Big insights from small samples: uncovering the potential of EUS-guided FNAs in pancreatic cancer through predictive genomics and transcriptomics

Over the past two decades, molecular biomarkers have transformed the treatment of many common cancer types by matching precision therapies to cancer subtypes that are known to respond to targeted inhibition. However, chemotherapy remains the dominant treatment modality for pancreatic ductal adenocarcinoma (PDAC), and only a small minority of patients qualify for targeted therapy.¹ Given that PDAC is the third leading cause of cancer-related deaths in the U.S. and has a 5-year survival rate of less than 3.5%, new treatment options are urgently needed.² Cumulatively, only approximately 10% of patients are eligible for targeted therapy. These indications include maintenance PARP inhibition for patients with germline defects in the homologous recombination DNA repair pathway and several molecular alterations that are targetable under pan-solid cancer therapeutic approvals, including mismatch repair deficiency, elevated tumour mutational burden, BRAF V600E mutation, and RET or NTRK rearrangement. However, recent advances in the ability to directly inhibit KRAS, which is altered in approximately 90% of PDACs, hold immense promise for redefining therapeutic approaches to a cancer type once broadly considered intractable for targeted therapy.^3,4

In the October 2024 issue of eBioMedicine, Nicolle et al.⁵ present a comprehensive approach for diagnosis and molecular profiling of pancreatic cancer using endoscopic ultrasound-guided fine-needle aspiration biopsies (EUS-FNABs) to assess primary PDAC. Notably, they study patients with all stages of disease, enabling direct comparison between resectable, borderline resectable, locally advanced, and metastatic cancers. The study's robustness is underscored by its multicentric and prospective design, encompassing a cohort of 397 patients with uniform clinical data collection and follow-up.

Endoscopic ultrasound-guided fine-needle aspiration biopsies of the pancreas are an effective initial diagnostic method for PDAC, allowing for successful diagnosis in most cases while preserving high quality DNA for ancillary testing.⁶ However, despite the safety and cost-effectiveness of EUS-FNABs for evaluation of PDAC, their reported diagnostic sensitivity is variable (65–95%).^7,8 In this context, numerous prior studies have demonstrated that incorporation of molecular techniques boosts sensitivity for diagnosing PDAC and helps distinguish between PDAC and benign pancreatic lesions.^7,9 Furthermore, such molecular approaches can identify actionable molecular alterations. Thus, up-front molecular profiling of PDAC not only increases diagnostic yield but can also inform targeted therapy options for patients at all stages of disease. As EUS-FNABs are the primary specimens obtained from patients during initial evaluation for pancreatic lesions, they provide a key opportunity for molecular workup to aid in diagnosis, prognosis, and treatment decisions, as demonstrated in the current study.

Genomic analysis identified the expected landscape of major driver alterations in pancreatic cancer, establishing the validity of targeted sequencing analysis using EUS-FNAB specimens. Furthermore, analysis of patients with localised versus metastatic disease revealed a higher frequency of TP53 mutations and a lower frequency of RNF43 mutations in metastatic versus non-metastatic primary cancers. These results raise the possibility that genomic features of primary cancers may be partially predictive of metastatic potential. Additionally, evaluation of KRAS mutation allele fraction as an estimate of cancer cellularity demonstrated comparable cellularity for surgical samples across different disease stages and underscored the utility of EUS-FNAB for obtaining representative samples.

Transcriptomic deconvolution was used to estimate the relative composition of major cell types in each cancer and identified a mixture of classical, basal-like, and so-called “proliferating” cancer cells in most specimens, consistent with recent single-cell studies of PDAC. Comparison of cancer composition across stages revealed that the basal-like cancer component was enriched in patients with metastatic disease. Similar to the genomic differences identified between patients with localised and metastatic disease, these results suggest that the transcriptomic profile of a primary cancer may be associated with metastatic potential.

In addition to baseline characterisation, the authors studied the prognostic and predictive utility of molecular biomarkers. Genomic analysis showed that patients with locally advanced or metastatic cancer and homologous recombination deficiency gene mutations respond better to first-line platinum-based chemotherapy, while those with locally advanced cancer and wild-type TP53 genes show improved outcomes with radio-chemotherapy. In a similar vein, a pre-defined transcriptomic signature was predictive of better survival for patients with metastatic PDAC that were treated with gemcitabine-based chemotherapy.

While this study provides a powerful demonstration that molecular biomarker testing is feasible from EUS-FNAB specimens, it also highlights the limitations of testing small specimens, as 45% of specimens yielded insufficient nucleic acid for analysis. While improved molecular analysis methods can enable successful testing of small quantities of nucleic acid, and retrieval of larger tissue quantities can boost nucleic acid yield, the low neoplastic cell content of pancreatic cancer also poses a significant challenge to molecular testing, especially for bulk transcriptional profiling. This is particularly relevant for assignment of classical and basal-like molecular subtype, as these two subtypes differ significantly in prognosis and may also exhibit differential sensitivity to the two main standard-of-care chemotherapy regimens for PDAC. If these results are confirmed, it is possible that molecular subtype testing will become part of the standard of care for all cancers treated with first-line chemotherapy. Notably, robust approaches for determination of subtype status, including the RNA-based PurIST algorithm, already exist and are suitable for clinical implementation.¹⁰ Beyond transcriptional profiling, it is also very possible that determination of KRAS status will be needed to guide first-line therapy selection in the future.

As the rationale for routine molecular profiling of PDAC becomes more firmly established, it will be increasingly important to set realistic expectations about the practical limits imposed by small specimens with low neoplastic cell content. Future research should not only aim to validate the findings of this study, but also identify approaches that increase the per-patient success rate of molecular profiling for pancreatic cancer.

Contributors

Conceptualisation: Melissa Zhao, Anirban Maitra, Jonathan Nowak.

Writing (original draft): Melissa Zhao.

Writing (editing and review): Melissa Zhao, Anirban Maitra, Jonathan Nowak.

All authors have read and approve this final manuscript.

Declaration of interests

A.M. is listed as an inventor on a patent relevant to pancreatic cancer early detection that has been licensed by Johns Hopkins University to Thrive Earlier Detection and serves as a consultant for Tezcat Biosciences. J.A.N. receives research funding from Natera and serves as a consultant to Leica Biosystems. M.Z. has no competing interests to declare.