Value of diffusion weighted imaging on predicting radiotherapy and concurrent chemoradiotherapy response in patients with advanced non-small cell lung cancer

Abstract

Objective To investigate the value of DWI using 3.0 T MRI to predict response to radiotherapy(RT) and concurrent chemoradiotherapy(CCRT) in patients with advanced non-small cell lung cancer (NSCLC). Methods From January 2014 to May 2015, 40 patients with stage Ⅲ(Ⅲa or Ⅲb) NSCLC underwent DWI using 3.0 T MRI before RT/CCRT were enrolled. The imaging quality of diffusion-weighted images were evaluated on 3-level grades as good, moderate and non-diagnostic.The patients with good or moderate image quality were underwent DWI at 2 weeks after starting therapy(total dose of 20 Gy), and at the end of therapy (total dose of 60 Gy). Apparent diffusion coefficient(ADC) of lung cancer with good and moderate image quality were calculated by Funtool. The following quantitative parameters were recorded and calculated: the mean pretreatment ADC value(ADCpre), the mean mid-treatment ADC value (ADCmid), the mean post-treatment ADC value(ADCpost), the rate of changes inmean ADC value at 2 weeks post therapy (ΔADCmid) and the rate of changes inmean ADC value at the end of therapy(ΔADCpost). The patients were classified into response group and non-response group according to the tumor response, which was assessed with revised response evaluation criteria in solid tumors (RECIST1.1) after CCRT. The Mann-Whitney U test was used to compare parameters between the two groups.The relationship between these obtained parameters and tumor response was evaluated by Spearman correlation analysis. The value of parameters on predicting tumor response was calculated by receiver operating characteristic curve. Results 96.4% (80/83) DW images were graded as good or moderate image quality. The responders had lower median ADCpre[1.32 (0.77—1.96) × 10- 3 mm2/s] than non-responders[1.60(1.12—2.33) × 10- 3 mm2/s], which had statistically significant difference (Z=-2.934,P=0.003).Tumor regression rate after treatment had negative correlation with ADCpre(r=-0.386, P=0.018).The responders had increased ADC [ΔADCmid: 38.9% (12.8% —139.0%), ΔADCpost: 48.3% (25.6% —148.1%)] than non-responders [ΔADCmid:-2.5% (-15% —29.4%), ΔADCpost: 14.2% (-28.1% —71.3% )], which had statistically significant difference (Z=-2.847,-2.221, respectively; P<0.05). Tumor regression rate after treatment had positive correlation with ΔADCmid(r=0.637, P=0.001) and ΔADCpost(r=0.631, P=0.005).From ROC analysis,when setting threshold on pretreatment ADCpre=1.38×10- 3 mm2/s, ΔADCmid=21.6%, ΔADCpost=38.8%, the area under curve was 0.782, 0.838 and 0.813. Conclusion The mean ADC value before RT/CCRT and its changes during treatment is likely to be a valuabletool for predicting the response after RT/CCRT in advanced NSCLC, which may be helpful to clinical decision on individualized therapy. Key words: Lung neoplasms; Magnetic resonance imaging; Comparative effectiveness research