Prediction of Lymphovascular invasion status in breast cancer based on magnetic resonance imaging radiomics features

In In The News by Barbara Jacoby

From: sciencedirect.com

Abstract

Objective

This study intended to investigate the feasibility and effectiveness of using clinical magnetic resonance imaging (MRI) radiomics features to predict lymphovascular invasion (LVI) status in breast cancer (BC) patients.

Methods

A total of 182 BCE patients were retrospectively collected and randomly divided into a training set (n = 127) and a validation set (n = 55) in a 7:3 ratio. Based on pathological examination results, the training set was further divided into LVI group (n = 60) and non-LVI group (n = 67), and the validation set was divided into LVI group (n = 24) and non-LVI group (n = 31). General data and MRI examination indicators were compared. Multivariate logistic regression was utilized to analyze MRI radiomics features and clinically relevant indicators that were significant in the baseline information of patients in training set, independent risk factors were identified, and a logistic regression model was built. The accuracy of logistic model was validated using ROC curves in training and validation sets.

Results

Age, pathohistological classification, tumor length, tumor width, presence or absence of Magnetic Resonance Spectroscopy (MRS) cho peak, presence or absence of spicule sign, peritumoral enhancement, and peritumoral edema were statistically significant (P < 0.05) between the two groups. Multivariate logistic regression analysis presented that spicule and peritumoral edema were independent risk factors for LVI in BC patients (P < 0.05). The ROC curve illustrated that AUC of the logistic regression model in the training set was 0.807 (95%CI: 0.730–0.885) and that in the validation set was 0.837 (95%CI: 0.731–0.944).

Conclusion

Radiomics features of spicule sign and peritumoral edema were independent risk factors for LVI in BC patients. A logistic regression model based on these factors, along with age, could accurately predict LVI occurrence in BC patients, providing data support for diagnosis and modeling of LVI in BC patients.

Introduction

With the improvement in living standards, breast cancer (BC) has become one of the most common malignant tumors in women worldwide, surpassing lung cancer, accounting for about one-quarter of female cancer cases, according to research data from 2020 [1]. Early BC is considered a curable disease, however, once the disease metastasizes, the 5-year overall survival rate is significantly reduced [2]. Therefore, preoperative identification of BC metastasis status has important reference value for prognostic evaluation, staging, and treatment plan formulation of BC.

BC metastasis pathways include local infiltration, hematogenous metastasis, and lymphatic metastasis [3], of which lymphatic metastasis involves continuous processes like tumor-related lymphatic blood vessel infiltration, lymphatic vessel generation, and distant organ micro-metastasis proliferation [4]. Lymphovascular invasion (LVI) is a pathological manifestation of thrombosis in the peripheral lymphatic and vascular system of invasive BC. Its occurrence and development process involve complex molecular pathways, which are not only related to the interaction of the surrounding microenvironment but also involve the changes of malignant epithelial cells, with strong invasive and migratory abilities [5]. LVI is an adverse prognostic factor of BC, and its prognostic value is independent of histological grade, which can assist in judging the staging and treatment plan selection of BC [6,7], and accurate identification of LVI status has practical significance for BC treatment. Currently, LVI status mainly relies on postoperative or puncture biopsy for pathological diagnosis, and there is a lack of effective tools for accurately predicting LVI status preoperatively. Therefore, studying effective preoperative prediction models is of great significance for the presence of BC LVI.

BC magnetic resonance imaging (MRI) radiomics can improve the accuracy of BC MRI diagnosis by high-throughput extraction and quantitative analysis of tumor information [8], and provide more basis for the development of scientific, accurate, and personalized treatment plans for BC patients. Multiple investigations presented that dynamic parameters such as the apparent diffusion coefficient (ADC) value in BC MRI [9], peritumoral edema [10], the presence of circular high signals in diffusion-weighted imaging (DWI) [11], background parenchymal enhancement, and type II or III time-signal intensity curves [11,12] are significantly related to LVI positivity. However, the features discovered between studies are inconsistent, and the research results are not yet unified, making it difficult to reach a consistent conclusion. In order to reach a consensus on the risk factors related to LVI diagnosis, more research using different methods is needed. Therefore, this study used multiple indicators and parameters of clinical general data, pathological indexes, and MRI radiomics to establish a multi-variable model and explored the effect of a multi-variable model based on MRI radiomics features in predicting BC LVI status, so as to provide relevant information for effective prediction of BC LVI status in clinical practice.

Study population

A retrospective analysis was conducted on 182 patients with BC admitted to our hospital from August 15, 2016, to November 24, 2022. Inclusion criteria: female patients diagnosed with BC by pathological examination, detection of the pathological status of lymphatic and vascular systems around the BC, and determination of the presence of LVI; age over 18 years; patients with sufficient detailed clinical and pathological information. Exclusion criteria: pregnancy or in the preconception period;

Patient characteristics

A total of 127 BCE patients were included in the training set of this study, with 60 patients in BC LVI group and 67 patients in BC non-LVI group. Baseline information of the two groups was compared. According to the results in Table 1, significant differences were seen in patient age and pathohistological classification between two groups (P < 0.05). With respect to MRI-related indicators, significant differences were seen between two groups in tumor length, tumor width, MRS cho peak, presence

Discussion

Imaging-based analysis using post-processing software to extract quantitative imaging features can more accurately assess tumor heterogeneity, tumor grading, and phenotypic characteristics in the spatial domain of tumors [8]. In this study, we included multiple indicators of clinical demographic data, and MRI imaging features closely related to LVI status in BC patients to establish a predictive model based on imaging features for preoperative evaluation of BC LVI status. Results showed that

Authors’ contributions

(I) Conception and design: Xinhua Li.

(II) Administrative support: Kangwei Wu.

(III) Provision of study materials or patients: Kangwei Luo and Na Zhang.

(IV) Collection and assembly of data: Bin Li and Wubiao Chen.

(V) Data analysis and interpretation: Zhendong Lu and Yixian Chen.

(VI) Manuscript writing: Xinhua Li.

(VII) Final approval of manuscript: All authors.

Ethics approval and consent to participate

The study was approved by the ethics committee of The Affiliated Hospital of Guangdong Medical University. The methods were carried out in accordance with the approved guidelines (NO.PJKT2023–087).

Funding

Project supported by the 2022 Zhanjiang City science and technology project (NO. 2022B01053).

CRediT authorship contribution statement

Xinhua Li: Writing – original draft, Formal analysis, Data curation, Conceptualization. Kangwei Luo: Writing – original draft, Investigation, Formal analysis, Data curation. Na Zhang: Writing – review & editing, Project administration, Methodology, Investigation. Wubiao Chen: Writing – review & editing, Project administration, Methodology, Investigation. Bin Li: Supervision, Software, Resources, Project administration. Zhendong Lu: Visualization, Validation, Supervision, Software. Yixian Chen:

Declaration of competing interest

The authors declare that they have no competing interests.

References (24)