Heat shock proteins create a signature to predict the clinical outcome in breast cancer

Marta Klimczak, Przemysław Biecek, Maciej ŻyliczScientific 2019 · DOI

Abstract

Utilizing genomic databases, researchers identified a six-gene heat shock protein (HSP) signature associated with breast cancer survival. While HSPA2 and DNAJC20 expression correlate with improved outcomes, overexpression of HSP90AA1, CCT1, CCT2, and CCT6A indicates a poor prognosis. A novel risk score model combining these HSPs with clinical staging was developed and validated, offering a potential tool for more precise patient stratification.

Introduction

Breast cancer is a highly heterogeneous disease, where patients at the same clinical stage often experience vastly different outcomes. Standard molecular markers like ER, PR, and HER2 provide a baseline, but there is a critical need for multigene classifiers that can more accurately predict individual risk.

Heat shock proteins (HSPs) are molecular chaperones essential for protein folding under cellular stress—a condition common in tumor environments. While traditionally viewed as pro-oncogenic "chaperones" that aid cancer cell survival, their specific prognostic value across different families (HSP70, HSP90, and Chaperonins) has remained complex. This study aims to define a validated HSP signature to refine breast cancer prognosis.

Methodology

The research team employed a multi-step bioinformatic workflow to identify and validate prognostic markers:

  1. Screening: Analysis of 96 HSP-encoding genes across The Cancer Genome Atlas (TCGA) and KM plotter databases.
  2. Statistical Filtering: Log-rank tests identified six genes (HSPA2, DNAJC20, HSP90AA1, CCT1, CCT2, and CCT6A) significantly associated with overall survival in both cohorts.
  3. Modeling: Univariate and multivariate Cox regression analyses were used to develop a risk score formula integrating five of these HSPs with clinical tumor stage.
  4. Validation: The model's predictive power was confirmed using the independent METABRIC dataset.
  5. Functional Analysis: Gene Set Enrichment Analysis (GSEA) identified biological pathways active in high-risk versus low-risk patient groups.

Key Results and Insights

The Dual Nature of Chaperones

A central finding of this study is that HSPs do not uniformly promote tumor progression. High expression of HSPA2 and DNAJC20 predicted significantly longer survival. In contrast, HSP90AA1 and the CCT subunits (CCT1, CCT2, CCT6A) were linked to unfavorable outcomes and aggressive clinical features.

Survival analysis and mRNA expression distributions for the six identified prognostic HSPs in breast cancer. Panels (A) show Kaplan-Meier curves for high vs. low expression, while panels (B) show the distribution of mRNA expression levels.

The Prognostic Signature

The researchers established a weighted risk score formula: Risk Score = (-0.4181 × HSPA2) + (-0.1813 × DNAJC20) + (0.6861 × HSP90AA1) + (0.0824 × CCT1) + (0.11 × CCT2) + (0.8427 × Stage).

This model achieved an Area Under the Curve (AUC) of approximately 0.66, demonstrating consistent accuracy in distinguishing high-risk from low-risk cases. In validation sets, the median survival for the high-risk group was 112.3 months, compared to 216.6 months for the low-risk group—a nearly twofold difference.

Biological Context and Pathway Divergence

The study revealed that "protective" HSPs like HSPA2 were more prevalent in tumors with wild-type TP53, whereas "unfavorable" HSPs were upregulated in TP53-mutant tumors. Furthermore, high-risk patients showed upregulation in cell-cycle regulators (E2F and MYC targets), while low-risk patients were associated with estrogen response pathways, typically found in less aggressive luminal subtypes.

Discussion and Future Directions

While the transcriptomic signature is statistically robust, it primarily reflects mRNA levels rather than protein activity or post-translational modifications. Additionally, pan-cancer analysis using the PRECOG tool suggested that the role of these HSPs is highly context-dependent; a protein that is protective in breast cancer may act as an oncogene in lung or bladder cancer.

Future work is focused on "wet lab" functional investigations to determine how manipulating these specific HSPs influences cancer cell proliferation and chemotherapy resistance. Understanding why certain chaperones act as tumor suppressors in specific contexts remains a vital frontier for personalized oncology.

Summary updated

Abstract

Utilizing The Cancer Genome Atlas (TCGA) and KM plotter databases we identified six heat shock proteins associated with survival of breast cancer patients. The survival curves of samples with high and low expression of heat shock genes were compared by log-rank test (Mantel-Haenszel). Interestingly, patients overexpressing two identified HSPs - HSPA2 and DNAJC20 exhibited longer survival, whereas overexpression of other four HSPs - HSP90AA1, CCT1, CCT2, CCT6A resulted in unfavorable prognosis for breast cancer patients. We explored correlations between expression level of HSPs and clinicopathological features including tumor grade, tumor size, number of lymph nodes involved and hormone receptor status. Additionally, we identified a novel signature with the potential to serve as a prognostic model for breast cancer. Using univariate Cox regression analysis followed by multivariate Cox regression analysis, we built a risk score formula comprising prognostic HSPs (HSPA2, DNAJC20, HSP90AA1, CCT1, CCT2) and tumor stage to identify high-risk and low-risk cases. Finally, we analyzed the association of six prognostic HSP expression with survival of patients suffering from other types of cancer than breast cancer. We revealed that depending on cancer type, each of the six analyzed HSPs can act both as a positive, as well as a negative regulator of cancer development. Our study demonstrates a novel HSP signature for the outcome prediction of breast cancer patients and provides a new insight into ambiguous role of these proteins in cancer development.
BibTeX
@inproceedings{klimczak2019heat,
  author = {Marta Klimczak and Przemysław Biecek and Maciej Żylicz and Maciej Żylicz},
  title = {{Heat shock proteins create a signature to predict the clinical outcome in breast cancer}},
  booktitle = {Scientific},
  year = {2019},
  doi = {10.1038/s41598-019-43556-1},
}
Figure