Safety and Feasibility of Using Magnetic Resonance Imaging Criteria to Identify Patients With “Good Prognosis” Rectal Cancer Eligible for Primary Surgery

Key Points

Question

Can magnetic resonance imaging (MRI) be used to select patients with “good prognosis” rectal tumors for primary surgery and safely avoid preoperative chemoradiotherapy?

Findings

In this nonrandomized phase 2 study, 82 patients with MRI-predicted good prognosis rectal cancer were included from 12 hospitals across Canada and underwent primary surgery. The positive circumferential resection margin rate was 4.9% (4 of 82).

Meaning

The use of MRI criteria to select good prognosis patients with rectal cancer for primary surgery results in a low rate of positive circumferential resection margin and suggests that chemoradiotherapy may not be necessary for all patients with stage II and III rectal cancer.

This nonrandomized phase 2 clinical trial evaluates the safety and feasibility of using magnetic resonance imaging criteria to select patients with “good prognosis” rectal tumors for primary surgery.

Abstract

Importance

Chemoradiotherapy (CRT), followed by surgery, is the recommended approach for stage II and III rectal cancer. While CRT decreases the risk of local recurrence, it does not improve survival and leads to poorer functional outcomes than surgery alone. Therefore, new approaches to better select patients for CRT are important.

Objective

To conduct a phase 2 study to evaluate the safety and feasibility of using magnetic resonance imaging (MRI) criteria to select patients with “good prognosis” rectal tumors for primary surgery.

Design, Setting, and Participants

Prospective nonrandomized phase 2 study at 12 high-volume colorectal surgery centers across Canada. From September 30, 2014, to October 21, 2016, a total of 82 patients were recruited for the study. Participants were patients newly diagnosed as having rectal cancer with MRI-predicted good prognosis rectal cancer. The MRI criteria for good prognosis tumors included distance to the mesorectal fascia greater than 1 mm; definite T2, T2/early T3, or definite T3 with less than 5 mm of extramural depth of invasion; and absent or equivocal extramural venous invasion.

Interventions

Patients with rectal cancer with MRI-predicted good prognosis tumors underwent primary surgery.

Main Outcomes and Measures

The primary outcome was the proportion of patients with a positive circumferential resection margin (CRM) rate. Assuming a 10% baseline probability of a positive CRM, a sample size of 75 was estimated to yield a 95% CI of ±6.7%.

Results

Eighty-two patients (74% male) participated in the study. The median age at the time of surgery was 66 years (range, 37-89 years). Based on MRI, most tumors were midrectal (65% [n = 53]), T2/early T3 (60% [n = 49]), with no suspicious lymph nodes (63% [n = 52]). On final pathology, 91% (n = 75) of tumors were T2 or greater, 29% (n = 24) were node positive, and 59% (n = 48) were stage II or III. The positive CRM rate was 4 of 82 (4.9%; 95% CI, 0.2%-9.6%).

Conclusions and Relevance

The use of MRI criteria to select patients with good prognosis rectal cancer for primary surgery results in a low rate of positive CRM and suggests that CRT may not be necessary for all patients with stage II and III rectal cancer.

Trial Registration

ISRCTN.com identifier: ISRCTN05107772

Introduction

It is well established that the main goal of rectal cancer treatment is to achieve a negative circumferential resection margin (CRM), which is the strongest predictor of local recurrence (LR). Currently, North American guidelines for the treatment of stage II and III rectal cancer recommend preoperative chemoradiotherapy (CRT), followed by surgery, because this approach has been shown to decrease the rate of LR in several well-designed randomized clinical trials (RCTs).^1,2,3,4,5,6 However, with improvements in the quality of rectal cancer surgery, rates of positive CRM and LR have decreased. Therefore, while CRT still decreases the risk of LR by approximately 50%, the number of patients needed to treat to avoid one LR has increased markedly. Although in the era of modern rectal cancer surgery fewer patients benefit from CRT, the long-term adverse outcomes of treatment, including bowel and sexual dysfunction, have not changed.^7,8,9,10,11 Therefore, a more selective approach to the use of CRT may allow for better functional outcomes, while maintaining oncologic outcomes. Both the MERCURY and OCUM study groups have tested more selective approaches to CRT to avoid radiotherapy in patients with stage I to III tumors at low risk of LR.^12,13,14,15 The indications for CRT in these studies are largely based on the predicted status of the CRM (ie, distance to the mesorectal fascia) as determined by magnetic resonance imaging (MRI). In these studies, patients with a predicted CRM greater than 1 mm underwent primary surgery, and CRT was avoided. Both the MERCURY and OCUM have reported excellent oncologic results, with rates of positive CRM and LR less than 5% in patients undergoing primary surgery. Despite the promising findings, these studies included fewer than 600 patients. Therefore, additional investigations evaluating more selective approaches to the use of CRT are needed. The objective of the present phase 2 study was to prospectively evaluate the use of MRI criteria to identify patients with “good prognosis” rectal tumors for primary surgery in Canada.

Methods

Study Overview

This was a prospective nonrandomized phase 2 study conducted at 12 high-volume colorectal surgery centers across Canada (Figure). Each participating center designated a site lead for surgery, radiology, radiation oncology, and pathology. A QuickSilver study investigators’ meeting with all site leads was conducted to obtain consensus on the final trial protocol (Supplement 1).¹⁶ Research ethics and data sharing agreements were obtained from each participating center before the start of the study. Surgeons involved in the study identified consecutive patients with primary rectal cancer who met the study inclusion criteria. Rectal cancer was defined as adenocarcinoma 0 to 15 cm from the anal verge on endoscopy and proximal extent of the tumor at or below the sacral promontory on MRI.

Figure. Study Flow Diagram^a.

CRM indicates circumferential resection margin; MRI, magnetic resonance imaging.

^aThe initial number of patients assessed and excluded were not recorded; the Limitations section offers an explanation.

Patient Sample

Patients were eligible for study enrollment if they were 18 years or older, expected to undergo radical resection (ie, not a local excision), had a good prognosis tumor based on MRI criteria, and were able to provide written informed consent. The MRI criteria for good prognosis tumors included (1) distance to the mesorectal fascia (ie, predicted CRM) greater than 1 mm from the primary tumor, discontinuous tumor nodule, or likely positive (ie, suspicious) lymph node; (2) T category of definite T2, T2/early T3, or definite T3 with less than 5 mm of extramural depth of invasion (EMD); and (3) absent or equivocal extramural venous invasion (EMVI). The status of the mesorectal lymph nodes was not considered a criterion for good prognosis tumors, and patients with N0-N2 disease were eligible for inclusion provided all other criteria were met. Among patients undergoing radical resection for rectal cancer, numerous studies^1,2,17 have demonstrated that risks of a positive CRM or local tumor recurrence are higher when the procedure is an abdominoperineal vs low anterior resection. Therefore, to be cautious in this first prospective national-level study to date of new criteria for CRT, we excluded patients with a planned intersphincteric dissection, coloanal handsewn anastomosis, and/or abdominoperineal resection. Other exclusion criteria included suspicious extramesorectal lymph nodes or other metastatic disease, previous pelvic radiotherapy, more than one primary tumor, or history of other major malignant cancer within the last 5 years.

Radiology, Surgery, and Pathology Protocols and Assessments

Before the start of the study, webinars were conducted with participating radiologists, surgeons, and pathologists to ensure standardization of the MRI, surgical, and pathologic protocols and reports. The treating surgeon completed the preoperative assessment, including (1) digital rectal examination and endoscopic examination of the primary tumor; (2) computed tomography of the chest, abdomen, and pelvis; (3) pelvic MRI; and (4) presentation at a multidisciplinary cancer conference (MCC). Primary surgery was performed as soon as possible after obtaining written informed consent from the patient to participate in the study. The surgical procedure was left to the discretion of the treating surgeon and involved a partial (ie, 5 cm of mesorectum removed distal to the leading edge of the tumor) or total mesorectal excision (TME). Participating surgeons were fellowship trained in colorectal surgery or surgical oncology and were encouraged to use a synoptic operative report template designed specifically for the study.

A standard MRI protocol based on the results from the MERCURY study group was used and required high-resolution axial oblique T2-weighted sequences, and it was recommended that participating radiologists use a synoptic MRI report endorsed by Cancer Care Ontario and the Radiologic Society of North America.^12,18 Preoperative MRI reports provided tumor location, N category, T category, and distance to the mesorectal fascia. Radiologists were not required to distinguish between T2 and T3 tumors with minimal extension beyond the bowel wall; therefore, designations were definite T2, T2/early T3, and definite T3 with less than 5 mm of EMD.

Macroscopic examination and processing of surgical specimens were performed according to the method by Quirke et al¹⁹ and included macroscopic evaluation of TME quality. Photographs of the intact specimens and cross-sectional sections were obtained. All mandatory data elements of the College of American Pathologists colorectal cancer protocol²⁰ were evaluated and included in pathology reports. All final pathology reports were reviewed by the pathology site lead, who completed a standardized pathologic summary report that was submitted to the central study office located at Mount Sinai Hospital, Toronto, Ontario, Canada, with the corresponding pathology report. Participating pathologists were encouraged to use the College of American Pathologists checklist.²⁰

Follow-up

The recommended adjuvant treatment of patients included in the study was (1) no further treatment for negative CRM and negative lymph nodes; (2) adjuvant chemotherapy for negative CRM and positive lymph nodes; and (3) postoperative CRT for positive CRM regardless of nodal status, followed by adjuvant chemotherapy, at the discretion of the treating center. Surveillance of all patients was recommended for a period of 5 years as per institutional protocol.

Data Collection

Participating surgeons were required to submit the MRI report, operative report, and pathology report to the central study office. Data from these reports were abstracted into the study database by a dedicated research coordinator.

Primary Outcome and Data Analysis

The primary outcome for the study was the proportion of patients with a positive CRM. A positive CRM was defined as any tumor, discontinuous tumor nodule, or positive lymph node located less than 1 mm from the CRM on final pathologic assessment. Descriptive and univariate analyses were used for patient, imaging, and tumor specimen characteristics. For the study, a baseline probability of a 10% positive CRM was selected and was considered the highest acceptable risk of positive CRM. This was based on the results of previous rectal cancer RCTs, in particular the MRC CR07 trial, which reported an overall positive CRM rate of 11.1% (134 of 1208).^5,21

Based on 30 high-volume rectal cancer surgeons treating a minimum of 10 new patients with rectal cancer over a 2-year period, the target sample size for the study was 75 patients, assuming 30% (90 of 300) would meet the inclusion criteria and an 80% participation rate. Assuming a 10% probability of a positive CRM, this sample size was estimated to yield a 95% CI of ±6.7%.²² A data safety monitoring committee consisting of a statistician, surgeon, radiation oncologist, and pathologist (who were not participating in the study) reviewed interim assessments that occurred after every 25 patients accrued. The stopping rule for the study was a positive CRM of greater than 10% reported at any interim assessment.

Results

From September 30, 2014, to October 21, 2016, a total of 82 patients from 12 hospitals across Canada were recruited for the study. All eligible patients invited to participate in the study agreed and underwent primary surgery. Surgery was performed by 32 fellowship-trained surgeons. All patients were presented at MCCs, and adequate image quality and the use of pathology methods by Quirke et al¹⁹ were confirmed in 87% (71 of 82) and 88% (72 of 82), respectively. The mean time from MRI to surgery was 4.9 weeks (range, 1 day to 14 weeks).

Participant Characteristics

Most of the patients were male (74% [61 of 82]), and the median age at the time of surgery was 66 years (range, 37-89 years). All patients underwent restorative surgery, and a diverting ileostomy was created in 72% (59 of 82). Sixty-six percent (54 of 82) of the cases were performed using a laparoscopic approach.

MRI Findings

The MRI results are listed in Table 1. The MRI data were reported by 30 gastrointestinal radiologists. Most of the tumors were midrectal (65% [n = 53]), T2/early T3 (60% [n = 49]), with no suspicious lymph nodes (63% [n = 52]). The mean distance of the tumor to the mesorectal fascia was reported as 10.4 mm (range, 1.9-24 mm).

Table 1. Magnetic Resonance Imaging (MRI) Results Among 82 Patients.

Variable	No. (%)
MRI tumor height
Low, 0-5 cm	6 (7)
Mid, 5.1-10 cm	53 (65)
High, 10.1-15 cm	23 (28)
MRI T category
Definite T2	16 (20)
T2/early T3	49 (60)
Definite T3 with <5 mm of EMD	17 (21)
MRI lymph node status
Not suspicious	52 (63)
Suspicious	30 (37)

Abbreviation: EMD, extramural depth of invasion.

Pathologic Findings

The results of the pathologic assessments are listed in Table 2. The pathologic assessments were performed by 47 gastrointestinal pathologists. The quality of the TME was graded as complete or near complete in 98% (80 of 82) of the specimens. The mean CRM distance was 12.8 mm (range, 0-70 mm), and the mean number of lymph nodes retrieved was 25 (range, 7-53). Thirty-three cases (40%) were T2, and 40 cases (49%) were T3. Twenty-nine percent (24 of 82) of cases had positive mesorectal lymph nodes, and EMVI was present in 16% (13 of 82). Percentage agreement for lymph nodes between MRI and pathology was 68.3%. Ten cases (12%) were understaged, and 16 cases (20%) were overstaged. The positive predictive value was 46.7%, and the negative predictive value was 80.7%.

Table 2. Pathology Results Among 82 Patients.

Variable	No. (%)
Quality of the TME
Complete	67 (82)
Near complete	13 (16)
Incomplete	2 (2)
pCRM status
Negative	78 (95)
Positive	4 (5)
pT category
T1	7 (9)
T2	33 (40)
T3	40 (49)
T4	2 (2)
pN category
N0	58 (71)
N1	18 (22)
N2	6 (7)
pStage, AJCC 7th ed
Stage I, T1-T2, N0	34 (41)
Stage II, T3-T4, N0	24 (29)
Stage III, any T, N1-N2	24 (29)
EMVI
Absent	69 (84)
Present	13 (16)

Abbreviations: AJCC, American Joint Committee on Cancer; CRM, circumferential resection margin; EMVI, extramural venous invasion; TME, total mesorectal excision.

On final pathology, 91% (n = 75) of tumors were T2 or greater, 29% (n = 24) were node positive, and 59% (48 of 82) had stage II or III disease. Four patients had a positive CRM, for a positive CRM rate of 4.9% (95% CI, 0.2%-9.6%). In addition, one patient had a positive distal margin.

Details of Positive CRMs

The clinical and pathologic details for the 4 positive CRMs are listed in eTable 1 in Supplement 2. The first positive CRM was due to a discontinuous tumor nodule. The MRI was rereviewed at the MCC, and the tumor deposit was not visualized on the preoperative MRI. The patient subsequently received postoperative CRT, followed by adjuvant chemotherapy. The second positive CRM was due to primary tumor at the level of the intersphincteric space requiring a handsewn anastomosis and represented a protocol variation. This patient subsequently received only adjuvant chemotherapy. The third positive margin was also due to primary tumor. The time between the MRI and surgery was 14 weeks due to patient comorbidity and represented a protocol variation. This patient subsequently received postoperative CRT, followed by adjuvant chemotherapy. The fourth positive margin was due to a positive lymph node. The MRI was rereviewed at the MCC, and the lymph node was not visualized on the preoperative MRI. The patient subsequently received only adjuvant chemotherapy.

Follow-up Treatment

In total, 30% (25 of 82) of patients received additional adjuvant treatment. Six patients received postoperative CRT and adjuvant chemotherapy (2 positive CRM, 2 negative CRM with positive lymph nodes, and 2 negative CRM with discontinuous tumor nodules). Nineteen patients received adjuvant chemotherapy (2 positive CRM, 12 negative CRM with positive lymph nodes, 1 negative CRM with discontinuous tumor nodules, and 4 with negative CRM and EMVI present). Among patients with stage II or III tumors, 88% (42 of 48) received no form of radiotherapy.

Discussion

These results of the QuickSilver study show that the use of MRI criteria to select good prognosis rectal tumors for primary surgery is safe and feasible. The findings resulted in a positive CRM rate of 4.9% (95% CI, 0.2%-9.6%).

This study performed in Canada adds to the reported international experience of the MERCURY and OCUM studies, which included 122 and 254 patients, respectively (eTable 2 in Supplement 2).^12,15 All 3 of these studies used high-resolution T2-weighted axial oblique MRI sequences. The common MRI criterion used to predict good prognosis rectal tumors was a predicted negative CRM, defined as distance of the tumor to the mesorectal fascia greater than 1 mm on MRI. Furthermore, none of the studies included the status of the mesorectal lymph nodes as an MRI criterion to predict good prognosis tumors. All of the studies had a similar proportion of patients with stage II or III disease and low rectal cancers. The positive CRM rates were 5% (4 of 82) in QuickSilver, 3% (4 of 122) in MERCURY, and 2% (2 of 134) in OCUM, and 5-year LR rates in the latter 2 studies were 3.3% and 2.0%, respectively. Taken together, the findings of these studies suggest that, in contrast to CRT for all stage II and III rectal tumors, a more selective approach to CRT based on predicted CRM status may also result in excellent oncologic outcomes.

The criteria to identify good prognosis patients will likely continue to evolve. We used the MERCURY study MRI criteria, which included both EMD and EMVI; however, we excluded patients with very low rectal cancers in which the surgeon recognized the need for an intersphincteric anastomosis or abdominoperineal resection. The OCUM study investigators considered distance to the mesorectal fascia as the only MRI criterion for their study and did not include either EMD or EMVI. Furthermore, patients in OCUM with T3 low tumors within 6 cm of the anal verge and all T4 tumors received CRT. Given that the long-term results for MERCURY and OCUM are similar, it seems that the use of the OCUM criteria may be more reproducible and feasible to adapt into clinical practice.

There is great interest in developing more selective approaches to CRT for patients with stage II and III rectal cancer. The MRC CR07/NCIC-CTG C016 was a multicenter RCT that compared short-course preoperative radiotherapy vs primary surgery, followed by selective postoperative CRT for patients with a positive CRM.^5,19 Although the rate of positive CRM was similar in both groups (10% vs 12%), the rate of LR at 5 years was significantly lower in the preoperative radiotherapy arm (5% vs 12%). However, only 52% of the surgical specimens were in the mesorectal plane; because the mesorectal plane was an independent predictor of LR, it is possible that preoperative therapy compensated for inadequate TME surgery. The PROSPECT (N1048) trial is randomizing patients with resectable stage II or III (T2N1, T3N0, and T3N1) rectal cancer with noninvolved mesorectal fascia to standard preoperative CRT and surgery vs preoperative full-dose FOLFOX (leucovorin calcium [folinic acid], fluorouracil, and oxaliplatin) chemotherapy, selective CRT, and then surgery.²³ In the experimental arm, preoperative CRT is reserved for tumors that have less than 20% reduction in tumor volume after full-dose chemotherapy. Most patients eligible for PROSPECT would meet the eligibility criteria for QuickSilver, MERCURY, and OCUM and thus avoid neoadjuvant full-dose chemotherapy or CRT. Response to full-dose chemotherapy, predicted CRM status, and/or clinical stage II or III disease are varying criteria that can inform recommendations for preoperative radiotherapy among patients with rectal cancer. Moving forward, it is likely that other criteria (eg, biomarkers for response to radiotherapy) will emerge and require testing to determine if such factors more effectively balance optimization of functional and oncologic patient outcomes vs current selection criteria for preoperative radiotherapy.

No patient declined to participate in the study even though the consent process provided a detailed outline of the risks and benefits of primary surgery vs standard treatment with CRT. This result is consistent with the existing literature indicating that patients value functional outcomes relative to oncologic outcomes.^{24,25,26,27,28,29} Kennedy et al²⁴ previously reported that 54% of future patients would not choose CRT until it offered an absolute reduction in LR from 15% to 5% or less. These results suggest that a significant proportion of patients may be willing to accept a higher risk of LR to avoid CRT to optimize functional results. Therefore, offering patients with good prognosis rectal cancer a choice between CRT and primary surgery may be more consistent with patient values.

Limitations

There are several limitations to this study. First, involved clinicians were from high-volume centers with focused expertise. Therefore, our identification of good prognosis patients and quality of treatment may not be generalizable. Second, our study excluded patients with an expected intersphincteric anastomosis or need for an abdominoperineal resection because the risk of positive CRM and LR is higher in these patients compared with low anterior resection. Moving forward, our next national study is testing the indications for patients with CRT with low tumors but with modifications for CRT that may be similar to those in the OCUM study. Third, we did not collect data for patients seen at participating centers who may have been eligible for the study but were not included. While our patient population included a high proportion of stage I tumors and a low proportion of node-positive tumors, our patient population is not considerably different from the MERCURY or OCUM study populations. Fourth, only the positive CRM rate is reported because long-term oncologic outcomes have not yet matured. However, we anticipate a low LR rate comparable to rates achieved in the MERCURY and OCUM studies given our similar rates of positive CRM.

Conclusions

Results of this study suggest that in high-volume centers, the use of MRI criteria to identify patients with good prognosis rectal cancer for primary surgery results in a positive CRM of less than 5%. These findings suggest that CRT may not be necessary for all patients with stage II and III rectal cancer. Further data will be required before this approach can be widely adopted into clinical practice.

Supplement 1.

Trial Protocol

Supplement 2.

eTable 1. Details of Positive CRMs

eTable 2. Results for QuickSilver, MERCURY, and OCUM

Supplement 3.

Data Sharing Statement