Abstract
Background/Aim: We compared the risk factors for locally advanced lower rectal cancer (LALRC) recurrence evaluated by preoperative magnetic resonance imaging (MRI) and pathological factors analysed via the longitudinal slicing method to identify high risk groups for recurrence. Patients and Methods: This retrospective single-institution cohort study analysed 45 consecutive patients who underwent curative surgery for LALRC. Data were analysed by an experienced radiologist and pathologist. Results: Final preoperative extramural venous invasion (EMVI) and extramural depth of invasion (EMD) determined via MRI were significantly associated with EMVI and EMD determined via pathological analysis. The log-rank test for disease-free survival based on initial preoperative factors showed significantly poor prognoses for circumferential resection margin (CRM)-positive, EMVI-positive, and EMD-positive patients. Conclusion: Final preoperative EMVI and EMD determined via MRI correlated with pathological EMVI and EMD, especially in patients who did not undergo preoperative treatment. CRM, EMVI, and EMD determined via preoperative MRI were significant risk factors for recurrence.
Total mesorectal excision (TME), a reproducible anatomical approach for pelvic dissection, was introduced in the late 1980s. Due to its ability to reduce local recurrence, it has been increasingly adopted as the standard surgical resection technique for treating rectal cancer (1). However, locally advanced lower rectal cancer (LALRC) has high local recurrence rates after curative surgery treatment (2). Further, based on large-scale randomised trials, preoperative chemoradiotherapy (CRT) followed by TME has been shown to decrease the risk of local recurrence (2, 3). Therefore, TME is performed after preoperative CRT in Western countries as the standard treatment for rectal cancer to improve the local control of LALRC. Unfortunately, CRT adversely affects bowel and sexual functions compared to surgery alone (4, 5). Thus, strategies for the appropriate selection of patients for CRT are necessary.
Circumferential resection margin (CRM), extramural venous invasion (EMVI), and extramural depth of invasion (EMD) have been reported to be associated with the prognosis of LALRC patients. Traditionally, these factors have been diagnosed in postsurgical pathology specimens (6-8). However, postoperative evaluations are not helpful for the preoperative treatment planning of rectal cancer patients. Recently, using magnetic resonance imaging (MRI) to evaluate these factors has been declared equivalent to evaluating these factors via pathological analysis. Additionally, it has been reported that these MRI-evaluated factors are associated with the prediction of prognosis of LALRC patients (8-11). Since MRI is an accurate and reproducible technique for preoperatively identifying these factors, it can be beneficial in formulating treatment strategies.
The National Comprehensive Cancer Network (NCCN) and European Society for Medical Oncology (ESMO) guidelines suggest that LALRC patients with CRM and EMVI are at high risk for LALRC recurrence, and preoperative treatments, such as CRT, are recommended for such patients (12, 13). However, these data are typically assessed by pathological analysis of rectal specimens via the transverse slicing method. Furthermore, although the longitudinal slicing method is recommended by the Japanese Classification of Colorectal Carcinoma guidelines (14), there have been no reports comparing preoperative MRI to pathological analysis via the longitudinal slicing method while simultaneously assessing CRM, EMVI, and EMD.
We believe that it is crucial to evaluate the risk factors of LALRC via preoperative MRI and pathological analysis, as these risk factors may potentially help predict prognoses postoperatively and help make decisions regarding the indication for preoperative treatment. Therefore, we aimed to identify high risk groups for recurrence among LALRC patients based on preoperative information by comparing risk factors for recurrence evaluated by preoperative MRI and pathological factors analysed via the longitudinal slicing method.
Patients and Methods
Patients and study design. In this retrospective single-institution cohort study, 45 consecutive patients were selected from the Department of General Surgical Science database, Gunma University Hospital, according to the following criteria. The presence of locally advanced lower rectal adenocarcinoma (TNM classification, T3/T4, any N, and M0), patients who underwent curative resection for rectal cancer as a primary surgery between July 2013 and December 2016, and patients who underwent MRI within one month before surgery.
Data regarding patients’ age, sex, anal verge (AV) distance, carcinoembryonic antigen (CEA) levels, clinical TNM classifications, clinical CRM (cCRM), clinical EMVI (cEMVI), clinical EMD (cEMD), operation type, surgical procedure, adjuvant chemotherapy, maximum tumour size, tumour histological type, pathological TNM classifications, lymph node yield, distal resection margins, pathological CRM (pCRM), pathological EMVI (pEMVI), and pathological EMD (pEMD) were analysed. Patients were followed up until June 2020.
Initial preoperative factors, such as initial CEA (iCEA) levels, initial cCRM (icCRM), initial cEMVI (icEMVI), and initial cEMD (icEMD), were assessed before patients underwent CRT and used for preoperative diagnosis in upfront surgery cases. Final preoperative factors, such as final CEA (fCEA) levels, final cCRM (fcCRM), final cEMVI (fcEMVI), and final cEMD (fcEMD), were assessed after CRT in patients who underwent CRT and used for preoperative diagnosis in upfront surgery cases.
Ethics approval and consent to participate. This study was approved by the ethics committee/institutional review board (Gunma University Hospital Approval no. HS2020-046) and was performed in accordance with the 1964 Declaration of Helsinki and its later amendments. Due to the retrospective nature of our study, the need for informed consent was waived.
Clinical TNM classifications, cCRM, cEMVI, and cEMD. Clinical TNM classifications, cCRM, cEMVI, and cEMD were evaluated before and after preoperative CRT. Clinical TNM classifications were assessed based on endoscopy, computed tomography (CT), and MRI findings. cCRM, cEMVI, and cEMD were evaluated by an experienced radiologist who was blinded to the patients’ clinical history or outcomes. The radiologist assessed T2-weighted MRI, contrast-enhanced MRI, and CT scans. Patients were considered cCRM-positive if the distance from the tumour to the mesorectal fascia (MRF) or the levator muscle was ≤1 mm (15, 16). EMVI was defined as the involvement of veins beyond the muscularis propria. cEMVI status was evaluated according to the 5-scale EMVI scoring system (17) and recorded as negative (EMVI score, 0-2) or positive (EMVI score, 3-4) (Figure 1A and E). cEMD was measured as the maximum tumour invasion depth beyond the muscularis propria, as determined via MRI (15). cEMD-positive was defined as a maximum tumour invasion depth of >5 mm.
Extramural venous invasion evaluations of magnetic resonance imaging (MRI) and pathological analysis. Case 1. Axial T2-weighted MRI (A). The tumor signal did not extend into the vascularis, and this case was scored as a 2. Macro long axis section corresponding to MRI (B and C) and the micro section using elastica van Gieson stain (D and E) showed tumor invading into perirectal venous (arrow heads). Case 2. Coronal T2-weighted MRI (F). The tumor signal extended into the vascularis outside the muscularis propria (arrows), and this case was scored as a 4. Macro long axis section corresponding to MRI (G) and the micro section using elastica van Gieson stain (H) showed tumor invading into perirectal venous (arrow heads).
Pathological TNM classifications, pCRM, pEMVI, and pEMD. Pathological TNM classifications, pCRM, pEMVI, and pEMD were evaluated via specimen and histopathological analyses by an experienced pathologist blinded to patients’ clinical history or outcomes. Pathological analysis was performed using the longitudinal slicing method for rectal specimens, as previously described (14) (Figure 1B, C, F and G). Patients were considered pCRM-positive if the distance from the tumour to the margin of surgical resection was ≤1 mm. Hematoxylin and eosin-stained sections of the tumour were initially examined, and elastic tissue-stained sections were reviewed to diagnose pEMVI. pEMVI diagnoses were confirmed if an adherent tumour was present within an extramural, well-defined tubular or rounded structure accompanying an artery (18) (Figure 1D and H). pEMD was measured as the maximum tumour invasion depth beyond the muscularis propria, as determined via histopathological analysis. pEMD-positive was defined as a maximum tumour invasion depth of >5 mm (11).
Preoperative treatment and surgery. Preoperative treatments were offered to patients with a high risk of local recurrence, such as patients with large tumours that invaded other organs, and decided upon in multidisciplinary team meetings. Hyperthermochemoradiation therapy was performed for preoperative CRT at our hospital, as previously described (19). Briefly, the clinical target volume for radiation encompassed the primary tumour and entire mesorectal tissue. The total radiation dose was 50 Gy, with daily fractions of 2.0 Gy administered for 5 consecutive days per week. Chemotherapy consisted of capecitabine (1,700 mg/m2 per day) administered 5 days a week for 5 weeks on the day of radiation. Five hyperthermia sessions were performed once a week using an 8-MHz radiofrequency capacitive heating device (Thermotron-RF 8; Yamamoto Vinita Co., Ltd., Osaka, Japan). Concerning surgical procedure, patients underwent TME, as previously described (1).
Postoperative treatment and follow-up. Postoperative treatment and follow-up were performed in accordance with the Japanese Society for Cancer of the Colon and Rectum guidelines (20). Generally, 5-fluorouracil-based chemotherapy was used for postoperative adjuvant chemotherapy; the treatment lasted for 6 months. Chest and abdominal CT were performed every 6 months, and blood tests, including CEA and CA19-9 level measurements, were performed at every 3 months post-operation for postoperative surveillance. If recurrence was suspected, pelvic MRI, gadolinium-ethoxybenzyldiethylenetriamine pentaacetic acid-enhanced MRI, and positron emission tomography were performed for confirmation.
Statistical analysis. Categorical factors were analysed using Fisher’s exact test or the Chi-square test. The 3-year disease-free survival (3y-DFS) rate was estimated using the Kaplan–Meier method, and differences were assessed using the log-rank test. A Cox proportional hazards model based on initial and final preoperative factors was used to analyse independent prognostic factors for DFS. If the p-value of a factor was <0.1 in the univariate analyses, it was included in the multivariate analyses. The Cox model analysis results were reported as hazard ratios (HRs) and 95% confidence intervals (CIs). All statistical analyses were performed using IBM SPSS Statistics for Windows, ver. 22.0 (IBM Corp., Armonk, NY, USA). A p-value of <0.05 was considered statistically significant.
Results
Patient characteristics and pathological findings. The patients’ characteristics are summarised in Table I. The median age of the patients was 61 years (range=28-86 years). Of the 45 patients, 26 (57.8%) underwent upfront surgery, 19 (42.2%) received preoperative CRT followed by surgery, and 26 (57.8%) received adjuvant chemotherapy. Pathological findings are summarised in Table II.
Patient clinical characteristics and performed treatment.
Pathological findings of the resected specimen.
Correlation between final preoperative factors and pathological findings. fcEMVI and fcEMD were significantly associated with pEMVI and pEMD (p=0.008 and <0.001, respectively) (Table III), especially in patients who did not receive preoperative CRT (upfront surgery, p=0.007 and 0.001, respectively; preoperative treatment, p=0.373 and 0.095, respectively). fcCRM did not correlate with pCRM (p=0.192). The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of the final preoperative factors and pathological EMVI, CRM, and EMD are shown in Table IV. The specificity and positive predictive value of EMVI were high. Moreover, the sensitivity and negative predictive value of EMD were high, especially in patients who did not receive preoperative CRT.
Correlation between final preoperative diagnoses and pathological findings in CRM, EMVI, and EMD.
Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy between final preoperative diagnoses and pathological findings in CRM, EMVI, and DME.
The 3y-DFS and prognostic preoperative factors associated with recurrences. The 3y-DFS rate was 64.4% (Figure 2A), and the mean follow-up period was 44 months (range=0-76 months). The 3y-DFS rate of icCRM-positive patients (44.3%) was significantly different from the 3y-DFS rate of icCRM-negative patients (80.4%) (p=0.021) (Figure 2B). The 3y-DFS rate of icEMVI-positive patients (34.6%) was significantly different from the 3y-DFS rate of icEMVI-negative patients (74.3%) (p=0.039) (Figure 2C). The 3y-DFS rate of icEMD-positive patients (49.1%) was significantly different from the 3y-DFS rate of icEMD-negative patients (82.5%) (p=0.038) (Figure 2D).
Disease-free survival in all patients and initial preoperative magnetic resonance imaging (MRI) diagnoses including extramural venous invasion (EMVI) and EMVI, circumferential resection margin (CRM), and extramural depth of invasion (EMD). (A) Kaplan–Meier curves for disease-free survival (DFS) in all patients. The 3y-DFS was 64.4%. (B) Kaplan–Meier curves for DFS in initial clinical CRM (icCRM)-positive and icCRM-negative. The 3y-DFS in icCRM-positive patients was 44.3%, whereas that in icCRM-negative patients was 80.4% (p=0.021). (C) Kaplan–Meier curves for DFS in initial clinical EMVI (icEMVI)-positive and icEMVI-negative. The 3y-DFS in icEMVI-positive patients was 34.6%, whereas that in icEMVI-negative patients was 74.3% (p=0.039). (D) Kaplan–Meier curves for DFS in icEMD-positive and icEMD-negative. The 3y-DFS in icEMD-positive patients was 49.1%, whereas that in icEMD-negative patients was 82.5% (p=0.038).
Table V summarises the results of our univariate and multivariate analyses. The multivariate analysis showed that initial preoperative factors were not independent factors for recurrence. However, being fcCRM-positive (HR=4.103; 95%CI=1.192-14.122; p=0.025) was an independent factor for recurrence.
Univariate and multivariate analyses of preoperative factors using cox proportional hazards model.
Relationship between changes in MRI findings before and after CRT and recurrence. Of the 19 patients who received preoperative CRT, the disease recurred in five patients (26.3%), including one patient (5.3%) who developed local recurrence and four patients (21.1%) who developed distant metastases. Although the incidence of local recurrence after preoperative CRT was low, distant metastases were relatively high. Additionally, local recurrence did not occur in patients whose risk factors improved after preoperative CRT. However, distant metastases were observed in one (33.3%) of three patients whose cEMVI status improved after preoperative CRT and two (66.7%) of the three patients whose cEMD status improved after preoperative CRT. Of the four patients whose cCRM status improved after preoperative CRT, none developed distant metastases.
Discussion
The present study revealed that fcEMVI and fcEMD significantly correlated with pEMVI and pEMD, especially in patients who did not receive CRT. fcCRM did not correlate with pCRM. To the best of our knowledge, this study is the first to compare preoperative MRI and pathological analysis via the longitudinal slicing method for rectal specimens in simultaneously assessing CRM, EMVI, and EMD. Furthermore, the log-rank test showed significantly poor DFS prognoses for icCRM-positive, icEMVI-positive, and icEMD-positive patients. Moreover, our multivariate analysis showed that fcCRM-positivity was a significant risk factor for recurrence in LALRC patients. The correlation between these preoperative factors and LALRC patient prognoses may be helpful in making preoperative treatment decisions.
Final preoperative EMVI and EMD significantly correlated with pathological EMVI and EMD. Interestingly, we found that pathological EMVI was likely to be positive if it was preoperatively positive and that pathological EMD was likely to be negative if it was preoperatively negative. Additionally, the specificity and positive predictive value of EMVI were high; however, its sensitivity and negative predictive value were relatively low, which is consistent with previous studies (17, 21). While tumour invasions visible during MRI are observable during pathological analysis, tumour invasions into tiny blood vessels can only be observed through pathological analysis because they are beyond the resolution limit of MRI (22). False-negative diagnoses caused by microscopic tumour invasion in small extramural vessels do not cause serious clinical consequences, and cEMVI status has been reported to more accurately predict the prognosis of LALRC than does pEMVI status (21). Therefore, tumour invasion in small vessels, which can be found via pathological analysis and is typically overlooked with MRI, may not be essential in determining the indication for preoperative treatments. Our study also showed that the sensitivity and negative predictive value of EMD was relatively high. In previous studies, EMD measurements obtained via MRI and pathological analysis were comparable and had a mean difference of ±0.05 mm; however, this difference increased when EMD exceeded 5 mm (23). Therefore, it is crucial to carefully interpret the significance of EMD measurements obtained via MRI when tumour invasion depth in the muscularis propria exceeds 5 mm.
The MRI assessment of cEMVI and cEMD after CRT is complex due to residual tumours from the desmoplastic reaction, radiation-induced fibrosis, and the misinterpretation of radiation-induced proctitis (23, 24). As such, it is necessary to determine treatment strategies through image evaluations before preoperative treatments.
Final preoperative CRM did not correlate with pathological CRM. CRM is determined by the extent of surgical resection that cannot be predicted with MRI because surgeons achieve CRM negativity via resecting tissue around the MRF. Therefore, there was no relationship between cCRM and pCRM because we defined cCRM as the distance between the tumour and the MRF. To avoid confusion when defining the true CRM, which can refer to the preoperative distance between the tumour and the MRF or the distance between the tumour and surgical resection margin, some experts recommend using the term “MRF” instead of “CRM” for MRI-based diagnosis (25).
We were able to compare factors that were assessed with pathological analysis via the longitudinal slicing method and preoperative MRI. The transverse slicing of rectal specimens, which has been performed by Quirke et al., is recommended in Western countries, and evaluations of pathological analysis can be performed in the same shape as the MRI axial images (26, 27). As such, the relationship between risk factors for recurrence, including CRM, EMVI, and EMD, assessed via preoperative MRI and pathological analysis has previously been reported in terms of pathological findings obtained through the transverse slicing method (10, 16). However, in Japan, the longitudinal slicing method is recommended by the Japanese Classification of Colorectal Carcinoma guidelines (20). The present study is the first to compare preoperative MRI and pathological analysis via the longitudinal slicing method for rectal specimens in simultaneously assessing CRM, EMVI, and EMD.
The log-rank test showed significant differences in DFS between icCRM, icEMVI, and icEMD positivity and negativity. These factors have been reported to be critical prognosticators of recurrence risk, consistent with our findings. They are emphasised in various treatment guidelines for rectal cancer and are used in deciding whether preoperative CRT should be performed (12, 13). However, multivariate analysis showed that these factors were not independent risk factors for recurrence. This might be attributed to our small study population size, which reduced the degree of power of our analyses. Although CRT has been shown to significantly decrease the risk of local recurrence after curative surgery for rectal cancer, CRT leads to poorer functional outcomes, such as reduced bowel and sexual function, than surgery alone (4, 5, 28). Selecting indicators for CRT, such as CRM, EMVI, and EMD, is vital to avoid radiation-induced side effects. Therefore, incorporating these factors into staging systems may lead to improved prognosis prediction and patient selection.
Furthermore, the present study also showed that fcCRM-positivity was a significant risk factor for recurrence in LALRC patients. Moreover, although the incidence of local recurrence was small, it did not occur if CRM involvement improved after CRT. This suggests that CRT contributes to decreasing local recurrence and may help identify true CRM-positive cases, excluding false-positive cases. Therefore, CRM involvement after CRT was a strong risk factor for recurrence. Several reports have shown that CRM involvement is a risk factor for both local recurrence and distant metastasis, and the risk of recurrence and distant metastasis is high if tumours invade other organs (29, 30). Distant metastases would result in poor prognoses even if additional resections achieved CRM negativity to reduce local recurrence incidence rates. Since our study suggests that CRM involvement after CRT can predict distant metastasis development, this may also indicate that controlling aggressive tumour growth, including local recurrence and distant metastasis, is the most effective way to improve LALRC patient prognoses. Previous reports have shown that CRT does not contribute to decreasing the incidence of distant metastases and does not affect survival compared to surgery alone (31, 32). Further, up to 15% of LALRC patients are at risk of developing distant metastasis, even with reasonable local tumour control (2, 3, 31, 32). Neoadjuvant chemotherapy to control distant metastases, in addition to CRT, may help improve the prognosis of patients who exhibit factors for recurrence (33), and new trial results regarding neoadjuvant chemotherapy are currently awaiting publication (34).
The present study had several limitations. Firstly, it had a small sample size, was retrospective in nature, and performed at a single institution. Further analysis of a larger population is needed to clarify the risk factors of recurrence for LALRC patients and the relationship between MRI and pathological findings. Secondly, because there are no criteria for implementing preoperative CRT in Japan, our study included patients who underwent both CRT and upfront surgery for LALRC treatment. Preoperative CRT is the standard therapy used in Western countries. However, the recommendation of preoperative CRT is not strong in Japanese guidelines, and using preoperative CRT for managing LALRC remains uncommon (20). However, the present study revealed that CRM, EMVI, and EMD, which are described in the NCCN and ESMO guidelines (12, 13), were significantly associated with recurrence. Therefore, these factors may be helpful in implementing preoperative CRT in Japan. Thirdly, we did not directly compare the longitudinal slicing and transverse slicing methods. However, we believe that our study is worth considering because our results were consistent with those of studies that used the transverse slicing method. A comparative study between the longitudinal slicing and transverse slicing methods is warranted to clarify whether they yield equivalent results. Despite these limitations, our results suggest that CRM, EMVI, and EMD are important prognostic factors for LALRC patients and may be used in identifying patients who benefit from preoperative treatment preoperatively.
In conclusion, final preoperative EMVI and EMD determined via MRI significantly correlated with pathological EMVI and EMD determined via the longitudinal slicing method for rectal specimens, especially in patients who did not receive preoperative CRT. Furthermore, icCRM, icEMVI, and icEMD were significant risk factors of recurrence for LALRC patients. The correlations between these factors and prognosis may suggest the usefulness of determining indicators for preoperative treatment.
Footnotes
Authors’ Contributions
TS collected data and wrote the manuscript. TS and YS prepared figures. SK and YS evaluated data. HO, KO, TO, RK, KH, AS, MS, TY, YT, TO, MS, KS, and HS revised the manuscript and provided comments on the structure and details of the article. All Authors read and approved the final manuscript.
Conflicts of Interest
The Authors declare that they have no competing interests in relation to this study.
- Received April 25, 2021.
- Revision received May 4, 2021.
- Accepted May 5, 2021.
- Copyright © 2021 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
