Impact of VMAT-IMRT compared to 3D conformal radiotherapy on anal sphincter dose distribution in neoadjuvant chemoradiation of rectal cancer

Background Neoadjuvant radio- or chemoradiation (nIRT) therapy is the standard treatment for loco-regional advanced rectal cancer patients of the lower or middle third. Currently, intensity modulated radiation therapy (IMRT) is not the recommended radiation technique even though IMRT has advantages compared to 3D-radiation regarding dose sparing to organs at risk like small bowel and urinary bladder. So far, the benefit of IMRT concerning the anal sphincter complex is not examined. With this study we intended to evaluate the dose distribution on the anal sphincters of rectal cancer patients treated with IMRT in comparison with 3D-techniques. Methods We selected 16 patients for the IMRT-group and 16 patients for the 3D-group with rectal cancer of the middle third who were treated in our institute. All patients received 45 Gy in a chemoradiation protocol. Patients in both groups were matched regarding stage, primary tumor distance to the anal verge and size of the tumor. We delineated the internal and external anal sphincters, the addition of both sphincters and the levator ani muscle in all patients. Subsequently, we evaluated and compared dose parameters of the different sphincters in both groups and analysed the configuration of the isodoses in the area of the caudal radiation field, respectively. Results Most of the relevant dose parameters of the caudal sphincters (Dmean, Dmedian, V10–V40) were significantly reduced in the IMRT-group compared to the 3D-group. Accordingly, the isodoses at the caudal edge of the target volume in the IMRT group demonstrated a steep dose fall. The levator ani muscle always was included into the planned target volumes and received the full dose in both groups. Conclusions The modern VMAT-IMRT can significantly reduce the dose to the anal sphincters for rectal cancer patients of the middle third who were treated with conventional chemoradiation therapy.


Background
Either neoadjuvant short term radiation therapy (nRT, 5 × 5 Gy) or conventional chemoradiation (nCRT) are the standard treatment protocols for patients with locally advanced rectal cancer (UICC-Stage II or III) of the lower and middle third [1]. In the last years, total mesorectal excision (TME) became the standard of treatment and can provide very good local control rates [2,3]. Even though local control rates were improved by TME alone, nIRT can significantly reduce the number of loco-regional relapses [2,4]. Next to improved local control sphincter preservation rates can be increased by nCRT especially for tumors of the lower third [2,[5][6][7][8][9][10][11].
Due to improved dose sparing of organs at risk (OAR), intensity modulated radiation therapy (IMRT) has become the sole or equivalent standard in tumors of the pelvic which are treated with (neo)adjuvant, definitive or palliative radiation [12,13]. This technique is more conformal than conventional radiation techniques. Several studies have compared IMRT of rectal cancer to 3D-conformal radiation therapy. IMRT is usually associated with less dose to the rectum, small bowel and the urinary bladder [14][15][16][17][18]. This also translated into better clinical outcome measured by grade 2 but also ≥3 acute gastrointestinal toxicity, genitourinary toxicity and skin side effects [19,20]. Despite these advantages of IMRT, due to marginal misses in the first experiences of IMRT in rectal cancer, up to now, 3D-RT is still the recommended technical standard of treatment, even though IMRT is acceptable for special cases that do not meet dose constraints with 3D-RT [21].
In the past it was shown that sphincter function worsened by conventional adjuvant or neoadjuvant radiotherapy of rectal cancer [22][23][24][25][26][27]. With neoadjuvant radiation before surgery, fecal incontinence was almost twice as high compared to surgery alone [26]. Though, unfortunately, studies which compared IMRT to 3D-radiation did not investigate the dose to anal sphincters. Furthermore, in previous published IMRT-studies, sphincter function (fecal incontinence) was not considered as a clinical endpoint [19,20]. Though, not in rectal cancer, but in prostate cancer, the relationship between dose and sphincter function in IMRT treated patients was performed [28]. Moreover, the correlation of dosimetric parameters of the anal sphincter with sphincter function in patients with locally advanced rectal cancer treated with nCRT was demonstrated [29]. The evidence for standardized constraints is low. The RTOG 0630 phase II trial, dealing with soft tissue sarcoma, required that less than 50% of the anus should receive 30 Gy [30].
So far, the impact of IMRT on nCRT treated rectal cancer patients regarding dose distribution of the anal sphincters has not been investigated. In addition, up to now IMRT has not been the recommended standard technique for treatment. In this study, we evaluated the effect of intensity modulated radiotherapy versus conventional 3D irradiation on the dose distribution of the anal sphincters to estimate the risk of anal incontinence.

Patient selection and radiation techniques
Between 2008 and 2016 a total of 154 rectal cancer patients were treated with neoadjuvant radiation therapy in our institution, 106 with IMRT (either RapidArc (88) or TOMO (28)) and 48 with 3D-RT. To compare IMRT with 3D-RT, we selected the subgroup which might has the greatest possible benefit of sphincter sparing. These are patients with rectal cancer of the middle third (5-12 cm) that were irradiated with a conventional regimen (total dose 45 Gy, single dose 1.8 Gy). In order to make the two groups as homogeneous as possible and thus comparable, we included only such patients in the study who had T3N+ disease.
Patients in the IMRT-group were treated on a Varian Clinac® DHX linear accelerator (Varian Medical Systems, Palo Alto, CA, USA) using IGRT (imageguided radiotherapy) with kilo-voltage cone-beam-CT scans (CBCT). Regularly 2 arcs were used (6 or 15 MV). Some patients received TOMO-therapy with the TomoTherapy Hi-ART-System (6 MV) (Accuray, Sunnyvale, USA). The dose was prescribed to the median of the PTV (ICRU83). Contouring and treatment planning for the intensity modulated radiation therapy was performed using the Eclipse 13.0 planning system (Varian Medical Systems, Palo Alto, CA, USA). Contouring was done on planning CT scans with 3 mm slice thickness. Patients in the 3D-group underwent chemoradiation between 2008 and 2012. All of those patients were treated with ONCOR -Digital Medical Linear Accelerator by Siemens and planned with Oncentra MasterPlan software Version 3.0 SP1, usually 4 fields were used. Daily image guidance was performed using 2D-MV imaging from 0°and 90°. The dose was prescribed to the reference point (ICRU50/62). Dose constraints for organs at risk (OAR) (femoral heads, small bowel, sigmoid colon, genitals and urinary bladder) orientated on Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) [31]. There were no constraints for the anal sphincter(s) in both groups.
In all patients, the clinical target volume (CTV) definition generally based upon the recommendations of the RTOG Consensus Panel Contouring [32]. In all patients, next to the primary tumor, the whole mesorectum, the peri-rectal-, pre-sacral-and internal iliac nodes were always included into the CTV. As recommended, the visible external iliac nodes were included in four patients in the IMRT group and three patients in the 3D-group. The lower border was at least 2 cm caudal of the gross disease. Further 0.7-1.0 cm were added to reach the PTV margin.

Contouring and definition of the anal sphincters
For dose-evaluation we defined the volume of the sphincters in two different ways: Most of the studies that analysed dose distribution to the anal sphincters used a unified approach ("sphincter region") and combined the internal and external anal sphincters as one, mostly filled volume [28,29,33]. The cranial border was usually 1 cm below the ampulla recti. The entire anal canal including the anal mucosa/skin was contoured, (CT hyperdensity). For comparison, we defined the internal as well as the external sphincter as "anal sphincter" (AS) in the same way.
As a second, anatomically more detailed approach, we defined the internal anal sphincter (IAS), the external sphincter (EAS), and the levator ani muscle according to the following rules ( Fig. 1 For the delineation of the sphincters, a matched MRIscan helped with most individuals, especially for the demarcation of the muscle to the mucosa or skin. For correct comparison, the final contouring was performed on the planning CT-scan in both groups.

Evaluation of dose and data
We examined the absolute max dose (D98%), absolute min dose (D2%) and the absolute mean and median dose to the entire volume of the different sphincters, respectively. Furthermore, we evaluated and compared the relative dose parameters (V10, V20, V30, V40, V45). In addition, we quantified the differences in dose distribution to the sphincters between the two groups by measuring the longitudinal distance between the 95 and 10% isodoses at the caudal level of the planar centre of the anal canal. For all dose parameters and the longitudinal distance between the isodoses, a Mann-Whitney U test was performed with SPSS 25.0 (SPSS Inc., Chicago, IL, USA) to identify significant differences between the IMRT-and the 3D-group (independent samples). With a p-value < 0.5, we assumed that there is a significant difference between the two groups.

Results
Between 2008 and 2016 a total of 154 rectal cancer patients were treated with neoadjuvant radiation therapy in our institution. Seventy-two (47%) of these patients received a nCRT-concept with a total dose of 45 Gy (single dose 1.8 Gy). About two thirds of this group (49) had a rectal cancer of the middle third. Sixteen patients who fulfilled these criteria were treated with 3D-RT and further 16 patients were matched to this group regarding tumor extension and TNM. This means that finally, a total of 32 rectal cancer patients of the middle third with UICC stadium III (T3N+) treated with long term nCRT between 2008 and 2016 were included and compared in this study. All patients received concomitant chemotherapy, with either 5-Fluoruracil (5FU) on day 1-4 and 29-32 with 1000 mg/qm body surface area (BSA) or Capecitabine 825 mg/qm BSA twice a day, 5 days per week during radiation. The main patients and tumor characteristics are listed in Table 1. The two groups were well balanced regarding tumor extension, distance from the anal verge and volume of the sphincters. The Dmax (D98%), Dmin (D2%), Dmean and Dmedian of the AS (internal + external muscle + mucosa) were significantly reduced in the IMRT-group compared to the 3D-group ( Table 2). The mean Dmax, Dmean and Dmedian in the IMRT group were about 10 Gy lower than in the 3D-group, whereas the mean Dmin was almost the same in both groups (~46 Gy). This indicates that at least a small part of the anal sphincter is always covered by the planned target volume (PTV). The dose distribution inside the volume of the anal sphincter shows the effect very clearly (Table 3). From 5 to 30 Gy, the volume of the anal sphincter in the IMRT-group decreased noticeably (99 ➔ 69.4%), whereas the volume changed just marginally from V30 to V45 (76.6 ➔ 73.4%). Furthermore, the V10, V20, V30 and V40 were significantly smaller in the IMRT-group. The dose distribution in the 3D-group changed inversely to the IMRT-group. A large portion of the volume was covered by up to 40 Gy (90.9%) with a break in the maximum high-dose range (45 Gy) (58%), which means that the V45 is significantly lower in the 3D-group than in the IMRT-group (p = 0.003).
This pattern of dose distribution was also observed with the internal and external anal sphincter. In the IMRT-group, the volume covered by 10 to 40 Gy was significantly lower, whereas it was significantly higher for 45 Gy.
The situation was different with the levator ani, which is located more cranial than the anal sphincter. It was   IMRT Intensity modulated radiation therapy, 3D 3D-conformal radiation therapy boldfaced p-values are statistically significant almost completely covered by the planned target volume in both groups, which has the consequence that the V40 was > 97% in both groups. Only the V45 (prescription dose) was significantly higher in the IMRT-group (97%) compared to the 3D-group (66%). This indicates homogeneous and therefore more adequate dose coverage of the PTV in the IMRT-group. In summary, except the area where the anal sphincter overlaps with the PTV (prescription dose), the volume of the sphincter (AS, IAS, EAS) can be significantly spared with IMRT. In the IMRT-group, the Dmean, Dmedian and V10-V40 of the caudal anal sphincters were reduced by about one quarter.
The detailed evaluation of the isodoses in the area of the caudal PTV and sphincter complex confirmed the results of the dose distribution on the lower anal sphincters. Overall, there was a steep dose gradient in the IMRT-group compared to the 3D radiation-group. Basically in this caudal area of the PTV at the transition to the sphincters, the dose coverage of the PTV was much more homogeneous and better than in the 3D-group. This is a reasonable explanation for significantly higher V45 values in the IMRT-group. Logically, the distance between the different isodoses in the 3D-group was much wider than in the IMRT-group. In the IMRT-group, the longitudinal distance of the 95% isodoses to the 10% isodoses was in mean 1.87 cm (1.59-2.04 cm), while in the 3D group it was 2.4 cm (2.1-2.77 cm) (Fig. 2). This difference was highly significant (p = 0.001). The largest distance in the IMRT-group (2.04 cm) was still below the shortest distance in the 3D-group (2.1 cm).

Discussion
We were able to show that in IMRT-treated rectal cancer patients of the middle third most dose parameters (Dmean, Dmedian, V10-V40) for the lower anal sphincter were significantly reduced using IMRT compared to 3D-technique. The mean Dmean of 30.5 Gy of the EAS in the IMRT-group is equal to the results of Chen et al. for 3D-treated patients (28 Gy), whereas the Dmean in our 3D-group is significantly higher (40.6 Gy) [33]. These differences can be explained by the higher mean tumor location of the patients in the cohort of Chen et al.

Dose and sphincter function
Long-term functionality of the anal sphincters correlates with the dose to the sphincter complex. Most studies with large patient collectives, in which sphincter dose distribution was correlated with sphincter function, dealt with prostate cancer. The dose prescription in a treatment of prostate cancer is usually much higher (68-78 Gy) than of a long term nCRT concept of the rectum (45 Gy). In an analysis of 414 patients treated with 70 Gy to the prostate or postoperative region, Alsadius et al. could show that long-term fecal leakage largly increases with ≥40 Gy compared to a mean dose < 40 Gy [28]. Buettner et al. quantified dose distribution to the anal-sphincter region and sphincter control of 388 prostate cancer patients treated with 3D-technique [34]. Dose-surface maps were created to estimate the dose of the surface of the anal canal. Furthermore, DVHs and DSHs were determined. Subjective sphincter control correlated significantly with the dose of the sphincters. Optimal cut-points of 47 Gy to the anal sphincter and 45 Gy to the anal surface were found. The relative risk of unsatisfactory subjective sphincter control was also higher for patients receiving ≥30 Gy. One study was found dealing with sphincter function and the correlation of dose in nCRT treated rectal cancer patients. In this study, Arias et al. compared dose distribution and fecal incontinence in rectal cancer patients treated with chemoradiation. Patients with V20 > 0 had significantly more often poor sphincter function (Mean Wexner score of 5.5) compared to those with V20 = 0 (p = 0.008). In median the V20 was 34.9% (4.5-85.2%). The authors recommend a sphincter dose constraint of V20 = 0, if possible. In our study, the median V20 was 85.8% in the IMRT-group and 100% in the 3D radiation-group. The lowest V20 in the IMRT-group was 14.9% although we used the same method in contouring of the AS. We assume that the great difference in the V20 is related to the differences in the patient population. Arias et al. selected patients, who received sphincter-preserving surgery and who were more than 2 years free of overall relapse at the time of the study. In our study the majority of the patients had tumors starting from 5 to 6 cm from the anal margin, which resulted in an overlapping of the PTV with the cranial aspect of the AS in most of the cases. Nevertheless, a big difference between the two techniques occurred in V20 of the AS (IMRT: 80.1% vs. 3D: 96.5%; p = 0,004). Ultimately, V20, V30 and V40 of the AS, which were found to be relevant in prostate cancer studies, were significantly reduced in the IMRT group compared to 3D-technique.

Target volume definition and the effect on the anal sphincters
In rectal cancer, a high inter-observer variability in the definition of the clinical target volume in rectal cancer patients exist [35][36][37]. The caudal border and the ischioanal fossa are two of the structures which usually deviate most from contouring standards [37].
The fact, that loco-regional recurrences of rectal cancer usually occur posteriorly or laterally [38,39] to the target volume, gives rise to the problem, that in rectal cancers, which are located near to the sphincter complex, an adequate CTV margin is needed. Subsequently, this results in a subtotal coverage of the anal sphincters by the PTV. On the other hand, especially if sphincter preservation is intended, a maximal protection of the sphincters has to be aimed. The common RTOG Consensus Panel recommendations define a caudal extension of the CTV of 2 cm to gross disease with an inclusion of the entire mesorectum to the pelvic floor, even in upper rectal cancers [32]. Roels et al. published recommendations for CTV delineation in advanced rectal cancer based on an analysis of local recurrences and lymph node involvement. They defined pelvic subsites for loco-regional recurrence. The inferior pelvic subsite, which includes the anal sphincter complex, was at risk of locoregional recurrence, when the tumor was within 6 cm from the anal margin or in those patients who were treated with abdominoperineal resection. For the reasons stated above the authors recommend an inclusion of this region in the CTV for patients with a tumor located ≤6 cm from the anal verge or in a situation where the tumor infiltrates the anal sphincter complex and an abdominoperineal resection has been planned. The current consensus guidelines (ESTRO35) differ for this caudal area. They recommend that the ischiorectal fossa should only be included in case of invasion of the external anal sphincter or the ischiorectal fossa [40].
The levator ani muscle is located more cranially to the caudal sphincter complex and is mostly completely covered by the PTV. This leads to a Dmean of about 45 Gy and a V40 of about 98% no matter which of the radiation techniques is used. Because of its cranio-caudal extension along the perirectal fat tissue, this structure is located in a high risk area for loco-regional recurrence. This means, that dose sparing is not desired or useful and more conformal radiation techniques couldn't improve any protection of this structure.
In our study, the majority of the patients in both groups had cancers starting from 5 to 6 cm above the anal verge. Analogous to the recommendations of Roels et al., in these patients a caudal extension of 2 cm plus about 1 cm PTV margin results either in an overlap or a total coverage of the caudal sphincter muscles with the PTV (Fig. 3).
Regarding dose coverage of anal sphincters, two groups of middle third rectal patients can be identified. In patients, where the primary tumor is in close proximity to the anal sphincter complex, the sphincters should be included into the CTV. This was the case in exactly 8 of 16 patients in each group. The AS was completely covered by the previous PTV, which leads to a V40 of about 100% of the caudal anal sphincters. These patients do not benefit from sphincter protection by IMRT. In patients with a primary tumor starting 1 to 2 cm above the sphincter (also 8 of 16), the sphincter complex only partially overlaps with the PTV. These patients benefit greatly from a steep dose drop of IMRT compared to the 3D-irradiation (Fig. 3). The significant improvement of Dmin, Dmean as well as the V10-V40 in the IMRTgroup compared to the 3D-group is supported by the fact that the 8 patients with total anal sphincter inclusion were part of the comparison. As it is known, that pathologic complete response is associated with higher radiation dose and finally improved outcomes, IMRT provides another advantage of local dose escalation (boost) without increased toxicity [41]. In rectal cancer patients of the middle third, this reduced target volume has often some distance to the sphincters and offers another opportunity of sphincter sparing.

Limitations
The main limitation of this study was the retrospective character. Furthermore, to make a comparison as accurate and meaningful as possible, many of the treated rectal cancer patients could not be included into the study. Unfortunately, this ultimately led to a relatively small number of patients for the final comparison. However, we were able to make the two groups very homogeneous, which ultimately led to a good comparability, in which significant differences in the dose at the sphincters could be shown.

Conclusion
In patients where the anal sphincter complex does not have to be included into the clinical target volume, IMRT can significantly reduce the mean dose distribution at the anal sphincters for rectal cancer of the middle third that were treated with long-term chemoradiation therapy.