Predicting oncologic outcomes by stratifying mesorectal extension in patients with pT3 rectal cancer: A Japanese multi-institutional study

Authors


Abstract

The goal of this study was to clarify the clinical significance of mesorectal extension in pT3 rectal cancer. This currently remains unclear. Data from 975 consecutive patients with pT3 rectal cancer that underwent curative surgery at 28 institutes were reviewed. The distance of the mesorectal extension (DME) was measured histologically. The optimal prognostic cut-off point of the DME for oncologic outcomes was determined using the receiver operating characteristic curve and Cox regression analysis. When patients were subdivided into two groups according to the optimal cut-off point, DME ≤ 4 mm and DME > 4 mm, DME was found to be a powerful independent risk factor for postoperative recurrence. A DME > 4 mm was significantly correlated with distant and local recurrences at Stage IIA and IIIB diseases. The recurrence-free 5-year-survival rate was significantly higher in patients with a DME ≤ 4 mm [86.6% at Stage IIA (p = 0.00015), and 68.7% at Stage IIIB (p < 0.0001)] than in patients with a DME > 4 mm (71.3% at Stage IIA and 49.1% at Stage IIIB). No significant difference was noted in the oncologic outcomes between the two groups at Stage IIIC. A value of 4 mm provides the best prognostic cut-off point for patient stratification and for the prediction of oncologic outcomes. A subclassification based on a 4-mm cut-off point may improve the utility of the TNM 7th staging system except for Stage IIIC. These findings warrant further prospective studies to determine the reliability and validity of this cut-off point.

The influence of the distance of mesorectal extension (DME) on prognosis in patients with pT3 rectal cancer remains unclear. In 1990, Cawthorn et al.1 advocated stratifying mesorectal extension (ME) using a cut-off point of 4 mm, and in 1993, the International Union Against Cancer (UICC) proposed optional cut-off points for ME in the context of pT3 and pT4 tumors.2 Thereafter, several studies have described prognostic heterogeneity in patients with pT3 rectal cancers,3–12 and they used different prognostic cut-off points to stratify the ME (e.g., microscopic invasion,5 2 mm,3 3 mm,9, 12 4 mm,1, 7, 8 5 mm4, 10, 11 or 6 mm6). Furthermore, the clinical significance, statistical appropriateness and reliability of these cut-off points remain controversial, partly because these studies had small samples sizes with underpowered statistical analyses and included cohorts from only a single institution. The goal of this study was to retrospectively analyze a large multi-institutional database from the Study Group of the Japanese Society for Cancer of the Colon and Rectum (JSCCR) to determine the optimal cut-off point for stratification of DME to predict the clinical outcomes in patients with pT3 rectal cancer.

Material and Methods

All protocols contained within this study were approved by the Ethics Committee of the JSCCR and by the local Institutional Review Board. Data were derived from 1,009 patients with pT3 rectal cancer from 28 member institutions of the Study Group of the JSCCR on Extramural ME of Rectal Cancer. All patients had primary rectal adenocarcinoma that was located in the lower two-thirds of the rectum. Patients with rectosigmoid colon cancer were not included in this study. None of the patients received radiotherapy or neoadjuvant chemotherapy before operative management in this study. Total mesorectal excision (TME) and histologically defined curative surgery were performed in each patient by well-trained colorectal surgeons strictly according to the standard technique13 between 1995 and 1999. Thirty-two colorectal surgeons took part in this study and all were trained in TME. The TME quality including longitudinal and circumferential resection margins (CRM) was independently evaluated by expert surgeons and local pathologists according to the rules defined by the JSCCR.14 The CRM positive case was not included in this study. Pelvic lymph node dissection was performed in 593 patients (60.8%). Of the 1,009 patients, clinicopathological information was available for 975 patients, which were eligible for analysis. Thirty-four patients were excluded because of insufficient clinical and follow-up information. Four hundred twenty-five patients (43.6%) received abdominoperineal resection and 550 patients (56.4%) received sphincter-saving operation. The median number of retrieved lymph nodes was 28 (range: 2–129).

The clinicopathological data and follow-up system were based on the rules defined by the JSCCR.14 Patients were restaged according to the pathological TNM classification (7th edition)15, 16 identifying 463 patients at Stage IIA, 422 patients at Stage IIIB and 90 patients at Stage IIIC. According to the postoperative adjuvant treatment protocol of each institution, peroral 5-fluorouracil (5-Fu)-based chemotherapy, such as doxifluridine (5′DFUR), 1-hexylcarbamoyl-5-fluorouracil (HCFU) or uracil-tegafur (UFT), were most frequently administered. One hundred and seventy-seven patients with Stage II (38.2%) and 270 patients with Stage III (52.7%) diseases received postoperative chemotherapy.

Follow-up studies were also conducted in patients and consisted of measurement of serum tumor marker, chest X-ray and abdominal ultrasound examination every 3 months for the first 3 years, and then every 6 months for the following 2 years. When recurrence was suspected based on the serum tumor marker, digital examination and/or ultrasonography, the final diagnosis was made using rectoscopy, computerized tomography (CT) and/or magnetic resonance imaging (MRI) and other diagnostic tools. Local recurrence was defined as the presence of radiologically confirmed or histologically proven tumor occurring via nonhematogenous mechanisms within the pelvis and within the field of the initial surgery. Distant metastasis included hematogenous metastases to the liver, lung, bone, brain, kidney or other organs. Peritoneal dissemination, intra-abdominal, para-aortic, subclavicular, mediastinal and inguinal lymph node metastases also qualified as other recurrences. The outcomes of all patients were precisely investigated. As of January 1995, the eligible surviving patients had been followed for a median period of 86 months (range: 1–166 months).

Measurement of DME

All surgically resected specimens were opened along the antimesenteric side. They were fixed in 20% formalin for at least 48 hr after pinning to a wooden or cork board. Next, one or more longitudinal sections of the tumor were sliced at the point of maximum extramural invasion. They were embedded in paraffin after division into blocks of suitable size and were then routinely processed for staining with hematoxylin and eosin and elastica Van Gieson. Using these sections, tumors in the pT3 category were subdivided based on the histological measurement of the maximum depth (mm) of invasion beyond the outer border of the muscular layer (i.e., DME). Pathological workshops were held by six specialized pathologists before this study to standardize the measurement of DME. Histological tumor depth measurements were carried out without prior knowledge of patient clinical information according to our methods previously reported.17 When the outer border of the muscular layer was completely identifiable (sometimes identifiable as fragments of muscle), the distance from the outer border of the muscular layer to the deepest part of the invasion was measured. When the outer border of the muscular layer was not entirely identifiable due to destruction by invasion or excessive inflammatory reaction, an estimate of the outer border was obtained by drawing a straight solid line between both break points in the muscular layer.

Statistical analysis

Statistical analysis was performed using StatView 5.0 and JMP 8.0 (SAS Institute Inc, Cary, NC, USA) for Windows. All clinicopathological independent variables (15 items) were coded for analysis. These were: gender (female: 0, male: 1); size of tumor (≤ 5 cm: 0, > 5 cm: 1); location of tumor (middle-third: 0, lower-third: 1); gross type (expansive: 0, infiltrative: 1); histology (well-differentiated adenocarcinoma [well]: 0, others :1); lymphatic invasion (negative-to-minimal [ly0-1]: 0, moderate-to-severe [ly2-3]: 1); venous invasion (negative-to-minimal [v0-1]: 0, moderate-to-severe [v2-3]: 1); circumferential resection margin (CRM) (> 1 mm: 0, ≤ 1 mm: 1); lymph node (LN) metastasis (negative: 0, positive: 1); number of retrieved LN (≥12: 0, < 12: 1); operative methods (sphincter-saving operation (SSO): 0, abdominoperineal resection (APR): 1); pelvic LN dissection (no: 0, yes: 1); autonomic-nerve-saving operation (no: 0, yes: 1); postoperative chemotherapy (no: 0, yes: 1) and DME (≤ X mm: 0, > X mm: 1). Overall recurrence (absent: 0, present: 1), distant metastasis (absent: 0, present: 1), local recurrence (absent: 0, present: 1) and survival (alive: 0, dead: 1) were coded as dependent variables. Univariate logistic regression analysis and multivariate Cox regression analysis were used to estimate the independent risk factors for overall postoperative recurrence. The receiver operating characteristic (ROC) curve and multivariate Cox regression analysis were used to determine the optimal cut-off point of the DME for recurrence-free survival. The Kaplan-Meier method and the log-rank test were used for calculating survival rates. The level for statistical significance was determined at p < 0.05, and the confidence interval (CI) was determined at the 95% level.

Results

Measurement of the DME

The mean DME for the 975 cases of pT3 rectal cancer was 4.8 ± 4.4 mm (median: 3.7 mm; range: 0.1–30 mm).

Postoperative recurrence after curative surgery

Postoperative recurrence occurred in 336 patients (34.5%), including 89 patients (19.2%) at Stage IIA, 183 patients (43.4%) at Stage IIIB and 64 patients (71.1%) at Stage IIIC. Eighty patients (8.2%) had local recurrence only, whereas 171 patients (17.5%) had distant metastasis only, and 24 patients (2.5%) had both local recurrence and distant metastases. The remaining 61 patients had other recurrence including peritoneal dissemination, intra-abdominal, para-aortic, subclavicular, mediastinal or inguinal lymph node metastases. The stage-specific local recurrence rate was 5.4% at Stage IIA, 8.8% at Stage IIIB and 20% at Stage IIIC. The recurrence rate of distant metastasis was 10.6% at Stage IIA, 22.2% at Stage IIIB and 31.1% at Stage IIIC.

Cut-off point for DME

To find an optimal prognostic cut-off point, continuous variable analysis of the DME was applied to the ROC curve. As shown in Figure 1, a cut-off value of 4 mm had the best point with higher true-positive (sensitivity) recurrence rate (0.5863), lower false-positive (1-specificity) recurrence rate (0.3709) and highest accuracy rate (0.6144) among all cut-off points (odds ratio [OR]: 2.4, 95% CI: 1.835–3.149, p <0.00001). The ROC curve analysis was valid as a statistical model (area under curve (AUC): 0.6296, OR: 1.08, 95% CI: 1.052–1.119, p <0.0001). Results from the log-rank and multivariate Cox regression analyses for recurrence-free survival are summarized in Table 1. A cut-off value of 4 mm was associated with the highest chi-square value (43.320), lowest p-value (p = 0.00000) and high hazard ratio (HR) of 2.11. The lower/upper limits of CI (L/U ratio) had higher reliability (0.6414) when this cut-off point was compared with other cut-off points. A cut-off value of 4 mm had the greatest influence on recurrence-free survival. Thus, the best prognostic cut-off point for DME was determined as 4 mm, and patients were stratified into two groups according to this value (≤ 4 mm and > 4 mm).

Figure 1.

Cut-off point of the distance of mesorectal extension using ROC curve analysis. A cut-off value of 4 mm showed the best point with higher true-positive (sensitivity) recurrence rate (0.5863), lower false-positive (1-specificity) recurrence rate (0.3709) and highest accuracy rate (0.6144) among all cut-off points (odds ratio: 2.4, 95% CI (1.835–3.149), p < 0.00001). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Table 1. Cut-off points of distance of mesorectal extension (DME) for recurrence-free (RF) survival using log-rank and multivariate Cox regression analyses
inline image

Independent risk factors for postoperative overall recurrence

Univariate and multivariate regression analyses showed that lymph node metastasis was the most powerful independent risk factor for overall postoperative recurrence. The DME was also validated as a powerful independent risk factor by multivariate Cox regression analysis (chi-square: 26.147, HR (95% CI): 1.80 (1.438–2.257), p <0.00001) (Table 2).

Table 2. Independent risk factors for postoperative overall recurrence
inline image

Distant metastasis and local recurrence

The distant metastasis rate was significantly higher in patients with DME > 4 mm at Stage IIA (16.7%, HR: 2.72, 95% CI: 1.529–4.820, p = 0.0006) and Stage IIIB (26.6%, HR: 1.87, 95% CI: 1.275–2.749, p = 0.0014) (Table 3). The local recurrence rate was higher in patients with DME > 4 mm at Stage IIA (7.7%, HR: 2.11, 95% CI: 0.960–4.614, p = 0.0632) and at Stage IIIB (10.6%, HR: 1.75, 95% CI: 1.007–3.036, p = 0.0471). No significant difference was noted in distant metastasis or in local recurrence when stratifying Stage IIIC according to the cut-off point of 4 mm (p = 0.4716 and p = 0.5003, respectively). In all Stage III, however, significant difference was noted in either distant or local recurrence at the cut-off point of 4 mm.

Table 3. Distant Metastasis and Local Recurrence at the Cut-off Value of 4 mm Using Cox Regression Analysis
inline image

Recurrence-free and cancer-specific survival rates

The recurrence-free 5-year-survival rate was significantly higher in patients with a DME ≤ 4 mm than in patients with a DME > 4 mm: 86.6% versus 71.3% (p = 0.00015; HR: 0.44; 95% CI: 0.286–0.683) at Stage IIA and 68.7% versus 49.1% (p < 0.0001; HR: 0.54; 95% CI: 0.399–0.730) at Stage IIIB (Fig. 2). In addition, the cancer-specific 5-year-survival rate of patients with a DME ≤ 4 mm was significantly higher when compared with patients with a DME > 4 mm: 91.3% versus 83.2% (HR: 0.52, 95% CI: 0.325–0.843, p = 0.0066) at Stage IIA and 79.2% versus 60.4% (HR: 0.54, 95%CI: 0.390–0.751, p = 0.0002) at Stage IIIB. However, no significant difference was noted in the recurrence-free and in the cancer-specific survival rates at Stage IIIC (p = 0.2679 and p = 0.0791, respectively).

Figure 2.

Recurrence-free survival. The recurrence-free 5-year-survival rate was significantly higher with a distance of the mesorectal extension (DME) ≤ 4 mm [86.6% (p=0.00015) at Stage IIA (a) and 68.7% (p<0.0001) at Stage IIIB (b)] than with DME > 4 mm. No significant difference was noted between the two groups at Stage IIIC [p = 0.2679, (c)].

Discussion

The 7th edition of TNM staging systems15, 16 is strong prognostic predictors in patients with colorectal cancer. Several reports each from a single institution have also shown that there was prognostic heterogeneity among patients with pT3 rectal cancers1, 3, 5–11 leading investigators to advocate subdivision of category pT3. In 1993, the UICC proposed optional subdivisions for pT3 and pT4 tumors.2 Unfortunately, proposed cut-off points for the DME as a prognostic measurement have varied from 3 to 6 mm among these different studies. A cut-off point of 3 mm for the DME had no prognostic significance,12 whereas a recent multi-institutional study carried out by our group demonstrated that a cut-off point of 4 mm could independently delineate adverse prognosis of pT3N0 rectal cancers (TNM 6th edition).17 This study was focused on pT3N0-2 rectal cancers based on the new TNM 7th staging system.15, 16 Another large multi-institutional study analyzed patient data from the Erlangen Registry for Colo-Rectal Carcinomas (ERCRC) and the Study Group Colo-Rectal Carcinoma (SGCRC) registries.4 The pT3 tumors were subdivided into pT3a (DME ≤ 5 mm) and pT3b (DME > 5 mm), and the prognostic heterogeneity was reported. Another author examined two different patient databases and reported the oncologic outcomes that varied, based on a cut-off point of 6 mm.6 In combination, data from these studies confirm the prognostic heterogeneity associated with different DMEs. However, the clinical significance and validity of the various proposed cut-off points remain unclear. This study utilized ROC curve and Cox regression analyses and demonstrated that a cut-off point of 4 mm produced the most useful predictor of oncologic outcomes in patients with pT3 rectal cancer.

Important risk factors for postoperative overall recurrence are summarized in Table 2. Although lymph node metastasis was the most powerful independent risk factor, DME was also a powerful independent risk factor. However, the circumferential resection margin, number of retrieved lymph nodes, venous invasion and postoperative chemotherapy were not extracted as independent risk factors for predicting outcomes. Therefore, the combination of lymph node status and DME was analyzed to predict prognosis and stratify the TNM 7th staging system.15, 16

Previous studies have reported that local recurrence at Stage II and III following TME for rectal cancer can vary from 4 to 21%18–22 and from 8 to 36%,18, 20, 22 respectively. A multicenter prospective randomized trial organized by the Dutch Colorectal Cancer Group23 reported that the 2-year local recurrence rate after surgery alone with TME was 5.7% in Stage II patients and 15.0% in Stage III patients, which is consistent with findings from this study. Some studies have investigated the relationship between DME stratification and local recurrence. The local recurrence rate was significantly higher in patients with pT3b tumors with a DME > 5 mm (Stage II: 15.4%, Stage III: 34.0%) than in patients with pT3a tumors with a DME ≤ 5 mm (Stage II: 5.5%, Stage III: 17.1%) in the ERCRC cohort.4 However, there was no significant difference when making the same comparison in the SGCRC cohort.4 No significant correlation was found between local recurrence and DME stratification around a cut-off value of 6 mm,6 and this finding is consistent with observations by other investigators.9, 12 However, in patients with overall Stage III rectal cancer, local recurrence seems to be associated with a DME > 4 mm (p = 0.0386, Table 3).

It has been reported that distant metastasis was significantly different when patients were stratified by a cut-off value of 3 mm (< 3 mm, 0% vs. ≥ 3 mm, 46.7%, p = 0.01), although the number of patients in that analysis was relatively small.9 In this study, DME was strongly associated with distant metastasis, even more so than local recurrence. An increased DME is presumably associated with undetectable lymphovascular invasion and microtumor deposits in the mesorectal adipose tissues that increase the risk of local recurrence and/or distant metastases. In the Stage IIIC based on the TNM 7th edition,15, 16 however, distant metastasis and/or local recurrence may occur regardless of the grade of ME because the malignant behavior is more aggressive.

Other authors have also reported that DME was an important predictor of recurrence-free and cancer-specific survival.1, 3, 4, 6 For example, the ERCRC reported that the cancer-related 5-year-survival rate was significantly higher for pT3a tumors than for pT3b tumors (91.2% vs. 77.2% at Stage II and 77.8% vs. 40.3% at Stage III), which was similar to our findings. Other similar outcomes were noted for Dukes B tumors (66% vs. 37%) and Dukes C tumors (30% vs. 18%) at a cut-off value of 4 mm, and for Stage II tumors (73% vs. 52%) and Stage III tumors (40% vs. 27%) at a cut-off value of 6 mm.1, 6 In our statistical analyses, DME was a powerful predictor for stratifying patients in Stages IIA and IIIB. However, DME may be not a useful predictor in patients with Stage IIIC because of extremely advanced disease. Thus, DME is a useful predictor of postoperative recurrence and survival, and improves the utility of the TNM 7th staging system except for Stage IIIC. In addition, these findings raise questions regarding the optimal management of rectal cancer patients with a DME > 4 mm. How does this apply to preoperative and/or postoperative treatments? Willett et al.3 recommended selecting patients with rectal cancer for postoperative adjuvant therapy according to the depth of tumor invasion into the perirectal fat. Diagnostic techniques using MRI enable accurate measurement of the DME, which correlates well with pathological measurements.24, 25 Indeed, use of a cut-off value that can be assessed by preoperative MRI would present an efficient strategy to select patients for pre- and/or postoperative adjuvant treatments including chemoradiotherapy (CRT). In this series, between 1995 and 1999, postoperative adjuvant chemotherapy was given perorally according to the local criteria at each institute.

In European countries, preoperative CRT is the standard strategy for T3 rectal cancer to control local recurrence. In Japan, prophylactic pelvic lymph node dissection has been often performed rather than using preoperative CRT for midlower rectal cancer to control local recurrence.26, 27 Recently, the Dutch Colorectal Cancer Group has reported that preoperative CRT has decreased local recurrence rate to 10.6% in Stage III rectal cancers,28 which is consistent with findings from this study without using preoperative CRT (55/512: 10.7%, Table 3). Moreover, there was no survival benefit in those who received irradiation.28 That is why, neoadjuvant CRT for Stage III rectal cancer has been seldom performed in Japan.

More intensive adjuvant treatments including CRT may be needed for patients with a DME > 4 mm or with Stage IIIC disease to eradicate isolated tumor cells, to prevent postoperative recurrence and to improve survival.

In conclusion, a DME value of 4 mm provides the best prognostic cut-off point to stratify patients with pT3 rectal cancer and predict oncologic outcomes. A subclassification based on a 4-mm cut-off point may improve the utility of the TNM 7th staging system except for Stage IIIC. Intensive chemotherapy is needed for patients with a DME > 4 mm or with Stage IIIC disease. These findings warrant further prospective studies to determine the reliability and validity of this cut-off point.

Acknowledgements

The authors have no conflicts of interest to declare. No external funding was obtained for this study. The authors are grateful to Kenta Murotani from the Department of Biostatistics, Kurume University Graduate School of Medicine, for help with statistical analyses. They also thank the following surgeons and pathologists: Koya Hida, Division of Gastrointestinal Surgery, Department of Surgery, Graduate School of Medicine, Kyoto University; Toru Inoue, Department of Surgical Oncology Osaka City University Graduate School of Medicine; Tomohisa Furuhata, First Department of Surgery, Sapporo Medical University; Tatsuro Yamaguchi, Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital; Tetsuro Higuchi, Department of Surgical Oncology, Graduate School, Tokyo Medical and Dental University; Tomoaki Mizobe, Department of Surgery, Kurume University School of Medicine; Yutaka Ogata, Department of Surgery, Kurume University School of Medicine; Yutaka Kawamura, Department of Surgery, Saitama Medical Center, Jichi Medical University; Toshimasa Yatsuoka, Division of Gastroenterological Surgery, Saitama Cancer Center; Seiichi Shinji; Department of Surgery, Chiba-Hokusoh Hospital, Nippon Medical School; Kimihiko Funahashi, Division of General and Gastroenterological Surgery, Department of Surgery (Omori), Faculty of Medicine, Toho University; Kazuhiko Yoshimatsu, Department of Surgery, Tokyo Women's Medical University, Medical Center East; Fumikazu Koyama, Department of Surgery, Nara Medical University; Takanori Goi, First Department of Surgery, Faculty of Medical Sciences University of Fukui; Shingo Kameoka, Department of Surgery 2, Tokyo Women's Medical University; Wataru Onozato, Kitasato University School of Medicine; Keiichiro Ishibashi, Department of Digestive Tract and General Surgery, Saitama Medical Center Saitama Medical University; Yoshihiro Kakeji, Department of Surgery and Science, Graduate School of Medical Sciences, Kyushu University; Akihiko Kataoka, Department of General Surgery, Hokkaido University Graduate School of Medicine; Yoshiro Kubo, Department of Surgery, National Hospital Organization Shikoku Cancer Center; Shunji Ogata, Coloproctology Center, Takano Hospital; Mitsugu Sekimoto, Department of Gastroenterological Surgery, Osaka University, Graduate School of Medicine; Masaki Kitazono, Department of Surgical Oncology and Digestive Surgery, Kagoshima University Graduate School of Medical and Dental Sciences; Shinichiro Yoshitani, Department of Surgical Oncology, Kanazawa Medical University; Takashi Yao, Department of Human Pathology, Juntendo University School of Medicine; Michiyo Higashi, Department of Human Pathology, Field of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences; Hirokazu Fukui, Department of Surgical and Molecular Pathology, Dokkyo Medical University; Yoichi Ajioka, Division of Molecular and Diagnostic Pathology, Niigata University Graduate School of Medicine and Dental Sciences; Tadakazu Shimoda, Center for Cancer Control and Information Services, National Cancer Center Hospital; and Atsushi Ochiai, Pathology Division, Research Center for Innovative Oncology, National Cancer Center Hospital East.

Ancillary