Endoscopy and MRI for restaging early rectal cancer after neoadjuvant treatment

Chemoradiotherapy (CRT) has great potential to downstage rectal cancer. Response assessment has been investigated in locally advanced rectal cancer but not in early stage rectal cancer. The aim is to characterize the diagnostic accuracy of endoscopy performed by surgical endoscopists compared to (diffusion‐weighted, DWI) MRI only and a multimodal approach combining (DWI‐)MRI and endoscopic information both analysed by an abdominal radiologist for response assessment in early rectal cancer after neoadjuvant CRT.


INTRODUC TI ON
Neoadjuvant chemoradiotherapy (CRT) in rectal cancer treatment has been proved to significantly downstage locally advanced rectal cancer (LARC) [1,2]. The downstaging effect varies between patients, with the potential to lead to a pathological complete response (pCR; ypT0N0) in 15%-25% of LARC as observed in total mesorectal excision (TME) specimens [3]. Although oncological outcomes are good, TME surgery causes substantial morbidity and occasionally mortality [4][5][6]. These findings have led to the exploration of organpreserving approaches in good or complete response patients aiming to reduce the morbidity of conventional rectal cancer treatment whilst maintaining oncological outcome and quality of life.
A watch-and-wait strategy instead of TME surgery in this specific group has been the subject of many studies [7,8]. Also, resection of the residual tumour mass after chemoradiation can be performed safely with local excision techniques, such as transanal excision, transanal endoscopic microsurgery (TEM) or transanal minimally invasive surgery [9][10][11]. Long-term follow-up with close clinical follow-up reveals that this strategy is also a safe alternative for TME [8,[12][13][14][15][16].
The key to selecting patients for an organ-preserving approach is a dedicated diagnostic work-up with high accuracy. For restaging rectal cancer after CRT, digital rectal examination (DRE) and endoscopy are essential [17,18]. However, these examinations only provide information on the (intra)-luminal presence of the tumour.
T2-weighted (T2W) MRI is the superior modality for (re)staging rectal cancer, providing the best anatomical relationship of the tumour to key surgical landmarks [19,20]. Together with diffusion-weighted imaging (DWI) a strong indication can be made whether there is remaining residual tumour and/or lymph node involvement. The combination of DRE, endoscopy and (DWI-)MRI has provided the most accurate assessment of the response in LARC [21].
Most publications on the accuracy of restaging rectal cancer after neoadjuvant CRT have focused predominantly on LARC treated with neoadjuvant chemoradiation. Nowadays, organ-preserving treatment is studied for less advanced stages such as early stage (cT1-3 N0) rectal cancer. So far, the performance of (DWI-)MRI and its additional value to endoscopy for the assessment of good and completely responding early rectal cancer patients is unknown and therefore the subject of the present study.

MATERIAL S AND ME THODS
This study was approved by the institutional review board of the participating hospitals. For inclusion in the current study, endoscopic images and/or (DWI-)MRI images before and after neoadjuvant CRT had to be available for reassessment. Patients who participated in the CARTS study (N = 55; registered at clini caltr ials.gov; NCT01273051) [15,22] and patients treated with a similar organ-sparing regime at the Laurentius Hospital (N = 22; a tertiary referral centre for local excision) in the period after the CARTS study were screened for eligibility.
Based on the availability of the clinical data of these 77 patients, three reassessment cohorts could be formed: (1) reassessment of endoscopy alone, (2) reassessment of (DWI-)MRI data alone and (3) a multimodal reassessment combining (DWI-)MRI with endoscopic information.
Informed consent for re-evaluation of the clinical examinations was obtained at initial inclusion in the CARTS study. Patients included at the Laurentius Hospital provided their written informed consent separately.

Patients
The selected patients for reassessment in this study were treated in nine Dutch referral centres for rectal cancer treatment with an expertise in TEM surgery. Patients with a histologically proven rectal cancer, staged cT1-3 N0 and located within the distal 10 cm of the rectum were evaluated for inclusion. All patients were treated between December 2010 and June 2017 and were scheduled for an organ-preserving treatment consisting of neoadjuvant CRT with a prolonged interval (≥5 weeks) to surgery. Patients were included if initial as well as restaging clinical endoscopy images and/or reports and MRI examinations were available. Exclusion criteria were (1) no availability of clinical investigations, (2) no TEM resection, (3) no DWI at restaging MRI, (4) use of endorectal gel as MRI intra-luminal contrast, (5) insufficient image quality or (6) presence of poor prognostic factors such as tumour budding, extramural venous invasion or lymphovascular invasion. Patients were analysed prospectively with standard investigations according to the guidelines for rectal cancer treatment [23].

Treatment regimen
Chemoradiotherapy consisted of a total dose of 50 or 50.4 Gy given in 25 fractions of 2 Gy or in 28 fractions of 1.8 Gy, respectively, with concomitant capecitabine 825 mg/m 2 twice daily on all days.
Restaging examinations were performed with a preferred interval of 6-16 weeks after completion of (C)RT. Residual tumour or scar tissue was removed by performing a TEM procedure with a preferred maximum interval of 8 weeks after restaging.

Clinical reassessment
In this study, the main outcome was the predictive value of (1) endoscopy alone, (2) (DWI-)MRI alone and (3) a multimodal approach for What does this paper add to the literature?
This study reveals that a multimodality approach using endoscopy and (DWI) -MRI is most accurate in the assessment of response after neoadjuvant chemoradiotherapy in early rectal tumours. The optimal identification of good and complete responders, will aid in the patient selection for organ-preserving treatment. response assessment after neoadjuvant CRT. The predictive value of DRE was not part of the analysis. The endoscopic and MRI readers selected for reassessment were not the patients' treating clinicians and thus DRE was not incorporated in the reassessment. For endoscopic, (DWI-)MRI reassessment and/or multimodal reassessment, availability of pre-and post-CRT images was essential. For the endoscopic reassessment, the images of the restaging endoscopy had to be available from the treating hospitals. For the MRI only cohort as well as for the multimodal reassessment cohort, at least the MRI images had to be available from the treating hospitals. For the multimodal approach, endoscopic images or endoscopic reports had to be available, so non-availability of endoscopic images was not an exclusion criterion for this analysis. In the case of incomplete data, patients were excluded from (one of the) cohort analyses. As a result, some patients could be included in either one reassessment cohort or in the case of all available data in all reassessment cohorts (endoscopy alone, (DWI-)MRI alone and multimodal reassessment). Abdominal Radiology guidelines, was used for revision of available primary and restaging MR datasets [24]. The re-evaluation focused on clinical T and N staging, tumour size diameter, location and morphological characteristics as well as extramural depth. On restaging (DWI-)MRI, the percentage of tumour size regression, absence or presence of residual tumour and/or fibrosis were evaluated, using five point confidence level scores.

Clinical assessment: multimodal reassessment using endoscopy and (DWI-)MRI
Directly after the MRI response reporting session the restaging endoscopy reports of the local hospitals including endoscopic images were shown to the expert abdominal radiologist while the (DWI-) MRI restaging images were still available to evaluate the additional value of (DWI-)MRI to endoscopy. By doing so, the radiologist was able to provide a multimodal response assessment by combining (DWI-)MRI with endoscopic information focusing on absence or presence of residual tumour using five point confidence level scores.

Histopathology
Histopathology of the TEM resection specimens was used as reference standard for the clinical reassessments. TEM specimens were evaluated by colorectal pathologists from participating centres according to the method described by Quirke et al. [25]. A tumour regression grading scale was not consistently used. Since only TEM resections were available for histological evaluation, only T stage was used as a reference for response evaluation. A ypT0 tumour was considered a complete response, any other ypT stages as residual tumour.

Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics  Table 1, in which the assessment groups are separately described. All patients received neoadjuvant CRT prior to TEM surgery. Histopathological findings of the TEM specimens are presented in Table 2.
Endoscopy images available for 36 patients were taken after a median interval of 6 (range 4-10) weeks after the final dose of (C)RT.
TEM resection in this cohort was performed after a median 3 (range 1-9) weeks after endoscopy. Table 3 shows the parameters scored during endoscopic re-evaluation of the five surgeons. Agreement for scoring the endoscopic features between all readers was moderate for elevated residual tumour (κ = 0.551), fair for residual ulcer F I G U R E 1 Flowchart of the included patients. The 77 potentially eligible patients consisted of 55 patients of the CARTS study and 22 patients from the Laurentius Hospital who were treated with a similar treatment scheme. Of the 77 potentially eligible patients, 36 patients had available endoscopic images pre-and post-CRT which could be used for endoscopic re-evaluation (left column). Twenty five patients had available (DWI-)MRI images for MRI response assessment (middle column). For the multimodal response assessment, 25 patients had available (DWI-)MRI images as well as available endoscopic data (right column). This led to two patient cohorts for three separate reassessment cohorts. Reasons for exclusion for the endoscopic cohort are mentioned in the above left box; reasons for exclusion for the MRI only cohort and the multimodal cohort are mentioned in the above right box. presence of risk factors a completion TME was recommended. Four patients with a ypT2 tumour underwent completion TME without evidence of residual tumour. Eleven patients refused completion TME.
Based on pathological staging, 67% of the patients (30 of the 44 included patients with a ypT0-1 after TEM) were successfully treated with the organ-preserving approach.

DISCUSS ION AND CON CLUS IONS
In the present study, both a single and multimodality approach were Literature on response assessment in rectal cancer patients after CRT using endoscopy alone or endoscopy combined with other modalities is scarce and focuses predominantly on LARC after chemoradiation. For restaging LARC patients, the most reliable information on response evaluation is obtained by combining (DWI-)MRI, endoscopy and DRE with a diagnostic AUC of 0.89 [21]. Recent studies showed similar diagnostic performances for the assessment of pCR in LARC patients with a diagnostic accuracy of 90% when combining both modalities [26,27]. When comparing these accuracies with the current study outcomes, these good diagnostic outcomes can be translated to tumours of an earlier stage which are treated with similar neoadjuvant treatment regimens.
Endoscopic assessment by five surgical endoscopists after CRT resulted in mean sensitivity and specificity rates of 54.5% and 79.1%, respectively. Comparing with (DWI-)MRI assessed by an expert radiologist, endoscopic evaluation by experienced surgeons performs equally well [28]. Endoscopic information is essential for detection of mucosal abnormalities, which can vary widely after CRT. White scar tissue (with or without telangiectasia) or normalized mucosa after irradiation has been suggested to represent a complete response. Other residual superficial ulceration or mucosal irregularity must alarm the observer for potentially viable tumour, but can still be a good responder with a tumour in remission [29,30]. Moreover, according to the current study there   Whether these residual lesions are tumours in remission or residual disease that will progress to recurrent disease remains to be seen on long-term follow-up.

RO
An adequate interval between neoadjuvant treatment and restaging the disease plays an important role in determining adequate treatment.
It has been shown that pCR rates increase when exceeding the classical interval of 6-8 weeks after neoadjuvant treatment, without compromising surgical or oncological outcomes [32,33]. Translating this to multimodality reassessment, residual mucosal abnormalities 6-8 weeks after CRT may still evolve and eventually turn into a complete response [34]. Therefore, when restaging early tumours after neoadjuvant treatment, one should not focus on a single observation but consider clinical findings in relation to initial staging and its time interval. formation about the tumour, but probably also causing discomfort and some loss of function [15,16]. Previous rectal preserving studies demonstrated good oncological outcomes after local excision, but functional outcomes might be better when local surgery can be avoided [8,37]. Therefore, time can also be a good diagnostic tool, and good responders can be followed up and reassessed at a later stage. The non-responding tumours can usually be identified in time for TME surgery. This suggests that, when aiming for a rectal preserving strategy in early rectal cancer patients, a restaging protocol with standard evaluation interval after 6-8 weeks post-neoadjuvant treatment can also be performed merely to distinguish good from bad responders. When a good responder or a potential complete responder is identified, no direct (local) intervention is needed and further response can be awaited [38]. A second restaging moment at least 6 weeks later may then be used to definitely opt for either watch-and-wait, local excision or completion TME which should be endorsed by an expert multidisciplinary team [39].
The current study has some limitations. It was a challenge to collect available clinical examinations from patients treated with this organ-preserving treatment due to the long inclusion period.
Fortunately, two overlapping cohorts were available for the evaluation of response assessment. The external validity of the study needs confirmation: the (DWI-)MRI images were evaluated by an expert in the field of restaging rectal cancer. It is unclear whether the performance of these evaluations can be extrapolated to any other radiologist who works in the field of abdominal or colorectal imaging. For endoscopic evaluation, not all primary endoscopy images were available, which made it more difficult to evaluate the response after neoadjuvant treatment. The endoscopic data were assessed by surgeons looking at images of lesions instead of in real time which would be more accurate. As with the radiological evaluation, the external validity is to be confirmed.
To conclude, the current study shows that a multimodality approach available information to identify good and complete responders after irradiation of early stage rectal cancer. The addition of (DWI-)MRI to endoscopy was especially helpful to improve the selection of patients who are potentially eligible for organ-preserving treatment.

AUTH O R CO NTR I B UTI O N S
All authors contributed to this paper with either conception and design of the study, literature review and analysis, drafting and critical revision and editing, or final approval of the final version.

FU N D I N G I N FO R M ATI O N
No funding received for this study.

CO N FLI C T O F I NTE R E S T
No conflicts of interest.

E TH I C S A PPROVA L
The study was approved by the medical ethical committee of the Radboud Medical Centre Nijmegen, The Netherlands. Permission to reproduce material from other sources.

I N FO R M ED CO N S ENT
All patients provided written informed consent.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author.