Xiaoqing Fan1, Kangna Cao1, Xu Steven Xu2, Xiaoyu Yan1
1The Chinese University of Hong Kong, 2Genmab Inc.
Introduction: Iron is a crucial component required for a variety of vital physiological processes, such as DNA synthesis and energy metabolism that is indispensable to cell growth and division[1]. In cancer cells, enhanced iron uptake and/or decreased iron efflux is essential to increase the intracellular iron amount to support their increased demands of survival, rapid cell division, and metastasis[2]. However, the amount of iron required by diverse tumors is different due to their remarkable heterogeneity. How the expression level of proteins that control iron uptake and efflux changes and coordinates to generate a unified effect on iron homeostasis is an important open question. These quantitative interconnections are important for guiding novel interventions for cancer therapy that involve iron homeostasis perturbations, such as iron-chelator-based therapy and ferroptosis-inducing drugs[3-5]. Moreover, investigating the biological relevance and clinical significance of iron homeostasis in larger patient cohorts to stratify high-risk individuals at an early clinical stage holds promise to facilitate precision oncology, turning the unique iron metabolic properties of cancer cells to therapeutic advantage[4, 6, 7]. Objectives: (1) To develop parametric time-to-event (TTE) models using the expression profile of the major proteins that control iron uptake and efflux to quantify the contribution of the expression level of these proteins on the outcome of cancer patients. (2) To create a cancer-specific iron risk score (CIRS) based on TTE modeling results to predict the long-term individual survival outcomes of patients. Methods: We hypothesized that the intracellular iron amount in cancer cells is increased due to either the increased expression of iron uptake proteins, decreased expression of iron efflux protein, or both. Based on the literature review, we found eight proteins is involved in iron transport, including transferrin receptor 1 (TFR1), divalent metal transporter 1 (DMT1), lipocalin-2 (LCN2), cluster determinant 44 (CD44), heme carrier protein 1 (HCP1), transmembrane ion channels Zrt-/Irt-like protein (ZIP) 8, and ZIP14 for iron uptake, and ferroportin (FPN) for iron efflux [8-12]. To investigate the major mediators for iron uptake or export in cancer patients, we collected human data from UALCAN (https://ualcan.path.uab.edu/index.html), a comprehensive database that integrates publicly available cancer OMICS data, including The Cancer Genome Atlas [TCGA] and Clinical Proteomic Tumor Analysis Consortium [CPTAC] database [13]. The RNA expressions of the iron transport-related proteins in cancer patients were compared with the corresponding normal subjects from TCGA. We only included cancer types that contained more than 10 paired normal samples to ensure the strength of statistical power. Therefore, 6903 TCGA patient samples and 709 normal subject samples across 16 cancer types were analyzed. To evaluate the impact of patient iron transport-related proteins (TFR1, CD44, and FPN) expression level on the observed overall survival (OS) and to stratify the iron homeostasis-based risk, parametric TTE models were developed and evaluated based on the RNA expression data, as the RNA data is more complete than the protein expression data[14, 15]. The RNA expression level of individual patients together with the corresponding demographics and characteristics were collected from the HPA database[16] and the National Cancer Institute’s Genomic Data Commons data portal (https://portal.gdc.cancer.gov/v1/). Using the final covariate model and the distribution of the TCGA datasets, simulation-based (1000 replicates) visual predictive checks (VPCs) were compared for different models. For each simulation, Kaplan–Meier proportions of the surviving patients were computed across time for evaluation of median OS and the 90% confidence intervals around simulations were included based on the 5th and 95th percentiles of the estimated covariate effect parameters. The parametric TTE analysis was performed using the first-order conditional estimation method with interaction algorithm in NONMEM (Version 7.5, Icon Development Solutions, Ellicott City, MD, USA). The use of NONMEM was facilitated by Perl-speaks-NONMEM (version 4.9.6, http://psn.sourceforge.net/docs.php). The ordinary differential equations were solved by the ADVAN6 subroutine. To exploit the potential of iron homeostasis profiles at baseline to predict long-term survival outcomes across various cancers, we developed a cancer-specific iron risk score (CIRS) based on the TTE modeling estimation results. We divided the patients into tertile groups based on stratified CIRS. Then, the relationship between CIRS and patient survival was plotted with the Kaplan–Meier survival curves using the Survival and Survminer R packages. Results: We found that, across tissue types, at least 1 iron uptake protein, either TFR1 or CD44, overexpressed, or the iron efflux protein FPN was decreased, or both, in the analysis of cancer cell lines compared to the corresponding normal tissue samples. Notably, for the proteins that were reported to participate in iron uptake, including DMT1, HCP1, LCN2, ZIP8, and ZIP14, their expression levels were not significantly increased in most cancers, and thus were excluded in the subsequent studies. TTE modeling results suggested that, despite the heterogeneous contribution of the three proteins to OS, an increase in TFR1 and CD44 raised the hazards of death and disease progression, whereas higher FPN was beneficial for survival. The magnitude of baseline hazards was quite different across the 15 cancers, in line with the observed event times. These results are consistent with the findings that iron homeostasis alterations and cancer progression are interrelated, i.e. higher iron content is associated with shorter survival. These results indicate that the expression level of TFR1, CD44, and FPN has the clinical significance of prognostic biomarkers in response to cancer progression and predicted clinical outcomes. The VPCs stratified by cancer type showed good agreement between the observed data and model prediction, demonstrating acceptable predictive performance of the TTE models across cancers. The Kaplan-Meier survival curves showed distinctive paths between the three risk groups stratified by CIRS across 15 cancers. The CIRS significantly stratified the cancer patients, and individuals in the high-risk group usually had notably worse outcomes and elevated risk of cancer progression and mortality. Conclusions: This study provided useful insights to identify cancer patient subgroups at higher risk based on predefined CIRS using RNA-seq data, and guide optimal clinical interventions regarding the significant predictors, which would constitute a novel strategy for cancer diagnosis, staging, and lead to potential therapeutic strategies to target these tumor-specific events.
1. Roemhild K, von Maltzahn F, Weiskirchen R, Knüchel R, von Stillfried S, Lammers T. Iron metabolism: pathophysiology and pharmacology. Trends Pharmacol Sci. 2021; 42: 640-56. 2. Torti SV, Manz DH, Paul BT, Blanchette-Farra N, Torti FM. Iron and Cancer. Annu Rev Nutr. 2018; 38: 97-125. 3. Ibrahim O, O’Sullivan J. Iron chelators in cancer therapy. Biometals. 2020; 33: 201-15. 4. Torti SV, Torti FM. Iron and Cancer: 2020 Vision. Cancer Res. 2020; 80: 5435-48. 5. Zhang DD. Ironing out the details of ferroptosis. Nat Cell Biol. 2024 Sep;26(9):1386-1393. 6. Gimeno M, Sada Del Real K, Rubio A. Precision oncology: a review to assess interpretability in several explainable methods. Brief Bioinform. 2023 Jul 20;24(4):bbad200. 7. Kim H, Villareal LB, Liu Z, Haneef M, Falcon DM, Martin DR, et al. Transferrin Receptor-Mediated Iron Uptake Promotes Colon Tumorigenesis. Adv Sci (Weinh). 2023; 10: e2207693. 8. Anderson GJ, Frazer DM. Current understanding of iron homeostasis. Am J Clin Nutr. 2017; 106: 1559s-66s. 9. Müller S, Sindikubwabo F, Cañeque T, Lafon A, Versini A, Lombard B, et al. CD44 regulates epigenetic plasticity by mediating iron endocytosis. Nat Chem. 2020; 12: 929-38. 10. Zhang YY, Han Y, Li WN, Xu RH, Ju HQ. Tumor iron homeostasis and immune regulation. Trends Pharmacol Sci. 2024; 45: 145-56. 11. El Hout M, Dos Santos L, Hamaï A, Mehrpour M. A promising new approach to cancer therapy: Targeting iron metabolism in cancer stem cells. Semin Cancer Biol. 2018; 53: 125-38. 12. Galy B, Conrad M, Muckenthaler M. Mechanisms controlling cellular and systemic iron homeostasis. Nat Rev Mol Cell Biol. 2024; 25: 133-55. 13. Chandrashekar DS, Karthikeyan SK, Korla PK, Patel H, Shovon AR, Athar M, et al. UALCAN: An update to the integrated cancer data analysis platform. Neoplasia. 2022; 25: 18-27. 14. Holford N. A time to event tutorial for pharmacometricians. CPT Pharmacometrics Syst Pharmacol. 2013; 2: e43. 15. Yao Y, Wang Z, Yong L, Yao Q, Tian X, Wang T, et al. Longitudinal and time-to-event modeling for prognostic implications of radical surgery in retroperitoneal sarcoma. CPT Pharmacometrics Syst Pharmacol. 2022; 11: 1170-82. 16. Uhlén M, Fagerberg L, Hallström BM, Lindskog C, Oksvold P, Mardinoglu A, et al. Proteomics. Tissue-based map of the human proteome. Science. 2015; 347: 1260419.
Reference: PAGE 33 (2025) Abstr 11429 [www.page-meeting.org/?abstract=11429]
Poster: Drug/Disease Modelling - Oncology