Introduction

Pancreatic acinar cell carcinomas (PACCs) are rare forms of pancreatic cancer, accounting for approximately 1–2% of pancreatic tumors in adults and 15% of those in children1. PACCs are more prevalent in men, with a male to female ratio of 2:1. Histologically, these neoplasms are characterized by tumor cells displaying features of acinar cell differentiation, with moderate amounts of eosinophilic cytoplasm containing periodic acid-Schiff (PAS)-positive diastase-resistant zymogen granules and uniform nuclei with distinct nucleoli, arranged in acinar, glandular, solid, or trabecular patterns1. Genomically, PACCs have a repertoire of alterations distinct from that of pancreatic ductal carcinomas, rarely displaying KRAS, TP53, CDKN2A, and SMAD4 somatic alterations, and more frequently harboring recurrent activating fusions affecting BRAF, RET, RAF1, and NTRK1/2/3, which are present in up to 30% of patients2,3,4,5,6,7. In addition, a subset of patients with PACCs harbor germline pathogenic alterations affecting DNA damage response (DDR)/homologous recombination (HR) DNA repair-related genes, including ATM, BRCA1, BRCA2, and PALB22,3,8.

The mainstay of treatment for patients with PACCs is surgical resection with negative margins followed by chemotherapy. Currently, no targeted therapies are approved by the US Food and Drug Administration specifically for patients with PACC. Patients whose tumors carry oncogenic fusions may be treated using therapies with tumor-agnostic labels, such as entrectinib or larotrectinib for NTRK fusions or selpercatinib for RET fusions9,10,11. Inhibitors of poly (ADP-ribose) polymerase (PARP) have been approved for patients with pancreatic cancer (including PACC) harboring germline BRCA1 or BRCA2 pathogenic mutations, and further studies are investigating the benefit of PARP inhibitors in pancreatic cancers with HR deficiency (HRD) due to alterations affecting other canonical HR-related genes (NCT04858334), and in unselected PACC (NCT05286827).

The nibrin gene (NBN, also known as NBS1 [Nijmegen breakage syndrome 1]) is an integral component of the DDR/HR DNA repair machinery12,13. Germline NBN loss-of-function (LOF) alterations have not only been causally implicated in the Nijmegen breakage syndrome, but have also been shown to result in an increased risk of multiple cancer types13. The impact of somatic pathogenic NBN alterations on the development and progression of pancreatic cancer development is uncertain.

In this report, we describe a patient with PACC whose tumor harbored both an activating SEL1L::NTRK1 fusion and somatic biallelic inactivation of NBN. Upon disease progression on treatment with folinic acid, fluorouracil, irinotecan, and oxaliplatin (FOLFIRINOX), a novel reversion mutation restoring the NBN reading frame was detected, suggesting a potential link between NBN LOF, PACC progression, and the evolution of therapeutic resistance to mainstay treatment. This novel reversion mutation affecting the NBN gene adds to the growing list of resistance mechanisms to DNA damaging- and DDR-targeted therapies.

Results

Clinical case

A 65-year-old White male patient presented with abdominal pain. Computed tomography showed a mild distension of the common bile duct, and further evaluation with an endoscopic ultrasound revealed a mass in the pancreatic head. Fine needle aspiration of this mass was performed, and cytology was consistent with adenocarcinoma (Fig. 1a). Staging evaluation indicated pT1c, pTN2, pM0 (AJCC 8th edition14), and the patient underwent pancreaticoduodenectomy. Pathologic examination revealed an acinar cell carcinoma measuring 2 cm with varied architecture, including solid trabecular and acinar formations that invaded the duodenum and intrahepatic common bile duct, with 4 of 16 regional lymph nodes involved. At the cellular level, cytoplasms were granular with monomorphic nuclei containing prominent nucleolei. Consistent with the diagnosis, immunohistochemical analysis were negative for synaptophysin, chromogranin, and MOC-31/EpCAM, while periodic acid-Schiff staining with diastase digestion (PAS-D) were positive for cytoplasmic granules. Given the rarity of the tumor type, the diagnosis was confirmed and agreed upon by multiple pathologists. Furthermore, the screening tissue used to determine molecular eligibility was retrospectively found to be BCL-10 positive consistent with the diagnosis1.

Fig. 1: Case study of a 65-year-old male diagnosed with pancreatic cancer who developed a reversion mutation in NBN after more than 1 year on FOLFIRINOX.
figure 1

a Therapy history and CA19-9 kinetics. Representative micrographs of the primary resection and metastatic liver biopsy specimens revealed the presence of acinar structures comprising neoplastic cells with eosinophilic and granular cytoplasm, and atypical nuclei, consistent with a diagnosis of pure acinar cell carcinoma of the pancreas (scale bar: 20 µm). Representative CT scans pre- and on-treatment with FOLFIRINOX. b NBN reversion mutation detected in the metastatic liver biopsy. The reference (top), primary mutation (middle), and secondary mutation (bottom) cDNA and amino acid sequences are shown. The combination of the primary and secondary mutations created a 3-amino acid deletion and 2-amino acid insertion resulting in a novel in-frame indel (c.1146_1155delinsGAAATCA; p.383_385delinsKS). c B allele frequency/log R ratio plots derived from WGS analysis of the metastatic liver biopsy. d Genomic and transcriptomic structure of the SEL1L::NTRK1 fusion. Skipping of NTRK1 exon 7 was observed in the targeted RNA sequencing data, reflective of the coding potential of different splice configurations. e Chromosome-based circos plot derived from WGS of the metastatic liver biopsy. f Distribution of mutational signatures as determined by WGS of the metastatic liver biopsy. CA cancer antigen, cDNA complementary DNA, CT computed tomography, FOLFIRINOX folinic acid, fluorouracil, irinotecan, and oxaliplatin, indel insertion and deletion, Mb megabase, WGS whole genome sequencing.

Full size image

Subsequently, the patient received adjuvant treatment with gemcitabine and capecitabine for 6 months, followed by radiation concurrent with capecitabine. After completion of this treatment, the patient was on surveillance for 16 months, at which time he was diagnosed with metastatic disease to the liver, histopathologically confirmed as metastatic acinar cell carcinoma (Fig. 1a).

At that time, a tissue specimen from the primary pancreaticoduodenectomy (herein referred to as the primary resection) was sent for clinical-grade next generation sequencing (NGS) as per standard of care, which identified a somatic pathogenic NBN mutation and an in-frame SEL1L::NTRK1 fusion (Table 1). The patient then received palliative chemotherapy with the modified FOLFIRINOX regimen for 13 months. The patient achieved an initial response with a marked decrease in cancer antigen (CA)19-9 and a reduction in hepatic tumor burden, but treatment was eventually discontinued due to the development of a new liver lesion and an increase in the size of prior liver lesions (Fig. 1a). Given the presence of an NTRK1 fusion in the primary resection specimen, the patient received targeted treatment with the NTRK inhibitor larotrectinib. However, the patient rapidly progressed with multiple new liver lesions and an increase in the size of prior liver lesions, resulting in treatment discontinuation 2 months later.

Table 1 Summary of genomic alterations identified
Full size table

A new liver biopsy (herein referred to as the metastatic liver biopsy) was taken at this time and sent for clinical-grade NGS (Table 1). Pending the results and based on the detection of a pathogenic NBN mutation in the primary resection specimen, the patient was enrolled on TRESR (NCT04497116) and received a therapeutic dose level of the ataxia telangiectasia and Rad3-related (ATR) inhibitor camonsertib (RP-3500, Fig. 1a). The patient had stable disease at 6 weeks, but treatment was discontinued after 12 weeks due to progression of the target liver lesions (32% increase). The patient was then treated with gemcitabine and paclitaxel (nab-paclitaxel was not available due to a nationwide shortage), on which he also experienced progression of disease after 2 months due to worsening peritoneal carcinomatosis. Subsequently, he was briefly treated on a clinical trial with TST001, an anti-claudin 18.2 monoclonal antibody, before discontinuing the study due to clinical progression after which the patient was enrolled in a hospice and passed away 1 month later.

Genomic analysis

Genomic analyses of the tissue specimens obtained during the course of treatment were performed with clinical-grade NGS assays, as well as with research-use-only NGS assays performed as part of TRESR (NCT04497116; Table 1). This included the use of the Synthetic Lethal Interactions for Precision Diagnostics (SNiPDx™) panel, a novel NGS-based diagnostic panel specifically designed to detect and distinguish between monoallelic and biallelic LOF alterations in genes determined to be synthetic lethal with camonsertib15. The primary resection specimen collected 3 months after diagnosis was initially evaluated by clinical-grade NGS and retrospectively by SNiPDx™, and anchored multiplex polymerase chain reaction (PCR) (AMP)-based RNA sequencing (Table 1)15,16,17,18. The metastatic liver biopsy specimen, collected after treatment with FOLFIRINOX and larotrectinib and just prior to initiation of camonsertib, was initially evaluated by clinical-grade NGS, and retrospectively by SNiPDx™ and whole genome sequencing (WGS)15,19,20,21.

DNA analysis of both tissue specimens and peripheral blood mononuclear cells (PBMCs) by multiple methodologies identified a somatic (absent in PBMCs) pathogenic 1-nucleotide deletion in exon 10 of NBN (c.1146del; p.E383Kfs*21; ClinVar Variation ID: 822251; Table 1 and Fig. 1b). This deletion results in a premature stop codon 21 codons downstream of p.E383 and is predicted to truncate the protein leading to nonsense-mediated decay. Analysis of the allele-specific copy number by SNiPDx™ and WGS revealed copy-neutral loss of heterozygosity of the NBN wild-type allele with subsequent duplication of the mutant allele, consistent with a clonal somatic biallelic inactivation of NBN (Fig. 1c and Supplementary Fig. 1). In addition, a subclonal RB1 (c.649 C > T; p.Q217*) pathogenic mutation was detected in the primary tumor resection. No germline or somatic pathogenic alterations affecting TP53, ATM, BRCA1, BRCA2, PALB2, or other DDR/HR-related genes included in the assays were detected. No RET, BRAF, or RAF1 fusions were detected by WGS, targeted DNA sequencing or targeted RNA sequencing; however, we detected a complex structural rearrangement involving exon 1 of SEL1L and exon 8 of NTRK1 resulting in an expressed, in-frame, likely oncogenic SEL1L::NTRK1 fusion containing the NTRK1 kinase domain (Table 1 and Fig. 1d, e).

Novel NBN reversion mutation in the metastatic biopsy

A genomic comparison of the biopsies from the liver metastasis obtained upon progression on FOLFIRINOX and the primary tumor resection revealed a novel SMAD4 homozygous deletion, an MCL1 amplification, and a second frameshift mutation in NBN. This NBN mutation, which was detected only in the liver metastasis specimen, but absent in the primary resection specimen, resulted in an additional 2-nucleotide deletion (c.1154_1155del; p.K385fs*27) 8 nucleotides downstream of the initial somatic mutation (Table 1 and Fig. 1b). Visual inspection of the genomic sequencing reads revealed that the NBN mutations were in cis (data not shown) and predicted to re-establish the NBN coding frame.

Multimodal detection of NTRK1 rearrangement and fusion

Following progression on FOLRIFINOX, the patient received larotrectinib due to the presence of an NTRK1 fusion. The patient, however, rapidly progressed within 2 months of initiating therapy. Given the lack of response to NTRK-targeted therapy, we aimed to validate the presence of the NTRK1 fusion. Targeted RNA sequencing of the primary resection confirmed the presence of a SEL1L::NTRK1 fusion and further delineated the breakpoints for the fusion transcript to be at exon 1 of SEL1L and exon 8 of NTRK1, resulting in an in-frame fusion containing the NTRK kinase domain. To understand the genomic mechanisms underlying this rare molecular anomaly, we examined the genomic rearrangements identified by WGS of the metastatic liver biopsy. No direct interaction was identified between SEL1L and NTRK1, rather, both genes rearranged with the same intron of CENPC (intron 17) resulting in a “bridged fusion” between SEL1L intron 1 and NTRK1 intron 6 (Fig. 1d, e)22. These data demonstrate that the skipping of NTRK1 exon 7 and usage of the next proximal splice site, exon 8, led to productive expression of the fusion transcript. Thus, the combined analysis of targeted RNA sequencing and WGS elucidated both the genomic and post-transcriptional details of this rare molecular event. The lack of larotrectinib antitumor activity in this case cannot be explained by the absence of an in-frame NTRK1 fusion, suggesting that another mechanism of primary resistance may be involved.

Characterization of genomic signatures

To identify additional genomic features of the tumor, we performed WGS on the metastatic liver biopsy specimen only, as the available primary resection specimen had insufficient tumor fraction and residual DNA. This analysis revealed the presence of 8789 single nucleotide variants (SNVs), 319 insertions and deletions (indels; 161 insertions and 158 deletions), and 37 structural variants (16 translocations, 8 deletions, 8 inversions, and 5 tandem duplications; Fig. 1e). The tumor specimen was found to be microsatellite-stable with a tumor mutational burden of 3.0 mutations per megabase, and an estimated tumor ploidy of 1.7. The dominant mutational signatures were related to aging (signatures 1 [12%] + 5 [29%]), followed by HRD (signatures 3 [27%] + 8 [11%]; Fig. 1f). The tumor, however, displayed only a partial set of the genomic features consistent with HRD. Despite the dominant HRD-related single nucleotide substitution signature and genome-wide loss-of-heterozygosity patterns, the enrichment for deletions with microhomology was modest, and rearrangement signatures associated with HRD were not observed. Consistent with these findings, the HRDetect score was 0.24, well below the cutoff used to classify a tumor as HRD (0.7). These results are consistent with genomic scars present in cancers with LOF alterations in genes upstream in the DDR pathway23,24,25.

Discussion

PACCs have been shown to harbor DDR/HR DNA repair defects, but somatic loss of NBN has not been linked to the development of pancreatic cancers in general, or specifically to PACC. In this case study, we performed detailed multimodal genomic profiling of multiple biopsies from the same patient with PACC revealing a comprehensive genomic characterization of the tumor evolution during multiple lines of therapy.

Our findings provide circumstantial evidence to suggest that the somatic NBN biallelic mutation likely played a role in the development and/or progression of this PACC. First, the biallelic inactivation of NBN through somatic mutation and copy-neutral loss of heterozygosity was linked to genomic signatures consistent with a partial HRD profile, but lacking long indels with (micro)homology, the main driver of HRDetect scores. This is expected, given that HRDetect was trained on genomes from patients with BRCA1/BRCA2 LOF in breast cancer, whereas NBN plays a role further upstream of BRCA1/BRCA2 in the repair of DNA double-strand breaks12,23. Although the genomic “scars” present in canonical HR-related genes (eg, BRCA1, BRCA2, PALB2, RAD51C, RAD51D, and RAD51B) and their respective absence in cancers with alterations affecting DDR-related genes (eg, ATM and CHEK2) have been described, the spectrum of genomic signatures in tumors with LOF of less common HR/DDR genes, including NBN, has yet to be fully characterized. Our findings emphasize the importance of further studies characterizing the genomic scars stemming from LOF of specific HR/DDR genes, which may not only provide mechanistic insights about back-up DNA repair mechanisms operating in absence of these genes, but also novel therapeutic opportunities based on synthetic lethality approaches26.

Second, an additional alteration predicted to restore the NBN reading frame was detected following initial treatment and therapeutic vulnerability to FOLFIRINOX. We interpreted this to be a novel reversion mutation based on (1) the potential selective pressure of the platinum-containing regimen FOLFIRINOX; (2) the biallelic nature of the primary NBN mutation in the tumor; (3) the proximity; and (4) the cis nature of the second NBN mutation to the first; and (5) the potential to re-establish the coding frame in the absence of an intervening stop codon. Identification of an emergent NBN reversion mutation in the post-therapy specimen might suggest an adaptive resistance mechanism mediated by restoration of NBN function. Reversion alterations in BRCA1/2, PALB2, RAD1C, and RAD51D have been documented as mechanisms of resistance to platinum-containing therapies and PARP inhibitors27. To our knowledge, this is the first report of a reversion in NBN following treatment with platinum or PARP inhibitor therapy in any tumor type.

Third, the patient did not respond to a selective NTRK inhibitor despite demonstrated expression of an in-frame NTRK1 fusion containing the kinase domain. This was unexpected, given the high response rates for NTRK inhibitors across tumor types including pancreatic cancers9,10. Most notably, a recent case report described a patient with PACC with the same SEL1L::NTRK1 fusion reported in this study, but lacking any other likely driver alterations, who had an exceptional response to the NTRK inhibitor, larotrectinib7. In contrast, one can posit that the lack of response to NTRK inhibition in our case might be explained by the presence of the NBN LOF mutation and genomic instability that has been associated with resistance to targeted therapies20,28.

Taken together, our findings are consistent with NBN LOF having played a role in the progression of this PACC, and provide evidence that LOF of DDR/HR-related genes may have a causative role in PACC more broadly. This observation warrants the consideration of other HR/DDR genes in ongoing studies of PARP inhibitors in pancreatic cancers, including PACC (NCT04858334 and NCT05286827). Finally, our report highlights the unique insights and the clinical impact stemming from comprehensive genomic analysis of rare tumors, such as PACC.

Methods

The patient reported in this case study was enrolled in a monotherapy arm of TRESR (NCT04497116), an ongoing, modular, phase 1/2a, first-in-human, multicenter, open-label, non-randomized, dose-escalation, dose-expansion study of camonsertib administered orally as a single agent, or in combination with talazoparib or gemcitabine in patients with advanced solid tumors. The study was conducted in accordance with the Declaration of Helsinki and Council for International Organizations of Medical Sciences International Ethical Guidelines, applicable International Conference on Harmonization Good Clinical Practice Guidelines, and applicable laws and regulations. The patient provided written informed consent which allowed for in-depth genomic investigation of the tumor and normal tissue specimens provided. The protocol was approved by the Institutional Review Board or ethics committee at the treating institution, Sarah Cannon Research Institute/Tennessee Oncology.

Tissue specimen review

Formalin-fixed paraffin-embedded (FFPE) tissue specimens corresponding to the primary resection (November 2018) and metastatic liver biopsy (July 2021) were retrieved. After biopsy, tissue specimens were immediately fixed in 10% neutral buffered formalin for 12–24 h. Genomic DNA and total RNA were extracted (Invitae; San Francisco, CA) using the AllPrep DNA/RNA FFPE kit (Qiagen, cat# 80234) and quantified using the Quant-iT dsDNA Assay Kit, broad range (ThermoFisher, cat# Q33120) at NeoGenomics using documented standard operating procedures as part of conduct of the trial (Supplementary Table 1). Corresponding hematoxylin and eosin sections were reviewed by a board-certified pathologist to review tumor cellularity and histology to confirm diagnoses. BCL-10 staining of the screening tissue to confirm the diagnosis was performed at NeoGenomics.

Biomarker monitoring

Serum CA19-9 levels were monitored routinely via commercial laboratory testing.

Clinical-grade NGS

Prior to enrollment into TRESR (NCT04497116), the patient’s tumor tissue specimens were analyzed by Tempus xT and Foundation Medicine FoundationOne CDx to guide therapy selection as part of standard of care at the treating institution16,19.

Retrospective central NGS testing

SNiPDx™ targeted sequencing

SNiPDx™ is a novel targeted sequencing panel capable of distinguishing monoallelic and biallelic LOF alterations in select DNA damage repair genes15. DNA (minimum of 30 ng) was extracted from 10 × 5 µm FFPE slides. DNA was analyzed on a custom AMP panel comprising 26 genes, referred to as SNiPDx™ 15. Libraries were quantitated using quantitative PCR (qPCR; Kapa Biosystems, Woburn, MA, USA) per manufacturer protocol. Amplicon sequencing was performed on the NovaSeq platform (Illumina, San Diego, CA, USA) according to the manufacturer’s standard protocol. Paired-end sequence data were processed using methods developed for AMP to align error-corrected reads18. AMP libraries were processed using the VariantPlex Pipeline from Archer Analysis Platform v6.2.8.

Genome-wide major and minor copy numbers were inferred by FACETS29. Briefly, copy number alterations and allelic imbalances in the 26 SNiPDx™ target genes and other genomic regions were calculated on the basis of the Log2R (ie, the Log2 ratio of single nucleotide polymorphism coverage in a tumor specimen to coverage in a matched panels of normal), and Log2 odds ratio (Log2OR; calculated from the number of reads reporting the alternative allele:number of reads reporting the reference allele), adjusted by tumor purity and ploidy.

Whole genome sequencing

All FFPE-isolated DNA was quantified by KAPA SYBR FAST qPCR quantification and quality control (QC) kit (Kapa Biosystems, Woburn, MA, USA). QC criteria required a minimum DNA concentration of >1 ng/µL and minimum amplifiable ratio of >0.1. WGS was performed on the NovaSeq targeting 60x coverage for tumor and 30x coverage for PBMCs. Genome sequencing covered a minimum of 95% of the genome with 20x coverage or higher. Nucleic acid from the submitted specimen was isolated and the library products were sequenced with 2 × 150 base pair reads using the Illumina NovaSeq sequencing instrument (Illumina, San Diego, CA). After alignment to the reference genome (GRCh37/hg19), off-target, low quality, and duplicate reads were removed from the analysis. The targeted regions were assessed for average depth of coverage and other data quality thresholds.

WGS analysis was performed as previously described20,21. Raw sequencing reads from FASTQ were aligned using the Burrow-Wheelers Aligner30 to the GRCh37/hg19 reference. SNVs were called using MuTect31. Short indels were called using VarScan232, Strelka33, Platypus34, and Scalpel35. Allele-specific copy number analysis was performed using FACETS29. Mutational signatures were decomposed using SIG NAL36. Structural variants were called using Manta37 and GRIDSS38. HRD scores were calculated using HRDetect23.

Targeted RNA sequencing

Targeted RNA sequencing was performed utilizing the Archer® FusionPlex® Custom Solid Panel with Anchored Multiplex PCR (AMP™; Invitae) at the Memorial Sloan Kettering Integrated Genomics Operation as previously described17,39. Results were analyzed utilizing the FusionPlex® standard analytical suite as previously described17,40.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.