Scholarlink

Background <p>The pan-AKT inhibitor MK2206 has shown improved but limited efficacy in achieving pathologic complete response (pCR) in patients with breast cancer receiving standard neoadjuvant chemotherapy (NACT). This study aimed to develop non-invasive tumor heterogeneity index (THI) from dynamic-contrast-enhanced magnetic resonance imaging (DCE-MRI) that can better select AKT-inhibition responders than immunohistochemistry (IHC)-based receptors and explore associated tumor microenvironment (TME).</p> Patients and methods <p>We included 987 patients who underwent surgery following neoadjuvant chemotherapy (NACT) in the I-SPY2 trial, which comprised three cohorts: THI discovery, THI validation, and TME exploration. THI was constructed by regression on the consensus clustering labels using repeatable, reproducible, and specifically predictive radiomics-based features. The predictive value of THI for treatment outcomes was evaluated using AUC-ROC with cross-validation, while its clinical utility was assessed through Bayesian logistic regression. Additionally, we explored differentially expressed phosphoproteins, proteins, genes, biological mechanisms, and the tumor immune microenvironment.</p> Results <p>Seven repeatable image features were identified to be specifically predictive in the MK2206 arm and yielded two consensus clusters. The developed THI accurately predicted the two clusters, achieving an AUC-ROC of 1 in the discovery cohort and 0.99 in the remaining patients. It was significantly associated with treatment response in the AKT inhibition cohort with a cross-validation AUC of 0.73 (95% CI 0.63 to 0.83), and in corresponding subgroups (HER2−, HER2+, HR−) where AUCs ranged from 0.73 (95% CI 0.63 to 0.83) to 0.85 (95% CI 0.73 to 0.96). Bayesian analysis indicated that patients with high THI had a higher estimated average pCR rate of 56.1% in the AKT inhibition arm compared to 18.7% in the control arm (<i>p</i> &lt; 0.001). Furthermore, high THI was associated with a "hot" tumor immune microenvironment, characterized by active immune pathways, extensive immune cell infiltration, and decreased cell cycle activity.</p> Conclusions <p>The Tumor Heterogeneity Index was found to be specifically and independently associated with the response to AKT inhibition, highlighting its potential to aid in responder prediction. Additionally, the associated tumor microenvironment enhances our understanding of the antitumor mechanisms underlying AKT inhibition.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Tumor heterogeneity index from DCE-MRI for AKT-inhibition responder identification and reveals hot immune microenvironment for patients with breast cancer: a multi-omics analysis of I-SPY2 trial

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月5日

查看全文
  • Xinzhi Teng,
  • Jiang Zhang,
  • Qingpei Lai,
  • Wen Li,
  • Xinyu Fan,
  • Xinyu Zhang,
  • Yu-hua Huang,
  • Jiahuan Wang,
  • Chenyang Liu,
  • Ge Ren,
  • Habib Zaidi,
  • Elaine Yuen Phin Lee,
  • Aya El Helali,
  • Lina Zhao,
  • Defeng Sun,
  • Jing Cai

摘要

Background

The pan-AKT inhibitor MK2206 has shown improved but limited efficacy in achieving pathologic complete response (pCR) in patients with breast cancer receiving standard neoadjuvant chemotherapy (NACT). This study aimed to develop non-invasive tumor heterogeneity index (THI) from dynamic-contrast-enhanced magnetic resonance imaging (DCE-MRI) that can better select AKT-inhibition responders than immunohistochemistry (IHC)-based receptors and explore associated tumor microenvironment (TME).

Patients and methods

We included 987 patients who underwent surgery following neoadjuvant chemotherapy (NACT) in the I-SPY2 trial, which comprised three cohorts: THI discovery, THI validation, and TME exploration. THI was constructed by regression on the consensus clustering labels using repeatable, reproducible, and specifically predictive radiomics-based features. The predictive value of THI for treatment outcomes was evaluated using AUC-ROC with cross-validation, while its clinical utility was assessed through Bayesian logistic regression. Additionally, we explored differentially expressed phosphoproteins, proteins, genes, biological mechanisms, and the tumor immune microenvironment.

Results

Seven repeatable image features were identified to be specifically predictive in the MK2206 arm and yielded two consensus clusters. The developed THI accurately predicted the two clusters, achieving an AUC-ROC of 1 in the discovery cohort and 0.99 in the remaining patients. It was significantly associated with treatment response in the AKT inhibition cohort with a cross-validation AUC of 0.73 (95% CI 0.63 to 0.83), and in corresponding subgroups (HER2−, HER2+, HR−) where AUCs ranged from 0.73 (95% CI 0.63 to 0.83) to 0.85 (95% CI 0.73 to 0.96). Bayesian analysis indicated that patients with high THI had a higher estimated average pCR rate of 56.1% in the AKT inhibition arm compared to 18.7% in the control arm (p < 0.001). Furthermore, high THI was associated with a "hot" tumor immune microenvironment, characterized by active immune pathways, extensive immune cell infiltration, and decreased cell cycle activity.

Conclusions

The Tumor Heterogeneity Index was found to be specifically and independently associated with the response to AKT inhibition, highlighting its potential to aid in responder prediction. Additionally, the associated tumor microenvironment enhances our understanding of the antitumor mechanisms underlying AKT inhibition.