Abstract
Introduction
Current clinical practice guidelines recommend molecular profiling for patients with pancreatic ductal adenocarcinoma (PDAC). However, endoscopic ultrasound–guided small specimens are often insufficient for next‐generation sequencing (NGS). In this study, the impact of a standardized protocol designed to optimize tissue acquisition for NGS was evaluated.
Methods
Quality assurance data from patients with suspected PDAC who underwent endoscopic ultrasound–guided tissue acquisition for NGS was retrospectively analyzed. A standardized workflow incorporating rapid onsite evaluation using predefined adequacy criteria, dedicated tissue procurement for cell block preparation, multidisciplinary communication, and prospective monitoring of molecular outcomes were evaluated using a successful NGS result as the primary endpoint.
Results
Of the 61 patients with confirmed PDAC by cytology, 49 specimens (80.3%) were deemed adequate for NGS during rapid onsite evaluation. Forty‐one specimens with a median tumor fraction of 30% underwent NGS. Median DNA and RNA yields were 20.8 ng/µL and 19.7 ng/µL, respectively. DNA‐based NGS was successful in 39 cases (95.1%), whereas RNA‐based fusion analysis succeeded in 22 cases (53.7%). Median turnaround time from specimen receipt to NGS reporting was 10 business days.
Conclusions
A standardized molecular adequacy protocol achieved high DNA sequencing success but did not substantially improve overall specimen adequacy compared with prior institutional experience. Intrinsic biologic features of PDAC, including low tumor cellularity and abundant desmoplastic stroma and nonneoplastic elements remain the principal barriers to molecular profiling. Future efforts that focus on improving tumor enrichment and optimizing RNA preservation may expand precision oncology access to patients with PDAC.
Keywords: comprehensive genomic profiling, molecular adequacy, next‐generation sequencing, pancreatic ductal adenocarcinoma, quality assurance, rapid on‐site evaluation, tissue acquisition
Short abstract
Implementing a standardized molecular adequacy protocol can achieve high sequencing success for DNA‐based next‐generation sequencing in patients with pancreatic ductal adenocarcinoma (PDAC). However, intrinsic biologic features of PDAC, including low tumor cellularity and abundant desmoplastic stroma and nonneoplastic elements, remain the principal barriers to molecular profiling and limit overall specimen adequacy for precision oncology.
INTRODUCTION
Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies, with a 5‐year survival rate of approximately 13%. 1 , 2 The increasing adoption of precision oncology has made comprehensive genomic profiling (CGP) and predictive biomarker testing an essential component of PDAC management. Current guidelines from the National Comprehensive Cancer Network recommend molecular profiling for all patients with PDAC, to identify potentially actionable alterations and eligibility for targeted therapies and clinical trials (Pancreatic Adenocarcinoma. National Comprehensive Cancer Network. Updated 2026‐03‐18., n.d.).
Endoscopic ultrasound (EUS)–guided tissue acquisition is the primary method for obtaining diagnostic material in patients with suspected PDAC. Although EUS‐guided fine‐needle aspiration (FNA) and fine‐needle biopsy (FNB) are highly effective for establishing a diagnosis, obtaining sufficient material for biomarker testing, including next‐generation sequencing (NGS) and immunohistochemical assays remain challenging. PDAC specimens are frequently characterized by extensive desmoplastic stroma, low tumor cellularity, necrosis, and admixture with benign pancreatic tissue, all of which limit tumor enrichment and molecular testing success. 3 , 4 , 5 These limitations have restricted the clinical utilization of EUS‐guided specimens for molecular profiling in routine practice, despite evidence that approximately 18% of patients undergoing CGP harbor potentially actionable mutations amenable to personalized therapeutic approaches. 6
A recent systematic review and meta‐analysis of 20 studies reported a pooled NGS adequacy rate of 89.9% for EUS‐guided specimens; however, when strict preanalytical screening criteria were applied, adequacy dropped to 79.7%, revealing substantial heterogeneity across studies and clinical settings. 7 Other studies have reported even lower success rates for CGP, with one study demonstrating an overall FoundationOne CDx success rate of only 67.9% from EUS‐guided specimens. 3 The broad heterogeneity of molecular sufficiency of PDAC specimens seen in the literature can be largely ascribed to the variations in testing practices. These include the ability to perform rapid onsite evaluation (ROSE) for adequacy assessment, use of FNB samples for formalin‐fixed paraffin‐embedded (FFPE) blocks, implementing direct smears in addition to FFPE cell blocks for NGS testing, tumor mapping for tumor enrichment, the ability to use microdissection to isolate tumor‐rich areas from the nontumor background, sensitivity of sequencing assay used for testing, and the molecular adequacy criteria required for the sequencing assay.
Some studies have reported that DNA yields from FNB samples are significantly higher with superior molecular testing adequacy compared to EUS‐FNA specimens of pancreatic lesions. 8 In a prior institutional study by our group, adequacy rates for biomarker testing and NGS in EUS‐acquired PDAC specimens were suboptimal in cell block preparations. 9 However, the ability to combine EUS‐FNA cytology smears with FFPE cell blocks allowed for improved molecular adequacy rates (85% vs 42%, respectively). As molecular profiling has become increasingly integrated into routine patient care for patients with PDAC, developing a standardized approach to tissue acquisition and specimen triage that would better support downstream molecular testing has emerged as a critical quality improvement objective at our institution.
To address this challenge, our institution implemented a standardized operating procedure designed to optimize collection of material for both NGS and immunohistochemical‐based biomarker testing. The protocol incorporated ROSE with specific molecular adequacy criteria, prospective communication between surgical, endoscopy, and cytopathology teams, and additional dedicated tissue acquisition for cell block preparation. The objective of this study was to evaluate the effectiveness of this protocol and determine whether implementation of a molecular adequacy‐focused workflow could improve downstream biomarker testing outcomes in patients with PDAC.
METHODS
Patient cohort
This study represents a retrospective analysis of prospectively collected quality assurance data from patients undergoing EUS‐guided tissue acquisition for suspected PDAC with an NGS‐based testing request at our institution between December 2023 and September 2025. Cases without an NGS test request were excluded. The study was approved by the institutional review board.
Before study initiation, a standardized institutional protocol was implemented for tissue acquisition intended for molecular profiling. Molecular test requests were placed prospectively by the referring clinical team, including gastrointestinal medical and surgical oncology through the institutional electronic medical record using a standardized order set, generally within two weeks before the procedure. If a prospective molecular test request was not located within the patient’s electronic medical record, an NGS request was placed by the gastrointestinal endoscopy team, at the time of the procedure. A peripheral blood sample was also collected for paired tumor/normal NGS for enhanced discrimination of germline events, as per institutional testing protocols.
EUS‐guided procedure and specimen handling
All EUS‐guided procedures were performed using the standardized institutional protocol that was implemented for tissue acquisition. EUS procedures were performed using a linear echoendoscope and FNA sampling was performed first using 25‐gauge needles. The aspirate was expressed on glass slides to prepare direct smears. Both air‐dried slides for Diff‐Quik staining and ethanol‐fixed, Papanicolaou‐stained slides were prepared and used to evaluate sample adequacy by ROSE. During ROSE, cytopathologists evaluated not only diagnostic adequacy, but also estimated molecular adequacy using predefined criteria, including tumor cellularity, tumor fraction, and overall specimen quality. Adequate specimens were defined as direct smears containing approximately 10,000 tumor cells with a tumor fraction of at least 20% and minimal contamination by benign or inflammatory elements. Tumor cellularity assessments were performed by visual estimation by the reviewing cytopathologist. If adequate tumor was identified on direct smears, two to three additional FNA passes (in rare cases) or a 22‐gauge core biopsy needle (FNB) (preferred option) were performed specifically for cell block generation to support downstream molecular and immunohistochemical testing. Adequate core sampling was defined as two visible cores in the collection media. The FNA passes and FNB cores were collected in Roswell Park Memorial Institute medium, transported immediately to the cytopathology laboratory, and processed by cytocentrifugation for a cell block. The choice of needle type, the total number of passes, and selection of FNB versus FNA passes were determined by the endoscopist on a case‐by‐case basis depending on lesion characteristics and procedural considerations.
Molecular testing by next‐generation sequencing
All cases were evaluated for adequacy for requested biomarkers and cell block sections and/or smear preparations were tumor mapped and sent to the molecular laboratory for testing, as described previously. 10 , 11 (Figure 1) Genomic profiling was performed using DNA‐ and RNA‐based NGS assays. The DNA‐based assay employs hybridization capture targeting a 2.1‐megabase region of the hg19 reference genome to detect single nucleotide variants and insertions/deletions (610 genes), copy number alterations (583 genes), selected gene rearrangements (34 genes), as well as microsatellite instability and tumor mutational burden, 12 , 13 whereas the RNA‐based assay uses targeted amplicon sequencing to detect clinically relevant gene fusions across 51 genes. 14 , 15 In brief, nucleic acid was extracted from cell block sections and/or smear preparations following macro‐/microdissection. DNA extracted from tumor specimens and paired normal (peripheral blood) controls underwent library preparation with unique molecular identifier adapters, followed by hybridization capture enrichment and bidirectional paired‐end sequencing on the Illumina NovaSeq 6000. For RNA‐based fusion analysis, cDNA was amplified using fusion‐targeted primers and aligned to a synthetic fusion reference genome for detection by coverage‐based methods. Somatic variants were identified using an in‐house bioinformatics pipeline incorporating dual‐duplex molecular barcoding for consensus error correction and comparison with matched normal controls. For interpretation and reporting, the following analytical limits of detection were used: ≥5% variant allele frequency for SNVs and indels; ≥5 fusion‐supporting molecules for DNA‐based rearrangements; and ≥6 copies for gene amplifications; ≥500,000 mapped cDNA target reads with an average read length ≥60 base pairs for RNA‐based fusions.
FIGURE 1.

Tumor mapping cytology smears and cell block sections for tumor enrichment and microdissection for tissue extraction for next‐generation sequencing.
Quality assurance
Cases were prospectively monitored as part of a quality assurance initiative, with subsequent review of specimen adequacy, NGS performance, and biomarker testing outcomes. The primary outcome was NGS success, defined as the generation of a clinically reportable genomic profiling result. NGS failure was defined as the inability to produce a reportable result due to suboptimal specimen characteristics (e.g., insufficient tumor cellularity, inadequate nucleic acid yield or quality) or technical sequencing limitations. DNA‐ and RNA‐based sequencing outcomes were evaluated separately, given their distinct analytic requirements and failure modes. Performance metrics were reviewed periodically, and protocol refinements including adjustments to adequacy thresholds and specimen handling practices were implemented as part of an ongoing quality improvement process.
RESULTS
A total of 65 patients underwent EUS‐guided procedures under the molecular adequacy protocol for clinically suspected PDAC during the study period. Of these, 61 were confirmed to have PDAC on final diagnosis and were included in the final analysis (Table 1). Four cases were excluded, including three intraductal papillary mucinous neoplasms with high‐grade dysplasia and one neuroendocrine carcinoma.
TABLE 1.
Clinicopathologic characteristics of patients with pancreatic ductal adenocarcinoma who underwent molecular testing by next‐generation sequencing.
| Characteristic | Value |
|---|---|
| Patient characteristics (n = 61) | |
| Age, years, median (range) | 68 (41–85) |
| Sex, n (%) | |
| Male | 37 (61) |
| Female | 24 (39) |
| ROSE for Molecular Adequacy (n = 61) (%) | |
| Adequate | 49 (80.3) |
| Inadequate | 12 (19.7) |
| Specimens submitted for NGS (n = 41) (%) | |
| Cell block only | 18 (43.9) |
| Smears only | 13 (31.7) |
| Both | 10 (24.4) |
| NGS performance (n = 41) (%) | |
| DNA‐sequencing success | 39 (95.1) |
| Cell block only | 18 |
| Smears only | 13 |
| Both | 8 |
| RNA‐sequencing success | 22 (53.7) |
| Molecular parameters, median (range) | |
| Tumor fraction (%) | 30 (20–90) |
| DNA yield (ng/µL) | 20.8 (0.8–118.4) |
| RNA yield (ng/µL) | 19.7 (2.4–198.0) |
Abbreviations: NGS, next‐generation sequencing; ROSE, rapid onsite evaluation.
Using the molecular adequacy criteria as outlined previously, 49 of 61 cases (80.3%) were deemed adequate for downstream molecular testing during ROSE. Twelve cases (19.7%) did not meet the minimum 20% tumor cellularity for NGS, either because of insufficient tumor cellularity or abundance of desmoplastic stroma and nonneoplastic elements and were considered inadequate for molecular testing (Figure 2). Of these 12 inadequate cases, in which the molecular test request was canceled, three cases were subsequently successfully sequenced using the surgical resection specimen, whereas one patient underwent liquid‐based testing for genomic profiling.
FIGURE 2.

Pancreatic fine‐needle aspiration (FNA) specimens evaluated for molecular adequacy demonstrating scant tumor cellularity representing inadequate specimens. (A) Diff‐Quik–stained smears, 4×; (B) Papanicolaou‐stained smears, 40×; and (C) cell block preparation (hematoxylin and eosin), 20× showing predominantly benign pancreatic parenchymal tissue.
Of the 49 cases deemed adequate, 41 EUS‐guided specimens were ultimately tested using NGS (Figure 3). Median tumor fraction was 30% (range, 20%–90%) with a median DNA yield of 20.1 ng/µL (range, 0.8–188.4 ng/µL) and a median RNA yield of 17 ng/µL (range, 0–198 ng/µL) (Table 2). DNA‐based sequencing was successful in 39 cases (95.1%); two cases failed because of suboptimal quality control metrics. In contrast, the RNA‐based sequencing demonstrated lower performance, with successful fusion analysis in 22 cases (53.7%). The remaining cases failed either due to suboptimal RNA yield (<10 ng/µL) or failed NGS quality control metrics. Mutations in KRAS G12D were most commonly detected (51%), followed by KRAS G12V (26%) and G12R (5%), frequently co‐mutated with TP53 (84% cases). One case detected a fusion in DMD:::BRAF (Table 2). Median turnaround time from specimen receipt in the laboratory to NGS reporting was 10 business days (range, 7–19 days).
FIGURE 3.

Pancreatic fine‐needle aspiration (FNA) specimens evaluated for molecular adequacy demonstrating adequate specimens. (A) Papanicolaou–stained smears, 40×; (B) cell block preparation (hematoxylin and eosin), 20×; and (C) IGV representation of tumor and normal paired next‐generation sequencing demonstrating a mutation in KRAS p.G12D.
TABLE 2.
Next‐generation sequencing (NGS)‐based analysis of pancreatic adenocarcinoma specimens (n = 39).
| Case | Tumor % | Testing material | DNA (ng/µL) | RNA (ng/µL) | Driver mutation detected (DNA‐based) | Fusion detected (RNA‐based) | ||
|---|---|---|---|---|---|---|---|---|
| KRAS | TP53 | Other alterations | ||||||
| 1 | 60 | Sm | 28.5 | <10 | KRAS p.G12D | TP53 p.V274A | ‐ | QNS |
| 2 | 40 | CB | 38 | 14 | KRAS p.G12V | TP53 p.? | ‐ | None |
| 3 | 50 | Sm | 10.9 | 20.9 | KRAS p.G12D | TP53 p.R213* | ‐ | None |
| 4 | 20 | CB | 3.05 | 2.4 | KRAS p.G12D | ‐ | ‐ | QNS |
| 5 | 90 | CB | 3.02 | 3.84 | KRAS p.G12D | TP53 p.R248W | CDKN2A p.R58* | QNS |
| 6 | 25 | CB + Sm | 21.5 | 5.02 | ‐ | TP53 p.R335fs* | SMAD4 p.R361H | QNS |
| 7 | 25 | CB | 4.97 | 3 | KRAS p.G12D | TP53 p.R282W | SMAD4 p.E330K | QNS |
| 8 | 20 | Sm | 6.39 | 4.09 | KRAS p.G12V | TP53 p.H168P | SMAD3 p.R80W | QNS |
| 9 | 30 | Sm | 26.2 | 16.8 | KRAS p.G12V | ‐ | SMAD3 p.G2595 | None |
| 10 | 60 | CB | 17.8 | 6.29 | KRAS p.G12D | ‐ | SMAD4 E538fs* | QNS |
| 11 | 55 | Sm | 46.3 | 22.3 | ‐ | TP53 p.Y220S | BRAF p.N486_P490del | QNS |
| 12 | 25 | Sm | 73.2 | 172 | KRAS p.G12V | TP53 p.V274G | ‐ | None |
| 13 | 25 | Sm | 13.7 | 16.6 | ‐ | ‐ | GNAS p.R201C; DMD::BRAF fusion | None |
| 14 | 60 | Sm | 14.9 | 19.1 | KRAS p.G12V | TP53 p.A159P | ‐ | None |
| 15 | 30 | CB + Sm | 15.70 | 56.8 | ‐ | CDKN1A p.R84Q; APC p? | None | |
| 16 | 25 | CB | 39.9 | 21 | KRAS p.G12V | TP53 p.Y234D | ‐ | None |
| 17 | 20 | CB + Sm | 32.4 | 3.33 | KRAS p.G12D | TP53 p? | ‐ | QNS |
| 18 | 25 | CB + Sm | 22.8 | 20.3 | KRAS p.G12D | TP53 p.R273H | ‐ | None |
| 19 | 20 | Sm | 7.32 | 8.89 | ‐ | ‐ | STK40 p.L264P; CTNNB1 p.A622del | QNS |
| 20 | 80 | CB | 1.83 | 5.52 | KRAS p.G12D | TP53 p.F113V | SMAD4 p.Q334* | QNS |
| 21 | 30 | CB | 37 | 5.24 | KRAS p.G12D | TP53 p.R248W | ‐ | QNS |
| 22 | 30 | CB | 35.2 | 10.8 | KRAS p.G12V | TP53 p.V147fs* | SMAD4 p.H92D | None |
| 23 | 70 | CB | 100.8 | 60.6 | KRAS p.G12V | TP53 p.L161F | ‐ | None |
| 24 | 21 | CB + Sm | 26.1 | 30.3 | KRAS p.G12V | TP53 p.G302fs* | ‐ | None |
| 25 | 21 | CB + Sm | 34.3 | 10.27 | KRAS p.G12D | TP53 p.M237I; TP53 p.E258D | ‐ | None |
| 26 | 25 | CB | 30.3 | 19 | KRAS p.G12R | TP53 p.P152R | SMAD4 p.A85delinsV* | None |
| 27 | 50 | CB | 10.9 | 0 | KRAS p.G12D | TP53 p.R306* | ‐ | QNS |
| 28 | 50 | Sm | 14 | 5.24 | KRAS p.G12D | ‐ | FBXW7 p.R465H | QNS |
| 29 | 21 | Sm | 0.793 | 5.5 | ‐ | TP53 p.R337L | GATA2 p.L305V | QNS |
| 30 | 21 | Sm | 6.89 | 4.29 | KRAS p.G12R | TP53 p.P151T | SMAD4 p.G352V; SMAD4 p? | QNS |
| 31 | 40 | CB | 33.8 | 198 | KRAS p.G12D | TP53p. K132R | ‐ | None |
| 32 | 60 | CB | 118.4 | 111.2 | KRAS p.G12D | TP53 p.W53* | ‐ | QNS |
| 33 | 60 | CB | 10.4 | 47.6 | KRAS p.G12D | TP53 p.V272L | MTAP (loss) CDKN2A/B (loss) | None |
| 34 | 21 | CB + Sm | 3.9 | 47.6 | KRAS p.G12V | TP53 p.R342* | ‐ | None |
| 35 | 60 | CB | 46.9 | 80 | KRAS p.G12D | ‐ | SMAD4 p.Q448fs* | None |
| 36 | 80 | CB | 38.3 | 17.2 | KRAS p.G12D | TP53 p.H214R | ‐ | None |
| 37 | 80 | Sm | 20.1 | 100 | KRAS p.G12D | TP53 p.R248Q | SMAD4 p.R361P | None |
| 38 | 70 | CB | 3.51 | 10.3 | KRAS p.G12D | TP53 p.W53* | ‐ | None |
| 39 | 60 | CB + Sm | 17.3 | 29.2 | ‐ | TP53 p.I195T | ERBB2 p.R678Q; NRAS p.Q61R ; | None |
| 40 | 70 | CB + Sm | 14.4 | 46.9 | QNS | QNS | ||
| 41 | 20 | CB + Sm | 32.9 | 30.4 | QNS | QNS | ||
Abbreviations: CB, cell block; QNS, quantity not sufficient; Sm, smears.
The remaining eight cases that were deemed adequate during ROSE but were not tested included five cases where NGS was performed on a prior biopsy from an outside institution or a subsequent surgical resection; therefore, testing on the cytology specimen was canceled (one patient without consent for testing; one patient who opted for alternative liquid biopsy testing; and one test was canceled because of laboratory issues).
DISCUSSION
The increasing adoption of precision oncology has made biomarker testing with NGS‐based genomic profiling a key component of PDAC management. However, there are major challenges for the molecular testing success of EUS‐guided small specimens, such as FNA and FNB samples. These specimens include an inherent low tumor cellularity due to extensive desmoplastic stroma, necrosis, and admixture with benign pancreatic tissue, limiting the ability for tumor enrichment and being able to meet cellular adequacy criteria for NGS analysis.
Recent systematic reviews and meta‐analyses report broad heterogeneity in molecular testing success rates. Pan et al. reported a pooled adequacy of 79.7% (95% CI, 67.3–90.0) across 20 studies of pancreatic malignancies, 7 whereas Ko et al. found a pooled sequencing success rate of 83.9% (95% CI, 75.8–89.7). 6 Buchberg et al. reported a technical yield of only 48% when applying stringent quantitative criteria (minimum 3‐mm2 tissue, 2‐mm2 tumor area, 20% relative tumor area). 16 The variability across studies reflects differences in sampling technique, testing methodologies, and molecular adequacy criteria and underscore the need for a standardized approach to tissue acquisition and specimen triage that would better support downstream genomic profiling.
The primary objective of this study was to improve specimen adequacy of EUS‐guided FNA/FNB specimens for NGS and biomarker testing through implementation of a standardized molecular adequacy protocol. The protocol incorporated predefined adequacy criteria during ROSE, dedicated collection of material (either dedicated FNA passes and/or FNB samples) for cell block preparation, and prospective monitoring of molecular outcomes. Although the adequacy assessment during ROSE was performed on FNA smears, the cell block preparation represented a pooled specimen, combining needle rinses from the initial FNA passes (with ROSE) as well as the additional FNA/FNB passes (without ROSE). The goal of this tissue collection protocol was based on the assumption that additional passes for the cell block would enhance the overall and tumor cellularity, provided the specimen met the predefined ROSE adequacy criteria. However, if the subsequent dedicated passes contained predominantly nonneoplastic elements (such as stroma, peripheral blood, or inflammatory cells), they could potentially dilute the overall tumor cellularity of the cell block, even in cases where the initial ROSE was deemed adequate on smears.
Our results show that an adequate sample for molecular testing was collected in 80.3% of sampled cases with a high success rate of 95.1% for DNA‐based NGS. These results align with published literature that report high molecular testing success rates when an adequate sample is collected. 17 ROSE is widely used during EUS‐guided procedures, primarily to assess cytomorphologic adequacy but not specifically to evaluate molecular sufficiency. 18 Al Zoughbi et al. reported that ROSE adequacy assessment predicted NGS success in majority of procedures, with success rates of 93% in specimens considered “adequate” versus 32% in inadequate groups. 19 In contrast, another study by Dbouk et al. reported that adequacy for genomic evaluation was similar between cases with and without ROSE (88.6% vs 85.5%), and that ROSE required significantly more procedure time and needle passes. 18 These findings suggest that although ROSE may inform molecular adequacy, its overall benefit may vary depending on individual practice settings, sampling technique, and molecular testing methodologies.
The results from our study indicate that while the molecular adequacy protocol was successfully implemented and resulted in high DNA sequencing success among the submitted specimens, overall adequacy rates remained remarkably similar to those observed in a previous study at our institution. Notably, the prior study evaluated molecular adequacy by a smaller panel of 50 genes using an amplicon‐based NGS assay that had an input requirement of 10 ng of DNA. In contrast, the current NGS assay used in our molecular laboratory is a custom hybrid capture‐based NGS assay interrogating 610 genes and requiring higher input DNA (40–50 ng). Therefore, although the overall success rates of the DNA‐based NGS remain largely similar between the two studies, the input requirements for the current study are significantly higher. Based on this, it is possible that the success rates of the current cohort of specimens could potentially be higher if a subset of cases could qualify for an assay with lower input requirements.
The main conclusions based on our results is that the principal barrier to successful biomarker testing in PDAC may not be related to tissue acquisition technique alone, but rather the intrinsic biology of the disease. PDAC specimens frequently contain limited neoplastic cellularity surrounded by abundant desmoplastic stroma, inflammatory cells, necrosis, and benign pancreatic tissue. As a result, even technically successful sampling procedures often yield material with insufficient tumor fraction with minimal ability to tumor‐map for enrichment to meet molecular testing thresholds. These biologic constraints create challenges for NGS‐based biomarker assessment and likely explain the persistent rate of inadequate specimens despite protocol‐driven optimization.
Our findings also provide insight into the ongoing discussion regarding FNA versus FNB for molecular testing. Multiple studies have demonstrated improved tissue yield and DNA concentration with FNB. Kandel et al. reported mean DNA concentrations of 5.93 versus 3.365 µg/mL for FNB versus FNA, whereas Mohamed et al. found that 71.4% of FNB samples were adequate for molecular testing versus 32.1% for FNA. 8 , 20 The updated 2026 National Comprehensive Cancer Network Pancreatic Adenocarcinoma Guidelines now recommend core biopsy for molecular profiling and suggest performing two extra needle passes beyond diagnostic passes for future NGS if needed (Pancreatic Adenocarcinoma. National Comprehensive Cancer Network. Updated 2026‐03‐18., n.d.). Our experience at our institution suggests that FNB may not necessarily provide a significant advantage for DNA‐based NGS when adequately cellular direct smears are available for testing. Molecular adequacy of direct smears can be evaluated during ROSE, and smear preparations may sometimes be more amenable to tumor mapping. In addition, cytology smears provide high‐quality substrates for molecular testing and avoid formalin fixation artifacts that can negatively affect nucleic acid quality. The primary advantage of FNB in our workflow may instead be related to generation of cell blocks for ancillary testing. Cell block material provides a useful substrate for immunohistochemical biomarker evaluation and may facilitate additional studies that are difficult to perform on smears alone. Consequently, the relative value of FNB should be considered in the context of the intended downstream testing platform rather than assuming a universal advantage for NGS.
An important observation from this study was the marked difference between DNA and RNA sequencing performance. Although DNA sequencing was successful in >95% of submitted specimens, RNA‐based fusion testing failed in 46.3% of cases. This finding likely reflects the greater susceptibility of RNA to degradation during specimen collection and processing and highlights an important area for future workflow optimization. This finding may have important clinical implications, as RNA‐based sequencing is superior for detecting gene fusions 21 (Pancreatic Adenocarcinoma. National Comprehensive Cancer Network. Updated 2026‐03‐18., n.d.). Michuda et al. reported that addition of RNA‐sequencing to DNA‐sequencing led to a 24% increase in patients eligible for matched therapies, with 29.1% of fusions detected only through RNA‐sequencing versus 4.8% only through DNA‐sequencing. 22 Therefore, for RNA‐based testing, additional optimization including dedicated RNA preservation protocols and RNAse‐free workflows, immediate specimen processing, and use of RNA‐stabilizing collection media may be required for improving test success rates. 23 , 24
This study is limited by its retrospective, single‐institution design, which introduces potential selection bias and limits generalizability. In a high‐volume tertiary cancer center, institutional expertise in specimen acquisition, processing, and molecular testing may not reflect community practice settings. Additionally, the use of pathologist estimated tumor cellularity as a quality metric for NGS specimen adequacy has inherent limitations. Zarabi et al. demonstrated poor correlation between pathologist estimated tumor cellularity and NGS‐derived variant allele fraction in lung and colorectal cancer patient specimens, with R 2 values < .25. 25 Similarly, a joint consensus guideline by the Association for Molecular Pathology and the College of American Pathologists noted that estimation of tumor percentages based on H&E review may have significant interobserver variability with underestimation of tumor cellularity, as nonneoplastic cells such as inflammatory infiltrates and endothelial cells may remain inconspicuous. 26 This underscores the need for objective quantification methods to replace subjective estimation of tumor cellularity by light microscopy alone. 26 Adoption of digital pathology tools for more accurate and consistent tumor estimation, rapid nucleic acid quantification, or automated cell counting may improve real‐time specimen triage, reduce interobserver variability and unnecessary specimen rejections to improve access to actionable genomic results. 27 , 28 , 29 , 30 Finally, although an association between specimen adequacy and NGS success was demonstrated, this study did not formally evaluate procedural and technical variables, such as needle type, number of passes, lesion characteristics, and operator experience in a multivariable model. As such, the independent contribution of these factors to sequencing outcomes remains undefined, and future prospective studies incorporating these variables are needed to optimize EUS‐FNA workflows for molecular testing.
In conclusion, implementation of a standardized molecular adequacy protocol incorporating ROSE‐guided specimen triage, dedicated tissue procurement, and prospective quality monitoring did not substantially improve adequacy rates compared with our prior institutional experience. Approximately one‐fifth of PDAC specimens remained unsuitable for comprehensive molecular testing despite optimized collection practices. These findings suggest that the major limitation to successful biomarker testing in PDAC is the intrinsic nature of the specimens themselves, including low tumor cellularity, abundant stromal tissue, and admixture with benign pancreatic elements. These factors continue to limit tumor enrichment and contribute to molecular testing failures and quantity‐not‐sufficient outcomes. Although FNB‐derived cell blocks may provide advantages for immunohistochemical biomarker testing, our data do not support a clear advantage of FNB over adequately cellular direct smears for NGS‐based molecular testing. Comparative studies evaluating FNB versus FNA in the context of comprehensive genomic profiling encompassing both DNA and RNA analysis would further inform optimal sampling strategies. 9 , 18 , 31 Future efforts should focus on improving tumor enrichment strategies, developing objective adequacy assessment tools, and optimizing workflows for RNA preservation and comprehensive biomarker testing in limited cytology specimens. Finally, development of institutional protocols incorporating quantitative adequacy thresholds aligned with national guideline recommendations may facilitate broader implementation of precision oncology for patients with PDAC. 32
AUTHOR CONTRIBUTIONS
Yan Gao: Writing—original draft; writing—review and editing; investigation; formal analysis; data curation. Camille Canty: Writing—review and editing; data curation. Laura R. Prakash: Data curation; writing—review and editing. Emmanuel Coronel: Writing—review and editing; investigation. Matthew H. G. Katz: Writing—review and editing. Rebecca A. Snyder: Writing—review and editing; investigation. Sinchita Roy‐Chowdhuri: Conceptualization; investigation; writing—original draft; writing—review and editing; project administration; supervision.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
ACKNOWLEDGMENTS
This study did not receive any funding.
REFERENCES
- 1. Siegel RL, Kratzer TB, Wagle NS, Sung H, Jemal A. Cancer statistics, 2026. CA Cancer J Clin. 2026;76(1):e70043. doi: 10.3322/caac.70043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Stoop TF, Javed AA, Oba A, et al. Pancreatic cancer. Lancet. 2025;405(10485):1182‐1202. doi: 10.1016/S0140-6736(25)00261-2 [DOI] [PubMed] [Google Scholar]
- 3. Doi T, Ishiwatari H, Ohike N, et al. Impact of tumor necrosis on comprehensive genomic profiling success in pancreatic cancer: a retrospective study. Jpn J Clin Oncol. 2026;56(1):48‐54. doi: 10.1093/jjco/hyaf156 [DOI] [PubMed] [Google Scholar]
- 4. Goswami RS, Luthra R, Singh RR, et al. Identification of factors affecting the success of next‐generation sequencing testing in solid tumors. Am J Clin Pathol. 2016;145(2):222‐237. doi: 10.1093/ajcp/aqv023 [DOI] [PubMed] [Google Scholar]
- 5. Ikeda, G , Hijioka, S , Takahashi, Y , et al. Tumor cell count and total cell count as surrogate indicators of comprehensive genomic profiling success in pancreatic cancer specimens collected by endoscopic ultrasonography‐guided tissue acquisition. J Hepato Biliary Pancreat. 2026;70106(7):576‐583. doi: 10.1002/jhbp.70106 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ko SW, Jo IH, Yoon SB. Feasibility and clinical utility of endoscopic ultrasound‐guided tissue acquisition for comprehensive genomic profiling in pancreatic cancer: a systematic review and meta‐analysis. Pancreatology. 2025;25(1):89‐97. doi: 10.1016/j.pan.2024.12.012 [DOI] [PubMed] [Google Scholar]
- 7. Pan Y, Ran T, Zhang X, et al. Adequacy of EUS–guided fine‐needle aspiration and fine‐needle biopsy for next‐generation sequencing in pancreatic malignancies: a systematic review and meta‐analysis. Endosc Ultrasound. 2024;13(6):366‐375. doi: 10.1097/eus.0000000000000097 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Mohamed WT, Jahagirdar V, Jaber F, et al. Endoscopic ultrasound‐guided fine‐needle biopsy versus aspiration for tissue sampling adequacy for molecular testing in pancreatic ductal adenocarcinoma. Cancers. 2024;16(4):761. doi: 10.3390/cancers16040761 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Gan Q, Roy‐Chowdhuri S, Duose DY, et al. Adequacy evaluation and use of pancreatic adenocarcinoma specimens for next‐generation sequencing acquired by endoscopic ultrasound–guided FNA and FNB. Cancer Cytopathology. 2022;130(4):275‐283. doi: 10.1002/cncy.22533 [DOI] [PubMed] [Google Scholar]
- 10. Roy‐Chowdhuri S, Goswami RS, Chen H, et al. Factors affecting the success of next‐generation sequencing in cytology specimens. Cancer Cytopathol. 2015;123(11):659‐668. doi: 10.1002/cncy.21597 [DOI] [PubMed] [Google Scholar]
- 11. Roy‐Chowdhuri S, Stewart J. Preanalytic variables in cytology: lessons learned from next‐generation sequencing‐the MD Anderson experience. Arch Pathol Lab Med. 2016;140(11):1191‐1199. doi: 10.5858/arpa.2016-0117-RA [DOI] [PubMed] [Google Scholar]
- 12. Dai M, San Lucas FA, Alvarez H, et al. Unraveling the nexus: tumor mutational burden, PD‐L1 expression, and oncogenic alterations in non‐small cell lung cancer cytology specimens. Cancer Cytopathol. 2026;134(4):e70090. doi: 10.1002/cncy.70090 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Yang RK, Alvarez H, Lucas AS, et al. Microsatellite instability and high tumor mutational burden detected by next generation sequencing are concordant with loss of mismatch repair proteins by immunohistochemistry. Cancer Genet. 2025;290–291:44‐50. doi: 10.1016/j.cancergen.2024.12.002 [DOI] [PubMed] [Google Scholar]
- 14. Ramani NS, Patel KP, Routbort MJ, et al. Factors impacting clinically relevant rna fusion assays using next‐generation sequencing. Arch Pathol Lab Med. 2021;145(11):1405‐1412. doi: 10.5858/arpa.2020-0415-OA [DOI] [PubMed] [Google Scholar]
- 15. Ramani NS, Chen H, Broaddus RR, et al. Utilization of cytology smears improves success rates of RNA‐based next‐generation sequencing gene fusion assays for clinically relevant predictive biomarkers. Cancer Cytopathol. 2021;129(5):374‐382. doi: 10.1002/cncy.22381 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Buchberg J, De Stricker K, Pfeiffer P, Mortensen MB, Detlefsen S. Mutational profiling of 103 unresectable pancreatic ductal adenocarcinomas using EUS‐guided fine‐needle biopsy. Endosc Ultrasound. 2024;13(3):154‐164. doi: 10.1097/eus.0000000000000072 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Dong S, Agarunov E, Fasullo M, et al. Somatic mutational analysis in endoscopic ultrasound–guided biopsy of pancreatic adenocarcinoma: assessing yield and impact. Am J Gastroenterol. 2024;119(8):1636‐1639. doi: 10.14309/ajg.0000000000002786 [DOI] [PubMed] [Google Scholar]
- 18. Dbouk M, Davis BG, Peller M, et al. EUS‐guided fine‐needle biopsy sampling of solid pancreatic masses with and without rapid onsite evaluation for commercial next‐generation genomic profiling. Gastrointest Endosc. 2025;102(6):855‐862.e1. doi: 10.1016/j.gie.2025.03.1208 [DOI] [PubMed] [Google Scholar]
- 19. Al Zoughbi W, Kim D, Alperstein SA, et al. Incorporating cytologic adequacy assessment into precision oncology workflow using telepathology: an institutional experience. Cancer Cytopathology. 2021;129(11):874‐883. doi: 10.1002/cncy.22441 [DOI] [PubMed] [Google Scholar]
- 20. Kandel P, Nassar A, Gomez V, et al. Comparison of endoscopic ultrasound‐guided fine‐needle biopsy versus fine‐needle aspiration for genomic profiling and DNA yield in pancreatic cancer: a randomized crossover trial. Endoscopy. 2021;53(4):376‐382. doi: 10.1055/a-1223-2171 [DOI] [PubMed] [Google Scholar]
- 21. Chakravarty D, Johnson A, Sklar J, et al. Somatic genomic testing in patients with metastatic or advanced cancer: ASCO provisional clinical opinion. J Clin Orthod. 2022;40(11):1231‐1258. doi: 10.1200/JCO.21.02767 [DOI] [PubMed] [Google Scholar]
- 22. Michuda J, Park BH, Cummings AL, et al. Use of clinical RNA‐sequencing in the detection of actionable fusions compared to DNA‐sequencing alone. J Clin Orthod. 2022;40(16_suppl):3077. doi: 10.1200/JCO.2022.40.16_suppl.3077 [DOI] [Google Scholar]
- 23. Anderson MA, Brenner DE, Scheiman JM, et al. Reliable gene expression measurements from fine needle aspirates of pancreatic tumors. J Mol Diagn. 2010;12(5):566‐575. doi: 10.2353/jmoldx.2010.090107 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Sura GH, Engel KB, Greytak SR, et al. Pre‐analytical best practices for RNA sequencing from small biopsies and cytologic specimens. Mol Diagn Ther. 2026;30(3):441‐461. doi: 10.1007/s40291-026-00837-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Zarabi SK, Zhai L, Cheng Y‐W. A challenging correlation between tumor cellularity and somatic variant allele fraction in lung and colorectal cancers—specimens of low tumor percentage should be analyzed with caution. Biomolecules. 2024;14(2):168. doi: 10.3390/biom14020168 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Jennings LJ, Arcila ME, Corless C, et al. Guidelines for validation of next‐generation sequencing–based oncology panels. J Mol Diagn. 2017;19(3):341‐365. doi: 10.1016/j.jmoldx.2017.01.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Lewis JE, Shebelut CW, Drumheller BR, et al. An automated pipeline for differential cell counts on whole‐slide bone marrow aspirate smears. Mod Pathol. 2023;36(2):100003. doi: 10.1016/j.modpat.2022.100003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Sakamoto T, Furukawa T, Pham HHN, et al. A collaborative workflow between pathologists and deep learning for the evaluation of tumour cellularity in lung adenocarcinoma. Histopathology. 2022;81(6):758‐769. doi: 10.1111/his.14779 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Sah S, Chen L, Houghton J, et al. Functional DNA quantification guides accurate next‐generation sequencing mutation detection in formalin‐fixed, paraffin‐embedded tumor biopsies. Genome Med. 2013;5(8):77. doi: 10.1186/gm481 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Cornish TC. Clinical application of image analysis in pathology. Adv Anat Pathol. 2020;27(4):227‐235. doi: 10.1097/PAP.0000000000000263 [DOI] [PubMed] [Google Scholar]
- 31. Razzano D, Bouza SJ, Hernandez PV, et al. Comprehensive molecular profiling of pancreatic ductal adenocarcinoma in FNA, biopsy, and resection specimens. Cancer Cytopathol. 2022;130(9):726‐734. doi: 10.1002/cncy.22589 [DOI] [PubMed] [Google Scholar]
- 32. Sohal DPS, Kennedy EB, Cinar P, et al. Metastatic pancreatic cancer: ASCO guideline update. J Clin Orthod. 2020;38(27):3217‐3230. doi: 10.1200/JCO.20.01364 [DOI] [PMC free article] [PubMed] [Google Scholar]
