Baltimore.—B. Frank, Robin Fox, and Ronald Brockmeyer, Johns Hopkins University School of Hygiene and Public Health; Richard D. Leavitt, University of Maryland Cancer Center.

Chicago.—John P Phair, Joan S. Chimiel, and David G. Ostrow, Howard Brown Memorial Clinic, Northwestern University Medical School.

Los Angeles.—Roger Detels, Barbara R. Visscher, John L. Fahay, Janis V Giorgi and Jan Dudley, University of California.

Pittsburgh.—Charles R. Rinaldo, Monto Ho, Lawrence A. Kingsley, David W. Lyter, Ronald O. Valdiserri, and Allen Winkelstein, University of Pittsburgh Graduate School of Public Health and School of Medicine.

National Institute of Health.—Richard A. Kaslow, Alfred J. Saah, Mark J. VanRaden, and Rachel E. Solomon, National Institutes of Allergy and Infectious Disease; Andrew A. Monjan and A. R. Patel, National Cancer Institute.

2507 homosexual men who were seronegative for human immunodeficiency virus (HIV) at enrollment were followed for six months to elucidate risk factors for seroconversion to HIV. 95 (3.8%) seroconverted. Of men who did not engage in receptive anal intercourse within six months before baseline and in the six-month follow-up period, only 0.5% (3/646) seroconverted to HIV By contrast, of men who engaged in receptive anal intercourse with two or more partners during each of these successive six-month intervals, 10.6% (58/548) seroconverted. No HIV seroconversions occurred in 220 homosexual men who did not practice receptive or insertive anal intercourse within twelve months before the follow-up visit. On multivariate analysis receptive anal intercourse was the only significant risk factor for seroconversion to HIV, the risk ratio increasing from 3-fold for one partner to 18-fold for five or more partners. Furthermore, data from the two successive six-month periods show that men who reduced or stopped the practice of receptive anal intercourse significantly lowered their risk of seroconversion to 3.2% and 1.8% respectively. Receptive anal intercourse accounted for nearly all new HIV infections among the homosexual men enrolled in this study, and the hazards of this practice need to be emphasized in community educational projects.

Epidemiological data on transmission of human immunodeficiency virus (HIV) within the major risk group, homosexual men, have been limited primarily to inferences drawn from cross-sectional studies. Several investigations have shown that receptive anal intercourse, receptive "fisting," and large numbers of male sexual partners are major risk factors for both the acquired immunodeficiency syndrome (AIDS) and seropositivity to HIV. [1-4] Other reports have suggested a low risk of infection due to oral-genital exposure.[5-7] However, direct estimates of the attributable risk for acquisition of HIV infection can be made only from longitudinal (cohort) studies. This paper focuses on risk factors for seroconversion to HIV among the 2507 initially seronegative homosexual men enrolled in the Multicenter AIDS Cohort Study (MACS).

The MACS is a collaborative cohort study of homosexual and bisexual men. Four institutions (the University of California, Los Angeles; Johns Hopkins University, Baltimore; Howard Brown Memorial Clinic and Northwestern University, Chicago; and the University of Pittsburgh,

Pittsburgh) are conducting investigations according to a standardized protocol that includes at serial visits an interviewer-administered questionnaire, physical examination, laboratory tests, and collection of specimens. Baseline measurements were obtained on 4955 men between April, 1984, and March, 1985. Follow-up examinations for the first six-month period were completed on 90% of the cohort (n=4452) by November, 1985. This report will deal with the 2507 men who were seronegative for HIV at entry into the study, completed their first follow-up visit, and were therefore at risk of seroconversion to HIV.

Sexual activity data reported here are based on respondents' answers to questions eliciting the number of partners with whom various sexual practices were performed six months before baseline and during the six-month follow-up. In these two time-frames the questions were identical, and elicited information on active masturbation of sexual partners, oral-genital intercourse to ejaculation, anilingus ("rimming"), digital-anal insertion, hand-anal insertion ("fisting"), dildo use, and enema/douche use. For oral intercourse, anal intercourse, "rimming," and "fisting," data were collected for both receptive and insertive numbers of partners. With dildo and enema/douche use, only recipients were considered.

Antibody to HIV (HTLV-III) was determined by two methods. Pairs of sera taken from each participant at entry and follow-up visits were tested by an enzyme-linked immunosorbent assay licensed by the FDA (DuPont HTLV-III ELISA, DuPont Company).[8] All specimens for which ELISA results suggested seroconversion (i.e., increase from <0.5 to >0.5) were examined by immunoblot techniques (Biotech Laboratories, Inc.). A score of 0 was assigned for a negative band, 1 for a weakly reactive band, 2 for a moderately reactive band, and 3 for a strongly reactive band to pl5, p24, p31, p45, p53, p64, or pl20. The scores of all bands were summed and a value of >3 was defined as positive, 2 as equivocal, and <1 as negative. The validity of this method has been evaluated by prospective assessment in this cohort. [9]

Statistical analysis used to support inferences where based on the chi-square test for 2x2 contingency tables (for medians or proportions), chi-square test for trend, 95% for binomial confidence limits, and multiple logistic regression.

Results Table I shows baseline demographic characteristics of the 2507 MACS participants who were seronegative for HIV at the initial examination and completed the six-month follow-up examination.

Men who seroconverted to HIV were slightly younger than those who remained seronegative. Despite similar durations of homosexual activity, the reported number of male sexual partners for a lifetime, two years, and six months before baseline examination were about 2-fold higher among those who subsequently seroconverted. The lifetime history of gonorrhoea (any site) and rectal gonorrhoea were also slightly higher in the seroconverter group, as was a six-month history of at least one episode of perianal bleeding noticed after sexual activity.

The overall six-month HIV seroconversion rate was 3.8% (95/2507). Six-month HIV seroconversion rates, by center, were: Pittsburgh, 2.5% (13/519); Baltimore-Washington, 3.5% (26/749); Chicago, 4.5% (25/554); and Los Angeles, 4.5% (31/685). The seroconversion rates did not differ significantly

Table II shows the gradient in seroconversion rate to HIV when MACS participants were stratified according to levels of receptive and insertive anal intercourse reported for the six months before documented seroconversion. HIV seroconversion was rare in those who did not participate in receptive anal intercourse in the previous six months (0.9%; 9/984), but increased steadily to 13.6% (30/221) among those who practiced receptive anal intercourse with at least five partners. Of the 9 seroconverters who had not practiced receptive anal intercourse during the six months before seroconversion, 6 had done so within the six months before enrolment. Thus, only 3 men reported no receptive anal intercourse in the year before documentation of their seroconversions. This cross-classification also shows that, when each level of insertive anal intercourse is controlled for, a strong trend (p< 0.001 by chi-square test for trend) is observed for increasing seroconversion rates with increasing number of partners with whom receptive anal intercourse was practiced. No such consistent trend is observed for insertive anal intercourse. Among the 513 men who reported just insertive anal intercourse during the six-month follow-up period, only 5 (1.0%) seroconverted to HIV compared with 7 of 228 (3.1%) for those who reported just receptive anal intercourse. 2 of the 5 seroconverters who practiced only insertive anal intercourse during the six-month follow-up did report receptive anal intercourse before the initial examination.

Table III stratifies the MACS participants by the reported number of partners with whom they engaged in receptive anal intercourse during the two six-month periods before recording seroconversion. Those who continued receptive anal intercourse with two or more partners showed the highest six-month seroconversion rate—10.6%. Significantly lower

seroconversion rates were observed in those who reduced (3.2%) or stopped (1.8%) receptive anal intercourse. These rates were similar to those in men who continued with only one partner (2.3%), "began" (1.7%), or increased the number of partners (5.4%) with whom receptive anal intercourse was reported during the subsequent six-month follow-up period.

Table IV details the HIV seroconversion rates for two mutually exclusive sexual practice sub-groups of the 646 men reporting no receptive anal intercourse within twelve months of measured seroconversion. Of the 147 men who engaged in oral receptive intercourse with at least 1 partner during the six-month follow-up but reported no receptive or insertive anal intercourse within twelve months, no seroconversions to HIV were observed. The only 3 men to seroconvert without reported receptive anal intercourse within twelve months were among those 344 men who reported anal intercourse as the insertive partner only—a seroconversion rate of 0.9%. Since no seroconversions to HIV were detected among those not reporting anal intercourse, the distribution of sexual practices among these 220 men is shown separately in Table V

Potential parenteral exposure was quite rare among those who sero-converted. Only 1 of 95 seroconverters (1%) gave a history in the six-month follow-up of shared needle use, and none of them gave a history of shared-needle use with an individual in whom AIDS had developed.

Multiple logistic regression was used to investigate the independence and strength of association between each exposure variable and seroconversion. Stepwise entry methods always indicated that receptive anal intercourse during the six-month follow-up was the most significant sexual-exposure variable. The adjusted odds ratio increased from 3.2 (1 receptive anal partner) to 9.5 (2-4 partners), to 18.0 (5 or more partners). When receptive anal intercourse was controlled for, only enema/douche use entered the model (odds ratio 1.5), despite liberal entry limits (p > 0.10). Thus, although not significant at the 5% level, there was a trend toward and an association between enema/douche use before sex with at least one partner and seroconversion. The variables used in the stepwise procedure were: insertive anal intercourse at 2 levels (1 and 2+ partners), receptive dildo use (1+ partners), enema/douche use before sex (1+ partners), insertive and receptive anilingus (1+ partners), insertive and receptive digital-anal contact (1+ partners), reported sexual contact with an AIDS case (1+ partner), episodes of perianal bleeding (1+ partners), age (< 32 or > 33), and the total number of reported sexual partners in the follow-up period at 4 levels (0, 1, 2-4, and 5+ partners).

The independence and strength of association between receptive anal intercourse and HIV seroconversion provides further support for the inferences drawn from Tables II-V.

Data obtained from initially seronegative MACS participants clearly demonstrates that receptive anal intercourse is the major mode for acquisition of HIV infection and that discontinuation of the practice sharply reduces the likelihood of seroconversion in the next six to twelve months. Receptive anal intercourse was the only sexual practice shown to be independently associated with an increased risk of seroconversion to HIV in this study, and could account for nearly all new infections. The gradient of risk for seroconversion accelerated in proportion to the number of receptive partners, from about 3-fold for one partner to about 18-fold for those with 5 or more partners during the observation period. Further, these risk calculations are conservative because they are based only on reported receptive anal intercourse in the six-month longitudinal follow-up period. In fact, 6 or the 9 seroconverters who denied receptive anal intercourse during the six-month follow-up did report having practiced this activity within the six months before the initial evaluation. Seroconversion in these 6 men may have resulted from exposure via receptive anal intercourse that had occurred before baseline.

Of the remaining 3 seroconverters who did not participate in receptive anal intercourse, all did participate in insertive anal intercourse, during both the pre-enrollment period and the six-month follow-up period. These data suggest a low (less than 1%) six-month risk for seroconversion due only to the practice of insertive anal intercourse. Alternately, this may reflect misclassification of men who actually did participate in receptive anal intercourse.

The most important finding of this study comes from a comparison of HIV seroconversion rates based on receptive anal intercourse before the initial assessment and during the six-month follow-up (Table III). Significantly lower seroconversion rates were noted in the men who reduced (3.2%) or stopped (1.8%) receptive anal intercourse than in men who continued the practice with at least 2 partners (10.6%). As previously noted, the fact that the seroconversion rate was not zero for those who stopped receptive anal intercourse very likely reflects exposure via receptive intercourse before study enrolment. The public health message from these findings is clear. Reduction of this high-risk practice by homosexual men dramatically reduces the risk of HIV infec-

tion. Furthermore, it is clear that the degree of modification of this high-risk behavior necessary to make a substantial impact can actually be achieved.

The absence of detectable risk for seroconversion due to receptive oral-genital intercourse is striking. That there were no seroconversions detected among 147 men engaging in receptive oral intercourse with at least 1 partner, but not receptive or insertive anal intercourse, accords with other data suggesting a low risk of infection from oral-genital (Receptive semen) exposure.[4-7] It must be mentioned that we were unable to determine the infection status of the sexual partners to whom these men were exposed. Perhaps these 147 men who practiced receptive oral intercourse were never or rarely exposed to HIV seropositive men. However, this explanation seems improbable. As noted, the 220 participants who did not engage in receptive or insertive anal intercourse within twelve months before the follow-up visit had ample opportunity to be exposed to HIV-infected men since 67% percent had engaged in receptive oral intercourse with at least one partner (median 2, range 1-60).

From a public health point of view, we can affirm to homosexual men that receptive anal intercourse is the principal route by which they may become infected with HIV. We must also communicate that a small but real risk from other exposure may have been undetectable, even in this large study. Receptive fisting, enema/douche use before sex, and perianal bleeding, as markers for rectal trauma, have all been strongly associated with prevalent HIV infection in the cross-sectional Multicenter AIDS Cohort Study of nearly 5000 men (38% HIV seropositive).[10] That none of these trauma indicators was significantly associated with seroconversion in the present study may indicate only that a smaller sample size (95 seroconverters) precluded their detection. Enema or douche use before sex did, however, show a trend toward association with seroconversion (odds ratio 1.5, p< 0.01). Although the prospective nature of this analysis makes it a more compelling assessment of risk factors, the potential importance of the traumatic practices in promoting HIV infection should not be overlooked.

The relative "safety" of the sexual practices not detected as risk factors for seroconversion in this report deserves comment. Oral intercourse with ejaculate introduced into the oral cavity, anilingus, "fisting," enema/douche use, and dildo use are all potentially unsafe. HIV infection apart, many of these practices have already been associated with other sexually transmitted diseases that present a public threat to male homosexuals. [11] These sexual practices should be considered in the

light of all the infections transmissible by these means—e.g., hepatitis B, cytomegalovirus, herpes simplex virus, amoebiasis, syphilis, and gonorrhoea. "Safe sex" guidelines should not apply only to prevention of HIV infection.

Among homosexual men avoidance of anal intercourse may be the only existing means of limiting future morbidity and mortality from HIV infection. A prudent course would be to stop anal intercourse entirely This study demonstrates a 3-fold greater risk of seroconversion for the MACS men who engaged in receptive anal intercourse with only 1 partner in six months. Even if exclusively monogamous men knew their serological status, the risk of exposure to HIV remains a concern, in view of the reports of virus-positive, antibody-negative homosexual men. [12,13] We cannot comment on whether the consistent and proper use of condoms (with or without spermicide) reduces or eliminates the risk of HIV infection. Although latex and natural condoms do not permit passage of HIV under laboratory conditions[14] and spermicidal nonoxynol-9 preparations are viricidal of HIV,[15] more data are needed to examine the efficacy of both condoms and spermicide in preventing infection.

Avoidance of anal intercourse must be the principal focus of efforts to reduce risk in the male homosexual community. [16] This education message must be given the highest public health priority.

We thank C. Perfetti, W. Amoroso, and S. Jones for assistance in data analysis and D. Laurie for preparation of the script. This study was supported by NIAID research contracts Al-32511, Al-32513, Al-32520, and Al-23535.

Correspondence should be addressed to L. A. Kingsley, Departments of Infectious Diseases and Microbiology-Epidemiology, Graduate School of Public Health, University of Pittsburgh, PO Box 7319, Pittsburgh, PA, 15213, USA.

Requests for reprints should be addressed to AIDS Program Office, National Institute of Allergy and Infectious Diseases, Westwood Bldg., Rm. 753, 5333 Westbard Avenue, Bethesda, MD 20892, USA.

[I] Jaffe HW, Choi K, Thomas PA, et al. National case-control study of Kaposi's sarcoma and Pneumocystis carinii pneumonia in homosexual men Part 1, epidemiologic results. Am Intern Med 1984; 99: 145-51.

[2] Goedert JJ, Samgadharan MG, Bigger RJ, et al. Determination of retrovirus (HTLV-III) antibody and immunodeficiency conditions in homosexual men. Lancet 1984: ii: 711-16.

[3] Marmor M, Friedman-Kien AE, Zolla-Pazner S, et al. Kaposi's sarcoma in homosexual men: a seroepidemiologic case-control study. Am Intern Med 1984; 108: 809-15.

[4] Melbye M, Bigger RJ, Ebbessen P, et al. Seroepidemiology of HTLV-III antibodies in Danish homosexual men: prevalence, transmission, and disease outcome. BrMedJ 1984; 289: 573-75.

[5] Jeffries E, Wiloughby B, Boyko WJ, et al. The Vancouver lym-phadenopathy-AIDS study-11 seroepidemiology of HTL V-III antibody. Can Med Assoc J 1985: 132: 1373-77.

[6] Sheuter MT, Boyko WJ, Douglas B, et al. Can HTLV-III be transmitted orally? Lancet 1986; i: 379.

[7] Lyman D, Winkelstein W, Ascher M, Levy J A. Minimal risk of AIDS transmission of AIDS-associated retrovirus infection by oral-genital contact. JAMA 1986: 255: 1703.

[8] Samgadharan MG, Popovic M, Bruch L, Schuphjach J, Gallo RC. Antibodies reactive with human T-lymphotropic retrovirus (HTLV-III) in the serum of patients with AIDS. Science 1984: 224: 506-08.

[9] Saah A, Farzadegan H, Fox R, et al. Sensitivity of the ELISA for human immunodeficiency virus antibodies during early stages of infection. Unpublished.

[10] Multicenter AIDS Cohort Study (MACS). Prevalence and correlates of HTL V-lll antibodies among 5000 gay men in 4 cities. Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC) October, 1985 (abstr).

[II] Ostrow DG, Alman NL, Sexually transmitted diseases and homosexuality. Sex Traum Dis 1983; 10: 208-15.

[12] Mayer K, Spader AM, McCukzer J, et al. Human T-lymphotropic virus type III in high-risk, antibody negative homosexual men. Am Intern Med 1986; 104: 194-96.

[13] Groopman JE, Hartzband PI, Schulman L., et al. Antibody seronegative human T lymphotropic virus type III (HTLV-III)-infected patients with acquired immunodeficiency syndrome or related disorders. Blood 1985; 66: 742-44.

[14] Conant M, Hardy D, Sernatinger J, et al. Condoms prevents transmission of AIDS-associated retrovirus. JAMA 1986; 255: 1706.

[15] Hicks RD, Martin SL, Getchell JP, et al. Inactivation of HTLV-III— LAV-infected cultures of normal human lymphocytes by nonoxynol-9 in vitro. Lancet 1985; 255: 1422-23.

[16] McKusick L, Conant M, Coates TJ. The AIDS epidemic: a model for developing intervention strategies for reducing high risk behavior in homosexual men. Sex Traum Dis 1983; 12: 229-34.

Seroconverters

n=95

n=2412

Median

(range)

Median (range)

Age at enrollment

(19-55)

33 (18-72)

Age at first sex with

another male

(5-31)

18 (2-25)

Age began regular sex

with males

(5-31)

21 (2-58)

Years of regular

homosexual activity

(1-37)

11 (1-52)

No. of male partners

during previous 6 mo.

10*

(1-500)

5 (1-356)

No. of male partners

during past 2 years

36*

(3-1000+)

20 (1-1000+)

No. of male partners

during lifetime

200*

(4-1000+)

100 (1-1000+)

% Reporting

Lifetime history of

any gonorrhea

64% t

50%

Lifetime history of

rectal gonorrhea

28% t

16%

History of perianal

bleeding (> 1 time)

during previous 6 mo.

31%t

18%_______________

For comparisons between Seroconverters and non-Seroconverters: *p< 0.001 by chi-square test for median difference; fp< 0.009 by chi-square test for proportional difference.

Insertive:

Row

95%

2-4

summary

confidence

Receptive

n %

limits

4/471

0.8

2/234

0.9

2/168 1.2

1/111

0.9

9/984 0.9

0.4-1.7

1/142

0.7

9/388

2.3

6/168 3.6

3/56

5.4

19/754 2.5

1..5-3.9

2-4

3/60

5.0

7/105

6.7

19/26 37.2

8/82

9.8

37/510 7.3

5.2-9.9

3/26

11.5

4/20

20.0

7/50 14.0

16/125

12.8

30/221 13.6

9.3-18.7

Column

summary

11/699

1.6

22/747

2.29

34/649 5.2

28/374

7.4

95/2469° 3.8

Partners with whom

receptive anal intercourse reported:

6 mo. before

6 mo.

Receptive anal

enrollment

before

Seroconversion

95%

intercourse status

visit 1

visit 2

rate (n)t

confidence limits

None

646

0.5% (3)

0.1-1.4%

Stopped anal

receptive

338

1.8% (6)

0.7-3.8%

Continued anal

receptive

346

2.3% (8)

1.0-4.5%

Reduced anal

receptive

283

3.2% (9)

1.5-6.0%

Began anal

receptive*

172

1.7% (3)

0.4-5.0%

Increased anal

receptive

130

5.4% (7)

2.2-10.8%

Continued anal

receptive

548

10.6% (58)

8.1-13.5%

^Refers only to responses for the two six-month periods; lifetime history of receptive anal intercourse was not obtained. nf =2463 because data were missing on 44 participents, 1 of them a seroconverter.

Seroconversion

95% confidence

Sexual practice*

rate (n)

limits

Oral receptive intercourse

> 1 partner (no receptive anal intercourse)

147

0.0% (0)

0.0-2.5%

Insertive anal intercourse

> 1 partner (no receptive anal intercourse)

344

0.9% (3)

0.2-2.5%

No. of partners

Sexual practice*

Median Range

Masturbation of partners (> 1)

150

68%

2 (1-60)

Oral intercourse

Receptive (> 1)

147

67%

2 (1-60) ;

Insertive (> 1)

160

73%

2 (1-50)

Anilingus ("rimming")

Receptive (> 1)

15%

1 d-5)

Insertive (> 1)

10%

1 (1-60)

Digital anal

Receptive (> 1)

12%

1 (1-5)

Insertive (> 1)

18%

1 (1-60)

Voyant Publishing's name derives from the term employed in Arthur Rimbaud's famous Lettre du Voyant, in which he described the self-inflicted tortures that await writers who wish to be among the poets of the age. Voyant Publishing was founded in the interest of literature that values the sentence in its own right rather than as a simple tool for moving a story along.

We'd like to encourage readers to step up to the plate and take a few swings at pitches from non-conventional writers — no matter how the ball might curve or snake beyond the bounds of readerly expectation.

Voyant Incarnations is the new imprint of Voyant Publishing devoted to restoring "lost classics" to print, like phoenixes risen from the ashes of the corporate publishing "industry."

The first two releases in this line are Paul West's The Place In Flowers Where Pollen Rests and Samuel R. Delany's The Mad Man. The next Voyant Incarnations book will be Walter Pater's On Plato and Platonism.

We're always looking for more of these out-of-print gems, so drop us a line with your suggestions.

"Adrift in a Vanishing City is a book that rewards multiple readings and demands to be quoted, as the multilayered construction of Czyz s prose enables Adrift to speak toward those depths of mind and memory that tend to elude language/'

"A wonderful complement to his autobiographical writings, these letters are as much literature as any of Delany s fiction. As ever, his intelligence, kindness, empathy, critical skills and intense interest in writing, art and the world around him shine through."

"[Tjhis video is a must-see not only for Delany fans but also for those unfamiliar with the man and his work, those who want to find out what it—and he—is all about."

"Paul West's epic touches upon the most powerful human themes —the meaning of home, the desecration of war, the quest for a curative path out of spiritual exile. A poignant masterpiece from the mind and heart of one of our indubitable masters."

Samuel R. Delany is a novelist and critic. His fiction includes Dhalgren (1975), the Return to Neveryon series (1979-1987) and Atlantis: Three Tales (1995). His nonfiction writing includes The Motion of Light in Water (1987), Village Voice bestseller Times Square Red, Times Square Blue (1999) and 1984: Selected Letters (2000).

Winner of the William Whitehead Memorial Award for a Life-times Contribution to Lesbian and Gay Literature, he currently teaches at Temple University in Philadelphia.