- Email: [email protected]
The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy
Radiotherapy and Oncology xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
Original article
The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy Lan Wang a, Lihong Liu a, Chun Han a,⇑, Shutang Liu a, Hua Tian a, Zhensheng Li a, Xuejiao Ren a, Gaofeng Shi b, Qi Wang b, Guangda Wang b a
Department of Radiation Oncology; and b Department of Radiology, The Fourth Hospital of Hebei Medical University, Shijiazhuang, China
a r t i c l e
i n f o
Article history: Received 14 April 2016 Received in revised form 7 October 2016 Accepted 24 October 2016 Available online xxxx Keywords: Esophageal neoplasms Radiotherapy Diffusion-weighted magnetic resonance imaging Apparent diffusion coefficient Therapeutic effect
a b s t r a c t Purpose: To investigate the predictive value of serial diffusion-weighted MRI (DWI) on tumor response of the concurrent chemoradiotherapy (CCRT) for esophageal squamous cell carcinoma (ESCC) and to determine the optimal time point of DWI measurements. Methods and materials: From June 2010 to October 2011, 38 ESCC patients were consecutively enrolled into this prospective cohort study. During their treatment, the longitudinal DWIs were acquired at beginning and every week during the course of CCRT. ADC maps were generated from there DWIs. The tumor responses were evaluated according to the RECIST. Results: (1) At completion of CCRT, 20 (52.6%) and 18 (47.4%) patients were evaluated as CR and PR, respectively. Over the time points of measures, the series of ADC values (10 3 mm2/s) in whole GTV were consistently characterized with higher (all p < 0.05) for these CR patients as their means (std) were 2.24 (0.49), 2.23 (0.51), 2.44 (0.57), 2.54 (0.52), 2.70 (0.46), 2.80 (0.55), 2.92 (0.62), compared with these PR patients as 1.83 (0.31), 1.79 (0.21), 1.87 (0.30), 1.97 (0.37), 2.15 (0.44), 2.26 (0.46), 2.32 (0.51), respectively. However, the ADC change rates (DADC) of two groups were found to be similar. These results were also supported by the multivariate ANOVA analyses. The same analysis results of DWI based GTV volumes were also found. (2) The comparisons of logistic regression analysis indicated that only the ADC values at Week 3 (15 fractions) were an independent prognostic factor of tumor response (OR = 0.303, p = 0.003). ROC curve analysis showed that Area Under Curve for ADC values of the end of 2nd and 3rd weeks were biggest (0.822) and the prediction efficacy was comparatively optimized. The corresponding cut-off values were 2.11 and 2.14 (10 3 mm2/s), respectively. (3) Additional analyses on those eight patients with tumor local recurrence within 1 year demonstrated the level-off after the continuously increased ADC values till Week 4. Conclusions: DWI can be used as a biomarker to predict TE of esophageal cancers in early time during CCRT. The treatment-induced change in ADC of whole GTV during the first 2–3 weeks can be highly predictive to TE. The unchanged ADC value in late period may indicate the high tendency of tumor recurrence after 1 year. Ó 2016 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology xxx (2016) xxx–xxx
Concurrent chemoradiotherapy (CCRT) with or without surgery is currently regarded as the standard treatment for locally advanced esophageal carcinoma (EC). A good evaluation, especially at early time, on the therapeutic effect (TE, i.e. tumor response) of CCRT for EC is essential for the choice and adjustment of subsequent therapeutic managements. However, traditionally image techniques such as CT and esophagogram had limitations in early
⇑ Corresponding author. E-mail addresses: [email protected] (L. Wang), [email protected] (C. Han).
predicting or evaluating TE. Recently, a functional imaging technique, diffusion-weighted magnetic resonance imaging (DWI) (particularly the apparent diffusion coefficient (ADC) obtained from DWI) has been actively studied for its potential in evaluating TE of CCRT for multiple cancers due to its capacity of detecting tumor changes at the molecular or cellular level [1–6]. Among these studies, one prospective study was initiated in our institution in 2010, with the study objective of investigating DWI/ADC as an image biomarker for CCRT of esophageal squamous cell carcinoma (ESCC) and, particularly, searching for optimal DWI acquisition time and ADC threshold.
http://dx.doi.org/10.1016/j.radonc.2016.10.021 0167-8140/Ó 2016 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
2
DWI in esophageal carcinoma
Materials and methods From 06/01/2010 to 10/31/2011, 38 patients who met the eligible criteria of the study (below) were identified. All patients underwent the institutional standard tests and procedures. Inclusion criteria The diagnosis of ESCC was confirmed by tumor tissue pathology. The Karnofsky performance status (KPS) score was P70. No distant metastasis was found under routine medical care. The patient had agreed to have DWI exams before and during the course of CCRT.
scan of b = 600 s/mm2 were close to the real tumor lengths based on surgical specimen with a high concordance with pathology. For each patient, the DWI-defined GTVs at various time points were manually delineated based on the DWIs acquired before, during and after CRT using a software tool (MIM, MIM Software Inc.). These DWIs were used to generate ADC maps. The ADC values reported in this paper are the mean ADC values calculated based the 3D ADC maps of the DWI-defined GTVs. The change rate of ADC (DADC) was calculated by DADC = (ADCx ADC0)/ ADC0 * 100%, where ADC0 was the ADC value from the baseline DWI (i.e. acquired before radiotherapy), x was a time point during or after CRT. For some patients whose DWI hyperintense disappeared at a time point during treatment, the DWI based GTV volume at the time point was recorded as 0.
Clinical data collection At baseline, the patient clinical characteristics including sex, age, tumor site, lesion length and TNM stages were collected for the analysis. Radiotherapy All patients had the following procedures for radiotherapy. First, one CT simulation was performed with scanning range covering the lung, neck and upper abdomen. The CT images were electronically transferred to a treatment planning system (TPS) through local network. Second, based on the CT images and esophagogram, the treating physician outlined the gross tumor volume (GTV), the clinical target volume (CTV), the planning target volume (PTV), and the organs at risk (OAR) (including both lungs, heart, and spinal cord) on the TPS. The target volumes were delineated based on the following guidelines. The GTV were determined by the wall of the esophagus expending P0.5 cm and the mediastinal LNs with a short-axis diameter of P1.0 cm [7,8]. The CTV was defined and contoured by extending 0.5 cm around the GTV in the axial direction and 2.0 cm in the superior and inferior directions. Following that, the PTV was outlined around the CTV with a positive margin of 0.5 cm in the axial direction and 1.0 cm in the superior and inferior directions. Third, the prescription dose for each patient was 60 Gy in 30 daily fractions and the dose received by 95% of the PTV (PTV D95) should be more than 100% of the prescription dose. Last, the OARs had the following dose restrictions: lungs V5 6 55–58%, V20 6 25–30%, V30 6 18–20%, the mean dose of heart 630 Gy, and the maximum dose (Dmax) of spinal cord <45 Gy. After the careful examination and verification on the treatment plan by the supervised radiotherapy physicist and physician, the treating physician approved the treatment plans, which was then delivered with either 3-DCRT or IMRT. Chemotherapy Chemotherapy was given concurrently at the first and fifth weeks of radiotherapy. The regimen of either FP (cisplatin 25 mg/ m2 3 days, 5-fluorouracil (5-FU) 450–500 mg/m2 5 days) or TP (paclitaxel 135 mg/m2, d1, cisplatin 25 mg/m2, d2, 3, 4) was universally given. DWI acquisition All the enrolled patients had seven MRIs taken before the first fraction of radiotherapy and every week during of the course of treatment. The sequences of each MRI acquisition included the conventional T1-weighted (T1W1) and T2-weighted (T2W1), and DWI. The b values (diffusion-sensitive factor) of 0 and 600 s/mm2 were selected according to a previous pathological study in our institution [9], showing that the tumor lengths measured by DWI
Follow-up and evaluation of TE Until 3/15/2015, all patients were followed-up for over 1 year (the longest for 53 months). The TE was evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST 1.1) [10] and the TE of the primary tumor immediately after CCRT was the analysis endpoint. Statistical tool and method One-Sample Kolmogorov–Smirnov method was used to test normal distribution. The independent two-samples T test and multivariate ANOVA statistical analysis for the repeated measurements were used to associate the change of DWI parameters during the course of CCRT between the complete response (CR) and partial response (PR) patients. Logistic Regression Model was performed to find the factors that may affect TE. Furthermore, ROC (receive operating characteristic) curve analysis was used to judge prognosis efficacy and to identify threshold. The survival rates were estimated by the Kaplan–Meier method with log-rank test for the differences between the groups of interest. The SPSS 16.0 statistical software was used. Results Clinical characteristics of the patients The study population included 38 patients (25 male and 13 female), with their ages ranging from 47 to 77 (median 61) years old. The tumor primary distributions by site were cervical (3 patients), upper-thoracic (9 patients), middle-thoracic (23 patients, 60.5%), and lower-thoracic (3 patients). The esophagogram-based tumor lengths before the CCRT ranged from 2.4 to 10.2 (median 5.7) cm. Among all the patients, 6, 10 and 22 patients were determined to be at the T2, T3 and T4 stages, respectively. A total of 35 patients were found to have local LN metastasis based on CT or MRI and were assigned to be at the N1–2 stages, with 1, 14 and 23 patients at TNM I, II and III stage, respectively. Description of therapeutic response All the enrolled patients had completed CCRT. Per the RECIST [10], there were 20 (52.6%) and 18 (47.4%) were determined as CR and PR, respectively. The overall response rate to CCRT was 100%, while the 1, 2, 3-year survival rates of CR and PR groups were 80.0%, 65.0%, 50.0% and 50.0%, 27.8%, 22.2%, respectively (v2 = 5.126, p = 0.024) (Fig. 2b). The mean ± standard deviation of ADC values (10 3 mm2/s) and GTV volumes (cm3) based on the 3D map at the 7 time points were 2.04 ± 0.46, 2.02 ± 0.45, 2.17 ± 0.54, 2.27 ± 0.53, 2.44 ± 0.53, 2.55 ± 0.57, 2.64 ± 0.64 and
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
L. Wang et al. / Radiotherapy and Oncology xxx (2016) xxx–xxx
3
Fig. 1. Changing curves of ADC values and DWI based GTV volumes (Fig. 1a shows ADC values of CR group and PR group at different time points, Fig. 1b shows DWI based GTV volumes of CR and PR groups, Fig. 1c shows ADC values of local recurrence group and local control group at different time points, Fig. 1d shows DWI based GTV volumes of control and recurrence groups).
38.17 ± 2.54, 31.53 ± 19.02, 19.81 ± 10.88, 11.24 ± 10.16, 8.77 ± 7.82, 6.25 ± 6.08, 4.93 ± 6.86, respectively. This shows that the ADC values increased gradually along with the increased radiation doses (i.e. time points). Meanwhile, the DWI based GTV volumes decreased gradually, while the GTV for 12 patients became zero GTV (the hyperintense disappeared completely on their DWIs), indicating their lesions could not be measured anymore from DWI. Pearson and Spearman correlation analysis manifested that the mean ADC values had inverse correlation relationships with T stages and GTV volumes (r = 0.344, 0.478; p = 0.034, 0.002) and the GTV volumes of CR group were generally lower as compared to the PR group (45.0 cm3 vs 53.1 cm3). Comparative analysis of DWI parameter changes between CR and PR patients Table 1 shows the comparisons of ADC and DADC values between the CR and PR patients. At each time point of DWI measurement, the CR patients had a significant higher ADC value than
the PR patients (p < 0.05). However, none of DADC comparisons showed the significant difference (e.g., DADC at the 2nd week was 8.93 ± 11.08% and 2.72 ± 10.11%, respectively, p = 0.08). To further define the difference in the change of ADC over time points between the CR and PR patient groups, the repeated measures define fact analysis and multivariate statistical analysis were performed. Fig. 1a and Fig. 1b show mean ADC values in the whole GTV (CT-defined) and DWI based GTV volumes. It is seen from Fig. 1a that the ADC values at the 1st weekend were similar to those before CCRT, and from the time point of 2nd weekend, mean ADC values of GTVs increased gradually. Particulary, the curve of the CR group increases sharply at the 2nd weekend. Indeed, the test of between-subjects effects demonstrated a significant difference (F = 11.028, p = 0.002 < 0.05). In multivariate test, the results of 4 statistic parameters were concordant and the p values of each time point were less than 0.001, which indicated a significant difference in ADC values and DWI based GTV volumes at different time points of radiation course. Table 2 shows a significant difference between CR and PR groups at each time point.
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
4
DWI in esophageal carcinoma
Fig. 2. (a) Overall survival rates of patients according to the ADC. Kaplan–Meier survival rate and p values are calculated according to a log-rank test. Survival probabilities are displayed as a red line (ADC < 2.14 10 3 mm2/s) and a green line (ADC P 2.14 10 3 mm2/s) (p = 0.061). Median survival times were 13 and 31 months, respectively. (b) Overall survival rates according to TE. Kaplan–Meier survival rates and p values are calculated according to a log-rank test. Survival probabilities are displayed as a green line (CR group) and a red line (PR group) (p = 0.024). Median survival times were 31 and 12 months, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Table 1 DWI parameters of CR group and PR group. Group
Measure value
ADC value (10
3
mm2/s)
DADC (%)
Before CRT
1st weekend
2nd weekend
3rd weekend
4th weekend
5th weekend
6th weekend
CR group PR group t p
2.24 ± 0.49 1.83 ± 0.31 3.027 0.005
2.23 ± 0.51 1.79 ± 0.21 3.400 0.002
2.44 ± 0.57 1.87 ± 0.30 3.754 0.001
2.54 ± 0.52 1.97 ± 0.37 3.891 0.000
2.70 ± 0.46 2.15 ± 0.44 3.754 0.001
2.80 ± 0.55 2.26 ± 0.46 3.215 0.003
2.92 ± 0.62 2.32 ± 0.51 3.236 0.003
CR group PR group t p
– – – –
0.19 ± 8.58 0.78 ± 14.14 0.158 0.875
8.93 ± 11.08 2.72 ± 10.11 1.799 0.080
14.71 ± 13.67 8.20 ± 13.41 1.477 0.148
23.00 ± 17.65 18.73 ± 21.06 0.680 0.501
27.71 ± 25.49 24.40 ± 19.42 0.447 0.658
32.71 ± 26.36 27.67 ± 24.84 0.604 0.549
The bold values indicated the change of ADC values and there was a significant difference between CR and PR group.
Table 2 Comparison of time points (multivariate analysis of variance). Source
Group
Dependent variable
Before CRT End of 1st week End of 2nd week End of 3rd week End of 4th week End of 5th week End of 6th week
ADC values
DWI based GTV volume
Type IV sum of squares
df
Mean square
F
p
Type IV sum of squares
df
Mean square
F
p
1.569 1.843 3.072 3.122 2.879 2.694 3.390
1 1 1 1 1 1 1
1.569 1.843 3.072 3.122 2.879 2.694 3.390
9.163 11.559 14.094 15.142 14.092 10.335 10.469
0.005 0.002 0.001 0.000 0.001 0.003 0.003
3149.53 1643.24 560.07 576.16 406.10 225.27 181.22
1 1 1 1 1 1 1
3149.53 1643.24 560.07 576.16 406.10 225.27 181.22
5.518 5.109 5.423 6.39 7.92 7.14 4.12
0.025 0.030 0.026 0.016 0.008 0.011 0.050
p was the test result of Multivariate analysis of variance, the bold values indicated that there was a significant difference between CR and PR group at each time point.
In this study, 8 patients were considered as local relapse(regional metastases was not included) within 1 year according to examination of esophagogram and CT scan, which were finally confirmed by pathology. We classified the eight patients as the local recurrence group, and the rests (i.e. 30 patients) as the local control group. The ADC changing curves of the two groups ascended from the beginning to the 4th week, but after the monitoring point of 5th week, the curve of recurrence group dropped to below the local control group (Fig. 1c), and at the same time, the
reduced ranges of DWI based GTV volumes were significantly decreased at the time of 3rd to 5th week (Fig. 1d). Logistic regression analysis of DWI parameters on TE Multivariable logistic regression analysis (Stepwise with entry p = 0.05 and removal p = 0.10) on ‘CR’ as the endpoint was performed in order to determine the best time-point of DWI series associating with ‘CR’ the most. The variable pool included sex,
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
5
L. Wang et al. / Radiotherapy and Oncology xxx (2016) xxx–xxx
age, tumor site, T stage, N stage, GTV volume, the ADC values of before CCRT and at the end of 1st, 2nd, 3rd, 4th, 5th weeks of treatment course. Because of the small sample size, we classified the continuous variables into 4 groups based on their percentile ranks of 25%, 50% and 75%. At end, the ADC values of the end of 3rd week remained in the model only, the corresponding values of B, S.E., Wald and OR were 1.195, 0.405, 8.721 and 0.303, respectively, and p = 0.003. The results of logistic regression analysis indicated that ADC value of the end of 3rd week was an independent prognosis factor of TE. The large ADC values associate with good chance of therapeutic response being ‘CR’ (OR = 0.303, 95%CI 0.137–0.669). ROC analysis To determine the cut point of ADC value over Week 1 to Week 5 used for ‘CR’ vs. ‘PR’, one ROC analysis was conducted. The results are shown in Table 3. The analysis indicated that, the Area Under Curve for ADC values of the end of 2nd and 3rd weeks were biggest, the values were 0.822 and the prediction efficacy was comparatively optimized. Based on the cut point choice of 2.11 and 2.14 (10 3 mm2/s), the Sensitivity and Specificity were 70.0%, 88.9%, and 85.0, 77.8%, respectively. Patients were divided into two groups according to the ADC threshold of the end of 2nd week, there were 22 patients whose ADC values <2.11 10 3 mm2/s, only 6 of them achieved the therapeutic response of CR (27.23%), on the other side, for the group whose ADC values P2.11 10 3 mm2/s (16 cases) at the end of 2nd week, 14 achieved CR (87.50%), the latter was significantly higher than the former, with v2 = 13.478, p = 0.000. At the end of 3rd week, there were 17 patients whose ADC values <2.14 10 3 mm2/s, only 3 of them achieved CR (17.65%), there were 21 patients whose ADC values P2.14 10 3 mm2/s, the CR rate was 80.95% (17/21), v2 = 15.102, p = 0.000. The 1, 2, 3-year survival rates of group P2.14 10 3 mm2/s and group <2.14 10 3 mm2/s were 76.2%, 61.9%, 47.6% and 52.9%, 29.4%, 23.5%, respectively, with v2 = 3.510, p = 0.061 (Fig. 2a) . Discussion Concurrent chemoradiotherapy with or without surgery is currently regarded as the standard treatment for locally advanced esophageal carcinoma. One good evaluation on tumor response of EC to CCRT especially at early time becomes essential for the choice and adjustment of subsequent therapeutic managements. Measurement of tumor volume was generally used to evaluate therapeutic response in clinic currently. However esophagus is a cavity organ, the esophageal tumor volume may not be measured precisely by conventional images. Furthermore, the accuracy of using tumor volume decrease or increase to predict TE at the early time of CCRT still needs further confirmation. DWI is a functional image modality, as it reflects the diffusion movement of hydrone in tissues, and may be used to measure the changes in both the shape and function of tumors. The use of DWI to monitor TE in the treatments of several cancer types including breast cancer and rectal cancer has been reported. However, the correlated studies carried out in the field of esophageal carcinoma are sparse.
There are a few studies which were aimed to use DWI to guide target delineation [11] or evaluating prognosis [12,13], and to investigate the feasibility of using DWI parameters to predict early response to CCRT for esophageal cancers [14,15]. So In this study, 38 patients with the diagnosis of ESCC were prospectively observed for 2 DWI-derived measurements: ADC value and DWI based GTV volume during the entire course of CCRT. The objective was to investigate the diagnostic efficiency of DWI parameters for early prediction of the TE of esophageal carcinoma treatment with chemoradiotherapy, and to find the optimal predicting time point and predicting threshold. According to multi-time points of DWI examination during the course of CCRT and analysis of the change of ADC values, DADC, DWI based GTV volumes, we found that there was a significant difference in ADC values between the CR group and PR group from the beginning to the end of treatment course, especially, the pre-CCRT ADC values of the CR group were significantly higher than those of PR group, and the difference continued to the end of course. However, there was no significant difference in DADC between the two groups. We could only observe from Fig. 1a that the curve of CR group stepped-up even more after the time point of 2nd weekend, with no significant difference observed between the two groups. The predictive value of ADC for TE and survival in esophageal cancer has not been well established. Two studies [13,16] from Japan indicated that ADC value of pre-treatment may be a useful marker to predict TE and high ADC values indicated a well TE or survive. This viewpoint was in accordant with ours, while in the study of van Rossum [15], there was no significant difference in initial ADC values between path CR group and no path CR group. Interestingly, in the study of De Cobelli [17], the authors observed that responders showed lower pre-NT ADC (1.32 vs 1.63 10 3 mm2/s; p = 0.002) than non-responders. We think this may be due to two reasons: (1) The presence of necrotic (poorly perfuse and hypoxic, but with high ADC value) areas, which are less sensitive to CRT, explaining the higher ADC values of poor responders [13]. (2) Generally speaking, ADC is a measure of tumor cell density. A higher ADC value represents less tumor cells, thus has higher probability for CR, as compared to a low ADC value. In this study, the ADC values have inverse correlation relationships with T stages and GTV volumes (r = 0.344, 0.478; p = 0.034, 0.002) and the GTV volumes of CR group are smaller (45.0 cm3 vs 53.1 cm3), we presumed that the larger the GTV volumes, the more and compacted the tumor cells, explaining the lower ADC values of poor TE. Because we found that there were significant differences of ADC values between different time points and different groups according to the description and statistics of DWI parameters, so a Logistic regression analysis was conducted with ADC values of 1st to 5th weekends and a series of correlative clinical factors. The objective of the analysis was to find significant parameters which had effects on TE of esophageal cancers. The results of multivariate analysis showed that only ADC values of the 3rd weekend were the independent prognostic factor, the OR value was 0.303, which indicated the third-weekend ADC values were a comparatively strong protection parameter of TE, the higher of ADC values, the superior of TE. The following ROC curve analysis found that the prognosis
Table 3 The results of ROC curve analysis. Time point
Area under curve
p
ADC threshold (10
1st weekend 2nd weekend 3rd weekend 4th weekend 5th weekend
0.786 0.822 0.822 0.810 0.769
0.003 0.001 0.001 0.001 0.005
1.87 2.11 2.14 2.34 2.50
3
mm2/s)
Sensitivity (%)
Specificity (%)
75.00 70.00 85.00 85.00 70.00
72.22 88.89 77.78 72.22 72.22
Bold values indicated that the Area under curve of 2nd and 3rd weekend was biggest, and the prediction efficacy of the two time points may be optimal.
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
6
DWI in esophageal carcinoma
efficiency of second and third-weekend ADC values was optimized (0.822), the result indicated that two time points were effective monitoring time points of TE. So we suggested that DWI scan could be performed at the end of 2nd or 3rd week and the measured ADC values could be used in determining the subsequent. In recent years, studies [14,15] from Japan and Netherlands also considered that treatment-induced ADC or DADC changing during the first 2– 3 weeks of treatment for esophageal cancer seemed to be highly predictive of therapeutic response. An additional point to make is there were 8 patients occurred in 1 year, the ADC values of the group ascended from the beginning to the 4th week, but subsequently, although the radiation doses still increase, the ADC values ‘‘dropped” steeply, at the same time, the curve of DWI based GTV volumes presented a ‘‘shoulder area” at the time of 3rd to 5th weeks, the reduced ranges of DWI based GTV volumes were significantly decreased. We think this phenomenon may indicate a high tendency of tumor recurrence. According to all the above studies, we conclude that DWI can be a useful biomarker to predict TE of esophageal cancers in early time, the treatment-induced change in ADC in GTV during the first 2–3 weeks seemed highly predictive to TE, the phenomenon of ADC value dropping in late treatment course or DWI based GTV volumes stopping shrinking maybe indicators for possible tumor recurrence. Conflict of interest statement None of the authors have financial or personal conflicts of interest to disclose. References [1] Hein PA, Kremser C, Judmaier W, et al. Diffusion-weighted magnetic resonance imaging for monitoring diffusion changes in rectal carcinoma during combined, preoperative chemoradiation: preliminary results of a prospective study. Eur J Radiol 2003;45:214–22. [2] Kremser C, Judmaier W, Hein P, et al. Preliminary results on the influence of chemoradiation on apparent diffusion coefficients of primary rectal carcinoma measured by magnetic resonance imaging. Strahlenther Onkol 2003;179: 641–9.
[3] Galbán CJ, Mukherji SK, Chenevert TL, et al. A feasibility study of parametric response map analysis of diffusion-weighted magnetic resonance imaging scans of head and neck cancer patients for providing early detection of therapeutic efficacy. Transl Oncol 2009;2:184–90. [4] Dzik-Jurasz A, Domenig C, George M, et al. Diffusion MRI for prediction of response of rectal cancer to chemoradiation. Lancet 2002;360:301–8. [5] Hayashida Y, Yakushiji T, Awai K, et al. Monitoring therapeutic responses of primary bone tumors by diffusion-weighted image: initial results. Eur Radiol 2006;16:2637–43. [6] Mardor Y, Pfeffer R, Spiegelmann R, et al. Early detection of response to radiation therapy in patients with brain malignancies using conventional and high b-value diffusion-weighted magnetic resonance imaging. J Clin Oncol 2003;21:1094–100. [7] van Overhagen H, Laméris JS, Berger MY, et al. Supraclavicular lymph node metastases in carcinoma of the esophageal and gastroesophageal junction: assessment with CT, US and US-guided fine-needle aspiration biopsy. Radiology 1991;179:155–8. [8] Takashi M, Yasumasa N, Yutaka S, et al. Optimal size criteria of malignant lymph nodes in the treatment planning of radiotherapy for esophageal cancer: evaluation by computed tomography and magnetic resonance imaging. Radiother Oncol 1996;36:1091–8. [9] Lan Wang, Chun Han, Shuchai Zhu, et al. A pathological comparative study of diffusion-weighted imaging and computed tomography in determination of lesion length for esophageal carcinoma. Chin J Radiat Oncol 2015;24:373–6. [10] Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228–47. [11] Hou DL, Shi GF, Gao XS, et al. Improved longitudinal length accuracy of gross tumor volume delineation with diffusion weighted magnetic resonance imaging for esophageal squamous cell carcinoma. Radiat Oncol 2013;8:169. [12] Wang L, Han C, Zhu S, et al. Investigation of using diffusion-weighted magnetic resonance imaging to evaluate the therapeutic effect of esophageal carcinoma treatment. Oncol Res Treat 2014;3:112–6. [13] Giganti F, Salerno A, Ambrosi A, et al. Prognostic utility of diffusion-weighted MRI in oesophageal cancer: is apparent diffusion coefficient a potential marker of tumour aggressiveness? Padiol Med 2016;121:173–80. [14] Imanishi S, Shuto K, Aoyagi T, et al. 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. [15] van Rossum PS, van Lier AL, van Vulpen M, et al. Diffusion-weighted magnetic resonance imaging for the prediction of pathologic response to neoadjuvant chemoradiotherapy in esophageal cancer. Radiother Oncol 2015;115:163–70. [16] Aoyagi T, Shuto K, Okazumi S, et al. Apparent diffusion coefficient values measured by diffusion-weighted imaging predict chemoradiotherapeutic effect for advanced esophageal cancer. Dig Surg 2011;28:252–7. [17] 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.
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
Original article
The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy Lan Wang a, Lihong Liu a, Chun Han a,⇑, Shutang Liu a, Hua Tian a, Zhensheng Li a, Xuejiao Ren a, Gaofeng Shi b, Qi Wang b, Guangda Wang b a
Department of Radiation Oncology; and b Department of Radiology, The Fourth Hospital of Hebei Medical University, Shijiazhuang, China
a r t i c l e
i n f o
Article history: Received 14 April 2016 Received in revised form 7 October 2016 Accepted 24 October 2016 Available online xxxx Keywords: Esophageal neoplasms Radiotherapy Diffusion-weighted magnetic resonance imaging Apparent diffusion coefficient Therapeutic effect
a b s t r a c t Purpose: To investigate the predictive value of serial diffusion-weighted MRI (DWI) on tumor response of the concurrent chemoradiotherapy (CCRT) for esophageal squamous cell carcinoma (ESCC) and to determine the optimal time point of DWI measurements. Methods and materials: From June 2010 to October 2011, 38 ESCC patients were consecutively enrolled into this prospective cohort study. During their treatment, the longitudinal DWIs were acquired at beginning and every week during the course of CCRT. ADC maps were generated from there DWIs. The tumor responses were evaluated according to the RECIST. Results: (1) At completion of CCRT, 20 (52.6%) and 18 (47.4%) patients were evaluated as CR and PR, respectively. Over the time points of measures, the series of ADC values (10 3 mm2/s) in whole GTV were consistently characterized with higher (all p < 0.05) for these CR patients as their means (std) were 2.24 (0.49), 2.23 (0.51), 2.44 (0.57), 2.54 (0.52), 2.70 (0.46), 2.80 (0.55), 2.92 (0.62), compared with these PR patients as 1.83 (0.31), 1.79 (0.21), 1.87 (0.30), 1.97 (0.37), 2.15 (0.44), 2.26 (0.46), 2.32 (0.51), respectively. However, the ADC change rates (DADC) of two groups were found to be similar. These results were also supported by the multivariate ANOVA analyses. The same analysis results of DWI based GTV volumes were also found. (2) The comparisons of logistic regression analysis indicated that only the ADC values at Week 3 (15 fractions) were an independent prognostic factor of tumor response (OR = 0.303, p = 0.003). ROC curve analysis showed that Area Under Curve for ADC values of the end of 2nd and 3rd weeks were biggest (0.822) and the prediction efficacy was comparatively optimized. The corresponding cut-off values were 2.11 and 2.14 (10 3 mm2/s), respectively. (3) Additional analyses on those eight patients with tumor local recurrence within 1 year demonstrated the level-off after the continuously increased ADC values till Week 4. Conclusions: DWI can be used as a biomarker to predict TE of esophageal cancers in early time during CCRT. The treatment-induced change in ADC of whole GTV during the first 2–3 weeks can be highly predictive to TE. The unchanged ADC value in late period may indicate the high tendency of tumor recurrence after 1 year. Ó 2016 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology xxx (2016) xxx–xxx
Concurrent chemoradiotherapy (CCRT) with or without surgery is currently regarded as the standard treatment for locally advanced esophageal carcinoma (EC). A good evaluation, especially at early time, on the therapeutic effect (TE, i.e. tumor response) of CCRT for EC is essential for the choice and adjustment of subsequent therapeutic managements. However, traditionally image techniques such as CT and esophagogram had limitations in early
⇑ Corresponding author. E-mail addresses: [email protected] (L. Wang), [email protected] (C. Han).
predicting or evaluating TE. Recently, a functional imaging technique, diffusion-weighted magnetic resonance imaging (DWI) (particularly the apparent diffusion coefficient (ADC) obtained from DWI) has been actively studied for its potential in evaluating TE of CCRT for multiple cancers due to its capacity of detecting tumor changes at the molecular or cellular level [1–6]. Among these studies, one prospective study was initiated in our institution in 2010, with the study objective of investigating DWI/ADC as an image biomarker for CCRT of esophageal squamous cell carcinoma (ESCC) and, particularly, searching for optimal DWI acquisition time and ADC threshold.
http://dx.doi.org/10.1016/j.radonc.2016.10.021 0167-8140/Ó 2016 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
2
DWI in esophageal carcinoma
Materials and methods From 06/01/2010 to 10/31/2011, 38 patients who met the eligible criteria of the study (below) were identified. All patients underwent the institutional standard tests and procedures. Inclusion criteria The diagnosis of ESCC was confirmed by tumor tissue pathology. The Karnofsky performance status (KPS) score was P70. No distant metastasis was found under routine medical care. The patient had agreed to have DWI exams before and during the course of CCRT.
scan of b = 600 s/mm2 were close to the real tumor lengths based on surgical specimen with a high concordance with pathology. For each patient, the DWI-defined GTVs at various time points were manually delineated based on the DWIs acquired before, during and after CRT using a software tool (MIM, MIM Software Inc.). These DWIs were used to generate ADC maps. The ADC values reported in this paper are the mean ADC values calculated based the 3D ADC maps of the DWI-defined GTVs. The change rate of ADC (DADC) was calculated by DADC = (ADCx ADC0)/ ADC0 * 100%, where ADC0 was the ADC value from the baseline DWI (i.e. acquired before radiotherapy), x was a time point during or after CRT. For some patients whose DWI hyperintense disappeared at a time point during treatment, the DWI based GTV volume at the time point was recorded as 0.
Clinical data collection At baseline, the patient clinical characteristics including sex, age, tumor site, lesion length and TNM stages were collected for the analysis. Radiotherapy All patients had the following procedures for radiotherapy. First, one CT simulation was performed with scanning range covering the lung, neck and upper abdomen. The CT images were electronically transferred to a treatment planning system (TPS) through local network. Second, based on the CT images and esophagogram, the treating physician outlined the gross tumor volume (GTV), the clinical target volume (CTV), the planning target volume (PTV), and the organs at risk (OAR) (including both lungs, heart, and spinal cord) on the TPS. The target volumes were delineated based on the following guidelines. The GTV were determined by the wall of the esophagus expending P0.5 cm and the mediastinal LNs with a short-axis diameter of P1.0 cm [7,8]. The CTV was defined and contoured by extending 0.5 cm around the GTV in the axial direction and 2.0 cm in the superior and inferior directions. Following that, the PTV was outlined around the CTV with a positive margin of 0.5 cm in the axial direction and 1.0 cm in the superior and inferior directions. Third, the prescription dose for each patient was 60 Gy in 30 daily fractions and the dose received by 95% of the PTV (PTV D95) should be more than 100% of the prescription dose. Last, the OARs had the following dose restrictions: lungs V5 6 55–58%, V20 6 25–30%, V30 6 18–20%, the mean dose of heart 630 Gy, and the maximum dose (Dmax) of spinal cord <45 Gy. After the careful examination and verification on the treatment plan by the supervised radiotherapy physicist and physician, the treating physician approved the treatment plans, which was then delivered with either 3-DCRT or IMRT. Chemotherapy Chemotherapy was given concurrently at the first and fifth weeks of radiotherapy. The regimen of either FP (cisplatin 25 mg/ m2 3 days, 5-fluorouracil (5-FU) 450–500 mg/m2 5 days) or TP (paclitaxel 135 mg/m2, d1, cisplatin 25 mg/m2, d2, 3, 4) was universally given. DWI acquisition All the enrolled patients had seven MRIs taken before the first fraction of radiotherapy and every week during of the course of treatment. The sequences of each MRI acquisition included the conventional T1-weighted (T1W1) and T2-weighted (T2W1), and DWI. The b values (diffusion-sensitive factor) of 0 and 600 s/mm2 were selected according to a previous pathological study in our institution [9], showing that the tumor lengths measured by DWI
Follow-up and evaluation of TE Until 3/15/2015, all patients were followed-up for over 1 year (the longest for 53 months). The TE was evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST 1.1) [10] and the TE of the primary tumor immediately after CCRT was the analysis endpoint. Statistical tool and method One-Sample Kolmogorov–Smirnov method was used to test normal distribution. The independent two-samples T test and multivariate ANOVA statistical analysis for the repeated measurements were used to associate the change of DWI parameters during the course of CCRT between the complete response (CR) and partial response (PR) patients. Logistic Regression Model was performed to find the factors that may affect TE. Furthermore, ROC (receive operating characteristic) curve analysis was used to judge prognosis efficacy and to identify threshold. The survival rates were estimated by the Kaplan–Meier method with log-rank test for the differences between the groups of interest. The SPSS 16.0 statistical software was used. Results Clinical characteristics of the patients The study population included 38 patients (25 male and 13 female), with their ages ranging from 47 to 77 (median 61) years old. The tumor primary distributions by site were cervical (3 patients), upper-thoracic (9 patients), middle-thoracic (23 patients, 60.5%), and lower-thoracic (3 patients). The esophagogram-based tumor lengths before the CCRT ranged from 2.4 to 10.2 (median 5.7) cm. Among all the patients, 6, 10 and 22 patients were determined to be at the T2, T3 and T4 stages, respectively. A total of 35 patients were found to have local LN metastasis based on CT or MRI and were assigned to be at the N1–2 stages, with 1, 14 and 23 patients at TNM I, II and III stage, respectively. Description of therapeutic response All the enrolled patients had completed CCRT. Per the RECIST [10], there were 20 (52.6%) and 18 (47.4%) were determined as CR and PR, respectively. The overall response rate to CCRT was 100%, while the 1, 2, 3-year survival rates of CR and PR groups were 80.0%, 65.0%, 50.0% and 50.0%, 27.8%, 22.2%, respectively (v2 = 5.126, p = 0.024) (Fig. 2b). The mean ± standard deviation of ADC values (10 3 mm2/s) and GTV volumes (cm3) based on the 3D map at the 7 time points were 2.04 ± 0.46, 2.02 ± 0.45, 2.17 ± 0.54, 2.27 ± 0.53, 2.44 ± 0.53, 2.55 ± 0.57, 2.64 ± 0.64 and
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
L. Wang et al. / Radiotherapy and Oncology xxx (2016) xxx–xxx
3
Fig. 1. Changing curves of ADC values and DWI based GTV volumes (Fig. 1a shows ADC values of CR group and PR group at different time points, Fig. 1b shows DWI based GTV volumes of CR and PR groups, Fig. 1c shows ADC values of local recurrence group and local control group at different time points, Fig. 1d shows DWI based GTV volumes of control and recurrence groups).
38.17 ± 2.54, 31.53 ± 19.02, 19.81 ± 10.88, 11.24 ± 10.16, 8.77 ± 7.82, 6.25 ± 6.08, 4.93 ± 6.86, respectively. This shows that the ADC values increased gradually along with the increased radiation doses (i.e. time points). Meanwhile, the DWI based GTV volumes decreased gradually, while the GTV for 12 patients became zero GTV (the hyperintense disappeared completely on their DWIs), indicating their lesions could not be measured anymore from DWI. Pearson and Spearman correlation analysis manifested that the mean ADC values had inverse correlation relationships with T stages and GTV volumes (r = 0.344, 0.478; p = 0.034, 0.002) and the GTV volumes of CR group were generally lower as compared to the PR group (45.0 cm3 vs 53.1 cm3). Comparative analysis of DWI parameter changes between CR and PR patients Table 1 shows the comparisons of ADC and DADC values between the CR and PR patients. At each time point of DWI measurement, the CR patients had a significant higher ADC value than
the PR patients (p < 0.05). However, none of DADC comparisons showed the significant difference (e.g., DADC at the 2nd week was 8.93 ± 11.08% and 2.72 ± 10.11%, respectively, p = 0.08). To further define the difference in the change of ADC over time points between the CR and PR patient groups, the repeated measures define fact analysis and multivariate statistical analysis were performed. Fig. 1a and Fig. 1b show mean ADC values in the whole GTV (CT-defined) and DWI based GTV volumes. It is seen from Fig. 1a that the ADC values at the 1st weekend were similar to those before CCRT, and from the time point of 2nd weekend, mean ADC values of GTVs increased gradually. Particulary, the curve of the CR group increases sharply at the 2nd weekend. Indeed, the test of between-subjects effects demonstrated a significant difference (F = 11.028, p = 0.002 < 0.05). In multivariate test, the results of 4 statistic parameters were concordant and the p values of each time point were less than 0.001, which indicated a significant difference in ADC values and DWI based GTV volumes at different time points of radiation course. Table 2 shows a significant difference between CR and PR groups at each time point.
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
4
DWI in esophageal carcinoma
Fig. 2. (a) Overall survival rates of patients according to the ADC. Kaplan–Meier survival rate and p values are calculated according to a log-rank test. Survival probabilities are displayed as a red line (ADC < 2.14 10 3 mm2/s) and a green line (ADC P 2.14 10 3 mm2/s) (p = 0.061). Median survival times were 13 and 31 months, respectively. (b) Overall survival rates according to TE. Kaplan–Meier survival rates and p values are calculated according to a log-rank test. Survival probabilities are displayed as a green line (CR group) and a red line (PR group) (p = 0.024). Median survival times were 31 and 12 months, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Table 1 DWI parameters of CR group and PR group. Group
Measure value
ADC value (10
3
mm2/s)
DADC (%)
Before CRT
1st weekend
2nd weekend
3rd weekend
4th weekend
5th weekend
6th weekend
CR group PR group t p
2.24 ± 0.49 1.83 ± 0.31 3.027 0.005
2.23 ± 0.51 1.79 ± 0.21 3.400 0.002
2.44 ± 0.57 1.87 ± 0.30 3.754 0.001
2.54 ± 0.52 1.97 ± 0.37 3.891 0.000
2.70 ± 0.46 2.15 ± 0.44 3.754 0.001
2.80 ± 0.55 2.26 ± 0.46 3.215 0.003
2.92 ± 0.62 2.32 ± 0.51 3.236 0.003
CR group PR group t p
– – – –
0.19 ± 8.58 0.78 ± 14.14 0.158 0.875
8.93 ± 11.08 2.72 ± 10.11 1.799 0.080
14.71 ± 13.67 8.20 ± 13.41 1.477 0.148
23.00 ± 17.65 18.73 ± 21.06 0.680 0.501
27.71 ± 25.49 24.40 ± 19.42 0.447 0.658
32.71 ± 26.36 27.67 ± 24.84 0.604 0.549
The bold values indicated the change of ADC values and there was a significant difference between CR and PR group.
Table 2 Comparison of time points (multivariate analysis of variance). Source
Group
Dependent variable
Before CRT End of 1st week End of 2nd week End of 3rd week End of 4th week End of 5th week End of 6th week
ADC values
DWI based GTV volume
Type IV sum of squares
df
Mean square
F
p
Type IV sum of squares
df
Mean square
F
p
1.569 1.843 3.072 3.122 2.879 2.694 3.390
1 1 1 1 1 1 1
1.569 1.843 3.072 3.122 2.879 2.694 3.390
9.163 11.559 14.094 15.142 14.092 10.335 10.469
0.005 0.002 0.001 0.000 0.001 0.003 0.003
3149.53 1643.24 560.07 576.16 406.10 225.27 181.22
1 1 1 1 1 1 1
3149.53 1643.24 560.07 576.16 406.10 225.27 181.22
5.518 5.109 5.423 6.39 7.92 7.14 4.12
0.025 0.030 0.026 0.016 0.008 0.011 0.050
p was the test result of Multivariate analysis of variance, the bold values indicated that there was a significant difference between CR and PR group at each time point.
In this study, 8 patients were considered as local relapse(regional metastases was not included) within 1 year according to examination of esophagogram and CT scan, which were finally confirmed by pathology. We classified the eight patients as the local recurrence group, and the rests (i.e. 30 patients) as the local control group. The ADC changing curves of the two groups ascended from the beginning to the 4th week, but after the monitoring point of 5th week, the curve of recurrence group dropped to below the local control group (Fig. 1c), and at the same time, the
reduced ranges of DWI based GTV volumes were significantly decreased at the time of 3rd to 5th week (Fig. 1d). Logistic regression analysis of DWI parameters on TE Multivariable logistic regression analysis (Stepwise with entry p = 0.05 and removal p = 0.10) on ‘CR’ as the endpoint was performed in order to determine the best time-point of DWI series associating with ‘CR’ the most. The variable pool included sex,
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
5
L. Wang et al. / Radiotherapy and Oncology xxx (2016) xxx–xxx
age, tumor site, T stage, N stage, GTV volume, the ADC values of before CCRT and at the end of 1st, 2nd, 3rd, 4th, 5th weeks of treatment course. Because of the small sample size, we classified the continuous variables into 4 groups based on their percentile ranks of 25%, 50% and 75%. At end, the ADC values of the end of 3rd week remained in the model only, the corresponding values of B, S.E., Wald and OR were 1.195, 0.405, 8.721 and 0.303, respectively, and p = 0.003. The results of logistic regression analysis indicated that ADC value of the end of 3rd week was an independent prognosis factor of TE. The large ADC values associate with good chance of therapeutic response being ‘CR’ (OR = 0.303, 95%CI 0.137–0.669). ROC analysis To determine the cut point of ADC value over Week 1 to Week 5 used for ‘CR’ vs. ‘PR’, one ROC analysis was conducted. The results are shown in Table 3. The analysis indicated that, the Area Under Curve for ADC values of the end of 2nd and 3rd weeks were biggest, the values were 0.822 and the prediction efficacy was comparatively optimized. Based on the cut point choice of 2.11 and 2.14 (10 3 mm2/s), the Sensitivity and Specificity were 70.0%, 88.9%, and 85.0, 77.8%, respectively. Patients were divided into two groups according to the ADC threshold of the end of 2nd week, there were 22 patients whose ADC values <2.11 10 3 mm2/s, only 6 of them achieved the therapeutic response of CR (27.23%), on the other side, for the group whose ADC values P2.11 10 3 mm2/s (16 cases) at the end of 2nd week, 14 achieved CR (87.50%), the latter was significantly higher than the former, with v2 = 13.478, p = 0.000. At the end of 3rd week, there were 17 patients whose ADC values <2.14 10 3 mm2/s, only 3 of them achieved CR (17.65%), there were 21 patients whose ADC values P2.14 10 3 mm2/s, the CR rate was 80.95% (17/21), v2 = 15.102, p = 0.000. The 1, 2, 3-year survival rates of group P2.14 10 3 mm2/s and group <2.14 10 3 mm2/s were 76.2%, 61.9%, 47.6% and 52.9%, 29.4%, 23.5%, respectively, with v2 = 3.510, p = 0.061 (Fig. 2a) . Discussion Concurrent chemoradiotherapy with or without surgery is currently regarded as the standard treatment for locally advanced esophageal carcinoma. One good evaluation on tumor response of EC to CCRT especially at early time becomes essential for the choice and adjustment of subsequent therapeutic managements. Measurement of tumor volume was generally used to evaluate therapeutic response in clinic currently. However esophagus is a cavity organ, the esophageal tumor volume may not be measured precisely by conventional images. Furthermore, the accuracy of using tumor volume decrease or increase to predict TE at the early time of CCRT still needs further confirmation. DWI is a functional image modality, as it reflects the diffusion movement of hydrone in tissues, and may be used to measure the changes in both the shape and function of tumors. The use of DWI to monitor TE in the treatments of several cancer types including breast cancer and rectal cancer has been reported. However, the correlated studies carried out in the field of esophageal carcinoma are sparse.
There are a few studies which were aimed to use DWI to guide target delineation [11] or evaluating prognosis [12,13], and to investigate the feasibility of using DWI parameters to predict early response to CCRT for esophageal cancers [14,15]. So In this study, 38 patients with the diagnosis of ESCC were prospectively observed for 2 DWI-derived measurements: ADC value and DWI based GTV volume during the entire course of CCRT. The objective was to investigate the diagnostic efficiency of DWI parameters for early prediction of the TE of esophageal carcinoma treatment with chemoradiotherapy, and to find the optimal predicting time point and predicting threshold. According to multi-time points of DWI examination during the course of CCRT and analysis of the change of ADC values, DADC, DWI based GTV volumes, we found that there was a significant difference in ADC values between the CR group and PR group from the beginning to the end of treatment course, especially, the pre-CCRT ADC values of the CR group were significantly higher than those of PR group, and the difference continued to the end of course. However, there was no significant difference in DADC between the two groups. We could only observe from Fig. 1a that the curve of CR group stepped-up even more after the time point of 2nd weekend, with no significant difference observed between the two groups. The predictive value of ADC for TE and survival in esophageal cancer has not been well established. Two studies [13,16] from Japan indicated that ADC value of pre-treatment may be a useful marker to predict TE and high ADC values indicated a well TE or survive. This viewpoint was in accordant with ours, while in the study of van Rossum [15], there was no significant difference in initial ADC values between path CR group and no path CR group. Interestingly, in the study of De Cobelli [17], the authors observed that responders showed lower pre-NT ADC (1.32 vs 1.63 10 3 mm2/s; p = 0.002) than non-responders. We think this may be due to two reasons: (1) The presence of necrotic (poorly perfuse and hypoxic, but with high ADC value) areas, which are less sensitive to CRT, explaining the higher ADC values of poor responders [13]. (2) Generally speaking, ADC is a measure of tumor cell density. A higher ADC value represents less tumor cells, thus has higher probability for CR, as compared to a low ADC value. In this study, the ADC values have inverse correlation relationships with T stages and GTV volumes (r = 0.344, 0.478; p = 0.034, 0.002) and the GTV volumes of CR group are smaller (45.0 cm3 vs 53.1 cm3), we presumed that the larger the GTV volumes, the more and compacted the tumor cells, explaining the lower ADC values of poor TE. Because we found that there were significant differences of ADC values between different time points and different groups according to the description and statistics of DWI parameters, so a Logistic regression analysis was conducted with ADC values of 1st to 5th weekends and a series of correlative clinical factors. The objective of the analysis was to find significant parameters which had effects on TE of esophageal cancers. The results of multivariate analysis showed that only ADC values of the 3rd weekend were the independent prognostic factor, the OR value was 0.303, which indicated the third-weekend ADC values were a comparatively strong protection parameter of TE, the higher of ADC values, the superior of TE. The following ROC curve analysis found that the prognosis
Table 3 The results of ROC curve analysis. Time point
Area under curve
p
ADC threshold (10
1st weekend 2nd weekend 3rd weekend 4th weekend 5th weekend
0.786 0.822 0.822 0.810 0.769
0.003 0.001 0.001 0.001 0.005
1.87 2.11 2.14 2.34 2.50
3
mm2/s)
Sensitivity (%)
Specificity (%)
75.00 70.00 85.00 85.00 70.00
72.22 88.89 77.78 72.22 72.22
Bold values indicated that the Area under curve of 2nd and 3rd weekend was biggest, and the prediction efficacy of the two time points may be optimal.
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021
6
DWI in esophageal carcinoma
efficiency of second and third-weekend ADC values was optimized (0.822), the result indicated that two time points were effective monitoring time points of TE. So we suggested that DWI scan could be performed at the end of 2nd or 3rd week and the measured ADC values could be used in determining the subsequent. In recent years, studies [14,15] from Japan and Netherlands also considered that treatment-induced ADC or DADC changing during the first 2– 3 weeks of treatment for esophageal cancer seemed to be highly predictive of therapeutic response. An additional point to make is there were 8 patients occurred in 1 year, the ADC values of the group ascended from the beginning to the 4th week, but subsequently, although the radiation doses still increase, the ADC values ‘‘dropped” steeply, at the same time, the curve of DWI based GTV volumes presented a ‘‘shoulder area” at the time of 3rd to 5th weeks, the reduced ranges of DWI based GTV volumes were significantly decreased. We think this phenomenon may indicate a high tendency of tumor recurrence. According to all the above studies, we conclude that DWI can be a useful biomarker to predict TE of esophageal cancers in early time, the treatment-induced change in ADC in GTV during the first 2–3 weeks seemed highly predictive to TE, the phenomenon of ADC value dropping in late treatment course or DWI based GTV volumes stopping shrinking maybe indicators for possible tumor recurrence. Conflict of interest statement None of the authors have financial or personal conflicts of interest to disclose. References [1] Hein PA, Kremser C, Judmaier W, et al. Diffusion-weighted magnetic resonance imaging for monitoring diffusion changes in rectal carcinoma during combined, preoperative chemoradiation: preliminary results of a prospective study. Eur J Radiol 2003;45:214–22. [2] Kremser C, Judmaier W, Hein P, et al. Preliminary results on the influence of chemoradiation on apparent diffusion coefficients of primary rectal carcinoma measured by magnetic resonance imaging. Strahlenther Onkol 2003;179: 641–9.
[3] Galbán CJ, Mukherji SK, Chenevert TL, et al. A feasibility study of parametric response map analysis of diffusion-weighted magnetic resonance imaging scans of head and neck cancer patients for providing early detection of therapeutic efficacy. Transl Oncol 2009;2:184–90. [4] Dzik-Jurasz A, Domenig C, George M, et al. Diffusion MRI for prediction of response of rectal cancer to chemoradiation. Lancet 2002;360:301–8. [5] Hayashida Y, Yakushiji T, Awai K, et al. Monitoring therapeutic responses of primary bone tumors by diffusion-weighted image: initial results. Eur Radiol 2006;16:2637–43. [6] Mardor Y, Pfeffer R, Spiegelmann R, et al. Early detection of response to radiation therapy in patients with brain malignancies using conventional and high b-value diffusion-weighted magnetic resonance imaging. J Clin Oncol 2003;21:1094–100. [7] van Overhagen H, Laméris JS, Berger MY, et al. Supraclavicular lymph node metastases in carcinoma of the esophageal and gastroesophageal junction: assessment with CT, US and US-guided fine-needle aspiration biopsy. Radiology 1991;179:155–8. [8] Takashi M, Yasumasa N, Yutaka S, et al. Optimal size criteria of malignant lymph nodes in the treatment planning of radiotherapy for esophageal cancer: evaluation by computed tomography and magnetic resonance imaging. Radiother Oncol 1996;36:1091–8. [9] Lan Wang, Chun Han, Shuchai Zhu, et al. A pathological comparative study of diffusion-weighted imaging and computed tomography in determination of lesion length for esophageal carcinoma. Chin J Radiat Oncol 2015;24:373–6. [10] Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228–47. [11] Hou DL, Shi GF, Gao XS, et al. Improved longitudinal length accuracy of gross tumor volume delineation with diffusion weighted magnetic resonance imaging for esophageal squamous cell carcinoma. Radiat Oncol 2013;8:169. [12] Wang L, Han C, Zhu S, et al. Investigation of using diffusion-weighted magnetic resonance imaging to evaluate the therapeutic effect of esophageal carcinoma treatment. Oncol Res Treat 2014;3:112–6. [13] Giganti F, Salerno A, Ambrosi A, et al. Prognostic utility of diffusion-weighted MRI in oesophageal cancer: is apparent diffusion coefficient a potential marker of tumour aggressiveness? Padiol Med 2016;121:173–80. [14] Imanishi S, Shuto K, Aoyagi T, et al. 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. [15] van Rossum PS, van Lier AL, van Vulpen M, et al. Diffusion-weighted magnetic resonance imaging for the prediction of pathologic response to neoadjuvant chemoradiotherapy in esophageal cancer. Radiother Oncol 2015;115:163–70. [16] Aoyagi T, Shuto K, Okazumi S, et al. Apparent diffusion coefficient values measured by diffusion-weighted imaging predict chemoradiotherapeutic effect for advanced esophageal cancer. Dig Surg 2011;28:252–7. [17] 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.
Please cite this article in press as: Wang L et al. The diffusion-weighted magnetic resonance imaging (DWI) predicts the early response of esophageal squamous cell carcinoma to concurrent chemoradiotherapy. Radiother Oncol (2016), http://dx.doi.org/10.1016/j.radonc.2016.10.021