pith. sign in

arxiv: 2604.16685 · v1 · submitted 2026-04-17 · 💻 cs.LG · cs.AI

Graph Transformer-Based Pathway Embedding for Cancer Prognosis

Koushik Howlader , Md Tauhidul Islam , Wei Le This is my paper

Pith reviewed 2026-05-10 08:24 UTC · model grok-4.3

classification 💻 cs.LG cs.AI
keywords cancer prognosisgraph transformerpathway embeddingmulti-omics datametastasis predictiongene embeddingpathway rewiringpatient-conditioned adaptation
0
0 comments X

The pith

PATH creates gene embeddings from a shared base then adapts them to each patient's mutations and copy changes for better cancer spread prediction.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces PATH as a modulation-based embedding that begins with one stable representation per gene across all patients and then modifies it using that individual's copy number and mutation data. This embedding feeds into a graph transformer that attends to interactions among pathways. The approach aims to handle patient heterogeneity in omics data more effectively than direct mapping or simple aggregation methods. A reader would care because accurate metastasis forecasts matter for treatment timing and because the model also surfaces how pathway connections shift with disease state. It reports an F1 score of 0.8766 on pancancer data, an 8.8 percent gain over prior multi-omics benchmarks.

Core claim

PATH represents a paradigm shift by starting from a shared base embedding for each gene, preserving a stable biological identity across the population, and then dynamically adapting it using patient-specific copy number variation (CNV) and mutation signals. This allows the model to capture subtle individual molecular variations while maintaining a consistent latent understanding of the gene itself. We integrate PATH into a graph transformer framework that models interactions among biologically connected pathways through pathway-guided attention. Across pancancer metastasis prediction, PATH achieves an F1 score of 0.8766, representing an 8.8 percent improvement over the current SOTA multi-om

What carries the argument

PATH, a modulation-based patient-conditioned gene embedding that starts from a shared base representation for each gene and adapts it with individual CNV and mutation signals inside a graph transformer using pathway-guided attention.

If this is right

  • Improved accuracy in predicting metastasis across multiple cancer types from combined omics inputs.
  • Discovery of pathways whose interactions change in specific disease states rather than staying fixed.
  • More interpretable models that keep a consistent view of each gene while allowing patient variation.
  • A template for other hierarchical pathway models that currently rely on raw mapping or statistical pooling.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same shared-base-plus-adaptation idea could extend to non-cancer diseases where molecular heterogeneity limits prediction.
  • The rewiring findings might point to stage-specific drug targets that act on changing pathway links rather than single genes.
  • Testing whether the learned gene embeddings transfer to new cancer types without retraining would check if the stable identity holds.
  • If the adaptation step proves robust, it could reduce the need for very large patient cohorts in future multi-omics studies.

Load-bearing premise

Starting from a shared base embedding for each gene and then adapting it with patient-specific signals keeps the gene's biological identity intact while still capturing real individual differences without adding noise or overfitting.

What would settle it

Run the trained PATH model on an independent pancancer cohort and check whether the F1 score drops below the prior SOTA by more than a few percent or whether the detected pathway rewiring patterns appear equally in random patient groupings.

Figures

Figures reproduced from arXiv: 2604.16685 by Koushik Howlader, Md Tauhidul Islam, Wei Le.

Figure 1
Figure 1. Figure 1: Overview of the PATH framework and model core. (a) For each cancer cohort, patient multi-omics profiles (somatic mutation and copy-number variation) are used as input. PATH performs gene-level representation learning to produce patient-specific gene features, aggregates these features into pathway embeddings using Reactome gene sets, and generates pathway tokens summarizing pathway activity. The pathway to… view at source ↗
Figure 2
Figure 2. Figure 2: PATH accurately predicts metastatic status in the pancancer cohort and learns progression-associated latent representations. (a) ROC curves comparing PATH with baseline models for primary versus metastatic classification in the pancancer dataset. (b) PR curves showing that PATH preserves favorable precision across a broad recall range under class imbalance. (c) Mean ± standard deviation of F1 score, precis… view at source ↗
Figure 3
Figure 3. Figure 3: (a) SHAP-based pathway analysis highlighting the top pathways contributing to primary versus metastatic classification, with mean absolute SHAP values summarizing overall pathway importance. (b) Circular network of model-inferred novel pathway–pathway interactions (FDR < 0.05) absent from the reference Reactome network, shown alongside existing curated connections. (c) Pathway–pathway attention heatmaps fo… view at source ↗
Figure 4
Figure 4. Figure 4: PATH shows strong performance and interpretable biological signals in prostate cancer metastasis classification. (a) F1 score, precision, and recall across models, showing the best overall performance by PATH. (b) ROC and precision–recall curves confirming the strong discriminative ability of PATH. (c) SHAP-based gene importance analysis identifying genes linked to primary and metastatic prostate cancer, w… view at source ↗
Figure 5
Figure 5. Figure 5: PATH improves BLCA stage classification and highlights FGFR-related pathway and gene signals. (a) Mean ± standard deviation of F1 score, precision, and recall across five-fold cross-validation for early versus late stage classification in BLCA. PATH achieves the best F1 (0.81±0.01) and perfect recall (1.00±0.00). (b) Pathway-level SHAP heatmap showing pairwise interactions among FGFR1–FGFR4 signaling axes … view at source ↗
read the original abstract

Accurate prediction of cancer progression remains a challenge due to the high heterogeneity of molecular omics data across patients. While biologically informed models have improved the interpretability of these predictions, a persistent limitation lies in how they encode individual genes to construct pathway representations. Existing hierarchical models typically derive gene features by directly mapping raw molecular inputs, whereas integration frameworks often rely on simple statistical aggregations of patient-level signals. These approaches often fail to explicitly learn a shared base representation for each gene, thereby limiting the expressiveness and biological accuracy of downstream pathway embeddings. To address this, we introduce PATH, a modulation-based, patient-conditioned gene embedding strategy. PATH represents a paradigm shift by starting from a shared base embedding for each gene, preserving a stable biological identity across the population, and then dynamically adapting it using patient-specific copy number variation (CNV) and mutation signals. This allows the model to capture subtle individual molecular variations while maintaining a consistent latent understanding of the gene itself. We integrate PATH into a graph transformer framework that models interactions among biologically connected pathways through pathway-guided attention. Across pancancer metastasis prediction, PATH achieves an F1 score of 0.8766, representing an 8.8 percent improvement over the current SOTA multi-omics benchmarks. Beyond superior predictive accuracy, our approach identifies biologically meaningful pathways and, crucially, reveals disease-state-specific pathway rewiring, offering new insights into the evolving pathway-pathway interactions that drive cancer progression.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Based solely on the abstract, the central claim rests on standard machine-learning assumptions about graph attention and embedding modulation; no explicit free parameters, additional axioms, or invented entities beyond the PATH strategy itself are quantified.

axioms (1)
  • domain assumption Graph transformers with pathway-guided attention can accurately model interactions among biologically connected pathways
    Invoked in the description of the framework that integrates the embeddings.
invented entities (1)
  • PATH modulation-based gene embedding no independent evidence
    purpose: To create a shared base representation for each gene that is then adapted by patient-specific signals
    Newly introduced strategy described as a paradigm shift over prior gene feature derivation methods.

pith-pipeline@v0.9.0 · 5557 in / 1302 out tokens · 53237 ms · 2026-05-10T08:24:08.056003+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

89 extracted references · 89 canonical work pages · 2 internal anchors

  1. [1]

    & Swanton, C

    McGranahan, N. & Swanton, C. Clonal heterogeneity and tumor evolution: Past, present, and the future.Cell168, 613–628 (2017)

    work page 2017

  2. [2]

    & Conesa, A

    Tarazona, S., Arzalluz-Luque, Á. & Conesa, A. Undisclosed, unmet and neglected challenges in multi-omics studies. Nature Computational Science1, 395–402 (2021)

    work page 2021

  3. [3]

    M.et al.An integrative deep learning framework for classifying molecular subtypes of breast cancer

    Islam, M. M.et al.An integrative deep learning framework for classifying molecular subtypes of breast cancer. Computational and Structural Biotechnology Journal18, 2185–2199 (2020). 20/25

    work page 2020

  4. [4]

    Communications Biology3, 502 (2020)

    Fu, Y .et al.A gene prioritization method based on a swine multi-omics knowledgebase and a deep learning model. Communications Biology3, 502 (2020)

    work page 2020

  5. [5]

    Wang, T.et al.MOGONET integrates multi-omics data using graph convolutional networks allowing patient classification and biomarker identification.Nature Communications12, 3445 (2021)

    work page 2021

  6. [6]

    B., Islam, M

    Tanvir, R. B., Islam, M. M., Sobhan, M., Luo, D. & Mondal, A. M. Mogat: A multi-omics integration framework using graph attention networks for cancer subtype prediction.International Journal of Molecular Sciences25, 2788 (2024)

    work page 2024

  7. [7]

    Liu, X.et al.Pathformer: a biological pathway informed transformer for disease diagnosis and prognosis using multi-omics data.Bioinformatics40, btae316 (2024)

    work page 2024

  8. [8]

    A.et al.Biologically informed deep neural network for prostate cancer discovery.Nature598, 348–352 (2021)

    Elmarakeby, H. A.et al.Biologically informed deep neural network for prostate cancer discovery.Nature598, 348–352 (2021)

    work page 2021

  9. [9]

    H.et al.PathCNN: interpretable convolutional neural networks for survival prediction and pathway analysis applied to glioblastoma.Bioinformatics37, i443–i450 (2021)

    Oh, J. H.et al.PathCNN: interpretable convolutional neural networks for survival prediction and pathway analysis applied to glioblastoma.Bioinformatics37, i443–i450 (2021)

    work page 2021

  10. [10]

    & Wang, S

    Ma, T. & Wang, S. GraphPath: a graph attention model for molecular stratification with interpretability based on the pathway-pathway interaction network.Briefings in Bioinformatics25, bbae112 (2024). 12.Gillespie, M.et al.The reactome pathway knowledgebase 2022.Nucleic Acids Research50, D687–D692 (2022)

    work page 2024

  11. [11]

    & Bernards, R

    Prahallad, A. & Bernards, R. Opportunities and challenges provided by crosstalk between signalling pathways in cancer. Oncogene35, 1073–1079 (2016)

    work page 2016

  12. [12]

    & Taniguchi, K

    Kawazoe, T. & Taniguchi, K. The sprouty/spred family as tumor suppressors: Coming of age.Cancer science110, 1525–1535 (2019)

    work page 2019

  13. [13]

    di Martino, E., Tomlinson, D. C. & Knowles, M. A. A decade of fgf receptor research in bladder cancer: past, present, and future challenges.Advances in urology2012, 429213 (2012)

    work page 2012

  14. [14]

    E., Kim, H

    Bahar, M. E., Kim, H. J. & Kim, D. R. Targeting the ras/raf/mapk pathway for cancer therapy: from mechanism to clinical studies.Signal transduction and targeted therapy8, 455 (2023)

    work page 2023

  15. [15]

    M., Ding, K., Chen, L

    Levine, K. M., Ding, K., Chen, L. & Oesterreich, S. Fgfr4: A promising therapeutic target for breast cancer and other solid tumors.Pharmacology & therapeutics214, 107590 (2020)

    work page 2020

  16. [16]

    Garcia-Recio, S.et al.Fgfr4 regulates tumor subtype differentiation in luminal breast cancer and metastatic disease.The Journal of clinical investigation130, 4871–4887 (2020)

    work page 2020

  17. [17]

    C., Baxter, E

    Tomlinson, D. C., Baxter, E. W., Loadman, P. M., Hull, M. A. & Knowles, M. A. Fgfr1-induced epithelial to mesenchymal transition through mapk/plcγ/cox-2-mediated mechanisms.PloS one7, e38972 (2012)

    work page 2012

  18. [18]

    A.et al.Genomic hallmarks and structural variation in metastatic prostate cancer.Cell174, 758–769 (2018)

    Quigley, D. A.et al.Genomic hallmarks and structural variation in metastatic prostate cancer.Cell174, 758–769 (2018)

    work page 2018

  19. [19]

    Abida, W.et al.Genomic correlates of clinical outcome in advanced prostate cancer.Proceedings of the National Academy of Sciences116, 11428–11436 (2019)

    work page 2019

  20. [20]

    C.et al.The translational landscape of mTOR signalling steers cancer initiation and metastasis.Nature485, 55–61 (2012)

    Hsieh, A. C.et al.The translational landscape of mTOR signalling steers cancer initiation and metastasis.Nature485, 55–61 (2012)

    work page 2012

  21. [21]

    Y ., Dass, M

    Shorning, B. Y ., Dass, M. S., Smalley, M. J. & Pearson, H. B. The PI3K-AKT-mTOR pathway and prostate cancer. International Journal of Molecular Sciences21, 4507 (2020)

    work page 2020

  22. [22]

    C.et al.Effects on prostate cancer cells of targeting RNA polymerase III.Nucleic Acids Research47, 3937–3956 (2019)

    Colis, L. C.et al.Effects on prostate cancer cells of targeting RNA polymerase III.Nucleic Acids Research47, 3937–3956 (2019)

    work page 2019

  23. [23]

    Proceedings of the National Academy of Sciences107, 14134–14139 (2010)

    Furic, L.et al.eIF4E phosphorylation promotes tumorigenesis and is associated with prostate cancer progression. Proceedings of the National Academy of Sciences107, 14134–14139 (2010)

    work page 2010

  24. [24]

    & McGraw, T

    Gonzalez, E. & McGraw, T. E. PTEN-regulated PI3K-p110 and AKT isoform plasticity controls metastatic prostate cancer progression.Oncogene43, 330–343 (2024)

    work page 2024

  25. [25]

    Lorente, D.et al.Clinical context shapes the relationship between genomic alterations and response to AR inhibitors and chemotherapy in metastatic prostate cancer.Cancer Research Communications(2025)

    work page 2025

  26. [26]

    & Xie, L

    Wo, Q., Shi, L., Shi, J., Mao, Y . & Xie, L. The mechanism by which hedgehog interacting protein (hhip) in cancer- associated fibroblasts regulate the secretion of inflammatory factors through the jak1/stat3 pathway affecting prostate cancer stemness.Journal of Inflammation Research8659–8680 (2024). 21/25

    work page 2024

  27. [27]

    McCall, P.et al.NF κB signalling is upregulated in a subset of castrate-resistant prostate cancer patients and correlates with disease progression.British Journal of Cancer107, 1554–1563 (2012)

    work page 2012

  28. [28]

    Nadiminty, N.et al.Inhibition of NF- κB signaling restores responsiveness of castrate resistant prostate cancer cells to anti-androgen treatment by decreasing androgen receptor variants expression.Oncotarget6, 6982–6992 (2015)

    work page 2015

  29. [29]

    G., Bhatt, C., Bhatt, R., Bhatt, T

    Arimbasseri, A. G., Bhatt, C., Bhatt, R., Bhatt, T. & White, R. J. A cancer-associated RNA polymerase III identity drives robust transcription and expression of snaR-A noncoding RNA.Nature Communications13, 3145 (2022). 32.Taylor, B. S.et al.Integrative genomic profiling of human prostate cancer.Cancer Cell18, 11–22 (2010)

    work page 2022

  30. [30]

    & Niu, R

    Zhang, J., Zhang, F. & Niu, R. Shp2 promotes metastasis of prostate cancer by attenuating the PAR3/PAR6/aPKC polarity protein complex and enhancing epithelial-to-mesenchymal transition.Oncogene35, 4862–4872 (2016). 34.Szybowska, P., Kostas, M., Wesche, J., Haugsten, E. M. & Wiedlocha, A. Negative regulation of fgfr (fibroblast growth factor receptor) sign...

    work page 2016

  31. [31]

    Fan, S.et al.Pharmacological and biological targeting of fgfr1 in cancer.Current issues in molecular biology46, 13131–13150 (2024)

    work page 2024

  32. [32]

    & Chen, Q

    Tang, S., Hao, Y ., Yuan, Y ., Liu, R. & Chen, Q. Role of fibroblast growth factor receptor 4 in cancer.Cancer Science109, 3024–3031 (2018)

    work page 2018

  33. [33]

    Streuli, C. H. & Meng, Q.-J. Influence of the extracellular matrix on cell-intrinsic circadian clocks.Journal of Cell Science132, jcs207498 (2019)

    work page 2019

  34. [34]

    & Streuli, C

    Meng, Q.-J. & Streuli, C. H. The circadian clock and extracellular matrix homeostasis in aging and age-related diseases. American Journal of Physiology–Cell Physiology325, C44–C54 (2023)

    work page 2023

  35. [35]

    Tates, J.et al.Cilium structure, assembly, and disassembly regulated by the cytoskeleton.Biochemical Journal475, 2329–2353 (2018)

    work page 2018

  36. [36]

    41.Holst, C

    Sfakianos, J.et al.Par3 functions in the biogenesis of the primary cilium in polarized epithelial cells.Journal of Cell Biology179, 1133–1140 (2007). 41.Holst, C. R.et al.Epithelial cell polarity and tumorigenesis.Acta Pharmacologica Sinica32, 552–564 (2011). 42.Wong, S. Y .et al.Primary cilia and cancer: a tale of many faces.Oncogene(2025)

    work page 2007

  37. [37]

    & Morris, D

    Masoumi-Moghaddam, S., Amini, A. & Morris, D. L. The developing story of sprouty and cancer.Cancer Metastasis Reviews33, 695–720 (2014)

    work page 2014

  38. [38]

    Yue, S.et al.Fgfr-tki resistance in cancer: current status and perspectives.Journal of Hematology & Oncology14, 23 (2021)

    work page 2021

  39. [39]

    & Hazan, R

    Suyama, K., Shapiro, I., Guttman, M. & Hazan, R. B. A signaling pathway leading to metastasis is controlled by n-cadherin and the fgf receptor.Cancer Cell2, 301–314 (2002)

    work page 2002

  40. [40]

    Xu, Y .et al.Sprouty2 correlates with favorable prognosis of gastric adenocarcinoma via suppressing fgfr2-induced erk phosphorylation and cancer progression.Oncotarget8, 4888–4900 (2017)

    work page 2017

  41. [41]

    & Kontek, R

    Kciuk, M., Gieleci´nska, A., Budzinska, A., Mojzych, M. & Kontek, R. Metastasis and mapk pathways.International journal of molecular sciences23, 3847 (2022)

    work page 2022

  42. [42]

    Y ., Bhatt, D

    Takeda, D. Y ., Bhatt, D. & Drapkin, B. J. Androgen receptor-mediated transcription in prostate cancer.Cancers14, 900 (2021). AR is the primary driver of PCa growth and differentiation in nearly all patients; vast majority of CRPC remains AR-dependent via amplification, mutation, or splice variants

    work page 2021

  43. [43]

    A.et al.Genomic hallmarks and structural variation in metastatic prostate cancer.Cell174, 758–769.e9 (2018)

    Quigley, D. A.et al.Genomic hallmarks and structural variation in metastatic prostate cancer.Cell174, 758–769.e9 (2018). AR gene body amplification and upstream enhancer amplification are the most common mechanisms sustaining AR activity at castrate testosterone levels in mCRPC

    work page 2018

  44. [44]

    Raptor and Rictor expression elevated in prostate cancer tissues; knockdown of Raptor attenuates migration and invasion; mTOR drives EMT

    Xu, Y .et al.mTOR regulate EMT through RhoA and Rac1 pathway in prostate cancer.Molecular Biology Reports 42, 67–72 (2015). Raptor and Rictor expression elevated in prostate cancer tissues; knockdown of Raptor attenuates migration and invasion; mTOR drives EMT

    work page 2015

  45. [45]

    Proceedings of the National Academy of Sciences107, 14134–14139 (2010)

    Furic, L.et al.eIF4E phosphorylation promotes tumorigenesis and is associated with prostate cancer progression. Proceedings of the National Academy of Sciences107, 14134–14139 (2010). Phospho-eIF4E levels correlate with prostate cancer disease progression in patient cohorts; eIF4E phosphorylation required for translational upregulation of oncoproteins. 22/25

    work page 2010

  46. [46]

    PI3K/Akt/mTOR and Ras/MAPK pathways converge on eIF4E as a shared translational control node in prostate cancer progression

    Dua, K.et al.eIF4E phosphorylation in prostate cancer.Frontiers in Genetics10, 14 (2019). PI3K/Akt/mTOR and Ras/MAPK pathways converge on eIF4E as a shared translational control node in prostate cancer progression

    work page 2019

  47. [47]

    PTEN inactivation in 20% of primary tumors and 50% of CRPC; loss activates PI3K–AKT; adverse oncological outcomes confirmed across large cohorts

    Cotter, K.et al.Clinical implications of PTEN loss in prostate cancer.Nature Reviews Urology17, 338–351 (2018). PTEN inactivation in 20% of primary tumors and 50% of CRPC; loss activates PI3K–AKT; adverse oncological outcomes confirmed across large cohorts

    work page 2018

  48. [48]

    Yoshimoto, M.et al.Genomic rearrangements of PTEN in prostate cancer.Frontiers in Oncology3, 240 (2013). PTEN is the most frequently deleted tumor suppressor in prostate cancer; homozygous loss linked to metastasis and androgen-independent progression; monoallelic loss in up to 60% of localized tumors

    work page 2013

  49. [49]

    Zhang, P.et al.Shp2 promotes metastasis of prostate cancer by attenuating the PAR3/PAR6/aPKC polarity protein complex and enhancing epithelial-to-mesenchymal transition.Oncogene35, 2473–2485 (2016). SHP2 (PTPN11) is upregulated in prostate cancer; promotes metastasis via EMT and cell polarity disruption; elevated expression correlates with metastasis and ...

    work page 2016

  50. [50]

    AR negatively regulates locally produced GH in PCa cells; GH induction following ADT enhances invasion, mesenchymal transition, and CRPC via JAK2/STAT5 and AR-V7 induction

    Pandey, V .et al.Androgen receptor regulation of local growth hormone in prostate cancer cells.Endocrinology158, 2255–2268 (2017). AR negatively regulates locally produced GH in PCa cells; GH induction following ADT enhances invasion, mesenchymal transition, and CRPC via JAK2/STAT5 and AR-V7 induction

    work page 2017

  51. [51]

    Chopin, L. K.et al.Growth hormone (GH) receptors in prostate cancer: Gene expression in human tissues and cell lines and characterization, GH signaling and androgen receptor regulation in LNCaP cells.Molecular and Cellular Endocrinology220, 109–123 (2004). GHR mRNA 80% higher in prostate adenocarcinoma than BPH; GH induces JAK2/STAT5 and Akt signaling and...

    work page 2004

  52. [52]

    Prolactin is a survival factor in prostate cancer; PRLR signaling expressed in PCa cells; PRLR blockade proposed as novel treatment strategy

    Goffin, V .et al.Unexploited therapies in breast and prostate cancer: Blockade of the prolactin receptor.Trends in Endocrinology & Metabolism22, 247–254 (2011). Prolactin is a survival factor in prostate cancer; PRLR signaling expressed in PCa cells; PRLR blockade proposed as novel treatment strategy

    work page 2011

  53. [53]

    T.et al.GPI-AP: Unraveling a new class of malignancy mediators and potential immunotherapy targets

    Lauber, D. T.et al.GPI-AP: Unraveling a new class of malignancy mediators and potential immunotherapy targets. Frontiers in Oncology10, 537311 (2020). GPI-anchored proteins play established roles in signal transduction, immune modulation, and cancer cell invasion and metastasis across multiple cancer types

    work page 2020

  54. [54]

    High SOAT1 expression associated with lymph node metastasis, Gleason score, and shorter BCR-free survival (HR 2.40, p<0.001) in 305 high-risk PCa patients

    Eckhardt, C.et al.High expression of sterol-o-acyl transferase 1 (SOAT1), an enzyme involved in cholesterol metabolism, is associated with earlier biochemical recurrence in high risk prostate cancer.Prostate Cancer and Prostatic Diseases25, 484–490 (2022). High SOAT1 expression associated with lymph node metastasis, Gleason score, and shorter BCR-free sur...

    work page 2022

  55. [55]

    SOAT1 elevated in PCa tissues, correlated with metastasis and Gleason score; knockdown suppresses SCD1-driven lipogenesis and inhibits tumor growth in vivo

    Liu, Y .et al.Knockdown of sterol o-acyltransferase 1 (SOAT1) suppresses SCD1-mediated lipogenesis and cancer procession in prostate cancer.Prostaglandins & Other Lipid Mediators153, 106537 (2021). SOAT1 elevated in PCa tissues, correlated with metastasis and Gleason score; knockdown suppresses SCD1-driven lipogenesis and inhibits tumor growth in vivo. 62...

    work page 2021

  56. [56]

    L.et al.Intratumoral androgens in castration-recurrent prostate cancer.Endocrine-Related Cancer18, R175–R182 (2011)

    Mohler, J. L.et al.Intratumoral androgens in castration-recurrent prostate cancer.Endocrine-Related Cancer18, R175–R182 (2011). Metastatic CRPC upregulates androgen-synthesizing enzymes (HSD3b1, CYP17A1, AKR1C3) to synthesize androgens de novo from cholesterol, maintaining intratumoral androgen levels sufficient to activate AR target genes. 64.Zheng, J.et...

    work page 2011

  57. [57]

    FGFR inhibitors in urothelial cancer: From scientific rationale to clinical development.Journal of Korean Medical Science39, e320 (2024)

    Kwon, W.-A. FGFR inhibitors in urothelial cancer: From scientific rationale to clinical development.Journal of Korean Medical Science39, e320 (2024)

    work page 2024

  58. [58]

    C., Knowles, M

    Tomlinson, D. C., Knowles, M. A. & Speirs, V . FGFR1-induced epithelial to mesenchymal transition through MAPK/PLCγ/COX-2-mediated mechanisms.PLoS ONE7, e38972 (2012)

    work page 2012

  59. [59]

    Knowles, M. A. & Hurst, C. D. A decade of FGF receptor research in bladder cancer: past, present, and future challenges. Nature Reviews Cancer15, 25–41 (2015)

    work page 2015

  60. [60]

    International Journal of Medical Sciences17, 213–220 (2020)

    Huang, C.-P.et al.Involvement of FGFR4 gene variants on the clinicopathological severity in urothelial cell carcinoma. International Journal of Medical Sciences17, 213–220 (2020). 23/25

    work page 2020

  61. [61]

    Tsichlis, P. N. & McNiel, E. A. Analyses of publicly available genomics resources define FGF-2-expressing bladder carcinomas as EMT-prone, proliferative tumors with low mutation rates and high expression of CTLA-4, PD-1 and PD-L1.Signal Transduction and Targeted Therapy1, 16045 (2017)

    work page 2017

  62. [62]

    Noeraparast, M.et al.FGFR3 alterations in bladder cancer: Sensitivity and resistance to targeted therapies.Cancer Communications44, 1189–1208 (2024)

    work page 2024

  63. [63]

    Chen, Y .et al.Targeting PSMB5-induced PANoptosis in bladder cancer: multi-omics insights and TCM candidate discovery.Frontiers in Immunology16, 1656682 (2025)

    work page 2025

  64. [64]

    C.et al.FGFR3, HRAS, KRAS, NRAS and PIK3CA mutations in bladder cancer and their potential as biomarkers for surveillance and therapy.PLoS One5, e13821 (2010)

    Kompier, L. C.et al.FGFR3, HRAS, KRAS, NRAS and PIK3CA mutations in bladder cancer and their potential as biomarkers for surveillance and therapy.PLoS One5, e13821 (2010)

    work page 2010

  65. [65]

    Larsson, P.et al.Pan-cancer analysis of genomic and transcriptomic data reveals the prognostic relevance of human proteasome genes in different cancer types.BMC Cancer22, 993 (2022)

    work page 2022

  66. [66]

    & Weinberg, R

    Valastyan, S. & Weinberg, R. A. Tumor metastasis: molecular insights and evolving paradigms.Cell147, 275–292 (2011)

    work page 2011

  67. [67]

    & Weinberg, R

    Dongre, A. & Weinberg, R. A. New insights into the mechanisms of epithelial–mesenchymal transition and implications for cancer.Nature Reviews Molecular Cell Biology20, 69–84 (2019)

    work page 2019

  68. [68]

    W., Pattabiraman, D

    Lambert, A. W., Pattabiraman, D. R. & Weinberg, R. A. Emerging biological principles of metastasis.Cell168, 670–691 (2017). 77.Creixell, P.et al.Pathway and network analysis of cancer genomes.Nature Methods12, 615–621 (2015). 78.Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: The next generation.Cell144, 646–674 (2011)

    work page 2017

  69. [69]

    Dent, P.et al.Crosstalk between oncogenic signalling pathways in cancer.Cancer Biology & Therapy2, S122–S129 (2003)

    work page 2003

  70. [70]

    & Nishida, E

    Hanafusa, H., Torii, S., Yasunaga, T. & Nishida, E. Sprouty1 and sprouty2 provide a control mechanism for the ras/mapk signalling pathway.Nature Cell Biology4, 850–858 (2002)

    work page 2002

  71. [71]

    82.Davies, H.et al.Mutations of thebrafgene in human cancer.Nature417, 949–954 (2002)

    Helsten, T.et al.The fgfr landscape in cancer: Analysis of 4,853 tumors by next-generation sequencing.Clinical Cancer Research22, 259–267 (2016). 82.Davies, H.et al.Mutations of thebrafgene in human cancer.Nature417, 949–954 (2002)

    work page 2016

  72. [72]

    Yao, Z.et al.Targeting RAF–MEK–ERK signalling in cancer: negative feedback via Sprouty induction by ERK.Signal Transduction and Targeted Therapy4, 5 (2019)

    work page 2019

  73. [73]

    Kiefel, H.et al.L1CAM: a major driver for tumor cell invasion and motility.Cell Adhesion & Migration6, 374–384 (2012)

    work page 2012

  74. [74]

    Ganesh, K.et al.L1CAM defines the regenerative origin of metastasis-initiating cells in colorectal cancer.Nature Cancer 1, 28–45 (2020)

    work page 2020

  75. [75]

    Xu, Y .-F.et al.Sprouty2 correlates with favorable prognosis of gastric adenocarcinoma via suppressing FGFR2-induced ERK phosphorylation and cancer progression.Oncotarget8, 4888–4900 (2017)

    work page 2017

  76. [76]

    Haglund, F.et al.Cancer mutations in FGFR2 prevent a negative feedback loop mediated by the ERK1/2 pathway.Cells 8, 518 (2019)

    work page 2019

  77. [77]

    Peng, W.et al.Combined inhibition of PI3K and STAT3 signaling effectively inhibits bladder cancer growth.Oncogenesis (2024)

    work page 2024

  78. [78]

    Saez-Ayala, M.et al.Nuclear pore complex protein RANBP2 and related SUMOylation in solid malignancies.Biochimica et Biophysica Acta – Reviews on Cancer(2024)

    work page 2024

  79. [79]

    J.et al.Visualizing and interpreting cancer genomics data via the Xena platform.Nature Biotechnology38, 675–678 (2020)

    Goldman, M. J.et al.Visualizing and interpreting cancer genomics data via the Xena platform.Nature Biotechnology38, 675–678 (2020)

    work page 2020

  80. [80]

    The cancer genome atlas pan-cancer analysis project.Nature Genetics45, 1113–1120 (2013)

    Cancer Genome Atlas Research Network. The cancer genome atlas pan-cancer analysis project.Nature Genetics45, 1113–1120 (2013)

    work page 2013

Showing first 80 references.