D'Souza et al1 in the Annals paper “Definition of the rectum: An International, expert-based Delphi consensus” report how a radiological approach for distinguishing the rectum and sigmoid using the “sigmoid take-off” was selected as the best of 11 pragmatic definitions for addressing global variation in the definition of the rectum. Variation in the definition of the rectum has perpetuated inconsistencies in clinical practice, trial participation, and outcomes reporting and we agree that consistency in rectal cancer care would be significantly improved if there were an unambiguous, accurate, and reproducible method of defining the rectum. The authors propose that cross-sectional magnetic resonance imaging, MRI, can identify the junction of the sigmoid mesocolon with the mesorectum (ie, the “sigmoid take-off”) and this can be used to classify tumors into rectal, rectosigmoid, or sigmoid based on location relative to the take-off. Survey results from experts support their definitional approach. We congratulate the authors for calling the question on this complex problem and we agree that using MRI certainly could address reproducibility and accuracy. We are not yet convinced, however, that the use of the sigmoid take-off can eliminate ambiguity. To frame our discussion, we address 2 key questions: why are there major differences in the definition of the junction between the rectum and sigmoid; and what will it take to resolve these differences and create a universally accepted definition?
To consider the first question, it is helpful to review the key drivers behind the natural evolution of divergent definitions of the proximal extent of the rectum. First, over time, each discipline of cancer care providers has contributed their own perspective based on what they need to know and/or what they can know about the location of a rectal cancer (ie, distance to anal verge, identification of surgical and/or imaging landmarks). These perspectives and diverse landmarks do not share any fixed points of reference, nor have they been calibrated against a fixed standard. These landmarks have no common interfaces, have not been harmonized, and so they are not interoperable or interchangeable. An additional complication is that gastrointestinal biology is not typically associated with discrete and/or abrupt transitions, which explains why even the sigmoid take off requires classification into 3 parts, the rectum, rectosigmoid, and the sigmoid. Furthermore, the rectum also has more than one feature that defines its anatomy and functions and the purpose of the definition makes a difference as to what feature should be measured. For example, lymphatic anatomy is of great significance to cancer outcomes, whereas proximity to the anal sphincter and pelvic nerves significantly impacts bowel and sexual outcomes following surgery. Below, we discuss how MRI could, if properly validated, address reproducibility and accuracy. We also discuss why MRI alone will likely not resolve the challenges associated with definitional ambiguity unless multiple anatomic features of the rectum can be accommodated and diverse landmarks can be harmonized.
Our optimism that MRI can bring a new level of reproducibility and accuracy is based both on the theoretical assets of MRI technologies and on the evidence generated by the MERCURY studies.2–4 We suggest that pelvic MRI offers many unique advantages toward solving the complex problem of defining the rectum, including the fact that it is not operator dependent and the technology lends itself to sharing, storage, and archiving. As many providers as may wish to view patient images can readily do so, and can do so across disciplines, institutions, and across periods of treatment and post-treatment surveillance. For clinical trials, MRI images can be preserved and expected to retain high-fidelity information, thereby facilitating auditing and cross-trial comparisons. When MRI was tested in the MERCURY trials, for example, it performed extremely well. These trials demonstrated that high-resolution MRI accurately predicts surgical margin clearance when tested against the accepted standard of histopathology. Even more impressive, preoperative MRI assessment of circumferential margin status outperformed standard staging for assessing risk of cancer relapse and survival outcomes. There appeared to be no problems with interobserver reproducibility of MRI in the 11 participating European sites. Although the MERCURY trial performed well at staging using an existing endoscopic definition, that does not mean it will perform well at defining the rectum. To define the rectum in a way that resolves definitional ambiguity is an entirely different and significantly more difficult matter, as it will require reconciliation of the diverse perspectives of providers and incorporation of multiple anatomic features.
As previously mentioned, a single definition has likely not emerged over the century of rectal cancer treatment largely due to the lack of a fixed point of reference between the various disciplines and diagnostic modalities. In essence, there has been no way to harmonize endoscopic measurements against the surgical, radiographic, or anatomic landmarks. The endoscopist sees the intraluminal rectum, and although they can measure distances from a fixed point, such as anal verge, they cannot simultaneously identify other surrounding anatomic landmarks. Surgeons, on the contrary, can identify landmarks, such as sacral promontory, tenial confluence, and peritoneal reflection, but they cannot know how far they are from the anal verge. There are assumptions about the overlap of these definitions, but no data exist to calculate differences. We tried to reconcile these differences in the 1990s during the laparoscopic colectomy trial by conducting a small IRB-approved study (unpublished) of 8 consented subjects (4 female), each of whom was scheduled for surgery and none of whom were known to have anatomic distortion or prior pelvic surgery or radiation. At surgery, before the pelvic dissection commenced, a plastic rigid proctoscope with radio opaque markers was inserted and a cross-table lateral x-ray was obtained and endoscopic measurements were compared against surgical and image-based landmarks. Surprisingly, the promontory and tenial confluence were quite proximal, a mean of 23 and 21 cm; the sacroiliac joint (S3) and mesorectum were more distal, a mean of 18 and 17 cm; while the anterior peritoneal reflection was most distal, a mean of 11.5 cm from the anal verge, respectively. Data from this small study provided insights into the degree of variability between patients and demonstrated the wide disparities that exist between commonly cited endoscopic thresholds, 12 to 15 cm, and commonly cited surgical thresholds, sacral promontory and tenial confluence, both consistently beyond 20 cm from the anal verge. In the absence of interoperable landmarks, it made no sense to go forward and validate divergent definitions against clinical outcomes. The fact that MRI is rich with anatomic detail and it can provide measurements calibrated against uniform standards, supports that there could be a new path forward to reconcile these differences and finally get everyone at multidisciplinary team meetings on the same page.
If MRI could harmonize the various existing endoscopic and surgical definitions in a reproducible manner, it would go a long way toward addressing existing ambiguities. That said, tackling the hardest part of ambiguity will require the issue of definitional utility be addressed. A universal definition must meet the needs of all key stakeholders. We already know that different providers and patients need to, and wish to know about the anatomy of the rectum and relative tumor location for different purposes. Medical oncologists consider tumor location relative to the peritoneal reflection to estimate the risk of peritoneal disease. Radiation oncologists focus on extent and distribution of lymphatic involvement when considering pelvic radiation. Surgeons use rectal tumor location to plan surgical approaches. Patients are interested in functional outcomes that are influenced by distance to anal verge and proximity to pelvic nerves. Clinical trialists and epidemiologists want to know that patients have been correctly assigned to the colon or rectum so that reliable comparisons can be made across trials and over time. We submit that describing a single feature of the rectum is analogous to describing a single feature of an elephant. One cannot expect to know an elephant by the description of its trunk or tusks. Similarly, a universal definition of the rectum and its anatomic parts will likely only succeed if complexity can be accommodated rather than reduced. Everyone's needs should be part of a multidimensional definition.
In summation, we believe that the authors of this Annals paper are on to a new approach that deserves serious consideration and future dedicated research. Using cross-sectional MRI to map rectal anatomy toward a more universal, evidence-based definition is entirely possible. Validating a multidimensional definition that incorporates diverse features, including the “sigmoid take off,” against meaningful outcomes that satisfies all stakeholders will deliver on the promise of reproducibility, accuracy, and unambiguous clarity.
1. D'Souza N, de Neree Tot Babberich MPM, d’Hoore A, et al. Definition of the rectum: an international, expert-based Delphi consensus. Ann Surg 2019 [Epub ahead of print].
2. Battersby NJ, How P, Moran B, et al. Prospective validation of a low rectal cancer magnetic resonance imaging staging system and development of a local recurrence risk stratification model: the MERCURY II study. Ann Surg
3. Taylor FG, Quirke P, Heald RJ, et al. Preoperative high-resolution magnetic resonance imaging can identify good prognosis stage I, II, and III rectal cancer best managed by surgery alone: a prospective, multicenter, European study. Ann Surg
4. Taylor FG, Quirke P, Heald RJ, et al. Preoperative magnetic resonance imaging assessment of circumferential resection margin predicts disease-free survival and local recurrence: 5-year follow-up results of the MERCURY study. J Clin Oncol