Value of apparent diffusion coefficient histogram analysis in the assessment of response to neoadjuvant chemotherapy in patients with in locally advanced breast cancer / 中华放射学杂志

ABSTRACT

Objective To investigate the value of ADC histogram analysis in the assessment of response to neoadjuvant chemotherapy (NACT) in patients with in locally advanced breast cancer (LABC). Methods Thirty?five female patients with invasive ductal carcinoma proved by pathology before NACT and treated with operation after NACT were retrospectively analyzed. All patients were received MR examination (including non?enhanced MRI, enhanced?MRI, and DWI) breast before NACT. After neoadjuvant chemotherapy, 19 of 35 patients were categorized as responders and 16 were categorized as non?responders according response evaluation criteria in solid tumors criteria. Per?patient weighted ADC histograms were generated. Mean ADC, mode ADC, maximum ADC, minimum ADC, median ADC, skewness, and kurtosis were analyzed by using t test between responders and non responders groups. ROC curves were constructed to determine the optimum threshold for each histogram parameter to differentiate non?responders and responders in breast cancers. The optimal threshold values, determined by maximal Youden index were selected when significant differences existed in two groups. Results Mean, minimum, skewness, and kurtosis of ADC between responders and non?responders group were(0.955 ± 0.135)× 10?3mm2/s,(0.535 ± 0.115)×10?3mm2/s,0.85±0.61, 2.93±0.17,and(1.103±0.233)×10?3 mm2/s,(0.650±0.104)×10?3mm2/s,-0.42± 0.17, 3.11 ± 0.25,respectively. Significant differences were found mean ADC, minimum ADC, skewness, and kurtosis (t=2.345, 3.096, 8.051 and 2.524,P0.05).We set the optimal threshold criteria of mean ADC (0.956×10?3mm2/s), minimum ADC (0.580×10?3mm2/s), skewness (0.890), sensitivity, specificity of three parameters for predicting responders in LABC were 73.7%,62.5%, 78.9%,68.8%, and 63.2%,75.0%, respectively, and the areas under ROC curve of mean ADC, minimum ADC, skewness was 0.678, 0.770, and 0.890, respectively. Kurtosis of responders and non?responders did not get cutoff value for much more overlap. Conclusion ADC histogram analysis is valuable in predicting LABC in patients with NACT effect, the minimum and skewness of ADC is highest sensitivity, specificity, respectively.