SEARCH

SEARCH BY CITATION

Keywords:

  • human papillomavirus (HPV);
  • anal neoplasms;
  • cigarette smoking;
  • sexual risk factors;
  • epidemiology

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

The incidence of anal cancer has increased among both men (160%) and women (78%) from 1973 to 2000 in the U.S. The authors conducted a population-based case–control study of anal cancer to examine factors that may account for this increase.

METHODS

Men (n = 119 patients) and women (n = 187 patients) who were diagnosed with anal cancer between 1986 and 1998 in the Seattle area were ascertained through the local Surveillance, Epidemiology, and End Results registry. Control participants (n = 1700) were ascertained through random-digit telephone dialing. Participants were interviewed in person and provided blood samples. Archival tumor tissue was tested for human papilloma virus (HPV) DNA, and serum samples were tested for HPV type 16 (HPV-16).

RESULTS

Overall, 88% of tumors (all histologies) in the study were found to be positive for HPV. HPV-16 was the most frequent HPV type detected (73% of all tumors), followed by HPV-18 (6.9%), regardless of gender. However, 97.7% of tumors from men who were not exclusively heterosexual contained HPV DNA. The risk of anal cancer increased among men (odds ratio [OR], 5.3; 95% confidence interval [95% CI], 2.4–12.0) and women (OR, 11.0; 95% CI, 5.5–22.1) who had ≥ 15 sexual partners during their lifetime. Among men who were not exclusively heterosexual and women, receptive anal intercourse was related strongly to the risk of anal cancer (OR, 6.8 [95% CI, 1.4–33.8] and OR, 2.2 [95% CI, 1.4–3.3], respectively). Current smokers among men and women were at particularly high risk for anal cancer, independent of age and other risk factors (OR, 3.9 [95% CI, 1.9–8.0] and OR, 3.8 [95% CI, 2.4–6.2], respectively).

CONCLUSIONS

The high proportion of tumors with detectable HPV suggests that infection with HPV is a necessary cause of anal cancer, similar to that of cervical cancer. Increases in the prevalence of exposures, such as cigarette smoking, anal intercourse, HPV infection, and the number of lifetime sexual partners, may account for the increasing incidence of anal cancer in men and women. Cancer 2004. © 2004 American Cancer Society.

The incidence of anal cancer in the U.S. has increased among both men (160%) and women (78%) from 1973 to 2000.1, 2 Although the incidence was consistently higher for women compared with men through 1995, the annual incidence from 1996 to 2000 was equal for men and women (2.1 per 100,000 population).1 Human papillomaviruses (HPV) have been implicated as a cause of cervical and other anogenital cancers; however, to our knowledge only three population-based case–control studies of anal cancer have been conducted that included the collection of tumor tissue for testing for HPV DNA,3–5 and only one study included both the collection of blood and tissue for both HPV antibody and HPV DNA testing.6 Approximately 85% of anal cancers are of squamous cell origin, and to our knowledge all studies published to date have focused on this histologic type.

We have been conducting a series of population-based case–control studies of anogenital cancers during the past 20 years. In the current study, we present the final results from our population-based case–control study of anal cancer using patients who were diagnosed between January 1986 and June 1998, including the results of the tumor tissue analyses for HPV for all histologic types of anal cancer and HPV type 16 (HPV-16) serologic testing. Patient characteristics and exposures from this more recent period that either predispose to HPV infection or are cofactors with HPV for disease also are described.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Patients in the case group (cases) included all men and women age < 75 years whose primary residence was in King, Pierce, or Snohomish Counties in western Washington State from January 1986 through June 1998 and who were diagnosed with in situ or invasive anal or perianal skin cancer during that time. The cases were identified through the Cancer Surveillance System, one of the population-based registries that is part of the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program.2 The types of cancer that potentially were eligible for the study included epidermoid cancer of the rectum (International Classification of Diseases for Oncology [ICD-O] site codes 154.0 and 154.1 and ICD-O histology codes 8010–8124) and all histologic tumors that occurred in the anal canal (ICD-O code 154.2); anus, not otherwise specified (ICD-O code 154.3); anorectum (ICD-O code 154.8); and perianal skin (ICD-O code 173.5). The registry ascertained 485 men and women with anal cancer, and 306 of those patients (63.1%) were interviewed, with 119 male patients and 187 female patients available for analysis. The median time from the diagnosis to the interview was 10 months.

Individuals in the control group (controls) were selected from the general population by random-digit dialing using the two-stage Waksberg–Mitofsky method.7 Control participants were frequency matched to the cases for all of our population-based anogenital studies that were being conducted at the same time (cervical, vulvar, vaginal, anal, and penile cancer) by age (5-year age groups), gender, and reference year (year of diagnosis of cases); and they had to live in the same 3 county area as the cases. Controls were ascertained continuously throughout the study period, and reference months were assigned randomly for the reference months currently being interviewed. Of 2413 men and women who were identified through a census of all individuals living in a randomly dialed area, 1700 individuals (71.5% of women and 64.4% of men) were interviewed successfully. After taking into account the proportion of screened residences (92.7% for women and 94.5% for men), the participation rate was 66.3% for female controls and 60.8% for male controls.

Trained interviewers conducted the in-person interviews in the homes of the participants. The information collected included the following: demographic characteristics, number of lifetime sexual partners (asked in the categories indicated), sexual practices, history of sexually transmitted diseases, tobacco and drug use, and medical history. Interview questions were designed to collect only information that applied prior to each participant's reference date (the diagnosis date of cancer for cases and a comparable time for controls).

At the time of the interview, an approximately 10-mL blood sample was drawn by routine venipuncture from 91.2% of cases and 89.1% of controls. Permission was obtained from cases to acquire paraffin embedded tumor tissues from the individual pathology laboratories, and 98.0% of patients signed a tissue release form. Tissue samples were available for 262 of the interviewed cases (85.6%). Seropositivity for HPV-16 capsid proteins was determined using an antigen-capture enzyme-linked immunosorbent assay.6 Antibody response to herpes simplex type 2 (HSV-2) was assessed using a Western blot assay to discriminate between immune response to HSV-1 and HSV-2.8

To classify HPV types, we used polymerase chain reaction (PCR) methods to amplify HPV DNA extracted from paraffin embedded tissue using methods previously described.9 All laboratory procedures were performed without the knowledge of tumor characteristics or patient outcomes. Tissue sections were cut from the tumor blocks for HPV testing using strict precautions to prevent intersample DNA contamination; sections from an HPV-negative heart tissue block were cut and tested for HPV as every 10th study sample. HPV-DNA negative results were tested for PCR amplification of a 536-base pair (bp) or 268-bp fragment of the human β-globin gene to determine whether DNA was present, and 14 of the tumor blocks (5.3%) were β-globin negative.

Only squamous cell (SCC) and cloacogenic or basaloid (cloacogenic) anal cancers (ICD-O histology codes 8010–8081 and 8123–8124, respectively) were included in the risk-factor analyses. Analyses in men combined both histologic types, because only eight men had cloacogenic anal cancer. Analyses in women included separate risk estimates for both histologic types. Multiple logistic regression analysis was used to obtain odds ratio (OR) estimates as measures of the relative risk of anal cancer associated with potential risk factors. All analyses were adjusted for age at reference date (continuous) and reference year (exact year). Some analyses were adjusted for the lifetime number of sexual partners (1, 2–4, 5–14, or ≥ 15 partners) and the presence of HPV-16 antibodies (negative or positive). Two-sided P values were calculated using the Wald test for interaction on a multiplicative scale and for differences between coefficients in a polytomous model.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

SCC was the most common histologic type of anal cancer diagnosed in the current study (72.2% excluding cloacogenic cancers, a unique subset of SCC tumors). Women were more likely than men to be diagnosed with cloacogenic histology (20.9% and 6.7%, respectively; P = 0.001) (Table 1). Women also were more likely to have their tumor localized to the anorectum (36.4% and 21.0%; P = 0.001). Overall, 87.9% of tumors in the study were HPV positive, with HPV-16 the most frequent HPV type detected (73.0%), followed by HPV-18 (6.9%). Neither the proportion of tumors that was positive for HPV nor the HPV type detected varied significantly by gender (P > 0.05). However, 97.7% of tumors in men who were not exclusively heterosexual contained HPV DNA, a rate that is higher than that observed in heterosexual men (78.0%; P = 0.004). The proportions of invasive tumors that were positive for HPV among women were somewhat higher but within the limits of chance compared with men (88.4% vs. 81.8%; P = 0.22). This was driven mainly by the cloacogenic tumors that were HPV positive (40 of 41 tumors; 97.6%). Only 8 of 20 specimens (40.0%) of anal adenocarcinoma were positive for HPV DNA. Two males had Kaposi sarcoma, and both were HPV DNA positive. One male (HPV DNA positive) and one female (HPV DNA negative) had a diagnosis of melanoma.

Table 1. Tumor Site, Histology, and Behavior of Anal Tumors by Human Papillomavirus DNA Positivity Rate and by Gender
CharacteristicMenWomen
No. (%)No. testedHPV positivity (%)No. (%)No. testedHPV positivity (%)
HPV+HPV-16+HPV-18+HPV+HPV-16+HPV-18+
  • HPV: human papillomavirus; HPV-16: HPV type 16; HPV-18: HPV type 18; SCC: squamous cell cancer; NOS: not otherwise specified.

  • a

    Numbers in parentheses indicate International Classification of Diseases for Oncology site codes.

  • b

    SCC of the rectum was included, because these tumors are more similar to anal cancer than to adenocarcinoma of the rectum.

Histologic type          
 SCC 94 (79.0)8192.677.87.4127 (67.9)9891.874.510.2
 Cloacogenic/basaloid  8 (6.7)6100.0100.00.0 39 (20.9)3597.194.30.0
 Adenocarcinoma 10 (8.4)825.012.512.5 18 (9.6)1250.016.70.0
 Basal cell cancer  3 (2.5)366.733.30.0  2 (1.1)10.00.00.0
 Melanoma  1 (0.8)1100.00.00.0  1 (0.5)00.00.00.0
 Kaposi sarcoma  2 (1.7)2100.0100.00.0  0 (0.0)00.00.00.0
 Sarcoma  1 (0.8)10.00.00.0  0 (0.0)00.00.00.0
 Overall11910286.371.66.918714689.074.06.9
Tumor sitea          
 Rectum (154.1)b  6 (5.0)5100.0100.020.0 11 (5.9)862.562.50.0
 Anal canal (154.2) 38 (31.9)3491.279.45.9 46 (24.6)3992.374.412.8
 Anus, NOS (154.3) 12 (10.1)988.977.811.1 12 (6.4)1080.080.010.0
 Anorectum (154.8) 25 (21.0)1984.263.20.0 68 (36.4)5488.977.81.9
 Skin (172.5, 173.5) 38 (31.9)3580.062.98.6 50 (26.7)3594.368.68.6
Tumor behavior          
 In situ 43 (36.1)3694.480.65.6 49 (26.2)3491.267.78.8
 Invasive 76 (63.9)6681.866.77.6138 (73.8)11288.475.96.3

Sexual orientation was related strongly to risk of anal cancer, with 47.1% of male cases reporting that they were not exclusively heterosexual (OR, 17.3; 95% confidence interval [95% CI], 8.2–36.1) (Table 2). The age distribution of the two groups of male cases (heterosexual and not exclusively heterosexual) differed vastly, and risk factors sometimes differed by sexual orientation; therefore, we estimated the risks overall and separately by sexual orientation using the entire male control group as the reference group, except as noted in the text and in Table 2. Men with ≥ 15 lifetime sexual partners were at high risk of anal cancer (heterosexual men: OR, 3.9 vs. and OR of 6.6 for men who were not exclusively heterosexual) independent of sexual orientation. Among men who were not exclusively heterosexual, receptive anal intercourse was related strongly to the risk of anal cancer (OR, 6.8; 95% CI, 1.4–33.8). A history of genital warts was related strongly to the risk of anal cancer, both overall (OR, 7.4; 95% CI, 3.2–17.1) and for each of the 2 groups of men. HSV-2-positive serology was found to be related to anal cancer risk only among men who were not exclusively heterosexual. The relation between seropositivity to HPV-16 and the risk of anal cancer did not appear to vary by sexual orientation in men. Seropositivity to HPV-16 was found to be related strongly to the risk of anal cancer (OR, 6.0; 95% CI, 3.3–10.9).

Table 2. Selected Characteristics in a Group of Men with Squamous Cell or Cloacogenic Anal Cancer (Cases) and in a Male Control Group (Controls) by Sexual Orientation
CharacteristicHeterosexualNonheterosexualAdjusted overall OR (95% CI)
Cases (n = 54) No. (%)Controls (n = 221) No. (%)Adjusted OR for cases (95% CI)Cases (n = 48) No. (%)Controls (n = 14) No. (%)Adjusted OR for cases (95% CI)
  • OR: odds ratio; 95% CI: 95% confidence interval; Ref: reference; HSV-2: herpes simplex virus type 2; HPV-16: human papillomavirus type 16.

  • a

    Adjusted for age at reference, reference year, and number of lifetime sexual partners.

  • b

    Cases and controls include all men, regardless of sexual orientation.

  • c

    Odds ratios compared cases with controls of similar sexual orientation, rather than comparing cases with all controls.

  • d

    Odds ratios were adjusted for age at reference and reference year.

  • e

    Adjusted for age at reference, reference year, number of lifetime sexual partners, and human papillomavirus type 16 serostatus.

Sexual orientationab       
 Heterosexual54 (52.9)221 (92.5)    Ref
 Bisexual/homosexual48 (47.1)14 (5.9)    17.3 (8.2–36.1)
 Missing04     
Age at reference       
 < 45 yrs14 (25.9)83 (37.6) 26 (54.2)7 (50.0)  
 45–59 yrs12 (22.2)69 (31.2) 12 (25.0)5 (35.7)  
 ≥ 60 yrs28 (51.9)69 (31.2) 10 (20.8)2 (14.3)  
Marital statusa       
 Married/living as married39 (72.2)184 (83.3)Ref8 (16.7)5 (35.7)RefRef
 Divorced4 (7.4)22 (10.0)0.6 (0.2–2.0)10 (20.8)3 (21.4)7.5 (2.6–21.5)1.9 (0.9–4.0)
 Widowed3 (5.6)3 (1.4)2.4 (0.4–13.4)0 (0.0)0 (0.0)2.8 (0.5–15.2)
 Never married8 (14.8)12 (5.4)2.5 (0.9–6.9)30 (62.5)6 (42.9)32.1 (12.2–84.3)8.7 (4.3–17.9)
No. of sexual partnerscd       
 16 (11.1)54 (24.5)Ref3 (6.3)3 (21.4)RefRef
 2–49 (16.7)47 (21.4)1.9 (0.6–5.8)5 (10.4)5 (35.7)0.8 (0.1–6.7)1.8 (0.7–4.4)
 5–1419 (35.2)56 (25.5)3.8 (1.4–10.6)9 (18.7)2 (14.3)3.9 (0.4–37.0)3.2 (1.4–7.5)
 ≥ 1520 (37.0)63 (28.6)3.9 (1.4–10.7)31 (64.6)4 (28.6)6.6 (0.9–47.1)5.3 (2.4–12.0)
 Missing01 00  
Age at first intercoursecd       
 > 20 yrs5 (20.8)35 (29.9)Ref10 (33.3)5 (55.6)RefRef
 18–19 yrs10 (41.7)49 (41.9)1.4 (0.4–5.2)8 (26.7)2 (22.2)1.0 (0.1–15.7)0.6 (0.3–1.6)
 < 18 yrs9 (37.5)33 (28.2)1.7 (0.4–7.0)12 (40.0)2 (22.2)10.0 (0.6–161.2)1.0 (0.4–2.5)
 Not asked30104 182  
Anal intercourseac       
 Never54 (100.0)  6 (12.5)7 (53.9)RefRef
 Ever0 (0.0)  42 (87.5)6 (46.1)6.8 (1.4–33.8)32.8 (12.4–86.8)
 Missing0  01  
Genital wartsa       
 Never48 (88.9)212 (95.9)Ref28 (59.6)13 (92.9)RefRef
 Ever6 (11.1)9 (4.1)3.5 (1.1–11.4)19 (40.4)1 (7.1)11.3 (4.4–28.9)7.4 (3.2–17.1)
  Ever on anus4 (7.4)0 (0.0)17 (35.4)0 (0.0) 
 Missing00 10  
HSV-2 antibodye       
 Negative35 (74.5)160 (79.6)Ref15 (32.6)7 (63.6)RefRef
 Positive12 (25.5)41 (20.4)1.0 (0.4–2.4)31 (67.4)4 (36.4)5.0 (2.2–11.4)2.4 (1.3–4.4)
 Not tested720 23  
HPV-16 L1 antibodya       
 Negative23 (48.9)169 (83.7)Ref23 (51.1)11 (100.0)RefRef
 Positive24 (51.1)33 (16.3)7.4 (3.5–15.8)22 (48.9)0 (0.0)6.0 (3.3–10.9)
 Not tested719 33  
Corticosteroid usea       
 Never48 (88.9)208 (94.5)Ref37 (77.1)13 (92.9)RefRef
 Ever6 (11.1)12 (5.5)2.1 (0.7–6.3)11 (22.9)1 (7.1)5.6 (2.1–14.8)3.2 (1.4–7.2)
 Missing01 00  

We evaluated ever use of corticosteroids to test the hypothesis that infection with HPV and progression to malignancy may be linked to immunosuppression. The overall risk of anal cancer in men who used corticosteroids was 3-fold (OR, 3.2; 95% CI, 1.4–7.2). Heterosexual cases reported an intermediate amount of corticosteroid use (11.1%) compared with not exclusively heterosexual cases (22.9%) and controls (5.5%).

The age distribution of women with SCC tumors differed from that of women with cloacogenic tumors, with the latter group being substantially older (Table 3). The risk of anal cancer with both histologic groups combined increased with the number of lifetime sexual partners, with a risk of 11.0 (95% CI, 5.5–22.1) associated with ≥ 15 partners relative to 1 partner. Young age at first intercourse (OR, 1.8; 95% CI, 1.1–2.8) and a history of anal intercourse (OR, 2.2; 95% CI, 1.4–3.3) also were found to be related to overall risk and varied only slightly by histology.

Table 3. Selected Characteristics of in a Group of Women with Squamous Cell or Cloacogenic Anal Cancer and in a Control Group of Women (Controls)
CharacteristicNo. of participants (%)OR (95% CI)
SCC (n = 127)Cloacogenic (n = 39)Controls (n = 1445)aSCCCloacogenicOverall
  • SCC: squamous cell cancer; OR: odds ratio; 95% CI: 95% confidence interval; Ref: reference; HSV-2: herpes simplex virus type 2; HPV-16: human papillomavirus type 16.

  • a

    Excluding 16 women who reported never having intercourse with a male.

  • b

    Adjusted for age at reference date, reference year, and number of lifetime sexual partners.

  • c

    Odds ratios were adjusted for age at reference and reference year.

  • d

    Adjusted for age at reference date, reference year, number of lifetime sexual partners, and human papillomavirus antibody status.

Age at reference      
 < 45 yrs32 (25.2)3 (7.7)790 (54.7)   
 45–59 yrs45 (35.4)11 (28.2)377 (26.1)   
 > 60 yrs50 (39.4)25 (64.1)278 (19.2)   
Marital statusb      
 Married/living as married57 (44.9)26 (66.7)1027 (71.1)RefRefRef
 Divorced41 (32.3)7 (18.0)216 (15.0)2.0 (1.3–3.2)0.9 (0.4–2.3)1.7 (1.1–2.6)
 Widowed18 (14.2)6 (15.4)77 (5.3)1.8 (0.9–3.5)0.7 (0.2–1.8)1.4 (0.8–2.4)
 Never married11 (8.7)0 (0.0)125 (8.7)2.5 (1.2–5.2)2.0 (1.0–4.2)
No. of male sexual partnersc      
 112 (9.5)9 (23.1)452 (31.4)RefRefRef
 2–444 (34.9)15 (38.5)417 (29.0)6.0 (3.1–11.9)3.5 (1.5–8.3)5.0 (2.9–8.5)
 5–1449 (38.9)13 (33.3)419 (29.1)9.7 (4.9–19.4)6.1 (2.4–15.7)7.8 (4.4–13.7)
 ≥ 1521 (16.7)2 (5.1)150 (10.4)15.9 (7.1–35.6)3.1 (0.6–15.7)11.0 (5.5–22.1)
Age at first intercourseb      
 > 20 yrs28 (22.2)11 (28.2)509 (35.4)RefRefRef
 18–19 yrs42 (33.3)12 (30.8)391 (27.2)1.6 (1.0–2.8)2.1 (0.9–5.3)1.7 (1.1–2.8)
 < 18 yrs56 (44.4)16 (41.0)540 (37.5)1.5 (0.9–2.6)3.1 (1.2–7.8)1.8 (1.1–2.8)
Anal intercourseb      
 Never83 (65.9)29 (74.4)1128 (78.5)RefRefRef
 Ever43 (34.1)10 (25.6)309 (21.5)2.1 (1.3–3.3)2.9 (1.2–7.1)2.2 (1.4–3.3)
Genital wartsb      
 Never98 (77.2)38 (97.4)1309 (90.8)RefRefRef
 Ever29 (22.8)1 (2.6)132 (9.2)3.1 (1.9–5.3)0.4 (0.0–3.0)2.6 (1.6–4.3)
 Ever on anus17 (13.4)1 (2.6)20 (1.4)15.1 (6.8–33.5)3.5 (0.4–31.9)13.6 (6.2–29.6)
HSV-2 antibodyd      
 Negative67 (55.8)22 (56.4)933 (73.1)RefRefRef
 Positive53 (44.2)17 (43.6)343 (26.9)1.1 (0.7–1.8)1.8 (0.8–4.0)1.3 (0.8–1.9)
 Not tested70169   
Seropositivity for HPV-16b      
 Negative63 (57.8)24 (61.5)1084 (85.0)RefRefRef
 Positive46 (42.2)15 (38.5)191 (15.0)3.9 (2.5–6.1)4.1 (2.0–8.6)4.1 (2.7–6.2)
 Not tested180175   
Corticosteroid usec      
 Never106 (83.5)29 (74.4)1329 (92.0)RefRefRef
 Ever21 (16.5)10 (25.6)116 (8.0)2.0 (1.1–3.4)3.0 (1.3–6.7)2.3 (1.4–3.7)

A history of genital warts was found to be related strongly to SCC tumors (OR, 3.1; 95% CI, 1.9–5.3), particularly for warts that occurred on the anus (OR, 15.1; 95% CI, 6.8–33.5) (Table 3). HSV-2-positive serology was associated somewhat with an excess risk among women with cloacogenic tumors (OR, 1.8; 95% CI, 0.8–4.0), but not SCC tumors, after adjusting for the number of partners and for HPV serology.

The risk for anal cancer associated with seropositivity for HPV-16 (OR, 4.1; 95% CI, 2.7–6.2) did not vary by histology compared with women who were seronegative. None of the cloacogenic tumors were positive for HPV-18 DNA. The risk of anal cancer among women, both overall and by histology (Table 3), was related to corticosteroid use (OR, 2.3; 95% CI, 1.4–3.7).

Cigarette smoking has been related to the risk of SCC anal cancer as well as squamous cell cervical, penile, vulvar, and vaginal cancers.10 However, concern remains that smoking may be an etiologic factor only among certain subgroups11 and that confounding by sexual factors may not have been ruled out. To explore this hypothesis, we analyzed the correlation between smoking and anal carinoma for the complete group of men and women in the current study population and then for various subgroups.

Overall, there was a strong relation between status as a current smoker at the time of study enrollment and the risk of anal cancer among men (adjusted OR, 3.9; 95% CI, 1.9–8.0), because 56.9% of cases were current smokers compared with 22.2% of controls (Table 4). The risks of developing anal cancer associated with status as a current smoker did not vary according to age, sexual orientation, stage of disease, number of sexual partners, HPV DNA positivity, HPV-16 seropositivity, or corticosteroid use.

Table 4. Correlation between Cigarette Smoking Status and the Risk of Squamous Cell and Cloacogenic Anal Cancer According to Patient and Tumor Characteristics in Men
CharacteristicNo. (row %)a 
Never smokersFormer smokersCurrent smokersOR (95% CI)b
CasesControlsCasesControlsCasesControlsFormer smokersCurrent smokers
  • OR: odds ratio; 95% CI: 95% confidence interval; HPV: human papillomavirus; OR: odds ratio; 95% CI: 95% confidence interval; HPV: human papillomavirus.

  • a

    Row percent indicates the percentages of never smokers, former smokers, and current smokers in the case group and the percentages of never smokers, former smokers, and current smokers in the control group.

  • b

    The reference group was is never smokers; odds ratios were adjusted for age at the reference date, except where noted.

  • c

    Adjusted for age at reference date, reference year, number of lifetime sexual partners, and human papillomavirus antibody status.

  • d

    Odds ratios represent the risk of being in a given case group relative to being in the control group.

Overallc17 (16.7)89 (37.2)27 (26.5)97 (40.6)58 (56.9)53 (22.2)0.9 (0.4–1.9)3.9 (1.9–8.0)
Age at reference        
 < 45 yrs6 (15.0)45 (47.9)6 (15.0)25 (26.6)28 (70.0)24 (25.5)1.8 (0.5–6.1)8.6 (3.1–23.8)
 45–59 yrs4 (16.7)21 (28.4)4 (16.7)33 (44.6)16 (66.7)20 (27.0)0.6 (0.1–2.9)4.9 (1.3–18.3)
 ≥ 60 yr7 (18.4)23 (32.4)17 (44.7)39 (54.9)14 (36.8)9 (12.7)1.5 (0.5–4.1)5.4 (1.6–18.1)
Tumor behaviord        
 In situ7 (16.7) 10 (23.8) 25 (59.5) 1.7 (0.6–4.8)6.1 (2.4–15.3)
 Invasive10 (16.7) 17 (28.3) 33 (55.0) 1.2 (0.5–2.9)5.6 (2.5–12.6)
Sexual orientation (exclusively heterosexual)        
 Yes11 (20.4)82 (37.1)16 (29.6)91 (41.2)27 (50.0)48 (21.7)1.1 (0.5–2.6)4.2 (1.9–9.3)
 No6 (12.5)3 (21.4)11 (22.9)6 (42.9)31 (64.6)5 (35.7)1.8 (0.4–9.1)7.1 (1.7–30.1)
No. of sexual partners        
 11 (11.1)29 (50.9)5 (55.6)21 (36.8)3 (33.3)7 (12.3)5.2 (0.5–50.8)10.3 (0.9–118.3)
 2–44 (28.6)19 (36.5)2 (14.3)19 (36.5)8 (57.1)14 (26.9)0.4 (0.1–2.8)2.8 (0.7–11.4)
 5–144 (14.3)14 (24.1)8 (28.6)30 (51.7)16 (57.1)14 (24.1)0.8 (0.2–3.3)4.0 (1.0–14.9)
 ≥ 158 (15.7)22 (32.8)12 (23.5)27 (40.3)31 (60.8)18 (26.9)1.2 (0.4–3.5)5.0 (1.8–13.5)
Ever had genital warts        
 Never13 (17.1)88 (38.4)24 (31.6)93 (40.6)39 (51.3)48 (21.0)1.5 (0.7–3.2)5.5 (2.7–11.4)
 Ever4 (16.0)1 (10.0)2 (8.0)4 (40.0)19 (76.0)5 (50.0)0.1 (0.0–2.0)0.9 (0.1–10.2)
 Ever on anus4 (19.1)0 (0.0)2 (9.5)0 (0.0)15 (71.4)0 (0.0)
Tumor positive for HPVd        
 No0 (0.0) 3 (50.0) 3 (50.0) 
 Yes13 (16.1) 20 (24.7) 48 (59.3) 1.3 (0.6–2.8)6.2 (3.1–12.6)
Tumor positive for HPV-16d        
 No2 (11.1) 7 (38.9) 9 (50.0) 3.2 (0.6–16.4)7.6 (1.6–36.4)
 Yes11 (15.9) 16 (23.2) 42 (60.9) 1.2 (0.5–2.8)6.4 (3.0–13.6)
Tumor positive for HPV-18d        
 No13 (16.1) 22 (27.2) 46 (56.8) 1.5 (0.7–3.1)6.0 (2.9–12.1)
 Yes0 (0.0) 1 (16.7) 5 (83.3) 
Seropositive for HPV-16        
 No10 (21.7)65 (35.5)13 (28.3)77 (22.4)23 (50.0)41 (42.1)1.0 (0.4–2.4)3.6 (1.6–8.4)
 Yes7 (15.2)15 (44.1)8 (17.4)13 (17.7)31 (67.4)6 (38.2)1.4 (0.4–5.3)11.3 (3.2–39.8)
Seropositive for HPV-18        
 No9 (17.7)52 (36.6)12 (23.5)61 (43.0)30 (58.8)29 (20.4)1.1 (0.4–3.0)6.0 (2.5–14.3)
 Yes8 (19.5)28 (38.4)9 (22.0)27 (37.0)24 (58.5)18 (24.7)1.1 (0.4–3.4)4.7 (1.7–12.9)
Corticosteroid use        
 Never15 (17.7)86 (38.2)26 (30.6)90 (40.0)44 (51.8)49 (21.8)1.5 (0.7–3.1)5.2 (2.6–10.3)
 Ever2 (11.8)3 (23.1)1 (5.9)7 (53.8)14 (82.4)3 (23.1)0.3 (0.0–5.1)6.7 (0.7–68.5)

Among the women in the current study, status as a current smoker at the time of study enrollment similarly was associated with an increased risk for anal cancer (adjusted OR, 3.8; 95% CI, 2.3–6.2), because 48.2% of female cases were current smokers compared with 23.8% of controls (Table 5). The risk for anal cancer associated with status as a current smoker did not appear to vary significantly by age; however, there was some indication that the risk associated with status as a current smoker was somewhat lower for women age ≥ 60 years (OR, 2.5; 95% CI, 1.3–4.6) compared with women age < 60 years (OR, 7.2; 95% CI, 4.0–13.0; P for interaction = 0.07).

Table 5. Correlation between Cigarette Smoking Status and the Risk of Squamous Cell and Cloacogenic Anal Cancer According to Patient and Tumor Characteristics in Women
CharacteristicNo. (row %)a 
Never smokersFormer smokersCurrent smokersOR (95% CI)b
CasesControlsCasesControlsCasesControlsFormer smokersCurrent smokers
  • OR: odds ratio; 95% CI: 95% confidence interval; HPV: human papillomavirus.

  • a

    Row percent indicates the percentages of never smokers, former smokers, and current smokers in the case group and the percentages of never smokers, former smokers and current smokers in the control group.

  • b

    The reference group was never smokers; odds ratios were adjusted for age at the reference date, except where noted.

  • c

    Adjusted for age at reference date, reference year, number of lifetime sexual partners, and human papillomavirus antibody status.

  • d

    Odds ratios represent the risk of being in a given case group relative to being in the control group.

Overallc41 (24.7)728 (50.4)45 (27.1)373 (25.8)80 (48.2)344 (23.8)1.5 (0.9–2.6)3.8 (2.3–6.2)
Age at reference        
 < 45 yrs6 (17.1)423 (53.5)7 (20.0)159 (20.1)22 (62.9)208 (26.3)2.7 (0.9–8.3)7.9 (3.1–19.9)
 45–59 yrs10 (17.9)172 (45.6)15 (26.8)130 (34.5)31 (55.4)75 (19.9)1.9 (0.8–4.3)6.7 (3.1–14.5)
 ≥ 60 yrs25 (33.3)133 (47.8)23 (30.7)84 (30.2)27 (36.0)61 (21.9)1.5 (0.8–2.8)2.5 (1.3–4.6)
Histologyd        
 Squamous cell31 (24.4) 25 (19.7) 71 (55.9) 1.4 (0.8–2.4)5.4 (3.4–8.5)
 Cloacogenic10 (25.6) 20 (51.3) 9 (23.1) 3.2 (1.5–7.1)2.1 (0.8–5.4)
Tumor behaviord        
 In situ8 (16.3) 10 (20.4) 31 (63.3) 2.2 (0.9–5.8)8.4 (3.8–18.6)
 Invasive33 (28.2) 35 (29.9) 49 (41.9) 1.8 (1.1–3.0)3.6 (2.2–5.9)
No. of sexual partners        
 110 (47.6)303 (67.0)8 (38.1)100 (22.1)3 (14.3)49 (10.8)1.9 (0.7–4.9)1.3 (0.3–5.1)
 2–421 (35.6)207 (49.6)12 (20.3)105 (25.2)26 (44.1)105 (25.2)0.8 (0.4–1.9)2.1 (1.1–4.1)
 5–148 (12.9)170 (40.6)17 (27.4)116 (27.7)37 (59.7)133 (31.7)2.0 (0.8–5.0)6.5 (2.8–15.5)
 ≥ 151 (4.3)45 (30.0)8 (34.8)49 (32.7)14 (60.9)56 (37.3)4.7 (0.5–41.0)12.4 (1.5–100.4)
Ever had genital warts        
 Never37 (27.2)672 (51.3)41 (30.1)334 (25.5)58 (42.6)303 (23.1)1.9 (1.2–3.0)3.9 (2.5–6.1)
 Ever4 (13.3)55 (41.7)4 (13.3)37 (28.0)22 (73.3)40 (30.3)1.4 (0.3–6.2)9.8 (2.9–33.6)
 Ever on anus2 (11.1)7 (35.0)2 (11.1)5 (25.0)14 (77.8)8 (40.0)1.2 (0.1–13.1)7.9 (1.1–54.9)
Tumor positive for HPVd        
 No1 (11.1) 4 (44.4) 4 (44.4) 6.6 (0.7–59.3)9.3 (1.0–84.0)
 Yes30 (24.2) 38 (30.6) 56 (45.2) 2.1 (1.3–3.5)4.4 (2.7–7.1)
Tumor positive for HPV-16d        
 No6 (22.2) 7 (25.9) 14 (51.9) 2.0 (0.7–5.9)5.3 (2.0–13.9)
 Yes25 (23.6) 35 (33.0) 46 (43.4) 2.3 (1.3–4.0)4.3 (2.6–7.3)
Tumor positive for HPV-18d        
 No27 (22.0) 41 (33.3) 55 (44.7) 2.5 (1.5–4.2)4.8 (2.9–7.9)
 Yes4 (40.0) 1 (10.0) 5 (50.0) 0.4 (0.0–4.0)2.7 (0.7–10.3)
Seropositive for HPV-16        
 No16 (18.4)555 (51.2)29 (33.3)266 (24.5)42 (48.3)263 (24.3)3.3 (1.7–6.3)6.7 (3.6–12.4)
 Yes22 (36.1)87 (45.5)12 (19.7)56 (29.3)27 (44.3)48 (25.1)0.7 (0.3–1.6)3.5 (1.7–7.5)
Seropositive for HPV-18        
 No31 (29.5)544 (50.3)30 (28.6)274 (25.3)44 (41.9)263 (24.3)1.6 (0.9–2.7)3.5 (2.1–5.8)
 Yes7 (16.3)96 (50.3)11 (25.6)49 (25.7)25 (58.1)46 (24.1)3.0 (1.1–8.6)10.9 (4.1–29.2)
Corticosteroid use        
 Never29 (21.5)666 (50.1)38 (28.1)345 (26.0)68 (50.4)318 (23.9)2.2 (1.3–3.7)5.6 (3.5–8.9)
 Ever12 (38.7)62 (53.4)7 (22.6)28 (24.1)12 (38.7)26 (22.4)1.0 (0.3–3.0)3.0 (1.1–8.0)

Women who had cloacogenic tumors had a somewhat lower risk associated with status as a current smoker at the time of study enrollment (OR, 2.1; 95% CI, 0.8–5.4) compared with women who had SCC tumors (OR, 5.4; 95% CI, 3.4–8.5); however, this difference was within the limits of chance (P for difference = 0.10). The risk estimates for anal cancer among women associated with status as a current smoker did not appear to vary according to HPV DNA, HPV antibody status, or corticosteroid use. In contrast to our findings in men, the risk of anal cancer associated with status as a current smoker was higher among women with in situ tumors (OR, 8.4; 95% CI, 3.8–18.6) than among women with invasive tumors (OR, 3.6; 95% CI, 2.2–5.9) (P for difference = 0.06). In addition, the risk associated with status as a current smoker was elevated significantly for women who had any number of lifetime sexual partners > 1, for whom the risk associated with current smoking was not elevated (OR, 1.3; 95% CI, 0.3–5.1).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

There were some limitations to the current study that should be considered in the interpretation of the results. We were only able to get a census from approximately 93.0% of the households we attempted to contact, and we were able to interview only 63.1% of patients in the case group and 65.5% of potential control participants. To the extent that those who did and did not participate differed with regard to exposures of interest, and such differences varied by case–control status, our results may be biased. This issue may be of concern if potential cases or controls declined to participate due to impaired health (e.g., because of human immunodeficiency virus [HIV] positivity). However, the proportion of male cases who never married (a marker for sexual orientation for which data were available in the registry data base) among those we were unable to interview (43.3%) did not differ substantially from that of male cases who we were able to interview (36.4%). In addition, the sensitive nature of the study may have contributed to the lower than average response rate in both men and women. It is difficult to predict the direction or the extent of any resulting bias from this source.

We were able to obtain tumor tissue for HPV DNA testing from 262 cancer cases (85.6%) and amplifiable DNA in tissue from 248 (81.0%) of those cases. We relied on expert pathologic review by the study pathologist (P.L.P.) for only 41% of the cases; however, the histologic determination matched that recorded in the local registry for all of the tumors that were reviewed.

Because anal cancer is a rare disease, and tumors with cloacogenic histology are less frequent than tumors with squamous cell histology, some of our subgroup analyses suffer from small numbers. These results need to be interpreted cautiously. In addition, among men, we grouped the 8 cloacogenic cases with the 94 squamous cases in the risk factor analysis. However, the risk factors for the two histologic types did not appear to differ among men in this study.

In the current study, HPV DNA was found in all histologic types of anal cancer, at all anatomic sites, and in both in situ and invasive tumors. The most common HPV DNA type detected was HPV-16, which was detected in 73% of tissues tested. HPV-18 was detected in 6.9% of the tissues tested. The proportion of HPV-positive tissues in the current study was similar to that found by Frisch et al.4 in a population-based study in Denmark and Sweden (90% vs. 88%), although that study found that lower percentages of anal cancers were HPV positive among males (63% vs. the 82% reported herein). The high proportion of tumors with detectable HPV suggests that infection with HPV is a necessary cause of anal cancer, similar to that of cervical cancer. If reporting of sexual orientation was accurate in both studies, and if HPV DNA prevalence is truly higher in anal tissues from homosexual men, then the difference in HPV detection rates between the two studies may be attributable to the higher percentage of cases in the current study that were not exclusively heterosexual (47.1% vs. 14.3% in the study by Frisch et al.). Frisch et al.4 also found that 100% of the tumors diagnosed in males who reported homosexual experience were positive for HPV DNA, a rate similar to that noted in the current study (98%). However, the HPV DNA prevalence was somewhat higher among heterosexual men in the current study compared with the Denmark/Sweden study (78% vs. 58%, respectively). Although Frisch et al. included only patients with high-risk HPV in their study results, it is unlikely that this accounts for the difference, because only 3.4% of the tumors tested in the current study were positive for a low-risk HPV type.

It is interesting to note that, in the current study, there was an absence of HPV-18 DNA in any of the cloacogenic tumors, and HPV-16 was found in both Kaposi sarcoma specimens. One of the two melanomas contained an unknown HPV type. Eight of the 20 adenocarcinoma tissue specimens (40%) were positive for HPV, with 3 specimens that contained HPV-16, 2 specimens that contained HPV-18, and 3 specimens that contained other HPV types.

In the current study, we were unable to test the collected blood samples for the presence of HIV or to inquire about infection with HIV because of the human subjects boards concerns regarding the availability of adequate counseling services. We did evaluate the relation between anal cancer and a marker of immunosuppression: the use of corticosteroids. The risk for anal cancer associated with corticosteroid use was found to be elevated significantly among men (OR, 3.2), particularly among men who were not exclusively heterosexual (OR, 5.6), and among women (OR, 2.3). Greater than 25% of women who had a diagnosis of cloacogenic anal cancer reported the use of corticosteroids, compared with 8% of controls. The most common reason (41.5%) for the use of corticosteroids was for the treatment of skin conditions (e.g., acne, eczema, psoriasis, skin rash, etc.), followed by allergies and asthma (14.8%). Although corticosteriods are only weakly immunosuppressive, and their use is transient, the consistent finding of an elevated risk among both heterosexual men and nonheterosexual men, as well as among women with both histologies, indicates that either the drug or the conditions the drug is used for may enhance the serious risk of HPV-related disease. Corticosteroid users may be a group of individuals who should be monitored more closely for HPV-related cancers, particularly if they have other risk factors.

Consistent with our previous report based on earlier years of diagnosis3 and on results from other population-based studies12–14 and nonpopulation-based studies,15–17 men who were not exclusively heterosexual were at increased risk of anal cancer. In this study, nearly half of the men with anal cancer (47.1%) were not exclusively heterosexual, compared with 6.0% of male controls (OR, 17.3; 95% CI, 8.2–36.1). This is a much higher proportion compared with that reported in our earlier study (36.8% of cases compared with 1.6% of controls) but is similar to that reported by Holly et al.11 (47.0% of cases compared with 10.1% of controls). Frisch et al.13 found that only 15% of men with anal cancer and none of the male controls reported homosexual contact.

In contrast to our previous study,3 women who reported experience with anal intercourse were at increased risk for the development of both SCC and cloacogenic anal cancer (overall adjusted OR, 2.2; 95% CI, 1.4–3.3). The prevalence of this exposure (32.1% of cases and 21.4% of controls) has increased compared with our previous results (16.9% of cases and 10.8% of controls).3 The latter levels are similar to the reported prevalence of anal intercourse among women in two other population-based studies conducted by Holly et al.12 and Frisch et al.13 The years of diagnosis in our previous study (1978–1985) and those in the study by Holly et al. (1979–1986) were similar, whereas Frisch et al. used patients who were diagnosed between 1991 and 1994.

We can only speculate with regard to the apparent increase in the practice of anal intercourse among women and the prevalence of homosexuality among men in the current study. Because our methods of case and control ascertainment have not changed, it is possible that, in recent years, these exposures have increased or that individuals are more likely to answer affirmatively to these sensitive questions. Since the onset of the acquired immune deficiency syndrome (AIDS) epidemic, individuals may be speaking more freely concerning sexual practices. Similar comments may apply to increased reporting of the lifetime number of sexual partners and a history of genital warts over that reported in prior studies.3–5, 12

Status as a current smoker at the time of study enrollment has been a consistent risk factor in most studies of anal cancer3, 5, 12, 18 and in the larger population-based studies of other HPV-related SCC anogenital malignancies.6, 10, 19–24 In this study, current smoker status was related to the risk of anal cancer in both men and women. The current study findings are in contrast to those of Frisch et al.,4 who found an increased risk of anal cancer attributable to cigarette smoking among women (OR, 1.9; (95% CI, 1.3–2.8) but not among men (OR, 1.6; 95% CI, 0.8–3.3). Furthermore, those authors found that the relation between current smoking and risk of anal cancer among women was confined to premenopausal women. Frisch et al.11, 25 speculated that the findings of a strong correlation between status as a current smoker and the risk of anal cancer may be due to the lack of adjustment for confounding by sexual factors and that smoking may represent an important risk factor only among women who are not estrogen deficient (i.e., premenopausal women). To explore these hypotheses further, we evaluated the effect of current smoking among men and women by age, histology, tumor behavior, and by all risk factors for anal cancer stratifying by levels of exposure.

Among men, the risk associated with current smoker status was found to be important in all age groups as well as all other risk factors for cancer, except for men who reported a history of genital warts. Similarly, we found that current smoker status among women was important at all levels of other exposures, with the exception of the small subgroup of women who had only one lifetime sexual partner. We did note that the magnitude of risk of being a current smoker was somewhat lower among women age ≥ 60 years (OR, 2.5; 95% CI, 1.3–4.6) and among women with cloacogenic tumors (OR, 2.1; 95% CI, 0.8–5.4). In the current study, one-third (25 of 75 patients) of the cases of SCC and cloacogenic anal cancer reported among women age ≥ 60 years had the cloacogenic subtype. Therefore, the high proportion of older women with cloacogenic tumors in the current study may account in part for the lower risk associated with smoking in that group. Likewise, if the study of Frisch et al. had included a large number of cloacogenic anal tumors among postmenopausal women, that factor may account for the reported absence of an association between current smoking and anal cancer.

The role that cigarette smoking plays in the etiology of anal and other anogenital tumors still has not been defined, despite the early observations of an association.5, 10, 12, 19, 23 Current smoking at the time of diagnosis has been associated with all HPV-related squamous cell anogenital tumors, regardless of whether HPV DNA is present in the tumor. Because it appears that it is most important if patients with cancer are current smokers at the time of diagnosis, current smoking likely has a promotional effect at late stages of disease progression. Garrett et al.26 found that HPV-18-immortalized human keratinocytes treated with a low dose of a compound similar to N-nitroso compounds found in tobacco and a single exposure to TPA, a tumor promoter, were converted to a malignant phenotype. It also has been shown that nicotine inhibits apoptosis, which may promote tumor growth.27–29 Cigarette smoking also has been related to immunosuppression.30–32

The prevalence of many of the exposures studied herein (including anal intercourse, cigarette smoking in women, a greater number of sexual partners, and a history of genital warts) has increased from the levels reported in prior studies among both patients with anal cancer and our population-based control group. The reported increases simply may reflect an increased willingness of study participants to speak freely regarding sensitive sexual issues, or they may represent true increases in exposure prevalence. If the latter hypothesis is correct, then we predict that the incidence of anal cancer in the U.S. will continue to rise in the coming decades.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES