npj | precision oncology Article

Published in partnership with The Hormel Institute, University of Minnesota

https://doi.org/10.1038/s41698-024-00723-6

Serial single-cell RNA sequencing unveils

drug resistance and metastatic traits in
stage IV breast cancer
Check for updates
1 1,2 3 4 4
Kazutaka Otsuji , Yoko Takahashi , Tomo Osako , Takayuki Kobayashi , Toshimi Takano ,
Sumito Saeki 5, Liying Yang5, Satoko Baba3,6, Kohei Kumegawa 1, Hiromu Suzuki7, Tetsuo Noda8,
Kengo Takeuchi3,6, Shinji Ohno 1, Takayuki Ueno 1,2 & Reo Maruyama 1,5

Metastasis is a complex process that remains poorly understood at the molecular levels. We profiled
1234567890():,;
1234567890():,;

single-cell transcriptomic, genomic, and epigenomic changes associated with cancer cell
progression, chemotherapy resistance, and metastasis from a Stage IV breast cancer patient.
Pretreatment- and posttreatment-specimens from the primary tumor and distant metastases were
collected for single-cell RNA sequencing and subsequent cell clustering, copy number variation (CNV)
estimation, transcriptomic factor estimation, and pseudotime analyses. CNV analysis revealed that a
small population of pretreatment cancer cells resisted chemotherapy and expanded. New clones
including Metastatic Precursor Cells (MPCs), emerged in the posttreatment primary tumors in CNV
similar to metastatic cells. MPCs exhibited expression profiles indicative of epithelial–mesenchymal
transition. Comparison of MPCs with metastatic cancer cells also revealed dynamic changes in
transcription factors and calcitonin pathway gene expression. These findings demonstrate the utility of
single-patient clinical sample analysis for understanding tumor drug resistance, regrowth, and
metastasis.

Breast cancer in its early stage is generally associated with a better prog- One of the crucial causes of ITH is the accumulation of different
nosis compared to other cancers, while recurrence and metastasis sig- genomic abnormalities in individual cells. The genetic diversity arises from
nificantly worsen its outcomes regardless of its subtype1–3. Intratumoral genetic and epigenetic alterations during tumor progression, creating
heterogeneity (ITH) is a key factor in drug resistance, cancer recurrence, multiple subclonal populations. These subclones can differ in their
and metastasis. Deciphering ITH is not only essential for improving responses to therapy, invasive potential, and metastasizing ability. This
cancer diagnostics and treatments but also holds the potential to improve evolutionary process is known as clonal evolution10,11. Similar to but dif-
patients’ prognoses4,5. The traditional bulk analysis has struggled to cap- ferent from clonal evolution, cancer cell plasticity also significantly con-
ture the heterogeneous characteristics of cancer; however, the advent of tributes to ITH. It represents a transient change within cancer cells, allowing
single-cell analysis has dramatically deepened our understanding of them to adapt to their environment and to adopt different phenotypic states,
ITH6,7. Some studies for breast cancer using single-cell analyses, for such as epithelial-to-mesenchymal transition (EMT), or acquire stem-like
instance, revealed subpopulations specific to subtypes, which may play a properties12–14. Studies on clonal evolution in cancer have primarily focused
crucial role in treatment resistance and cancer progression8,9. These on genomic data. Genetic research in breast cancer has explored transitions
advancements in our understanding could pave the way for more effective from ductal carcinoma in situ to invasive cancer or progression from the
and targeted cancer treatments in the future. primary site to lymphatic metastasis in the axilla15–19. However, genomic

1
Cancer Cell Diversity Project, NEXT-Ganken Program, Japanese Foundation for Cancer Research, Tokyo, Japan. 2Breast Surgical Oncology, Breast Oncology
Center, Cancer Institute Hospital, Japanese Foundation for Cancer Research, Tokyo, Japan. 3Division of Pathology, Cancer Institute, Japanese Foundation for
Cancer Research, Tokyo, Japan. 4Breast Medical Oncology, Breast Oncology Center, Cancer Institute Hospital, Japanese Foundation for Cancer Research,
Tokyo, Japan. 5Project for Cancer Epigenomics, Cancer Institute, Japanese Foundation for Cancer Research, Tokyo, Japan. 6Pathology Project for Molecular
Targets, Cancer Institute, Japanese Foundation for Cancer Research, Tokyo, Japan. 7Department of Molecular
Biology, Sapporo Medical University School of Medicine, Sapporo, Japan. 8Director’s room, Cancer Institute,
Japanese Foundation for Cancer Research, Tokyo, Japan. e-mail: reo.maruyama@jfcr.or.jp

npj Precision Oncology | (2024)8:222 1

https://doi.org/10.1038/s41698-024-00723-6 Article

analysis evaluates clonal patterns and does not fully assess the functional significant upregulation of VIM (an EMT marker) and CD44 (a cancer
characteristics of those clones or the causes of genetic alterations. Single‐cell stemness marker). Reduced expression of TACSTD2 in the posttreatment
RNA-sequencing (scRNA-seq), on the contrary, enables us to interpret the samples (Fig. 1D, Supplementary Figs. 3D, E) might be clinically significant,
significance of expressional changes and estimate copy numbers with effi- as this gene encodes TROP2, the target protein of the therapeutic agent
cient tools like inferCNV20 at single-cell resolution. datopotamab-deruxtecan (Dato-DXd)24, used as the third-line drug treat-
Whether analyzing the clonal evolution of cancer or examining its ment in this case (Supplementary Fig. 1). However, no report has directly
plasticity, it is challenging to clarify these processes with a sample at a single demonstrated an association between TACSTD2 gene expression and Dato-
time and location. Longitudinal observations of cancer progression are Dxd therapeutic efficacy. Expression levels of ESR1 and FOXA1 were
crucial; however, especially in human clinical samples, collecting samples markedly decreased after treatment, although no statistically significant
from the same patient’s multiple lesions over time presents significant difference from pretreatment was observed, likely because of the initially low
challenges. While it is relatively easier to access biopsy samples from pri- expression levels (Supplementary Fig. 3D, E). This observation appears to be
mary tumors, approaching metastatic sites in other organs remains difficult. closely related to immunohistopathology showing loss of estrogen and
In close collaboration with clinicians, we have managed to undertake progesterone receptor expression in resected samples after drug treatment
temporal and spatial multi-sampling from a patient with Stage IV breast compared with pretreatment biopsy samples (Supplementary Table 1).
cancer. The primary tumor at diagnosis exhibited a heterogeneous popu- Differential gene expression analysis also revealed that CALCA, SNAI2,
lation of clones, one of which was estimated to have resisted drug treatment and BMP5 were upregulated, whereas KLK6, KLK7, LTF, and TACSTD2
and subsequently proliferated. Furthermore, a specific clone that emerged in were downregulated in cancer cells from posttreatment samples (Fig. 1D,
the primary tumor following treatment resembled clones at distant meta- Supplementary Data 1). Pathway enrichment analysis of these genes
static sites. These unique clones also expressed cancer stem cell markers and revealed increased expression levels of genes related to HALLMARK_E-
genes related to epithelial–mesenchymal transition (EMT), as well as altered PITHELIAL_MESENCHYMAL_TRANSITION and decreased expression
activation of specific transcription factors (TFs). Collectively, these findings levels of genes related to HALLMARK_ESTROGEN_RESPONSE_EARLY
reveal the molecular evolution underpinning breast cancer progression, in posttreatment samples compared with those in the pretreatment speci-
drug resistance, and metastasis, and may offer clues for the development of men (Fig. 1E), in accordance with a previous study reporting association
more effective therapeutic strategies. between these hallmark processes and treatment efficacy25. Close inspection
revealed some overlap between terms enriched positively and negatively in
Results the posttreatment cancer cells (Fig. 1E). However, despite the overlap, the
scRNA-seq analysis of the four merged samples revealed sig- composition of the genes was different before and after treatment (Fig. 1F
nificant changes in cancer characteristics after initial and Supplementary Fig. 3F).
chemotherapy
The clinical specimens analyzed in this study were collected during the Copy number variation analysis revealed a small population of
lactation period from a 40-year-old patient with de novo stage IV breast cancer cells within the primary tumor that exhibited drug treat-
cancer. The first sample, referred to as “Pre,” was obtained from biopsy ment resistance
specimens of the breast tumor collected for diagnostic purposes at the initial To investigate clonal heterogeneity in our samples, we estimated copy
consultation. The second and third samples, named “Post1” and “Post2,” number variations (CNVs) in each sample using the R package inferCNV20
were specimens collected during surgical resection for the local control of (Supplementary Figs. 4A–D). The CNV patterns inferred from the scRNA-
rapidly growing breast cancer despite ongoing drug therapy. The fourth seq data were closely aligned with the results from CNV profiling analysis
sample, “Meta,” was obtained from surgical specimens of a newly developed performed using genomic DNA sequencing data from the same samples
metastatic lesion in the peritoneum, 6 months after the surgery of the (Supplementary Figs. 4E, F). The cancer clusters C1, C4, C6, C9, and C14 of
primary tumor. We conducted scRNA-seq on these four samples using the Pre (labeled Pre-C1, -C4, -C6, -C9, and -C14, respectively, in Fig. 2A and
10x Chromium platform, performing analyses individually and in an Supplementary Fig. 2A) demonstrated some shared CNVs, such as the
integrated manner (Fig. 1A). Detailed clinical outcomes are presented in overall loss of chromosome 6 and gain of the chromosome 8 long arm, but
Supplementary Fig. 1 and Methods section, and the pathological diagnostic there were differences in chromosomes 3, 5 and 9 (Fig. 2B). The tumor
results for each sample are listed in Supplementary Table 1. subclustering method of inferCNV segregated the pretreatment cancer cell
Gene expression profiles of the four clinical samples were obtained populations into four distinct clonal subgroups (Pre-CNV1–4). Notably, the
from scRNA-seq using Seurat21 and Uniform Manifold Approximation and majority of cells in clone Pre-CNV1 were from UMAP clusters Pre-C0, -C9,
Projection (UMAP) plots were generated for visualization (Supplementary and -C14, those in Pre-CNV3 and 4 mainly from Pre-C4, and Pre-CNV2
Fig. 2A). Subsequently, a total of 16,368 cells from the four samples with primarily from Pre-C6 cells (Fig. 2C and Supplementary Fig. 5A).
sufficient quality reads were first simply merged and plotted on UMAP While most cancer cells in Pre exhibited copy number gain in the long
(Supplementary Fig. 3A). We then performed data integration using an R arm of chromosome 11 (Fig. 2B and Supplementary Fig. 4A), almost all
package STACAS22,23 based on the library production batches for each posttreatment specimens demonstrated copy number loss within this
sample (Fig. 1B, C, Supplementary Fig. 3B). To clarify the distinction genomic region (Supplementary Fig. 4B–D). However, upon closer exam-
between clusters before and after data integration, we designated clusters ination, a subset of cells within Pre-C9 also exhibited copy number loss in
before integration as “All-Unintegrated-C” followed by a number and those the long arm of chromosome 11 (Fig. 2B). To further analyze the clinical
after integration as “All-Integrated-C” with a subsequent number. Cell-type significance of this pretreatment change in copy number profile, we
annotation confirmed that UMAP clustering effectively depicted cell type extracted Pre-C9 cells and repeated the clustering and inferCNV. Again,
variation within the tumors (Fig. 1B, Supplementary Figs. 2A, B, 3A, C). only a subset of the cancer cells in Pre-C9 exhibited loss on the long arm of
Focusing on the cancer cells, the integrated clusters formed two distinct chromosome 11, and the entire CNV pattern in this small subset was very
groups, one comprising mainly clusters All-Integrated-C3 and -C14 from similar to that of Post2 and a subpopulation of Post1 (Fig. 2D), indicating
Pre and another including clusters All-Integrated-C1, -C2, -C5, -C6, -C9, that a subset of cells in Pre-C9 survived drug treatment and expanded
and -C10 from the posttreatment specimens (Post1, Post2, and Meta) (Fig. further. We named this treatment-resistant cluster as Pre-C9R, while
1B, C, Supplementary Fig. 3B). The separation of pre- and posttreatment categorizing the remaining treatment-sensitive cells as Pre-C9S, and this
samples in the plot indicates that the initial drug treatments markedly information was overlaid on the UMAP plot in Fig. 2A (Fig. 2E). Using
altered cancer cell expression characteristics. For instance, when extracting CNV scores with single-cell resolution obtained through inferCNV, we
only cancer cells, posttreatment samples exhibited significantly decreased calculated the differences in gene-specific scores between Pre-C9R and other
expression levels of the marker genes EPCAM and TACSTD2 as well as clusters. We then assessed how genes with significant CNV score differences

npj Precision Oncology | (2024)8:222 2

https://doi.org/10.1038/s41698-024-00723-6 Article

B
D
Pretreatment Posttreatment

C
UMAP2

UMAP1

E F

y
n

wa
tio

th
g n o in te nsi

Pa
An ge Ph al La ra

e on

es
tro ve ign se l T

r ly
ea p is s ti

ci
Ea
Es ati a S on yma

R Res es on ryla

pe
U iog Re sph g

si s S
xi et s h

V en s o
O −b Re enc

Ap og Ox Dn
to esi en
p

op en yg
F n s

ip e e
TG ge Me

Ad tiv ns o

Po t1
M t2
tro ial

a
s
s
et
e
Es hel

Po
Pr
c
d
it
Ep

Sample
CD44
VCAN
SNAI2
BGN
CALU
LOX
FN1
CYP26B1
SLC7A5
COX6C
FABP5
ID1
JUNB
ID3
UQCRQ
COX4I1
ATP5PD
RHOBTB3
FGFR1
IGFBP5
PRDX2
APOE
CHCHD10
Enrichment −log10(p−value) Scaled Expression

0 2 4 6 0 0.5 1

Fig. 1 | Experimental scheme, and changes in breast cancer cell gene expression gene expression in cancer cells before and after chemotherapy. E Pathway enrich-
induced by chemotherapy. A Schema of sample naming, processing, and analysis. ment analysis of these differentially expressed genes (D). F Clustergram of positively
(Created by BioRender) B, C UMAP plots for merged cells after integration, color- enriched terms in posttreatment cancer cells (left) and heatmap displaying the
coded by cluster cell type (B) and sample (C). D Volcano plot comparing differential average expression of genes related to enriched terms in cancer cells of each sample.

npj Precision Oncology | (2024)8:222 3

https://doi.org/10.1038/s41698-024-00723-6 Article

A B

Chromosome

10
11

12
13
14
15
16
17
18
19
20
21
22
X
1

3
4
5
6
7
8
9
UMAP2

V1
N
C
e-
Pr
UMAP1

V2
N
C
e-
Pr

3
N NV
e- e - C
V4
Pr P r
C
Seurat Clusters
UMAP2

Post2
Chromosome 11
UMAP1

D
Chromosome 11
UMAP2

UMAP1 Post1
Chromosome 11

Seurat Clusters

E
G
UMAP2

UMAP1

F H
HE E-Cadherin N-Cadherin Vimentin

I 11q22.1/CEP11
UMAP2

UMAP1

correspond to genes specifically expressed in Pre-C9R compared to other identified as DEGs that met the criteria of an adjusted P value < 0.05 and
clusters. The results revealed that there were 1,153 genes with a mean log2FC < −1.0 (Supplementary Data 2). After merging Pre and Post1, we
difference (MeanDiff) > 0 and significant score differences (−log10 False reperformed inferCNV analysis with annotations based on each sample’s
Discovery Rate (FDR) > 2.0); of these, 47 genes met the criteria for differ- original clusters (Seurat-identified Pre and Post1 clusters), identifying four
entially expressed genes (DEGs) with an adjusted P value < 0.05 and log2 new clones: Pre/Post1-CNV1–4. Unlike other Pre clusters, most cells in Pre-
Fold Change (FC) > 1.0. Conversely, among the 814 genes with a MeanDiff C9R were classified into the same clone Pre/Post1-CNV3 as cells in some
< 0 and significant score differences (−log10FDR > 2.0), 29 genes were Post1 clusters (Supplementary Fig. 5B and C). Notably, Post1-C6 cancer

npj Precision Oncology | (2024)8:222 4

https://doi.org/10.1038/s41698-024-00723-6 Article

Fig. 2 | Genomic, transcriptomic, and pathological characteristics in cancer cells Post2) specimens (right). E UMAP plot of cancer cells of Pre, overlayed the infor-
of Pre sample. A UMAP plot of cancer cell clustering of Pre (numbering of Seurat- mation of Pre-C9S and -C9R onto that in A. F UMAP of integrated cancer cells
identified Pre clusters is based on the UMAP clustering in Supplementary Fig. 2A). displaying Pre clusters, including -C9S and -C9R. G Violin plot and box plot of gene
B Heatmap of cancer cell CNV signals by inferCNV in Pre. Horizontal lines divide signature scores in each Pre cluster. Box plots represent the following: center line,
subclones defined by inferCNV. A subset of cells mostly comprised of Pre-C9 are median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range;
highlighted with a red square. C UMAP plot of cancer cells in Pre, color-coded points, outliers. EMT, epithelial–mesenchymal transition; ER-Early, estrogen
according to CNV clone. D Comparison of CNV patterns in cancer from primary response early. H Hematoxylin and eosin staining and immunohistochemical
site before and after drug therapy. UMAP plot (left) and heatmap of CNV signals analysis of clinical biopsy specimens of Pre. Scale bar: 250 µm. I Fluorescence in situ
(middle) after extracting and reclustering only cells from Pre-C9 in A. Cells with loss hybridization analysis of clinical biopsy specimens from Pre. Image shows the loss of
on the long arm of chromosome 11 correspond mainly to cells in newly formed 11q22.1 (spectrum red signals) in some cancer cells.
cluster C2 and show CNV patterns (blue bar) similar to most surgical (Post1 and

cells did not align with the cancer cell clones of Pre. When displaying Pre posttreatment samples. As shown in Fig. 2D, the Post2 sample consisted
clusters on the UMAP of integrated cancer cells in Fig. 1B and C using the R primarily of uniform clones, whereas Post1 exhibited varying copy number
package SCpubr26, only cells in Pre-C9R were distributed to All-Integrated- patterns. Notably, some CNV patterns in the Post1 sample closely matched
C1 and -C2, which mainly comprised cancer cells from Post1 and Post2 (Fig. that in the distant metastasis sample Meta (Supplementary Figs. 4B, D).
2F and Supplementary Fig. 5D). These findings indicate that Pre-C9R cells, To elucidate cancer heterogeneity and metastatic recurrence
genomically similar to posttreatment cancer cells, also closely resemble mechanisms, we will now focus primarily on Post1. UMAP clustering
Post1 and Post2 in terms of gene expression. Overall, cells in the Pre-C9R classified Post1 cancer cells into four clusters, namely, Post1-C2, -C3, -C4,
cluster can be classified as drug-tolerant persister cells (DTPs)27–30 in this and -C6 (Fig. 3A). Meanwhile, the inferCNV tumor subclustering method
breast cancer case. divided the Post1 sample into four major clones, Post1-CNV1–4 (Fig. 3B).
Comparing the expression of several marker genes among Pre clusters, Of these, Post1-CNV3 and -CNV4 were almost identical to Post1-C6,
first, the expression levels of epithelial marker genes EPCAM and TACSTD2 whereas Post1-CNV1 and 2 did not overlap with any UMAP clusters (Fig.
were significantly reduced in Pre-C9R compared with cells in other clusters 3C and Supplementary Fig. 6A). We compared inferCNV-derived gene-
(Supplementary Fig. 5E, Supplementary Data 3). These findings strongly specific CNV score differences and transcriptomic DEGs between Post1-C6
indicate that the expression of the Dato-DXd therapeutic target gene and other clusters in Post1. Among 3644 genes with a MeanDiff > 0 and
TACSTD2 was already lost in DTPs before initiation of treatment rather than significant CNV score differences (−log10FDR > 2.0), 244 genes met the
downregulated during the treatment course. In other words, the primary DEG criteria with an adjusted P value < 0.05 and log2FC > 1.0. Conversely,
tumor included a cell population that was potentially less sensitive or resistant of the 1347 genes with a MeanDiff < 0 and significant CNV score differences,
to Dato-DXd before treatment. When scoring Pre clusters based on three 296 satisfied the DEG criteria with an adjusted P value < 0.05 and log2FC <
epithelial marker gene types in normal mammary glands31 (Supplementary −1.0 (Supplementary Data 5). We combined Post1 and Meta samples and
Data 4), the expression of genes associated with luminal hormone-sensing (L- performed inferCNV analysis, annotating based on each sample’s original
Hor) cells and luminal alveolar (L-Alv) cells was significantly lower in Pre- clusters (Seurat-identified Post1 and Meta clusters), resulting in three new
C9R than in the other clusters (Supplementary Fig. 5F), consistent with clones: Post1/Meta-CNV1–3. In contrast to other Post1 clusters, most cells
reduced expression of luminal markers KRT8 and 18 in Pre-C9R (Supple- in Post1-C6 were categorized into Post1/Meta-CNV2, identical to cells in
mentary Fig. 5E, Supplementary Data 3). However, the basal markers KRT5 Meta clusters (Supplementary Fig. 6B, C). When Post1 clusters and Post1-
and KRT14 were also downregulated in Pre-C9R compared with Pre-C9S CNV clones were displayed alongside Meta cancer cells on the UMAP of
(Supplementary Fig. 5E), despite a high basal score in Pre-C9R (Supple- integrated cancer cells in Fig. 1B and C using SCpubr, cells from Post1-C6
mentary Fig. 5F), indicating a substantial loss of the epithelial nature of cancer and Post1-CNV3/4 almost overlapped those from Meta (Fig. 3E and Sup-
cells in Pre-C9R. Moreover, some EMT markers VIM, COL6A1, and SNAI2, plementary Fig. 6D). These genomic and transcriptomic similarities
as well as the stem cell marker CD44 were upregulated in the Pre-C9R cluster, strongly indicate that cancer cells in Post1-C6 and Post1-CNV3/4 acquired
strongly indicating that these potentially drug-resistant cells had undergone metastatic capacity in the primary tumor and ultimately spread to sub-
EMT and acquired stem cell characteristics before treatment (Supplementary sequent metastatic lesions in the mesentery. We termed these cells meta-
Fig. 5E, Supplementary Data 3). Scorings of gene signatures showed that, in static precursor cells (MPCs).
Pre-C9R cancer cells, the HALLMARK_EPITHELIAL_MESENCHY- Comparing the expression of several marker genes among Post1
MAL_TRANSITION score was higher and the HALLMARK_ESTRO- clusters and Post1-CNV clones, we observed that the expression levels of
GEN_RESPONSE_EARLY score was lower than those of the other cells in EPCAM and certain keratin-related genes were higher in MPCs than in
the pretreatment samples (Fig. 2G), which are patterns very similar to those of other cancer cells in Post1, although statistically unsignificant. Conversely,
cancer cells in posttreatment samples (Fig. 1E, F). These findings provide some keratin-related genes, such as KRT14 and 19, were significantly down-
supporting evidence that cells in the Pre-C9R cluster served as the precursors expressed in MPCs (Supplementary Fig. 6E, Supplementary Data 6 and 7).
for the subsequently re-emerging primary tumor. When scoring Post1 clusters and Post1-CNV clones based on the three
To validate the findings from scRNA-seq analysis, we performed epithelial marker gene types in normal mammary glands, MPCs tended to
immunohistochemistry (IHC) staining and fluorescence in situ hybridiza- exhibit higher scores for all three breast epithelial markers compared with
tion (FISH) on clinical biopsy specimens of Pre. IHC staining revealed other cancer cells in Post1. In MPCs, Post1-CNV4 showed significantly
expression of the EMT markers N-cadherin and vimentin in a small subset higher L-Hor and L-Alv scores than Post1-CNV3 (Supplementary Fig. 6F).
of cancer cells from biopsy samples, indicating EMT-like changes in these DEGs were subsequently identified for each Post1-CNV clone
cells (Fig. 2H). FISH analysis of the biopsy specimens showed two signals of (according to thresholds log2FC > 1.0 and P value < 0.05) (Supplementary
the chromosome 11 centromere (green) and only one signal of 11q22.1 (red) Fig. 7A, Supplementary Data 7) and subjected to enrichment analysis. The
in a few cells, indicating loss of the long arm of chromosome 11 in these most prominent enriched pathways in Post1-CNV1 were associated with
cancer cells (Fig. 2I). HALLMARK_OXIDATIVE_PHOSPHORYLATION, whereas the most
predominant pathways in Post1-CNV2 were related to HALLMAR-
Exposure to hypoxia within the primary tumor during drug K_HYPOXIA, and those in Post1-CNV3 and -CNV4 to HALLMARK_E-
treatment conferred metastasis capacity to some cancer cells PITHELIAL_MESENCHYMAL_TRANSITION (Supplementary Fig. 7B).
To examine the molecular mechanisms conferring cancer cell metastatic Gene signatures related to EMT were expressed at much higher levels in
capacity, we conducted similar CNV and clustering analyses of MPCs, whereas hypoxia-related genes were upregulated as much or more in

npj Precision Oncology | (2024)8:222 5

https://doi.org/10.1038/s41698-024-00723-6 Article

A B

Chromosome

10
11

12
13
14
15
16
17
18
19
20
21
22
X
1

3
4
5
6
7
8
9
V2
N
C
1-
UMAP2

st
Po
UMAP1

V1
C

N
C
1-
st
Po

V3 4
N NV
1- - C
st s t 1
C
Po Po
Post1 Seurat Clusters
UMAP2

D Post1 Meta
Chromosome 5 17 Chromosome 5 17
UMAP1
V3 4
N NV
1- -C
st st1

Post1 Seurat Cluster

C
Po Po

E F
UMAP2

UMAP1

G H I
Spot A
HE CA IX GLUT-1
5q22.1/CEP5

E-Cadherin N-Cadherin Vimentin

Spot B
HE CA IX GLUT-1 5q22.1/CEP5

E-Cadherin N-Cadherin Vimentin

MPCs than in Post1-CNV2 (Fig. 3F), indicating that Post1-CNV2, 3, and 4 upregulated as the DEGs of Post1-CNV2, and their expression was higher in
may have evolved from Post1-CNV1 under the strong influence of hypoxia. Post1-CNV2 than in MPCs, indicating that Post1-CNV2 may contribute to
The second most enriched term in Post1-CNV3 and the third most enriched angiogenesis within the primary tumor, facilitating metastasis indirectly by
term in Post1-CNV4 were related to angiogenesis, and MPCs scored higher aiding MPC infiltration into blood vessels.
in angiogenesis than other cancer cells in Post1 (Supplementary Fig. 7B, C). According to cell-to-cell communication analysis conducted using the
Meanwhile, the genes related to angiogenesis, VEGFA, and ADM, were R package Connectome32, several cell–cell interactions were identified

npj Precision Oncology | (2024)8:222 6

https://doi.org/10.1038/s41698-024-00723-6 Article

Fig. 3 | Genomic, transcriptomic, and pathological characteristics in cancer cells interquartile range; points, outliers. EMT, epithelial-mesenchymal transition.
of Pre sample. A UMAP plot of Post1 cancer cells. The numbering of Seurat- G Circos plots visualizing the cell–cell connectivity among Post1-CNV clones.
identified Post1 clusters is based on the clustering of Post1 in Supplementary Fig. 2A. Ligands occupy the lower semicircle, and corresponding receptors the
B Heatmap of Post1 cancer cell CNV signals by inferCNV. Horizontal lines divide upper semicircle. Subclones are color-coded by edge, ligand, and receptor.
subclones defined by inferCNV. C UMAP plot of Post1 cancer cells color-coded H Hematoxylin and eosin staining and immunohistochemical analysis of clinical
according to CNV clone. D Heatmap of CNV signals of Post1 (left) and Meta (right). surgically resected specimens from Post1 at two locations on the same slide: Spot A
Cells with loss on the long arm of chromosomes 5 and 17 correspond mainly to (upper) and Spot B (lower). Scale bar: 250 µm. I Fluorescence in situ hybridization
Post1-CNV3 and -CNV4, and show CNV patterns (red bar) similar to those of Meta. analysis of clinical surgically resected specimens from Post1 at two locations on the
E UMAP of integrated cancer cells displaying Seurat-identified Post1 clusters plus same slide: Spot A (upper) and Spot B (lower). At Spot A, cancer cells showed normal
Meta (upper) and Post1-CNV clones plus Meta (lower). F Violin plot and box plot of copy numbers with two spectrum red signals (5q22.1) and two spectrum green
gene signature scores in each Post1-CNV clone. Box plots represent the following: signals (CEP5). At Spot B, cancer cells exhibited the loss of 5q22.1 (spectrum red
center line, median; box limits, upper and lower quartiles; whiskers, 1.5× signals).

among Post1-CNV2, 3, and 4 (Fig. 3G; Supplementary Fig. 8A–C), whereas during breast cancer progression. These genes and associated signaling
minimal interactions were observed between Post1-CNV1 and each of the pathways are shown in Fig. 5A. Among the cancer clusters obtained from
other three clones. These findings indicate that Post1-CNV2 and MPCs Post1 and Meta integration (Fig. 4A), the expression of CALCA and CALCB
form an independent cell population within Post1 tumors. was markedly elevated in Post1/Meta-Integrated-C2 and -C3, ADM was
To validate the scRNA-seq analysis results, we performed IHC staining highly expressed in -C3 and -C4, and CALCRL expression was high expli-
and FISH on surgically resected clinical specimens of Post1 in two distinct citly in -C3 (Fig. 5B). The expression levels of RAMP1/2/3 and ACKR3 were
areas (Spots A and B) on the same slide. Hypoxia markers CA IX and Glut-1 generally low (Fig. 5B), and the expression levels of CALCR and MRGPRX2
showed diffusely high expression in spots A and B. However, protein levels were undetected in all samples (data not shown).
associated with EMT and signaling on the long arm of chromosome 5 To investigate the coexpression of calcitonin-related genes, specifically
exhibited significant differences between the two areas. In Spot A cancer CALCA, ADM, and CALCRL, within individual cells, we performed a cor-
cells showed high E-cadherin expression but minimal N-cadherin and relation analysis of their expression at the single-cell level across samples,
vimentin expression. FISH analysis revealed two pairs of signals at 5q22.1 Seurat-identified Post1 clusters, and Post1-CNV clones. Given that scRNA-
(colored red) in most cancer cells in this area. Conversely, the cancer cells in seq is susceptible to pseudo-negative results in gene expression because of
Spot B showed low E-cadherin expression levels and higher N-cadherin and limited sequencing depth34,35, which can underestimate correlation assess-
vimentin levels. Additionally, FISH analysis indicated the loss of one of the ments, we smoothed the data using the Rmagic package36 in advance.
5q22.1 signals (red) in many cells. Combining the results from single-cell According to the correlation analysis per cell, we found a positive correlation
analysis with pathology validations indicated that Spot A primarily com- between CALCA and CALCRL expression levels in Post1-CNV3/4 and
prised cancer cells with Post1-CNV1 and -CNV2, whereas MPCs were Meta. In contrast, expression levels of ADM and CALCA, and those of ADM
found in Spot B (Fig. 3H, I). and CALCRL, were negatively correlated in Post1-CNV3/4, and Meta
(Supplementary Fig. 10A, B). These findings indicate that while many
The activities of transcription factors, including ATF3 and SNAI2, cancer cells in MPCs and Meta coexpressed CALCA and CALCRL, simul-
appeared altered in MPCs taneous expression of ADM with CALCA or ADM with CALCRL was rare.
To examine the molecular mechanisms contributing to metastatic capacity, To examine the trajectory of these gene expression changes from MPC
we integrated the single-cell data of Post1 and Meta, and selectively isolated to Meta cell in greater detail, we performed pseudotime analysis of cancer
cancer cells from them (Fig. 4A, B, Supplementary Fig. 9A). The new cluster clusters obtained from Post1 and Meta integration (Fig. 4A) using R package
Post1/Meta-Integrated-C3, consisted mainly of cells in the G2M and S Monocle 337–39. This time, only G1-stage cancer cells were included in the
phases of the cell cycle (Supplementary Fig. 9B). Post1 MPCs and Meta- analysis to remove cell cycle effects (Fig. 5C, D). Pseudo-temporal kinetic
derived cells were plotted in nearly identical positions on the UMAP (Fig. analysis revealed that, while ADM expression levels decreased at the end of
4C), though cancer cells in Post1/Meta-Integrated-C9 were mostly derived the trajectory, CALCA, CALCB, and CALCRL expression levels increased
from Post1-C6, while approximately two-thirds of cells in Post1/Meta- almost simultaneously from MPCs to Meta. Furthermore, RAMP1
Integrated-C4 were from Meta-derived cells (Fig. 4A, C and Supplementary expression was practically absent in MPCs, but elevated in metastases at the
Fig. 9C). end of pseudotime (Fig. 5B, E, and F).
To explore the gene expression regulation mechanisms in MPCs, we
calculated TF activity alteration scores for integrated cancer cell data from Discussion
Post1 and Meta using the R package BITFAM33. Based on a heatmap of TF Since the widespread adoption of next-generation sequencing, there has
target scores for each cancer cell grouped by Post1/Meta-Integrated cancer been considerable progress in studies on the clonal evolution of breast
clusters (Fig. 4D), several TFs showed altered activation in a cell subgroup cancer, mainly through genomic analysis, reaching a resolution down to the
suspected of cancer metastasis (Post1/Meta-Integrated-C4 and -C9 in Fig. single-cell level15,16. Although genomic mutations provide critical insights
4A). Among the cluster-specific TF activities inferred through the BITFAM into cancer’s evolution, it is challenging to fully understand the specific
algorithm, the activities of ATF3, SNAI2, KLF4, and EGR1 were highly functional changes and variations in cancer phenotypes caused by these
altered in Post1/Meta-Integrated-C9, while those of SPI1 and SPIB was mutations. scRNA-seq allows for detailed analysis of expression differences
altered mainly in Post1/Meta-Integrated-C4 (Fig. 4D, E). These variations in at the cellular level, deepening our understanding of functional abnormal-
TF activity indicate dynamic changes in the gene expression regulation ities in cancer cells, including breast cancer40–53. Moreover, tools like
landscape during cancer progression, from the formation of MPCs in the inferCNV also enable the estimation of genomic changes, adding depth to
primary tumor to circulation in the bloodstream and eventual infiltration our understanding of clonal dynamics. However, integrated analysis using
and growth at distant sites. multiple samples necessitates cautious interpretation due to considerable
variability in gene expression among patients. Also, it is crucial to compare
Cancer cells with metastatic potential exhibited upregulated different lesions within the same patient when studying cancer clonal
expression of calcitonin gene-related peptide receptors evolution with scRNA-seq.
Finally, we investigated the impact of calcitonin-related gene upregulation In this study, we analyzed multiple samples from a single patient both
on metastasis as the expression levels of these genes, including CALCA, were temporally and spatially, avoiding issues of interpatient variability and
elevated in posttreatment cancer cells (Fig. 1D) but have not been examined providing a detailed overview of the cancer’s functional evolution by

npj Precision Oncology | (2024)8:222 7

https://doi.org/10.1038/s41698-024-00723-6 Article

A C
UMAP2

UMAP1
B

UMAP2
UMAP1
UMAP2

UMAP1

D C3 C10C12 C6 C2 C9 C4
E
Integrated Clusters
Post1−CNV&Meta
ZNF331
CREB3L1
MEF2C
NFIA
ID1
BATF3
MYC
SPI1
SOX17
MYBL2
FOXP3
HES1
SPIB
ID3
ELF3
PBX1
KLF4
CREM
SNAI2
ATF3
NR2F1
NR2F2
UMAP2

EGR1
IRF1
NR4A1
MAF UMAP1
BATF

Activity Post1/Meta−Integrated Clusters Post1−CNV & Meta

C2 C4 C9 C12 Post1−CNV1 Post1−CNV3 Meta
0 0.25 0.5 0.75 1 C3 C6 C10 Post1−CNV2 Post1−CNV4

Fig. 4 | Transcription factors contributing to MPC transition and metastasis. alteration scores inferred from gene expression. TFs were vertically aligned by
A, B UMAP plots of integrated Post1 and Meta with only cancer cells. Cells are color- hierarchical clustering. Horizontal bars at the top of the heatmap depict cluster
coded according to cluster number (A), and sample (B). C UMAP of integrated numbers in A and Post1-CNV clones plus Meta. E TFs whose activity was highly
Post1- and Meta-cancer cells displaying Seurat-identified Post1 clusters plus Meta altered in Post1/Meta-Integrated-C4 and -C9 are featured on the UMAP plot.
(upper) and Post1-CNV clones plus Meta (lower). D Heatmap of the TF activity

comparing conditions before and after drug treatment and between primary were highly similar to those of the post-treatment primary tumor. scRNA-
and metastatic sites. This approach allowed us to detect particular cell types seq analysis revealed that the re-enlarged primary tumor was heterogeneous
like DPTs and MPCs, which are often difficult to identify with bulk analysis at both the expression and genomic levels. In the Post1 sample, the inferred
or when analyzing data integrated across many patients. scRNA-seq and CNV pattern was divided into four subclones, two of which (named Post1-
clustering analyses revealed that the primary breast tumor of our patient was CNV3/4 in Fig. 3, nearly synonymous with Post1-C6) almost precisely
highly heterogeneous at the genomic level even before treatment and con- matched the CNV pattern of the Meta sample. Conversely, the CNV pattern
tained a small number of cancer cells with basal-like features (within mostly of Post1-C6 did not match any cluster of pretreatment cancer cells. We
cells with luminal breast epithelial characteristics) that appeared to survive corroborated these findings by conducting inferCNV analyses on merged
drug treatment and expand, reforming the tumor. This small subpopulation scRNA-seq data from Pre + Post1 and Post1 + Meta, alongside individual
may correspond to previously reported DTP cells27–30 and exhibited gene sample-specific inferCNV assessments. Additionally, identical results were
expression profiles associated with EMT and stemness in the present case. achieved when inferCNV was applied to a merged dataset of all four sam-
Clinically, the patient initially responded to drug therapy; however, only the ples, although these data are not shown. Consequently, we estimated that in
primary tumor gradually enlarged. We speculate that the enlarged cancer the tumor at Post1, new clones originated from DTP-derived cells, and those
cells were DTPs existing before treatment (designated as Pre-C9R in Fig. 2). that acquired metastatic capability (Post1-CNV3/4 or Post1-C6) migrated
We demonstrated that the CNV pattern and gene expression of Pre-C9R to the mesentery around the time of primary lesion surgery (Fig. 6A).

npj Precision Oncology | (2024)8:222 8

https://doi.org/10.1038/s41698-024-00723-6 Article

A B

Gene Expression Level

Post1/Meta-Integrated Clusters
C D
F
UMAP2

UMAP1

E
UMAP2

UMAP1

Fig. 5 | Potential contributions of calcitonin signaling to progression from MPC which does not interact with the CLR but binds to MRGPX2 and ACKR365. (Created
to metastatic cell. A Genes, products, and their receptors for the human calcitonin by BioRender) B Violin plots visualizing the expression levels of calcitonin-related
family. CALCA, located on chromosome 11p15, encodes two hormones, calcitonin genes in integrated cancer clusters of Post1 and Meta (Fig. 4A). C, D Pseudotime
(CT) and calcitonin gene-related peptide (CGRP), produced through alternative analysis of cells in integrated cancer clusters of Post1 and Meta (Fig. 4A). Only cancer
mRNA processing54. CT binds to the calcitonin receptor (encoded by CALCR), while cells in G1 cell-cycle phase are contained in this analysis. C Developmental trajec-
CGRP binds to the calcitonin receptor-like receptor (CLR, coded by CALCRL). ADM tories (color-coded by Post1-CNV clones plus Meta). D Pseudotime values.
encodes adrenomedullin (AM), which shares structural similarities with CGRP and E Expression level changes of ADM, CALCA, CALCB, RAMP1, and RAMP2 in
also binds to the CLR. The specific ligand to which CLR binds is regulated by cancer cells of Post1 and Meta (only in G1 cell-cycle phase) during pseudotime
complex formation with receptor activity-modifying proteins. If RAMP1 forms a analysis, plotted on the UMAP. F Pseudo-temporal kinetics of ADM, CALCA,
complex with CLR, CGRP is the ligand; if RAMP2 or RAMP3 forms the complex, CALCB, RAMP1, and RAMP2 expression.
AM is the ligand64. In addition to AM, ADM encodes a protein called PAMP-12,

However, we cannot definitively prove that the enlarged primary site after metastases nearly undetectable in imaging. The only confirmed recurrence
treatment metastasized to Meta, considering that multiple mesenteric after primary tumor surgery was at the sampled mesenteric metastases,
metastases were already identified in imaging at initial diagnosis and the located in different areas from those at initial screenings. Since their surgical
sampled metastatic sites may have existed at that time. Nevertheless, the removal, no further distant metastases have been found in this patient.
clinical course supports our hypothesis. Drug treatment rendered distant Additionally, axillary lymph nodes excised for sampling during primary

npj Precision Oncology | (2024)8:222 9

https://doi.org/10.1038/s41698-024-00723-6 Article

Fig. 6 | Cancer progression estimation in the studied case. A Putative cancer cell have contributed to angiogenesis within the primary tumor by expressing ADM and
evolutionary pathways in the studied case. B Schematic representation illustrating VEGFA, thereby prompting MPC survival and migration within Post1-CNV3 and
the generation of subclones within Post1 and their respective roles in cancer pro- -CNV4. C Putative contributions of calcitonin gene-related peptide (CGRP)
gression. Post1-CNV2–4 may have originated from Post1-CNV1 owing to hypoxic secretion and expression of calcitonin receptor-like receptor (CLR) and RAMP1 in
conditions within the tumor core. Cancer cells within the Post1-CNV2 clone may breast cancer cell evolution. (Created by BioRender).

npj Precision Oncology | (2024)8:222 10

https://doi.org/10.1038/s41698-024-00723-6 Article

tumor surgery were pathologically undiagnosed as metastatic. It is plausible the changes in cancer cell biology induced by selection pressure and
that all cancers, including the axillary lymph nodes, metastasized de novo genomic instability.
and were susceptible to drug treatment and elimination. The major limitation of this study is that we analyzed only a single case;
Following drug treatment, multiple subclones were identified at the thus, these findings cannot be extrapolated to the broader breast cancer
primary tumor site, and hypoxia emerged as a factor in our analysis. population. Additionally, constraints related to sample collection include
Notably, the surgical specimen of the primary lesion was large and exhibited obtaining biopsy specimens solely from primary lesions at the patient’s first
extensive internal necrosis, indicating a hypoxic environment within the consultation. Distant metastases were not biopsied by clinicians before
tumor. Immunostaining of pathological specimens from the primary lesion treatment, rendering these sites inaccessible for analysis. Consequently, the
revealed diffuse expression of the hypoxia markers CA IX and Glut-1. Two outlined cancer progression pattern in this patient includes speculative
out of three subclones, Post1-CNV3 and 4, exhibited CNV profiles similar elements. Moreover, single-cell analyses typically offer a limited number of
or identical to a clone isolated from distant metastases, indicating the pre- cells for examination, indicating that our findings may not fully represent all
sence of specific cancer cells in the primary tumor with latent capacity to cancer cells within this patient. Additionally, there were no analyses of local
metastasize, which we named MPCs. MPCs’ properties closely resemble microenvironmental influences on cancer cell gene expression profiles.
those of cancer cells in metastatic sites in terms of copy number profiles, and Nonetheless, we identified a subpopulation of cells (MPCs) that may be
gene expression levels, including EMT characteristics. In pathological spe- critical for cancer metastasis but difficult to detect by tissue-level analysis of
cimens from primary lesion surgical samples, EMT marker expression of multiple patients. Understanding the characteristics of DTPs and MPCs
was highly heterogeneous. The other subclone, Post1-CNV2, although could lead to the development of treatments for targeted eradication, greatly
apparently lacking metastatic capacity, strongly expressed genes related to reducing the risk of metastasis. The reproducibility of our findings must be
angiogenesis, such as VEGFA and ADM, and thus may have supported confirmed by conducting the same single-cell analyses in multiple breast
metastasis by allowing migration of MPCs into the bloodstream (Fig. 6B). It cancer cases with diverse clinical outcomes.
remains uncertain whether hypoxia promoted the branching of clones. In summary, we compared scRNA-seq results from pretreatment
However, our findings indicate that the subclones Post1-CNV2–4 exhibited primary breast tumor, posttreatment primary breast tumor, and distant
notably higher hypoxia marker expression levels than the main clone Post1- tumors to identify gene expression and cell phenotype changes underlying
CNV1. Additionally, observable cell–cell communication occurred among the development of treatment resistance and metastasis. We identified
these subclones, whereas minimal communication with Post1-CNV1 was specialized cell populations with drug resistance at baseline and another
observed. These findings indicate that the newly branched subclones were with metastatic potential after drug treatment. Identifying cells contributing
closely associated with hypoxia and strongly interacted with each other. to poor therapeutic response and the associated mechanism of metastasis
Furthermore, there were dynamic changes in transcriptional modulator may aid in the development of targeted ablation treatments.
expression during cancer cell progression to metastasis, indicating epige-
nomic plasticity during this metastatic transition. Methods
In the course of breast cancer evolution, cells began to exhibit high Patient characteristics and sample collection
expression levels of calcitonin-related genes, including CALCA, although it Pretreatment and posttreatment specimens were obtained from a lactating
is unclear if this change was induced or emerged by chance mutation. patient with breast cancer aged 40 years at initial diagnosis. A midwife
Changes in the expression of calcitonin gene-related peptide (CGRP), a noticed a mass in the patient's left breast while instructing on lactation. She
splicing variant of calcitonin54, have been reported in various cancers, visited the Cancer Institute Hospital to undergo detailed examinations. A
including breast cancer55. Initially, in our case, posttreatment cancer cells did subsequent needle biopsy confirmed the tumor to be invasive ductal car-
not express receptors for CGRP; thus, CGRP could not influence cancer cell cinoma. Imagin tests revealed metastases in the axillary lymph nodes, liver,
behavior. However, the MPCs that emerged in the posttreatment primary lung, and peritoneum. The patient initially received paclitaxel and bev-
tumor began to express CALCRL (coding CGRP’s receptor, calcitonin acizumab treatment but discontinued after 1.5 months because of tumor
receptor-like receptor (CLR)44), potentially allowing CGRP signaling and progression at the primary site, lymph nodes, and liver, as revealed by CT
associated phenotypic changes. However, expression of RAMP1, which scan. Subsequently, carboplatin, gemcitabine, and pembrolizumab therapy
form complex with CLR54, was still largely absent in posttreatment primary were initiated, which resulted in significant tumor and metastasis reduction.
tumor cells, including MPCs. Surprisingly, as MPCs metastasized to distant However, after a few months, the primary tumor gradually regrew,
sites, RAMP1 was expressed (in addition to further CALCRL upregulation), prompting treatment cessation. The patient then participated in a clinical
indicating that these metastatic cancer cells had developed the capacity for trial for Dato-DXd, an antibody-drug conjugate targeting TROP2. How-
autocrine CGRP responses. This sequence of gene expression changes ever, the tumor continued to grow after the first dose, reaching 10 cm in
indicates that cancer cells can dynamically adapt to promote survival diameter, which required surgical resection of the left breast and axillary
(Fig. 6C). lymph nodes. After the mastectomy, the patient did not wish to undergo
In this study, we obtained tumor samples for gene expression and further treatment and opted for follow-up observation only. However, new
clonal analyses from the same patient because the phenotypic hetero- lesions were subsequently found in the peritoneum, necessitating surgical
geneity of tumors from different patients can obscure the unique mole- intervention. In total, four specimens were collected from the patient for
cular characteristics of small subpopulations relevant to tumor scRNA-seq: the pretreatment needle biopsy sample (Pre), a posttreatment
progression56,57. Conversely, when analyzing various samples from the surgical specimen near the central necrotic tissue (Post1), a sample con-
same patient, the background remains consistent, facilitating the detec- taining viable cancer cells collected near the tumor margins (Post2), and a
tion of important molecular similarities and differences among clones, surgical specimen of peritoneal metastasis (Meta). The patient gave written
even when based on single-cell analysis. This study confirms the dramatic informed consent before specimen collection. The protocol was approved
changes in tumor clonal structure induced by drug treatment due to by the institutional ethics committee of Cancer Institute Hospital, Japanese
differential sensitivity. Furthermore, we found a subpopulation of DTP- Foundation for Cancer Research (No. 2018–1168) in accordance with the
like cells with high basal drug resistance, including cells with high meta- ethical guidelines of the institutional ethics committee and with the 1964
static capacity (termed MPCs). Such findings could only be obtained by Declaration of Helsinki and its later amendments or comparable ethical
single-cell analysis at multiple stages of cancer development within the standards.
same patient. A distinctive feature of this study is the successful descrip-
tion of cancer progression in a patient through integration of omics Single-cell preparation from clinical specimens
analysis with clinical course and pathological data. The findings presented Biopsy specimens for clinical diagnosis and surgically resected tumor tissues
here indicate that effective cancer treatment strategies must account for were dissociated into single cells using a MACS Tumor Dissociation Kit and

npj Precision Oncology | (2024)8:222 11

https://doi.org/10.1038/s41698-024-00723-6 Article

a gentleMACS Dissociator (Miltenyi Biotec, North Rhine-Westphalia, Data integration

Germany) following the manufacturer’s instructions. Aliquoted cells were Library preparation and sequencing of samples were performed in three
resuspended in the freeze preservation solution Bambanker (CS-02-001, batches: Pre and Post2 in the first batch, Post1 in the second batch, and Meta
GCLTEC), and frozen at –80 °C. in the third batch. To avoid batch effects after merging multiple sample data,
we performed data integration using the R package STACAS (v2.2.2)22,23.
10x Genomics Chromium library construction and sequencing Briefly, after merging normalized data using the merge() function in Seurat,
We performed single-plex analysis of four clinical samples acquired, we segregated the object by batch information. Anchors were identified
respectively, by needle biopsy of the primary breast tumor prior to treat- using FindAnchors.STACAS (…, dims = 1:15, anchor.features = 1000). The
ment (Pre), surgical resection of the primary tumor after third-line drug integration order was calculated using SampleTree.STACAS(). Finally,
treatments (Post1 and Post2), and from surgical resection of distant integration was conducted with the anchors using the Inte-
metastases (Meta). In the initial batch, library preparations were conducted grateData.STACAS() function. Downstream analysis, graph-based clus-
for samples labeled as Pre (approximately six months post-cryopreserva- tering, visualization, and differential gene expression analyses were
tion) and Post2 (approximately one-week post-cryopreservation). Subse- performed using Seurat. To visualize cluster or CNV-clone information on
quently, the library for the Post1 sample (approximately two months post- the UMAP plot of integrated data, we employed the do_DimPlot() function
cryopreservation) was prepared in the second batch. Finally, in the third in the SCpubr package (v2.0.0)26.
batch, we prepared the library for the Meta sample (two days post-cryo-
preservation). Approximately 20,000 to 30,000 indexed cells per sample Breast epithelial cell signature score comparisons
were loaded in a single microfluidic channel of the 10x Genomics Chro- Normal breast epithelial cells are differentiated into three cell types,
mium system for single-cell capture and cDNA preparation according to “luminal hormone-sensing (L-Hor) cells,” “luminal-alveolar (L-Alv)
the single-cell 3’ v3 Protocol recommended by the manufacturer (10x cells,” and “basal cells.” We scored our samples’ cancer clusters with breast
Genomics, Pleasanton, CA, USA). The libraries were then sequenced on the epithelial-cell signatures using published gene sets generated from human
Illumina NextSeq 550 platform (Illumina, California, USA) with paired-end scRNAseq data31 and via the AddModuleScore() function in Seurat.
reads (read1, 28 bp; index1, 10 bp; index2, 10 bp; read2, 90 bp). Raw data Differences in scores within each cluster were detected using the Wilcoxon
from the 10x Genomic platform were processed by Cell Ranger (v6.1.2) and Rank Sum test.
mapped to the human reference genome GRCH38 (accession
GCA_000001405.15). Enrichment analysis and gene set scoring
Molecular Signature DataBase Hallmark 2020 (MSigDB Hallmark 2020)
Quality check and preprocessing of the single-cell RNA-seq data pathway enrichment analysis was performed using the enrichR package
Quality control, normalization, and unbiased clustering of single-cell (v3.2) interface of the Enrichr database59. For the top-ranked hallmarks,
transcriptomes were performed using R and Seurat package (v5.0.2)21, gene sets from MSigDB60 were downloaded, and gene signature expression
unless otherwise specified. As the first step, low-quality barcodes with a gene scores were calculated using the Seurat “AddModuleScore()” function to
count less than 400 (nFeature_RNA < 400), unique molecular identifier compare expression intensities among clusters and subclones. Differences in
count more than 30,000 to 40,000 (nCount_RNA > 30,000 or > 40,000), or scores within each cluster were detected using the Wilcoxon Rank Sum test.
mitochondrial genes fraction greater than 20–25% (percent.mt > 20% for
Pre, Post1, and Meta, and >25% for Post2) were removed. Transcript count Copy number variation prediction with inferCNV
matrices from high-quality scRNA-seq data were normalized to the total Somatic large-scale chromosomal CNV of each sample and some merged
number of counts per cell and multiplied by a scale factor of 10,000. The samples were calculated using the R package “inferCNV” (v1.15.3)20.
normalized values were subsequently natural-log transformed using the InferCNV is a tool used to deduce CNV from tumor single-cell RNA
Seurat “NormalizeData()” function, and a linear transformation was applied sequencing data, identifying signs of large-scale chromosomal CNV in
using the Seurat “ScaleData()” function. The impact of cell cycle-related somatic cells, such as expansions or deletions of whole chromosomes or
gene expression heterogeneity on subsequent analyses was eliminated by substantial chromosomal segments. By comparing these data to a reference
cell-cycle scoring according to canonical marker expression and regressing set of “normal” cells, the variation in gene expression across different regions
out these markers using Seurat. Principal component analysis was per- of the tumor genome can be analyzed to identify areas of chromosomal
formed using “RunPCA()”, and the top 2000 variable features were iden- amplification or deletion. Subsequently, the relative expression intensity
tified using the Seurat “FindVariableFeatures()” function and “vst” selection across each chromosome is visualized as a heatmap. A raw counts matrix,
method. Seurat standard clustering procedures were performed using annotation file, and gene/chromosome position file were prepared
“FindNeighbors()” and “FindClusters()” with 1-50 dimensions and a according to data requirements (https://github.com/broadinstitute/
resolution of 0.5. scRNA-seq datasets were then projected onto UMAP inferCNV). The detailed settings were as follows: cutoff = 0.1, clus-
embedding space using “RunUMAP()” with 1-50 dimensions. DEGs in ter_by_groups = F, and analysis_mode = “subcluster.” The inferCNV
each cluster were identified using the FindAllMarker() or FindMarker() package “random_trees” option was used for partitioning the hierarchical
function within the Seurat package, with corresponding P value determined clustering tree into subclusters. Fibroblasts and myeloid cells were selected
using the Wilcoxon Rank Sum test followed by a Bonferroni correction. as reference normal cells.

Cell-type annotation Cell-to-cell communication analysis

Following UMAP clustering, we annotated cell types for each cluster. To analyze and visualize ligand-receptor interactions among CVN
First, we identified marker genes for each cluster, setting |log2 (fold clones in Post1, we used the R package Connectome (v1.0.0)32. Briefly,
change)|> 1, and adjusted P value < 0.05. Additionally, we examined the we applied the CreateConnectome() function with a min.cells.per.ident
expression levels of established marker genes for each cluster as follows: cutoff of 75, followed by data filtering using the FilterConnectome()
breast epithelial markers EPCAM, KRT8, and KRT14; B lymphocyte function to include only edges with a ligand and receptor z-score
marker CD79A; T lymphocyte marker CD3D; natural killer lymphocyte exceeding 0.25. This Filtration narrowed the 38,752 edges down to 260.
markers KLRB1 and KLRD1; myeloid cell markers FCER1G, CD1C, and Centrality analysis was conducted across all signaling families by setting
CELC10A; fibroblast markers FAP, ACTA2, and MYL9; and endothelial the parameters as weight.attribute = “weight_sc” and group.by =
cell markers PECAM1 and THBD. Finally, the SingleR58 (v2.4.1) “mode.” The CircosPlot() and NetworkPlot() functions were used to
package was used for cell cluster annotation with the reference dataset visualize the interactions. In making circos plots, the top 5 signaling
HumanPrimaryCellAtlasData49. vectors for each cell–cell vector were selected.

npj Precision Oncology | (2024)8:222 12

https://doi.org/10.1038/s41698-024-00723-6 Article

Transcription factor activity estimation centromere 5 (CEP5, FITC) for Post1 surgical specimens. BAC clone
TF activities of posttreatment surgical specimens were estimated from DNA was extracted using PI-80X (Kurabo, Osaka, Japan) and fluores-
normalized scRNA-seq data using the R package BITFAM (v1.2.0) 33. A cently labeled using a nick translation kit (Abbott Molecular, Des Plaines,
heatmap of inferred TF activities was then generated using the Complex- IL, USA). Unstained 4-μm-thick FFPE sections were hybridized with
Heatmap package (v2.14.0). Some single-cell TF activity results were fluorescent DNA probes, and the hybridized slides were stained with 4,6-
merged into the Seurat object using the “AddMetaData()” function, and diamidino-2-phenylindole before being examined under a BZ-X800
plotted in UMAP space using the “FeaturePlot()” function. fluorescence microscope (KEYENCE). The names of the BAC clones used
are available upon request.
Correlation analysis for gene expression levels per cell
Before performing correlation analysis of gene expression levels per cell, we Data availability
used the MAGIC() function of the Rmagic package (2.0.3)36 with its default Processed scRNA-seq data have been deposited at GEO (Accession ID:
parameters to reduce the sparsity of single-cell data. Pearson correlation GSE264205).
coefficients were calculated between the expression of two selected genes
using the cor() function. Code availability
R code for reproducing the results can be found at https://github.com/
Pseudotime analysis KazutakaOtsuji/scRNA-StageIV_BC. Data analysis was performed using R
To examine genetic variation during the transition from MPCs to metastatic version 4.2.2, and Seurat version 5.0.2. Default variables and parameters
cancer cells, we used Post1/Meta-Integrated cancer cells in only G1 cell were used unless otherwise specified in the Method section.
phase to perform trajectory/pseudotime analysis with the Monocle 3
(v1.2.9) algorithm36–39. An inferred branched pseudotime trajectory was Received: 16 January 2024; Accepted: 24 September 2024;
constructed using the Monocle 3’s “learn_graph()” function. For ordering
the cells according to pseudotime, the point considered furthest from Post1/
Meta-Integrated-C4 or -C9 was defined as the root node. The “plot_gen- References
es_in_pseudotime()” function was used to display ADM, CALCA, CALCRL, 1. Parker, J. S. et al. Supervised risk predictor of breast cancer based on
RAMP1, and RAMP2 genes in the pseudotime trajectory. intrinsic subtypes. J. Clin. Oncol. 41, 4192–4199 (2023).
2. Kast, K. et al. Impact of breast cancer subtypes and patterns of
DNA extraction and copy number variation analysis metastasis on outcome. Breast Cancer Res Treat. 150,
Total genomic DNA was isolated from dissociated cells using the QIAamp 621–629 (2015).
UCP DNA Micro Kit (QIAGEN) following the manufacturer’s instructions. 3. Tarighati, E., Keivan, H. & Mahani, H. A review of prognostic and
The sequencing library was prepared using Nextera DNA Flex Library Prep predictive biomarkers in breast cancer. Clin. Exp. Med 23,
Kit (Illumina) and sequenced on a NovaSeq 6000 system (Illumina). To 1–16 (2023).
ensure quality control and remove adaptor sequences, we employed 4. Marusyk, A., Janiszewska, M. & Polyak, K. Intratumor heterogeneity:
TrimGalore (version 0.6.10; https://github.com/FelixKrueger/TrimGalore), The Rosetta stone of therapy resistance. Cancer Cell 37,
a Perl wrapper of Cutadapt and FastQC. Subsequently, reads were mapped 471–484 (2020).
to the hg38 genome using BWA MEM61. Copy number analysis was per- 5. Ramon, Y. C. S. et al. Clinical implications of intratumor heterogeneity:
formed using CNVkit62. challenges and opportunities. J. Mol. Med. 98, 161–177 (2020).
6. Ortega, M. A. et al. Using single-cell multiple omics approaches to
Correlation analysis of copy number variation levels resolve tumor heterogeneity. Clin. Transl. Med 6, 46 (2017).
We compared CNV scores obtained from inferCNV with those from 7. Nath, A. & Bild, A. H. Leveraging single-cell approaches in cancer
genomic DNA for each sample. Since the bulk DNA sequencing results precision medicine. Trends Cancer 7, 359–372 (2021).
provided by CNVkit were based on regions, genes overlapping these regions 8. Pang, L. et al. Single-cell integrative analysis reveals consensus
were annotated using the hg38 reference genome63 (TxDb.Hsa- cancer cell states and clinical relevance in breast cancer. Sci. Data 11,
piens.UCSC.hg38.knownGene) and formatted as log2-transformed cover- 289 (2024).
age at the gene level. Correlation plots were generated by mapping the genes 9. Inayatullah, M. et al. Basal-epithelial subpopulations underlie and
identified in the inferCNV results to those in the CNVkit output. predict chemotherapy resistance in triple-negative breast cancer.
We also compared CNV scores derived from inferCNV data between EMBO Mol. Med. 16, 823–853 (2024).
two groups: Pre-C9R and other cancer clusters, and Post1-C6 and other 10. Greaves, M. & Maley, C. C. Clonal evolution in cancer. Nature 481,
cancer clusters. We assessed the differences in scores per cell using the 306–313 (2012).
Wilcoxon rank-sum test. The false discovery rate (FDR) was calculated for 11. Dang, H. X. et al. The clonal evolution of metastatic colorectal cancer.
each gene. Sci. Adv. 6, eaay9691 (2020).
12. Biswas, A. & De, S. Drivers of dynamic intratumor heterogeneity and
Immunohistochemistry phenotypic plasticity. Am. J. Physiol. Cell Physiol. 320,
For ICH, the 4-μm-thick formalin-fixed paraffin-embedded (FFPE) sec- C750–C760 (2021).
tions of clinical samples were stained with antibodies against E-cadherin 13. Burkhardt, D. B., San Juan, B. P., Lock, J. G., Krishnaswamy, S. &
(NCH-38; 1:200 dilution; DAKO), N-cadherin (6G11; 1:100 dilution; Chaffer, C. L. Mapping phenotypic plasticity upon the cancer cell state
DAKO), Vimentin (V9; 1:4 dilution; DAKO), CA IX (Polyclonal; 1:500 landscape using manifold learning. Cancer Discov. 12,
dilution; abcam), and Glut-1 (Polyclonal; 1:100 dilution; IBL). 1847–1859 (2022).
14. Whiting, F. J. H., Househam, J., Baker, A. M., Sottoriva, A. & Graham,
FISH analysis T. A. Phenotypic noise and plasticity in cancer evolution. Trends Cell
To detect the characteristic genetic features of 11q loss in pretreatment Biol 34, 451–464 (2023).
biopsy sample and 5q loss in posttreatment surgical sample from the 15. Wang, K. et al. Archival single-cell genomics reveals persistent
primary site, we performed FISH assays using bacterial artificial chro- subclones during DCIS progression. Cell 186, 3968–3982
mosome (BAC) probes. Probes were designed as follows: targeting e3915 (2023).
ARHGAP42 at 11q22.1 (Texas Red) and centromere 11 (CEP11, FITC) 16. Nishimura, T. et al. Evolutionary histories of breast cancer and related
for Pre biopsy specimens; targeting APC at 5q22.1 (Texas Red) and clones. Nature 620, 607–614 (2023).

npj Precision Oncology | (2024)8:222 13

https://doi.org/10.1038/s41698-024-00723-6 Article

17. Kumar, T. et al. A spatially resolved single-cell genomic atlas of the 41. Ruan, H. et al. Single-cell RNA sequencing reveals the characteristics
adult human breast. Nature 620, 181–191 (2023). of cerebrospinal fluid tumour environment in breast cancer and lung
18. Lomakin, A. et al. Spatial genomics maps the structure, nature and cancer leptomeningeal metastases. Clin. Transl. Med 12, e885 (2022).
evolution of cancer clones. Nature 611, 594–602 (2022). 42. Liu, Y. et al. Intercellular communication reveals therapeutic potential
19. Harvey, J. M., Clark, G. M., Osborne, C. K. & Allred, D. C. Estrogen of epithelial-mesenchymal transition in triple-negative breast cancer.
receptor status by immunohistochemistry is superior to the ligand- Biomolecules 12, 1478 (2022).
binding assay for predicting response to adjuvant endocrine therapy 43. Gambardella, G. et al. A single-cell analysis of breast cancer cell lines
in breast cancer. J. Clin. Oncol. 17, 1474–1481 (1999). to study tumour heterogeneity and drug response. Nat. Commun. 13,
20. Tickle, T. I., Georgescu, C., Brown, M. & Haas, B. inferCNV of the 1714 (2022).
Trinity CTAT Project. https://github.com/broadinstitute/inferCNV. 44. Carpen, L. et al. A single-cell transcriptomic landscape of innate and
21. Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell adaptive intratumoral immunity in triple negative breast cancer during
184, 3573–3587 e3529 (2021). chemo- and immunotherapies. Cell Death Discov. 8, 106 (2022).
22. Andreatta, M. et al. Semi-supervised integration of single-cell 45. Zhou, S. et al. Single-cell RNA-seq dissects the intratumoral
transcriptomics data. Nat. Commun. 15, 872 (2024). heterogeneity of triple-negative breast cancer based on gene
23. Andreatta, M. & Carmona, S. J. STACAS: sub-type anchor correction regulatory networks. Mol. Ther. Nucleic Acids 23, 682–690 (2021).
for alignment in Seurat to integrate single-cell RNA-seq data. 46. Xu, K. et al. Single-cell RNA sequencing reveals cell heterogeneity and
Bioinformatics 37, 882–884 (2021). transcriptome profile of breast cancer lymph node metastasis.
24. Okajima, D. et al. Datopotamab Deruxtecan, a novel TROP2-directed Oncogenesis 10, 66 (2021).
antibody-drug conjugate, demonstrates potent antitumor activity by 47. Wu, S. Z. et al. A single-cell and spatially resolved atlas of human
efficient drug delivery to tumor cells. Mol. Cancer Ther. 20, breast cancers. Nat. Genet 53, 1334–1347 (2021).
2329–2340 (2021). 48. Vishnubalaji, R. & Alajez, N. M. Transcriptional landscape associated
25. He, B. et al. The prognostic landscape of interactive biological with TNBC resistance to neoadjuvant chemotherapy revealed by
processes presents treatment responses in cancer. EBioMedicine 41, single-cell RNA-seq. Mol. Ther. Oncolytics 23, 151–162 (2021).
120–133 (2019). 49. Ren, L. et al. Single cell RNA sequencing for breast cancer: present
26. Blanco-Carmona, E. Generating publication ready visualizations for and future. Cell Death Discov. 7, 104 (2021).
Single Cell transcriptomics using SCpubr. bioRxiv, 50. Pal, B. et al. A single-cell RNA expression atlas of normal,
2022.2002.2028.482303 (2022). preneoplastic and tumorigenic states in the human breast. EMBO J.
27. Pu, Y. et al. Drug-tolerant persister cells in cancer: the cutting edges 40, e107333 (2021).
and future directions. Nat. Rev. Clin. Oncol. 20, 799–813 (2023). 51. Hu, L. et al. Single-cell RNA sequencing reveals the cellular origin and
28. Dhanyamraju, P. K., Schell, T. D., Amin, S. & Robertson, G. P. Drug- evolution of breast cancer in BRCA1 mutation carriers. Cancer Res 81,
tolerant persister cells in cancer therapy resistance. Cancer Res 82, 2600–2611 (2021).
2503–2514 (2022). 52. Bassez, A. et al. A single-cell map of intratumoral changes during anti-
29. Mikubo, M., Inoue, Y., Liu, G. & Tsao, M. S. Mechanism of drug tolerant PD1 treatment of patients with breast cancer. Nat. Med 27,
persister cancer cells: the landscape and clinical implication for 820–832 (2021).
therapy. J. Thorac. Oncol. 16, 1798–1809 (2021). 53. Ding, S., Chen, X. & Shen, K. Single-cell RNA sequencing in breast
30. Ramirez, M. et al. Diverse drug-resistance mechanisms can emerge cancer: Understanding tumor heterogeneity and paving roads to
from drug-tolerant cancer persister cells. Nat. Commun. 7, individualized therapy. Cancer Commun. (Lond.) 40, 329–344 (2020).
10690 (2016). 54. Fila, M., Sobczuk, A., Pawlowska, E. & Blasiak, J. Epigenetic
31. Saeki, K. et al. Mammary cell gene expression atlas links epithelial cell connection of the calcitonin gene-related peptide and its potential in
remodeling events to breast carcinogenesis. Commun. Biol. 4, migraine. Int. J. Mol. Sci. 23, 6151 (2022).
660 (2021). 55. Sanchez, M. L., Rodriguez, F. D. & Covenas, R. Peptidergic Systems
32. Raredon, M. S. B. et al. Computation and visualization of cell-cell and Cancer: Focus on Tachykinin and calcitonin/calcitonin gene-
signaling topologies in single-cell systems data using Connectome. related peptide families. Cancers. 15, 1694 (2023).
Sci. Rep. 12, 4187 (2022). 56. Mahalanabis, A. et al. Evaluation of single-cell RNA-seq clustering
33. Gao, S., Dai, Y. & Rehman, J. A Bayesian inference transcription factor algorithms on cancer tumor datasets. Comput Struct. Biotechnol. J.
activity model for the analysis of single-cell transcriptomes. Genome 20, 6375–6387 (2022).
Res 31, 1296–1311 (2021). 57. Fan, J., Slowikowski, K. & Zhang, F. Single-cell transcriptomics in
34. Bouland, G. A., Mahfouz, A. & Reinders, M. J. T. Consequences and cancer: computational challenges and opportunities. Exp. Mol. Med
opportunities arising due to sparser single-cell RNA-seq datasets. 52, 1452–1465 (2020).
Genome Biol. 24, 86 (2023). 58. Aran, D. et al. Reference-based analysis of lung single-cell
35. Zhang, M. J., Ntranos, V. & Tse, D. Determining sequencing depth in a sequencing reveals a transitional profibrotic macrophage. Nat.
single-cell RNA-seq experiment. Nat. Commun. 11, 774 (2020). Immunol. 20, 163–172 (2019).
36. van Dijk, D. et al. Recovering gene interactions from single-cell data 59. Kuleshov, M. V. et al. Enrichr: a comprehensive gene set enrichment
using data diffusion. Cell 174, 716–729.e727 (2018). analysis web server 2016 update. Nucleic Acids Res 44,
37. Qiu, X. et al. Reversed graph embedding resolves complex single-cell W90–97 (2016).
trajectories. Nat. Methods 14, 979–982 (2017). 60. Liberzon, A. et al. The Molecular Signatures Database (MSigDB)
38. Qiu, X. et al. Single-cell mRNA quantification and differential analysis hallmark gene set collection. Cell Syst. 1, 417–425 (2015).
with Census. Nat. Methods 14, 309–315 (2017). 61. Li, H. Aligning sequence reads, clone sequences and assembly
39. Trapnell, C. et al. The dynamics and regulators of cell fate decisions contigs with BWA-MEM, arXiv:1303.3997; 10.48550/
are revealed by pseudotemporal ordering of single cells. Nat. arXiv.1303.3997 (2013).
Biotechnol. 32, 381–386 (2014). 62. Talevich, E., Shain, A. H., Botton, T. & Bastian, B. C. CNVkit: genome-
40. Dave, A. et al. The Breast Cancer Single-Cell Atlas: Defining cellular wide copy number detection and visualization from targeted DNA
heterogeneity within model cell lines and primary tumors to inform sequencing. PLoS Comput Biol. 12, e1004873 (2016).
disease subtype, stemness, and treatment options. Cell Oncol. 63. Casper, J. et al. The UCSC Genome Browser database: 2018 update.
(Dordr.) 46, 603–628 (2023). Nucleic Acids Res 46, D762–D769 (2018).

npj Precision Oncology | (2024)8:222 14

https://doi.org/10.1038/s41698-024-00723-6 Article

64. Hay, D. L., Garelja, M. L., Poyner, D. R. & Walker, C. S. Update on the Correspondence and requests for materials should be addressed to
pharmacology of calcitonin/CGRP family of peptides: IUPHAR Reo Maruyama.
Review 25. Br. J. Pharm. 175, 3–17 (2018).
65. Meyrath, M. et al. Proadrenomedullin N-Terminal 20 Peptides Reprints and permissions information is available at
(PAMPs) are agonists of the chemokine scavenger receptor ACKR3/ http://www.nature.com/reprints
CXCR7. ACS Pharm. Transl. Sci. 4, 813–823 (2021).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional
Acknowledgements claims in published maps and institutional affiliations.
The authors would like to thank Enago for the English language review.
Open Access This article is licensed under a Creative Commons
Author contributions Attribution-NonCommercial-NoDerivatives 4.0 International License,
K.O., K.K., H.S., and R.M. performed data analysis. K.O., Y.T., S.S., S.O., and which permits any non-commercial use, sharing, distribution and
T.U. recruited patients and obtained clinical specimens. K.O. processed reproduction in any medium or format, as long as you give appropriate
tumor samples and L.Y. performed single-cell experiments. T.O., S.B., and credit to the original author(s) and the source, provide a link to the Creative
T.K. performed pathological analyses. T.K. and T.T. provided clinical infor- Commons licence, and indicate if you modified the licensed material. You
mation. T.N., S.O., T.U., and R.M. supervised the research. K.O. and R.M. do not have permission under this licence to share adapted material
wrote the manuscript. All authors discussed the results and commented on derived from this article or parts of it. The images or other third party
the manuscript. material in this article are included in the article’s Creative Commons
licence, unless indicated otherwise in a credit line to the material. If material
Competing interests is not included in the article’s Creative Commons licence and your intended
The authors declare no competing Interests. use is not permitted by statutory regulation or exceeds the permitted use,
you will need to obtain permission directly from the copyright holder. To
Additional information view a copy of this licence, visit http://creativecommons.org/licenses/by-
Supplementary information The online version contains nc-nd/4.0/.
supplementary material available at
https://doi.org/10.1038/s41698-024-00723-6. © The Author(s) 2024

npj Precision Oncology | (2024)8:222 15