Diagnostic Value of Next-Generation Sequencing (NGS) and Microarray in Characterizing Tumor Origin: A Challenging and Educational Case

Introduction

A solitary malignant pancreatic lesion in a patient with a history of lung adenocarcinoma always raises the question of whether the lesion is a primary pancreatic carcinoma or a metastasis from the lungs. Molecular studies, including next-generation sequencing (NGS) and microarray, are often the best methodology in determining tumor classification or typing when clinical and histologic correlations, as well as immunohistochemical characterization, fail to reach definite conclusions. NGS allows for rapid sequencing of the entire genome or targeted genomic regions, providing a comprehensive and detailed view of genetic variations or anomalies, including mutations and rearrangements. NGS-based comprehensive genomic profiling (CGP) technology is commercially available. The one used in our study examines 324 cancer genes in solid tumors, and reports known and likely pathogenic short variants (SVs), copy number alterations (CNAs), and select rearrangements, as well as complex biomarkers, including tumor mutational burden (TMB) and microsatellite instability (MSI) [1,2]. In contrast, microarray is a high-throughput technique used to detect and measure the expression levels and variations of target genes in thousands of transcripts, providing crucial insights into the molecular landscape and a snapshot of the molecular makeup of a sample [3]. The resulting data are representative of the cumulative expression of all cell types found in the sample. The microarray assay used in the present case uses real-time RT-PCR to measure the expression of approximately 90 genes in a tumor sample. It classifies the tumor by comparing the gene expression profile to a comprehensive database of over 2000 known tumor types and subtypes. The test reports the cancer type with the highest probability match and provides a list of tumor types that can be ruled out with 95% confidence. The pros and cons of these two methods in tumor diagnosis and classification are under continuous exploration. This report describes the case of an 81-year-old woman with a history of resected right-lung adenocarcinoma, presenting with a solitary nodule in the head of the pancreas. NGS and immunocharacterization of the tumor were most consistent with metastasis of lung adenocarcinoma, whereas the microarray analysis favored the gastrointestinal tract to be the tumor origin, and the possibility of lung metastasis was largely excluded.

Case Report

An 81-year-old woman with a history of surgically resected right-lung adenocarcinoma in December 2016, followed by adjuvant chemotherapy, presented with recurrence in a mediastinal lymph node in July 2021. She received radiation and chemotherapy and showed excellent cancer response, with no evidence of further recurrence or metastasis on surveillance until July 2024. A computed tomography (CT) scan in November 2024 revealed no suspicious mass in the lungs, but showed a new 16-mm nodule in the head of pancreas with mild extrahepatic bile duct dilation to 10 mm in caliber towards the porta hepatis, and abrupt caliber change from dilated to non-dilated duct in the midportion (Figure 1A). Positron emission tomography-computed tomography (PET CT) revealed the nodule was hypermetabolic with SUV of 10.2 (Figure 1B). No suspicious hypermetabolism was identified elsewhere.

This pancreas lesion was biopsied via endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA), indicating moderately differentiated adenocarcinoma (Figure 2A, 2B). It showed overlapping histologic morphology with the lung adenocarcinoma from 2016 (Figure 2C, 2D), although the comparison study was limited by the aspiration nature of the FNA specimen. Characterization by immunohistochemical stains showed the tumor cells were positive for CK7, CA19.9, and CDX2 (weak and patchy), and negative for CK20 (Figure 3A-3D). Lung immunohistochemical marker of TTF-1 mostly showed non-specific cytoplasmic staining (Figure 3E), but napsin A was positive in tumor cells (Figure 3F). This immunostaining profile, although not entirely specific, was consistent with metastasis from the lungs, but not primary pancreatic adenocarcinoma. NGS analysis was performed on this pancreatic lesion FNA material and revealed mutations in KRAS G13C, STK11 splice site 920+1G>T, and KEAP1 V155F.

Due to the solitary presence of this pancreatic lesion without other metastasis, as well as the patient’s desire, a Whipple procedure was performed. On the resected pancreas, a 2.1-cm moderately-to-poorly-differentiated invasive adenocarcinoma was identified on pathology evaluation, which involved the pancreatic head and 2 of 21 lymph nodes. Similar to the EUS-FNA biopsy, the tumor again showed morphology histologically overlapping with the patient’s prior adenocarcinoma of the lung (Figure 4A, 4B). The tumor material was sent for microarray analysis utilizing real-time RT-PCR for tumor typing and was classified as “indeterminate,” as the tumor showed “an expression profile that is not sufficiently similar to a tumor type in the reference database.” However, “several tumor types in the reference database” did show “some gene expression profile similarity to the sample and cannot be excluded from consideration.” These included “gastrointestinal adenocarcinoma (67%), pancreaticobiliary (25%), among others.” Unexpectedly, the probability of lung adenocarcinoma was reported to <5%.

Discussion

Various ancillary methodologies are available for determining tumor origin, with variable diagnostic specificity and strength. Histology comparison, if available, is a convenient and practical way to determine the cancer type. However, carcinoma does not always have histologic features specific to their origin. In our case, the pancreatic lesion showed histology overlapping with the lung adenocarcinoma diagnosed in 2016, but this morphology alone is not sufficient and reliable for definite conclusions on tumor origin.

Immunohistochemical staining for lineage markers is more specific for identification and classification of carcinomas of uncertain primary site [4]. Its simplicity and effectiveness have made it an essential and often first-line tool in practice. Thyroid transcription factor 1 (TTF1) and napsin A are the most frequently used immunohistochemical markers for confirming pulmonary lineage. TTF-1 is a nuclear transcription factor that promotes embryogenic pulmonary differentiation and is expressed by most but not all lung adenocarcinomas [5]. The sensitivity of TTF1 decreases in poorly-differentiated (~50%) compared with well-differentiated (~100%) lung adenocarcinomas [6]. Napsin A is a valuable complementary and alternative stain to TTF-1 and is believed to be more specific for pulmonary adenocarcinoma than TTF-1 [7,8]. As most markers, the expressions of TTF-1 and napsin A are not 100% specific to lung lineage. TTF-1 was reported to be positive in 2% of pancreatic adenocarcinomas [8], and both TTF-1 and napsin A have been reported in a subset of cholangiocarcinoma [9]. However, napsin A positivity has not been reported in pancreatic adenocarcinoma [10,11]. In the present patient’s pancreatic tumor, although TTF-1 was negative, napsin A positivity favors it to be metastatic lung carcinoma over primary pancreatic adenocarcinoma. Additionally, the pancreatic lesion showed a staining pattern (positive for napsin A and negative for nuclear TTF-1) similar to the metastatic adenocarcinoma in the mediastinal 4R lymph node, further supporting the tumor origin as lung over pancreas primary carcinoma.

Gene alterations are important in understanding the molecular basis of carcinoma and provide valuable insights into tumor profiling. KRAS (Kirsten rat sarcoma), EGFR (epidermal growth factor receptor), BRAF (v-raf murine sarcoma viral oncogene homolog B), ALK (anaplastic lymphoma kinase), and MET are the most common mutated genes in lung adenocarcinoma. Up to 30% of non-small cell lung carcinoma (NSCLC) harbor a mutation in the KRAS oncogene, making KRAS the most commonly detected oncogenic driver in lung cancer, with the most frequent mutations involving a substitution in codon 13 or 12 [12]. Although KRAS mutations are the most common mutations found in non-small cell lung cancer, they are also the most frequent gain-of-function alterations in cancers overall, including pancreas cancer. Four canonical mutations – KRAS (~85%), TP53 (60–70%), CDKN2A (>50%), and SMAD4 (~50%) – are commonly detected in primary pancreatic adenocarcinomas [13]. Studies indicate that KRAS mutation is an early event present in pancreatic intraepithelial neoplasia (PanIN), and acquisition of mutations in CDKN2A, TP53, and SMAD4 are associated with PanIN progression and the development of invasive adenocarcinoma. Among KRAS mutant adenocarcinomas, a tissue-specific pattern of codon mutation is observed. For example, G12 mutations are the most common in pancreatic adenocarcinoma, but these mutations are rarely found in lung cancer [14]. KRAS G13C mutation, as identified in the present patient’s tumor, is more frequently found in lung cancer (3%) than in pancreatic adenocarcinoma (less than 1.2%) [13,15,16].

KRAS-mutant lung tumors have distinct co-mutational profiles [17]. In a recent analysis of 1078 KRAS-mutant NSCLCs, 53% of them have at least 1 additional genomic alteration. The most common mutations involved are TP53 (39%), STK11 (20%), and KEAP1 (13%). This was consistent with a separate study that used RNA sequencing to identify 3 distinct common co-mutation clusters: (1) STK11, often with KEAP1 co-mutation, (2) TP53, and (3) CDKN2A/B inactivation with low TTF-1 [18].

The NGS analysis of the pancreas lesion FNA material in the present case revealed mutations, including KRAS G13C, STK11 splice site 920+1G>T, and KEAP1 V155F. This unique co-mutation pattern further supports that the pancreatic lesion was lung metastasis rather than pancreatic primary, consistent with immunohistochemistry. Surprisingly, although not entirely conclusive, the microarray data on RNA expression profile done on the resected tumor specimen almost excluded the possibility of pulmonary origin (<5%) and instead favored gastrointestinal and pancreaticobiliary adenocarcinoma. This result seems to conflict with the NGS data.

Choosing between NGS and microarray in cancer typing is an interesting topic. Studies have shown that when metastasis occurs, the tumor cells travel from their origin tissue to target tissues and adapt there through a complex cascade of molecular alterations [19]. While primary tumors overall remain metabolically similar to the normal origin tissues in their expression profiles, studies have shown that they show metabolic adaptation, including altered fatty acid metabolism and changes in energy supply, suggesting different genomic rewiring during growth from metastases [20,21]. Robust studies by Sanghiv et al revealed that the metastatic tumors of several cancer types demonstrated an RNA-seq expression profile closer to their target tissue than their origin tissue when compared to the primary tumors [22]. This study suggested that the metastatic tumor may show transcriptomic adaptations specific toward their target tissue due to environmental pressures like hypoxia or treatment exposure. Microarray detects RNA expression profiles and can reflect specific alterations by the microenvironment of the target organ, together with that of the origin tumor. In the present case, this could be the reason for the “indeterminate result,” as well as listing gastrointestinal and pancreaticobiliary as the most likely primary origins by microarray analysis. In contrast, driver mutations, once detected by NGS in the primary tumor, will stay mutated and unaffected on the DNA level by the microenvironment that the tumor metastases to. From this prospective, NGS might be superior to microarray in confirming cancer typing and origin. Of course, microarray still provides valuable genetic information when NGS is inconclusive.

Conclusions

Differentiating between primary versus metastasis from the lung can be challenging in an isolated pancreatic adenocarcinoma in a patient with a history of lung adenocarcinoma. Histologic comparison is most easily applied, but often is unreliable in reaching a definite conclusion. Immunohistochemical study is accessible to most institutions and is most frequently used in tumor classification. However, a clear diagnosis is not always possible when the immunostaining pattern is not typical, as most of the markers do not have 100% specificity for specific organs. NGS and microarray are both powerful technologies for understanding genetic variations and functions. They provide extensive genetic information by detecting genetic mutations and expressions, respectively, but they can lead to inconsistent conclusions in rare cases, as demonstrated in this report. Driver gene mutations, which can be detected by NGS when present, may be more reliable in predicting tumor origin/type. The advantages and disadvantages of each of these two methods are under active discussion, and correlations with clinical case studies will enhance further research.

Data Availability Statement

The data supporting the findings of this case report are securely stored in the electronic medical records of Cooper University Hospital and consist of patient-specific information. Due to privacy and ethical restrictions, the data are not publicly available. Further inquiries can be directed to the corresponding author.