Although definitive chemoradiotherapy with mitomycin C and 5-FU remains the standard of care for SCCA of the anal canal,9-11 the acute and late toxicities of treatment with standard pelvic radiotherapy can be significant.12 In the RTOG 98-11 randomized trial, a 74% rate of grade 3-4 nonhematologic toxicity was reported for the overall cohort.13 In addition, patients in the mitomycin C arm had a 61% rate of grade 3-4 hematologic toxicity. The current study aimed to determine whether IMRT could reduce treatment toxicity and improve outcome for anal cancer patients by comparing IMRT patients to a group treated with CRT from a single institution. To our knowledge, this is the first article to directly compare non-IMRT with IMRT in this disease site from a single institution.
Salama et al have previously reported on IMRT in a pooled experience from the University of Chicago and Mayo Clinic on 53 patients. In that study, with a median follow-up of 14.5 months, the investigators found that 37% of patients experienced grade 3 dermatologic toxicity and 15% experienced grade 3 GI toxicity.7 In a study by Pepek et al from Duke University, 47 patients were treated with IMRT with a median follow-up of 14 months. Their study reported that no patients with SCCA of the anal canal experienced grade 3 dermatologic toxicity. However, 16% did experience grade 3 GI toxicity (Table 2).6 Although significant, both of these studies have toxicity rates that compare favorably with those from RTOG 98-11. Ultimately, the results of the RTOG 0529 phase 2 trial will provide prospective data on rates of acute toxicity in patients treated with IMRT and concurrent 5-FU/mitomycin C.
Our study had a median follow-up that was twice as long as previously reported studies (32 months for IMRT patients). Our results confirmed reduced toxicity rates compared with those reported in RTOG 98-11. When compared with a cohort of patients treated with CRT from the same institution, we found significantly reduced rates of grade 3-4 acute nonhematologic toxicity. Table 3 shows how the current series compares to prior studies. Also noteworthy is the finding that, in the current study, even when higher doses were delivered in the IMRT group to the regional nodes (Table 1), treatment toxicity was still reduced. This finding further highlights the advantage of IMRT for dose escalation to improve outcome without increasing treatment toxicity.
An important limitation within this study is that fact that, despite the use of a standardized toxicity grading scale, significant subjectivity existed in retrospectively grading toxicity during treatments. As a result, the objective measurement of treatment breaks, duration of treatment breaks, and overall treatment time were also examined. The results showed that the reduction in toxicity was associated with a decrease in treatment breaks (34.5% vs 88%, P = .001), reduced treatment break duration (12 days vs 1.5 days, P < .0001), and an overall reduction in treatment time (57 vs 40 days, P < .0001). Salama et al reported that 41% of patients treated with IMRT required treatment breaks,7 whereas Pepek et al reported treatment breaks in 18%,6 both of which are consistent with the current study. As a caveat, it must be noted that although these numbers are objective, the decision to give a treatment break is far from objective, and different physicians will invariably have different thresholds at which to offer a break. It is likely that given the awareness of the detrimental effects of interrupting radiation, especially in disease sites such as the head and neck, and cervix,14-18 physicians are less willing to offer breaks due to toxicity.
Dosimetrically, we found that IMRT enhanced normal tissue sparing. Our dosimetric results coincide with the findings of other investigators.5, 19, 20 This improvement in normal tissue sparing appears to have lead to reduced grade 3 or 4 nonhematologic toxicities in the IMRT group compared with the CRT group.
The median time to complete clinical response was 6 weeks, and only 12% of patients had complete response at the end of therapy. The length of time before complete regression was a maximum of 56 weeks. Findings from other series have demonstrated similarly long regression times. Cummings et al reported a range of 2-36 weeks for complete clinical tumor response.21 Similarly, Schlienger et al found a mean time to complete regression of 3 months, with the longest extending to 1 year.22 These results support continued observation of persistent abnormalities at the completion of chemoradiation, because regression will continue. It is our policy to observe patients with frequent clinical examinations unless signs of progression are noted, at which point biopsy and additional workup are pursued. It is less clear how to manage patients with persistent abnormalities that neither improve or worsen. In this study, 4 of 12 patients with persistent abnormal clinical examination findings had positive biopsies. Imaging studies such as FDG PET could be helpful in assessing residual active disease.
We found improved 3-year OS (88% vs 52%, P < .01), LRC (92% vs 57%, P < .01), and PFS (84% vs 57%, P < .01) in the IMRT group compared with the CRT group. These excellent outcomes are similar to those reported by the other IMRT studies. Indeed, the outcomes in the CRT groups are worse than historical results.10, 11, 13 Although there was no significant difference in TNM stage between the 2 groups, the CRT group had a trend toward more T3-4 disease (47% vs 28%, P = .18), which could have impacted local control. However, the percentages of N+ patients (29% CRT vs 24% IMRT) and overall stage I-II patients (53% CRT vs 66% IMRT) were similar. The chemotherapy regimen was similar as well.
Whereas IMRT should not be dosimetrically superior to CRT in terms of target coverage, the advantage with IMRT is likely that it can reduce overall treatment time by decreasing normal tissue doses resulting in significantly reduced toxicity and treatment duration, thereby possibly improving outcome. In this study, we found that prolonged treatment times and interruptions in treatment were associated with poorer outcomes, similar to previous reports.11, 23-26 However, other studies have shown that treatment breaks have not been associated with worse disease control or survival.27, 28 A recent pooled data analysis of 937 patients treated on RTOG 87-04 and RTOG 98-11 provides more insight on the role of treatment time.29 This analysis found no correlation between duration of radiation therapy and local control. Conversely, total treatment time (stratified as ≤53 days vs >53 days), which was prolonged particularly in the subset of patients from RTOG 98-11 that received neoadjuvant chemotherapy, was associated with higher local failure and colostomy rates. Although this study strongly suggests that neoadjuvant chemotherapy can have a detrimental effect on disease control, presumably by delaying chemoradiation and allowing accelerated repopulation and chemotherapy resistance, delays during chemoradiation may be of less importance. This study offers some reassurance that treatment breaks during radiotherapy may be given without a significant impact on treatment outcome. However, it should be noted that in our patient population, none of whom received induction chemotherapy, the median treatment duration for patients in the CRT group was 57 days compared with 40 days for the IMRT group, while the median RT time in the RTOG pooled analysis was 45 days. Given the finding of worse local control with total treatment duration of >53 days in the latter study, the large disparity in treatment time between the IMRT and CRT groups in the current study could still have partly contributed to the differences in outcome. Therefore, we believe that caution should still be applied when allowing prolonged breaks.
Certainly technical factors could have contributed to the differences in outcome. First of all, we observed that relatively few patients were treated during the CRT time period (17 patients over the course of 10 years). Perhaps the experience of the treating physician could have impacted the quality of the treatment delivered.
Second, the CRT patients were treated during an earlier time period before the widespread use of PET or image-guided radiotherapy to account for setup uncertainty. It is possible that PET-based treatment planning in IMRT patients could have identified previously undetected regions of disease involvement, allowing proper dose delivery.
Finally, in addition, the dose delivery to the pelvic nodes differed. In the CRT group, the upper pelvic nodes received <40 Gy in 76% of patients compared with 38% in the IMRT group. Also all IMRT patients, with 1 exception, received 45 Gy to the lower pelvic nodes versus only 59% of CRT patients. These dose differences might have led to better disease control in the IMRT group. Indeed, 2 nodal failures were observed in the CRT group compared with none seen in the IMRT group. However, one was an inguinal nodal failure, and the other had simultaneous pelvic and para-aortic failures. Currently, there is no evidence that doses higher than 36 Gy are needed for these elective nodal regions. We, therefore, do not believe differences in dose or using PET-based planning had a significant impact in the locoregional outcomes difference. However, we believe further studies on the optimal dosing for anal canal cancer and the use of PET for treatment planning are necessary.
Even though this study has the inherent limitations of being retrospective with small patient numbers and potential imbalances in prognostic factors, our results support further investigation of IMRT for this disease site. The results of RTOG 0529 are awaited to provide survival data in a group of patients treated prospectively with IMRT. Although outcomes are generally excellent, locoregional failure remains a problem, particularly for larger tumors, and dose escalation could be used selectively in those at higher risk. In addition, given the potential morbidity of treatment IMRT could significantly diminish toxicity and improve the therapeutic ratio. Longer follow-up is needed to accurately determine late effects with this relatively new technique.