Diseases of the Esophagus (2017) 0, 1–7 DOI: 10.1093/dote/dox121 Original Article Prediction of pathologic responders to neoadjuvant chemoradiotherapy by diffusion-weighted magnetic resonance imaging in locally advanced esophageal squamous cell carcinoma: a prospective study Q.-W. Li,1,† B. Qiu,1,† B. Wang,1 D.-L. Wang,2 S.-H. Yin,2 H. Yang,3 J.-L. Liu,4 J.-H. Fu,3 M.-Z. Liu,1 C.-M. Xie,2,‡ H. Liu1,∗ 1 2 3 4 Departments of Radiation Oncology, Departments of Medical Imaging, Thoracic Oncology, and Medical Oncology, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Sun Yat-sen University Cancer Center Guangzhou, Guangdong, China SUMMARY. This study aims to investigate the role of diffusion-weighted magnetic resonance imaging (DWMRI) in ESCC patients receiving neoadjuvant concurrent chemoradiotherapy (CCRT), and the efficacy of apparent diffusion coefficient (ADC) values in predicting pathologic response to neoadjuvant CCRT. Twenty-eight locally advanced ESCC patients treated with neoadjuvant CCRT followed by radical resection were prospectively enrolled. DW-MRI was recommended to be performed within 2 weeks before and 4–6 weeks after neoadjuvant CCRT. The calculated ADCs pre- (ADC1) and post- (ADC2) neoadjuvant CCRT, the definite (ADC) and percentage changes (ADC%) were analyzed for the efficacy of predicting pathologic response to neoadjuvant CCRT. Twenty patients had been identified as responders (tumor regression grade 1–2). Among them, ADC2 (3.02 ± 0.84 vs. 2.12 ± 0.44 × 10−3 mm2 /s, P = 0.001) and ADC (1.22 ± 0.78 vs 0.64 ± 0.26 × 10−3 mm2 /s, P = 0.007) were significantly higher than those of nonresponders (tumor regression grade: 3–5). Receiver operating characteristic analysis revealed that ADC2 exhibited an overall accuracy of in 71.4% in predicting pathologic response, with a sensitivity of 60.0%, a specificity of 100%, a positive predictive value of 100%, and a negative predictive value of 50.0%, when 3.04 × 10−3 mm2 /s was used as the cutoff value. ADC value could be useful in predicting pathologic response to neoadjuvant CCRT in ESCC patients. High postneoadjuvant CCRT ADC is a predictive indicator for good response. KEY WORDS: apparent diffusion coefficient, esophageal squamous cell carcinoma, neoadjuvant chemoradiotherapy, tumor regression grade. INTRODUCTION Neoadjuvant concurrent chemoradiotherapy (CCRT) is currently regarded as a standard treatment option ∗ Address correspondence to: Hui Liu, MD, PhD, Department of Radiation Oncology, Sun Yat-sen University Cancer Center, No. 651 Dongfeng Road East, Guangzhou 510060, China. Email: firstname.lastname@example.org † Dr. Qi-Wen Li and Dr. Bo Qiu contributed equally to this work. ‡ Dr. Chuan-Miao Xie was the co-corresponding author. Specific author contributions: QWL and BQ analyzed and interpreted data, as well as drafted the article. DLW and SHY contributed in data acquisition, image analysis, and manuscript revision. BW, HY, JLL, JFH, MZL, CMX participated in data acquisition, data interpretation, and revised the manuscript. HL designed the study, interpreted data and revised the article. All authors gave final approval of the version to be submitted and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. for locally advanced esophageal cancer, yielding a greater chance of locoregional control and longer overall survival compared with surgery alone.1,2 However, response to neoadjuvant CCRT varies among individuals. It was reported that 29% of surgical specimen were completely free of tumor cells in pathologic examination.3 Forty-two percent of patients obtained tumor regression grade (TRG) 1–2, recognized as pathologic complete remission (pCR) or almost-pCR, while approximately half were found of poor response (TRG 3–5).4 The relationship between pathologic response and long-term prognosis has been widely investigated. Pathological CR is becoming a marker of favored survival and provided potential chance of omission of surgery.5 On the contrary, nonresponders are under the risk of disease progression, delayed resection, and unnecessary exposure to CCRT toxicities.6,7 It C The Authors 2017. Published by Oxford University Press on behalf of International Society for Diseases of the Esophagus. All rights reserved. For permissions, please e-mail: email@example.com 1 2 Diseases of the Esophagus is important to discriminate responders from nonresponders at an early stage, for the sake of individualizing the following treatments. Conflicting results have appeared when current modalities are assessed for response evaluation or prediction. Computed tomography (CT) and endoscopic ultrasonography (EUS) are routinely adopted but often suspected of difficulties in distinguishing residual tumor from fibrosis, tumor necrosis, or post-CCRT edema.8–9 18 fluorine-2-deoxy-D-glucose positron emission tomography-computed tomography (18 F-FDG PET/CT) also showed insufficient accuracy of only around 44%, along with the criticisms of high cost and radiation exposure.10 In contrast, diffusion-weighted magnetic resonance imaging (DW-MRI), a functional MR technique, allows a noninvasive way for assessing tumor cellularity quantitatively. The results are measured quantitatively by apparent diffusion coefficient (ADC), a parameter measuring the degree of restriction to diffusion of water protons within tissues in a certain volume.11 The indicative value of ADC on tumor response has been verified under several circumstances.12–15 In our previous study, the addition of DW-MRI to CT and endoscopy overwhelmed traditional modalities in response assessment for locally advanced ESCC treated by definitive chemoradiotherapy.16 Although a few studies have investigate the utility of DW-MRI in the neoadjuvant chemotherapy of esophageal cancer or gastro-oesophageal cancer,17,18 we focused our study on esophageal squamous cell carcinoma, and assess the predictive effect of DW-MRI in this radiosensitive pathologic subtype. The primary goal of this study is to investigate the role of DW-MRI-based ADC, both before and after neoadjuvant CCRT, in the prediction of pathologic response to neoadjuvant CCRT in ESCC. therapy, and without distant metastasis. Patients with contraindications for MRI, chemoradiotherapy, or surgery were excluded. Comprehensive work-ups consisting of contrast-enhanced CT of chest and abdomen, EUS and barium esophagram were performed at diagnosis for clinical staging. The staging procedure was based on American Joint Committee on Cancer 2009 standard. MATERIALS AND METHODS DW-MRI and imaging analysis Ethics, consent, and permissions All patients underwent two series of DW-MRI scanning, recommended within 2 weeks before and 4–6 weeks after neoadjuvant CCRT. A 1.5-T MR imaging unit (GE signaHDx 1.5T, General Electric, Milwaukee, WI, USA) with eight-channel phased array body coil was used. All patients were scanned in supine position and trained for shallow slow breath before the examination. Respiratory and electrocardiographically gated T2-weighted fast spin echo images were obtained. Parameters included repetition time (TR) 8500 ms, echo time (TE) 85 ms, thickness 5 mm, spacing 1 mm, field of view (FOV) 40 cm, matrix 320 × 224, incentive times (NEX) 1.00. Esophageal sagittal T2-weighted scans were adopted using the fast spin echo sequence with the following parameters: TR 5400 ms, TE 85 ms, thickness 3 mm, spacing 1mm, FOV 40 cm, matrix 320 × 224, NEX 1.00. This study was in accordance with the ethical standards of Helsinki Declaration, reviewed, and approved by the Ethic Committee of Sun Yat-sen University Cancer Center. Written informed consent was obtained from each participant. All records were anonymized and deidentified before analysis. Study population Twenty-eight consecutive patients pathologically diagnosed with locally advanced ESCC and hospitalized in our institution from November 2012 to May 2016 were prospectively enrolled. The following criteria were met: patients were ≥18 of age, without previous history of any malignancy or anticancer Neoadjuvant chemoradiotherapy Radiation of 40–44 Gy/20 fractions/4 weeks, planned and delivered by three-dimensional conformal radiotherapy, intensity modulated radiotherapy, or volumetric modulated arc therapy technique was administrated. The gross tumor volume (GTV) covered primary tumor and suspected metastatic lymph nodes determined by pretreatment examinations. The clinical target volume (CTV) included: 3 cm expansion of GTV to craniocaudal boundaries and 5 mm towards other directions, and area of regional lymph nodes. Planning target volume (PTV) was created from CTV by adding a uniform margin of 6 mm. One of the two regimens was selected for concomitant chemotherapy: 2 cycles of vinorelbine ditartrate (25 mg/m2 , d 1, d 8, d 22, d 29) and cisplatin (25 mg/m2 , d 1–4 and d 22–25); or weekly docetaxel (25 mg/m2 , d 1, d 8, d 15, d 22) and cisplatin (25 mg/m2 , d 1, d 8, d 15, d 22). Radical surgery was performed 4–8 weeks following the completion of neoadjuvant CCRT. The en bloc resection with two-field lymphadenectomy was performed, consisting of esophagectomy through right-sided posterolateral thoracotomy, gastric tube mobilization through midline laparotomy and anastomosis through left-side cervical incision. Alimentary tract reconstruction was achieved by a retrosternal or posterior mediastinal route. Response prediction in esophageal cancer 3 Table 1 Baseline clinicopathologic characteristics and their predictive effects Characteristics Sex Male Female Age (Mean ± SD) Length on pretreatment MRI (Mean ± SD) Location Proximal third Middle third Distal third Gastroesophageal junction Differentiation Low Middle High Missing cT cT2 cT3–4 cN cN0–1 cN2–3 Stage II III Dose 44 Gy 40 Gy Radiation technique 3D-CRT IMRT/VMAT Chemotherapy docetaxel + platinum vinorelbine + platinum All patients Responders Nonresponders p 25 3 56.86 ± 6.47 47.11 ± 16.04 18 2 57.85 ± 6.12 43.43 ± 15.47 7 1 54.38 ± 7.09 56.30 ± 14.4 >0.99 4 16 7 1 3 11 5 1 1 5 2 0 0.92 6 15 1 6 4 11 1 4 2 4 0 2 0.91 6 22 6 14 0 8 0.14 20 8 16 4 4 4 0.17 4 24 4 16 0 8 0.30 12 16 9 11 3 5 >0.99 12 16 7 13 5 3 0.23 12 16 8 12 4 4 0.69 0.21 0.053 Values of continuous variables are presented as means ± standard deviation (SD). ADC, apparent diffusion coefficient; 3D-CRT, threedimensional conformal radiotherapy; IMRT, intensity modulated radiotherapy; TRG, tumor regression grade; MRI, magnetic resonance imaging; VMAT, volumetric modulated arc therapy. DW-MRI were then obtained by a respiratory gated single-shot echo-planar imaging sequence and array spatial sensitivity encoding technique, with b values of 0 and 700 s/mm2 (TR 7500 ms, TE 70 ms, thickness 5 mm, spacing 1 mm, FOV 40 cm, matrix size 128 × 128, NEX 4.00). All original DW-MRIs were transferred to a Workstation (AW4.4; GE Healthcare, Milwaukee, Wis) and processed by Functool workstation, in order to create ADC maps. Two experienced radiologists who were blinded to clinical data were responsible for the interpretation of pictures. The lesion location was determined on the T2-weighted images. The lesion length was measured on sagittal T2-weighted images with a caliper tool. Meanwhile, T2-weighted images were used as a slice selection reference for ADC value assessment. DW-MRI reconstructed images were furtherly evaluated for image analysis. An axial slice showing the most predominant tumor size corresponding to T2-weighted images was selected. Region of interest (ROI) was contoured manually encompassing the entire essence of tumor displayed on the ADC map, excluding areas of necrotic, cystic, or hemorrhagic change. In case no residual tumor could be differentiated on the ADC map, the ROI was delineated according to the apparent tumor bed. The average ADC was extracted from the ADC map from the ROI, labeled as ADC1 (preneoadjuvant CCRT) and ADC2 (post neoadjuvant CCRT). The other parameters were calculated as follows: ADC = ADC2 − ADC1 ADC% = ADC/ADC1 Response evaluation An experienced pathologist, who was blinded to the DW-MRI data, evaluated the pathologic response based on TRG classification system:4 TRG 1, complete disappearance of tumor cells; TRG 2, rare residual tumor cells scattering throughout the fibrosis; TRG 3, an increased number of residual tumor cells outgrown by fibrostic tissue; TRG 4, residual cancer outgrowing fibrotic tissue; and TRG 5, no regressive signs. Patients were then divided into two groups, responders (TRG 1–2) and nonresponders (TRG 3–5), for statistical analysis. 4 Diseases of the Esophagus Table 2 ADCs and their predictive effects ADC (Mean ± SD) ADC1 (× 10−3 mm2 /s) ADC2 (× 10−3 mm2 /s) ADC (× 10−3 mm2 /s) ADC% (%) All patients Responders Nonresponders p 1.72 ± 0.44 2.77 ± 0.85 1.05 ± 0.72 64.17 ± 46.44 1.81 ± 0.41 3.02 ± 0.84 1.22 ± 0.78 71.25 ± 51.39 1.48 ± 0.44 2.12 ± 0.44 0.64 ± 0.26 46.49 ± 25.52 0.077 0.001 0.007 0.10 ADC, Apparent diffusion coefficient. Statistical methods Compared t-test was performed to assess the change of ADC pre- and postneoadjuvant CCRT. T-test of independent sampler was conducted to evaluate the association between ADCs and TRG classifications. ADC values significantly correlated with TRG were furtherly processed by receiver operating characteristic (ROC) analysis to calculate the best cutoff values. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated and reported. Comparisons of continuous and categorical clinic-pathologic factors were achieved by t-test and Fisher’s exact test, respectively. A statistically significant difference was identified when pvalue <0.05 (two-sided). All tests were performed using SPSS 16.0, except for ROC analysis, which was performed by MedCalc software 10.1.6.0. RESULTS Baseline clinicopathologic characteristics and their statistical association with TRG were detailed in Table 1. The median interval was 35 (range: 12–53) days between the end of neoadjuvant CCRT and the time of postneoadjuvant CCRT DW-MRI, and 48.5 (range: 37–72) days between the end of neoadjuvant CCRT and surgery. Fifteen patients were identified as TRG 1 (pCR), 5 as TRG 2, 4 as TRG 3, 3 as TRG 4 and 1 as TRG 5 in pathologic evaluation, or 20 (73.9%) as responders and 8 (26.1%) as nonresponders. As detailed in Table 2, the mean value of ADC increased significantly after neoadjuvant CCRT (1.72 ± 0.44 vs 2.77 ± 0.85 × 10−3 mm2 /s, P < 0.0001), and this trend was consistent among each participant, with 1.05 ± 0.72 × 10−3 mm2 /s as mean ADC and 64.17 ± 46.44% as ADC%. ADC1 was significantly inversely associated with clinical stage (2.13 ± 0.37 × 10−3 mm2 /s in stage II vs 1.65 ± 0.41 × 10−3 mm2 /s in stage III, P = 0.035). ADC2 (3.02 ± 0.84 vs 2.12 ± 0.44 × 10−3 mm2 /s, P = 0.001, Fig. 1) and ADC (1.22 ± 0.78 vs 0.64 ± 0.26 × 10−3 mm2 /s, P = 0.007) were significantly higher in patients with a good response compared with nonresponders. ADC1 advocated a marginally significant increase in responders (1.81 ± 0.41 vs 1.48 ± 0.44 × 10−3 mm2 /s, Fig. 1 ADC2 values in responders and nonresponders. ADC2 was significantly higher in patients with a good response compared with non-responders (3.02 ± 0.84 vs. 2.12 ± 0.44 × 10−3 mm2 /s, P = 0.001). ADC, apparent diffusion coefficient. P = 0.077), while neither ADC% (71.25 ± 51.39 vs 46.49 ± 25.52%, P = 0.21) nor other baseline clinicopathologic characteristics was statistically associated with pathologic endpoint (Table 1). Typical DW-MRI images for different tumor responses are presented in Fig. 2. ROC analysis was performed to determine the best cutoff value for prediction of pathologic response. With an area under curve (AUC) of 0.791, ADC2 exhibited a sensitivity of 60.0%, a specificity of 100%, a PPV of 100%, a NPV of 50.0% and an overall accuracy of 71.4%, when 3.04 × 10−3 mm2 /s was used as the cutoff value. The absence of statistical significance precluded ROC analysis for other ADC values. DISCUSSIONS Our study investigated the role of DW-MRI-based ADC values and its dynamic change in predicting early pathologic response to neoadjuvant CCRT in ESCC. Generally, neoadjuvant CCRT has been regarded as the standard therapy of locally advanced ESCC. In the era of precision medicine, early and accurate predictive markers for therapeutic response are highly desirable. Though efforts has been made on searching for clinical, molecular as well as radiological indicators, such as microRNA19 or PET/CT,10 none of them satisfactorily strikes a balance among Response prediction in esophageal cancer 5 Fig. 2 Typical magnetic resonance imaging pictures of responder and nonresponder. In a patient diagnosed with cT2N2 esophageal squamous cell carcinoma who obtained good pathologic response (TRG = 1), the mean ADC increased from 1.58 to 3.22 × 10−3 mm2 /s after neoadjuvant concurrent chemoradiotherapy (CCRT). T2-weighted, DW-MRI images and ADC maps with contoured regions of interest, before (a, b, c) and after (d, e, f) neoadjuvant CCRT were shown. In another case, staged cT3N1 identified with poor response (TRG = 5), the mean ADC did not change obviously during neoadjuvant CCRT (from 1.37 to 1.94 × 10−3 mm2 /s).T2-weighted, DW-MRI images and ADC maps before (g, h, i) and after (j, k, l) CCRT were presented. ADC, apparent diffusion coefficient; DW-MRI, diffusion-weighted magnetic resonance imaging; TRG, tumor regression grade. 6 Diseases of the Esophagus accuracy, harmlessness and convenience. Thus, ADC in DW-MRI was investigated in our study as a potential predictor of therapeutic response to neoadjuvant CCRT. In this study, 73.9% of patients were pathologically determined as responders, with 65.2% confirmed as pCR. The promising results might be related to the pathological type (squamous cell carcinoma),20 higher fractionated dose (44 Gy/20 fractions), sufficient chemotherapy and relatively longer interval between neoadjuvant CCRT and surgical intervention (mostly 6–8 weeks). Elevated ADC implied the change of characteristics of tumor cells and stroma, including loss of tumor cell membrane integrity, loss of stroma and angiogenesis, tumor apoptosis and necrosis,21 In our previous study, ADC value > 2.64 × 10−3 mm2 /s has become an additional criteria to CT and endoscopy, which significantly improved the accuracy of response assessment in ESCC patients treated by CCRT.16 In this study, ADCs had been identified as the response predictor as well. ADC2 > 3.04 × 10−3 mm2 /s was in a strong relation to good response, with overall accuracy of 71.4% and specificity of 100%. Similarly, De Cobelli et al. analyzed post-CCRT ADC with ROC curves, suggesting 1.84 × 10−3 mm2 /s as a favorable cut-off value.17 Another peer study included patients receiving definitive CCRT and radiological response assessment. When post-CCRT ADC was higher than 2.6 × 10−3 mm2 /s, not only a favored complete remission rate (81.1% vs 45.9%), but also an increased overall survival (2-year overall survival: 81.5% vs 16.9%) were identified.22 The various cutoff values suggested by different studies called for several considerations. First, the measurement of ADC2 relied primarily on method of ROI delineation. The contouring of residual tumor on DW-MRI was sometimes confusing, especially when the tumor disappeared completely on the image. In that case, the contouring of tumor bed was applied instead. Second, the postneoadjuvant CCRT ADC value might be dependent on the interval between the end of neoadjuvant CCRT and the time of postneoadjuvant CCRT DWMRI.18 In our study, the median interval was 35 days, which was comparable with that in van Rossums’ study and longer than that in De Cobelli’s.17,18 The most appropriate timing of second DW-MRI remains to be investigated. The predictive value of initial ADC was marginally significant in our study but varied among previous researches. According to Aoyagi et al., both radiological response and long-term survival were significantly better in patients with initially higher ADC, which was explained by the correlating increasing expression of vascular endothelial growth factor (VEGF) protein.15,21 In contrast, some researchers documented the opposite, yet with majority of participants diagnosed with gastric cancer and mostly consisted of adenocarcinoma.17 Another study focusing on definitive CCRT denied any statistical association between pre-treatment ADC and clinical outcomes.22 Therefore, the instability of predictive effect of ADC1 suggested the influence of heterogeneous study population, various adopted b-values and neoadjuvant CCRT protocols. Investigations with larger sample size are required for further explanation. ADC revealed its potential in response prediction, though less accurate than ADC2 in our study. The incompetence of prediction by ADC% was also noticed. It seemed that tumor regression was better reflected by postneoadjuvant CCRT DW-MRI directly, regardless of initial tumor status. Interestingly, in spite of contradictory results about ADC%,17,18 some suggested the increased rate of ADC early during CCRT, i.e. two weeks after treatment start, as an important indicator.18,23 It could be explained by the fact that incomplete pathologic response at early time-point might facilitate accurate contouring of residual tumor and increased the reliability of picture interpretation. ADC1 in stage II patients was found significantly higher than those of stage III in the current study. According to Van Rossum et al., a similar inverse association was noticed between T stage and initial ADC, which implied stratification of stage might be considered in the design of on-going studies. Besides, initial ADC was found to be higher in esophageal adenocarcinomas and poorly differentiated tumors, compared with squamous cell carcinomas and moderately differentiated tumors.18 Although the potential utility of ADC in squamous cell carcinoma and adenocarcinoma was demonstrated separately,17 subdivision of clinic-pathologic characteristics in future researches will reduce the complexity in interpretation of results. Growing interest has emerged in the use metabolic imaging techniques, such as PET/CT, in the assessment and prediction for tumor response. Although PET/CT and DW-MRI have not been compared directly, DW-MRI is known as a noninvasive method of apparent insensitivity to inflammation, with no exposure to radiation, no injection of contrast agent, short examination duration, and less cost. The measurement of ADC, no matter by manual or by semiautomated volumetric methods, presented with satisfactory interobserver reproducibility.24 The selection of imaging techniques and the optimal way to combine them is still an open question. Small sample size limited us from validating the results and cut-off points. Intervals between neoadjuvant CCRT and surgery covered a wide range in this study, inferring potential confounding factors. Another interesting research topic for future investigation is whether ADC remains its effectiveness in predicting progression-free survival or overall survival, which requires long-term follow-up. Response prediction in esophageal cancer 7 CONCLUSIONS In conclusion, this study showed that ADC postneoadjuvant CCRT, and the change of ADC after neoadjuvant treatment, were potential predictors of pathologic response in ESCC. Postneoadjuvant CCRT ADC showed remarkable accuracy in prediction of good response. Further prospective studies with larger sample size, homogenous population, and controlled procedures are required for validation. 10 11 12 13 ACKNOWLEDGMENTS 14 This work was supported by Medical Scientific Research Foundation of Guangdong Province, China [Grant Number A2015120]. We appreciate the assistance of Mr. Zhan Yihang for figure preparation. References 1 Shapiro J, van Lanschot J J, Hulshof M C et al. Neoadjuvant chemoradiotherapy plus surgery versus surgery alone for oesophageal or junctional cancer (CROSS): long-term results of a randomised controlled trial. Lancet Oncol 2015; 16: 1090–8. 2 Sjoquist K M, Burmeister B H, Smithers B M et al. Survival after neoadjuvant chemotherapy or chemoradiotherapy for resectable oesophageal carcinoma: an updated meta-analysis. Lancet Oncol 2011; 12: 681–92. 3 van Hagen P, Hulshof M C, van Lanschot J J et al. Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 2012; 366: 2074–84. 4 Mandard A M, Dalibard F, Mandard J C et al. Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma. Clinicopathologic correlations. Cancer 1994; 73: 2680–6. 5 Maas M, Beets-Tan R G, Lambregts D M et al. Wait-and-see policy for clinical complete responders after chemoradiation for rectal cancer. J Clin Oncol 2011; 29: 4633–40. 6 Chao Y K, Chang C B, Chuang W Y et al. Correlation between tumor regression grade and clinicopathologic parameters in patients with squamous cell carcinoma of the esophagus who received neoadjuvant chemoradiotherapy. Medicine 2015; 94: e1407. 7 Donahue J M, Nichols F C, Li Z et al. Complete pathologic response after neoadjuvant chemoradiotherapy for esophageal cancer is associated with enhanced survival. Ann Thorac Surg 2009; 87: 392–8; discussion 398–9. 8 Schneider P M, Metzger R, Schaefer H et al. Response evaluation by endoscopy, rebiopsy, and endoscopic ultrasound does not accurately predict histopathologic regression after neoadjuvant chemoradiation for esophageal cancer. Ann Surg 2008; 248: 902–8. 9 Westerterp M, van Westreenen H L, Reitsma J B et al. Esophageal cancer: CT, endoscopic US, and FDG PET for 15 16 17 18 19 20 21 22 23 24 assessment of response to neoadjuvant therapy—systematic review. Radiology 2005; 236: 841–51. Joye I, Deroose C M, Vandecaveye V, Haustermans K. The role of diffusion-weighted MRI and (18)F-FDG PET/CT in the prediction of pathologic complete response after radiochemotherapy for rectal cancer: a systematic review. Radiother Oncol 2014; 113: 158–65. Koh D M, Collins D J. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR Am J Roentgenol 2007; 188: 1622–35. Decker G, Mürtz P, Gieseke J et al. Intensity-modulated radiotherapy of the prostate: dynamic ADC monitoring by DWI at 3.0 T. Radiother Oncol 2014; 113: 115–20. Vandecaveye V, Dirix P, De Keyzer F et al. Diffusion-weighted magnetic resonance imaging early after chemoradiotherapy to monitor treatment response in head-and-neck squamous cell carcinoma. Int J Radiat Oncol Biol Phys 2012; 82: 1098–107. Lambrecht M, Vandecaveye V, De Keyzer F et al. Value of diffusion-weighted magnetic resonance imaging for prediction and early assessment of response to neoadjuvant radiochemotherapy in rectal cancer: preliminary results. Int J Radiat Oncol Biol Phys 2012; 82: 863–70. Aoyagi T, Shuto K, Okazumi S, Shimada H, Kazama T, Matsubara H. Apparent diffusion coefficient values measured by diffusion-weighted imaging predict chemoradiotherapeutic effect for advanced esophageal cancer. Dig Surg 2011; 28: 252–7. Qiu B, Wang D, Yang H et al. Combined modalities of magnetic resonance imaging, endoscopy and computed tomography in the evaluation of tumor responses to definitive chemoradiotherapy in esophageal squamous cell carcinoma. Radiother Oncol 2016; 121: 239–45. De Cobelli F, Giganti F, Orsenigo E et al. Apparent diffusion coefficient modifications in assessing gastro-oesophageal cancer response to neoadjuvant treatment: comparison with tumour regression grade at histology. Eur Radiol 2013; 23: 2165–74. van Rossum P S, van Lier A L, van Vulpen M et al. Diffusionweighted magnetic resonance imaging for the prediction of pathologic response to neoadjuvant chemoradiotherapy in esophageal cancer. Radiother Oncol 2015; 115: 163–70. Wen J, Yang H, Liu M Z et al. Gene expression analysis of pretreatment biopsies predicts the pathologic response of esophageal squamous cell carcinomas to neochemoradiotherapy. Ann Oncol 2014; 25: 1769–74. van Heijl M, van Lanschot J J B, Koppert L B et al. Neoadjuvant chemoradiation followed by surgery versus surgery alone for patients with adenocarcinoma or squamous cell carcinoma of the esophagus (CROSS). BMC Surg 2008; 8: 21. Aoyagi T, Shuto K, Okazumi S et al. Apparent diffusion coefficient correlation with oesophageal tumour stroma and angiogenesis. Eur Radiol 2012; 22: 1172–7. Wang L, Han C, Zhu S et al. Investigation of using diffusionweighted magnetic resonance imaging to evaluate the therapeutic effect of esophageal carcinoma treatment. Oncol Res Treat 2014; 37: 112–6. Imanishi S, Shuto K, Aoyagi T, Kono T, Saito H, Matsubara H. Diffusion-weighted magnetic resonance imaging for predicting and detecting the early response to chemoradiotherapy of advanced esophageal squamous cell carcinoma. Dig Surg 2013; 30: 240–8. Kwee R M. Prediction of tumor response to neoadjuvant therapy in patients with esophageal cancer with use of 18F FDG PET: a systematic review. Radiology 2010; 254: 707–17.