Application of simultaneous multi-slice accelerated readout-segmented echo planar diffusion-weighted imaging in...

To explore the role of simultaneous multi-slice (SMS) accelerated readout-segmented echo planar (RESOLVE) diffusion-weighted imaging (DWI) in assessing tumor response to neoadjuvant chemoradiotherapy (nCRT) in locally advanced rectal cancer (LARC). 42 clinically diagnosed LARC patients, who received nCRT were prospectively enrolled. All patients underwent pre-operative rectal MRI, including both RESOLVE and SMS-RESOLVE DWIs with respective scan time of 3min50s and 1min47s. Regions of interest covering the three largest slices of residual lesions were drawn independently by two radiologists to calculate mean ADC values. Two radiologists also independently determined T stage (ymrT) and tumor regression grade (mrTRG) using a 5-point grading system on T2WI with and without SMS-RESOLVE DWI on pre-operative MRI. According to the pathological data, patients were grouped into pathological complete response (pCR) and non-pCR, good response and poor response, low T stage and high T stage groups. The diagnostic performance of SMS-RESOLVE DWI was analyzed and compared with RESOLVE DWI and T2WI. The interobserver agreement was both excellent for ADC value measurements in RESOLVE (ICC = 0.908) and SMS-RESOLVE (ICC = 0.894) DWIs between two readers. Patients who achieved pCR or good response showed significantly higher ADC values than non-pCR (1.459 ± 0.173 vs. 1.205 ± 0.107*10^- 3mm²/s and 1.368 ± 0.106 vs. 1.177 ± 0.127*10^- 3mm²/s, respectively, both P < 0.001) or poor responders (1.411 ± 0.164 vs. 1.186 ± 0.115*10^- 3mm²/s and 1.333 ± 0.115 vs. 1.161 ± 0.125*10^- 3mm²/s, respectively, both P < 0.001) in both RESOLVE and SMS-RESOLVE DWIs. Low T stage group after nCRT also demonstrated higher ADC values than high T stage group (1.403 ± 0.211 vs. 1.205 ± 0.107*10^- 3mm²/s, p < 0.001 in RESOLVE, and 1.341 vs. 1.201*10^- 3mm²/s, p = 0.002 in SMS-RESOLVE DWI). Mean ADC values from SMS-RESOLVE DWI showed comparable good diagnostic power to RESOLVE DWI in identifying pCR (AUCs of 0.884 vs. 0.919, P = 0.344), good responders (AUCs of 0.84 vs. 0.9, P = 0.216), and low T stage patients (AUCs of 0.798 vs. 0.846, P = 0.231). Besides, SMS-RESOLVE DWI improved the interobserver agreement for evaluation of both ymrT stage (weighted kappa value of 0.693 vs. 0.528) and mrTRG (weighted kappa value of 0.644 vs. 0.53) on pre-operative MRI. Combining SMS-RESOLVE DWI and T2WI showed higher sensitivities (60% vs. 30%, 80% vs. 53.3%, and 81.8% vs. 50%) and negative predictive values (87.1% vs. 80%, 85% vs. 73.1%, and 71.4% vs. 57.7%) in selecting pCR, good responders and low T stage than T2WI alone. SMS-RESOLVE DWI demonstrated diagnostic performance comparable to conventional RESOLVE for quantitatively assessing T stage and tumor regression grade after nCRT in LARC. When added to T2WI, it showed potential to improve inter-reader consistency in qualitative restaging evaluation.