Highlights
-
•
The rate of pathological complete response (pCR) following NACRT was 14.7 %.
-
•
Combination of morphology and DWI patterns was 89–96 % accurate in predicting pCR.
-
•
Radiologists of varying experience agreed best with combined approach.
-
•
Radiologists were more confident of their response using the combined approach.
Abbreviations: NACRT, neoadjuvant chemoradiotherapy; LCCRT, neoadjuvant long course chemoradiotherapy; MR, TRG MR-tumour regression grade; p-TRG, pathological tumour regression grade; DWI, diffusion weighted imaging; T2-HR MRI, T2 high resolution MRI; cCR, clinical complete response; pCR, pathological complete response; IR, incomplete response; AUC, area under the curve; AJCC, American Joint Committee on Cancer; MERCURY, Magnetic Resonance Imaging and Rectal Cancer European Equivalence Study
Keywords: MRI, Diffusion weighted imaging, Rectal cancer, Complete response, Neoadjuvant chemotherapy
Abstract
Purpose
To assess the diagnostic performance, interobserver agreement and confidence level for determining response to neoadjuvant chemoradiotherapy (NACRT) using morphology based MR-tumour regression grade (MR TRG), diffusion weighted imaging (DWI) patterns and their combination in patients with locally advanced rectal cancer.
Methods
This was a retrospective study including patients with locally advanced rectal cancer treated with NACRT and subsequent surgery. Two independent radiologists blinded to the histopathology reviewed staging and restaging MRI. Diagnostic performance of morphology based MR-TRG, DWI patterns and their combination for determining complete (CR) and incomplete (IR) response was assessed with pathological response as the reference. Likert’s scale was used to assess the radiologist’s level of confidence. Interobserver agreement was determined using Kappa statistics.
Results
The study included 251 patients (mean age of 47.9+/-14 (range 19–86) years, M:F = 164:87). Rate of pathological CR was 14.7 % (n = 37). Pattern based interpretation of DWI and combined approach (DWI + T2-HR) had superior diagnostic performance than morphology based assessment alone with area under curve (AUC) for T2HR, DWI and their combination being 0.531, 0.887, 0.874 respectively for observer 1 and 0.558, 0.653, 0.678 respectively for observer 2, p < 0.001. Interobserver agreement was substantial (k = 0.688) for combined approach, moderate (k = 0.402) for DWI patterns and fair (k = 0.265) for T2 HR MRI with both observers exhibiting highest level of confidence for determining response with the combined approach.
Conclusion
Complete response to neoadjuvant chemoradiotherapy can be determined with excellent accuracy, substantial interobserver agreement and high level of confidence by combined interpretation of DWI and T2 high resolution MRI.
1. Introduction
In patients with rectal cancer, excellent long-term outcomes have been demonstrated for the ‘watch and wait’ strategy, after a complete clinical response (cCR) to neoadjuvant chemoradio therapy (NACRT). This has led to the concept of organ preservation in rectal cancer [[1], [2], [3]]. Patients with rectal cancer receiving NACRT have a pathological complete response (pCR) rate varying from 10 to 20% [2,4]. MRI is being increasingly performed following NACRT in order to assess the response and to identify the local stage since these have important bearing on subsequent treatment plan and the outcome [5]. The five-point MR-tumour regression grade by the Magnetic Resonance Imaging and Rectal Cancer European Equivalence Study (MERCURY) study group, adapted from pathological tumour regression grade, and similar three-point scoring system by ESGAR consensus meeting, both of which are based on assessment of T2 high resolution MR images, are being used to assess response to NACRT [[6], [7], [8], [9]]. However, the currently available methods of determining complete response to NACRT are suboptimal in predicting pCR, with a diagnostic accuracy between 50.2 % to 88 %; correlate poorly with pCR and only have fair interobserver agreement [8,10]. Post treatment related edema, fibrosis, necrosis and poor interface between tissues often lead to over staging and over estimation of tumour extent. Diffusion weighted imaging (DWI) has shown promising results in this area [[10], [11], [12], [13], [14]]. Through this large retrospective study, we aimed to assess the diagnostic performance, interobserver agreement and the level of radiologist’s confidence for determining complete response using T2 high resolution (T2-HR) MRI morphology based MR-tumor regression grade (MR-TRG), DWI patterns matching MR-TRG and a combination of both.
2. Materials and methods
2.1. Setting
This was an institutional review board approved IRB min no: 12011 retrospective review conducted on patients who underwent surgery for rectal cancer between April 2014 – March 2018 in a dedicated Colorectal Surgery unit of a 3000 bedded tertiary care teaching hospital.
2.1.1. Patients
Patients with locally advanced rectal adenocarcinoma who underwent neoadjuvant long course chemoradiotherapy (LCCRT) followed by surgery, and had a staging MRI and post LCCRT restaging MRI, were included in the study. Patients who underwent upfront surgery, or received other forms of neoadjuvant therapy apart from LCCRT, as well as those not having appropriate or optimal imaging were excluded. Patients with mucinous and signet ring cell tumours were also excluded. Standard LCCRT protocol included 45−50 Gy radiation in 25–28 fractions with a 5.4 Gy boost with intravenous 5-fluorouracil infusion or oral capecitabine in standard radio sensitising doses during NACRT. Fig. 1 shows the consort flow chart of patients included and excluded from the study.
Fig. 1.
Consort flow chart showing patients included in the study.
2.1.1.1. Sample size
For a 10 % precision, 80 % power, 5% error and assumption that there might be 40 % discordance between morphology based assessment on T2 HR MRI and DWI patterns, a sample size of 40 pathological complete responders and 145 pathological incomplete responders was calculated. The number of patients included in the study matched the sample size calculation.
2.1.2. MRI protocol
All the patients underwent MRI scans in one of the following two MRI scanners: Intera 22 Achieva 3.0 T™ (Philips Healthcare, Best, Netherlands) with a 16 channel phased-array body coil and Magnetom Avanto fit, 1.5 T (Siemens Healthcare, Erlangen, Germany) with an 18 channel body coil using Tim 4G coil technology. Table 1 shows the imaging protocol used in these scanners. No bowel preparation or spasmolytics were used in our patients. Standard T2 HR MRI of the pelvis was performed in sagittal, oblique axial (perpendicular to the rectum) and oblique coronal (parallel to the rectum) planes. Axial DWI was obtained using respiratory-triggered, single-shot echoplanar imaging with b-values of 0, 400 and 800 s mm−2. Depending on the respiratory efficiency of each patient, the acquisition time for this sequence ranged from 3 to 5 min. ADC maps were automatically generated in the system.
Table 1.
MRI imaging protocol used in 3T and 1.5T scanners for both T2 high resolution (T2-HR) MRI and diffusion weighted imaging (DWI).
| Scan parameters | Intera 22 Achieva 3.0 T |
Magnetom Avanto fit, 1.5T |
||
|---|---|---|---|---|
| T2-HR | DWI | T2-HR | DWI | |
| Repetition Time | 3500 | 3750 | 4000 | 3000 |
| Echo time | 90 | 75 | 105 | 61 |
| Slice thickness (mm) | 3 | 5 | 3 | 5 |
| FOV (cm) | 20 | 25 | 18 - 20 | 22 |
| Matrix | 368 × 290 | 128 × 116 | 325 × 250 | 128 × 116 |
| Sensitivity encoding factor | 2-2.5 | 1.7 | 2 | 1.9 |
| Echo train length | 25 | 1 | 12 | 1 |
| No. of signal averages | 2-6 | 4-6 | 2-6 | 4-6 |
| No. of slices | 20-40 | 20-30 | 20-40 | 20-30 |
| Acquisition time (minutes) | 3-4 | 3-4 | 4-6 | 5 |
| B-values | – | 0, 400, 800 | – | 0, 400, 800 |
| Echo planar imaging factor | – | 77 | – | 108 |
| Fat suppression technique | – | SPAIR | – | SPAIR |
2.1.3. Image interpretation and response evaluation
Images were reviewed on picture archiving and communication system (GE Health system, Barrington, IL) by two independent radiologists with 8 years and 2 experience in interpreting rectal cancer MRI. Radiologists were blinded to the surgical findings and histopathology reports.
The readers assessed response on post NACRT restaging MRI by using T2 high resolution MRI, diffusion weighted imaging and their combination in a pre-decided order. Staging MRI was available for comparison while assessing response on post NACRT restaging MRI. T2 HR sagittal images of staging and post NACRT restaging MRI were compared in order to accurately locate the site of the rectal growth. This was followed by review of T2 HR images, DWI and subsequently both together, with documentation of response assessment and the level of confidence for each. Response on T2 HR images were assessed according to 5-point MR-TRG scores of the MERCURY study group [7,15].
We used the patterns of response described on high b-value DWI by previous workers [9,13] and compared these patterns to Mandard’s pathological TRG (p-TRG) in a cohort of 64 patients of our previous study [16] who were outside of the current study population. We did this exercise in order to correctly identify the DWI patterns that best identifies with p-TRG. Following this, two observers independently read the DWI images to determine response based on DWI patterns. ADC maps were used to facilitate pattern based response assessment on DWI. Table 2 summarises the description of MR-TRG and DWI patterns. Fig. 2, Fig. 3, Fig. 4 show examples comparing the responses on T2HR images and DWI on restaging MRI.
Table 2.
Criteria for response assessment on post neoadjuvant chemoradiotherapy (NACRT) MRI using T2-high resolution (T2-HR) MRI based MR-TRG by MERCURY study group [7], diffusion weighted imaging (DWI) patterns and their combination.
| MR-TRG | T2-HR MRI aMERCURY (2012) | DWI | Combination of T2 HR and DWI |
|---|---|---|---|
| 1 | Normal rectal wall or thin band of fibrosis | No foci of restricted diffusion | T2 HR 1/2/3 + DWI 1 |
| 2 | Thick band of fibrosis with doubtful residual tumor | Few scattered foci of restricted diffusion | T2 HR 2/3 + DWI 2 |
| 3 | Fibrosis/ mucin > 50 % with tumor | C-shaped band or nodular focus of restricted diffusion | T2 HR 2/3 + DWI 3 |
| 4 | Little fibrosis, mostly tumour | Smaller than pre-Rx MR | T2 HR 4 + DWI 4 |
| 5 | No response or progression | No change since previous | T2 HR 5 + DWI 5 |
MERCURY - Magnetic Resonance Imaging and Rectal Cancer European Equivalence Study (7).
Fig. 2.
Restaging MRI of six different patients with pathological complete response show the spectrum of appearances of complete response on T2 HR images and DWI. (A–C) In three different patients who had features of complete response on both T2 HR images and DWI, there was with thin hypointense band of fibrosis in patient 1(A) and patient 2(B) and there was normalisation of rectal wall in patient 3(C). All three patients had no restricted diffusion on DWI. (D–F) MRI of three different patients who had features of complete response on DWI (dwi-TRG-1), but were either called as mrTRG-2 (D,E) due to thick band of fibrosis or mrTRG-3 (F) due to thick band of predominant fibrosis interspersed by intermediate signal areas on T2 HR images.
Fig. 3.
Restaging MRI of two different patients with pathological incomplete response (pTRG-2). (A,B) Patient 1 was called as complete response based on T2 HR image (mr-TRG1), but DWI showed few scattered foci of restricted diffusion suggestive of incomplete response (dwi-TRG-2). (C,D) Patient 2 had thick band of fibrosis on T2 HR images (mr-TRG-2) and the DWI showed few scattered foci of restricted diffusion suggestive of dwi-TRG 2.
Fig. 4.
Restaging MRI of three different patients with incomplete response (pTRG-3). (A) Patient 1 showed thick band of fibrosis (mr-TRG-2) on T2 HR images and DWI showed a C-shaped restricted diffusion along the left wall of the rectum suggestive of dwi-TRG-3. (B) Patient 2 had greater than 50 % fibrosis with intermediate signal intensity areas along the left wall of rectum (mrTRG-3). DWI showed a nodular focus of restricted diffusion suggestive of dwi-TRG-3. (C) Patient 3 had less than 50 % tumor signal areas than fibrosis on T2 HR images and was called mr-TRG-4. DWI showed curvilinear C shaped focus of restricted diffusion from 4-10 O’clock suggestive of dwi-TRG3.
A five point Likert’s scale was used to document the level of confidence in assessing response using MRI criteria in Table 2, with 5 being very confident and 1 not confident at all.
2.1.4. Reference standard
Surgical histopathology including pathological tumor regression grade (pTRG) by Mandard et al. [17] was the reference standard. The pathology specimens were reviewed by a single pathologist with 10 years of experience in pathology of colorectal cancer. Staging was performed in accordance with the 7th edition of American Joint Committee on Cancer (AJCC) guidelines. Patients with no viable tumour at the primary location or in the lymph nodes (ypT0N0) were considered to have pathological complete response (ypCR). All other patients including those with positive nodes (ypN+) and positive extramural vascular invasion (ypEMVI+) on histopathology were considered as incomplete response (ypIR).
2.2. Statistical analysis
Statistical analysis was performed using IBM SPSS Analytics 22.0 software. Descriptive statistics were reported as mean +/- 2SD and range for continuous variables and number with percentage for categorical variables. The rate of complete response was determined and given as a percentage. Diagnostic performance of determining response using T2-HR MRI morphology, DWI patterns and their combination was determined using two by two contingency tables with pathology as the reference standard. Interobserver agreement between the two radiologists was determined by Kappa statistics and was interpreted as follows: k < 0, poor agreement; k 0–0.2, slight agreement; k 0.21–0.40, fair agreement; k 0.41–0.60, moderate agreement; k 0.61–0.80, substantial agreement; and k 0.81–1.00, almost perfect agreement. Mode of Likert’s scale was determined to assess the most common level of confidence entered for each category of response assessment using T2 HR MRI based MR-TRG, pattern based interpretation of DWI and their combination.
3. Results
3.1. Patients and tumour characteristics
A total of 251 patients (164 males, 87 females) with a mean age of 47.9 +/- 14 years (range 19–86 years) were included in the study. Restaging MRI following NACRT was performed after a mean duration of 8 weeks (range of 6–14 weeks) following NACRT and within a mean of 7 days (range of 1–44 days) prior to the surgery. Table 3 compares the patient and tumour characteristics between pathological complete response (pCR) and incomplete response (pIR). Of the 251 patients, 14.7 % (N = 37) had pCR and the rest had pIR (N = 214). Experienced radiologist gave an assessment of y-mrT0 in 30 (11.9 %) patients. There was significant association between y-mrT and y-mrN stage with the y-pT and y-pN stage (p < 0.001).
Table 3.
Comparison of patient and tumor characteristics between pathological complete (pCR) and incomplete responders (pIR).
| Characteristics Total N = 251 | Pathological complete response (pCR) N = 37 | Pathological incomplete response (pIR) N = 214 | p-value |
|---|---|---|---|
| Age | 47.7 +/-12.1 years | 48.3+/-14.3 years | 0.301 |
| (22–70 years) | (19–86 years) | ||
| Male: Female | 26:11 | 138:76 | 0.495 |
| Pre-treatment biopsy: | 0.584 | ||
| Well to moderately differentiated | 8 (21.6 %) | 48 (22.4 %) | |
| Moderately differentiated | 24 (64.8 %) | 147 (68.6 %) | |
| Poorly differentiated | 5 (13.5 % | 19 (8.7 %) | |
| Length of tumor | 4.6 +/- 1.8 cm (1.8–10.6 cm) | 5.3 +/-1.9 cm (2–14 cm) | 0.066 |
| Location | <0.001 | ||
| Low/ low mid | 26 (70.2 %) | 92 (42.9 %) | |
| Mid/ mid high | 9 (24.4 %) | 59 (27.5 %) | |
| High | 2 (5.4 %) | 42 (19.6 %) | |
| Long segment | – | 21 (9.8 %) | |
| Annular | 17 (45.9 %) | 124 (57.9 %) | 0.174 |
| Semi-annular | 20 (54.1 %) | 90 (42.1 %) | |
| Intermediate signal intensity | 32 (86.5 %) | 181 (84.6 %) | 0.949 |
| Hyperintense | 2 (5.4 %) | 14 (6.5 %) | |
| Mixed signal intensity | 3 (8.1 %) | 19 (8.9 %) | |
| DWI appearance on pretreatment MRI: | 0.237 | ||
| Entire tumor showed diffusion restriction | 22 (59.5 %) | 130 (60.7 %) | |
| Few scattered foci of diffusion restricted | 11 (29.6 %) | 49 (22.9 %) | |
| Facilitated diffusion with no foci of diffusion restricted | 4 (10.8 %) | 8 (3.7 %) | |
| No restricted diffusion or facilitated diffusion | 0 | 27 (12.6 %) | |
| T-stage on pretreatment staging MRI: | 0.211 | ||
| mrT2 | 5 (13.5 %) | 17 (8 %) | |
| mrT3 | 26 (70.3 %) | 147 (68.7 %) | |
| mrT4 | 6 (16.2 %) | 50 (23.4 %) | |
| mrN0 | 24 (64.9 %) | 72 (33.6 %) | 0.001 |
| mrN1 | 6 (16.2 %) | 97 (45.3 %) | |
| mrN2 | 7 (18.9 %)) | 45 (21.0 %) | |
| CRM =0 mm | 24 (64.9 %) | 135 (63.1 %) | 0.467 |
| EMVI positive | 7 (18.9 %) | 81 (37.9 %) | 0.003 |
| Pelvic sidewall disease | 3 (8.1 %) | 18 (8.4 %) | 0.951 |
| y-mrT-stage: | ymrT0 = 30 (30/251 = 11.9 %) | ||
| ymrT1/2 = 101/251 (40.2 %) | |||
| ymrT3 = 86/251 (34.2 %) | |||
| ymrT4 = 34/251 (13.5 %) | |||
| y-mrN -stage | y-mr- N0 = 167 (66.5 %) | ||
| y-mr- N1 = 71 (28.3) | |||
| y-mr- N2 = 13 (5.2 %) | |||
| y-pT-stage | ypT0 = 37(14.7 %) | ||
| ypTis = 5 (2%) | |||
| ypT1/T2 = 84 (33.5 %) | |||
| ypT3 = 113 (45 %) | |||
| ypT4 = 12 (4.8 %) | |||
| y-pN-stage | pN0 = 175 (69.7 %) | ||
| pN1 = 57 (22.7 %) | |||
| pN2 = 19 (7.6 %) | |||
3.2. Comparison of response assessment using MRI with reference standard
Table 4 shows the comparison of the response assessment using T2 HR MRI based MR-TRG, patterns of DWI and their combination with the pathological response. Though there was significant association between the response assessed using T2-HR MRI morphology and pathological response (chi square = 63.8, p < 0.001), there was no agreement between the two (k = 0.008, p = 0.033). There was significant association (chi square = 148.8, p < 0.001) and substantial agreement (kappa = 0.765, p < 0.001) between response assessed using DWI patterns and the pathological response. Similarly, there was a significant association (chi square = 170.02, p < 0.001) and excellent agreement (kappa = 0.811, p < 0.001) between the response assessed using a combination of T2HR MRI based MR-TRG plus patterns on DWI and the pathological response.
Table 4.
Comparison between the reference standard and the response assessment using T2 high resolution MRI, diffusion weighted imaging and their combination.
| MR-TRG | T2-HR MRI |
DWI |
Combination of T2 HR and DWI |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Observer 1 |
Observer 2 |
Observer 1 |
Observer 2 |
Observer 1 |
Observer 2 |
|||||||
| pCR N = 37 | pIR N = 214 | pCR N = 37 | pIR N = 214 | pCR N = 37 | pIR N = 214 | pCR N = 37 | pIR N = 214 | pCR N = 37 | pIR N = 214 | pCR N = 37 | pIR N = 214 | |
| 1 | 3 | 4 | 5 | 4 | 30 | 8 | 12 | 4 | 28 | 2 | 14 | 5 |
| 2 | 24 | 26 | 13 | 12 | 6 | 72 | 10 | 38 | 6 | 52 | 6 | 19 |
| 3 | 8 | 94 | 10 | 93 | 0 | 61 | 7 | 90 | 1 | 78 | 6 | 75 |
| 4 | 2 | 77 | 9 | 101 | 1 | 61 | 8 | 77 | 2 | 70 | 11 | 109 |
| 5 | 0 | 13 | 0 | 4 | 0 | 12 | 0 | 4 | 0 | 12 | 0 | 4 |
3.3. Diagnostic performance and interobserver agreement
Table 5 compares the diagnostic performance of both the observers using the three MRI strategies. Both the observers were able to call complete response in significantly more number of patients on DWI and the combination of T2 HR + DWI when compared to only T2 HR MRI. A similar trend was also observed in the sensitivity and the positive predictive value (PPV). Sensitivity and PPV for determining complete response using MR-TRG on T2 HR MRI alone was very low (8% and 42.8 % respectively for observer 1 and 13.5 %, 55.5 % respectively for observer 2). On the other hand, the sensitivity and PPV for DWI patterns was 81 % and 79 % respectively for observer 1; 75 % and 73.5 % respectively for observer 2. The specificity for determining complete response on MRI was very high (∼ 98 %) for all three methods. However, the NPV was higher for DWI patterns and combined T2HR + DWI when compared to T2 HR MRI alone. The area under the curve (AUC) for T2 HR MRI, DWI and their combination was 0.531, 0.887, 0.874 respectively for observer 1 and 0.558, 0.653, 0.678 respectively for observer 2 with p < 0.001.
Table 5.
Diagnostic performance of response assessment using T2 high resolution MRI, diffusion weighted imaging and their combination.
| Observer 1 |
Observer 2 |
|||||
|---|---|---|---|---|---|---|
| T2HR | DWI | T2+DWI | T2HR | DWI | T2+DWI | |
| CR | 7 | 38 | 30 | 9 | 16 | 19 |
| IR | 244 | 213 | 221 | 242 | 235 | 232 |
| Sensitivity | 8.1 | 81.1 | 75.6 | 13.5 | 32.4 | 37.8 |
| Specificity | 98.1 | 96.2 | 99 | 98.1 | 98.1 | 97.6 |
| PPV | 42.8 | 78.9 | 93.3 | 55.5 | 75 | 73.6 |
| NPV | 86 | 96.7 | 95.9 | 86.7 | 89.3 | 90 |
| Accuracy | 84.8 | 94 | 95.6 | 85.6 | 88.4 | 88.8 |
| AUC | 0.531 | 0.887 | 0.874 | 0.558 | 0.653 | 0.678 |
The interobserver agreement was only fair (0.231, p < 0.001) for determining response using T2 HR MRI based MR-TRG; moderate (0.402, p < 0.001) for DWI patterns and substantial (0.688, p < 0.001) for the combination of both.
3.4. Level of confidence
The level of confidence for each assessment category for both observers combined is shown in Table 6. In all the three methods, the radiologists called MR-complete response (MR-CR) only when confident. However, there was significant difference in the total number of MR-CR for both readers combined using the three methods with 16, 54 and 49 MR-CR using T2 HR MRI morphology, DWI patterns and their combination respectively. Similarly, higher percentage of responses matched the most common level of confidence with DWI (77 %) and combined approach (79.6 %) when compared to T2 HR images (62.5 %). From being not at all confident and less confident with MR-TRG 2 and 3 responses, the confidence level rose to ‘confident’ and ‘very confident’ with a combined approach in over 60 %. Addition of DWI and the combined approach did not add value in MR-TRG 4 and 5 responses since there was a high level of confidence with all three approaches.
Table 6.
Mode of the level of confidence for each assessment category for both the observers combined. Likert’s scale used for the level of confidence is as follows: 5 - very confident, 4 - confident, 3 - not sure, 2 - less confident and 1 - not at all confident.
| MR-TRG | T2 HR MRI |
DWI |
T2 HR + DWI |
|||
|---|---|---|---|---|---|---|
| Mode of the level of confidence | (N/TR) and % | Mode of the level of confidence | (N/TR) and % | Mode of the level of confidence | (N/TR) and % | |
| 1 | 4 | (10/16) 62.5 | 4 | (42/54) 77 | 5 | (39/49) 79.6 |
| 2 | 1 | (52/75) 69.3 | 2 | (52/126) 41.2 | 4 | (51/83) 61.4 |
| 3 | 2 | (63/205) 30.7 | 4 | (113/158) 71.5 | 5 | (106/160) 66.2 |
| 4 | 5 | (153/189) 80.9 | 5 | (128/147) 87 | 5 | (143/192) 74.4 |
| 5 | 5 | (15/17) 88.2 | 5 | (12/16) 75 | 5 | (15/16) 93.7 |
*(N/TR) and %= number of responses for the mode of level of confidence divided by the total number of responses in that category for both observers combined and their percentage.
4. Discussion
The growing interest in organ preserving treatment strategy as an alternative to surgery in patients with locally advanced rectal cancer and complete clinical and radiological response to neoadjuvant chemoradiotherapy (NACRT) has created the need for accurate response assessment on post NACRT restaging MRI. While response assessment on T2 high resolution MRI using MR-TRG has suboptimal accuracy and interobserver agreement, DWI has shown promising results [10,11,13,16]. We assessed the diagnostic performance; inter reader agreement and level of confidence of determining response to NACRT on restaging MRI using MR-TRG of MERCURY study group, DWI patterns and their combination.
We found substantial agreement (k = 0.765) between DWI patterns and pTRG and excellent agreement (k = 0.811) between the combined approach (T2 HR MRI + DWI) and pTRG. This finding is supported by previous work by Iennicelli E et al. which showed good correlation (⍴ = 0.700) between DWI incorporated MR-TRG and pTRG [9]. We have shown that the response assessment using DWI patterns and the combination of T2 HR MRI + DWI patterns outperformed T2 HR MRI based MR-TRG. There were large improvements in sensitivity and PPV for DWI patterns and the combined approach when compared to T2 HR MRI based assessment alone. These differences were more marked with higher experience of the radiologist. Specificity was very high (>95 %) for all the three approaches. Our results are supported by the results of several previous studies, which showed similar improvements in the diagnostic accuracy for determining treatment response following NACRT using DWI [16,[18], [19], [20], [21], [22], [23]]. A recent work on pattern based approach of assessing response to NACRT using a combination of T2 HR MRI and DWI patterns was shown to have a sensitivity, specificity, PPV, NNV and accuracy of 94 %, 77 %, 88 %, 87 % and 88 % respectively, which is similar to our results [13]. A simplified 3-point DWI incorporated mrTRG by Lee MA et al. showed accuracy of 70 % [9]. We found moderate (0.402) and substantial agreement (k = 0.688) between the two observers for pattern based assessment using DWI and the combined approach similar to the other study which had substantial agreement (k = 0.77) using DWI patterns [9].
It was observed that both readers called complete response (MR-TRG 1) and no response (MR-TRG 5) with high levels of confidence using T2 HR MRI morphology. But the confidence level for MR-TRG 2 or 3 assessments were low. This explains the low sensitivity and the tendency for over staging using T2 HR MRI morphology alone. Radiologists had better levels of confidence for determining the response with a combined approach, especially in MR TRG1 to 3 scenarios. This could be due to the complementary nature of T2 HR and DWI. While it is easier to locate the site of treated rectal tumour on T2 HR MRI, DWI is capable of identifying tumour positive fibrosis from tumour negative fibrosis. Moreover, interpreting these together would help with better assessment of tumor when there is fluid in the lumen, mucin reaction and bowel wall oedema, which are the most common causes of incorrect assessments and lack of radiologist’s confidence while interpreting restaging MRI.
We did not use bowel preparation or spasmolytics. However, considering the total number of patients studied, only 13/268 patients (4.8 %) who had poor quality DWI had to be excluded. We used both 1.5 T and 3.0 T magnets from different vendors and used B-800 for high B-value DWI. Though we did not separately study the effect of different equipments and magnetic strengths on our results, in real practice these factors had no effect on the image interpretation. Our results emphasise the usefulness of incorporating DWI as a part of the standard imaging protocol as also recommended by recent expert consensus guidelines of European society of gastrointestinal and abdominal radiology (ESGAR) [8].
This study has a few limitations related to its retrospective nature. Post NACRT T2 HR images, DWI and their combination was reviewed at the same sitting and this may have introduced bias and may have positively contributed to the interpretation using the combined approach. The experience of the two observers was unequal. The value of experience and training in the interpretation of post NACRT MRI was clearly seen in the results of our study. While a large multi-center study is justified to prove the usefulness of the approach described by us, our results reflect the real life situation where radiologists of varying experience are likely to be interpreting MRI studies following NACRT.
5. Conclusion
There is significant association and excellent agreement between pathological and MRI response assessment using DWI patterns and its combination with T2HR MRI based MR-TRG. Combination of DWI patterns and T2 HR MRI based MR-TRG improves diagnostic performance of MRI for predicting complete pathological response. It also improves interobserver agreement and the level of confidence of the interpreting radiologists.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
CRediT authorship contribution statement
Anuradha Chandramohan: Conceptualization, Methodology, Validation, Investigation, Resources, Data curation, Writing - original draft, Writing - review & editing, Visualization, Supervision, Project administration. Umar M. Siddiqi: Methodology, Investigation, Resources, Data curation, Writing - original draft, Writing - review & editing, Visualization, Supervision, Project administration. Rohin Mittal: Conceptualization, Methodology, Validation, Investigation, Resources, Data curation, Writing - original draft, Writing - review & editing, Visualization, Supervision, Project administration. Anu Eapen: Methodology, Validation, Investigation, Resources, Data curation, Writing - review & editing, Project administration. Mark R. Jesudason: Methodology, Validation, Investigation, Resources, Writing - review & editing, Project administration. Thomas S. Ram: Methodology, Validation, Investigation, Resources, Writing - review & editing, Project administration. Ashish Singh: Methodology, Investigation, Resources, Writing - review & editing, Project administration. Dipti Masih: Methodology, Validation, Investigation, Resources, Writing - review & editing, Project administration.
Declaration of Competing Interest
None of the authors have any conflict of interests.
Contributor Information
Anuradha Chandramohan, Email: anuradhachandramohan@gmail.com.
Umar M. Siddiqi, Email: umarmukhtarsiddiqi@hotmail.com.
Rohin Mittal, Email: rohinmittal@gmail.com.
Anu Eapen, Email: anuepn@yahoo.com.
Mark R. Jesudason, Email: ranjanbernice@cmcvellore.ac.in.
Thomas S. Ram, Email: tomsamram@cmcvellore.ac.in.
Ashish Singh, Email: todrashish@gmail.com.
Dipti Masih, Email: dlondz@gmail.com.
References
- 1.Habr-Gama A., Perez R.O., Nadalin W., Sabbaga J., Ribeiro U., Silva e Sousa A.H. Operative versus nonoperative treatment for stage 0 distal rectal cancer following chemoradiation therapy: long-term results. Ann. Surg. 2004;240(October (4)):711–717. doi: 10.1097/01.sla.0000141194.27992.32. discussion 717-718. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Hartley A., Ho K.F., McConkey C., Geh J.I. Pathological complete response following pre-operative chemoradiotherapy in rectal cancer: analysis of phase II/III trials. Br. J. Radiol. 2005;78(October (934)):934–938. doi: 10.1259/bjr/86650067. [DOI] [PubMed] [Google Scholar]
- 3.van der Valk M.J.M., Hilling D.E., Bastiaannet E., Meershoek-Klein Kranenbarg E., Beets G.L., Figueiredo N.L. Long-term outcomes of clinical complete responders after neoadjuvant treatment for rectal cancer in the International Watch & Wait Database (IWWD): an international multicentre registry study. Lancet Lond. Engl. 2018;391(10139):2537–2545. doi: 10.1016/S0140-6736(18)31078-X. [DOI] [PubMed] [Google Scholar]
- 4.2017 European Society of Coloproctology (ESCP) collaborating group Evaluating the incidence of pathological complete response in current international rectal cancer practice: the barriers to widespread safe deferral of surgery. Colorectal Dis. Off. J. Assoc. Coloproctol. G B Irel. 2018;20(Suppl. 6):58–68. doi: 10.1111/codi.14361. [DOI] [PubMed] [Google Scholar]
- 5.Patel U.B., Taylor F., Blomqvist L., George C., Evans H., Tekkis P. Magnetic resonance imaging–detected tumor response for locally advanced rectal cancer predicts survival outcomes: MERCURY experience. J. Clin. Oncol. 2011;29(August (28)):3753–3760. doi: 10.1200/JCO.2011.34.9068. [DOI] [PubMed] [Google Scholar]
- 6.Bhoday J., Smith F., Siddiqui M.R., Balyasnikova S., Swift R.I., Perez R. Magnetic resonance tumor regression grade and residual mucosal abnormality as predictors for pathological complete response in rectal cancer postneoadjuvant chemoradiotherapy. Dis. Colon Rectum. 2016;59(October (10)):925–933. doi: 10.1097/DCR.0000000000000667. [DOI] [PubMed] [Google Scholar]
- 7.Patel U.B., Blomqvist L.K., Taylor F., George C., Guthrie A., Bees N. MRI after treatment of locally advanced rectal cancer: how to report tumor response—the MERCURY experience. Am. J. Roentgenol. 2012;199(October (4)):W486–95. doi: 10.2214/AJR.11.8210. [DOI] [PubMed] [Google Scholar]
- 8.Beets-Tan R.G.H., Lambregts D.M.J., Maas M., Bipat S., Barbaro B., Curvo-Semedo L. Magnetic resonance imaging for clinical management of rectal cancer: updated recommendations from the 2016 European Society of Gastrointestinal and Abdominal Radiology (ESGAR) consensus meeting. Eur. Radiol. 2018;28(4):1465–1475. doi: 10.1007/s00330-017-5026-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lee M.A., Cho S.H., Seo A.N., Kim H.J., Shin K.-M., Kim S.H. Modified 3-point MRI-based tumor regression grade incorporating DWI for locally advanced rectal cancer. Am J Roentgenol. 2017;209(October (6)):1247–1255. doi: 10.2214/AJR.16.17242. [DOI] [PubMed] [Google Scholar]
- 10.Schurink N.W., Lambregts D.M.J., Beets-Tan R.G.H. Diffusion-weighted imaging in rectal cancer: current applications and future perspectives. Br. J. Radiol. 2019;92(April (1096)) doi: 10.1259/bjr.20180655. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sclafani F., Brown G., Cunningham D., Wotherspoon A., Mendes L.S.T., Balyasnikova S. Comparison between MRI and pathology in the assessment of tumour regression grade in rectal cancer. Br. J. Cancer. 2017;117(November (10)):1478–1485. doi: 10.1038/bjc.2017.320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.van der Paardt M.P., Zagers M.B., Beets-Tan R.G.H., Stoker J., Bipat S. Patients who undergo preoperative chemoradiotherapy for locally advanced rectal cancer restaged by using diagnostic MR imaging: a systematic review and meta-analysis. Radiology. 2013;269(October (1)):101–112. doi: 10.1148/radiol.13122833. [DOI] [PubMed] [Google Scholar]
- 13.Lambregts D.M.J., Delli Pizzi A., Lahaye M.J., van Griethuysen J.J.M., Maas M., Beets G.L. A pattern-based approach combining tumor morphology on MRI with distinct signal patterns on diffusion-weighted imaging to assess response of rectal tumors after chemoradiotherapy. Dis. Colon Rectum. 2018;61(March (3)):328–337. doi: 10.1097/DCR.0000000000000915. [DOI] [PubMed] [Google Scholar]
- 14.van der Sande M.E., Beets G.L., Hupkens B.J.P., Breukink S.O., Melenhorst J., Bakers F.C.H. Response assessment after (chemo)radiotherapy for rectal cancer: why are we missing complete responses with MRI and endoscopy? Eur. J. Surg. Oncol. 2019;45(June (6)):1011–1017. doi: 10.1016/j.ejso.2018.11.019. [DOI] [PubMed] [Google Scholar]
- 15.Patel U.B., Brown G., Rutten H., West N., Sebag-Montefiore D., Glynne-Jones R. Comparison of magnetic resonance imaging and histopathological response to chemoradiotherapy in locally advanced rectal cancer. Ann. Surg. Oncol. 2012;19(September (9)):2842–2852. doi: 10.1245/s10434-012-2309-3. [DOI] [PubMed] [Google Scholar]
- 16.Sathyakumar K., Chandramohan A., Masih D., Jesudasan M.R., Pulimood A., Eapen A. Best MRI predictors of complete response to neoadjuvant chemoradiation in locally advanced rectal cancer. Br. J. Radiol. 2016;89(February (1060)) doi: 10.1259/bjr.20150328. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mandard A.M., Dalibard F., Mandard J.C. Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma. Clinicopathologic correlations. Cancer. 1994;73(June (11)):2680–2686. doi: 10.1002/1097-0142(19940601)73:11<2680::aid-cncr2820731105>3.0.co;2-c. [DOI] [PubMed] [Google Scholar]
- 18.Kim S.H., Lee J.M., Hong S.H., Kim G.H., Lee J.Y., Han J.K. Locally advanced rectal cancer: added value of diffusion-weighted MR imaging in the evaluation of tumor response to neoadjuvant chemo- and radiation therapy. Radiology. 2009;253(October (1)):116–125. doi: 10.1148/radiol.2532090027. [DOI] [PubMed] [Google Scholar]
- 19.Lambregts D.M.J., Vandecaveye V., Barbaro B., Bakers F.C.H., Lambrecht M., Maas M. Diffusion-weighted MRI for selection of complete responders after chemoradiation for locally advanced rectal cancer: a multicenter study. Ann. Surg. Oncol. 2011;18(August (8)):2224–2231. doi: 10.1245/s10434-011-1607-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Park M.J., Kim S.H., Lee S.J., Jang K.M., Rhim H. Locally advanced rectal cancer: added value of diffusion-weighted MR imaging for predicting tumor clearance of the mesorectal fascia after neoadjuvant chemotherapy and radiation therapy. Radiology. 2011;260(September (3)):771–780. doi: 10.1148/radiol.11102135. [DOI] [PubMed] [Google Scholar]
- 21.Sassen S., de Booij M., Sosef M., Berendsen R., Lammering G., Clarijs R. Locally advanced rectal cancer: is diffusion weighted MRI helpful for the identification of complete responders (ypT0N0) after neoadjuvant chemoradiation therapy? Eur. Radiol. 2013;23(December (12)):3440–3449. doi: 10.1007/s00330-013-2956-1. [DOI] [PubMed] [Google Scholar]
- 22.Song I., Kim S.H., Lee S.J., Choi J.Y., Kim M.J., Rhim H. Value of diffusion-weighted imaging in the detection of viable tumour after neoadjuvant chemoradiation therapy in patients with locally advanced rectal cancer: comparison with T2 weighted and PET/CT imaging. Br. J. Radiol. 2012;85(May (1013)):577–586. doi: 10.1259/bjr/68424021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Napoletano M., Mazzucca D., Prosperi E., Aisa M.C., Lupattelli M., Aristei C. Locally advanced rectal cancer: qualitative and quantitative evaluation of diffusion-weighted magnetic resonance imaging in restaging after neoadjuvant chemo-radiotherapy. Abdom Radiol N Y. 2019;19(April) doi: 10.1007/s00261-019-02012-4. [DOI] [PubMed] [Google Scholar]