Impact of advancing age on treatment and outcomes in anal cancer.

Abstract Background. Chemoradiotherapy (CRT) for squamous cell carcinoma of the anus (SCCA) may cause significant toxicity, and concerns exist about its tolerability in the elderly. The authors compared tolerability and outcomes across the age groups following CRT for SCCA. Methods. Single-institution retrospective analysis of patients with localized SCCA treated with CRT. CRT was standardized at 50.4–54 Gy, with concurrent infusional 5-fluorouracil and mitomycin C. Patients were arbitrarily categorized into three groups: Group 1 – age < 50 years; Group 2 – age ≥ 50 and < 70 years; and Group 3 – age ≥ 70 years. Results. Of 284 patients identified, 278 were evaluable. The number of patients in each age group was: Group 1 – 51; Group 2 – 140; and Group 3 – 93. Baseline and treatment characteristics, tumor stage, rates of overall acute toxicity, need for unplanned treatment breaks and chemotherapy delivery were largely similar across the age groups. However, nine patients in Group 3 did not complete CRT, compared with five and none in Groups 1 and 2, respectively (p = 0.006). In addition, five patients in Group 3 had diarrhea requiring treatment break, compared with none in the other two groups (p = 0.004). At a median follow-up 5.3 years, there was no significant difference in overall survival (p = 0.11), disease-free survival (p = 0.22) or local-recurrence free survival (p = 0.34), across the three age groups. Conclusions. CRT is safe and tolerable in the elderly age group, and provides equivalent disease control rates compared with the younger age group. Age alone should therefore not preclude aggressive curative treatment.

Concurrent chemoradiotherapy (CRT) has the potential to cure a signifi cant proportion of patients with non-metastatic anal cancer [1]. However, curative treatment comes at the expense of signifi cant acute toxicity, even in the young and fi t patient, not only from the frequently large radiation volumes for nodal prophylaxis, but also the added toxicities of concurrent chemotherapy.
There are concerns about the ability of the elderly age group ( Ն 70 years) to tolerate aggressive treatment, many of whom have signifi cant comorbidity, frailty or poor performance status. These factors have implications with regards to chemotherapy toxicity and tolerability, including prolonged myelosuppression with mitomycin C (MMC), renal toxicity and prehydration requirements of cisplatin.
Various treatment modifi cations have been employed in an attempt to improve treatment tolerability for elderly patients, including reduced radiation dose, omitting prophylactic nodal irradiation, chemotherapy dose reduction, single chemotherapy agent with 5-fl uorouracil  alone, as well as implementing a planned treatment break. These measures may potentially compromise tumor control, but are often considered reasonable options with the presumption that this group of patients may have a limited lifespan.
There is a lack of studies evaluating the treatment of elderly patients with anal cancer. The direct extrapolation of benefi t of curative CRT for elderly patients is hampered not only by the fact that some studies excluded the elderly subgroup [2,3], but also that the median age of patients in the large randomized trials ranged from 55 to 65 years, with an underrepresentation of patients aged Ն 70 years [1,2]. The burden of anal cancer in this population is not insubstantial, with Australian Cancer Registry statistics [4] showing over 40% of new anal cancer diagnoses occurring in patients over the age of 70 years, with 16% occurring above 80 years.
This study describes our experience treating patients with anal cancer across all age groups using a uniform treatment policy for the past 25 years. The aims are to compare the toxicity, outcomes and overall survival of patients in different age groups treated with curative intent CRT.

Material and methods
This is a single-institution retrospective review. Consecutive patients with non-metastatic squamous cell carcinoma of the anus treated with curative intent CRT between February 1983 and February 2008 were analyzed. Patients were staged by physical examination, examination under anesthesia/proctoscopy, chest radiograph or computed tomography (CT) of the chest, CT of abdomen and pelvis, and in the latter part of the study period, endorectal ultrasound, pelvic magnetic resonance imaging (MRI), and positron emission tomography (PET). Suspected groin node involvement was confi rmed with fi ne needle aspiration or excisional biopsy. Details of data collection and treatment have been described elsewhere [5]. A protocol and an analysis plan were established before commencement of this study. The study was approved by the institutional ethics board.
For the purposes of this study, patients were arbitrarily categorized into three age groups, which were pre-determined prior to data analysis: Group 1 -age Ͻ 50 years; Group 2 -age Ն 50 and Ͻ 70 years; and Group 3 -age Ն 70 years.
All patients were managed by the gastrointestinal team of the cancer center. The unit policy remained consistent during the study period. Patients were treated with CRT to a total dose of 50.4 -54 Gy in 1.8 Gy per fraction with concurrent infusional 5-FU and MMC.
Radiotherapy (RT) was generally delivered in a three-phase technique, with the fi rst phase encompassing the primary tumor, lower mesorectal, presacral, lower internal and external iliac, and inguinal lymph nodes. The standard technique utilized wide antero-posterior/postero-anterior photon beams to 36 Gy in 20 fractions, with the superior fi eld border 1 cm above the inferior sacroiliac joints or 5 cm proximal to the primary tumor, whichever was more proximal. The second phase then treated the primary tumor, distal internal iliac and mesorectal nodal regions, usually with a three-fi eld arrangement of posterior and parallel-opposed lateral photon beams, a further 9 Gy in 5 fractions, to 45 Gy in 25 fractions. The third phase boosted the primary tumor to 5.4 -9 Gy in 3 -5 fractions (bringing the total dose of 50.4 -54 Gy in 28 -30 fractions). Patients with T1 N0 tumors did not routinely have elective inguinal radiation, and were treated with a technique akin to the second phase described above (to 45 Gy in 25 fractions with a three-fi eld arrangement), followed by a boost to the primary tumor similar to the third phase above.
Patients were reviewed weekly during the course of CRT, fortnightly after completion of CRT until resolution of acute toxicity, three monthly for three years, six monthly for another two years and then yearly. Routine reviews consisted of history and clinical examination including digital rectal examination. Biopsies were performed if there was clinical suspicion of persistent or recurrent disease.
Data on the type of acute toxicity which resulted in interruption of treatment was collected, and data from subsequent visits about survival and recurrence status recorded. Local recurrence was defi ned as persistence (at least 12 weeks following CRT) or recurrence of disease following a complete response .
The differences in baseline and treatment characteristics between the ordered three age groups were assessed by the p-value corresponding to the Score test in ordered logistic regression (with ordered age group as the response) or the Type I sum of squares F-value in ANOVA, as appropriate. Median follow-up was assessed by the reverse Kaplan-Meier method. Local recurrence free, distant recurrence free and overall survival from the date of CRT completion were assessed with Kaplan-Meier curves, with the Cox proportional hazards regression unadjusted and adjusted for T stage and sex used to compare survival between the three age groups.
In addition, post-hoc sensitivity analyses were performed using varying age cut-offs ( Ͼ 75 years and Ͼ 80 years) for Group 3. These were done following the initial analysis to determine whether the initial conclusions were likely to change when using varying age parameters.

Results
There were 284 patients identifi ed with localized squamous cell carcinoma of the anal canal treated with defi nitive CRT at the cancer center between 1983 and 2008. There were fi ve patients without any response or follow-up data following radiotherapy or the post-treatment PET scan, and one patient with response data but no further follow-up, thus leaving 278 patients for the outcome analyses. When grouped by age at diagnosis, 18% of patients (n ϭ 51) were aged Ͻ 50 years, 50% of patients (n ϭ 140) aged 50 -69 years, and 33% (n ϭ 93) of patients were aged 70 years or older. There were 25 patients aged 80 years or older.

Baseline characteristics
Baseline characteristics of all patients (n ϭ 284) are outlined in Table I, and were similar across patients of different age groups with regards to sex, primary tumor site, diameter, differentiation, T stage and N stage. There was a suggestion that younger patients were more likely to have higher AJCC staging (p ϭ 0.05), yet this trend was not observed when the sensitivity analyses were done (p ϭ 0.24) (Supplementary Table I

Treatment characteristics
Treatment characteristics are listed in Table II. Ninetyfi ve percent of all patients completed CRT. Most patients {63% [95% confi dence interval (CI) 57 -69%]} completed their radiotherapy with no break. However, nine patients aged Ͼ 70 years did not complete CRT, compared with fi ve in the 50 -70-year-old age group, and none in the younger Ͻ 50 years age group; this difference was statistically signifi cant (p ϭ 0.006). Only two of 25 patients aged 80 years or above did not complete CRT. Otherwise, treatment details were largely similar across the different age groups with respect to the need for unplanned treatment breaks and the length of treatment breaks. There were no signifi cant differences in terms of chemotherapy regimens or dose intensity among the age groups.
Sensitivity analyses confi rmed a signifi cant association of age with completion of RT even when a cut-off of Ͼ 75 years was used (Supplementary Table II

Acute toxicity
Thirty-four percent (29 -40%) of patients experienced treatment-related acute toxicity necessitating a treatment interruption, the most common being skin related toxicity: 22% (17 -27%) (Table III). There were differences in the patterns of toxicity observed by age group: older patients were more

Tumor control and survival
Tumor control and survival rates were analyzed for 278 patients with follow-up data. At a median follow-up of 5.3 (95% CI 5.0 -6.1) years, there were 38 colostomies, 54 deaths, 43 locoregional and 17 distant recurrence events reported. Patterns of failure are outlined in Table IV and (Supplementary  Table IV (Figure 1).
Age at diagnosis was not a signifi cant predictor of overall survival when analyzed as a continuous (p ϭ 0.15) or categorical (p ϭ 0.25) variable on univariate analysis, nor when added to a multivariable model including T stage and sex. The fi ve-year local recurrence free survival (LRFS) for Groups 1, 2, and 3 were 80% (69 -93), 73% (65 -81) and 79% (70 -89), respectively (three-arm comparison log-rank p ϭ 0.34) (Figure 2). Corresponding percentages for fi ve-year distant metastases-free survival were 92% (83 -100), 83% (77 -91) and 82% (74 -91), respectively (p ϭ 0.21). Age at diagnosis was not a significant predictor of either local-or distant-recurrence free survival, when analyzed as a continuous or categorical variable, in univariate or multiple variable models, nor when added to a multivariate model including T stage and sex (as categorical, p ϭ 0.22).

Discussion
The signifi cant toxicity of CRT for anal cancer presents challenges for oncologists when dealing with the elderly age group. When attempting to ^patients may have a combination of loco-regional and distant recurrences; * patients with any combination of local, pelvic and/or inguinal nodal recurrence. Figure 1. Overall survival by age group, using age cut-off of Ͼ 70 years for Group 3.
[6] and the US [7] previously defi ned elderly as over 65 years, US Census data now defi ne the Ͼ 65s as the " older population " [8], a defi nition which is similarly used by the Australian Bureau of Statistics [9]. It is worth noting that despite an increasing population in the Ͼ 65 and Ͼ 75 age group, they determine an older patient ' s fi tness for aggressive treatment, clinicians are often faced with the uncertainty around basing treatment decisions on a patient ' s biological versus chronological age. Adding to this complexity is the inconsistent and varying defi nition of this so-called " elderly " age group. Whereas Japan Figure 3. Disease-free survival by age group, using age cut-off of Ͼ 70 years for Group 3. Figure 2. Local-recurrence free survival by age group, using age cut-off of Ͼ 70 years for Group 3. represent only about 14% and 5.8% of the total population, respectively. Thus, for the purposes of the current study, given the median-age at diagnosis for anal cancer is 60 years [10], and to avoid the possibility of too few patient numbers in the later age group, an age-limit and cut-off of 70 years was selected at the planning of this study.
Our results show that elderly patients had comparable tumor control rates and outcome following curative CRT, with fi ve-and eight-year disease-free survival rates of 75% [95% CI 66 -86%] and 72% [95% CI 62 -84%], respectively, which are similar to the fi ve other studies reported so far [11 -15]. In order to determine if these results could vary depending on the specifi c age cut-off selected, we also performed post-hoc sensitivity analyses using upper age cut-offs of Ͼ 75 and Ͼ 80 years. As alluded to in the results section, these sensitivity analyses for local recurrence free survival and disease-free survival confi rm the initial conclusions that there is no signifi cant effect of age on these outcomes.
The curves for overall survival demonstrate a p-value of 0.016 when a cut-off of Ͼ 75 years was used (Supplementary Figure A1 available online at http://informahealthcare.com/doi/abs /10.3109/0284 186X.2013.876513). These fi gures are interesting, as if the data were to suggest a difference in outcome with increasing age, one would expect the curves would move slightly apart, but in the same order and direction compared with the other age cut-offs. This is not demonstrated in the curves shown -the order of the curves changes for different age group cut-offs, and the uncertainty around each point estimate considered and hazard ratio suggests no evidence of a difference in outcome by age group. However, if the p-values alone are considered, as can be seen from (Supplementary Figure A1 available online at http:// informahealthcare.com/doi/abs/10.3109/0284186X. 2013.876513), the p-value of 0.016 suggests there may be a difference in overall survival at some time within 13 years from end of radiotherapy when using a cut-off of Ͼ 75 years. We are cautious about overinterpreting this signifi cant p-value (in line with best practice in interpretation of results as suggested in Gardner and Altman [16], as this interpretation is not supported by the other analyses using this cutoff. The confi dence intervals for the point estimates of outcome rates at all time-points considered overlap each other, as do the confi dence intervals for the hazard ratios. The suggestion of an association is also not supported with the sensitivity analyses with the alternate cut-offs for age group. Thus, our conclusion from this remains and supports that from the main analysis, that there is no evidence for a signifi cant effect of age group on outcome. Elderly patients also presented with similar stages of disease to that seen in the younger age groups, experienced similar rates of overall acute toxicity, and required no more frequent treatment breaks for toxicity. In this study, frequency of treatment break was selected as a surrogate endpoint for severe acute toxicity, where in practice these breaks would often be necessary for severe toxicity. For this cohort, no statistically signifi cant difference in the proportion of patients requiring a treatment break was observed (39%, 28% and 33% of patients in the Ͻ 50, 50 -70 and Ͼ 70 years age group, respectively). Likewise, the duration of unplanned break for each corresponding group was also similar at 10, 13 and 11 days, respectively (p ϭ 0.36), the majority of which was due to skin toxicity. The acute toxicity reported in our cohort overall is comparable to that seen in RTOG 9811, which reported rates of grade 3 acute gastrointestinal (GI) and skin toxicity of 32 -43% and 39 -43%, respectively [17].
Elderly patients seemed, however to have more severe acute gastro-intestinal toxicity, in particular, compared to younger patients. There were fi ve patients in the Ͼ 70 years group whom developed diarrhea requiring a treatment break, compared to none in the other two groups. In addition, there was a greater proportion of patients in the Ͼ 70 years group (10%) whom failed to complete RT. The reasons for RT cessation were not reported in this study, however a possible explanation could be that the treating physicians may have potentially been more inclined to either institute a treatment break or even cease RT in the elderly age group. It is also worth noting that that the tumor control outcomes in the older age group are comparable despite a greater proportion of them not completing CRT. We postulate a possible explanation for this is that the total dose of 50.4 -54 Gy used likely provides some " buffer " , in that the earliest studies by Nigro et al. [18] demonstrated signifi cant proportions of complete pathological responses even with doses as low as 30 Gy. Thus, it is possible (and/or likely) that a proportion of patients whom did not complete RT were potentially cured, despite receiving doses below 50.4 Gy. However, our data should not be interpreted so as to suggest that a lower radiation dose is always appropriate, as this would obviously require formal testing in a randomized fashion.
The retrospective nature of this analysis of course carries with it inherent biases and limitations, however, its strengths lie in the fact that, fi rstly, over the past 25 years, our institutional and unit policy has been to offer a uniform treatment regimen for anal cancer regardless of age, provided the patient is in good performance status and deemed to be fi t for all components of the therapy. At the outset, at least, there were no planned modifi cations made in terms of 5-FU and MMC dosing and intensity. In addition, for the fi rst 10 years or so, our center was the sole radiotherapy provider in the state of Victoria, as evidenced by the fact that the distribution of anal cancer across the age groups in this study is not dissimilar to that described in Australian population statistics.
In conclusion, our results show similar acute toxicity, disease control and overall survival at fi ve years among different age groups. We would therefore advocate that advancing age alone should not preclude comprehensive assessment and staging, nor should it exclude curative standard CRT. It is likely that techniques such as IMRT would improve tolerability for all age groups without compromising tumor control.

Declaration of interest:
The authors report no confl icts of interest. The authors alone are responsible for the content and writing of the paper.