XML Template (2015) [26.1.2015–10:15am] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/ACRJ/Vol00000/150003/APPFile/SG-ACRJ150003.3d
(ACR)
[1–9] [PREPRINTER stage]
Acta Radiol OnlineFirst, published on January 30, 2015 as doi:10.1177/0284185114568908
Original Article
Diffusion-weighted imaging for evaluating lymph node eradication after neoadjuvant chemoradiation therapy in locally advanced rectal cancer
Acta Radiologica 0(0) 1–9 ! The Foundation Acta Radiologica 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0284185114568908 acr.sagepub.com
Kyeong Hwa Ryu1, Seung Ho Kim1, Jung-Hee Yoon1, Yedaun Lee1, Jin Ho Paik2, Yun-Jung Lim1 and Kwang Hwi Lee1
Abstract Background: As lymph node (LN) eradication is the prerequisite for clinical surveillance or local excision for patients who have achieved a complete response after preoperative chemoradiation therapy (CRT), the radiological evaluation of LN eradication is important. Purpose: To evaluate the added value of diffusion-weighted imaging (DWI) in the evaluation of LN eradication after CRT in patients with locally advanced rectal cancer (LARC). Material and Methods: Ninety-five consecutive patients (64 men, 31 women; mean age, 59 years; range, 32–82 years) who underwent pre- and post-CRT 1.5-T MRI with DWI (b ¼ 0, 1000 s/mm2) were enrolled. To evaluate the added value of DWI in the evaluation of LN eradication after CRT, two radiologists first independently read the pre- and post-CRT T2-weighted (T2W) images and then read the combined T2W imaging set and the pre- and post-CRT DWIs with a 4week interval. The radiologists recorded their confidence scores for LN eradication using a 5-point scale on a per-patient basis. The diagnostic performances were compared between the two reading sessions for each reader with pair-wise comparisons of receiver-operating characteristic curves. Histopathological reports served as the reference standards for LN eradication. Results: The study population consisted of an LN-eradicated group (n ¼ 66) and a non-eradicated group (n ¼ 29). The diagnostic performances did not significantly differ between the two reading sessions for the two readers (AUCs for reader 1, 0.770 and 0.774, P ¼ 0.8155; for reader 2, 0.794 and 0.798, P ¼ 0.8588). Conclusion: Adding DWI to T2W imaging provided no additional diagnostic benefit for the evaluation of LN eradication following CRT in patients with LARC.
Keywords Diffusion-weighted image, lymph node, lymphatic metastasis, rectum, neoplasm Date received: 14 February 2014; accepted: 31 December 2014
Introduction As lymph node (LN) metastasis is a significant predictor of local recurrence, the evaluation of LN status is important for restaging following neoadjuvant chemoradiation therapy (CRT) (1). Although there are controversies related to the treatment strategy for patients with a clinical complete response (CR), this selected group might have the opportunity to undergo clinical surveillance based on the ‘‘wait-and-watch’’ policy or local excision (2–4). However, the prerequisite
1 Department of Radiology, Inje University College of Medicine, Haeundae Paik Hospital, Busan, Republic of Korea 2 Department of Pathology, Seoul National University Bundang Hospital, Gyeonggi-do, Republic of Korea
Corresponding author: Seung Ho Kim, Department of Radiology, Inje University College of Medicine, Haeundae Paik Hospital, Haeundae-ro 875, Haeundae-gu, Busan 612-030, Republic of Korea. Email:
[email protected] Downloaded from acr.sagepub.com at CHINESE UNIV HONG KONG LIB on February 11, 2015
XML Template (2015) [26.1.2015–10:15am] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/ACRJ/Vol00000/150003/APPFile/SG-ACRJ150003.3d
(ACR)
[1–9] [PREPRINTER stage]
2
Acta Radiologica 0(0)
for those treatments is the eradication of regional LN metastases. Therefore, the radiological evaluation of LN eradication following CRT is of great importance for the proper selection of patients who might benefit from sphincter-saving based on the ‘‘wait-and-watch’’ policy or local excision. Modern imaging techniques, such as computed tomography (CT), magnetic resonance imaging (MRI), and endoluminal ultrasound, all lack sufficient accuracy to identify LN metastases (5–7). The accuracy of MRI in the detection of LN metastasis varies widely and is thought be in the range of 62–85% because the size of the LN alone cannot aid the characterization of LNs (6,8,9). Diffusion-weighted imaging (DWI) has recently emerged as a functional imaging modality that can aid conventional morphologic imaging tools and has produced promising results in the field of oncology (10–13). Regarding LN metastasis, there are few studies of the restaging of LN status following CRT using the apparent diffusion coefficient (ADC) (14). Although the ADC benefits LN characterization, the overlap in ADC values is challenging, and the measurement of the ADC in small LNs (less than 2 mm in diameter) is a technically difficult, time-consuming and demanding procedure. In contrast, the visual assessment of LNs is simple and intuitive for radiologists in clinics. However, the major drawback of visual assessment is its poor specificity, which results from the high signal intensities (SI) of LNs on DWI (14–16). However, to the best of our knowledge, there are no studies of the imaging characteristics of LNs on DWI following CRT. Furthermore, the diagnostic value of DWI has not been evaluated in terms of the identification of LN eradication following CRT. Therefore, the purpose of this study was to evaluate the diagnostic value of additional DWI in the detection of LN eradication after CRT compared to that of the use of T2weighted (T2W) images alone in patients with locally advanced rectal cancer (LARC).
excision. The patients received concomitant radiation treatments (45 Gy/25 fractions þ 5.4 Gy/3 fractions) and two cycles of simultaneous chemotherapy with 5-fluorouracil (500 mg/m2/day). Four to six weeks after the neoadjuvant CRT was complete, the patients underwent surgery. Among the patients, 42 were excluded for the following reasons: (i) missing pathologic data (i.e. missing designation of the LN locations [n ¼ 7] and missing fractions [n ¼ 11]); and (ii) no available pre-CRT DWI (n ¼ 24). Therefore, 95 consecutive patients (64 men, 31 women; mean age, 59 years; range, 32–82 years) were included in the study. All patients underwent surgical excisions. The average interval between the post-CRT MRI and the surgery was 7 days (range, 1–14 days). The detailed case accrual process is summarized in Fig. 1. Fifty-eight patients (61%) achieved LN downstaging, and 66 patients (70%) achieved LN eradication. Therefore, 66 patients were in the LN-eradicated group, and 29 patients were in the non-eradicated group. The pre-CRT LN staging based on the MRI findings and the post-CRT pathologic LN staging of the study population are summarized in Table 1.
Histopathological evaluation Based on the histopathology reports, a total of 1105 LNs in 95 patients were identified (1030 non-metastatic LNs and 75 metastatic LNs). Of these 1105 LNs, 199 LNs (30 metastatic LNs and 169 non-metastatic LNs) were matched with radiological-pathological 1:1 correlations and were included in the further analyses. The degrees of tumor response to neoadjuvant CRT were complete (n ¼ 13), near complete (n ¼ 23), moderate (n ¼ 37), and minimal response (n ¼ 22). Mucinous carcinoma or total mucin pools following neoadjuvant CRT were present in eight of 95 cases (8%).
MRI technique Material and Methods Our institutional review board approved this retrospective study and waived informed patient consent.
Patients and patient selection criteria Between April 2008 and March 2010, 137 consecutive patients who met the inclusion criteria were selected for this study. The inclusion criteria were as follows: (i) patients with endoscopic-biopsy proven rectal adenocarcinoma; (ii) patients with pre-therapy MRIs; (iii) patients with neoadjuvant CRT; (iv) patients with post-therapy MRI; and (v) patients with surgical
All of the MR images, including the pre- and post-CRT T2W images with DWI, were taken using a 1.5-T scanner (Signa Excite, GE Healthcare, Milwaukee, WI, USA) with a phased-array torso coil (USA instruments, Aurora, OH, USA). The patient lay supine after the administration of approximately 80–100 mL of ultrasound gel (Supersonic, Sungheung Co. Ltd, Seoul, PR Korea) into the rectum (17). The oblique axial, coronal, and sagittal planes were imaged with T2W fast-spin echo sequences. The oblique axial and coronal images were scanned as orthogonal and parallel, respectively, to the long axis of the tumor (18). Immediately after the acquisition of the T2W images, the oblique axial plane
Downloaded from acr.sagepub.com at CHINESE UNIV HONG KONG LIB on February 11, 2015
XML Template (2015) [26.1.2015–10:15am] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/ACRJ/Vol00000/150003/APPFile/SG-ACRJ150003.3d
(ACR)
[1–9] [PREPRINTER stage]
Ryu et al.
3
Fig. 1. Flowchart of the case accrual process.
was imaged with free-breathing DWI via a single-shot echo planar imaging sequence. Two b factors of 0 and 1000 s/mm2 were used (19,20). The total scan time was 30 min. Table 2 illustrates the routine MRI protocol and the sequence parameters.
CRT DWIs (b ¼ 1000 s/mm2) to identify the imaging characteristics of LNs on post-CRT DWIs. Four criteria were used for the classification of the imaging characteristics of the metastatic and non-metastatic LNs on the post-CRT DWI. The imaging characteristics were composed of border, shape, internal heterogeneity, and SI. Several imaging characteristics were extracted from the image analyses. The categorized imaging characteristics from the post-CRT DWIs are illustrated in Fig. 3. The metastatic LNs had ill-defined, amorphous, and intermediate SIs (P < 0.0001) or well-defined, lobulating, and heterogeneous intermediate SIs (P < 0.0001); in contrast, the non-metastatic LNs had well-defined, homogeneous, high SI, dot appearances on the post-CRT DWIs (P < 0.0001). The imaging characteristics of the LNs on the post-CRT DWIs are summarized in Table 3.
Image analysis
Evaluation of diagnostic performance
First, we looked at the imaging characteristics of the LNs on DWI to acquire an idea of which LNs would be positive upon resection after CRT. Next, we selected the significant ones as criteria for calling LNs positive on post-CRT DWI. Last, to evaluate the added value of DWI for the identification of LN eradication following CRT, an interval reading was performed; the first session was based only on the T2W images, and the second session was based on the T2W images and DWIs. The study process is summarized in Fig. 2.
After the determination of the imaging characteristics of the LNs on the post-CRT DWIs, two additional radiologists (with 5 and 4 years of clinical experience in interpreting rectal MRIs) who were blinded to the histopathologic results independently performed readings of the T2W images and, after a 4-week interval, readings of the combined T2W imaging with DWI image sets to evaluate the added value of the imaging characteristics on the post-CRT DWIs in the evaluation of LN eradication. During the first reading session, the readers independently reviewed both the pre- and post-CRT T2W images simultaneously. They recorded their confidence levels regarding LN eradication after CRT using a 5-point scale. Metastatic LNs on the pre-CRT T2W images were predefined as LNs with
Table 1. Pre-therapy radiologic and final pathologic lymph node stages of the study population. Pre-CRT radiologic stage N1 N2 Total
Final pathologic stage
N0
N0 N1 N2 Total
20
34 12
20
46
12 12 5 29
66 24 5 95
Imaging characteristics of the LNs on post-CRT DWIs. Two radiologists (with 4 and 2 years of clinical experience in reading rectal MRIs) who were blinded to the histopathologic results reviewed in agreement the post-
Downloaded from acr.sagepub.com at CHINESE UNIV HONG KONG LIB on February 11, 2015
XML Template (2015) [26.1.2015–10:15am] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/ACRJ/Vol00000/150003/APPFile/SG-ACRJ150003.3d
(ACR)
[1–9] [PREPRINTER stage]
4
Acta Radiologica 0(0)
Table 2. MR sequence parameters. Parameter
T2W axial, sagittal, and coronal FSE
T1-weighted axial and sagittal SE
DWI (b ¼ 0, 1000 s/mm2)
TR TE ETL Slice thickness Slice gap Matrix size NEX FOV Acquisition time Slices (n)
4500 107.5 16 5 1.0 384 224 4 240 240 4 min 17 s 23–26
500–600 11 1 5 1.0 320 192 2 240 240 3 min 20 s – 3 min 59 s 23–26
8000 85.2 1 5 1.0 160 160 4 300 300 2 min 8 s 26
The diffusion-weighted imaging (DWI) was performed using a single-shot echo planar imaging technique. ETL, echo train length; FOV, field of view; FSE, fast-spin echo; NEX, number of excitations; TE, echo time; TR, repetition time.
Fig. 2. Flowchart of the overall study design.
short axis diameters greater than 8 mm or indistinct, spiculated borders with mottled heterogeneous SIs regardless of the short axis diameter (21). LN eradication was radiologically predefined according to the following criteria: (i) when all LNs on the pre-CRT T2W imaging were assumed to be benign, all of the LNs on the post-CRT T2W imaging were considered to be eradicated; (ii) when the LNs on the pre-CRT T2W imaging were presumed to be metastatic, a lack of residual LNs or tiny residual LNs exhibiting completely dark SIs (i.e. SIs lower than that of the muscle, which were assumed to indicate the replacement of the LN with fibrosis) on the post-CRT T2W imaging were regarded as eradicated; and (iii) when presumed metastatic LNs on the pre-CRT T2W imaging did not exhibit completely dark SIs on the post-CRT T2W imaging, the LNs were considered to be not eradicated. LN
eradication was assessed on the following 5-point scale: 1, definitely absent; 2, probably absent; 3, possibly present; 4, probably present; and 5, definitely present. At the second reading session, the readers were asked to score their confidence levels based on the combined image set (pre- and post-CRT T2W images þ pre- and post-CRT DW images) using the same 5-point scale. The criteria for residual metastatic LNs on the post-CRT DWIs were based on the results of the imaging characteristics on the post-CRT DWIs. The decisions regarding LN eradication on the second reading session were rendered via an integration of the two image sets. When the decisions based on the two image sets were discordant or the DWI results were equivocal, the readers were asked to base their decisions on the T2W image set.
Downloaded from acr.sagepub.com at CHINESE UNIV HONG KONG LIB on February 11, 2015
XML Template (2015) [26.1.2015–10:15am] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/ACRJ/Vol00000/150003/APPFile/SG-ACRJ150003.3d
(ACR)
[1–9] [PREPRINTER stage]
Ryu et al.
5 Table 3. Imaging characteristics of the lymph nodes on DWI after CRT. Histopathology Imaging characteristic Amorphous Heterogeneous lobulating Dense dot Ill-defined dot Homogeneous lobulating Crescent Ill-defined halo
Benign LN
Metastatic LN
P
2 0
19 4