Definitive Intensity-Modulated Chemoradiation for Anal Squamous Cell Carcinoma: Outcomes and Toxicity of 428 Patients Treated at a Single Institution

Abstract Background Although intensity-modulated radiation therapy (IMRT) is considered the standard of care for the treatment of squamous cell carcinoma of the anus (SCCA), few large series have reported oncologic outcomes and toxicities. In this retrospective report, we aim to describe outcomes and toxicities after IMRT-based chemoradiation (CRT) for the treatment of SCCA, evaluate the impact of dose escalation (>54 Gy), and compare concurrent fluoropyrimidine in combination with either mitomycin or with cisplatin as chemosensitizers. Methods Patients treated at The University of Texas MD Anderson Cancer Center between January 1, 2003 and December 31, 2018 with IMRT-based CRT were included. Median time to locoregional recurrence, time to colostomy, and overall survival were estimated using the Kaplan–Meier method. Results A total of 428 patients were included; median follow-up was 4.4 years. Three hundred and thirty-four patients (78.0%) were treated with concurrent cisplatin and fluoropyrimidine, and 160 (37.4%) with >54 Gy. Two- and 5-year freedom from locoregional failure, freedom from colostomy failure, and overall survival were 86.5% and 81.2%, respectively, 90.0% and 88.3%, respectively, and 93.6% and 85.8%, respectively. Neither dose escalation nor mitomycin-based concurrent chemotherapy resulted in improved outcomes. Mitomycin-based concurrent chemotherapy was associated with in approximately 2.5 times increased grade 3 or greater acute toxicity. Radiation dose >54 Gy was associated with approximately 2.6 times increased Grade 3 or greater chronic toxicity. Conclusions Our results suggest IMRT-based CRT with concurrent fluoropyrimidine and cisplatin is a safe and feasible option for patient with SCCA and may cause less acute toxicity. The role for radiation dose escalation is unclear and requires further study.


Introduction
Although rare, the incidence of squamous cell carcinoma of the anus (SCCA) has continued to rise over the past 20 years. 1,2 Definitive chemoradiation (CRT) is the standard of care due to high cure rates established by two randomized trials, Radiation Therapy Oncology Group (RTOG) 9811 and UK ACT II. 3,4 Cure rates were high overall, but patients with primary tumors >5 cm or locoregional lymph node involvement at presentation had higher rates of locoregional failure (LRF). 5 In ACT II, no statistically significant differences in survival outcomes for patients with locally advanced anal cancer were noted between cisplatin/5-fluorouracil (5FU) and mitomycin C/5FU when used as chemosensitizers with concurrent radiation.
Both RTOG 9811 and UK ACT II utilized 2D/3D conformal radiation techniques, which result in high doses to adjacent normal tissue and significant toxicity. 6,7 The preferred radiation technique shifted to intensity-modulated radiation therapy (IMRT) after the publication of RTOG 0529, which showed IMRT resulted in reduced acute toxicity compared with patients treated with 2D and 3D conformal techniques on RTOG 9811. Additionally, patients treated with IMRT had fewer radiation treatment breaks. Oncologic outcomes were not reported in the RTOG 0529 publication, 8 but a retrospective study of 43 patients using a similar IMRT technique reported 95% 2-year local control, 92% distant metastasisfree survival, 90% colostomy-free survival, and 94% 2-year overall survival (OS). 9 Existing publications on IMRT-based CRT for SCCA are small and/or have limited follow-up. [10][11][12][13][14] Most include a radiation dose of 50-54 Gy and concurrent mitomycin-C (MMC)/5FU as a chemosensitizer. Our aims are to (1) report LRF, colostomy failure (CF), OS, and toxicity data for patients treated with IMRT-based CRT at our institution, (2) evaluate any potential impact of dose escalation (>54 Gy), and (3) evaluate any differences between MMC-based versus cisplatin-based concurrent chemotherapy regimens.

Methods
We obtained a waiver of consent and approval from the institutional review board for this study. All consecutive patients treated at our institution from January 1, 2003 until December 31, 2018 with IMRT-based definitive CRT for nonmetastatic SCCA were included.

Treatment Details
All patients received definitive CRT using an IMRT technique that has been described elsewhere. 13 The primary tumor dose and fractionation were selected based on size; 50 Gy in 25 fractions for T1 tumors, 54 Gy in 27 fractions for T2 tumors, and 58 Gy in 29 fractions for T3 and T4 tumors. The majority of patients received weekly cisplatin (20mg/m 2 intravenously once weekly) and daily 5-FU (300 mg/m 2 /day infused continuously on days of radiation) as previously reported 15 . A minority of patients were treated with MMC (10 mg/m 2 on days 1 and 28). Occasionally, patients received capecitabine (825 mg/m 2 twice daily orally on days of radiation) instead of 5FU. While concurrent cisplatin and 5-FU was the preferred chemotherapy regimen by our multidisciplinary treatment group, MMC was often chosen for patients with baseline renal dysfunction, significant neuropathy or hearing loss. Additionally, patients treated at our some of our regional cancer care centers were co-managed by a radiation oncologist from our institution and a medical oncologist from outside our institution. Patients managed in this way more often received concurrent MMC.
During treatment, all patients were seen weekly for toxicity assessment. Laboratory tests, including a complete blood count with differential, were also obtained weekly. Acute toxicities were graded weekly by the attending physician according to the Common Terminology Criteria for Adverse Events version 4 (CTCAEv4) and reported in weekly treatment management notes in the medical record. After treatment, patients were seen every 3 to 6 months for 5 years.
Toxicities documented up to 6 weeks postcompletion of CRT were recorded as acute toxicities. Toxicities reported thereafter were recorded as late toxicities.

Statistical Analysis
Oncologic endpoints were defined as follows: time to LRF (time from the first day of radiation to either recurrence of disease in the anal canal and/or regional lymph nodes after complete clinical response (cCR) or biopsy-proven persistence of disease at least 6 months after completion of CRT); time to colostomy (from the first day of radiation to colostomy placement either due disease recurrence or treatment-related toxicity); and OS (time from the first day of radiation to the date of death, where applicable). Patients lost to follow-up were considered censored. From these data, median time-to-event outcomes were determined using the Kaplan-Meier method. Tests for univariate time-to-event analysis included the logrank test and Cox proportional hazards regression. Cox regression was used for the multivariate analysis for event endpoints to assess the relationship between prognostic factors and oncologic endpoints of interest. Stepwise regression was used to determine the most informative set of variables, with the Bayesian Information Criterion as the complexitypenalizing criterion. Toxicity endpoints were assessed with multivariate and variable-selected logistic regression models. Standard dose radiation was defined as ≤54 Gy and doseescalated radiation was defined as >54 Gy. P-values of < .05 were considered statistically significant. Software used for analysis was R version 4.0.5 (R Core Team, 2021).  Table 2. In the multivariable model, factors significantly associated with LRF included being HIV positive (HR: 3.146 (95% CI 1.501-6.595); P = .008), being a current smoker (HR: 2.206 (95% CI 1.272-3.825); P = .02) and receiving >54 Gy (HR: 3.348 (95% CI 2.076-5.399); P < .001).

Colostomy Failure
Seven patients required a diverting colostomy prior to treatment initiation due to fistula, obstruction, or pain. Fortyseven patients (11.0%) had a colostomy at last follow-up, either for recurrent or persistent disease (N = 39) or for the management of side effects of radiation (N = 8). Estimated 2and 5-year freedom from colostomy were 90.0% and 88.3%, respectively. Univariate analysis is shown in Table 2. In the multivariable model, only receiving >54 Gy (HR 3.082 (95% CI 1.693-5.610); P < .001) was associated with CF.

Discussion
In this retrospective analysis of 428 consecutive SCCA patients treated with IMRT-based CRT, over 90% achieved a cCR after CRT; at 5 years, freedom from LRF was 80%, freedom  16 , our institution has favored using weekly low dose cisplatin and daily 5FU concurrent with radiation 15 . Although retrospective results are no substitute for a randomized trial to establish the standard of care, we are reassured that our outcomes are similar to smaller reports outlining outcomes after IMRT [10][11][12][13][14] .
T-stage is established as a prognostic factor for LRF and DFS 5,6,17 . Interestingly, radiation dose >54 Gy was a stronger predictor than T-stage for worse LRF, CF, and OS in our analysis. T-stage is colinear with dose in our cohort given our practice of prescribing 50 Gy for T1, 54 Gy for T2, and 58 Gy for T3-T4 tumors. After removing radiation dose from the multivariate models, T-stage was significantly associated with worse outcomes. That dose demonstrated a greater strength of correlation with adverse outcomes than T-stage suggests that residual confounding likely impacted the dose variable. In our practice, there is some variation in radiation dose selection for individual patients. For example, larger T2 tumors with other adverse features may have received >54 Gy, while small (eg, just exceeding 5 cm), more favorable T3/T4 tumors may have received ≤54 Gy. Additionally, patients who received partial excisional may have received radiation doses based on their T-stage at presentation rather than postexcision.
Although we do not believe radiation >54 Gy directly causes worse LRF, CF, and OS, in our study it was also not associated with improved outcomes. Ours is not the first study demonstrating a lack of obvious benefit for dose escalation for advanced tumors 11,18 . We cannot make a clear comment on the benefit of dose escalation based on our data and await the results from PLATO ACT5 which is currently evaluating dose escalation for locally advanced tumors either to 61.6 or 58.5 Gy compared with 53.2 Gy 19 . Additionally, the Eastern Cooperative Oncology Group (ECOG) EA2165 study is evaluating adjuvant nivolumab after CRT as a means of treatment escalation for patients with locally advanced disease 20 .
We found current smoking, living with HIV and receiving an unplanned CRT break were associated with worse oncologic outcomes. Others have also demonstrated the association of smoking with worse LRF and OS 21 . Although studies in the pre-antiretroviral therapy era suggest worse mortality and morbidity for patients living with HIV 22 , modern data have shown comparable outcomes to patients without HIV 23 . Fewer than 5% in our series were living with HIV, which may have influenced our results. Database studies have shown the relationship with prolongation of CRT and worse survival 24,25 . The finding that pre-CRT excision was associated with improved OS is likely due to the fact that receipt of excision is associated with early-stage disease without prognostic significance on its own. Indeed, over 85% of patients who had excision had T1-T2 disease. Although beyond the scope of this manuscript, we acknowledge the controversy regarding the role of local excision in the management of patients with very early stage SCCA 26 .
RTOG 0529 showed lower rates of G3+ acute genitourinary (2%), dermatologic (23%), and gastrointestinal toxicities (21%) in patients treated with IMRT compared with patients treated with conformal radiation on RTOG 9811. In our series, G3+ acute genitourinary and dermatologic toxicities were similar at 2.6% and 21.3%, while rates of G3+ acute gastrointestinal toxicities were lower at 13%. Receiving MMC-containing concurrent chemotherapy was associated with worse acute G3+ non-hematologic toxicities, G3+ neutropenia, unplanned breaks in radiation and hospitalizations. The only factor significantly associated with increased late G3+ non-hematologic toxicities was receipt of >54 Gy. Prospectively collected patient-reported-outcomes are needed for optimal assessment of the toxicity profile, however. Ongoing cooperative group trials are evaluating treatment de-escalation with reduced radiation dose and/or volume as a means of reducing toxicity. 19,27  Limitations of this study include a lack of prospective toxicity collection throughout the duration of the study as well as a lack of patient-reported outcomes. There may also be selection bias of those able to receive treatment at a tertiary cancer center which may lead to improved outcomes. Despite these limitations, this report adds meaningfully to the existing literature, representing the largest modern cohort of patients treated with definitive IMRT-based CRT. Furthermore, unlike previously published studies, most patients in our cohort were treated with concurrent cisplatin/5FU without loss of treatment efficacy.
In conclusion, patients in this retrospective study treated with IMRT and weekly low dose cisplatin and daily 5FU had similar outcomes to those treated with MMC and 5FU but with less acute toxicity, fewer hospitalizations and fewer radiation treatment breaks. Prospective data are needed to validate these observations. In this study, dose escalation >54 Gy did not yield superior outcomes and resulted in increased chronic toxicity. Further study is needed to evaluate whether selective dose escalation may have a benefit patients at increased risk for local recurrence.

Funding
This work was supported in part by the National Institutes of Health through Cancer Center Support Grant P30CA016672.