, particularly in invasive mucinous adenocarcinoma (IMA) of the lung and PDAC. Moreover, NRG1 fusions have been detected with several fusion partners [63]. The NRG1 gene encodes for a member of the epidermal growth factor (EGF) family, which binds to ERBB3, thereby causing its heterodimerization with ERBB2. NRG1 gene fusions are generally in-frame and generate fusion proteins that maintain the extracellular EGF domain of NRG1 and the transmembrane domain of the rearrangement partner. Thus, the EGF domain of the fusion protein can constitutively bind to its partner and activate signaling through MAPK, PI3K-AKT, and NF-kB, increasing tumor proliferation and survival [63, 64]. In patients affected by PDAC, NRG1 fusions are detected only in KRAS wild-type tumors and usually in the absence of other concomitant driver gene mutations, suggesting that the rare NRG1 fusions act as an oncogenic driver in this tumor. In addition, some studies have demonstrated the actionability of NRG1 fusions in PDAC. Indeed, when patients affected by heavily pretreated PDAC carrying NRG1 fusion received matched therapy with afatinib, an irreversible ERBB1–4 inhibitor, clinical benefit and objective responses were documented [65, 66]. However, Drilon at al recently reported four patients affected by IMA of the lung with NRG1 fusion that did not respond to afatinib, but showed an extraordinary clinical response to treatment with a new monoclonal antibody targeting ERBB3 [67].

From a molecular pathology perspective, the necessity to identify fusion genes further highlights the importance of introducing NGS into clinical practice. For PDAC patients, this point represents an urgent need, at least for KRAS wild-type PDAC. The final pathology report for PDAC should integrate morphology with molecular pathology, following the model of a next-generation histopathologic diagnosis [68].