Article Text

Download PDFPDF
Neuroendocrine tumours of the breast: a genomic comparison with mucinous breast cancers and neuroendocrine tumours of other anatomic sites
  1. Fresia Pareja1,
  2. Mahsa Vahdatinia1,
  3. Caterina Marchio2,3,
  4. Simon S K Lee1,
  5. Arnaud Da Cruz Paula4,
  6. Fatemeh Derakhshan1,
  7. Edaise M da Silva1,
  8. Pier Selenica1,
  9. Higinio Dopeso1,
  10. Sarat Chandarlapaty5,
  11. Hannah Y Wen1,
  12. Anne Vincent-Salomon6,
  13. Edi Brogi1,
  14. Britta Weigelt1,
  15. Jorge S Reis-Filho1
  1. 1 Department of Pathology, Memorial Sloan Kettering Cancer Center, New York City, New York, USA
  2. 2 Department of Medical Sciences, University of Turin, Turin, Italy
  3. 3 Unit of Pathology, Candiolo Cancer Institute, FPO IRCCS, Candiolo, Italy
  4. 4 Department of Surgery, Memorial Sloan Kettering Cancer Center, New York City, New York, USA
  5. 5 Department of Medicine, Memorial Sloan Kettering Cancer Center, New York City, New York, USA
  6. 6 Department de Medicine Diagnostique et Theranostique, Institut Curie, Paris, France
  1. Correspondence to Dr Fresia Pareja, Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, NY 10065, USA; parejaf{at}mskcc.org; Dr Jorge S Reis-Filho, Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, NY 10065, USA; reisfilj{at}mskcc.org

Abstract

Aims Breast neuroendocrine tumours (NETs) constitute a rare histologic subtype of oestrogen receptor (ER)-positive breast cancer, and their definition according to the WHO classification was revised in 2019. Breast NETs display histologic and transcriptomic similarities with mucinous breast carcinomas (MuBCs). Here, we sought to compare the repertoire of genetic alterations in breast NETs with MuBCs and NETs from other anatomic origins.

Methods On histologic review applying the new WHO criteria, 18 breast tumours with neuroendocrine differentiation were reclassified as breast NETs (n=10) or other breast cancers with neuroendocrine differentiation (n=8). We reanalysed targeted sequencing or whole-exome sequencing data of breast NETs (n=10), MuBCs type A (n=12) and type B (n=11).

Results Breast NETs and MuBCs were found to be genetically similar, harbouring a lower frequency of PIK3CA mutations, 1q gains and 16q losses than ER-positive/HER2-negative breast cancers. 3/10 breast NETs harboured the hallmark features of ER-positive disease (ie, PIK3CA mutations and concurrent 1q gains/16q losses). Breast NETs showed an enrichment of oncogenic/likely oncogenic mutations affecting transcription factors compared with common forms of ER-positive breast cancer and with pancreatic and pulmonary NETs.

Conclusions Breast NETs are heterogeneous and are characterised by an enrichment of mutations in transcription factors and likely constitute a spectrum of entities histologically and genomically related to MuBCs. While most breast NETs are distinct from ER-positive/HER2-negative IDC-NSTs, a subset of breast NETs appears to be genetically similar to common forms of ER-positive breast cancer, suggesting that some breast cancers may acquire neuroendocrine differentiation later in tumour evolution.

  • neuroendocrine tumours
  • pathology
  • molecular
  • breast

Data availability statement

Data are available from the authors upon reasonable request.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

INTRODUCTION

Breast neuroendocrine tumours (NETs) are special histologic types of oestrogen receptor (ER)-positive breast cancer accounting for less than 1% of breast cancers.1 In an effort to standardise the taxonomy of NETs across different organ systems, a revised classification of breast neuroendocrine neoplasms has been put forward in the latest (fifth) Edition of WHO Classification of Breast Tumors,1 following an expert consensus proposal.2 Based on this revised classification system,1 neuroendocrine neoplasms of the breast are catalogued as neuroendocrine carcinomas, including small cell and large cell carcinomas, or breast NETs, defined as invasive carcinomas of histologic grade 1 or 2 displaying neuroendocrine morphology and extensive expression of neuroendocrine markers.1 Importantly, special histologic subtypes of breast cancer expressing neuroendocrine markers, such as solid papillary carcinomas or mucinous breast carcinomas (MuBCs) type B, which were included in the ‘carcinoma with neuroendocrine features’ category as per the fourth WHO Edition,3 have been excluded from the group of breast neoplasms in the current (fifth) WHO classification.1 It should be noted, however, that NETs and MuBCs display overlapping histologic features, as NETs are heterogeneous and may harbour foci of mucin production, and MuBCs type B may display neuroendocrine differentiation.4 5

Breast NETs also phenotypically resemble NETs arising in different anatomic locations. The molecular basis of neuroendocrine differentiation has been investigated in different organ systems,6 for instance the transcription factor BRN4 has emerged as a driver of neuroendocrine differentiation in castration-resistant prostate cancer.7 Whether a common molecular basis for the neuroendocrine phenotype exists has yet to be determined. The identification of genetic similarities and/or differences of breast NETs and NETs from other organ systems might aid in overcoming diagnostic challenges posed by their overlapping phenotypes and provide a rational basis for their taxonomy and a foundation for future studies to determine whether targeted treatments for patients with breast NETs should be aligned with those of NETs arising in other organ systems.

Given the overlapping histologic features between breast NETs compared and other breast tumours with neuroendocrine differentiation, including those cancers recently reclassified outside the group of neuroendocrine neoplasms and MuBCs,1 4 5 8 we hypothesised that these entities might display similarities also at the genetic level. We have previously shown that breast NETs and MuBCs are similar at the transcriptional level, but different from common forms of ER-positive breast cancers.9 Here, we sought (1) to compare the repertoire of somatic genetic alterations in breast NETs, other breast tumours with neuroendocrine differentiation and MuBCs; (2) to evaluate whether breast NETs differ from ER-positive/HER2-negative invasive ductal carcinomas of no special type (IDC-NSTs) and (3) to determine whether breast NETs genetically resemble NETs arising in other organ systems.

Materials and methods

Cases and reanalysis of massively parallel sequencing data

Our cohort included 18 breast cancers reported by Marchiò et al,10 which following the prior (fourth Ed) WHO classification (2012)3 had been previously classified as neuroendocrine breast carcinomas.10 Following histopathologic re-review (CM) according to criteria put forward by the latest (fifth edition) WHO classification (2019; table 1),1 these were reclassified as breast NETs (10/18; 55%) or excluded from this category and classified as other breast tumours with neuroendocrine differentiation (n=8), including IDC-NST with neuroendocrine features (n=3), MuBC type B (n=2), invasive lobular carcinoma with neuroendocrine features (n=1) or solid papillary carcinomas (n=2; table 2). Twenty-one MuBCs (12 type A and 9 type B) from the study by Pareja et al 11 were re-reviewed (FP). All MuBCs were composed of >90% of mucinous carcinoma areas and, following the criteria put forward by the current 2019 WHO classification,1 correspond to pure MuBCs. Of note, the two cases from the study by Marchiò et al 10 reclassified as MuBCs type B were grouped together with MuBCs type B from Pareja et al.11 All tumours were graded according to the Nottingham grading system.12 ER and HER2 status were retrieved from the original publications,10 11 where these were assessed by immunohistochemistry and/or fluorescence in situ hybridisation following the American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) guidelines.13 14 The raw targeted sequencing or whole-exome sequencing data (ie, FASTQ files) of all cases were reprocessed using our validated bioinformatics pipeline, as described15 16 (online supplemental methods).

Supplemental material

Table 1

Histologic characteristics and classification of neuroendocrine tumours and other tumours with neuroendocrine differentiation included in this study according to the WHO criteria

Table 2

Histologic and immunophenotypic characteristics of the breast neuroendocrine tumours and other breast cancers with neuroendocrine differentiation included in this study

Comparisons with breast cancers from TCGA and NETs from other anatomic locations

The frequency of mutations affecting the 254 genes from Marchiò et al 10 and copy number alterations in breast NETs (n=10) were compared with those of ER-positive/HER2-negative IDC-NSTs from The Cancer Genome Atlas (TCGA) breast cancer study (n=310). The publicly available MC3 dataset was retrieved from the 2018 TCGA Pan-Cancer Atlas study17 (https://gdc.cancer.gov/about-data/publications/mc3-2017). Somatic mutations of 98 pancreatic NETs subjected to whole-genome sequencing by Scarpa et al 18 were retrieved using cBioPortal19 and those of 29 pulmonary carcinoids subjected to targeted sequencing20 using MSK-Integrated Mutation Profiling of Actionable Targets (MSK-IMPACT)21 were retrieved from the publication by Laddha et al.20 The frequency of oncogenic/likely oncogenic mutations affecting genes common to the panels included in the study by Marchiò et al 10 and to MSK-IMPACT used in the study by Laddha et al 20 in breast NETs was compared with that of pancreatic NETs18 and lung carcinoids.20

Statistical analysis

Statistical analyses were performed using R V.1.2. Fisher’s exact test was used for comparisons between categorical variables. All tests were two-sided and p values <0.05 were considered statistically significant.

Results

Breast NETs and other breast cancers with neuroendocrine differentiation

All breast NETs (n=10) and other breast cancers with neuroendocrine differentiation (n=8) from Marchiò et al 10 were found to express neuroendocrine markers (ie, chromogranin A and/or synaptophysin) in at least 50% of their area10 22 and were ER-positive and HER2-negative (figure 1A,B and table 2).

Figure 1

Repertoire of somatic mutations and copy number alterations in breast neuroendocrine tumours (NETs) and in other breast cancers with neuroendocrine differentiation. (A–B) Representative photomicrographs of (A) H&E-stained breast NET CMNE11 and (B) corresponding chromogranin A immunohistochemical expression. Scale bars, 50 µm. (C) Heatmap depicting non-synonymous somatic mutations affecting the 254 genes included in the targeted sequencing panel from the study by Marchiò et al 10 in breast NETs (n=10) and other breast cancers with neuroendocrine differentiation (n=8). Cases are shown in columns and genes in rows. Clinicopathologic characteristics are depicted in the phenotype bars (top). Somatic mutations are color-coded according to the legend, and loss of heterozygosity is represented by a diagonal bar. (D) Heatmaps depicting non-synonymous somatic mutations affecting transcription factor-encoding genes, chromatin remodelers and PI3K genes in breast NETs (n=10) and other breast cancers with neuroendocrine differentiation (n=8). Cases are shown in columns and genes in rows. Cases with oncogenic/likely oncogenic mutations affecting these genes are indicated in a phenotype bar (top). Mutations are color-coded according to the legend, and loss of heterozygosity is represented by a diagonal bar. (E) Copy number alterations in breast NETs (n=10) and other breast tumours with neuroendocrine differentiation (n=8). The frequency of gains (green bars) or losses (purple bars) for each gene is plotted on the y-axis according to their genomic position (x-axis). Inverse Log10 values of Fisher’s exact test p values are plotted according to genomic position (lower panel). Chrom remod, chromatin remodelers; ER, oestrogen receptor; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; SNV, single nucleotide variant; TF, transcription factor.

Reanalysis of the repertoire of somatic genetic alterations affecting the 254 genes investigated by Marchiò et al 10 revealed that the most frequently mutated genes in the breast NETs (n=10) were DCHS2, TBX3, KMT2C, FOXA1, ARID1A, PIK3CA, MACF1 and CTCF (2/10; 20% each), and in the other breast cancers with neuroendocrine differentiation (n=8) were GATA3, AKT1, CDH1 and CACNA1C (2/8; 25% each; figure 1C). No statistically significant differences were observed in single-gene comparisons (p>0.05; Fisher’s exact test; figure 1C), however, the sample size of each group was small; therefore, we cannot rule out type II or β errors. Notably, we observed a low frequency of mutations in PIK3CA, a gene commonly affected in ER-positive invasive ductal and lobular carcinomas, as well as a lack of TP53 mutations (figure 1C). Two solid papillary carcinomas (CMNE04 and CMNE05) were included in the group of other breast cancers with neuroendocrine differentiation. CMNE04 harboured a DCHS2 truncating mutation, a ANK3 H438Y missense mutation and a BRCA1 (S146F) missense mutation of uncertain significance, and CMNE05 harboured a TBX3 frameshift mutation and splice-site mutations affecting CACNA1C and AK9 (figure 1C). The single nucleotide variants identified in these two cases, however, likely constitute passenger events. We next sought to compare the frequency of cases harbouring oncogenic/likely oncogenic mutations affecting transcription factor-encoding genes, chromatin remodelers and PI3K pathway genes. We detected a similarly high proportion of breast NETs (50%; 5/10) and other breast cancers with neuroendocrine differentiation (50%; 4/8) harbouring oncogenic/likely oncogenic mutations affecting transcription factor-encoding genes (p>0.05; Fisher’s exact test; figure 1D). Likewise, the fraction of cases affected by oncogenic/likely oncogenic mutations targeting chromatin remodelling genes (20% vs 25%) or PI3K pathway genes (10% vs 25%) was comparable between the two groups (p>0.05, Fisher’s exact test; figure 1D). No differences in the gene copy number alterations of breast NETs (n=10) and other breast cancers with neuroendocrine differentiation (n=8) were observed (figure 1E).

Breast NETs and mucinous breast cancers have similar genetic profiles

We next sought to compare the repertoire of genetic alterations of breast NETs (n=10) with those of MuBC type A (n=12) and MuBC type B (n=11), obtained from Pareja et al 11 (n=9) and two cases from Marchiò et al 10 reclassified as MuBCs type B (CMNE13 and CMNE10).

Our reanalysis of the repertoire of somatic genetic alterations affecting the 254 genes interrogated in the study by Marchiò et al 10 revealed that FRG1B and SF3B1 were the most frequently mutated genes in MuBC type A (2/12; 17% each), whereas the transcription factors GATA3 (4/11; 36%) and the chromatin remodelling gene KMT2C (3/11; 27%) were the most frequently mutated genes in MuBC type B. No statistically significant differences in the mutational frequency at the single-gene level were observed (p>0.05; Fisher’s exact test; figure 2A). Oncogenic/likely oncogenic mutations in transcription factors affected a similar proportions of breast NETs (50%; 5/10) and MuBC type B (6/11; 55%; p>0.05, Fisher’s exact test), which were numerically higher than that of MuBC type A (17%; p>0.05, Fisher’s exact test; figure 2B). Breast NETs, MuBC type A and MuBC type B were affected by oncogenic/likely oncogenic mutations in chromatin remodelling (20% vs 17% vs 27%) and PI3K pathway genes (10% vs 17% vs 9%) at a similar rate (p>0.05, Fisher’s exact test; figure 3B). Copy number analyses revealed that concurrent 1q gains/16q losses, the hallmark genetic alteration of common forms of ER-positive breast cancer23 were present in 2/10 breast NETs (20%) and in none of the MuBCs type A (n=12) or MuBCs type B (n=11; figure 2A). No significant differences in the frequency of gene copy number alterations between the two groups were observed (figure 2D).

Figure 2

Repertoire of somatic mutations in breast neuroendocrine tumours (NETs) compared with mucinous breast cancers. (A) Heatmaps depicting non-synonymous somatic mutations identified in breast NETs (n=10), mucinous breast cancers (MuBC) type A (n=12) and MuBC type B (n=11) affecting the 254 genes included in the targeted sequencing panel from the study by Marchiò et al.10 Cases are shown in columns and genes in rows. Clinicopathologic characteristics are depicted in the phenotype bars (top). Non-synonymous somatic mutations are color-coded according to the legend, and loss of heterozygosity is represented by a diagonal bar. (B) Heatmaps depicting non-synonymous somatic mutations affecting transcription factors, chromatin remodelers and PI3K genes in breast NETs (n=10), MuBCs type A (n=12) and MuBCs type B (n=11). Cases are shown in columns and genes in rows. Cases with oncogenic/likely oncogenic mutations affecting these genes are indicated in a phenotype bar (top). Mutations are color-coded according to the legend, and loss of heterozygosity is represented by a diagonal bar. (C) Copy number alterations in breast NETs (n=10), MuBCs type A (n=12) and MuBCs type B (n=11). Cases are depicted in rows and chromosomes are shown along the x-axis. Histopathologic characteristics are shown in the phenotype bar (left) and color-coded according to the legend. Dark red, amplification; light red, copy number gain; dark blue, homozygous deletion; light blue, copy number loss; white, copy neutral. (D) Frequency plots and Fisher’s exact test comparisons corrected for multiple testing of copy number gains and losses between breast NETs (n=10) and MuBCs type A+B (n=22). The frequency of gains (green bars) or losses (purple bars) for each gene is plotted on the y-axis according to their genomic position (x-axis). Inverse Log10 values of Fisher’s exact test p values are plotted according to genomic position (lower panel). Chrom rem, chromatin remodelers; ER, oestrogen receptor; SNV, single nucleotide variant; TF, transcription factor; WES, whole-exome sequencing.

Figure 3

Comparison of breast neuroendocrine tumours (NETs) to oestrogen receptor-positive/HER2-negative breast cancer and NETs of other organ systems. (A) Heatmaps depicting the most frequently mutated genes identified in breast NETs (n=10), and ER-positive/HER2-negative invasive ductal carcinomas of no-special type (IDC-NSTs; n=310) from TCGA. Cases are shown in columns and genes in rows. (B) Frequency plots and Fisher’s exact test comparisons corrected for multiple testing of copy number gains and losses between breast NETs from this study (n=10) and ER-positive/HER2-negative IDC-NSTs from TCGA (n=310). The frequency of gains (green bars) or losses (purple bars) for each gene is plotted on the y-axis according to their genomic position (x-axis). Inverse Log10 values of Fisher’s exact test p values are plotted according to genomic position (lower panel). (C–D) Representative photomicrographs of H&E-stained (C) breast NET CMNE21 and (D) corresponding synaptophysin immunohistochemical expression. Scale bars, 100 µm. (E–G) Frequency of breast NETs (n=10), pancreatic NETs from the study by Scarpa et al 18 (n=98) and lung carcinoids from the study by Laddha et al 20 (n=29) harbouring oncogenic/likely oncogenic mutations affecting (E) transcription factor-encoding genes, (F) chromatin remodelers and (G) PI3K pathway genes. *p<0.05; **p<0.01; ***p<0.001, Fisher’s exact test. ER, oestrogen receptor; indel, small insertion/deletion; SNV, single nucleotide variant; TCGA, The Cancer Genome Atlas; TF, transcription factor.

Breast NETs differ from common forms of ER-positive breast cancer

Next, we sought to compare the repertoire of genetic alterations in breast NETs to that of common forms of breast cancer. In comparison with ER-positive/HER2-negative IDC-NSTs from TCGA (n=310), breast NETs (n=10) harboured a statistically significantly higher frequency of mutations targeting the transcription factors TBX3, FOXA1 and CTCF (20% vs 3%, each; p<0.05; Fisher’s exact test; figure 3A) and, akin to MuBCs,11 24 a numerically lower frequency of PIK3CA mutations. Moreover, as described in MuBCs,11 24 breast NETs (n=10) had a lower frequency of 1q gains and 16q losses compared with ER-positive/HER2-negative IDC-NSTs from TCGA (n=310; figure 3B).

A subset of breast NETs (3/10; CMNE12, CMNE15 and CMNE21) harboured genomic features characteristic of common forms of ER-positive breast cancers (ie, PIK3CA mutations and/or concurrent 1q gains/16q losses; figures 2C and 3A). Interestingly, these cases diffusely expressed synaptophysin but were negative for chromogranin A, in contrast to the remaining breast NETs that expressed both markers (n=7; table 2). Notably, one of these cases, CMNE21, a PIK3CA-mutant NET with a concurrent 1q gain/16q loss, showed focal areas of ductal carcinoma lacking neuroendocrine histologic features and synaptophysin or chromogranin A expression (figure 3C,D).

Comparative analysis of NETs of the breast and other anatomic locations

Finally, we compared the frequency of breast NETs from this study, pancreatic NETs subjected to whole-genome sequencing by Scarpa et al,18 and pulmonary carcinoids subjected to targeted sequencing using MSK-IMPACT21 by Laddha et al 20 harbouring oncogenic/likely oncogenic mutations affecting transcription factors, chromatin remodelling or PI3K pathway genes common to the targeted sequencing panel used by Marchiò et al 10 and MSK-IMPACT used in the study by Laddha et al.20 Our analyses revealed that 50% (5/10) of breast NETs harboured oncogenic/likely oncogenic mutations in transcription factor-encoding genes compared with none of the pancreatic NETs (0/98; p<0.001, Fisher’s exact test) or pulmonary carcinoids (0/29; p<0.001, Fisher’s exact test; figure 3E). Breast NETs, pancreatic NETs and pulmonary carcinoids were affected by oncogenic/likely oncogenic mutations in chromatin remodelling (10% vs 12% vs 14%) and PI3K pathway genes (10% vs 9% vs 3.4%) at similar frequencies (p>0.05, Fisher’s exact test; figure 3F,G).

Discussion

Here, through the comparative reanalysis of massively parallel sequencing data, we show that breast NETs as defined by the latest WHO classification1 and other breast tumours with neuroendocrine differentiation classified outside the realm of neuroendocrine neoplasms according to this revised criteria1 have a similar repertoire of genetic alterations, including an enrichment in genetic alterations affecting transcription factors. These findings suggest that despite their different taxonomies, these tumours might be genetically related.

We have previously shown that the transcriptomic profiles of breast cancers with neuroendocrine differentiation closely resemble those of MuBCs type A and MuBCs type B.9 Here, we observed that breast NETs, MuBCs type A and MuBCs type B display similar patterns of somatic genetic alterations, which differ from that of common forms of ER-positive breast cancer, including a low frequency of oncogenic/likely oncogenic mutations in PI3K pathway genes and of 1q gains/16q losses, consistent with previous studies.10 11 24 These findings support the notion that these entities might belong to the same spectrum of disease.

Notably, a subset of the breast NETs studied here displayed genomic features characteristic of common forms of ER-positive breast cancer (PIK3CA mutations and/or concurrent 1q gains/16q losses). In tumours of other anatomic sites, neuroendocrine differentiation may be acquired late in the development or progression; for instance, prostate cancers may develop a neuroendocrine phenotype in the post-therapy/metastatic setting.25 Although all breast NETs analysed here were treatment-naive primary breast cancers, we posit that some of these tumours may constitute de novo breast NETs and would lack the hallmark genomic characteristics of common forms of ER-positive disease, whereas in others, the neuroendocrine phenotype may be acquired late or even constitute an epiphenomenon. Consistent with this hypothesis, the breast NETs harbouring concurrent 1q gains/16q losses and/or PIK3CA mutations differed immunophenotypically from the remaining breast NETs, in that synaptophysin was consistently expressed but chromogranin A was not, whereas in the remaining breast NETs both neuroendocrine markers were consistently detected. Whether these differences suggest partial neuroendocrine differentiation due to acquisition of a neuroendocrine phenotype later in tumour evolution warrants further study. In addition, CMNE21, a breast NET with a concurrent 1q gain/16q loss a clonal PIK3CA H1047R hotspot mutation, displayed focal IDC-NST areas lacking neuroendocrine morphology or immunophenotype. It is conceivable that CMME21 may correspond to an IDC-NST harbouring genomic features of ER-positive breast cancer that acquired neuroendocrine differentiation later in its evolution.

Consistent with previous studies reporting mutations in chromatin remodelling genes in NETs of other organ systems,26 27 we did observe a similar proportion of cases affected by oncogenic/likely oncogenic mutations targeting chromatin remodelling genes in pulmonary carcinoids, and pancreatic and breast NETs. As compared with NETs from other anatomic origins, breast NETs displayed a significant enrichment in oncogenic/likely oncogenic mutations in transcription factor-encoding genes. It is worth noting, however, that this comparative analysis was restricted to the set of genes common to the targeted sequencing panels used in the studies by Marchiò et al 10 and Laddha et al.20 Indeed, the chromatin remodelling gene MEN1, which was the most frequently mutated gene in pulmonary carcinoids (14%) in the series of Laddha et al, 20 was not interrogated in the breast NETs of our study. Hence, it is possible that the proportion of breast NETs with genetic alterations affecting chromatin remodelers might be conservatively estimated in our series.

Our study has important limitations, such as the small number of cases analysed given the rarity of breast NETs, the limited number of genes assessed and the fact that mutational signatures were not analysed due to the small gene panel employed. Despite these limitations, our findings suggest that breast NETs show an enrichment in mutations affecting transcription factor-encoding genes compared with NETs in other organs and that MuBCs and most breast NETs might constitute a spectrum of genomically related histologic subtypes, distinct from common forms of ER-positive breast cancer. Based on our findings, we posit the existence of two subsets of breast NETs, a de novo subset that lacks genomic features of common forms of ER-positive/HER2-negative breast cancers, and another where the neuroendocrine phenotype may develop later in cancer evolution or merely constitute an epiphenomenon.

Take home messages

  • Breast neuroendocrine tumours and mucinous breast cancers are genetically similar and differ from common forms of oestrogen receptor-positive breast cancer.

  • Breast neuroendocrine tumours display an enrichment of genetic alterations affecting transcription factors compared with common forms of oestrogen receptor-positive breast cancer and with neuroendocrine tumours arising in other anatomic locations.

  • A subset of breast neuroendocrine tumours genetically resemble common forms of oestrogen receptor-positive breast cancers and could potentially represent tumours that acquired a neuroendocrine phenotype later in tumour evolution.

Data availability statement

Data are available from the authors upon reasonable request.

Ethics statements

Patient consent for publication

References

Footnotes

  • Handling editor Runjan Chetty.

  • Contributors BW and JSR-F conceived the study. FP, CM and JSR-F reviewed the cases. SSKL and ADCP performed the bioinformatic analysis. FP, MV, CM, SSKL, ADCP, FD, EMdS, PS, HD, SC, HYW, AV-S and EB analysed and interpreted the data. FP, MV and JSR-F wrote the first manuscript, which was reviewed by all coauthors.

  • Funding This study was funded by the Breast Cancer Research Foundation. Research reported in this publication was funded in part by a Cancer Center Support Grant of the National Institutes of Health/National Cancer Institute (grant No P30CA008748). FP is partially funded by a K12 CA184746 grant, BW by Cycle for Survival and Stand Up To Cancer grants.

  • Disclaimer The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

  • Competing interests JSR-F reports receiving personal/consultancy fees from Goldman Sachs and REPARE Therapeutics, membership of the scientific advisory boards of VolitionRx and Page.AI, and ad hoc membership of the scientific advisory boards of Roche Tissue Diagnostics, Ventana Medical Systems, Novartis, Genentech and InVicro, outside the scope of this study. All other authors declare no conflicts of interest. CM has received personal/consultancy fees from Roche, Bayer, Daiichi Sankyo, MSD, Thesaro, COR2ED, outside of the scope of the present work. SC has received research support from Daichi Sankyo and consulting fees from Novartis, Sermonix, BMS, Context Therapeutics, Revolution Medicine, Paige AI and Eli Lilly.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.