The Impact of Male Circumcision on HIV Transmission

Anatomical Basis for Protection 

In the flaccid state the inner foreskin (inner prepuce) of an uncircumcised penis is exposed to the surface of the glans and shaft of the penis, creating a moist, protected microenvironment for microbial flora. If not routinely cleaned the inner prepuce may harbor a reservoir for the proliferation of pathogens.6 Unfortunately clinical trials have failed to identify a microbicide effective against the acquisition of HIV.7 Additionally, inner foreskin mucosa contains little or no keratin compared to the highly keratinized outer foreskin.8 In vitro culture studies demonstrate that the inner surface of the human foreskin is highly susceptible to HIV-1 infection, significantly more so than the cervix and the highly resistant heavily keratinized outer foreskin.9

HIV-1 infection depends on several events. The primary HIV-1 target cells include Langerhans cells, CD4+ T cells and macrophages. Each of these cells expresses chemokine receptors (CCR5 or CXCR4), likely facilitating HIV-1 entry in human genital tissue.9 The basic model for heterosexual transmission of HIV comes from the intravaginal inoculation of rhesus macaques with simian immunodeficiency virus.10, 11 After inoculation the Langerhans cells located in the vaginal mucosal epithelium are selectively targeted by the virus and infected. They subsequently fuse with subepithelial CD4+ lymphocytes and migrate through the lymphatic chain leading to systemic infection. A similar sequence of events is thought to occur after penile inoculation of male macaques.12

In histopathological studies Langerhans cells are present in significantly greater quantities in the epithelium of human foreskin compared to the other HIV-1 target cells and appear to be the main portal of entry into the penis.9, 13 In addition to disrupting the mucosal continuity ulcerative or inflammatory lesions have been shown to increase the number of Langerhans cells in the foreskin, potentially facilitating entry of HIV-1.9, 14

Sexually Transmitted Infections, Genital Ulcer Disease and Male Circumcision 

It has been widely recognized that ulcerative and nonulcerative STIs have been strongly linked biologically and epidemiologically to HIV infection.15, 16 An association between male circumcision and a reduced risk of genital ulcer disease was noted in the Jewish population in London as early as 1855.17 In 1998 Moses et al reviewed the literature on circumcision status as well as the risk of chancroid, syphilis, genital herpes, urethritis and genital warts.18 The majority of epidemiological studies reviewed showed a positive association between lack of circumcision, and presence of genital ulcer disease and gonococcal urethritis. There were no data suggesting an increased risk of genital ulcer disease or gonococcal urethritis after circumcision. In 1999 Van Howe reviewed 31 studies regarding circumcision status, and the risk of urethritis and genital ulcer disease.19 Because the studies lacked sufficient control groups he concluded that while uncircumcised men are more likely to present with ulcerative disease and circumcised men are more likely to present with urethritis, no inference regarding disease prevention with circumcision could be made. Weiss et al subsequently performed a systematic review and meta-analysis in 2006, and reported that while there was significant study heterogeneity, there was strong evidence of the reduction of syphilis and chancroid with MC.20 However, the association with herpes simplex virus type 2 was less robust.

In the Ugandan RCT the rate of self-reported episodes of genital ulcer disease after randomization was 3.1% in the intervention (circumcised) group and 5.8% in the control (uncircumcised) group (p <0.0001).21 Serum analysis of participants in parallel RCTs in Uganda 24 months after study enrollment demonstrated a 2.5% reduction in herpes simplex virus type 2 seroconversion in the circumcised group (7.8% intervention vs 10.3% control, p = 0.008) and 9.9% reduction in high risk human papillomavirus genotypes in the circumcised group (18% intervention vs 29% control, p = 0.009). No significant difference was seen in the incidence of syphilis.22 In followup data from the South African RCT uncircumcised males were at higher risk for having high risk human papillomavirus genotypes in their urethral swabs (22.3% control vs 14.8% intervention, p = 0.002).23 In an intent to treat model polymerase chain reaction analysis of urine samples from selected participants in this trial did not demonstrate a significant reduction in Neisseria gonorrhoeae, Chlamydia trachomatis or Trichomonas vaginalis (although a 1.4% reduction in T. vaginalis infection became significant on an as treated analysis, p = 0.03).

Regardless of circumcision status, community level randomized trials in sub-Saharan Africa on STI control and prevention have produced mixed results on HIV incidence. Targeted, improved management of STI cases in the Mwanza region, Tanzania yielded a 38% reduction in HIV infection.24 However, periodic mass treatment programs of STIs in Rakai, Uganda did not reduce HIV-1 incidence despite similar STI reduction levels.25 A Cochrane systematic review of population based interventions in 2004 demonstrated that currently there is limited evidence supporting STI control as an effective HIV prevention strategy.26

Observational Data on Male Circumcision and HIV Prevention 

To further study the association between male circumcision and HIV incidence Moses et al described the geographic distribution of MC practices in Africa and their association with the geography of HIV seroprevalence.27 Of 140 geographically distinct locations surveyed 68 had a general adult HIV seroprevalence less than 1% and 17 had a seroprevalence greater than 10%. Of these selected populations 97% with a seroprevalence less than 1% practiced circumcision. Only 6% of those areas with a seroprevalence greater than 10% practiced circumcision. Since then more than 40 observational studies have been published on circumcision status and HIV risk in heterosexual men, many with differing conclusions. In addition, multiple reviews and meta-analyses have been performed. The first meta-analysis by Weiss et al included 27 studies and found an odds ratio of 0.42 (95% CI 0.60–0.68), demonstrating that there was compelling evidence for the use of MC as an HIV prevention tool.28 A subsequent Cochrane review of 37 observational studies found a large degree of heterogeneity among studies and concluded that there was insufficient evidence to support circumcision as an intervention for the prevention of HIV in heterosexual men.29 The authors acknowledged the need for the results from the 3 randomized, controlled trials that were under way.

Randomized, Controlled Trials 

Beginning in 2002, 3 separate randomized controlled trials on delayed vs immediate circumcision were initiated in South Africa, Kenya and Uganda.21, 30, 31 The primary end point of all 3 studies was HIV incidence and a summary of each is shown in the table. A total of 5,497 men older than 15 years were randomized in the 3 trials to the control (delayed circumcision) arms and 5,411 males to the intervention (immediate circumcision) arms. HIV positive patients were not randomized in the Kenyan or Ugandan trials, and they were randomized but not included in the final analysis in the South African trial. After approximately 2 years of followup all 3 trials were stopped early by the data safety monitoring boards as there was overwhelming evidence that circumcision provided significant protection from HIV acquisition. The control arms had 140 HIV seroconversions while the intervention arms had 61 seroconversions. Circumcision provided a relative risk reduction of acquiring HIV of 54% at approximately 2 years. Interestingly these findings were almost identical to the meta-analysis of observational studies found by Weiss et al.28

Summary of randomized, controlled trials on adult male circumcision

		South Africa	Kenya	Uganda
	Technique	Forceps guided	Forceps guided	Sleeve
	Operative setting	Local general practicioner office	Study clinic	Hospital
	Cautery	No	No	Yes (bipolar)
	No.	3,128	2,784	4,996
	No. intervention/control	1,546/1,582	1,391/1,393	2,474/2,522
	Age range	18–24	18–24	15–49
	Median age	21	20	Not available
	% Married	1.80	7	47
	Randomized if HIV pos	Yes⁎	No	No
	Planned mos followup	21	24	24
	% Completion of followup	63	87	73
	% Adverse events (immediate)	3.8	1.7	7.6
	% Total seroconversions in followup	2.2	3.1	1.3
	No. seroconversions (intervention/control)	20/49	22/47	20/47
	Adjusted incidence rate ratio (95% CI)	0.39(0.23–0.66)	0.41(0.28–0.78)	0.49(0.29–0.81)

⁎ Not included in final analysis.

Weiss et al subsequently performed a random effects meta-analysis on the results of the 3 trials.32 There was no evidence of heterogeneity among trials and the cumulative protective effect from all 3 studies was 58%. Again, this finding is identical to that found in the meta-analysis on observational studies published before the initiation of the 3 RCTs.33 A third Cochrane review including the results of the 3 RCTs reversed the previous review's conclusion, stating that there is now strong evidence that at 24 months adult male circumcision reduces HIV acquisition between 38% and 66%.34

It has been suggested that early termination of the studies resulted in an overestimate of effect.33 Recent data presented on long-term followup from the Kenyan study refute this claim.35 The relative risk of HIV infection in circumcised men at 42-month followup was 0.36 (0.23–0.57), yielding a 64% (43–77) protective effect.

Sexual Function After Circumcision 

Whether sexual function is affected by circumcision is a subject of considerable debate among advocates and opponents of circumcision. The only prospective, randomized, controlled data on long-term sexual function and satisfaction after circumcision come from the 3 RCTs. In the Ugandan trial variables including sexual desire, satisfaction and erectile function were collected at baseline from the 2,210 men in the intervention arm and 2,246 men in the control arm, and then assessed at 6, 12 and 24-month followup visits. Compared to the control group the circumcised group reported slightly more difficulty with penetration (1.6% circumcised vs 0.6% control, p = 0.02) and slightly less pain with intercourse (0.6% circumcised vs 1.2% control, p = 0.05) at 6 months. There were no statistically significant differences in any domain at 12 and 24 months.36 Circumcised men in the Kenyan trial demonstrated no difference in sexual dysfunction compared to controls. A majority of circumcised men did report more penile sensitivity and ease of achieving orgasm after their circumcision.37

Adverse Events After Adult Male Circumcision 

Surgical AEs from MC can include pain, bleeding, wound or systemic bacterial infection, urinary fistula, inferior cosmesis, partial to complete penile loss and death. In the Eastern Cape of South Africa ritual circumcision performed typically by a traditional healer is still widely practiced, especially in the Xhosa population. A review of 10,609 circumcisions performed through initiation schools in June 2005 revealed 288 hospital admissions, 9 amputations or mutilations and 23 deaths directly related to the circumcision ritual.38 Circumcisions performed in a clinical setting by experienced providers are thought to be associated with limited numbers of AEs. However, Bailey et al performed retrospective interviews, and prospectively monitored patients undergoing circumcision in clinical and traditional settings in the Bungoma district, Kenya.39 Their data suggest a much higher rate of complications in both settings compared to the clinical trials (35% in traditional setting, 18% in medical setting). The Orange Farm, South Africa trial reported a 3.6% rate of AEs within the first month after circumcision, the most common being pain (0.8%), swelling (0.6%) and bleeding (0.6%).30 By the end of followup (21 months) 1% continued to report issues with urination, appearance and erectile function (the preoperative rates were not reported). The rate of moderate to severe surgical related AEs from the Uganda trial was 3.5%.21 Krieger et al updated the AE outcomes from the Kenyan trial in 2007.40 A total of 1,475 procedures were performed with an AE rate of 1.8%. The most common AEs were wound disruption or infection (1%) and bleeding (0.3%). The number of AEs per surgeon decreased significantly as surgeons gained more experience.

Perhaps evidenced by the variability in reported AE rates among the clinical trials, the reporting and classification of each event are extremely important and have yet to be completely standardized. The reduction and monitoring of AEs remain a critical element in wide scale implementation.

Acceptability, Potential Epidemic Impact, Economic and Resource Issues 

Despite strong physiological and clinical evidence supporting circumcision, the magnitude of public health impact depends on further considerations regarding acceptance of the procedure, epidemic dynamics and costs. Even before the RCT evidence, which generated excitement about the benefit of MC, acceptability was found to be high. In a review of 13 studies from 9 countries in sub-Saharan Africa where circumcision is not traditionally practiced, a median 65% of uncircumcised men would be willing to undergo circumcision if it could be performed safely, at no or minimal cost, and shown to be an effective HIV prevention strategy; 71% of men and 81% of women would be willing to circumcise their son.41 Forecasting actual demand for circumcision based on acceptability is challenging. A variety of factors can prevent actual seeking of services by those who consider MC acceptable, including proximity of services, costs (of the procedure itself, transport and lost wages), possible stigma, and the reputation that the clinical service has for quality and friendliness. Research on demand forecasting for MC is in its infancy, and data being gathered by scale-up programs will ultimately be used to reduce barriers and generate demand.

The potential benefits of large scale implementation of MC are huge. Multiple epidemic models have been developed that demonstrate significant reductions in HIV incidence in sub-Saharan Africa with MC scale-up. Williams et al estimated that providing circumcision services to all uncircumcised adult males in sub-Saharan Africa (100% coverage) would prevent 2 to 8 million HIV infections in the next 10 years.42 Other mathematical models assuming MC coverage of less than 100% confirm the findings that substantial reductions in HIV incidence are possible among men and women in regions with a high prevalence.43, 44

While the economics of MC are also favorable, the resources required for implementation are considerable. MC could be made significantly more feasible through efficient service delivery. The cost and cost-effectiveness of adult male circumcision as an intervention depend mainly on the nature of the delivery service model and local HIV incidence. The Orange Farm, South Africa trial delivered circumcision services in a general practitioner's office with a per service cost of $47. Using a model based on the prevalence of HIV in men in that province (25.6%), for every 1,000 adult MCs performed 308 infections would be averted during 20 years with a net savings to the health care system of $2.4 million. The cost per infection averted in this model was estimated to be $181.45 Other studies have demonstrated comparable results, when adjusted for epidemic setting, local costs and time frame. Male circumcision is a highly cost-effective strategy compared to all other HIV prevention interventions in the developing world.44, 46

A recent study examined MC scale-up in the 16 African settings (15 countries and 1 province) with a high HIV prevalence (average 15.6%) and low MC prevalence (average 36%). This study assumed 10 MCs per surgeon per day, as in the RCTs, and a cost of $37 per MC, incorporating modest government salary scales. In a 5-year push to offer circumcision to all men 18 to 49 years old, 2,300 full-time practitioners would be needed, decreasing to 500 subsequently to keep up with new cohorts of young men. The estimated 5-year cost of rolling out adult MC in the public sector was $919 million, with a cumulative net cost during the first 10 years of $672 million and a cumulative net savings in 20 years of $2.3 billion.47 Thus, the study highlighted 3 points. 1) Scale-up will take substantial up-front financial and human resources. 2) There will be substantial net financial payback but this may take more than 10 years to realize. 3) The potential to reduce critical human resource needs (health care workers qualified to perform surgery are in shorter supply than funding) is great if surgical productivity can be increased while maintaining low AEs. The exact method used to increase productivity, whether through the use of a novel device or operative task specialization with shifting of services to lower level providers, remains controversial and an area of intense research.

Sexual Practices After Circumcision 

As information on circumcision and HIV prevention disseminates through at-risk communities, serious concern about risk compensation (a change in behavior with a perceived change in risk) exists. In the South African trial men randomized to circumcision showed a trend toward high risk behaviors and were significantly more likely to have more sexual contacts than uncircumcised men during the 21-month followup (7.5 vs 6.4, p = 0.002). This finding is in contrast to the trials from Kenya and Uganda which showed no significant behavioral risk compensation (disinhibition). In fact, 12-month followup data from 2,008 participants in the Kenyan RCT demonstrate a significant decrease in sexual risk behavior in both groups compared to baseline (p = 0.05).

The 42-month followup from the Kenyan RCT recently presented suggests that the protective effect of MC is stable and perhaps strengthened in the long term despite widespread acknowledgment of the benefits.35 There is widespread agreement that MC services should be packaged with proper education and counseling to minimize any effect of risk compensation.

Conclusions 

With a protective effect of approximately 60%, adult male circumcision is a highly effective HIV prevention strategy. A majority of the international health care community, including the WHO and Joint United Nations Programme on HIV/AIDS, have adopted adult male circumcision as an effective tool for HIV prevention. Delivery of safe circumcision services where HIV prevalence is high and MC prevalence is low could save millions of lives and billions of dollars during the next 20 years.