Palefsky, Joel M.1,2,5; Holly, Elizabeth A.3; Ralston, Mary L.3; Jay, Naomi2; Berry, J Michael2; Darragh, Teresa M.4
The incidence rate for anal cancer among men with a history of receptive anal intercourse is several times higher than the current rate of cervical cancer in women in the United States, with an estimated incidence of 35 out of 100 000 prior to the HIV epidemic [1,2]. Anal cancer shares many biological characteristics with cervical cancer, including a similar histopathologic appearance and association with human papillomavirus (HPV) infection [3–5]. There is histopathologic similarity between cervical squamous intra-epithelial lesions, the precursor lesions to cervical cancer, and anal squamous intra-epithelial lesions and both are associated with HPV infection [4,5]. Several recent studies have shown that both anal squamous intra-epithelial lesions and anal HPV infection are more common in HIV-positive homosexual and bisexual (hereafter called homosexual) men than in HIV-negative homosexual men [6–11] raising the possibility that HIV-positive men may be at even higher risk of developing anal cancer than are HIV-negative homosexual men. Data are conflicting on the degree of excess incidence of anal cancer in HIV-positive homosexual men compared with HIV-negative homosexual men [12,13].
Whereas cervical squamous intra-epithelial lesions span a wide range of histopathologic abnormalities ranging from low-grade to high-grade, most cervical cancers are believed to arise from high-grade squamous intraepithelial lesions (HSIL). Detection of cervical HSIL through routine cytologic screening and treatment is believed to be responsible for much of the decline in the cervical cancer incidence in the last few decades. Although there are no data to demonstrate that anal HSIL directly progresses to invasive anal cancer, the strong similarities with cervical HSIL and the high incidence of anal cancer in populations known to have high rates of anal squamous intra-epithelial lesions suggest that anal HSIL is the precursor lesion for anal cancer. Little is known about the incidence and risk factors for anal HSIL in HIV-positive and HIV-negative homosexual men. This report describes the high incidence and risk factors for anal HSIL in a 4-year prospective study of HIV-positive and HIV-negative homosexual men.
This study was conducted with the approval of the Committee on Human Research of the University of California, San Francisco. Homosexual and bisexual participants were recruited into this prospective cohort study between November 1991 and March 1994 from three sources: the San Francisco Men's Health study (SFMHS) , the San Francisco General Hospital cohort (SFGH) study , and at the University of California at San Francisco (UCSF) using newspaper advertisements. Three hundred and forty-six HIV-positive and 262 HIV-negative men were enrolled at baseline. At baseline and at each follow-up visit, an anal cytology sample was obtained . Briefly, two Dacron (Hardwood Products, Guilford, Maine, USA) swabs were consecutively inserted for HPV testing, followed by anoscopy with colposcopic magnification after application of 3% acetic acid as described previously . An anal biopsy was performed if one or more lesions were visualized with the colposcope. If more than one lesion was seen, biopsies were obtained from areas within the lesions that had different colposcopic appearances. Anal biopsies were fixed in 10% formalin and embedded in paraffin for routine histopathologic examination. Some visible lesions were not biopsied due to the patient's refusal or medical contra-indications such as recent acetylsalicylic acid intake, neutropenia (fewer than 1000 cells × 106/l), thrombocytopenia (platelet count less than 65 000 cells × 106/l), or concurrent intra- or perianal bacterial or herpes simplex virus infection.
HIV-positive subjects with normal cytology and no lesions seen at baseline anoscopy underwent anal examinations every 6 months as described above. HIV-negative subjects with normal cytology and no lesions seen at baseline were examined every 12 months to maximize subject retention. Both HIV-positive and HIV-negative subjects who had baseline anal squamous intra-epithelial lesions were examined every 3 months. All HIV-positive and HIV-negative subjects who had low-grade squamous intra-epithelial lesions (LSIL) detected at any visit were followed every 3 months thereafter. Subjects with histologically-confirmed HSIL were referred for treatment.
Three HIV-negative men seroconverted to HIV-positive during the 4-year follow-up period. One man was diagnosed with atypical squamous cells of undetermined significance (ASCUS) at baseline. At his 1-year visit he was HIV-positive and was diagnosed with LSIL. Nine months later he was diagnosed with HSIL. The two other men were normal at baseline. One was HIV-positive at his 2-year visit with a normal anal diagnosis at that time. He was followed for one more year and was normal except for one indeterminate diagnosis. The final man was HIV-positive at his 3-year visit and had ASCUS at that time. One year later he was diagnosed with LSIL. He developed HSIL at 4.5 years, but this was not included in the data analysis since it was beyond the 4-year period of analysis.
Interpretation of anal cytology and histology
Anal cytology and histology were evaluated without knowledge of previous pathology results, HIV status of the subject, HPV DNA results, results of other diagnostic tests, or of any questionnaire data. Anal cytology results were classified as benign, ASCUS, LSIL or HSIL using current criteria for evaluation of cervical cytology . Anal histology results were classified as benign, LSIL or HSIL. Subjects were classified as normal if they had benign cytology and no lesions visible at anoscopy or if they had benign cytology and a lesion that was determined to be histologically benign. If a lesion was visible at anoscopy but was not biopsied, and the accompanying cytology was normal, the diagnosis was considered indeterminate.
Measurement of HIV status and CD4 levels
Enzyme-linked immunosorbent assay (ELISA) screening was used to determine HIV status at baseline in all subjects, and positive ELISA tests were confirmed with a Western blot assay. HIV-negative subjects were retested for HIV infection every 12 months. Different techniques were used over the course of this study to measure CD4 levels and appropriate adjustments made to correct for differences in these techniques using calibration curves that compared the results of the different methods using the same samples. For HIV-positive subjects CD4 levels were classified as greater than 500, 200–500, and less than 200 cells × 106/l.
Because of its high sensitivity, polymerase chain reaction (PCR) was used in this study to detect low-level HPV infection. Since a positive PCR result does not discriminate between low-level and high-level infection, ‘hybrid capture’ (HC; Digene Diagnostics Inc., Silver Spring, Maryland, USA), a non-amplification-based test that measures high levels of HPV infection, was used in addition. The quantity of HPV DNA in the specimen can be determined by measuring the HC relative light unit (RLU) ratio, computed by dividing the chemiluminescent signal of the specimen by that obtained with a control sample containing 10 pg/ml of HPV 16 DNA.
Two hundred microlitres of specimen in Sample Transport Medium (Digene Diagnostics, Inc.,) were used for PCR and the remainder was used for HC. The PCR was performed using MY09/MY11 consensus HPV L1 primers as well as primers for amplification of the human ß-globin gene . After 30 amplification cycles, 6 µl of amplification mixture were applied to a nylon membrane and probed with a biotin-labelled HPV L1 consensus probe mixture . A separate membrane was probed with a biotin-labelled probe to the human ß-globin gene. Specimens positive with the HPV consensus probe mixture were then studied to determine the specific HPV type by preparing membranes as described above with 6 µl of specimen. The membranes were studied with probes to 29 different HPV types (6,11, 16, 18, 26, 31, 32, 33, 35, 39, 40, 45, 51, 52, 53, 54, 55, 56, 58, 59, 61, 66, 68, 69, 70, 73, AE2, Pap 155, Pap 291, as well as the following 10 types together in a probe mixture: HPV 2, 13, 34, 42, 57, 62, 64, 67, 72, and W13B). Specimens negative for ß-globin gene amplification were excluded from analysis. Negative controls for each experiment consisted of amplification of the solution containing all of the above components except for sample DNA. Positive controls included amplification of cloned HPV DNA. To ensure reproducibility of results, five randomly chosen specimens were amplified and blotted twice on different parts of each membrane. For analysis, the results of baseline HPV testing and testing at 12-month intervals were used for all subjects. A subject was considered always to have a specific type if he tested positive for that type on all visits for which PCR was available, never to have the type if all of his tests were negative, and sometimes to have it if he had both positive and negative tests for that type.
HC was conducted according to the manufacturer's recommendations. Two separate tests were performed on each anal specimen, consisting of 50 µl for the HPV types associated with a low risk of cervical cancer in Probe Group A (6/11/42/43/44), and 50 µl for detection of the HPV types associated with intermediate and high risk of cervical cancer in Probe Group B (16/18/31/33/35/45/51/52/56). The RLU ratio was determined for each specimen and classified as negative (< 1.0) or positive (> 1.0). For some analyses, positive specimens were divided into those with an RLU ratio of 1–10, 10−100 and > 100. A subject's HPV status using the RLU ratio was summarized as follows: negative, always < 1; low, sometimes > 1 and sometimes < 1, never > 100; medium, usually > 10, or sometimes > 100 and sometimes < 10; and high, always > 10 and sometimes > 100.
HIV-positive subjects were interviewed every 6 months during the study and asked about their current status regarding HIV-related diseases and conditions, medications, sexually transmitted diseases, anal conditions such as haemorrhoids and fissures, and behavioural topics such as smoking, drinking, substance use, and sexual practices that might be associated with anal disease. HIV-negative subjects were interviewed every year. A subject was categorized as exposed to the risk factor during the study if he responded affirmatively on any of the questionnaires administered to him after baseline up to and including the time of his first diagnosis of HSIL. Otherwise he was considered unexposed. For many behavioural variables, almost all subjects had engaged in the activity at least occasionally. For these potential risk factors, summary variables were developed to characterize exposure to the activity . Different measures of duration and quantity of activity were used depending upon the activity.
Cutpoints were determined from the combined distributions of the HIV-positive and HIV-negative men. The summary risk factor exposure categories created by this process were used to divide study subjects into the following exposure groups during the 4-year follow- up: never, men who were not exposed to the risk factor during the 4-year follow-up; low exposure, men who occasionally engaged in the activity during the study; moderate exposure, men who engaged in the activity during most of the study; and high exposure, men who engaged in the activity throughout the study and usually engaged in it extensively.
The subject's status at 4 years was recorded as the highest grade of anal lesion identified either by cytology or histology during the 4-year interval. If biopsy results were available, the highest grade of lesion from cytology and histology was used to determine the diagnosis. All data up to a subject's 4-year visit or diagnosis of HSIL were included. The 4-year visit was defined as the visit closest to 4 years after baseline among visits less than 4.25 years after baseline. Time to develop HSIL was defined as the time after baseline that a subject was first diagnosed with HSIL either by cytology or histology. The life-table method was used to estimate the survival curve with 1-year time intervals to describe the course of HSIL development . The log-rank test was used to compare survival curves in different strata. Person-years analysis was used to evaluate potential risk factors for HSIL . Length of follow-up was rounded to the nearest 0.25 years in this analysis. Person-years at risk and HSIL incidence allocation was based on a subject's most recent HIV and CD4 tests. To evaluate HPV infection and questionnaire topics, subjects were assigned to a constant rather than time-varying risk category based on overall exposure to the risk factor during the follow-up period. Subjects with usable HPV results at fewer than two visits (including baseline) were excluded from analyses involving HPV infection. The SAS statistical package was used for all data analysis .
The three sets of men recruited for this study had similar demographic characteristics. 91% were Caucasian not Hispanic, 7% were Hispanic, 2% were African-American, and 1% were other. Nine per cent reported 12 or fewer years of education, 54% reported 13–16 years, and 36% reported greater than 16 years of education. There were no differences between the HIV-positive and HIV-negative men by race, ethnicity or education. The mean age of the HIV-positive men was 42 years (range, 24–64 years) and the mean age of the HIV-negative men was 45 years (range, 26–73 years).
Of the 346 HIV-positive and 262 HIV-negative men originally enrolled in the study, 277 (80%) HIV-positive and 221 (84%) HIV-negative men were seen after their baseline visit. Of the 69 HIV-positive men not seen after baseline, 17 were diagnosed with HSIL at baseline, 17 were too ill to participate in the study after baseline, 14 moved out of the area or to an unknown location, and 21 declined to participate further in the study for other reasons. Of the 41 HIV-negative men not seen after baseline, one was diagnosed with HSIL at baseline, one was too ill to continue in the study, 10 moved and 29 declined to participate further.
Table 1 summarizes the cumulative incidence of HSIL among the remaining HIV-positive and HIV-negative subjects. One hundred and six of 277 (38%) HIV-positive subjects and 34 of 221 (15%) HIV-negative subjects developed HSIL during the study. Forty-seven of 90 (52%) HIV-positive subjects with LSIL progressed to HSIL during the follow-up period, as did seven of 17 (41%) HIV-negative subjects.
Life-table analysis was used to compare the time course of HSIL development in HIV-positive and HIV-negative men. The 4-year incidence among all HIV-positive men was 49% [95% confidence interval (CI), 41–56]. In contrast, the 4-year incidence of HSIL among HIV-negative men was 17% (95% CI, 12–23). Figure 1 presents this analysis for HIV-positive men stratified by baseline CD4 count and for HIV-negative men. In HIV-positive men, lower baseline CD4 counts were associated with earlier development of HSIL (P = 0.007). HIV-positive men who had no anal lesions at baseline had a lower 4-year incidence of HSIL (38%) than did HIV-positive men who had an indeterminate diagnosis, ASCUS or LSIL at baseline (57%; P = 0.001). HIV-negative men who had no anal lesions at baseline had a lower 4-year incidence of HSIL (14%) than did HIV-negative men who had an indeterminate diagnosis, ASCUS or LSIL at baseline (33%) (P = 0.001).
Table 2 describes the relationship between incident HSIL, HIV status and CD4 level using person-years of observation in analysis. HIV-positive men were more likely to develop HSIL than were HIV-negative men. Among HIV-positive men, the risk for incident HSIL increased with lower baseline CD4 levels.
The role of HPV infection in incident HSIL was examined. HPV testing by PCR was available for 178 HIV-positive and 174 HIV-negative men, and testing by HC was available for 204 HIV-positive and 201 HIV-negative men. Table 3 presents relative risks and condfidence intervals for developing HSIL when HPV-positive using PCR and for the number of HPV types detected by PCR over time. Among both HIV-positive and HIV-negative men, consistent detection of HPV infection was associated with HSIL as was intermittent detection of HPV. A higher number of HPV types detected, whether persistent or intermittent, was associated with HSIL development in both HIV-positive and HIV-negative men. The role of individual HPV types could not be determined because of the small proportion of subjects with single HPV types. Similarly, using HC, the relative importance of A and B types in the development of HSIL was difficult to determine since 81% of the HIV-positive men were positive for both A types and B types. However, higher levels of HPV Group B RLU ratios over time were more strongly associated with incident HSIL among both HIV-positive and HIV-negative men than were higher Group A RLU ratios, suggesting a more important role for the former.
Figure 2 shows the life-table estimates of the cumulative proportions for incident HSIL among HIV-positive men stratified by presence or absence of persistent HPV infection (i.e., detection of a specific HPV type at each 12-month visit) with one or more HPV types. The 4-year incidence of HSIL among HIV-positive men with persistent HPV infection was 63% (95% CI, 51–74), whereas the incidence of HSIL among men with no persistent infection was 23% (95% CI, 13–34; P = 0.0001).
The relative importance of HIV status, baseline CD4 count, and HPV infection as risk factors for HSIL were evaluated using two different measurements of HPV; testing positive for HPV using PCR and presence of Group A and or B types using HC. For each comparison the reference group was HIV-negative subjects who were negative for the HPV measure used in the analysis. The highest relative risk was found for incident HSIL among HIV-positive men with CD4 counts less than 200 cells × 106/l who were positive for one or more types of HPV using PCR (Table 4). The highest relative risk was found for incident HSIL among HIV-positive men with CD4 counts less than 200 cells × 106/l who were positive for both Groups A and B using HC (Table 5). HIV-negative men with detection of both A and B groups were at a substantial risk of developing HSIL, similar to the risk of HIV-positive men with CD4 counts greater than 500 × 106cells/l. Some of the HIV-negative men developed HSIL without ever testing positive for HPV either by PCR or HC; three of 55 (5%) with no HPV detected using PCR developed HSIL, as did two of 63 (3%) with no HPV detected by HC (Tables 4 and 5).
Data from follow-up questionnaires on 204 HIV-positive men was used to evaluate the relationship between incident HSIL and possible behavioural and medical risk factors. New episodes of haemorrhoids since baseline (RR, 0.54; 95% CI, 0.34–0.80) and use of marijuana (RR, 0.64; 95% CI, 0.43–0.98) were associated with a reduced risk for HSIL. History of haemorrhoids at baseline only but not during follow-up was not associated with reduced risk of incident HSIL (RR, 0.97; 95% CI, 0.55–1.6). Other factors were examined and found to have RR close to unity. These included anal conditions (such as anal fissures, fistulas, anal infection or anal discharge, rectal itching or burning, use of laxatives, and constipation), history of sexually transmitted diseases other than warts (gonorrhea, chlamydia, syphilis and herpes), sexual practices (receptive anal intercourse, rimming, fisting), alcohol use, tobacco use, recreational drugs, recent sun exposure, radiography of the bowel, other AIDS-related symptoms, diseases and medications. A longer follow-up period is needed to assess the role of these factors in the development of HSIL in HIV-negative men who are still being followed.
These data demonstrate that the 4-year incidence of anal HSIL is high among both HIV-positive men and HIV-negative homosexual men. Thirty-eight per cent of the HIV-positive men and 17% of the HIV-negative men developed HSIL during the observation period of the study. Had it been possible to follow all HIV-positive men for the full 4 years, we estimate that 49% would have developed HSIL, based on our life table analysis. These results should be interpreted with caution since they were performed in highly sexually active urban populations with a mean age of over 40 years, and it is not known if they are applicable to the populations of homosexual men with different levels of sexual activity or at different ends of the age spectrum.
A key feature of this study was the routine use of anoscopy and biopsy of visible disease. While the diagnosis of HSIL was based on cytology or histology, 85% of the cases were histologically confirmed and the remainder were based on cytology. In a separate study of men participating in this cohort study, when HSIL was found on cytology and a biopsy was obtained from a visible lesion, 70% of these men also had HSIL on histology . The biopsy may have missed the highest grade of disease in the lesion resulting in an understimate of its true grade. Conversely, we may have overestimated the incidence of HSIL if some of the cytologies were false positive for HSIL. However, the potential impact of cytologies that were false positive for HSIL would be minimal since most were confirmed histologically, and the proportion of cases of HSIL diagnosed by cytology alone was similar for the HIV positive and HIV-negative men.
In this study, the incidence of HSIL among HIV-negative men with no anal lesions at baseline may have been slightly underestimated relative to the HIV-positive men since HIV-negative men with no lesions were followed at longer time intervals than the HIV-positive men. In some of these men, development and regression of HSIL during this longer time interval may have been missed. However, once detected, HSIL rarely regresses among either HIV-positive or HIV-negative men (unpublished data) minimizing the likelihood that an HSIL case was missed in either group. In addition, differences in incident HSIL among those with ASCUS or LSIL detected at any time would not have been underestimated among HIV-negative men relative to HIV-positive men since all men with anal lesions were followed at the same intervals regardless of HIV status. Thus it is unlikely that HSIL would have been missed in this study.
Analyses of risk factors for progression of anal lesions were conducted separately for HIV-positive and HIV-negative men in the cohort because of possible biological differences between the two groups. In univariate analyses among HIV-positive men, the most important risk factors for incident HSIL included lower CD4 levels, persistent anal HPV infection, anal infection with multiple HPV types, and high-level HPV infection with the more oncogenic HPV types as determined using HC. Among HIV-negative men, the same risk factors for incident HSIL were important, with the exception of lower CD4 counts.
The role of HPV infection in incident HSIL is consistent with the association between HPV infection and anal cancer [4,5]. The observation of increased risk for incident HSIL associated with multiple HPV types suggests the possibility of cooperation between specific HPV types in disease pathogenesis. Alternatively, the higher number of types may reflect more advanced immunosuppression with higher levels of replication of HPV DNA facilitating detection of types whose level was previously below the limits of detection by PCR. The absence of HPV detection in a small number of cases of incident HSIL suggests that HPV types other than those detectable using current PCR methodology may play a role, or that HPV may not be involved in those cases.
Analyses of risk factors for incident HSIL in HIV-positive men revealed reduced relative risks for haemorrhoids and use of marijuana. Although the mechanisms of protection by these factors are not known, haemorrhoids are usually accompanied by a local inflammatory immune response that may modulate the course of HPV infection. The effect of haemorrhoids on incident HSIL was not mediated through reduced practice of receptive anal intercourse, since a higher frequency of receptive anal intercourse was not associated with incident HSIL in this study. Marijuana use has been shown in another study to be protective against non-Hodgkin's lymphoma among HIV-positive men, possible through immune modulation . In addition, in our study, marijuana was used primarily as a recreational drug and was used more commonly among men with higher CD4 levels. Use of marijuana may have reflected a less advanced state of immunosuppression in this group rather than a direct effect on mitigating anal lesion progression.
Only one previous study has addressed incident HSIL . In that study, men were followed for an average of 21 months and 15% of HIV-positive men and 5% of HIV-negative men with no baseline lesions developed HSIL . In our study, 32% of HIV-positive men with no baseline lesions developed HSIL within 4 years and an even higher proportion of those with low-grade anal cytologic or histologic lesions progressed to HSIL. Fourteen per cent of HIV-negative men with no baseline lesion developed HSIL. The higher incidence of HSIL in our study among both HIV-positive and HIV-negative men may reflect the longer follow-up period. Among the HIV-positive men, it may also reflect a higher proportion of men with lower CD4 levels at baseline.
No studies have followed patients with untreated HSIL to determine the rate at which HSIL progresses to invasive cancer. However, the circumstantial evidence for HSIL as a precursor lesion to anal cancer is strong, and individuals with anal HSIL may be at risk for developing anal cancer. Several years may be required for HSIL to progress to invasive cancer, and the risk of progression of HSIL to cancer among HIV-positive men may have been mitigated in the past by mortality from other causes such as opportunistic infections. Protease inhibitors used for HIV infection are likely to prolong the survival of HIV-positive men . Moreover, a preliminary analysis of 50 subjects with HSIL adminstered protease inhibitor therapy for at least 3 months has demonstrated that most (38 subjects, 76%) showed no regression of HSIL, while 11 (22%) regressed to LSIL and one (2%) regressed to ASCUS (unpublished data). Together these observations suggest that progression from HSIL to cancer may increase. The incidence of HSIL was high in HIV-negative men, although lower than in HIV-positive men. These HIV-negative men with HSIL may have more time for progression to cancer given the likelihood that they will have normal lifespans.
If HSIL is an anal cancer precursor, then screening protocols for anal HSIL may be of value in anal cancer prevention, in a similar manner to cervical cancer prevention protocols. As in the cervix, cytology is a useful tool to identify lesions in the anal canal . Similarly, anoscopic evaluation of anal lesions using magnification and acetic acid may be used to identify HSIL lesions for biopsy and treatment . Further data are needed to define optimal screening algorithms and treatment regimens for HSIL before routine screening should be implemented in high-risk populations.
1. Daling JR, Weiss NS, Hislop TG, et al.
: Sexual practices, sexually transmitted diseases, and the incidence of anal cancer
. N Engl J Med
2. Qualters JR, Lee NC, Smith RA, Aubert RE: Breast and cervical cancer surveillance, United States, 1973–1987
3. Melbye M, Sprogel P: Aetiological parallel between anal cancer and cervical cancer
4. Palefsky JM, Holly EA, Gonzales J, Berline J, Ahn DK, Greenspan JS: Detection of human papillomavirus DNA in anal intraepithelial neoplasia and anal cancer
. Cancer Res
5. Zaki SR, Judd R, Coffield LM, Greer P, Rolston F, Evatt BL: Human papillomavirus infection and anal carcinoma. Retrospective analysis by in situ hybridization and the polymerase chain reaction
. Am J Pathol
6. Melbye M, Palefsky J, Gonzales J, et al.
: Immune status as a determinant of human papillomavirus detection and its association with anal epithelial abnormalities
. Int J Cancer
7. Caussy D, Goedert JJ, Palefsky J, et al.
: Interaction of human immunodeficiency and papilloma viruses: association with anal epithelial abnormality in homosexual men
. Int J Cancer
8. Critchlow CW, Holmes KK, Wood R, et al.
: Association of human immunodeficiency virus and anal human papillomavirus infection among homosexual men
. Arch Intern Med
9. Kiviat NB, Critchlow CW, Holmes KK, et al.
: Association of anal dysplasia and human papillomavirus with immunosuppression and HIV infection among homosexual men
10. Palefsky JM, Shiboski S, Moss A: Risk factors for anal human papillomavirus infection and anal cytologic abnormalities in HIV-positive and HIV-negative homosexual men
. J Acquir Immune Defic Syndr
11. Breese PL, Judson FN, Penley KA, Douglas Jr J: Anal human papillomavirus infection among homosexual and bisexual men: prevalence of type-specific infection and association with human immunodeficiency virus
. Sex Transm Dis
12. Rabkin CS, Yellin F: Cancer incidence in a population with a high prevalence of infection with human immunodeficiency virus type 1
. J Natl Cancer Inst
13. Melbye M, Cote TR, Kessler L, Gail M, Biggar RJ: High incidence of anal cancer among AIDS patients. The AIDS/Cancer Working Group
14. Winkelstein Jr W, Lyman DM, Padian N, et al.
: Sexual practices and risk of infection by the human immunodeficiency virus. The San Francisco Men's Health Study
15. Moss AR, Bacchetti P, Osmond D, et al.
: Seropositivity for HIV and the development of AIDS or AIDS related condition: three year follow up of the San Francisco General Hospital cohort
. Br Med J
16. Palefsky JM, Gonzales J, Greenblatt RM, Ahn DK, Hollander H: Anal intraepithelial neoplasia and anal papillomavirus infection among homosexual males with group IV HIV disease
17. Jay N, Holly EA, Berry M, Hogeboom CJ, Darragh TM, Palefsky JM: Colposcopic correlates of anal squamous intraepithelial lesions
. Dis Col Rectum
18. Kurman RJ, Solomon D: The Bethesda System for Reporting Cervical/vaginal Cytologic Diagnoses: Definitions, Criteria and Explanatory Notes for Terminology and Specimen Adequacy
. New York: Springer-Verlag; 1994.
19. Palefsky JM, Holly EA, Ralston ML, Jay N: Prevalence and risk factors for human papillomavirus infection of the anal canal in HIV-positive and HIV-negative homosexual men
. J Infect Dis
1998 (in press).
20. Collett, D: Modelling Survival Data in Medical Research
, London: Chapman and Hall; 1994.
21. Breslow NE, Day NE: Statistical Methods in Cancer Research
. Vol 2. The Design and Analysis of Cohort Studies
. International Agency for Research on Cancer Scientific Publications No. 82; 1987.
22. SAS Institute Inc.: SAS/STAT User's Guide Version 6, 4th Edition, Vol 1 and 2
. Cary: SAS; 1989.
23. Palefsky JM, Holly EA, Hogeboom CJ, Jay N, Berry M, Darragh TM: Anal cytology as a screening tool for anal squamous intraepithelial lesions
. J Acquir Immune Defic Syndr
24. Holly EA, Lele C, Bracci PM, McGrath MS: Non-Hodgkin's lymphoma in HIV-positive and HIV-negative homosexual men in the San Francisco Bay area. Allergies, prior medication use, sexual practices
. J Acquir Immune Defic Syndr
25. Critchlow CW, Surawicz CM, Holmes KK, et al.
: Prospective study of high grade anal squamous intraepithelial neoplasia in a cohort of homosexual men: influence of HIV infection, immunosuppression and human papillomavirus infection
26. MacDougall DS: Ritonavir: first to prolong survival
. J Int Assoc Phys AIDS Care
© Lippincott-Raven Publishers.