Superior performance in classification of breast cancer molecular subtype and histological factors by radiomics based on ultrafast MRI over standard MRI: evidence from a prospective study

Abstract

Purpose: To compare the performance of ultrafast MRI with standard MRI in classifying histological factors and subtypes of invasive breast cancer among radiologists with varying experience.

Methods: From October 2021 to November 2022, this prospective study enrolled 225 participants with 233 breast cancers before treatment (NCT06104189 at clinicaltrials.gov). Tumor segmentation on MRI was performed independently by two readers (R1, dedicated breast radiologist; R2, radiology resident). We extracted 1618 radiomic features and four kinetic features from ultrafast and standard images, respectively. Logistic regression algorithms were adopted for prediction modeling, following feature selection by the least absolute shrinkage and selection operator. The performance of predicting histological factors and subtypes was evaluated using the area under the receiver-operating characteristic curve (AUC). Performance differences between MRI methods and radiologists were assessed using the DeLong test.

Results: Ultrafast MRI outperformed standard MRI in predicting HER2 status (AUCs [95% CI] of ultrafast MRI vs standard MRI; 0.87 [0.83-0.91] vs 0.77 [0.64-0.90] for R1 and 0.88 [0.83-0.91] vs 0.77 [0.69-0.84] for R2) (all P < 0.05). Both ultrafast MRI and standard MRI showed comparable performance in predicting hormone receptors. Ultrafast MRI exhibited superior performance to standard MRI in classifying subtypes. The classification of the luminal subtype for both readers, the HER2-overexpressed subtype for R2, and the triple-negative subtype for R1 was significantly better with ultrafast MRI (P < 0.05).

Conclusion: Ultrafast MRI-based radiomics holds promise as a noninvasive imaging biomarker for classifying hormone receptors, HER2 status, and molecular subtypes compared to standard MRI, regardless of radiologist experience.

Keywords: Breast cancer; Histological factor; Magnetic resonance imaging; Radiomics; Subtype.

Fig. 1

Flowchart of the study participants. A total of 233 breast cancers of 225 participants were included

Fig. 2

Full MRI protocol used in our study. T1 T1-weighted, T2 T2-weighted, DWI diffusion-weighted imaging

Fig. 3

Ultrafast and standard MRI in a 30-year-old woman with triple-negative invasive ductal carcinoma. A Ultrafast MRI taken 16.8 s after contrast agent injection. B Standard MRI taken 58.8 s after contrast agent injection. Both images show an oval-shaped, irregular marginated, rim enhancing mass in the left outer breast. For three-dimensional tumor segmentation, the entire enhanced tumor margins were drawn from top to bottom of each tumor in axial views of U2 phase postcontrast ultrafast MRI and the initial phase postcontrast standard MRI. Among the 14 ultrafast images, U1 is the first phase in which the signal intensity of the hotspot region of interest of the tumor is 10% higher than the average signal intensity of unenhanced images. U2 is the immediate next phase of U1 and the well-established tumor enhancement time

Fig. 4

Illustration of the radiomics workflow. Three-dimensional tumor segmentation was performed on ultrafast MRI and standard MRI by two readers (Reader 1: a dedicated breast radiologist, Reader 2: a radiology resident), independently. The tumor segmentation agreement between two readers was evaluated using Dice and Jaccard similarity coefficients. After segmentation, 1618 radiomic features (first-order statistical features, shape and volume features, texture features, and wavelet-transformed features) were extracted. Kinetic features obtained from ultrafast MRI and standard MRI were added to the radiomic features. MRI feature selection was performed using the LASSO method. We tested the associations between MRI radiomic features and histological factors and subtypes in two models of ultrafast and standard MRI. The model performance of each reader was evaluated using the AUC analysis. LASSO least absolute shrinkage and selector operator, AUC the area under the receiver-operating characteristic curve

Fig. 5

Tumor segmentation agreement between two readers

Fig. 6

Radiomics heat map. Heat maps show associations between radiomic features from ultrafast MRI (A) and standard MRI (B) in Reader 1 and Reader 2, respectively, and histological factors and molecular subtypes. Maps illustrate the variation in signal intensity across multiple features, with each row representing a specific feature. A color intensity scale from low (blue) to high (yellow) is provided. Histograms below each heat map detail the distribution of specific histological markers, including HR and HER2 status, and luminal, HER2-overexpressed, and triple-negative subtypes. These markers were dichotomized into binary categories, denoted as 0 or 1