Age is in the Eye of the Beholder: Distinguishing Molecular Signatures in Early-onset Pancreatic Adenocarcinomas

In this issue of Clinical Cancer Research, Tsang and colleagues (1) examine four available datasets to identify differences in genetic signatures of pancreatic ductal adenocarcinomas (PDAC) between early-onset cases (<55 years of age) versus nonearly-onset cases (intermediate category defined as 55–70 years; advanced onset >70). RNA sequencing analysis determined that differences exist in the frequency of CDKN2A indels between the early-onset and other populations. Their overarching conclusion is that early-onset pancreatic carcinomas (EOPC) have a distinct mutational landscape than nonearly-onset cases, and that these findings provide the basis for hypothesis-generating studies.

Pancreatic cancer, like many solid tumors in adults, is generally considered to be a disease of aging. In recent years, there has been attention paid to young adult, or early onset, forms of other gastrointestinal cancers, most notably cancers of the colon and rectum. There is less information about forms of PDAC diagnosed in adults substantially younger than the median age of 71.

A previous study by Ben-Aharon and colleagues [also published in Clinical Cancer Research (2)] identified age 55 as a valid cut-off for patients labeled as “early-onset,” compared with “advanced onset” cases of PDAC (identified as cases in patients >70 years). The notion that molecular signatures differ significantly between these age-stratified populations is intriguing and opens the door to tailoring future therapeutic strategies to molecular-driven subcategories that may be more prevalent in some age-based populations than others. However, there remain many challenges and barriers before this could justifiably be translated into the clinical setting. This is true most especially because PDAC, like other stroma-rich carcinomas (e.g., ovarian) and others that are less driven by actionable, targetable mutations, and more so driven by a cellular landscape that makes the tumor microenvironment difficult to effectively target. Moreover, there is increasing evidence that the biology of PDAC tumors may differ significantly by anatomic location and stage, that is, localized resectable versus borderline resectable or locally advanced versus metastatic to distant organs. Tsang and colleagues. parsed out their analysis by comparing genomic and transcriptomic correlates in 277 resectable versus 269 advanced (locally advanced or metastatic) PDAC cases (1). Early-onset cases demonstrated notable differences in expression of CDKN2A and also of FOXC2 (forkhead box protein C2). Specifically, EOPC cases harbored lower frequencies of somatic single-nucleotide variants and indels of CDKN2A, but showed higher transcription of FOXC2. One especially pertinent aspect of this signature is enrichment of gene sets associated with epithelial-to-mesenchymal transition, a phenotype associated somewhat controversially with “stemness” properties in other contexts, but which has not been traditionally been strongly correlated to metastatic potential and chemoresistance in PDAC per se. However, despite the noted differences in molecular signatures between the EOPC and other cases, there were no significant differences in survival seen between these populations in this study (1). So what is the potential relevance of this study if there is no difference in overall survival? We can use it as a starting point for more in-depth evaluation to uncover molecular heterogeneity of pancreatic carcinomas to determine how the driver and passenger mutations in EOPC cases differ from tumors in older patients; determine whether associations exist with known, or even yet-to-be discovered, familial syndromes that increase risk of developing PDAC in younger-than-average populations; and determine whether any of the differentially expressed drivers are targetable with existing or in-discovery phase molecular or immunotherapeutic agents.

How familial syndromes and germline mutations associate with the molecular signatures identified in this study is one interesting path for future study. There is a broad spectrum of hereditary conditions that include but are not limited to hereditary pancreatitis, Peutz–Jeghers syndrome, and familial atypical multiple mole melanoma syndrome (FAMMM) that have similar molecular aberrations that may align. Germline mutations CDKN2A have been associated with and are a cause of FAMMM, However, mutations in CDKN2A are more prevalent in sporadic forms of pancreatic cancer than in germline cases that increase risk of this disease (3). Ultimately, hereditary cases of pancreatic adenocarcinoma in general make up a small proportion of all cases, but further investigation of the early-onset population may uncover a higher proportion of familial cases that will also shed better light on these and other molecular drivers and also passenger mutations of this subpopulation. In the short-term clinical care of these patients in our oncology clinics, as standard-of-care recommendations have veered toward recommending cancer genetics consultation for all patients with new diagnosis of pancreatic carcinomas (4), regardless of age, germline testing of these patients may likewise evolve and uncover patterns of gene penetration of CKDN2A and additional relevant genes.

In terms of accuracy of genomic profiling and identification of mutations in these tumors, an important biologic principle that will need to be addressed in future studies is how intertumoral heterogeneity (within the same patient) could account for any differences, and any specifications of whether the “advanced” cases that were metastatic sequenced metastatic lesions, or entirely focused on sequencing primary tumors (Fig. 1). If one were to analyze sequencing of, for example, metastatic tumors to the liver in the advanced cases, and primary tumors in resectable cases, how does one account for any potential heterogeneity unless we directly compare primary tumors from both groups? The question of intratumoral heterogeneity of these and other molecular factors further underscores the complexity that arises in accurate assessment and analysis of variability across populations. Additional questions include whether the dataset shown here is representative across the spectrum of pancreatic adenocarcinomas, or more highly applicable to subgroups, such as those patients whose carcinomas arise in the setting of chronic pancreatitis or similar states of long-term inflammation. From a mechanistic standpoint, the data provided by Tsang and colleagues. provide basis for exploring all of the above questions in more detail.

How will these data impact diagnosis of pancreatic cancer and/or care of patients in the years to come? As CDKN2A normally impairs the inhibitor functions of proteins that regulate the G₁–S-phase of the cell cycle, which in turn usually inhibits cyclin-dependent kinases CDK4 and CDK6, therapeutic strategies targeting the downstream pathway are potential candidates for future clinical trials in patients stratified by this mutation. Novel direct inhibitors of CDKN2A are being investigated in current clinical trials focused on advanced solid tumors, as are inhibitors of CDK 4 and 6 (e.g., abemaciclib in breast cancer) targeting the downstream targets affected by that mutation. As the authors correctly point out, loss of cell-cycle regulator p16ink4 encoded by CDKN2A is implicated in progression of KRAS-mutant cells. With recent, albeit very early, reports of modest response rates of solid tumors harboring the specific KRAS-variant G12C with a small-molecule inhibitor (5), the prospect of combining downstream inhibitors with inhibitors of corresponding variants in KRAS may present an additional novel therapeutic strategy. Nonetheless, a great deal of questions remain and reproducible preclinical evidence is needed to justify this speculation.

These are just a few examples of how to leverage the results of this study to craft a new identity for the early-onset pancreatic cancer population in a way that can inform molecularly targeted trials. In sum, there are both bench-to-bedside as well as bedside-to-bench translational implications of these data. The authors should be congratulated on this important effort which recognizes that molecular heterogeneity of PDAC may fall into stratifications based on patient age, as just one factor that can set the stage for future epidemiologic and molecular biology-based studies.

E. Lou reports research grants from the American Association for Cancer Research (2019 AACR-Novocure Tumor-Treating Fields Research Grant) and the Minnesota Ovarian Cancer Alliance; honorarium and travel expenses for a research talk at GlaxoSmithKline (2016); honoraria and travel expenses for lab-based research talks, and equipment for laboratory-based research from Novocure (2018–20); consultant, Nomocan Pharmaceuticals (unpaid); scientific advisory board member, Minnetronix, LLC (2018–present; unpaid); consultant and speaker honorarium, Boston Scientific US (2019); institutional principal investigator for clinical trials sponsored by Celgene, Novocure, Intima Biosciences, and the NCI; and University of Minnesota membership in the Caris Life Sciences Precision Oncology Alliance (unpaid).

E. Lou was supported by the 2019 AACR-Novocure Tumor Treating Fields Research Grant, grant number 1-60-62-LOU.