Anal cancer incidence and survival: The Surveillance, Epidemiology, and End Results experience, 1973–2000

Anal cancer is a rare disease, with only 4010 incident cases predicted to arise in the United States in 2004.1 However, the physical and emotional morbidities associated with treatment and disease progression can be great.2, 3 Anal cancer appears to be more similar etiologically to genital malignancies than to malignancies of the gastrointestinal tract. For example, previous reports have found associations between anal cancer incidence and female gender,4 infection with human papillomavirus (HPV),5-9 lifetime number of sexual partners,10-12 cigarette smoking,7, 12-14 genital warts,10, 11 receptive anal intercourse,9-11 and infection with the human immunodeficiency virus (HIV).9, 15-17 Recent work by our group18 suggests that secular changes in sexual behavior may have a significant impact on anal cancer incidence and survival. Data from the Surveillance, Epidemiology, and End Results (SEER) Program was used to explore changes over time in anal cancer incidence and survival in the United States between 1973 and 2000.

MATERIALS AND METHODS

The SEER Program has collected population-based data on diagnoses and outcomes for all incident cases of invasive and in situ anal cancer in five states (Connecticut, Hawaii, Iowa, New Mexico, and Utah) and four metropolitan areas (Atlanta, GA; Detroit, MI; San Francisco, CA; and Seattle, WA) since 1973 (although systematic and comprehensive data collection did not begin until 1974 in Atlanta and Seattle). The program collects selected data on demographics, tumor characteristics, and extent of disease. Although individual SEER registries use different follow-up methods, follow-up typically is initiated within 1 year of diagnosis via contact with the hospital or provider associated with the diagnosis. SEER registries have a target rate of 95% for successful overall follow-up of ascertained cases for assessment of survival.19

All histologic types except for sarcomas of the anus, anal canal, and anorectum (International Classification of Diseases for Oncology, Third Revision, [ICD-O-3] codes C210–C218) were included. Age-adjusted incidence rates and relative survival following diagnosis of anal cancer among individuals age ≥ 20 years between 1973 and 2000 were calculated.20 Age adjustment of rates was based on the United States census population in 2000.

All analyses were stratified according to gender. Correlations involving race, age, tumor histology, diagnosis year, and anal cancer incidence and survival were investigated. For analyses that focused on individual histologic types, squamous cell carcinoma (SCC) was defined by ICD-O histology codes 8010 and 8051–8081, cloacogenic (and basaloid) cancers were defined by codes 8120–8124, and adenocarcinoma (including mucinous adenocarcinoma) was defined by codes 8140–8263 and 8480–8481. Because long-term changes in the incidence of anal cancer represented the primary focus of the analysis, data were aggregated into calendar-year groups to reduce fluctuations caused by interyear variability in rates as a result of limited numbers of cases. Anal cancer incidence rates in these analyses generally refer to both in situ and invasive disease, although time trends in incidence were stratified according to tumor behavior.

For the survival analyses, only patients who had a first primary anal tumor that was confirmed microscopically, had invasive disease, and were followed actively after diagnosis were included. Relative survival, a net survival variable that takes into consideration other competing causes of death based on life tables that account for age, gender, race, and calendar year, was used as the outcome variable. In addition to being analyzed according to demographic factors and tumor histology and behavior, survival statistics also were analyzed according to tumor stage, grade, size, and site. Stage was defined as the historical SEER stage (local, regional, or distant)21; staging information was available for 89% of eligible patients. Further information regarding these data items is available from the SEER website (http://www.seer.cancer.gov).

SEER*Stat software was used to calculate incidence and relative survival counts and rates, and STATA statistical software (Stata Corp., College Station, TX) was used to calculate P values. Two-sided P values for trend were calculated using Poisson maximum-likelihood regression to model the selected characteristic against a linear year variable.22 P values for comparisons of incidence rates were calculated using the large-sample normal theory test.23

RESULTS

Survival

Between 1973 and 2000, men had poorer overall survival than did women with anal cancer (relative 5-year survival, 58% vs. 64%) (Table 3). Five-year survival improved substantially over time among women, from 59% in the earliest period to 73% in the most recent period. However, the 5-year survival rate for men with anal cancer was consistently lower than the corresponding rate for women and was nearly unchanged (∼60%) over the same span.

Table 3. Gender-Specific 5-Year Relative Survival among Patients Diagnosed with Anal Carcinoma by Race, Histology, and Year of Diagnosis: Surveillance, Epidemiology, and End Results Program, 1973–2000

Variable	Men		Women
	No. (%)a	5 yr relative survival	No. (%)a	5 yr relative survival
Overall	1875 (100)	0.58	2874 (100)	0.64
Period
1973–1979	321 (17.1)	0.60	517 (18.0)	0.59
1980–1986	414 (22.1)	0.55	722 (25.1)	0.64
1987–1993	550 (29.3)	0.56	838 (29.2)	0.62
1994–2000	590 (31.5)	0.61	797 (27.7)	0.73
Race
White	1576 (84.1)	0.61	2524 (87.8)	0.65
Black	228 (12.2)	0.38	259 (9.0)	0.57
Other	71 (3.8)	0.40	91 (3.2)	0.68
Histology
Squamous cell	1029 (56.8)	0.62	1560 (56.1)	0.67
Cloacogenic	320 (17.7)	0.55	786 (28.3)	0.71
Adenocarcinoma	463 (25.6)	0.51	433 (15.6)	0.48
Age (yrs)
20–49	474 (25.3)	0.56	418 (14.5)	0.72
50–64	671 (35.8)	0.62	869 (30.2)	0.69
≥ 65	730 (38.9)	0.54	1587 (55.2)	0.59
Race/period
White
1973–1979	284 (18.0)	0.63	467 (18.5)	0.59
1980–1986	358 (22.7)	0.57	637 (25.2)	0.65
1987–1993	451 (28.6)	0.59	732 (29.0)	0.63
1994–2000	483 (30.6)	0.66	688 (27.3)	0.73
Black
1973–1979	30 (13.2)	0.45	39 (15.1)	0.59
1980–1986	44 (19.3)	0.36	68 (26.3)	0.55
1987–1993	74 (32.5)	0.39	75 (29.0)	0.53
1994–2000	80 (35.1)	0.27	77 (29.7)	0.57

• a Only patients with anal carcinoma who were followed actively were considered.

Black men with anal cancer were likely to die within 5 years of diagnosis (relative 5-year survival rate, 38%) (Table 3). Even for black men with local-stage disease, the 5-year relative survival rate was only 62%, compared with 79% for white men with localized disease (data not shown). Black women had slightly poorer relative survival (57% at 5 years postdiagnosis) compared with white men and white women (61% and 65% at 5 years postdiagnosis, respectively). For women with regional or distant disease, survival did not vary according to race, although among women with local-stage disease, white women had somewhat better survival at 5 years compared with black women (81% vs. 68%; data not shown).

Five-year relative survival rates were similar for men and women who had SCC (62% and 67%, respectively) and for men and women who had adenocarcinoma (51% and 48%, respectively), but survival rates were substantially poorer for men with cloacogenic anal cancer than for their female counterparts (55% vs. 71%) (Table 3). Five-year survival also was poorer for individuals age > 65 years at the time of diagnosis (54% for men and 59% for women) compared with younger patients of both genders. Among individuals ages 20–49 years at diagnosis, the relative survival rate was 56% for men at 5 years postdiagnosis and 72% for women. Survival patterns according to disease histology and patient age were relatively consistent over time. There was, however, a substantial difference in gender-specific and race-specific relative survival rates over time. Survival decreased over time among black men (from 45% in the earliest period to 27% in the most recent period) and decreased slightly among black women (from 59% to 57%), whereas survival improved slightly among white men (from 63% to 66%) and improved substantially among white women (from 59% to 73%).

More advanced disease stage was inversely associated with survival, and in each disease stage category, men had poorer survival than did women (Fig. 1). Overall, at 5 years after diagnosis, 78% of patients who had local-stage anal cancer were alive, compared with 56% of patients who had regional-stage disease and 18% of patients who had distant disease. Men were less likely than were women and white patients were less likely than were black patients to be diagnosed with disease that had spread beyond the local primary site; in addition, stage-specific relative survival was poorer at 5 years for black patients compared with white patients (local disease, 65% vs. 80%; regional or distant disease, 38% vs. 49%). Both black women and white women had an increased proportion of local-stage disease in the most recent period compared with the earliest period, but this finding did not hold true for either black men or white men.

DISCUSSION

Between 1973 and 2000, anal cancer incidence in the SEER registries increased for both genders, with the increase being more dramatic among men. This led to parity in the age-adjusted, gender-specific annual incidence rates, which increased from 1.06 per 100,000 in 1973–1979 to 2.04 per 100,000 in 1994–2000 among men and from 1.39 per 100,000 in 1973–1979 to 2.06 per 100,000 in 1994–2000 among women. Although the rates of anal cancer incidence did not vary greatly between black women and white women, black men had substantially higher incidence rates compared with white men in the more recent period. Most increases in the incidence of anal cancer appeared to be attributable to increases in anal SCC incidence.

Relative survival for patients with anal cancer was inversely associated with disease stage and was consistently poorer for men compared with women and for black patients compared with white patients. Survival also was poorer in the oldest age group for both genders. Although survival improved over time for white women, it remained relatively constant over time for white men and black women and decreased appreciably for black men.

There has been concern regarding error of closure, which occurs near the end of a decade, when projected estimates from the previous national census may not accurately predict the actual make-up of the population. This phenomenon may have affected the denominator data in our incidence and relative survival calculations and may have caused these figures to be inflated, particularly in stratified analyses dealing with geographic regions or racial groups. Recently, SEER began using the National Center for Health Statistics' bridged estimates of the population (which are based on the 2000 census) to predict the population more accurately for the years 1991–1999.24 This change should increase the accuracy of population composition estimates in the SEER database for the 1990s, and particularly for the later stages of the decade, adding confidence that our rate estimates are not unduly biased by errors in population estimates.

The 2000 census also allowed respondents to choose more than one race category. This new feature may affect population estimates for specific racial groups and may subsequently affect race-specific incidence rate estimates after 1990. Although SEER has begun collecting data on multiple races for individuals with malignant disease, it is unclear as to how this data will be reconciled with race data obtained by the census.

In 1994, Melbye et al. examined anal cancer incidence in the SEER registry over the period 1973–1989, focusing on epidermoid anal tumors (ICD-O codes 8051–8052, 8070–8076, 8090–8095, and 8120–8124).4 In that analysis, the investigators found that women had higher rates of anal malignancy compared with men and that incidence rates were highest among black women. Since that analysis, anal cancer incidence increased more rapidly among men compared with women, eliminating the gender-based difference in incidence; in addition, black women had the lowest rates of anal cancer incidence between 1994 and 2000 (Fig. 1).

Unlike the report by Melbye et al., which focused on invasive disease, the current study included in situ anal cancer. Because there was an increase in the rate of in situ disease over the period studied, it is possible that various sources of bias accounted for some of the increase in anal cancer incidence. Beginning in the mid-1980s, pathologists gradually began classifying severe dysplasia at any anogenital site as in situ disease, increasing the number of in situ lesions diagnosed.25 Therefore, an increase in the rate of in situ cancer at anogenital sites over time would be expected, even if no true increase occurred. This expectation was consistent with our findings, as the incidence of in situ anal cancer increased over time for both genders.

Increased rates of in situ disease also may be attributable in part to detection bias as a result of increased surveillance, particularly in high-risk populations. During the 1990s, screening of high-risk men using anal cytology became more common in certain areas.26, 27 It is possible that some of the early lesions (e.g., in situ disease) that were detected by screening would not have progressed to invasive disease or would not have been detected otherwise. The number of male patients with in situ disease may have increased as a result, because such patients would not have been identified without screening. The increase in the rate of in situ disease also may be attributable to screening that identified early disease, which would have progressed to invasive disease without screening in previous years. The finding that in situ disease was much less common (9%) in the earliest period compared with the most recent period (25%) (Table 1) supports the possibility of a stage shift in anal cancer incidence. However, there has not been a concomitant drop in the rate of invasive anal cancer—the rate of invasive disease has increased over time, although not to the same degree that in situ disease incidence has increased. Therefore, at least some of the observed increase in incidence is likely to be attributable to a true increase in anal cancer incidence.

Although the screening of high-risk men may have been partially responsible for the gender differences in tumor behavior among patients with anal cancer, it also is possible that biologic differences cause more women to have disease that becomes invasive rapidly or avoids detection for a longer period. For example, anal tumors in women may occur deeper in the anal canal, making them less likely to be detected at an early phase of growth on routine examination. Women were more likely than men to be diagnosed with cloacogenic anal cancer, a subtype of SCC that occurs at the anorectal junction and was found to be primarily invasive at diagnosis. The preponderance of cloacogenic anal cancer in women relative to men also may explain why women in the oldest age group had a higher overall rate of anal cancer compared with men in that age group (5.11 per 100,000 vs. 3.73 per 100,000).

The overall increase in anal cancer incidence rates also may be partially attributable to an increase in the average number of lifetime sexual partners (a risk factor for anal cancer10-12) among individuals in the United States. This increase in average number of sexual partners can be seen by comparing two case–control studies of anal cancer etiology in the Seattle metropolitan area. In 1987, Daling et al.10 reported that 8.9% of a female control group had > 4 lifetime sexual partners, compared with 39.5% of a female control group in a more recent analysis.18 This increase may have led to an increase in the number of individuals at risk for infection with HPV; HPV infection is strongly associated with anal cancer and may be a necessary step in its carcinogenesis.11, 18, 28 It also is plausible that there have been increases in the number of individuals engaging in receptive anal intercourse, another risk factor for anal carcinoma.9-11, 18 Again, comparison of the two studies conducted by Daling et al.10, 18 revealed that in the 1987 analysis, 10.8% of female control patients had ever had receptive anal intercourse, compared with 21.5% of female control patients in the more recent study. Furthermore, a national survey study (the General Social Survey [GSS]) found that the number of men who reported intercourse with another man in the 12 months before the survey had increased significantly (from 1.8% to 3.5%; P < 0.05) between the first half of the study (1988–1994) and the second half (1995–2000).29 The GSS also reported that black men, a group with the highest anal cancer incidence rate in the current analysis, were more likely than white men to report having had intercourse with another male in the last year (although the survey did not specifically address receptive anal intercourse). Whether these changes in reporting represented true changes in levels of exposure or were attributable to societal changes and their effects on responses to questions regarding sexual history is unclear.

Although cigarette smoking repeatedly has been associated with an increased risk of anal cancer,7, 12-14 the prevalence of smoking has decreased steadily since 1965 across gender, racial, and age groups.30 Thus, it is unlikely that smoking is a significant factor in the general increase in anal cancer incidence over time, as the association between smoking and anal cancer is primarily restricted to current smokers. It is possible that individuals with other risk factors for anal cancer, such as a relatively high number of sexual partners, continue to smoke at high rates, although the prevalence of smoking in the overall population has decreased.

The increase in anal cancer rates among men also may be attributable to the gender imbalance in terms of HIV infection, a risk factor for anal cancer17, 31, 32; most individuals living with AIDS in the US are men.33 Furthermore, 43% of new HIV infections in men occur in black men, despite the fact that African Americans make up 12% of the general population.34 Unlike incidence rates among black men, anal cancer rates among black women were similar to those among white women and white men, although black women also are affected disproportionately by HIV (64% of new HIV infections in women occur in black women34). Because the racial discrepancy in anal cancer incidence is confined to men, it is possible that differences in incidence between white men and black men may be due to factors other than HIV positivity (e.g., number of sexual partners, intravenous drug use, smoking, etc.).

Poor access to care may be a contributor to the poor relative survival of black men and women with anal cancer, as African-American race is one of the predictors that patients with AIDS will not receive highly-active antiretroviral therapy (HAART), the current treatment of choice.35 It is possible that the decrease in survival over time among black with anal cancer is due to AIDS-related mortality, whereas the slightly improved survival among white men with anal cancer in recent years is partially attributable to the use of life-extending HAART regimens.

Improved survival over time in white patients with anal cancer may be due at least in part to the introduction and acceptance of combined chemotherapy and radiation for the treatment of anal cancer; this treatment has been shown to be effective in reducing morbidity and mortality among patients with anal malignancies.36 It is possible that black patients with anal cancer are not receiving adequate treatment because of factors such as a higher rates of poverty (23% [African Americans] vs. 10% [Caucasians]37). One study reported that when African-American women received the same level of care for cervical cancer as did their white counterparts, they had similar 5-year survival rates.38 This finding suggests that inequality in care may play a substantial role in the observed inequality in survival rates for patients with anal cancer, assuming that the relation between treatment and survival is similar for patients with cervical cancer and patients with anal cancer. Furthermore, a recent study reported that patients with anal cancer who were HIV positive were more likely to have treatment-related toxicity and may have been more likely to die of malignant disease after treatment.2 Because the black population has a higher rate of HIV infection, differences in treatment response may also play a role in observed racial differences in survival.

Although anal cancer is a rare malignancy, it is possible that improved surveillance and targeting of high-risk groups for participation in prevention campaigns will be worthwhile in reducing morbidity and mortality associated with the disease. Furthermore, given the importance of HPV in the development of anal cancer and the finding that HPV type 16 (HPV16) is the most common type of HPV in anal tumors (unpublished results), the advent and widespread use of a vaccine against HPV16 and other oncogenic HPV types may lead to a reduction in the number of individuals affected by anogenital malignancies.39 Earlier diagnosis of anal cancer may lead to improved survival among patients with anal cancer if adequate screening and treatment can be made more widely available.