Predictive Value Of MR Multimodal Imaging For Molecular Subtypes Of Breast Cancers

Objective To explore the predictive value of multimodal MR imaging for different molecular subtypes of breast cancers,and to find MRI features and ADC thresholds that have significant significance in the prediction of molecular subtypes of breast cancers.Materials and Methods Retrospective analysis of 256 patients who were diagnosed as non-special invasive breast cancer by pathology at the Hospital of Ningxia Medical University from July 2017 to November 2019.All patients were performed Magnetic resonance examination(including T2 WI,DCE-MRI and DWI)before pathological biopsy or treatment.According to the results of pathological immunohistochemistry,breast cancer molecular subtypes was divided into five groups: luminal A-like,luminal B-like(HER-2negative),luminal B-like(HER-2 positive),HER-2 overexpression and triple negative breast cancers,is referred to as Group A,Group B1,Group B2,Group H,and Group T in order.Observe the T2 WI sequence to record the T2 WI signal intensity of the lesions and whether there is chest edema.Observe the DCE-MRI sequence to record the lesion type(mass /non-mass),and focus on the morphological characteristics of the mass-type lesions including tumor maximum diameter,number of lesions,shape,boundary,presence or absence of spicule sign and internal enhancements of lesions.At the same time,the dynamic characteristics(ie,time-signal intensity curve)of the DCE-MRI sequence were evaluated at the post-processing workstation.The distribution of the above characteristics in different molecular types was analyzed using chi-square tests and Fisher's exact probability test,using multiple logistic regression analysis on statistically significant MRI features,and calculating the odds ratio(OR)of each factor.the average ADC(ADCmean)value,minimum ADC(ADCmin)and maximum ADC(ADCmax)values of the whole lesion were measured in the post-treatment workstation according to the DWI sequence.All data verify normality and homogeneity of variance,satisfiers use single-factor ANONA to analyze whether there is a statistical difference in ADC values of different molecular subtypes.If no satisfaction use Kruskal-Wallis H test,then use the receiver operate characteristic curves(ROC curves)to analyzed the ADC value with statistical difference and the threshold value is obtained.Results Among the 256 breast cancer patients,there were 63(24.6%)patients in group A,77(30.1%)patients in group B1,19(7.4%)patients in group B2,40(15.6%)patients in group H,and 57(22.3%)in group T,B1 group is the most.(1)There is a correlation between breast cancer molecular typing and histological grade(r =0.256,P value = 0.000),but there is no correlation with age and menopausal status(P>0.05).(2)molecular subtypes of breast cancer were significantly associated with breast edema,tumor maximum diameter,spicule sign,internal enhancement of lesions,TIC curve(P <0.05).Multivariate logistic regression analysis found that group A and B1 were more likely to have uniform and uneven enhancement than group T,and group A,B1,and B2 were more likely to have glitches than group H and T;group H and T The possibility of chest edema was higher than that in group A;the TIC curve type II(platform type)in group A and group B1 was higher than that in group H;group A,B1 and group T were more likely to have a tumor with a maximum diameter of 20 mm or less.(3)The minimum ADC value,average ADC value,and maximum ADC value of different molecular types of breasts were statistically different(P <0.05),and the pairwise comparison analysis found that the minimum ADC(ADCmin)value of group B1 was different from that of group H,There was a statistical difference in group T(P = 0.000,P = 0.002).The minimum ADC(ADCmin)value in group A was statistically different from that in group H(P = 0.003).The average ADC(ADCmean)and maximum ADC(The ADCmax value was statistically different from the B2 group,the H group and the T group(P <0.05).The ADC values of theabove statistically significant groups were analyzed by ROC curve,except for ADCmean(B1group vs T group)and Except for ADCmin(B1 group vs T group),the area under the curve of the other groups is greater than 0.7,which has higher accuracy.Conclusions(1)The characteristics of chest edema based on T2 WI sequence of the fat suppression axis evaluation can be used as MRI features different from morphological features to evaluate the molecular subtypes of breast cancers.(2)The morphological features include spicule sign,internal enhancement of lesions,tumor maximum diameter and TIC curve based on DCE-MRI evaluation,which have significant significance for different molecular subtypes of breast cancers.(3)The average ADC value,the maximum ADC value,and the minimum ADC value are significant for different molecular subtypes of breast cancers.And compared with luminal B-like(HER-2 negative),luminal B type(HER-2 positive),HER-2 overexpression and triple negative breast cancers,the maximum ADC value have significant singnificance;the minimum ADC value was found to be more discriminating for luminal A type and HER-2overexpression type.(4)Multimodal imaging has higher predictive value for breast cancer molecular subtypes assessment.