Clinical next-generation sequencing successfully applied to fine-needle aspirations of pulmonary and pancreatic neoplasms




Next-generation sequencing was performed on pulmonary and pancreatic fine-needle aspirations (FNAs) and on paired FNAs and resected primary tumors from the same patient.


DNA was isolated in formalin-fixed, paraffin-embedded cell blocks from 16 pulmonary FNAs, 23 pancreatic FNAs, and 5 resected pancreatic primary tumors. Next-generation sequencing was performed for 4561 exons of 287 cancer-related genes and for 47 introns of 19 genes on indexed, adaptor-ligated, hybridization-captured libraries using a proprietary sequencing system (the Illumina HiSeq 2000).


Genomic profiles were generated successfully from 16 of 16 (100%) pulmonary FNAs, which included 14 nonsmall cell lung cancers (NSCLCs) and 2 small cell lung cancers (SCLCs). The NSCLC group included 6 adenocarcinomas, 5 squamous cell carcinomas, and 3 NSCLCs not otherwise specified. Genomic profiles were successfully obtained from 23 of 23 (100%) pancreatic FNAs and from 5 of 5 (100%) matched pancreatic primary tumors, which included 17 ductal adenocarcinomas, 3 mucinous adenocarcinomas, 2 adenocarcinomas NOS, and 1 neuroendocrine tumor. Eighty-one genomic alterations were identified in the 16 pulmonary FNAs (average, 5.1 genomic alterations per patient); and the most common genomic alterations were TP53, RB1, SOX2, PIK3CA, and KRAS. Eighty-seven genomic alterations were identified in the 23 pancreatic tumor FNAs (average, 3.8 genomic alterations per patient); and the most common genomic alterations were KRAS, TP53, CDKN2A/B, SMAD4, and PTEN. Among the pancreatic tumors, there was 100% concordance of 20 genomic alterations that were identified in 5 patient-matched FNA and surgical primary tumor pairs.


The authors were able to perform next-generation sequencing reliably on FNAs of pulmonary and pancreatic tumors, and the genomic alterations discovered correlated well with those identified in matched resected pancreatic tumors. Cancer (Cancer Cytopathol) 2013;121:688–694. © 2013 American Cancer Society.


After it achieved widespread adoption in Europe, the fine-needle aspiration (FNA) biopsy was introduced and achieved common use in the United States in the early 1980s.[1-3] The FNA method became particularly attractive for the diagnosis of deep-seated tumor masses that otherwise would have required open procedures to obtain biopsy material.[1-3] In particular, because of patient safety considerations, FNA has continued to be an important method for obtaining a diagnosis in masses that involve the lung[4-7] and pancreas.[8-13] For pulmonary neoplasms, such as nonsmall cell carcinoma (NSCLC) and small cell undifferentiated carcinoma (SCLC), the risk of causing a significant pneumothorax when an incisional or larger bore, rigid-needle biopsy has favored the use of the safer FNA technique. For pancreatic tumors, the risk of causing acute pancreatitis or serious bleeding has similarly favored using the FNA method for selected high-risk individuals.

Notably, FNA became a standard diagnostic approach before the ever increasing number of genomic alterations started driving the use of targeted therapy in practice or in clinical trials for these and other common cancers. For example, the development of targeted therapies for lung cancer based on the presence of a therapy-specific biomarker has highlighted the need to perform molecular diagnostic tests on multiple genes from a single tumor sample to determine the optimum therapy.[14-17] However, the small size of a tumor sample obtained using the FNA technique has challenged the ability to obtain adequate biomarker coverage for therapy decision making when multiple single-gene hotspot diagnostic tests are done sequentially, because the limited sample is consumed rapidly by each successive diagnostic test.[18] For pancreatic tumors, a major change in the approach to both solid and cystic masses has been the introduction of image-guided, endoscopic-based FNA.[10-13] This also recently has become an issue for patients with pancreatic cancer diagnosed using the FNA technique given the increasing number of biomarker-based clinical trials now open for patients with newly diagnosed pancreatic cancer who seek the opportunity to receive treatment with novel agents.[19-21]

In the last 3 years, the next-generation sequencing (NGS) approach has been introduced to greatly enhance the capability of performing multiple genomic tests on a single, small, formalin-fixed, paraffin-embedded (FFPE) tumor specimen, including core-needle biopsies.[22-25] In addition to allowing clinicians to obtain clinically useful information for therapy selection by sequencing hundreds of genes simultaneously, some methods also allow for a complete assessment of all 5 classes of genomic alterations that are known to impact human malignancies (base substitution, short insertion and deletion, gene amplification, homozygous deletion, and gene rearrangement). We performed comprehensive genomic profiling using an NGS-based assay on a series of clinical pulmonary FNA samples, including both NSCLC and SCLC, and on a series of pancreatic FNA samples, including ductal, mucinous, and neuroendocrine tumors; and, in a subset of cases, we compared the pancreatic FNA-based genomic alterations identified with results obtained from matched primary tumor resections from the same patients.


This study was approved by the Institutional Review Board of Albany Medical Center. A retrospective, nonconsecutive series of FNAs was obtained from 16 patients with pulmonary tumors and from 23 patients with pancreatic tumors; and, after initial fixation in CytoLyt (Hologic, Inc., Marlborough, Mass), the FNA material was processed into FFPE cell blocks using a standard procedure.[26] For 5 patients with pancreatic tumors, a matched FFPE sample from a Whipple resection specimen also was obtained. All pulmonary and pancreatic tumors were evaluated by a single pathologist (J.S.R.) to confirm that a minimum of 20% of nucleated cells that were forwarded for DNA extraction were malignant. Each sample contained a minimum of 15,000 total benign and malignant cells and yielded the minimum 50 ng of DNA necessary to perform the DNA sequencing assay. DNA was isolated without microdissection or macrodissection from unstained FFPE sections of FNA cell blocks using 40 μm total sections for pulmonary tumors and 40 μm total sections for pancreatic tumors. Paired normal tissues or blood samples were not available to formally exclude the possibility that some of the genomic alterations identified could have arisen from contaminating nontumor material. DNA sequencing was performed for 4561 exons of 287 cancer-related genes and for 47 introns of 19 genes that frequently are rearranged in cancer on indexed, adaptor-ligated, hybridization-captured libraries using the Illumina HiSeq 2000 (Illumina, Inc., San Diego, Calif). The tumor samples were evaluated for genomic alterations, including base substitutions, short insertions and deletions, amplifications, homozygous deletions, and gene rearrangements, as previously described.[25] Actionable genomic alterations were defined as those that had an impact on anticancer therapies currently on the market or that could be used as guides to direct patients toward registered clinical trials. A combination of publicly available and newly validated analysis tools (Foundation Medicine, Inc., Cambridge, Mass) was used to analyze the data and assign DNA alteration calls.


Genomic profiles were successfully generated from 16 of 16 (100%) pulmonary FNA cases, which included 14 NSCLCs and 2 SCLCs. Samples were sequenced to a median coverage depth of 931X for the pulmonary tumors, 416X for the pancreatic FNAs, and 717X for the pancreatic primary tumors. The NSCLC group included 6 adenocarcinomas, 5 squamous cell carcinomas, and 3 NSCLCs not otherwise specified (NOS). Genomic profiles were similarly obtained from 23 of 23 (100%) pancreatic FNAs and from 5 of 5 (100%) matched pancreatic primary tumors, which included 18 ductal adenocarcinomas, 2 mucinous adenocarcinomas, 2 adenocarcinomas NOS, and 1 pancreatic neuroendocrine tumor. In total, 81 genomic alterations were identified in the 16 combined NSCLC and SCLC FNAs (range, 2-9 genomic alterations per patient; average, 5.1 genomic alterations per patient) (Fig. 1A) obtained from 8 men (50%) and 8 women (50%) with a median age of 69 years (age range, 48-84 years). The most common alterations were observed in tumor protein 53 (TP53) (88%), retinoblastoma 1 (RB1) (37.5%), sex-determining region Y-box 2 (SOX2) (25%), phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit α (PIK3CA) (25%), and v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) (19%). Of the 3 RB1 mutations, 1 was identified in an SCLC, and 2 were observed in pulmonary adenocarcinomas. Eighty-eight percent of the combined NSCLC and SCLC patients (14 of 16) harbored an actionable alteration, including alterations in v-raf murine sarcoma viral oncogene homolog B1 (BRAF), cyclin D1 (CCND1), epidermal growth factor receptor (EGFR), v-erb-b2 erythroblastic leukemia viral oncogene homolog 2, neuroblastoma/glioblastoma derived oncogene homolog (avian) (ERBB2), fibroblast growth factor receptor 1 (FGFR1), KRAS, myeloid cell leukemia sequence 1 (BCL2-related) (MCL), neurofibromin 1 (NF1), PIK3CA, phosphatase and tensin homolog (PTEN), and serine/threonine kinase 11 (STK11).

Figure 1.

Tile plot of genomic alterations in pulmonary and pancreatic fine-needle aspiration (FNA) samples. (A) Genomic alterations in 16 pulmonary FNA samples. AC indicates adenocarcinoma; SCC, squamous cell carcinoma; NSCLC NOS, nonsmall cell lung cancer not otherwise specified; SCLC, small cell undifferentiated lung cancer. (B) Genomic alterations on 23 pancreatic FNA samples. (C) Genomic alterations on 5 matched pancreatic FNA and surgical resection samples. Genomic profiles of both the FNA sample and the matched surgical resection specimen are identical in all 5 of these cases. AC, adenocarcinoma; Mucinous, mucinous AC; NOS, not otherwise specified; Neuroendo, neuroendocrine carcinoma.

In total, 87 genomic alterations were identified in the 23 pancreatic FNAs (range, 1-9 per patient; average, 3.8 per patient) (Fig. 1B) obtained from 10 men (43%) and 13 women (57%) with a median age of 66 years (age range, 43-88 years). Of the 23 sequenced pancreatic FNA samples, the most common alterations were observed in KRAS (78%), TP53 (74%), cyclin-dependent kinase inhibitors 2A and 2B (CDKN2A/B) (35%), SMAD family member 4 (SMAD4) (17%), and PTEN (13%) (Fig. 1B). Of the 18 ductal pancreatic adenocarcinomas, 15 (83%) harbored KRAS mutations, and 8 (44%) harbored CDKN2A alterations. For the 5 patients who had pancreatic tumors with matching tissue biopsy and surgical resection specimens, there was 100% concordance between the 20 total genomic alterations identified in both FNAs and tissue biopsy/resection specimens (Fig. 1C).

Examples of completely concordant pancreatic tumor FNAs and surgical resection specimens are provided in Figure 2A,B. For case PAFM32 (Fig. 2A), both the FNA specimen and the matched surgical resection specimen had the same genomic alterations: KRAS_c.34G→C_p.G12R, TP53_c.548C→G_p.S183*. For case PAFM12 (Fig. 2B), both the FNA, which was originally interpreted as an adenocarcinoma based on the presence of signet ring cells, and the matched Whipple resection specimen, which revealed a primary pancreatic neuroendocrine tumor (that was confirmed by positive immunostaining with synaptophysin and chromogrannin) with lymph node metastasis, featured the same genomic profile, which included an alteration in the STK11 gene: STK11_c.894C→A_p.F298L. An example of a pulmonary FNA that yielded an interesting genomic alteration is provided in case LFM25 (Fig. 3). This pulmonary squamous cell carcinoma harbored a base substitution in the targetable ERBB2 (HER2) gene (S310F), which is a known activation mutation and demonstrates in vitro sensitivity to Erbb2 inhibition.[27]

Figure 2.

(A) Case PAFM32: ductal adenocarcinoma of the pancreas. The fine-needle aspiration (FNA) specimen on the left demonstrates cohesive clusters of adenocarcinoma associated with a tumor diathesis featuring necrosis and acute inflammation. The matched surgical resection specimen on the right demonstrates a moderately differentiated pancreatic ductal adenocarcinoma invading the wall of the duodenum. Both the FNA sample and the surgical resection specimen exhibit the same genomic alterations: KRAS_c.34G→C_p.G12R, TP53_c.548C→G_p.S183*.(B) Case PAFM12: neuroendocrine tumor of the pancreas. FNA sample shown on the left was processed into a formalin-fixed, paraffin-embedded cell block. Signet ring cell features were originally interpreted as adenocarcinoma. The Whipple resection specimen shown on the right revealed a neuroendocrine tumor with lymph node metastases. Both the FNA sample and the surgical resection specimen exhibited the same genomic alterations: STK11_c.894C→A_p.F298L.

Figure 3.

Case LFM25: primary squamous cell carcinoma of the lung with a v-erb-b2 erythroblastic leukemia viral oncogene homolog 2, neuroblastoma/glioblastoma derived oncogene homolog (avian) (ERBB2 [HER2]) mutation. This pulmonary fine-needle aspiration sample was obtained from an unresectable primary squamous cell carcinoma in a man aged 68 years. The next-generation sequencing assay identified a potentially targetable S310F base substitution in the ERBB2 (HER2) gene involving the extracellular domain of the ERBB2 (HER2) protein.


Despite the introduction of the image-guided, spring-loaded, core-biopsy technique, the FNA biopsy remains a commonly used diagnostic procedure in patients with pulmonary and pancreatic masses. In addition, this technique has been adopted to endoscopic procedures, and endobronchial ultrasound and transgastroduodenal FNA approaches are now used frequently to diagnose pulmonary tumors and pancreatic tumors, respectively.[5, 8, 11-13] However, given the modern need to both diagnose and obtain a sample adequate for biomarker testing, genomic procedures have had to be customized to yield more and more detailed therapy-directing information on smaller and smaller tumor samples. One particular need has been the requirement that critical biomarkers for selecting therapy, such as EGFR sequencing and EML4:ALK fusion testing, must be obtained successfully on 1 biopsy procedure for patients with NSCLC.[14-17] Because a wide variety of additional actionable genomic alterations, including gene fusions (ROS1, RET), also must be determined for patients with NSCLC, the current study was undertaken to confirm that FNA would be able to generate sufficient DNA for accurate and sensitive, clinical-grade, and therapy-directing genotyping.[26, 28-30] It should be emphasized that performing biomarker studies, including NGS, on FNA samples is adapted to and requires the use of the FFPE cell blocks that are created from the aspirated material at the time of biopsy. In a previous study that applied targeted NGS to small NSCLC samples, the simultaneous detection of multiple mutations in various genes in a single test was compared with commonly used real-time polymerase chain reaction (PCR) methods to detect base substitutions (mutations) in the EGFR, KRAS, and BRAF genes.[23] The results from that study indicated that NGS-based genomic profiling was more sensitive than conventional PCR-based sequencing methods, supporting the use of targeted NGS for the screening of EGFR, KRAS, and BRAF mutations in FFPE tissue material.[23]

For pancreatic cancer, and especially for ductal adenocarcinomas, to date, the emerging targeted therapy approach has not led to US Food and Drug Administration drug approvals, although numerous clinical trials are ongoing.[31-34] However, anecdotal experience suggests that patients with pancreatic cancer who have selected genomic alterations that are associated with clinical benefit in other disease indications, such as ERBB2 (HER2) gene amplification and the receipt of anti-HER2–targeted therapies, also may benefit from those alterations.[35, 36] Thus, pancreatic cancer FNA samples likewise must be handled carefully, allowing for comprehensive genomic profiling to determine potential unanticipated or novel targets of therapy.

In this study, the feasibility of performing a sensitive and specific, clinical-grade, NGS-based genomic profiling assay on FNA samples was tested, and the results indicated that the test could readily be run on the these small tumor tissues. However, 1 real concern in assaying such small specimens is the potential for intratumor heterogeneity in the primary disease and intertumor heterogeneity in the metastatic disease. Although heterogeneity clearly exists in many tumor types, a growing body of data has demonstrated that the frequency of heterogeneity is much greater in passenger mutations than in the genomic alterations driving the disease and that, when heterogeneity occurs in driver alterations within the same tumor, convergent evolution can occur and can generate different driver alterations with the same functional consequence. In addition, more sensitive and quantitative methodologies like NGS, which has been optimized to generate high, uniform coverage, have a greater capacity to detect genomic alterations when subclonality occurs. In 5 of our pancreatic FNA cases, the generated genomic profile demonstrated high concordance with the resected tissue from the same patient, with 20 of 20 (100%) total genomic alterations identified in both specimens. Thus, based on these results, we can conclude that FNA is an acceptable sample for determining the presence of potential targets of therapy in patients with pulmonary and pancreatic tumors and that the extremely small size of these samples will not limit the ability to obtain a sensitive, specific, and comprehensive genomic profile of these tumors sufficient for use as a director of therapy selection.


Foundation Medicine supported this study and performed the next-generation sequencing procedures described in the article. Dr. Ross has received grant support from Foundation Medicine.


Geneva Young, Kai Wang, Je He, Geoff Otto, Matthew Hawryluk, Zac Zwirco, Tina Brennan, Michelle Nahas, Amy Donahue, Roman Yelensky, Doron Lipson, Philip J. Stephens, Vincent A. Miller, and Jeffrey S. Ross are employees of Foundation Medicine and hold stock in this commercial entity. Christine E. Sheehan and Ann B. Boguniewicz are not Foundation Medicine employees or stock holders.