|Home | About | Journals | Submit | Contact Us | Français|
To develop a Radiation Therapy Oncology Group (RTOG) atlas of the elective clinical target volume (CTV) definitions to be used for planning pelvic intensity-modulated radiotherapy (IMRT) for anal and rectal cancers.
The Gastrointestinal Committee of the RTOG established a task group (the nine physician co-authors) to develop this atlas. They responded to a questionnaire concerning three elective CTVs (CTVA: internal iliac, pre-sacral and peri-rectal nodal regions for both anal and rectal case planning; CTVB: external iliac nodal region for anal case planning and for selected rectal cases; CTVC: inguinal nodal region for anal case planning and for select rectal cases), and to outline these areas on individual computed tomography images. The imaging files were shared via the Advanced Technology Consortium. A program developed by one of the co-authors (IEN) utilized binomial maximum-likelihood estimates to generate a 95% group consensus contour. The computer-estimated consensus contours were then reviewed by the group and modified to provide a final contouring consensus atlas.
The panel achieved consensus CTV definitions to be used as guidelines for the adjuvant therapy of rectal cancer and definitive therapy for anal cancer. The most important difference from similar atlases for gynecologic or genitourinary cancer is mesorectal coverage. Detailed target volume contouring guidelines and images are discussed.
This report serves as a template for the definition of the elective CTVs to be used in IMRT planning for anal and rectal cancers, as part of prospective RTOG trials.
Intensity-modulated radiotherapy (IMRT) enables the delivery of complex radiation therapy (RT) plans that previously could not be accomplished with conventionally planned two to four field techniques or more sophisticated three-dimensional (3D) conformal RT (3D-CRT). The advent of IMRT provides an opportunity to spare critical normal tissue, which is of significant importance for cancers that are managed with concurrent chemoradiation. For patients receiving pelvic radiation for anal or rectal cancer, normal tissue can often be better protected with IMRT than other conformal techniques as demonstrated by dosimetric investigations 1–8. IMRT has also been associated with reduced acute toxicity for anal cancer 6, 9, 10.
Critical to the use of IMRT, is a clear understanding of the elective CTV targets. IMRT demands much more detailed knowledge of target structures than the conventionally planned two to four field technique. The “four-field box” technique is based on bony landmarks and does not lend itself to customized conformal treatment planning based on an individual’s anatomy. Target volumes for anal and rectal cancer also differ substantially from those appropriate for gynecologic (GYN) or genitourinary cancer (GU). The most striking differences arise from the need for proper coverage of the peri-rectal and pre-sacral regions. While the rectum and its associated mesentery are avoidance structures for GYN or GU malignancies, they represent first echelon drainage for both the anus and rectum.
The elective CTV atlas that follows was produced by a consensus panel of nine radiation oncologists (RA, PD, MG, LG, TH, LK, JK, RM, CW) who were assigned by the Gastrointestinal Committee of the Radiation Therapy Oncology Group-RTOG. Co-chairs were MG, LK, RM. The formation of this panel was motivated, in part, by what was felt to be inadequate contouring in a large number of cases enrolled on RTOG 0529 (A Phase II Evaluation of Dose-Painted IMRT in Combination with 5-Fluorouracil and Mitomycin-C for Reduction of Acute Morbidity in Carcinoma of the Anal Canal), 11. Because of a rapid submission and review process, patient care was not compromised, but an educational need was identified. This report provides the recommendations of this consensus panel, and serves as a template for the definition of the elective CTVs to be used in IMRT planning for anal and rectal cancers.
After the development of the RTOG anorectal contouring consensus panel, multiple informal discussions were held by both electronic mail and telephone conference to develop three elective CTVs for the purpose of this atlas. CTVA, defined as the internal iliac, pre-sacral and peri-rectal nodal groups; CTVB: the external iliac nodal region, and CTVC: the inguinal nodal region.
For anal cancers, the elective regional target volume would include all three. For rectal cancers, in most cases, CTVA would be the only volume to receive elective radiation. However, for certain presentations, discussed below, one could consider adding CTVB and even CTVC.
Ultimately, a sample set of computed tomography (CT) images were distributed to the members of the panel, along with a questionnaire to answer further detailed questions concerning the three elective CTVs. The case utilized for this atlas was a clinical T3 N2 rectal cancer located 7 cm above the verge. The case was selected because of the presence of multiple peri-rectal lymph nodes, without major distortion of the mesorectum. For reasons unrelated to the rectal cancer, this case also had multiple small inguinal and external iliac lymph nodes, which inform the process of outlining CTVB and CTVC. The patient was simulated in the prone position (with a styrofoam bowel displacement device 12, 13 incorporated into the alpha cradle), with a flexible endorectal tube placed at the distal edge of palpable disease and a skin marker placed 4 cm below the distal edge of palpable disease.
It should be noted that the use of the prone position is not mandatory in the RTOG anal cancer protocols. Nonetheless, it should be considered in cases where diagnostic imaging in the supine position demonstrates a substantial volume of small bowel near the target volumes. Similarly the use of small bowel contrast and the placement of markers for the distal edge of disease, while not mandatory, can be very helpful. The identification of the iliac vessels is also important. Unlike this patient, who had arterial calcifications and a clear fat plane between the ileopsoas muscles and the iliac vessels, many patients could also benefit from IV contrast at the time of simulation.
The consensus generating process consisted of answering the questionnaire as well as contouring the three elective target structures. The imaging files were shared via the Advanced Technology Consortium, with each participant using his/her own treatment planning system to contour. A software program developed by one of the co-authors (IEN) utilized the binomial distribution to generate a 95% confidence level group consensus contour. The computer-estimated consensus contours and questionnaire responses were then reviewed by the group at a formal consensus conference sponsored by the RTOG and held in San Diego on January 17, 2008. At this meeting, a general consensus regarding the three CTV’s was obtained, and one of the chairs (RM) has presented the panel with final modified contouring images.
We have developed a MATLAB (Mathworks Inc., Natick, MA) software program for estimating consensus contours from given individual experts contours. The software is an imputation method utilizing Expected Maximum (EM) algorithms for simultaneous truth and performance level estimation (STAPLE) 14. In this approach, the true contouring decisions at each voxel are formulated as maximum-likelihood estimates from the observed contours by optimizing sensitivity and specificity parameters of each expert’s performance using the EM algorithm, assuming a binomial distribution. Estimated sensitivities/specificities for the different CTVs are listed in Table 1. We estimated overall agreement using generalized kappa statistics 15. The goodness of agreement was evaluated according to Landis and Koch criteria 16. According to these criteria, the agreement on CTVA was considered substantial and on CTVB and CTVC was considered moderate (see Table 1. and discussion below).
In Figure 1, the individual panel submissions are superimposed and displayed on four representative slices. In general, there is good agreement on the location of the core portions of the mesorectum, as well as the iliac and femoral vessels. For CTVA, differences between individual contourers were primarily a matter of margin. Therefore, for this target volume, the group agreed to accept the computer generated consensus contours. For CTVB and C, the panel felt that small lymph nodes, if present, should be incorporated into the elective target volume, even if it was felt that they were probably uninvolved reactive nodes. Including nodes that were not covered by some of the members led to modifications of the computer generated contours. The agreed upon extensions was primarily into the lateral inguinal regions. The resulting consensus CTV contours are displayed in Figure 2. Specific CTV definition details, as decided through this consensus process and other planning recommendations are described.
The caudad extent of this elective clinical target volume should be a minimum of 2 cm caudad to gross disease. In addition there should be coverage of the entire mesorectum to the pelvic floor (located ~ at slice −41.08 in the sample case) even for upper rectal cancers. It should be understood that, after accounting for PTV margin and beam penumbra, a 2 cm CTV margin is equivalent to a ~ 4 cm block edge margin. For anal canal cancers, the 2 cm CTV margin should extend at least 2 cm around the anal verge into apparently normal peri-anal skin (in RTOG 0529, the requirement is 2.5 cm). For cancers with peri-anal skin involvement, the CTV margin should extend at least 2 cm beyond areas of involvement.
For this sample case of a mid-rectal cancer, the caudad extent of CTVA is at slice −42.58, because that slice is 2 cm caudad to the distal extent of palpable disease (demarcated by the tube placed at simulation). If this had been an upper rectal cancer, the caudad extent of CTVA would be at the pelvic floor (i.e. CTVA would extend down to slice −41.08).
Unless there is radiographic evidence of extension into the ischiorectal fossa, the group felt that extension of CTVA does not need to go more than a few millimeters beyond the levator muscles. For very advanced anal or rectal cancers extending through the mesorectum or the levators, the panel’s recommendation is to add ~1–2 cm margin up to bone, wherever the cancer extends beyond the usual compartments. Similarly, if the tumor is invading a neighboring organ (T4 disease) CTVA should include a 1–2 cm margin around the identified areas of invasion. An MRI and/or PET/CT scan is strongly recommended in such cases.
In the mid-pelvis, CTVA includes the rectum and its mesentery, the internal iliac region, and a margin for bladder variability. The posterior and lateral margins of CTVA should extend to pelvic sidewall musculature or, where absent, bone. Anteriorly, the consensus group recommended extending CTVA ~1 cm into the posterior bladder, to account for day-to-day variation in bladder position 17.
The panel also recommended including at least the posterior portion of the internal obturator vessels (which lie between the external and internal iliacs in the mid pelvis) with CTVA.
The recommended superior extent of the peri-rectal component of CTVA was whichever is more cephalad: the rectosigmoid junction or at least 2 cm proximal to the superior extent of macroscopic disease in the rectum/peri-rectal nodes. The full length of the rectum should always be in CTVA, but more of the large bowel should be incorporated, if the 2 cm proximal margin requires that.
The most cephalad extent of CTVA will be higher than the rectum, in order to properly cover the internal iliac and pre-sacral regions. The most cephalad aspect of CTVA should be where the common iliac vessels bifurcate into external/internal iliacs (approximate boney landmark: sacral promontory). At midline, CTVA should extend at least 1cm anterior to the sacrum, to provide proper coverage of the pre-sacral region.
For all areas of CTVA, care should be taken to avoid contouring into uninvolved bone. In general the CTVA contour does not extend into uninvolved pelvic sidewall muscles but does include the levators. Note that when small bowel fell into the region normally taken by rectal mesentery (see slices −30.88 to −34.18 of this sample case) the panel opted to keep it in the elective target volume CTVA—since the location of small bowel could vary from day to day.
In principle, the day-to-day variation of adjacent organs like the bladder and bowel should be incorporated in the PTV margin. It is more practical, however, to assign a uniform PTV margin and account for physiologic variability by adjusting the CTV. The panel opted for this approach.
The consensus group felt that elective coverage of the inguinal and external iliac regions should be routine for anal carcinoma. There was some disagreement as to the indications for covering these regions for rectal carcinomas. For rectal carcinomas extending into GYN or GU structures (the most common manifestation of T4 disease), the panel agreed that the external iliac region should be added (elective nodal coverage = CTVA + CTVB for these cases). Some, but not all, of the panel would also include the external iliacs for rectal cancers that extend into the anal canal. Similarly, the group was divided on whether to electively irradiate the inguinal nodal region for rectal adenocarcinomas that extend to the anal verge, peri-anal skin, or lower one-third of the vagina.
The consensus panel recommended that the caudad extent of the inguinal region (CTVC) should be 2 cm caudad to the saphenous/femoral junction. The transition between inguinal and external iliac regions (CTVC to CTVB) is somewhat arbitrary, but the group recommended that this should be at the level of the caudad extent of the internal obturator vessels (approximate boney landmark: upper edge of the superior pubic rami).
The group recommended at least a 7–8 mm margin in soft tissue around the iliac vessels, but one should consider a larger, 10+ mm, margin anterolaterally—especially if small vessels or nodes are identified in this area. The inguinal/femoral region should be contoured as a compartment with any identified nodes (especially in the lateral inguinal region) included. If using automated expansions, the CTVs should be trimmed off uninvolved bone and muscle.
The consensus panel opted not to include boost target volumes in this atlas. It was felt that boosts might be defined differently in different protocols, might be affected by evolving imaging capabilities, and might vary sharply between patients. The group did recommend that any boost CTV extend to the entire mesorectum and pre-sacral region at involved levels, including ~1–2 cm cephalad and caudad in the mesorectum and ~2 cm on gross tumor within the anorectum.
Similarly this atlas does not present planning target volumes (PTVs). It was generally agreed that the PTV margin should be ~0.7 to 1.0 cm, except at skin, where planning system requirements mandate it be trimmed to ~2–5 mm within the skin surface.
Although normal tissues were not contoured in this atlas, there were several general recommendations by the consensus panel. The femoral head and neck should be avoidance structures. The small bowel and large bowel are important structures to consider when planning treatment. To avoid unnecessary contouring effort, they only need be contoured to ~ 1 cm above the PTV. This, in turn, implies that absolute volume of bowel (in cc) is more important than relative volume (in %). Otherwise, cases with good exclusion of small bowel from the pelvis (e.g. with a bowel displacement device) would be unfairly penalized.
The group felt that it is important that dose volume histograms (DVH) be consistent from one contourer to the next. Therefore, we recommended that bowel be contoured tightly, rather than with a broad, ill-defined margin. It is recognized that the location of bowel could vary from one day to the next, but the dose volume histogram (DVH) from the simulation should remain representative. It was suggested that a broader avoidance structure could be used for IMRT planning purposes (e.g. anterior pelvic contents above the bladder and ~ 1 cm outside the PTVs), while the tightly contoured bowel would remain the structure evaluated in DVHs.
With regard to large bowel, it is very important to recognize that all of the rectum and much of the rectosigmoid will be part of CTVA and, therefore, should NOT be treated as an avoidance structure. Therefore, it is recommended that “uninvolved colon,” defined to be that part of the large bowel outside the CTVs, be contoured separately from rectum.
On the other hand, if small bowel happens to lie within a CTV, the CTV is NOT modified and the portion of small bowel that fell within the target volume is NOT extracted from the DVHs, as discussed above.
The panel opted not to set specific DVH recommendations for normal tissue, because this is still under active investigation. There are differing DVH constraints in the two recent RTOG IMRT protocols: 0529 (anal carcinoma) and 0822 (rectal carcinoma) (www.rtog.org). In 0822, major deviations are generated if criteria for bladder, small bowel, and femoral heads are exceeded. In 0529, bladder is not a deviation generating structure but criteria for femoral heads and small bowel are stricter.
Research into the potential benefits of IMRT in the treatment of anal and rectal cancer has only recently been undertaken. In terms of definitive chemoradiation for anal canal cancer, several recent studies, including RTOG 98-11, support the administration of higher radiation doses in the treatment of anal cancer in attempts to improve local-regional disease control 18–21. The current standard dose recommendations with conventional external beam RT range from 45 Gy for early lesions to 59.4 Gy for T2–T4 disease. The combination of these increased radiation doses with 5-fluorouracil (5FU) and mitomycin-C therapy have resulted in encouraging local control rates, but have been tempered by significant acute morbidity, most notably grade 2+ hematologic, dermatologic and gastrointestinal, as well as late toxicity including femoral head and neck fractures 18, 19, 22. IMRT in the treatment of anal canal cancer has tremendous potential for reducing these toxicities, while allowing for these higher radiation doses to the gross tumor volume. With IMRT, radiation dose to the normal structures (particularly small bowel, skin, femoral heads, bladder and external genitalia) is reduced compared to conventional 2D and more sophisticated 3D treatment planning 1, 5, 6. Small pilot series show IMRT in this population to be well tolerated, with most patients experiencing only mild to moderate acute symptoms, and few patients experiencing treatment breaks 6, 9, 10. Moreover, the use of IMRT has not compromised elective target coverage, as locoregional control in these reports appears favorable, albeit with limited follow-up.
Concerning preoperative chemoradiation for clinically staged muscle invasive or node positive rectal cancer, 5FU-based chemoradiotherapy has been associated with moderate rates of acute gastrointestinal toxicity 23–25. More recently, the addition of irinotecan to 5 FU and radiation has yielded impressive pathologic complete response rates, but at the expense of marked small bowel morbidity 26. The risk of grade 3 or greater GI toxicity has been shown to increase with both radiation dose and volume of irradiated bowel, for rectal cancer patients undergoing radiation in combination with 5FU or 5FU and oxaliplatin 27, 28. Therefore, IMRT could potentially reduce the radiation dose to the small bowel and, consequently, reduce gastrointestinal side effects for rectal cancers. Dosimetric analyses comparing 3D-CRT and IMRT planning have shown that IMRT improved small bowel sparing without compromising target volume coverage 2–4, 7, 8. Moreover, IMRT in the prone position with a bowel displacement device was more effective than IMRT in the supine position at minimizing the radiation dose to small bowel 4. To date, there is only one clinical report concerning the clinical use of IMRT for rectal cancer 29. This phase I effort examined the mean tolerated dose of preoperative radiotherapy using IMRT and an integrated boost (2 Gy to the gross tumor volume daily) with concurrent capecitabine in patients with locally advanced rectal cancer. Eight patients completed radiation at the initial dose level of 55 Gy, but the study was discontinued because six grade 3 toxicities occurred in the eight, with no patient achieving a pathologically complete response. Details of IMRT target volumes and planning directives were not described.
One of the significant hurdles facing implementation of IMRT for pelvic malignancies has been the complexity of target and elective lymphatic definition. The standardization of the CTV target volume definition will not only provide an important basis for the prospective study of IMRT for anorectal malignancies in the multi-institutional setting, but will also establish contouring guidelines for the radiation oncology community, if IMRT proves efficacious in reducing normal tissue toxicities while not compromising outcome. Herein, the RTOG anorectal contouring consensus panel demonstrated good concordance in their CTV definitions. This is understandable, considering that the panel members are physicians who specialize in the delivery of radiation for these malignancies. However, this was not the case in RTOG 0529 (A Phase II Evaluation of Dose-Painted IMRT in Combination with 5-Fluorouracil and Mitomycin-C for Reduction of Acute Morbidity in Carcinoma of the Anal Canal), despite the availability of perhaps a less robust anal contouring atlas 11. As part of this effort, real-time quality assurance of all IMRT treatments was successfully performed prior to treatment start by the principal investigators. Many cases required revisions due to incomplete coverage of the peri-rectal and pre-sacral regions (mesorectum).
While this report serves as the first contouring guideline consensus for anal cancer, there has been one previous publication providing target volume directives for rectal carcinoma 30. The RTOG panel recommendations are quite consistent with the CTV definitions of target structures for rectal cancer provided by Roels and colleagues 30. However, there is some disagreement on the anterior border for the mesorectum. The RTOG anorectal group was more generous on this border to account for day-to-day variability in the location of structures immediately anterior to the rectum (including bladder, gynecologic structures, and low-lying loops of small bowel and sigmoid colon) 17. In addition the group felt that failure in the ischiorectal fat in the absence of frank invasion on presentation was very infrequent and opted for more conformal coverage of the low anorectum.
This is the first report of the Radiation Therapy Oncology Group anorectal consensus panel. The guidelines and images should serve as a template for the definition of the elective clinical target volumes to be used in conformal planning for anal and rectal cancers. As there is no long-term outcome data demonstrating the efficacy of IMRT in this setting, this atlas will be utilized as a contouring guideline for two prospective IMRT trials, RTOG 0822, A phase II evaluation of preoperative chemoradiotherapy utilizing IMRT in combination with capecitabine and oxaliplatin for patients with locally advanced rectal cancer, and a developing phase II follow-up study to RTOG 0529, incorporating cetuximab with standard 5FU and mitomycin-C for anal canal cancer. In these studies, particular attention will be warranted to the patterns of recurrence, to ensure that these CTV consensus panel recommendations, and the use of IMRT for the management of anorectal cancers, are appropriate.
LAK and RJM co-wrote the manuscript. Supported by RTOG U10 CA21661, CCOP U10 CA37422, Stat U10 CA32115 NCI grants. The contents are the sole responsibility of the authors and do not necessarily represent official views of the NCI. Advanced Technology QA Consortium supported by NIH/NCI U24 Grant CA81647.
Conflict of Interest Notification
There are no conflicts of interest associated with this manuscript by any of the authors.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.