Triple-negative breast cancer (TNBC), lacking the expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER-2), accounts for about 10–15% and is the most aggressive molecular subtype of all breast tumors (Sung et al., 2021; Wang et al., 2021). Due to its invasiveness, early relapse, strong heterogeneity and limited therapeutic options, TNBC patients usually have distant metastasis at the diagnosis and worse long-term clinical outcomes compared with patients with other subtypes of breast cancer (Gadi and Davidson, 2017; Vagia et al., 2020). In recent years, advances in the landscape of diagnosis and treatment have elicited survival benefits for patients with TNBC. Nevertheless, TNBC remains a huge threat to life due to its recurrence and relatively high mortality (Dent et al., 2007; Bianchini et al., 2016; Garrido-Castro et al., 2019). Thus, identification of novel, accurate biomarkers, and exploration of individualized therapeutic targets for women with TNBC is very necessary.

Iron is a critical trace element for the activity of many proteins and enzymes. Iron is involved in cell respiration, oxygen transport, energy metabolism, DNA repair, and different signaling pathways (Torti et al., 2018). It is essential for human health, but excess iron or iron overload due to disorders of iron metabolism can induce severe toxicity even tumorigenesis in humans (Wu et al., 2004; Adams, 2015).

Increasing studies have demonstrated an association between consumption of red meat, intake of heme iron, or dietary intake of iron and initiation of breast cancer (Ferrucci et al., 2009; Guo J. et al., 2015; Inoue-Choi et al., 2016). Also, dysregulation of systemic iron homeostasis is a risk factor for the initiation, growth, progression, and metastasis of tumor cells (Bingham et al., 2002; Radulescu et al., 2012; Guo et al., 2015). Besides, iron accumulation has an important role in multiple cell-death pathways, including iron-dependent cell death, i.e., ferroptosis, which suggests a potential therapeutic target to inhibit tumor development in cancer patients (Dixon et al., 2012; Basuli et al., 2017). Several preclinical/clinical studies have explored the anti-tumor activity and safety of depleting iron overload in tumors by means of iron chelators (Nutting et al., 2009; Yamasaki et al., 2011; List et al., 2012; Neufeld et al., 2012; Kalinowski et al., 2016). Targeting an increased iron level instead of iron chelators could also be a novel treatment option (Stockwell et al., 2017; von Hagens et al., 2017). Usually, researchers obtain values for iron by measuring circulating levels of iron-bound proteins (e.g., transferrin, ferritin) (Hambidge, 2003; Fonseca-Nunes et al., 2014; Morales and Xue, 2021). However, direct measurement of iron is more accurate to reflect the iron level in the body, but is poorly understood.

Thus, we explored the association between the baseline serum iron level at the diagnosis and clinical prognosis of women with early-stage TNBC. We aimed to establish a model on basis of direct measurement of the serum iron level for individualized prognosis predictions and treatment guidance.

In this study, we determined a cut-off value of 17.84 μmol/L for the serum iron level to stratify heterogeneous female with early-stage TNBC into low- and high-iron groups according to maximally selected rank statistics. Patients in the high-iron group had a significantly shorter median survival than those in the low-iron group. Multivariate Cox regression analysis revealed that a high serum iron level continued to be an independent predicator of poor survival in patients with early-stage TNBC. Then, a prognostic model combining the serum iron level and two clinicopathological factors (T stage, N stage) was established and represented graphically as a nomogram. The latter showed satisfactory discriminative accuracy and good predictive consistency between the actual survival probability and nomogram-predicted clinical outcome in the training cohort and SYSUCC-001 cohort, respectively.

Iron is essential for the activity or inhibition of various proteins and enzymes involved in many biological processes (Adams, 2015; Torti et al., 2018). However, iron also contributes to oxidative stress, which can result in damage to DNA. Increasing numbers of studies have demonstrated that homeostatic dysregulation of iron metabolism and changes in distribution of iron in serum are found in different types of cancers, including breast cancer (Galaris and Pantopoulos, 2008; Torti et al., 2018). Excess iron or iron overload due to dysregulation of iron homeostasis can promote the development, progression, and metastasis of tumor cells (Bingham et al., 2002; Radulescu et al., 2012; Guo W. et al., 2015).

The main source of biologically available iron comes from dietary intake. More and more studies have explored a positive relationship between the intake of red meat, heme iron, and initiation of breast tumors (Ferrucci et al., 2009; Guo J. et al., 2015; Inoue-Choi et al., 2016). Also, iron accumulation might have a significant role in multiple pathways of programmed cell death, including apoptosis, necroptosis, ascorbate-mediated death, and ferroptosis (Dixon et al., 2012; Basuli et al., 2017; Torti et al., 2018). Therefore, iron chelators, because of depleting iron levels in the body, have been investigated as a potential therapeutic strategy with promising outcomes for cancer patients (Nutting et al., 2009; Yamasaki et al., 2011; List et al., 2012; Neufeld et al., 2012; Kalinowski et al., 2016).

Excess iron can lead to lipid peroxidation, DNA/protein damage, as well as the initiation and progression of tumors (Galaris and Pantopoulos, 2008; Radulescu et al., 2012; Guo W. et al., 2015; Torti et al., 2018). Conversely, the toxicity of iron accumulation can promote lethal damage to tumor cells by peroxidation of membrane lipids, subsequently, contribute to interruption of tumorigenesis and tumor development (Wang et al., 2018; Chang et al., 2019). Hence, iron overload or iron depletion might provide potential targets for anti-tumor treatment. Therefore, exploring the relationship between iron levels in the body and cancer is a rational approach.

Most studies have assessed iron levels in the body by measuring circulating levels of iron-bound proteins (e.g., transferrin, ferritin) (Hambidge, 2003; Fonseca-Nunes et al., 2014; Morales and Xue, 2021). However, this strategy might generate errors in reflecting the actual iron level, and few studies have measured the iron level directly (Torti et al., 2018). In current study, we creatively measured the serum iron level rather than levels of transferrin or ferritin to represent the iron level in the body, and explored the prognostic value of serum iron levels, on which researches remain not to reach a consensus up to date. Feng et al. evaluated trace element levels in serum for patients with different types of cancer, they failed to find a significant difference of serum iron between liver, kidney tumors and normal tissues (Yang et al., 2021). Others researchers demonstrated that, compared with normal cells, tumor cells were more dependent upon iron, and that they remodeled iron-metabolism pathways to acquire, store, and efflux iron during their development and replication (Wang et al., 2018). Patients with cancer suffering from anemia due to cachexia or therapeutic drugs tend to have a low serum iron level. This phenomenon has been explored in several tumor types, including breast cancer (Torti et al., 2018; Yang et al., 2021). In this study, we collected baseline blood samples < 1 week of the diagnosis, and defined an optimal cutoff of 17.84 μmol/L for the serum iron level. Based on the latter, we stratified patients with early-stage TNBC into two groups with significantly different survival outcomes.

Based on classification of the serum iron level, a prognostic model incorporating the serum iron level as well as the traditional T stage and N stage was developed. The common 21-gene recurrence score, 70-gene MammaPrint Assay, and the PAM50 prognostic model are limited to a specific subtype or lymph node-negative breast cancer or patients at high clinical risk from breast cancer with limited predictive accuracy of the C-index (Gnant et al., 2015; Wallden et al., 2015; Ibraheem et al., 2020; Poorvu et al., 2020). Our predictive nomogram was accurate, cost-efficient, convenient, and readily available in hospitals in developing countries. The TNM staging system is used commonly for risk stratification and therapeutic recommendations. However, TNM criteria are based on a limited number of clinical factors, and their discriminative accuracy is limited due to differences between patients (Bareche et al., 2018; Grosselin et al., 2019). According to our time-dependent ROC curves, the predictive accuracy of our prognostic model was higher than that of the traditional TNM staging system in the training cohort and SYSUCC-001 cohort, which suggests that our nomogram might be a potential supplement to the traditional TNM staging system. Besides, except from the TNM staging system, a series of prognostic models based on inflammatory status, tumor marker, stromal tumor-infiltrating lymphocytes, and kinds of gene signature have been explored with C-index ranging 0.69–0.77 (Shi et al., 2019; Yang et al., 2019; Zheng et al., 2020), compared with them, our prognostic models achieved a comparative predictive accuracy, and was more cost-efficient and convenient. As far as we know, our study is the first to propose a predictive model integrating the impact of trace element iron with clinicopathological features. But it should be noted that a further exploration about the preliminary mechanisms is warrant.

Our study had three main limitations. First, a retrospective study will have a selection bias. Nevertheless, we tried our best to enroll all eligible TNBC patients to minimize a selection bias, and validated our prognostic models in a cohort from previous randomized trial SYSUCC-001. Second, we measured only the baseline serum iron level at the diagnosis. It would have been preferable to monitor the dynamic change in the serum iron level during therapy and adjust the therapeutic strategy. Third, we included early-stage TNBC patients only from China. Hence, the availability and predictive accuracy of our prognostic nomogram to women from other geographic regions are required to be warrant in future study.

We proposed a cutoff of the serum iron level to stratify patients with early-stage TNBC into high- and low-iron groups. On basis of the serum iron level, we established a predictive model for individualized survival prediction and validated it in the SYSUCC-001 cohort. The prognostic nomogram showed good predictive performance and satisfactory consistency compared with the actual clinical outcome.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Ethics Committee of Sun Yat-sen University Cancer Center. The ethics committee waived the requirement of written informed consent for participation.

ZY designed this study. XH, FD, and JH collected, primarily analyzed, and interpreted data. XH, FD, JH, CS, LW, and CJ participated in the drafting of the manuscript. XH, FD, JH, WX, XB, and ZY contributed to administrative, technical, or material support. All authors revised this manuscript and approved the final submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.