The estimated 3-year rates of locoregional control, distant control, disease-free survival, and overall survival were 81%, 88%, 67%, and 84%, respectively.

Treatment

All patients were treated with definitive concurrent chemoradiation, without a planned treatment break. The median dose of radiotherapy was 5500 cGy (range, 1800–7029 cGy), with 95% of patients receiving between 5040 and 6040 cGy. The median duration of radiotherapy was 42 days (range, 16–87 days). The following technique was used in 153 (92%) patients: Patients were treated initially with anterior and posterior pelvic fields with a dose of 3060 cGy in 180-cGy fractions. The superior border of the pelvic field was placed at the bottom of the sacroiliac joints before the year 1999, and at the L5/S1 interspace from 1999 onwards, based on a departmental policy change. The lateral borders covered the medial inguinal nodes, and the inferior border was placed at least 3 cm below the inferior border of the tumor or the anal verge. Patients were then treated with a three-field technique, with posterior, right lateral, and left lateral fields, with a cumulative dose of either 5040 cGy (before 2003) or 4500 cGy (from 2003 onwards), in 180-cGy fractions. In this three-field portion of the treatment, the superior border was placed at the bottom of the sacroiliac joints and the inferior border was unchanged from the initial field. For the posterior field, the lateral borders were placed 1.5–2.0 cm outside the pelvic brim. For the lateral fields, the posterior border was placed behind the sacrum, and the anterior border was placed 0–2 cm behind the pubic symphysis. Finally, a boost was delivered to the primary tumor with a 2–3-cm margin, typically to a cumulative dose of 5500 cGy. An open tabletop (belly board) device was used for bowel displacement in some patients. Involved inguinal nodes were treated to a cumulative dose of 5500 cGy, using supplemental appositional electron fields, and uninvolved inguinal regions received a dose of 3060 cGy from the initial pelvic field. In addition to the technique described before, only anterior and posterior fields were used for 8 (5%) patients, and other techniques were used for 6 (4%) patients.

Concurrent chemotherapy was administered with protracted infusional 5-FU and cisplatin, without induction chemotherapy in 112 (67%) patients. Five (3%) patients received induction 5-FU and cisplatin, followed by concurrent 5-FU and cisplatin with radiotherapy. Concurrent chemotherapy was administered with 5-FU and mitomycin C in 24 (14%) patients, capecitabine and cisplatin in 18 (11%) patients, and other regimens in 8 (5%) patients.

Results: The estimated 3-year rates of locoregional control, distant control, disease-free survival, and overall survival were 81%, 88%, 67%, and 84%, respectively. Multivariate analysis showed that higher T stage and N stage independently predicted for a higher rate of locoregional failure; higher N stage and basaloid subtype independently predicted for a higher rate of distant metastasis; and higher N stage and positive human immunodeficiency virus status independently predicted for a lower rate of overall survival. Among the patients who had locoregional failure, 18 (75%) had failure involving the anus or rectum, 5 (21%) had other pelvic recurrences, and 1 (4%) had inguinal recurrence. The 5 pelvic recurrences all occurred in patients with the superior border of the radiotherapy field at the bottom of the sacroiliac joint.

Conclusions: Trials of more aggressive and innovative locoregional and systemic therapies are warranted in high-risk patients, based on their T and N stages. The majority of locoregional failures involve the anus and rectum, whereas inguinal recurrences occur rarely. Placing the superior border of the radiotherapy field at L5/S1 could potentially reduce pelvic recurrences.

Discussion

The goal of this study was to identify predictors and patterns of recurrence in patients who underwent definitive chemoradiation for anal cancer. T stage and N stage independently predicted for a higher rate of locoregional failure, N stage and basaloid subtype independently predicted for a higher rate of distant metastasis, and N stage and positive HIV status independently predicted for a lower rate of overall survival. Patterns of failure analysis showed that 75% of locoregional failures involved the anus or rectum, 21% occurred elsewhere in the pelvis, and only 4% occurred in inguinal nodes.

Information on patterns of locoregional failure can help guide optimal design of radiotherapy fields. In this study, 5 patients had locoregional recurrences in the presacral or iliac regions, and all 5 recurrences occurred in patients in whom the superior border of the pelvic radiotherapy field was placed at the bottom of the sacroiliac joint. Patients with the superior border of the pelvic field at the bottom of the sacroiliac joint had a 7% risk of pelvic regional recurrence, and no pelvic regional recurrence occurred in patients in whom the superior border of the radiotherapy field was placed at the L5/S1 interspace. Based on this finding, we recommend that the superior border of the pelvic field be placed at L5/S1 in patients with anal cancer.

The risk of inguinal recurrence after definitive chemoradiation was very low. Among patients with inguinal nodal involvement at presentation, only 1 (2%) developed an inguinal recurrence, on the ipsilateral side. Among those with negative inguinal nodes at presentation, none developed an inguinal recurrence. Thus, our current practice of treating negative inguinal regions to 3060 cGy, positive nodes to 5500 cGy, and contralateral noninvolved inguinal regions to 3060 cGy, results in excellent disease control in the inguinal regions. The importance of electively treating negative inguinal regions cannot be overemphasized. Studies have shown 8–15% risk of inguinal recurrence in anal cancer patients who did not receive elective inguinal radiotherapy

A better knowledge of predictive factors may eventually help in the development of risk-adapted treatment strategies for anal cancer. In our study, T stage and N stage independently predicted for locoregional failure and primary failure. In particular, the 3-year rates of primary failure were high for patients with T4 (31%) and for those with N3 (35%) disease. Of note, patients with N3 disease had a low risk of inguinal recurrence, but a high risk of primary failure after chemoradiation. Future studies need to explore methods of improving local control in these subgroups of patients, such as with radiation dose escalation, or with newer concurrent chemotherapeutic and targeted agents. Phase II studies have been reported on chemoradiation with higher doses of radiotherapy. The recently reported Intergroup Radiation Therapy Oncology Group (RTOG) 98–11 trial recommended higher doses of radiotherapy (55–59 Gy) for T3, T4, and node-positive patients Continued efforts are necessary to improve local control in anal cancer patients at high risk for local failures.

N stage was significantly associated with rates of distant metastasis and overall survival. Hence, investigations of newer chemotherapeutic and targeted agents are warranted for patients with node-positive anal cancer. The Intergroup RTOG 98–11 trial investigated whether induction chemotherapy could improve disease-free survival in anal cancer patients, but did not show any improvement in disease-free survival with induction 5-FU/cisplatin followed by concurrent chemoradiation with 5-FU/cisplatin, compared with concurrent chemoradiation with 5-FU/mitomycin C5. Investigations on tumor biology and patient genetics will be critical to devise more innovative therapies for high-risk patients, to improve rates of relapse and survival.

In addition to N stage, basaloid subtype was associated with a higher risk of distant metastasis, and positive HIV status was associated with a higher risk of death. Basaloid subtype has been previously thought to have clinical features similar to other squamous cell carcinomas of the anal canal (12). The number of patients with basaloid subtype was relatively small in this study (N = 27), and the subtype was determined based on pathology reports and not a dedicated pathology review. Our finding about basaloid subtype should, therefore, be regarded as merely hypothesis-generating and needs to be verified in other studies. Although positive HIV status was associated with lower overall survival, the poor survival was largely the result of intercurrent disease, which caused 2 of the 3 deaths in HIV-positive patients.

A recent retrospective study showed that radiotherapy dose >50 Gy was associated with a higher rate of local control. Radiation dose was not significantly associated with rates of locoregional recurrence in our study, likely because patients were treated with a small range of doses, with 95% receiving between 5040 and 6040 cGy. Multiple studies have shown that the duration of radiotherapy is significantly associated with the risk of local failure . We did not detect a significant association between treatment time and locoregional control, likely because patients were treated without a planned treatment break, with a median treatment duration of only 6 weeks and 96% of patients completing treatment in less than 7 weeks. Studies have also demonstrated the role of other prognostic factors in anal cancer including age, anal margin location, response to radiotherapy, and dose intensity of chemotherapy

Our study has certain limitations. This study was performed retrospectively, based on data collected from hospital records. Long-term follow-up information was not available on all patients; hence, rates of recurrence may have been underestimated. Only a small proportion of patients (14%) received concurrent chemotherapy with 5-FU and mitomycin C, the regimen that has been accepted as the standard of care at many institutions based on the results of an Intergroup trial. The most commonly used concurrent chemotherapy regimen in this study was 5-FU and cisplatin, which was used in 70% of patients, with (3%) or without induction (67%) chemotherapy. Previous studies at our institution demonstrated that patients treated with concurrent 5-FU and cisplatin had outcomes comparable with those treated with concurrent 5-FU and mitomycin C . The Intergroup RTOG 98–11 trial compared chemoradiation with concurrent 5-FU/mitomycin C to induction chemotherapy with 5-FU/cisplatin, followed by chemoradiation with concurrent 5-FU/cisplatin. There was no significant difference in disease-free survival between the 2 arms, but the rate of colostomy was significantly higher in patients treated with induction and concurrent 5-FU/cisplatin (20%) compared with patients treated with concurrent 5-FU/mitomycin C (10%). It remains unclear whether the higher rate of colostomy in the 5-FU/cisplatin arm was a result of the type of chemotherapy or a result of the use of induction therapy, which was not used in most patients in the current study. It is possible that patterns and predictors of recurrence may differ in patients treated with 5-FU/cisplatin compared with those treated with 5-FU/mitomycin C. However, the rates of recurrence, disease-free survival, and overall survival in this study are comparable with those reported with 5-FU/mitomycin C5.

This retrospective study showed that after definitive chemoradiation for anal cancer, T stage and N stage independently predicted for a higher rate of locoregional failure, and N stage independently predicted for a higher rate of distant metastasis and death. Trials of newer and more aggressive locoregional and systemic therapies are warranted in high-risk patients, based on their T and N stages. Studies on tumor biology and patient genetics could help in the development of innovative therapies for such high-risk patients. The patterns of failure observed in this study indicate that placing the superior border of the radiotherapy field at L5/S1 may help lower the risk of pelvic failures. Moreover, because the majority of locoregional failures involved the anus and rectum, efforts are warranted to increase primary control in patients at high-risk for local failure, such as T4 and N3 patients.