Whole-lesion histogram analysis of apparent diffusion coefficient for the prediction of pathological complete response to neoadjuvant chemotherapy in different subtypes of breast cancer

Objective To investigate the value of whole-lesion histogram parameters of apparent diffusion coefficient (ADC) in evaluating and predicting the pathological complete response(PCR) to neoadjuvant chemotherapy (NAC) in different subtypes of breast cancer. Methods This retrospective study included 117 patients with breast cancer who underwent MRI examination before NAC prior to surgery from January 2016 to December 2017 in the First Affiliated Hospital of Nanjing University. All cases were divided into Luminal B, HER2 positive (n=21) and triple negative (n=26) groups. The surgical pathology after chemotherapy was evaluated by Miller-Payne (M-P) system and the patients were divided into PCR group and non-PCR (nPCR) group. Firevoxel software was used to generate the whole-lesion ADC histogram. The parameters included mean (ADCmean), skewness, kurtosis, the minimum (ADCmin), the maximum (ADCmax), 10th percentile(ADC10%), 50th percentile (ADC50%) and 90th percentile (ADC90%). The two independent samples t test or Mann-Whitney U test was used to compare the differences between PCR and nPCR groups in each subtype. The diagnostic performance of statistically different ADC parameters for predicting PCR was evaluated by receiver operating characteristic (ROC) curve. Results Kurtosis was significantly higher in PCR group than that in nPCR group in HER2 positive subtype (P=0.039). It achieved an area under the curve (AUC) of 0.813 with sensitivity of 100% and specificity of 68.7% at the optimal cutoff value (1.861) for differentiating PCR from nPCR cases. In triple negative subtype, ADCmean and ADC50% were smaller in PCR group than those in nPCR group (P=0.028,0.013). They achieved AUCs of 0.800, 0.842, respectively. When ADCmean of 1.030×10-3 mm2/s and ADC50% of 0.976×10-3 mm2/s were used as cutoff value to differentiate PCR from nPCR, the sensitivities were 75.0%, 80.0% and the specificities were 83.3%, 83.3%, respectively. Conclusion Kurtosis can predict post-NAC PCR in patients with HER2 positive breast cancer, while ADC50% has a high value in predicting post NAC PCR of triple negative breast cancer patients. Key words: Breast neoplasms; Molecular subtype; Neoadjuvant chemotherapy; Magnetic resonance imaging