Abdominoperineal Resection is Associated with a Poor Outcome Compared with Anterior Resection for Rectal Cancer: A Meta-Analysis

Research Article

Austin J Radiat Oncol & Cancer. 2015; 1(3): 1014.

Abdominoperineal Resection is Associated with a Poor Outcome Compared with Anterior Resection for Rectal Cancer: A Meta-Analysis

Yen-Chien Lee1*, Chung-Cheng Hsieh2 and Jen- Pin Chuang1

1Department of Oncology, Tainan Hospital, Ministry of Health and Welfare, Tainan, Taiwan

2Department of Cancer Biology, University of Massachusetts Medical School, Worcester, UK

*Corresponding author: Yen-Chien Lee, Department of Oncology, Tainan Hospital, Ministry of Health and Welfare, ROC.No. 125, Jhongshan Rd, Tainan, Taiwan

Received: November 16, 2015; Accepted: December 01, 2015; Published: December 03, 2015


Background: It’s controversies whether Anterior Resection (AR) is superior to Abdominoperineal Resection (APR) for rectal cancer. To solve this question, a meta-analysis was performed.

Methods: Studies published from 1990 to May 2014 evaluating outcomes after AR and APR for rectal cancer were identified by an electronic literature search. A meta-analysis was performed to compute Hazard Ratio (HR) along with 95% Confidence Intervals (CI). We extracted HRs of Disease-Free Survival (DFS) and Overall-Survival (OS) compared these two methods using a randomeffects model.

Results: Twenty four studies met the inclusion criteria, yielding a total of 21,221 patients. Pooled adjusted HRs were 1.276 (95% CI, 1.266 -1.287) for DFS and 1.130 (95% CI, 1.126-1.133) for OS that compared APR with AR. Sensitivity analyses were performed and APR was still associated with poor prognosis.

Conclusion: APR has been associated with poor prognosis compared with AR. While possible, a more radical operation, extralevator abdominoperineal resection should be considered.

Keywords: Rectal cancer; Meta-Analysis; Abdominoperineal resection; Disease free survival


Colorectal cancer is the third most commonly diagnosed cancer in males and second in females [1]. It is also one of the leading causes of cancer mortality worldwide, and rectal cancer accounts for 30 to 35% of these cases [1].

Miles, who introduced Abdominoperineal Resection (APR) in 1908, based his work on that the spread of tumors of the lower rectum occurred through the lymph nodes outside the levator ani muscles. APR was thought to address this problem. Anterior Resection (AR) was later introduced for proximal rectal cancers. Improvements in stapling technology over the following decades have resulted in an increase in the number of sphincter-saving operations [2]. The more important recent advance in rectal cancer surgery has been the Total Mesorectal Excision (TME). TME is defined as sharp pelvic dissection between the parietal and visceral planes of the pelvic fascia and it has been introduced since 1982 by Heald [3]. TME and precise perimesorectal plane dissection were introduced in all the major countries of Europe in 1995 [4]. Local recurrence rate can decrease to less than 10% for middle and lower rectal carcinomas with this procedure [5]. Following several workshops and operative demonstrations around the world, TME is now generally accepted as the standard procedure for rectal cancer [6].

Previously APR was thought to be able to eradicate border field than AR and achieve better outcomes in disease free survival or even overall survival. Others suggest otherwise [7-10]. It’s not clear if the worse outcome associated with APR is a result of surgical procedure or patient- and tumor-related factors. Many studies have tried to explicate these relationships. Possible reasons for high local recurrence of APR compared with AR include the higher incidence of inadequate margin [11] or lymph node involvement that may follow a different pattern in low rectal carcinomas. Very low cancers have a dual lymphatic drainage [12]. Tumor or rectal perforation are other important factors that might contribute to poorer outcomes in APR.

However, due to tumor nature and anatomic reason, further randomized-controlled trials would not be feasible to compare these two methods. This question has been further complicated by different neoadjuvant, adjuvant, and chemotherapy agents. Since prospective randomized studies comparing APR with AR are ethically unavailable, we try to answer this question from previous studies through metaanalysis. This is the first meta-analysis that attempts to answer this question.


Search strategy and Selection criteria

We systematically searched Pub Med and Medlines for relevant articles published up to May12, 2014, with the following search terms, “anterior” and “rectal cancer, “abdominoperineal” and “rectal cancer” without language restriction. Additional search methods included manual review of the reference lists of relevant studies. Studies were included if the following were true: (1) they were published with extractable information on disease free survival (DFS) or overall survival (OS); (2) they were primary rectal cancer. Chemotherapy was not used in patients with rectal cancer during the late 1990s. Whether adjuvant was used or not are included as well. Studies before 1990 were excluded. Kaplan-Meier DFS or OS graph with three or fewer outcome events was considered as inadequate and was excluded. Only original articles were included; posters, abstracts, and conference reports were excluded.

Information on age and stage were extracted when available and taken into account in the analysis. Abdominoperineal excision was often involved positive circumferential resection margins [13] which has been associated with a poor prognosis [14]. Due to variable factors adjustments in each studies, if any one of stage, age or margin were included in original studies for adjustment, the HR were considered as multivariate analysis.

We assessed the methodologic quality of included studies using the Newcastle-Ottawa Scale [15] for determining the quality of individual cohort studies included in the meta-analysis.

Statistical analysis

We extracted HR by direct or indirect methods [16] based on the available information provided in the selected articles. If information for applying these methods was not available, a graph methods, which stratifies the published survival curve into nonoverlapping intervals to estimate the log HR, was used to extract the HR [16]. Summary HRs were obtained using the random-effect model. Forest plots were used to display the study-specific HRs and the summary estimates. Heretorgenetiy among the studies was tested with Q statistics. Analyses were conducted with Stata version 12.0 (Stata Corp, College Station, TX). Two-sided p values <0.05 were regarded as statistically significant.


Literature search

The steps of our literature search are shown in (Figure 1). After reviewing titles and abstracts, 104 potential papers were extracted to review in full. Two papers [17,18] were from two different but overlap periods of the same hospital. Due to one paper provided information to overall survival and the other to disease-free survival without double weighting. Both of the papers were included.

Four articles were excluded due to language problem [19-22]. Four other articles were excluded due to unavailable in our setting [23-26]. Two of the four were from the same Chinese journal. As for the rest, one is excluded due to a short report [27]. One is excluded due to abdominosacral resection instead of abdominoperineal resection [28]. One is excluded due to use chi-square test for survival data [29]. Some paper with hazard ratio of local recurrence only [30]. Two paper is excluded due to less or equal to 3 events and was unable to extract hazard ratio from survival curve [31, 32]. Two were excluded due to more than 3 survival curve and unable to extract [33,34]. Others were excluded due to no extractable hazard ratio (HR) of DFS or OS. Methodological quality scores ranged from 4 to 9 on a scale of 10 (Table 1). With regard to the quality of the studies, 21 trials (87.5%) showed a high-quality score (6 points). The distribution of total scores for the 24 studies were as follows: 4 (1 study), 5 (2 studies), 6 (2 studies), 7 (4 studies), 8 (12 studies) and 9 (3 studies). Because most of the studies received a score of 6 or above and the standard criteria have not been established, we considered the studies to be of adequate quality for the analysis.