Patterns of chromosome 18 loss of heterozygosity in multifocal ileal neuroendocrine tumors

Abstract Ileal neuroendocrine tumors (NETs) represent the most common neoplasm of the small intestine. Although up to 50% of patients with ileal NETs are diagnosed with multifocal disease, the mechanisms by which multifocal ileal NETs arise are not yet understood. In this study, we analyzed genome‐wide sequencing data to examine patterns of copy number variation in 40 synchronous primary ileal NETs derived from three patients. Chromosome (chr) 18 loss of heterozygosity (LOH) was the most frequent copy number alteration identified; however, not all primary tumors from the same patient had evidence of this LOH. Our data revealed three distinct patterns of chr18 allelic loss, indicating that primary tumors from the same patient can present different LOH patterns including retention of either parental allele. In conclusion, our results are consistent with the model that multifocal ileal NETs originate independently. In addition, they suggest that there is no specific germline allele on chr18 that is the target of somatic LOH.


| INTRODUCTION
Small intestinal neuroendocrine tumors (SI-NETs) represent the most common neoplasm of the small bowel, accounting for 40% of all small intestinal malignancies. 1 The estimated annual age-adjusted incidence is 1.05/100 000 persons in the United States and has increased approximately 4-fold since 1973. 2,3 SI-NETs are generally slow-growing, well-differentiated tumors that are mostly found in the terminal ileum (ileal NETs), but show often high metastatic potential. About 50% of the patients have distant metastases at diagnosis, and the 5-year survival rate for those patients is around 50%. [3][4][5] For patients with metastatic disease, effective treatment modalities are limited. 4 Previous high-throughput sequencing studies have revealed that ileal NETs have a low somatic mutation rate. 6,7 To date, the most frequent genomic event identified in these tumors is loss of heterozygosity (LOH) at chromosome (chr) 18, occurring in approximately 60% of cases, and the only recurrent mutations identified to date are loss-offunction mutations in cyclin-dependent kinase inhibitor 1B in 8% to 10% of tumors. [6][7][8] Indeed, in a recent whole genome sequencing study of 2520 metastatic cancer/normal pairs, SI-NETs were found to be the one tumor type in which a candidate driver gene was rarely found; 18 of 37 SI-NETs had no identified candidate driver compared to 16 of 2483 other cancers. 9 The majority of previous studies have focused on single primary tumors from each patient. However, up to 50% of patients with ileal NETs are diagnosed with multiple synchronous primary tumors. [10][11][12] The molecular genetic mechanisms by which these multifocal lesions develop remain mostly unknown. One hypothesis is that the multifocal tumors may arise as a result of an unrecognized germline mutation. 13 By analyzing LOH patterns at six candidate genetic loci using PCR-based method, Katona et al. have suggested multifocal ileal NETs may originate independently. 14 We report here the first high-throughput sequencing-based copy number profiling of multifocal primary ileal NETs. We not only confirm that chr18 LOH is the most common somatic copy number change in multifocal ileal NETs, but also reveal distinct patterns of chr18 allelic loss in individual tumors from the same patient. Our results corroborate the hypothesis that multiple primary ileal NETs may develop independently and provide novel insights into the molecular mechanism of chr18 LOH.

| Patients and tissue samples
Our sample set consisted of 40 de-identified synchronous primary tumors and matched adjacent normal ileal mucosa specimens from three ileal NET patients ( Figure 1; Table S1). A piece of each tissue sample was obtained, freshly frozen in liquid nitrogen, and stored at -80 C. All tumor specimens were 2.2 | Whole-exome and linked-read wholegenome sequencing Genomic DNA (gDNA) was extracted from fresh frozen tumor and adjacent normal ileum samples. Tissue specimens from patient 1 were sequenced using the 10× Genomics linked-read WGS approach (10× Genomics, Pleasanton, California), as detailed before. 15 Briefly, gDNA libraries were sequenced on an Illumina HiSeq platform to generate 151-bp paired-end reads with a mean depth of 60× coverage for tumor specimens, and 30× coverage for the tumor adjacent normal tissue. Whole-exome sequencing (WES) was performed for patients 2 and 3, as previously described. 16 In brief, exonic sequences were enriched using the Agilent V2 capture probe set (Agilent, Santa Clara, California) and sequenced by 76-bp paired-end reads using the Illumina Genome Analyzer IIx system (Illumina, San Diego, California) with a mean coverage of 80× for each base. Sequencing reads were aligned to the human reference genome GRCh37 (hg19).

| Chr18 LOH is common in multifocal ileal NETs
Copy number analysis identified hemizygous loss of chr18 (log2 [copy number/2] < -0.1) to be the most common genomic alteration in multifocal ileal NETs ( Figure S1), consistent with previous reports. 6,8 The frequency of chr18 LOH was 55% (6 of 11 primary tumors) in tumors from patient 1, 39% (7 of 18 primary tumors) in tumors from patient 2, and 27% (3 of 11 primary tumors) in tumors from patient 3. Other chromosomal copy number changes included gains of chromosomes 4, 5, 7, 14, and 20; none of these changes were present in tumors from more than one patient.

| DISCUSSION
Ileal NETs represent the most common neoplasm of the small intestine with a high incidence of multiple synchronous primary tumors at diagnosis. Our understanding of the tumor multifocality, however, remains limited with respect to the molecular mechanisms underlying their tumorigenesis and effects on clinical management. 11,12,20 Here, we utilized high-throughput sequencing to characterize copy number alterations present in multifocal ileal NETs to better understand the molecular genetic background of these lesions.
Most large-scale sequencing studies to date have concentrated on sequencing single primary tumors from ileal NET patients. 6,7 This is, to our knowledge, the first comprehensive copy number profiling study of multifocal primary ileal NETs. The observed copy number changes F I G U R E 2 Schematic representation of chr18 LOH patterns occurring in multifocal tumors from three ileal NET patients. Red and blue dots in figure inset represent parental chr18 alleles included loss of chr18 and gains of chr4, chr5, chr7, chr14, and chr20.
These findings are in agreement with what others and we have previously reported when studying single primary ileal NETs from different individuals. 6,7,21 More importantly, our data corroborate that chr18 loss is the most prominent copy number alteration in ileal NETs, regardless of whether tumors are unifocal or multifocal. Of note, not all primary tumors from the same patient displayed chr18 loss.
The origin of multifocal ileal NETs is not well understood. Previous studies have proposed that multifocal ileal NETs may arise due to germline predisposition or as independent clones. 13,14 Sei et al. used WES to screen for single nucleotide germline variants in 33 SI-NET families with multiple synchronous tumors but was unable to identify a common germline mutation in these families. 13  NETs, it would be reasonable to infer that the reason for chr18 LOH is to delete a wild type copy of a tumor suppressor gene in the setting of a germline loss-of-function mutation. However, the loss of either parental allele, in tumors from the same patient, rules out this possibility.
Thus, our data suggest that it is improbable that there exists a germline loss-of-function mutation in a tumor suppressor gene on chr18. Rather, we conclude that it is more likely that the oncogenic effect of chr18 allelic loss may be due to a dosage impact on the expression of one, several or many genes within the chromosome.
There are some limitations in our current study. First, our interpretation of LOH pattern assumed that the hemizygous loss of chr18 in ileal NET is more likely to occur in a single parental allele as previously reported. 6,7,22,23 Tissue karyotyping and fluorescence in situ hybridization analysis may be required for further confirmation.
Second, we could not map the LOH pattern for all tumors displaying chr18 LOH because of a small number of analyzable LOH-informative SNPs found in some tumors. This may be explained by a lower amount of tumor cells in those tissue samples. Last, our current data were not able to pinpoint the exact mechanism of how a tumor (T2 from patient 1) lost different parental alleles in p and q arms of chr18. One possibility is that homologous recombination occurs at chr18, and one of the recombined copies is lost during tumorigenesis.
Alternatively, this tumor might undergo loss of the two arms of chr18 separately, indicating that the chr18 LOH of ileal NETs may represent two independent genomic events ( Figure 2). Harnessing long-read sequencing technologies to generate ultra-long reads that cover both chromosomal arms may be required to reveal the exact mechanism. 24,25 In conclusion, our data demonstrate that chr18 LOH is commonly found in multifocal ileal NETs. The distinct LOH patterns in individual tumors from the same patients suggest that there is not a particular germline allele on chr18 that predisposes somatic LOH.