Scientific evidence
Key scientific evidence
The peer-reviewed literature behind each hypothesis in the case, grouped by theme, with its population and what the study does not say. Every DOI was checked on 29 September 2026.
Neuroendocrine Transformation and RB1 Loss
NE transformation3 RB1 variants (p.V622Yfs*33, p.R661W, p.F226*) in Guardant360 ctDNA (April 2026). Not confirmed in the July 2026 bone biopsy: the tissue FoundationOne finds no RB1 or TP53 alteration
- Park et al., 2018 — Science (DOI: 10.1126/science.aat5749)
A five-factor cocktail, including RB1 and p53 inactivation, reprograms prostate and lung epithelium into small-cell neuroendocrine carcinoma
An experimental study that starts from normal human prostate and lung epithelial cells, not breast: external validity to a breast tumour is indirect. The authors need all five factors together (dominant-negative p53, myristoylated Akt1, RB1 shRNA, c-Myc and Bcl2, which they call PARCB); without c-Myc or without Akt1 no tumour grows, and dropping the RB1 shRNA or the dominant-negative p53 yields a poorly differentiated adenocarcinoma rather than a neuroendocrine tumour. So RB1 and p53 are indispensable but not sufficient. Here 3 RB1 mutations were found in the Guardant360 liquid biopsy (April 2026), while the bone FoundationOne (July 2026) finds no RB1 or TP53 alteration: today the hypothesis rests on the ctDNA and on histology (positive synaptophysin and chromogranin A), not on the tissue genome.
- Aggarwal et al., 2018 — Journal of Clinical Oncology (DOI: 10.1200/JCO.2017.77.6880)
Castration-resistant prostate cancer: 17% treatment-emergent small-cell neuroendocrine carcinoma
A prospective multi-institutional study with metastatic biopsies in 202 patients, all with castration-resistant PROSTATE cancer (73% had already progressed on abiraterone or enzalutamide): the 17% belongs to that population, not to hormone-dependent tumours in general, and nothing in the study is breast. The paper reports incidence, a transcriptional signature and shorter survival; it does not attribute the transition to RB1 loss. It is cited because it documents the phenomenon (lineage switch to neuroendocrine under prolonged hormonal pressure) in the tumour where it has been best measured; carrying it over to breast is an analogy, not a datum.
- Schwartz et al., 2024 — Modern Pathology (DOI: 10.1016/j.modpat.2024.100541)
RB1 alterations in ER-positive breast carcinoma: under 1% of cases, associated with neuroendocrine differentiation
This one is breast, and oestrogen-receptor positive, which is where this case starts. Across 6,026 sequenced breast carcinomas, pathogenic RB1 alterations appear in fewer than 1% (55 cases): 23 primaries and 32 recurrent or metastatic. In almost every evaluable case the inactivation is biallelic, and every case expressing neuroendocrine markers has biallelic RB1 inactivation and loss of Rb protein. TP53 was co-mutated in 53% and PIK3CA in 45%. It gives the real frequency of what is suspected here, and the yardstick for reading an RB1 that shows up in blood but not in bone.
Biology of Breast Carcinoma with Neuroendocrine Differentiation
BC-NED biologyWhat is known about the subtype, and where it does and does not match this tumour's profile
- Marchiò et al., 2017 — The Journal of Pathology (DOI: 10.1002/path.4837)
The mutational repertoire of breast carcinoma with neuroendocrine differentiation: GATA3, FOXA1, TBX3, ARID1A, and less PIK3CA than common luminal tumours
Genetic characterisation of 18 ER+/HER2− breast carcinomas with extensive neuroendocrine differentiation (over 50% of cells positive for chromogranin A or synaptophysin). What comes out enriched are transcription-factor genes (FOXA1 and TBX3 at 17% versus 2%, ARID1A 17% versus 2%), with LESS PIK3CA than in common luminal cancers and no TP53 mutations at all. Worth stating plainly, because it runs against the usual reading: of this case's amplifications, the 8p11 one (FGFR1, ZNF703) appears in 2 of the series' 18 tumours; CCND1 and 11q13 are not among their findings. Eighteen cases, with sequencing targeted at 254 genes.
- Pareja et al., 2022 — Journal of Clinical Pathology (DOI: 10.1136/jclinpath-2020-207052)
Breast neuroendocrine tumours are genomically close to mucinous carcinomas, and some acquire the neuroendocrine phenotype late
Applying the WHO 5th edition criteria (2019), only 10 of 18 breast tumours with neuroendocrine differentiation remained classified as breast NETs. Compared with mucinous carcinomas and with pancreatic and pulmonary NETs, the breast ones are genetically heterogeneous and enriched for transcription-factor mutations. The sentence that matters here is theirs: 3 of the 10 carried the hallmarks of common ER-positive disease, suggesting that some breast cancers acquire neuroendocrine differentiation later in tumour evolution. Ten cases: it describes a possibility, not a frequency.
Receptor Conversion Between Primary and Metastasis
ER conversionThe 2024 primary was ER 85-100%; the liver metastasis biopsied in August 2026 reads ER 0%, PR 5%, HER2 0, with positive chromogranin A and synaptophysin
- Schrijver et al., 2018 — Journal of the National Cancer Institute (DOI: 10.1093/jnci/djx273)
Meta-analysis of 39 studies: 22.5% of distant metastases lose the oestrogen receptor, and 49.4% the progesterone receptor
This is the reference figure, from primary-metastasis pairs across 39 studies. Positive-to-negative conversion: 22.5% for ER (95% CI 16.4-30.0), 49.4% for PR and 21.3% for HER2; the reverse also happens (21.5% for ER). In liver, ER discordance is the lowest of the sites studied (14.3%, against 20.8% in the central nervous system and 29.3% in bone), while PR discordance there is among the highest (47.0%). The authors conclude that receptors should be reassessed in the metastasis. Here that has been done, and going from ER 85-100% to ER 0% is more than a threshold conversion. The meta-analysis measures receptor conversion, not lineage switch to neuroendocrine, which is a different question.
- Yan et al., 2025 — Discover Oncology (DOI: 10.1007/s12672-025-03342-0)
542 patients with breast cancer liver metastases: 21.0% ER discordance between primary and liver
A single-centre series in Beijing, 2008-2024: the population is not this case's (a Chinese cohort, with no breakdown by neuroendocrine differentiation). It is useful for one specific thing, liver-specific discordance. Of 453 cases with complete receptor records, overall discordance was 23.6%, and by receptor ER 21.0%, PR 38.6%, HER2 10.8%. It confirms with liver data what the meta-analysis gives in aggregate.
SSTR2 Pathway and Radioligand Therapy (PRRT)
SSTR2 · PRRTThe 68Ga-DOTATOC PET (26 May 2026) showed somatostatin receptor overexpression in the bone metastases; the liver has not been studied, and SSTR2 IHC on tissue has not been done
- Terlević et al., 2019 — Annals of Diagnostic Pathology (DOI: 10.1016/j.anndiagpath.2018.11.002)
SSTR2A across 31 primary neuroendocrine breast carcinomas: 71% positive, 35.5% with moderate or strong intensity
An immunohistochemical study of 31 cases, all PRIMARY: the authors themselves ask for validation in series that include metastatic disease, which is this case's situation. Of the 31, 9 were SSTR2A-negative (29%), 11 weak, 6 moderate and 5 strong; the moderate-to-strong band, 35.5%, is the one considered relevant for PRRT. It is the argument for measuring SSTR2 by IHC on tissue, because functional imaging alone says nothing about membrane intensity.
- Batty et al., 2022 — Internal Medicine Journal (DOI: 10.1111/imj.15961)
Seven patients with metastatic neuroendocrine breast carcinoma over ten years: all 6 studied with Ga-68-DOTATATE PET were avid, and first-line PFS was 5.8 months
Seven cases from a single Australian centre between 2011 and 2021: this is what exists in print, and it is a series of seven. Six of seven expressed synaptophysin and chromogranin and were also ER and PR positive, with a median Ki67 of 50%; the 6 who underwent Ga-68-DOTATATE PET showed significant uptake. Median overall survival was 31.8 months, first-line PFS 5.8 months, and every patient received chemotherapy at some point. It puts the May 2026 PET in context: avidity is often seen; what does not exist is a series measuring what PRRT does in this subtype.
- Küper et al., 2026 — Journal of Nuclear Medicine (DOI: 10.2967/jnumed.126.272521)
Of 30 breast cancer patients imaged with somatostatin-receptor PET, one reached a Krenning score of 3, the technical threshold for PRRT eligibility
A retrospective single-centre German series, 2023-2025, with 37 breast cancer patients (26 ER-positive, 11 triple negative) and final analysis on 30. SSTR-positive tumour, with at least one lesion, in 80% of the ER-positive patients: expression is common. But uptake was mostly moderate, FDG outperformed SSTR PET, and only ONE patient reached Krenning 3. It is the counterweight to an enthusiastic reading of a positive DOTATOC PET: avid does not mean candidate. The population is breast cancer in general, not selected for neuroendocrine differentiation.
- Liu et al., 2020 — Clinical Nuclear Medicine (DOI: 10.1097/RLU.0000000000003005)
One case of neuroendocrine breast carcinoma treated with 177Lu-DOTATOC, with partial remission
A single patient with invasive ductal carcinoma and large-cell neuroendocrine breast carcinoma, bone and nodal metastases, very high uptake on Ga-68-DOTATOC PET and partial remission after PRRT with 177Lu-DOTATOC. It is the lowest tier of evidence there is (one published case, no control) and it is here to mark where the evidence on this route ends: as far as could be verified, there is no PRRT series in neuroendocrine breast cancer that would support talking about response rates.
FGFR1/CCND1 Axis and CDK4/6i Resistance
FGFR1 / CCND18p11 amplification (FGFR1, NSD3, ZNF703, ≈33 copies) and 11q13 amplification (CCND1, FGF3, FGF4, FGF19, ≈37 copies) in the July 2026 bone tissue
- Formisano et al., 2019 — Nature Communications (DOI: 10.1038/s41467-019-09068-2)
Aberrant FGFR signalling mediates resistance to CDK4/6 inhibitors in ER-positive breast cancer
It proposes the mechanism; it does not prove it in patients. It starts from a kinase screen in MCF-7 cells treated with ribociclib plus fulvestrant, which pulls out FGFR1; cells with amplified or transduced FGFR1 become resistant to fulvestrant with or without ribociclib or palbociclib, and adding an FGFR inhibitor reverses that resistance, with complete responses in xenografts. The axis they describe is RAS/RAF/MEK/ERK with CCND1 and E2F downstream. The clinical part is supporting evidence: among 34 patients with ctDNA after progressing on a CDK4/6i, 14 carried FGFR1/2 amplification or activating mutations, and in MONALEESA-2 FGFR1 amplification was associated with shorter PFS. The authors close by proposing combination trials, not a treatment.
- Chae et al., 2020 — Journal of Clinical Oncology (DOI: 10.1200/JCO.19.02630)
NCI-MATCH subprotocol W: with a selective FGFR1-3 inhibitor, responses occurred only with point mutations or fusions, and in none of the 20 patients whose alteration was an amplification
The nuance that decides whether this tumour's FGFR1 axis is a target or just a finding. A histology-agnostic, multi-tumour basket trial, not an HR+ breast cohort with neuroendocrine differentiation, though breast was the commonest tumour (33.3%). There were 48 evaluable patients: FGFR1 or FGFR2 amplification (20), FGFR2 or FGFR3 single-nucleotide variants (19) and FGFR1 or FGFR3 fusions (9). Confirmed partial responses came to 8% and were observed only in tumours with point mutations or fusions: none among the 20 with amplification. With fusions, 22%. Here the alteration is an 8p11 amplification, the subgroup that did not respond.
Anti-TROP-2 ADC: Datopotamab Deruxtecan
TROP-2 · ADCShe is expected to enter TROPION-Breast06 (NCT07205822) in October 2026, a single-arm phase IIIb trial, with no prior chemotherapy; the trial the drug rests on required 1-2 prior chemotherapy lines
- Bardia et al., 2025 — Journal of Clinical Oncology (DOI: 10.1200/JCO.24.00920)
TROPION-Breast01: lower risk of progression or death than investigator's choice of chemotherapy (HR 0.63)
The trial the drug rests on, and the difference in population from this case belongs in the same sentence: its 732 patients with inoperable or metastatic HR+/HER2− breast cancer had received ONE OR TWO prior lines of chemotherapy; here there are none. Dato-DXd cut the risk of progression or death by 37% (HR 0.63; 95% CI 0.52-0.76) by blinded independent central review. Grade 3 or higher treatment-related adverse events were less frequent than with chemotherapy (20.8% versus 44.7%), but nausea and stomatitis affected around half the patients.
- Pistilli et al., 2026 — Annals of Oncology (DOI: 10.1016/j.annonc.2025.12.017)
Final overall survival analysis of TROPION-Breast01: no difference (HR 1.01; 95% CI 0.83-1.22)
The other half of the result, the one not to leave out. At 22.8 months of median follow-up, overall survival did not reach statistical significance: HR 1.01 (95% CI 0.83-1.22; p = 0.9445). The authors point to an imbalance in subsequent treatment (rescue ADCs in 12.3% of the Dato-DXd arm versus 24.0% of the chemotherapy arm) as a possible explanation, and the remaining secondary endpoints continued to favour Dato-DXd. Same population caveat: 1-2 prior chemotherapy lines.
- Royce et al., 2025 — Clinical Cancer Research (DOI: 10.1158/1078-0432.CCR-25-1388)
FDA approval summary: approved on 17 January 2025 without an overall survival benefit
It puts the tension in the file on the record, from the regulator: approval rested on the PFS improvement, with median overall survival of 18.6 versus 18.3 months (HR 1.01). The FDA reasons that, absent a survival benefit, no clear trend toward detriment was seen either. The approved indication follows prior endocrine therapy AND chemotherapy for metastatic disease, which is not this case's entry situation.
Liver Metastases and Systemic Immunity
Liver · immunityThe disease stopped being bone-confined in July 2026, and that bears on any personalised immunotherapy route, not just on the next line
- Yu et al., 2021 — Nature Medicine (DOI: 10.1038/s41591-020-1131-x)
Liver metastases siphon activated CD8 T cells into the liver, where they die on meeting monocyte-derived macrophages
A mechanism described in melanoma, non-small-cell lung cancer and mouse models, NOT in HR+ breast cancer: that is the first thing to say when reading it against this case. Liver metastases pull activated CD8 T cells out of circulation; inside the liver, antigen-specific Fas-positive CD8 cells undergo apoptosis after interacting with FasL-expressing monocyte-derived macrophages, and in the models the result is a systemic immune desert. In patients, those with liver metastases derive less benefit from immunotherapy, independently of other biomarkers, and have fewer peripheral T cells. It matters here because it bears on the whole personalised immunotherapy route (vaccine, TIL, TCR-T), not on a single modality; and it is a mechanism in models plus a correlation in other tumours, not a result in breast cancer.
- Lee et al., 2020 — Science Immunology (DOI: 10.1126/sciimmunol.aba0759)
Same phenomenon by another route: tumour antigen in the liver activates regulatory T cells and suppresses antitumour immunity in an antigen-specific way
A dual-tumour mouse model: the population is neither human nor breast. The suppression they describe is antigen-specific and comes with regulatory T cell activation. The detail that makes it useful alongside the previous paper: the dysfunctional state was not reversed by anti-PD-1 monotherapy, only by depleting regulatory T cells with anti-CTLA-4 or destabilising them with an EZH2 inhibitor. Two independent groups, two different mechanisms, the same direction: the liver is not just one more metastatic site when immunotherapy is what is being planned.