MRI of early rectal cancer; bisacodyl micro-enema increases submucosal width, reader confidence, and tumor conspicuity

Local excision (LE) of early rectal cancer (ERC) offers a definitive treatment option using organ-sparing techniques, avoiding the disadvantages of major surgery such as genito-urinary dysfunction, low anterior rectal resection syndrome, or permanent stoma after abdomino-perineal excision.

Still, early stage cancers have a risk for lymph node metastasis reported to increase with the depth of tumors submucosal invasion [1]. Newer research, however, points out high-risk factors for lymph node metastasis as poor differentiation, lymphovascular invasion, and high-grade tumor budding more important than deep submucosal invasion [2, 3]. In contrast to the high-risk factors found in histopathology after LE, the depth of submucosal tumor invasion can be determined preoperatively, and the division of submucosa into thirds, T1sm1-T1sm3, as described by Kikuchi et al. [1], is still important for the preoperative selection of candidates for LE [4]. Endorectal ultrasound (ERUS), evaluating the depth of submucosal tumor invasion, is recommended by several guidelines [4‐6] but is hampered by interobserver variations [7], more recently though, the supplement of ERUS elastography has shown high interreader agreement [8]. Tumor morphology, as presented at endoscopy and stratified into the Paris classification or classification for lateral spreading tumors (LST), has been used to predict the risk for malignancy but has shown only moderate agreement among experts [9, 10]. Higher diagnostic performance has been achieved with endoscopic optical analysis using Narrow-band Imaging (NBI), evaluating micro-vessel and surface patterns as in the NICE (NBI International Colorectal Endoscopic) classification and the JNET (Japan NBI Expert Team) classification [11]. All these techniques are operator-dependent and may have limited availability.

There has been an international agreement that LE is usually curative if tumor invasion does not exceed the superficial third of the submucosa (Kikuchi T1sm1) [4, 12], while major resection is recommended for tumors invading the outer third of the (Kikuchi T1sm3) due to a 15–20% prevalence of lymph node metastases. The opinions and guidelines vary regarding the treatment of T1sm2 cancers and treatment options after LE if histopathology of the resected specimen shows high-risk factors for nodal metastasis [13].

MRI reports often do not provide sufficient information for ERC treatment decisions [14] and tend to overestimate tumor invasion in early stages, representing a severe weakness as overstaging benign or early invasive tumors can lead to unnecessary major resection or permanent stoma [15, 16]. With the increasing incidence of ERC due to the introduction of screening programs [17], it is increasingly important for radiologists to recognize benign and early-stage tumors. MRI, however, is assumed to be unable to differentiate early rectal cancers as the submucosa is often incompletely displayed on MRI [18]. Finding a method to enhance the visualization of submucosa on MRI could improve the staging of ERC.

Bisacodyl is a stimulating laxative approved for treating constipation and cleansing the bowel before surgical or radiological procedures [19]. We routinely apply a bisacodyl micro-enema before rectal MRI examinations and have observed that the enema generates a widening of the submucosa due to submucosal edema Fig. 1.

In this study, we wanted to investigate if a bisacodyl micro-enema can increase submucosal width in the tumor-bearing wall, improve readers' confidence and tumor conspicuity, and finally influence the diagnostic performance of MRI to identify ERC suitable for LE.

The presented data in this study follow the STARD 2015 guidelines for reporting diagnostic accuracy studies [20].

In total, 50 participants underwent both MRin and MRex and were included in the study. The median time interval between the examinations was 18 (2–41) days. Except for one participant, the time interval between the interpretation of MRex and MRin was at least five weeks.

In total, 70% (35/50) were local tumors, of which 89% (31/35) were Tis and 11% (4/35) were T1sm1-sm2. Participant and tumor characteristics are shown in Table 1.

In this study, we used bisacodyl micro-enema as a preparation before MRI of ERC. The micro-enema led to a significant widening of the submucosa in the tumor-bearing wall, increased readers' confidence and interreader agreement in T-staging, tumor conspicuity and morphology. Finally, this resulted in a slight but insignificant improvement in diagnostic performance for the reader with less experience in MRI of ERC. However, when combining diagnostic performance with reader confidence, the proportions of tumors staged with both high confidence and correct T-stage increased significantly for both readers.

The use of rectal filling with fluid or gas to enhance tumor visualization and staging has been investigated by several authors [26‐29]. There has been a concern that using rectal filling media could lead to distension of the rectum and reduced distance between the tumor and mesorectal fascia (MRF), thereby causing unnecessary preoperative therapy [27]. However, neither glycerin enema nor gel was found to reduce the distance between the tumors and MRF [28, 29]. Rectal filling with water or glycerin enema improved the visualization and T-staging of rectal tumors [26, 28], whereas the use of gel tended to overstaging, especially in early tumors [29]. These partly conflicting results may be related to differences in the medium and volume used for rectal filling.

More recently, rectal micro-enemas have been advocated in the response evaluation after chemoradiotherapy (CRT) [30, 31] as they remove bowel gas and decrease susceptibility artifacts, thereby improving diffusion-weighted imaging (DWI). Still, there is no consensus on using pre-procedure preparations as micro-enemas or rectal filling, and they are not routinely advised by the ESGAR recommendation [5].

Bisacodyl micro-enema, used in this study, is a prodrug metabolized in the intestinal mucosa and has a dual action by stimulating bowel motility and fluid secretion [19]. We report on a third effect, the induction of submucosal edema, a finding already registered by Ciggaar et al. investigating MRI of endometriosis [32].

In our study, the fluid secretion induced by the micro-enema significantly increased the number of tumors surrounded by fluid, enhancing tumor conspicuity. In line with Coskun et al. [33], we found no increased distension of the rectum at tumor height, diminishing the risk for compression of the mesorectum and, subsequently, overtreatment of tumors due to the shortening of the distance to MRF.

The main aim of the study was to evaluate if the widening of the submucosa could improve the T-staging of ERC by better displaying the layers of the rectal wall and improving the depiction of the tumor's invasive front. Although both readers measured a significant increase in the width of the submucosa in the tumor-bearing bowel wall on MRin, there was no improvement in the diagnostic performance for Reader1. Reader2 achieved a slight but statistically insignificant improvement. Nevertheless, for Reader2, this may imply a clinical effect as the number of correct staged tumors increased from 78% (39/50) to 88% (44/50), and a significant increase in diagnostic performance could potentially have been achieved with a larger study selection. In addition, the interreader agreement for T-stage, tumor conspicuity and tumor morphology improved on MRin, indicating better visualization of the tumor and factors influencing T-staging.

MRI is usually not performed on benign to early stage tumors, as it is hampered by overstaging [4, 15]. Overtreatment, though, is still a problem [16, 34] and as screening is bringing a shift to earlier stages, it is essential to raise the focus on early stages and MRI, as some of these patients probably will undergo an MRI examination. To be able to record the effect of a micro-enema on overstaging, we aimed for a study population containing a specter of tumors from benign and early stages to invasive cancers ≤ mrT3b. Achieving the optimal study population is demanding, and to obviate the small clear-cut benign tumors, we only included polyps larger than 15 mm. Finally, the consecutive inclusion resulted in a slight overweight of benign tumors.

The diagnostic performance and rate of misclassification achieved by the readers on MRin were in line with comparable ERUS studies investigating similar T-stage selections, with sensitivities ranging from 71–95%, specificities 62–100%, and overstaging 5–27% [35‐40]. Misclassification of ERC is problematic, especially for overstaging, as overstaging can lead to overtreatment and an unnecessary rectal amputation. Both readers overstaged some tumors; however, for Reader2, the micro-enema reduced the number of overstaged tumors from nine to four on MRin. The number of understaged tumors was low for both readers and was not influenced by the micro-enema.

One possible reason for the micro-enema's low impact on overall diagnostic performance could be the evaluation method used. The readers were asked to select a T-stage without an alternative for indeterminate lesions and their uncertainty regarding T-stage, may not have been depicted.

Another reason for the absence of a significant improvement in diagnostic performance may be that both readers were experienced in interpreting MRIs of locally advanced rectal cancers, as the use of inexperienced readers may better demonstrate the effect of new technologies [41]. This applies especially to Reader1, who was also experienced in ERC and had no effect of the micro-enema on diagnostic performance in the study.

The reader's confidence was measured separately, and the micro-enema significantly increased the reader's true confidence for T-staging, and the proportions of tumors staged correctly with high confidence increased for both readers. In addition, the readers confidence for tumor conspicuity and tumor morphology raised from fair to moderate or good. Reader's confidence is essential as it can impact several factors, such as reading time, clarity of reports, and diagnostic performance [42]. Clinicians may also be more inclined to act on radiology reports expressed with higher confidence [43]. On the other hand, several conditions may influence readers' confidence, such as experience, image quality, clinical information, and CAD- assistance, and several studies have shown improvement in reader confidence by refinement of imaging techniques [44‐47]. As a reader's confidence is crucial in communicating radiology reports, a higher confidence achieved after a bisacodyl micro-enema may improve the quality of the reports and increase the impact on clinical practice.

We found a connection between specific tumor morphologies and tumor histology, with semiannular morphology indicating non-local tumors and sessile morphology local tumors. This suggests that tumor morphology, as shown on MRI, could play a role in tumor staging. Furthermore, the micro-enema increased the number of tumors with sessile morphology proven as local at histopathology, especially for Reader2, meaning that morphology after a micro-enema potentially can contribute to less overstaging.

The readers noted some tumors as pedunculated, but considering the stalks being only 3–6 mm long, measured on MRin in retrospect, the polyps rather qualify as subpedunculated, which may explain the lack of agreement between the readers.

A strength of the study is the use of histopathology as the gold standard and the prospective inclusion with paired participant and reader design. This reduces the number of confounding factors and raises the level of evidence for comparison.

In this study, we investigated the reader's confidence in T-staging and visual evaluation of tumor conspicuity, which are all subjective parameters difficult to reproduce among readers. To minimize disagreement, we tried to align the Reader's evaluation by defining criteria for visual scoring and using an fVAS with absolute minimum and maximum scores and two reference points in between.

Our study has some limitations. Twenty-three participants refused MRex, which may have led to a bias, but as we believe the reasons were random, we consider the bias minimal. Further, we did not achieve the required sample size, as only 50 participants admitted undergoing a second MRI examination and we included two readers only, a number lower than recommended for multiple-reader studies [41]. Both conditions may limit the statistical power to detect clinically significant differences.

Furthermore, because of the apparent signs of a rectal micro-enema, such as a clean bowel, submucosal edema, and fluid in the lumen, it was impossible to hide the exposure to an enema, and the study cannot be regarded as truly blinded. However, we tried to minimize the reader's recall bias by assigning MRex and MRin to two reading sessions separated by a minimum interval of four weeks. In addition, within each reading session, MRin and MRex were presented in a random order.

Moreover, as both readers were experienced readers of MRI of rectal cancer, the study's outcome cannot be generalized to the overall population of radiologists but merely to radiologists working in specialist centers in close cooperation with referring clinicians and pathologists.

As bisacodyl micro-enema enhances the layers of the bowel wall, the method may help image various conditions affecting the rectum. Further research is needed to explore its potential in imaging endometriosis to assess the possibility of easier identification and grading of deep infiltration. In combination with DWI, the method may also be advantageous in response evaluation of rectal cancer after CRT to help identify complete responders or in follow-up for early detection of regrowth or detection of recurrency after LE of rectal tumors. The improved display of polyp morphology could facilitate the development of an MRI classification system for polyps and ERC with possible correlation to clinical classifications and histopathology.

In conclusion, the supplement of a rectal micro-enema induced a significant increase in the submucosal width in the tumor-bearing wall. The micro-enema also led to higher reader confidence and interreader agreement for diagnostic performance, tumor conspicuity and morphology, and tumor morphology as depicted on MRI may contribute to correct T-staging. Diagnostic performance alone was not improved; however, the micro-enema significantly raised the proportion of tumors correctly staged with high confidence, which in a clinical setting may lead to more precise reports and a greater impact on clinical practice.