- © 2011 by American Society of Clinical Oncology
Cause-Specific Colostomy Rates After Radiotherapy for Anal Cancer: A Danish Multicentre Cohort Study
- Kåre G. Sunesen⇓,
- Mette Nørgaard,
- Lilli Lundby,
- Hanne Havsteen,
- Steen Buntzen,
- Ole Thorlacius-Ussing and
- Søren Laurberg
- Kåre G. Sunesen, Lilli Lundby, Steen Buntzen, and Søren Laurberg, Aarhus Hospital; Kåre G. Sunesen and Ole Thorlacius-Ussing, Aalborg Hospital; Kåre G. Sunesen and Mette Nørgaard, Clinical Institute, Aarhus University Hospital, Aarhus; and Hanne Havsteen, Herlev Hospital, Copenhagen University, Copenhagen, Denmark.
- Corresponding author: Kåre G. Sunesen, Department of Gastrointestinal Surgery A, Aalborg Hospital, Aarhus University Hospital, Hobrovej 18-22, PO Box 365, DK-9100 Aalborg, Denmark; e-mail: k.sunesen{at}rn.dk.
Abstract
Purpose In anal cancer, colostomy-free survival is a measure of anal sphincter preservation after treatment with radiotherapy or chemoradiotherapy. Failure to control anal cancer and complications of treatment are alternative indications for colostomy. However, no data exist on cause-specific colostomy rates. We examined this in a cohort study.
Patients and Methods Through national registries and review of medical records, we identified patients with anal cancer diagnosed from 1995 to 2003 who had curative-intent radiotherapy or chemoradiotherapy in four Danish centers. We computed cumulative incidence of tumor-related colostomy and therapy-related colostomy, treating colostomy and death as competing events. Follow-up started at completion of radiotherapy and continued throughout 2008. We used competing risk regression to compute hazard ratios (HRs) to compare the cumulative incidence of cause-specific colostomies between age, sex, tumor size, chemotherapy, and local excision before radiotherapy.
Results We included 235 patients with anal cancer. The 5-year cumulative incidences of tumor-related and therapy-related colostomy were 26% (95% CI, 21% to 32%) and 8% (95% CI, 5% to 12%), respectively. Tumor size greater than 6 cm versus less than 4 cm was a risk factor for tumor-related colostomy (adjusted HR, 3.8; 95% CI, 1.7 to 8.1), and local excision before radiotherapy was a risk factor for therapy-related colostomy (adjusted HR, 4.5; 95% CI, 1.5 to 13.5).
Conclusion After curative-intent radiotherapy or chemoradiotherapy, one third of patients had a colostomy, of which one third were related to therapy. Large tumor size was associated with a higher risk of tumor-related colostomy, whereas history of prior excision was associated with an increased incidence of therapy-related colostomy.
INTRODUCTION
Anal cancer is primarily treated by radiotherapy or chemoradiotherapy with radical surgery by abdominoperineal resection reserved for salvage therapy.1 The dual objective of treatment is to remove the tumor completely while preserving anal function.2 Efficacy of radiotherapy and chemoradiotherapy for anal cancer in terms of long-term survival and tumor control is well established.3–8 Randomized studies and large patient series have demonstrated 5-year overall survival rates of 56% to 75% and 5-year disease-free survival rates of 51% to 60%.5–7,9–18
Colostomy-free survival or cumulative colostomy rates are frequently used end points in clinical studies of radiotherapy or chemoradiotherapy for anal cancer as a measure of anal sphincter preservation.4–7,11,12,14,16–19 The 5-year cumulative colostomy rates are 10% to 29%.18,19 Large tumor size (> 5 cm), high tumor grade (T3/4), and neoadjuvant and concomitant cisplatin-based chemoradiotherapy predict a shorter time to colostomy.14,16,19 Failure to control the anal cancer and complications of treatment are two alternative indications for colostomy after radiotherapy or chemoradiotherapy.3,14,19,20 Up to 15% to 20% of colostomies are necessary because of treatment.3,19 However, minimal, if any, data exist on cause-specific colostomy rates, and existing studies of time to colostomy have not taken different indications for the procedure into account.
Therefore, we examined cause-specific colostomy rates, either tumor related or therapy related, following curative-intent radiotherapy or chemoradiotherapy for anal cancer and evaluated the prognostic impact of clinicopathologic variables in relation to cause-specific colostomy rates.
PATIENTS AND METHODS
Setting and Design
During the study period, oncologic treatment of anal cancer in Denmark took place in six public hospitals, of which four were included in our study. The Danish National Health Service provides tax-supported health care, guaranteeing free access to hospital-based medical care. Unique civil registration numbers, assigned to every Danish citizen, code for age, sex, and date of birth and allow unambiguous individual-level data linkage.21
Study Population
We included patients with anal cancer who were treated with curative-intent radiotherapy or chemoradiotherapy at Aalborg Hospital (Aarhus University Hospital), Aarhus Hospital (Aarhus University Hospital), Herlev Hospital (Copenhagen University), or Vejle Hospital from 1995 to 2003. We identified patients with anal cancer through the Danish Cancer Registry, which contains data on all patients with incident cancer in Denmark since 1943.22 Tumors were classified according to a modified version of the International Classification of Diseases (ICD) seventh edition until the end of 2003 and then according to the 10th edition thereafter. Topography and histology were classified according to the ICD for Oncology (ICD-O) first edition from 1978 to 2003 and then according to the third edition thereafter.22
We considered a tumor as anal cancer if topography was rectum, anal canal, or anorectum (ICD-O1 codes: 1541 to 1548), and histology was invasive squamous cell carcinoma (ICD-O1 codes: 80503, 80513, 80523, 80703, 80713, 80723, 80733, 80743, 80753, 80763, 80943, 81203, 81213, 81223, 81233, 81243, 84303, and 85603).
Through the National Patient Registry (NPR), we identified patients with anal cancer with a hospital contact to a surgical and/or oncologic department at one of the four centers. The NPR contains information about all admissions to nonpsychiatric hospitals in Denmark since 1977.23 The data include civil registration number, hospital and department, dates of admission and discharge, surgical procedure(s) performed, and up to 20 diagnoses from each hospital contact. Diagnoses were classified according to ICD-8 until the end of 1993 and according to ICD-10 thereafter. Surgical procedures were classified according to a Danish classification system until the end of 1995 and according to a Danish version of the Nordic Classification of Surgical Procedures thereafter.23
Through review of the medical records, we identified patients with anal cancer treated with curative-intent radiotherapy or chemoradiotherapy with a total dose of more than 40 Gy. We included patients treated with a sphincter-sparing local excision or excisional biopsy who subsequently received curative-intent radiotherapy or chemoradiotherapy but not patients with postoperative radiotherapy after an abdominoperineal rectal excision (APR) or with radiotherapy intended as a preoperative procedure before an APR. We verified the diagnosis of anal squamous cell carcinoma using pathology reports.
Staging and Treatment
We retrieved data regarding disease characteristics and treatments from medical records. The pretreatment evaluation included physical examination, endoanal ultrasonography and/or magnetic resonance imaging of the pelvis, ultrasonography of the inguinal region, computed tomography or conventional radiography of the chest, and computed tomography or ultrasonography of the liver and retroperitoneum. If suspicious inguinal lymph nodes were encountered, a fine-needle aspiration was performed. Fine-needle aspiration of perirectal lymph nodes was only performed occasionally. We assigned initial disease stage according to the 2002 (sixth edition) American Joint Committee on Cancer TNM staging system for cancer of the anal canal.24 The tumor (T) classification was primarily based on the recorded physical examination. Location of the anal cancer was classified as anal margin or anal canal including overlapping lesions. Node (N) classification was based on all available information from the initial physical, radiologic, and pathologic examinations.
Data on prescribed and received radiotherapy doses, radiotherapy technique, and date of the end of radiotherapy were retrieved from radiotherapy charts. Information on chemotherapy included the type of chemotherapy (neoadjuvant, concomitant, or both), chemotherapy drugs, and adherence to the prescribed treatment.
Outcome Data
Through the medical records, we identified patients who after anal cancer diagnosis had a surgical procedure involving creation of a colostomy or ileostomy, hereafter collectively referred to as colostomy because of the rarity of the latter. We additionally reviewed all surgical procedures recorded in the NPR to identify patients who had a colostomy at another hospital. We retrieved information on date, type, and cause of the surgical procedure from hospital records and pathology reports. We classified a colostomy as tumor related, including pretreatment colostomy, if the patient had evidence of anal cancer at time of surgery and as therapy related if the colostomy was created after radiotherapy and the patient had no evidence of cancer. A colostomy created as a part of an APR was classified as therapy related if there was no evidence of malignancy in the resection specimen. Vital status (dead or alive) and date of death or emigration for the patients in our cohort were retrieved from the Civil Registration System, which contains daily updated information on all residents.21
Statistical Analysis
We constructed cumulative incidence curves of tumor-related colostomy, therapy-related colostomy, and death assuming each of these events to be competing causes of failure.25 Patients were observed from the date of the end of radiotherapy until tumor-related colostomy, therapy-related colostomy, death, emigration, or December 31, 2008, whichever came first. A pretreatment colostomy was considered a tumor-related colostomy on day 0. The 5-year cumulative incidence of tumor-related colostomy and therapy-related colostomy was computed with corresponding 95% CIs.25
We used Fine and Gray26 competing risks regression to estimate the effect of age, sex, tumor size, nodal status, and chemotherapy on the cumulative incidence of tumor-related colostomy and the effect of age, sex, tumor size, and local excision before radiotherapy on the cumulative incidence of therapy-related colostomy. The relative risk or subdistribution hazard ratio (HR) for a categorical covariate was the ratio of subdistribution hazards for the actual group with respect to the baseline, with all other covariates being equal.26,27
The cmprsk package of the statistics software R (http://www.R-project.org) was used for competing risk analyses.25,27 We used Stata/MP 10.1 (Stata, College Station, TX) for other analyses. The study was approved by the Danish Data Protection Agency and the National Board of Health in Denmark.
RESULTS
Patient Characteristics and Oncologic Treatment
We identified 235 patients with anal cancer treated with curative-intent radiotherapy or chemoradiotherapy. The demographics and clinical characteristics are listed in Table 1. Median ages at date of the last radiotherapy treatment were 61 years (range, 37 to 89 years) among 186 women (79%) and 57 years (range, 32 to 84 years) among 49 men (21%). Median maximum tumor diameter was 40 mm (range, 5 to 130 mm), and 28 patients (12%) had infiltration of adjacent organs. Fifty-nine patients (25%) had involvement of regional lymph nodes including inguinal lymph nodes in 44 patients (19%). Three patients had unilateral external iliac node metastasis, and one patient had vulvar metastasis.
Demographics and Clinical Characteristics of Patients With Anal Cancer Who Received Curative-Intent Radiotherapy or Chemoradiotherapy
The treatments actually received are listed in Table 1. Treatment was primarily based on computed technology. Radiotherapy was most often external-beam radiotherapy (EBRT) with prescribed dose to the tumor targets of 2 Gy per fraction, 5 fractions per week, to a total dose of less than 60 Gy in five patients (2%), 60 to 64 Gy in 190 patients (81%), and more than 64 Gy in six patients (3%). Only one patient with a prescribed EBRT dose of ≥ 60 Gy received less than 60 Gy. Elective radiotherapy of 46 to 51.2 Gy to regional lymph nodes was given to 184 patients (78%), whereas 17 patients (7%) received radiotherapy to tumor without covering regional nodes. Usually, four-field technique was used for uninvolved pelvic and inguinal nodes, and three-field technique was used for pelvic nodes. Of 160 patients with EBRT and uninvolved inguinal nodes, 121 patients (76%) had inguinal irradiation. This volume usually comprised pelvic nodes proximally to the promontory, inguinal nodes distally to the level of trochanter minor, and tumor with a minimum distal margin of 2 cm. Tumor and involved lymph nodes were treated with a consecutive boost, concomitantly, or both. EBRT never included a gap.
EBRT (44 to 46 Gy, 2 Gy per fraction) to tumor and pelvic nodes followed by brachytherapy (15 to 20 Gy to a total dose of 61 to 66 Gy) was prescribed in 32 patients (14%); brachytherapy was given interstitially in all except one patient who received intracavitary brachytherapy. Median time between end of EBRT and start of brachytherapy was 10 days (range, 1 to 36 days). Two patients (1%) had only interstitial brachytherapy and received the prescribed doses of 50 Gy and 55 Gy.
In patients with T3-4 or N2-3 (N1 anal margin) tumors, the proportion who received chemotherapy increased from nine (38%) of 24 patients diagnosed from 1995 to 1997 to 29 (69%) of 42 patients diagnosed from 2001 to 2003. The types of chemotherapy are listed in Table 1.
Outcome
The cumulative incidences of tumor-related colostomy, therapy-related colostomy, and death are shown in Figure 1. The median follow-up times were 5 years (range, 0 to 14 years) in all patients and 7 years (range, 4 to 14 years) in survivors without a colostomy. The 5-year incidences of tumor-related colostomy and therapy-related colostomy were 26% (95% CI, 21% to 32%) and 8% (95% CI, 5% to 12%), respectively. The combined 5-year incidence of colostomies and death was 49% (95% CI, 43% to 56%).
Cumulative incidence of tumor-related colostomy, therapy-related colostomy, and death after curative-intent radiotherapy or chemoradiotherapy for anal cancer.
A total of 57 (89%) of 64 patients received their tumor-related colostomy within 2 years after the end of radiotherapy, and 61 patients (95%) received their colostomy within 5 years. Tumor-related colostomies included 20 pretreatment colostomies, 34 colostomies that were part of an APR, and 10 colostomies that were not part of an APR. The main indications for a pretreatment colostomy were vaginal involvement (n = 6), circumferential tumor (n = 4), anorectal abscess (n = 4), and disturbances in anorectal function (n = 5); the indication was unknown in one patient. One year after the end of radiotherapy, only eight patients with pretreatment colostomies were alive with no residual or recurrent anal cancer and no APR performed. This included the only patient with reversal of a pretreatment colostomy who had a second colostomy as a result of fecal incontinence.
All 18 therapy-related colostomies were created within 5 years after radiotherapy, and 15 therapy-related colostomies (83%) were created within 2 years after radiotherapy. In six patients, the therapy-related colostomy was part of an APR with no evidence of malignancy in the resection specimen. These APRs were performed 2 to 6 months after radiotherapy as a result of bothersome local symptoms and suspicion of residual tumor. Indications for 12 therapy-related colostomies not part of an APR included chronic anorectal ulcer (n = 4), fecal incontinence (n = 3), bowel obstruction (n = 2), rectovaginal fistula (n = 1), perianal skin injury (n = 1), and anal canal fibrosis (n = 1). The therapy-related colostomy was reversed after 22 months in a patient who had spontaneous healing of the perianal skin and after 20 months in a patient who had excision of fibrotic anal tissue followed by several plastic reconstructive procedures.
Tumor size ≥ 6 cm was a risk factor for tumor-related colostomy, compared with tumor size less than 4 cm (adjusted HR, 3.8; 95% CI, 1.7 to 8.1; Table 2). Tumor size ≥ 4 to less than 6 cm was associated with a nearly two-fold increased risk of tumor-related colostomy compared with tumor size less than 4 cm; however, the precision of the estimate was low (adjusted HR, 1.9; 95% CI, 0.9 to 3.7). On the basis of small numbers, we observed a substantial increased risk of therapy-related colostomy among patients who had a local excision or excisional biopsy before radiotherapy (adjusted HR, 4.5; 95% CI, 1.5 to 13.5). Age, sex, and nodal status were not associated with either tumor-related or therapy-related colostomy.
Crude and Adjusted Subdistribution HRs of the Effect of Risk Factors on the Cumulative Incidence of Tumor-Related Colostomy and Therapy-Related Colostomy
DISCUSSION
The colostomy rate is considered a relevant end point in anal cancer because it captures both locoregional failure and toxicity compromising anal sphincter preservation. We conducted a cohort study to assess the cumulative incidence of cause-specific colostomy after radiotherapy or chemoradiotherapy for anal cancer and found that a considerable proportion of patients required a colostomy, mostly as a result of the tumor. Risk factors for colostomy varied according to the indication for the procedure.
The existing studies of time to colostomy after radiotherapy for anal cancer–handled patients who died without a colostomy in different ways, rendering direct comparisons of results difficult. Thus, some studies4,11,16 estimated colostomy-free survival assuming both colostomy and death without colostomy as failures, whereas other studies5,12,14,17 only considered colostomy as a failure. The latter strategy assumed that patients who died without a colostomy would have had the same risk of colostomy as the patients still alive, had they not died. However, this assumption is unlikely to be true considering that within the first years after treatment, both death and colostomy are most often caused by persistent or recurrent anal cancer.3 A few other studies treated death as a competing risk of colostomy.4,6,18,19 Using this method, the US Gastrointestinal Intergroup Radiation Therapy Oncology Group (RTOG) trial 98-11 reported 5-year incidences of colostomy of 19% and 10% after cisplatin-based and mitomycin-based chemoradiotherapy, respectively.19 In comparison, our 26% 5-year incidence of tumor-related colostomy alone seemed high. However, a recent nonrandomized study reporting 5-year incidences of colostomy of 22% after cisplatin-based chemoradiotherapy and 29% after mitomycin-based chemoradiotherapy suggests that our finding simply reflects outcome of treatment in the course of routine clinical practice.18
Chemoradiotherapy is considered standard treatment for anal cancer primarily because of the colostomy-sparing potential. Still, only few studies have actually examined prognostic factors for colostomy.4,14,16,19,28 In a secondary analysis of 644 patients in the RTOG 98-11 trial, Ajani et al19 found tumor size of more than 5 cm to independently predict a higher 5-year cumulative incidence of colostomy, whereas neither nodal status nor sex was associated with colostomy. This finding was supported by two small observational studies.14,28 In contrast, nodal involvement, but not tumor classification, was associated with a higher 4-year cumulative incidence of colostomy in the RTOG/Eastern Cooperative Oncology Group trial (28% v 13% for nodal v no nodal involvement, respectively).4 We extended previous findings by demonstrating that tumor size was a strong predictor for the colostomies related to tumor, whereas no clear association was noted between tumor size and the cumulative incidence of therapy-related colostomy.
On the basis of small numbers, we found that patients with a local excision or excisional biopsy before radiotherapy had a more than four-fold higher incidence of therapy-related colostomy after radiotherapy compared with patients without these procedures. This association has been noted by other authors as well. Allal et al29 found that colostomy, alone or as part of an APR, was more frequent in patients who had a local excision or excisional biopsy before radiotherapy for anal cancer compared with patient who did not (five of 19 patients v five of 125 patients, respectively). In other studies, local excision of small anal cancers before radiotherapy or chemoradiotherapy did not appear to decrease local failure.30,31 Taken together, these findings suggest that a local excision should not be the primary therapeutic option, even for small tumors.
Most of our patients received a high radiotherapy dose (median, 64 Gy) without chemotherapy. Consequently, results may not apply to patients treated with chemotherapy and lower radiotherapy doses.4,6 The ACCORD-03 trial found no difference in 3-year colostomy-free survival for advanced anal cancer treated with chemoradiotherapy (45 Gy) plus a 15-Gy boost compared with a higher boost dose.7 Our finding of an association between large tumor size and increased incidence of tumor-related, but not therapy-related, colostomy, suggests that toxicity did not outweigh efficacy in patients with advanced disease despite high radiotherapy doses. Among patients with smaller tumors, in particular those with a local excision or excisional biopsy, reducing radiotherapy dose could perhaps reduce morbidity.32,33 However, because 97% of patients received a radiotherapy dose of ≥ 60 Gy, we could not examine this further.
A pretreatment colostomy was created in 9% of patients in our study, which is within the range of 4% to 13% reported in previous studies.10,14,34 None of the patients who had a pretreatment colostomy in our study ultimately became free from colostomy. This is similar to the findings in a Dutch study on 83 patients with anal cancer, of whom seven (8%) had a colostomy created before treatment and only one was permanently reversed.14 In another study from the United Kingdom, seven of 35 patients had their pretreatment colostomy reversed within 6 to 30 months after curative-dose chemoradiotherapy.34 A wider indication for a pretreatment colostomy could explain this higher proportion of patients with subsequent colostomy reversal.34 Nevertheless, the conclusion remains that colostomies created before initiation of radiotherapy or chemoradiotherapy for anal cancer are rarely permanently reversed.14,34
Our study regarding cause-specific colostomy rates had some strengths and limitations. We conducted one of the larger cohort studies of anal cancer, although our sample size was still limited. To obtain this sample size, we had to include patients during a 9-year period in which treatment changed a great deal. Hence, because of the rather heterogeneous treatment regimens, it was not possible to include specific treatment variables in our analyses. Our main strength was our use of nation-wide registries allowing complete follow-up for the competing outcomes of death and colostomy.
In conclusion, one third of patients with anal cancer had a colostomy after curative-intent radiotherapy or chemoradiotherapy. Most colostomies were tumor related, although one third were therapy related. Large tumor and prior local excision were associated with an increased risk of tumor-related and therapy-related colostomy, respectively. Patients should be informed that a pretreatment colostomy is usually permanent.
AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
The author(s) indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS
Conception and design: Kåre G. Sunesen, Mette Nørgaard, Lilli Lundby, Steen Buntzen, Ole Thorlacius-Ussing, Søren Laurberg
Collection and assembly of data: Kåre G. Sunesen, Hanne Havsteen
Data analysis and interpretation: Kåre G. Sunesen, Mette Nørgaard
Manuscript writing: All authors
Final approval of manuscript: All authors
Footnotes
-
Supported by the Danish Cancer Society.
-
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
- Received March 31, 2011.
- Accepted June 7, 2011.
