van de Wijgert, Janneke H H M PhD, MPH*†; Braunstein, Sarah L MPH*; Morar, Neetha S MMedSc‡; Jones, Heidi E MPH*; Madurai, Lorna PhD‡§; Evans Strickfaden, Tammy T MPH∥; Moodley, Manivasan MBChB, FCOG, MMeD¶; Aboobaker, Jamila MBChB, FFDerm¶; Ndlovu, Gugulethu MBA, BA(Cur)#; Ferguson, Taja M MPH*; Friedland, Barbara A BA*; Hart, Clyde E PhD∥; Ramjee, Gita PhD‡
Microbicides are being developed for topical application inside the vagina or rectum to prevent infection with HIV and possibly other sexually transmitted infections (STIs).1 Carraguard gel is the Population Council's lead noncontraceptive candidate microbicide. Carraguard consists of a mixture of λ-carrageenan and κ-carrageenan derived from red seaweed. Carrageenan is inexpensive and stable, with negligible mucosal absorption, and is widely used as a food thickener and emulsifier in topical preparations. Carraguard binds to cells, differentiating it from surfactants such as nonoxynol-9 (N9), which disrupt cell membranes, including those of host epithelial cells. Preclinical studies indicated that Carraguard and its matching placebo, methylcellulose gel, are nonirritative and nonteratogenic and have no effect on lactobacilli and spermatozoa, and that Carraguard is effective against HIV and other STI pathogens.2-6 In a series of 3 phase 1 and 2 phase 2 trials in South Africa, Thailand, and elsewhere, Carraguard was found to be safe and acceptable in HIV-negative women and men when used with or without latex condoms.7-10 A phase 3 trial to determine the effectiveness of Carraguard to prevent male-to-female HIV transmission was recently completed at 3 sites in South Africa.
If proven safe and effective, a microbicide may be introduced in communities with high HIV prevalence. Many users will be HIV positive and unaware of their status, or wish to protect their partners or to avoid reinfection with HIV or other STIs. We therefore assessed the safety of Carraguard in HIV-positive women and men in Durban, South Africa.
This randomized, placebo-controlled, triple-blinded clinical trial of Carraguard was a collaboration between the Population Council (New York and Johannesburg), the Medical Research Council (Durban), and the US Centers for Disease Control and Prevention (Atlanta). The objectives were to assess the safety of Carraguard when applied vaginally once per day for 14 intermenstrual days by sexually abstinent and sexually active HIV-positive women; and when applied directly to the penis once per day for 7 days by sexually abstinent HIV-positive men. The primary safety endpoints were genital mucosal safety in women and men, vaginal flora changes in women, HIV-1 genital shedding in women, and self-reported symptoms in women and men. Acceptability was also assessed, but results are not described here. Twenty participants were enrolled in each of the 3 cohorts. The sample size was not based on statistical considerations, and analyses were exploratory in nature.
Participants were recruited from sites in and around Durban, including King Edward VIII Hospital and support groups for people living with HIV/AIDS. Sexually abstinent women were eligible if they were confirmed HIV positive; aged 18 to 45; in good health (defined by CD4 cell count >200 cells/mL and absence of opportunistic infections); willing to maintain any current treatments while in the study; willing to abstain from sex, masturbation, and vaginal product and spermicide use for 48 hours before enrollment and while in the study; not pregnant, menopausal, lactating, or planning a pregnancy; having a regular or no menstrual cycle; and planning to remain in Durban for at least 6 months. Women were excluded if they had had a delivery, miscarriage, or abortion within the previous 6 weeks; genital surgery or nonmenstrual vaginal bleeding during sex within the previous 3 months; an abnormal Papanicolaou smear, STI, symptomatic bacterial vaginosis (BV) and/or yeast infection, or genital abnormality with epithelial disruption at baseline; used a spermicide within the previous week; or participated in another clinical trial within the past month. Eligibility criteria were identical for sexually active women, but they also had to agree to have sex at least twice per week with only 1 male partner, who was confirmed HIV positive, 18 years or older, and willing to have sex only with the participant while in the study; to use contraception (oral contraceptive, injectable, IUD, implant, sterilization, or condoms); and to have no known latex allergy. Sexually abstinent men were eligible if they were confirmed HIV positive; 18 years or older; in good health (defined as above); willing to abstain from sex and masturbation while in the study; and planning to remain in Durban for at least 6 months. They were excluded if they had an STI, genital abnormality with epithelial disruption, or contact dermatitis on the genitalia at baseline; used a spermicide within the previous week; or participated in another clinical trial within the past month. Participants could be screened for eligibility up to 3 times, with no more than 35 days between the final screening visit and enrollment.
Consenting women and men in each cohort were randomized at enrollment to Carraguard, placebo, or no product (1:1:1). The placebo looks, feels, smells, and tastes the same as Carraguard; both are clear gels. Both gels came in a single-use Microlax-type applicator (Norden Pac, Kalmar, Sweden), designed to dispense about 4 mL of gel. Sexually abstinent women were asked to insert 1 dose into the vagina every evening for 14 days. Sexually abstinent men were asked to apply 1 dose directly onto the penis every evening for 7 days. Once safety was established in the sexually abstinent cohorts by an interim data review, sexually active women were enrolled. They were asked to insert 1 dose vaginally no more than 1 hour before sex, each time they had sex during the 14-day follow up, and to insert gel on the evening of those days when sex did not take place. Non-N9-lubricated latex condoms were offered free of charge. Participants were given daily diaries to record sexual activity and symptoms, and were asked to bring the diaries and all used and unused applicators to the clinic at each visit. They were told that the gels were not to be used orally or rectally. The unit of randomization was the individual woman or man, and the randomization sequence was determined centrally and in advance by computer. Investigators and data managers/analysts were blinded throughout the trial and data analysis. Study participants and staff knew which participants were not receiving gel but were blinded to the 2 gel groups.
Female participants were asked to visit the study clinic at screening, enrollment, and days 7, 14, and 21 (7 days after cessation of gel use); visits were planned to avoid menses during gel use. Male participants were seen at screening, enrollment, and day 7. All participants were tested for HIV at baseline using the AxSYM HIV 1/2 gO (Abbott Diagnostics, Abbott Park, IL) and Vironostika HIV Uniform 11 plus O (BioMérieux, Marcy l'Etoile, France) assays. All participants underwent a CD4 count (TriTest CD4 FITC/CD8 PE/CD3 perCP; Becton Dickinson Immunocytometry Systems, San Jose, CA) and plasma viral load testing (Roche Amplicor HIV-1 Monitor v1.5; Roche Molecular Systems, Branchburg, NJ) at screening. For women, plasma viral load was also determined at day 14. Sexual activity, gel and condom use, symptoms, and acceptability were assessed by structured interviewing (aided by consulting daily diaries) and by counting used and unused applicators at each visit. Mucosal safety was evaluated by visual inspection of the genitalia by a dermatologist (men) or gynecologist (women). Women had a pelvic examination at each visit and colposcopy at the enrollment and day 14 visits. Genital findings were documented according to the WHO Manual for the Standardization of Colposcopy for the Evaluation of Vaginally Administered Products, Update 2000.11 To monitor changes in vaginal flora in women, Gram stains were scored according to the Nugent criteria, and wet mounts were evaluated for the presence of yeast.12 At screening, everyone was tested for Treponema pallidum (Rapid Plasma Reagin card; Randox, Antrim, UK; and Immutrep T. pallidum Haemagglutination Assay, Omega Diagnostics, Alva, Scotland), Neisseria gonorrhoeae and C. trachomatis (COBAS Amplicor PCR; Amplicor, Branchburg, NJ), and Trichomonas vaginalis (wet mount or urine microscopy); these tests were repeated during follow-up if clinically indicated. In addition, women were tested for T. vaginalis (wet mount) and pregnancy (QuickVue hCG test; Quidel, San Diego, CA; or Inverness Medical Testpack hCG Combo; Unipath, Galway, Ireland) at each visit, and they received a Papanicolaou (Pap) smear at screening.
Cervicovaginal lavage (CVL) specimens (3 mL of normal saline directed at the cervical os and 3.5 mL each at the left and right vaginal walls) were collected at each visit and were tested for HIV-1 RNA using the Roche Amplicor HIV-1 Monitor v1.5 test, hemoglobin and leukocytes using Bayer Reagent Strips for Urinalysis (Bayer Diagnostics, Bridgend, UK), and semen using the OneStep ABAcard p30 Test (Abacus Diagnostics, West Hills, CA). Male urine samples were also tested for leukocytes using the Bayer dipstick. CVL specimens were centrifuged, and epithelial cells stained and counted under a microscope. Hemoglobin and semen in CVL specimens were considered potential contaminants of CVL HIV-1 RNA levels. Leukocyte concentrations in CVL specimens and urine were considered markers of genital inflammation in women and men respectively, and the number of epithelial cells and presence of hemoglobin in CVL specimens markers were markers for genital epithelial sloughing and microhemorrhage in women.13
Data were double-entered into a Microsoft Access database. Adverse events (AEs) were coded by an independent physician using the WHO Adverse Reaction Dictionary adapted before this study to describe genital findings in greater detail.14 All primary analyses were by intent-to-treat, were exploratory in nature, and were not adjusted for multiple comparisons. Data were analyzed using SPSS version 10.0 (SPSS, Chicago, IL). Differences between enrolled and nonenrolled women/men and differences among women/men in the 3 product use groups were tested using the Mann-Whitney nonparametric test for continuous variables and the Fisher exact or χ2 tests for categorical variables. All tests were 2-sided at the 0.05 significance level. Log10 transformations were used in viral load and epithelial cell count analyses. CVL HIV-1 RNA levels were categorized as nondetectable (ND), detectable but not quantifiable (NQ; below 50 copies/mL), and quantifiable (Q). Hemoglobin contamination of CVLs was negligible and viral loads were therefore not adjusted; 5 CVL specimens tested positive for semen and were excluded. A significant change in CVL viral load was defined as a change in category (ND, NQ, or Q) and, within the quantifiable category, a ≥0.5 log10 change in viral load. A significant change in leukocytes and hemoglobin was defined as a change from 1 concentration category to another (13 categories in total) and, in epithelial cells, a ≥0.5 log10 change in cell count.
The protocol was approved by the ethical review committees of the Population Council, the South African Medical Research Council, the US Centers for Disease Control and Prevention, the US Food and Drug Administration, and the South African Medicines Control Council. Study procedures were in accordance with the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practices Guidelines.15,16
Between May 2002 and July 2003, 98 sexually abstinent women, 45 sexually abstinent men, and 46 sexually active women were screened, and 20 of each were randomized (Fig. 1). One additional sexually abstinent man was incorrectly randomized; he was excluded from the analyses and replaced. Reasons for ineligibility and nonenrollment are shown in the flow diagrams in Figure 1. Retention was high: 1 sexually abstinent woman withdrew early and 1 man was lost to follow-up (both at the day 7 visit).
Table 1 shows baseline characteristics of women and men. Randomization seems to have been successful, with few differences among product use groups in the 3 cohorts. None of the participants used antiretroviral therapy during the study.
Reported adherence with gel use was high: 25 of 27 women in gel-use groups used at least 14 applicators and two used 13 (they were given 16 applicators and asked to use 1 per day for 14 days). All but 3 women used the applicators on consecutive days starting at the enrollment visit: 2 women delayed initiation of gel use due to menses and 1 woman interrupted gel use for 4 days because she ran out of supplies due to a scheduling conflict. All but 1 of the men used at least 7 applicators (they were given 8 applicators and asked to use 1 per day for 7 days); 1 male participant was unable to comply because 3 of the 8 applicators he received were empty due to a manufacturing error.
Adherence with other protocol requirements was also high. Among the sexually active women, the median number of sex acts during the 14-day gel-use period was 6.2 for all product use groups combined (range, 2 to 19), with no significant difference among groups. Only 1 woman in the sexually abstinent cohort reported having sex, and 3 women reported vaginal cleansing practices, during the study period. All sexually active women reported condom use during 100% of sex acts. However, while almost all sexually active women inserted gel once per day (see above), 71% of women in the Carraguard group and 86% of women in the placebo group reported that sex took place more than 60 minutes after gel insertion at least once during the study.
Few genital findings were observed in women during follow-up. There were 5 genital findings with epithelial disruption in 3 women, all of which were superficial (stroma not exposed), on the external genitalia, and in the sexually abstinent cohort. These findings included an ulcer in the Carraguard group that persisted for 3 visits (day 7, 14, and 21), an ulcer in the placebo group at day 7, and excoriation in the Carraguard group at day 21. There were 6 findings with intact epithelium in 4 women; all occurred in the sexually abstinent cohort. These findings included a recurrent perianal wart in the placebo group (at day 0, 7, and 21), and 2 cases of cervical erythema in the placebo (one each at day 7 and 14) and one in the Carraguard group (at day 0, 7, and 21). None of the women in the no-product groups, and none of the men, ever had a genital finding during follow-up.
Other Clinical Findings
Abnormal clinical exam findings in women during follow-up were rare, with few clinical diagnoses made, medications prescribed, or referrals given. The most common exam findings during follow-up were palpable bilateral inguinal lymph nodes in women and men, and abnormal vaginal discharge in women (data not shown). However, these clinical findings were also common before initiation of gel use, and were equally common in the 3 product use groups, with 1 exception: sexually active women in the Carraguard group at the day 21 visit were significantly less likely to have palpable inguinal lymph nodes than women in the placebo (Fisher exact P < 0.01) and no-product (P < 0.01) groups. None of the women became pregnant during the study. There were no incident T. pallidum, N. gonorrhoeae, C. trachomatis, and T. vaginalis infections in any of the cohorts.
Vaginal Flora Changes
Seven of 19 sexually abstinent women had an incident BV infection during the study (defined as a Nugent score of 7 to 10 and regardless of symptoms); 6 of these were diagnosed during the 14 days of gel use. BV rates did not differ statistically significantly among the 3 product use groups. One woman (Carraguard group) had an incident yeast infection during follow-up. In the sexually active cohort, 10 of 20 women ever had an incident BV infection during the study, but only half of these were diagnosed during the 14 days of gel use. There was no evidence for differences among groups. Two women (1 in the placebo group and 1 in the no-product group) had a yeast infection during follow-up.
Almost half (48%) of the CVL specimens had no detectable HIV-1 RNA levels, and 20% had detectable but nonquantifiable levels. Table 2 shows that there was no evidence for differences among product use groups in viral load changes between study visits. At all visits, the number of women with no change or a significant decrease in shedding was larger than the number of women with a significant increase.
Markers of Genital Inflammation, Epithelial Sloughing, and Microhemorrhage
Almost all women in both cohorts had leukocytes and hemoglobin in their CVL specimens during follow-up, but concentrations were low (data not shown). In contrast, none of the men had leukocytes in their urine during follow-up. Table 3 shows that there was no evidence for differences in leukocyte or hemoglobin level changes between visits among women groups. The median number of epithelial cells found in CVL specimens for all women combined was 6.35 log10 cells per CVL. Very few women had significant increases in the number of epithelial cells shed between visits, with no significant variation among groups (Table 3).
All women were asked about a list of symptoms at each study visit, and all answered affirmatively to at least 1 during the study. The most common symptoms reported during follow-up were vaginal discharge (25%), lower abdominal pain (23%), increased urinary frequency (20%), pain when urinating (18%), genital itching (18%), genital burning (13%), and lower back pain (13%). There were no significant differences in self-reported symptoms among product use groups, but women in the Carraguard group (the 2 cohorts combined) were slightly more likely to report abnormal nonmenstrual vaginal bleeding than women in the placebo and no-product groups (Fisher exact P = 0.06). Male participants reported very few symptoms, with no differences among the 3 groups. The most commonly reported symptoms in men were increased urinary frequency (11%) and fatigue (11%).
In this study, adverse events included all structurally assessed endpoints described above and spontaneous reports. No serious AEs occurred. A total of 179 AEs were recorded, 165 in women and 14 in men (Table 4A and B). The majority of AEs in women and men were judged by study clinicians to be mild (79%) and unrelated or probably unrelated to study gel use (67%). In women, more AEs were reported in the sexually abstinent (111) than the sexually active cohort (54), but the number of AEs was equally distributed among product use groups in the 2 cohorts overall (31% in the Carraguard, 39% in the placebo, and 30% in the no-product group) (Table 4A). In men, there were more AEs in the no-product group compared to the two gel-use groups (Table 4B). In women, only 15 of the moderate and none of the severe AEs were judged as possibly or probably related to study gel use; these included 5 cases of urinary tract infection and 2 cases each of abdominal pain, BV, genital ulcers, abnormal menstrual onset, and abnormal vaginal discharge. In men, none of the moderate or severe AEs were judged to be related to study gel use.
To date, few microbicide safety studies have been conducted in HIV-positive women and men.16-19 In this study, in addition to traditional microbicide trial safety endpoints, the effects of microbicide use on vaginal shedding of HIV-1 RNA and markers of genital inflammation, epithelial sloughing, and microhemorrhage were explored. The inclusion of a no-product group allowed us to compare Carraguard with both no product use and placebo use. This is important because use of placebo may be associated with study endpoints, particularly genital irritation. Of note, the manner in which male participants used the gels in this study (ie, by direct application to the penis) is not the expected route by which men will be exposed to microbicide gels during actual use; interpretation of results should take this into account.
Our study showed that Carraguard was well tolerated by HIV-positive women and men. Genital findings were rare in the sexually abstinent women's cohort, with no evidence of differences among product use groups; there were no genital findings detected in the sexually active women or men. While incident BV was relatively common among women during follow-up (35% of sexually abstinent and 50% of sexually active women), there were no differences among groups in the 2 cohorts. Yeast was rarely detected at baseline and during follow-up. There were no STIs during follow-up in any of the cohorts, and other clinical exam findings and self-reported symptoms were equally distributed among product use groups. Furthermore, there were no significant increases in the amount of HIV-1 RNA, leukocytes, hemoglobin, or epithelial cells in CVL specimens between visits or across groups (keeping in mind that it is not known what changes are clinically relevant13). No leukocytes were detected in urine from men after 7 days of gel use. No serious AEs occurred in this study, and AEs were equally distributed among groups in the 3 cohorts.
The relatively large number of AEs recorded in this study most likely reflects intense participant monitoring, with targeted genital assessments, as has been noted in other published microbicide trials.9,17-20 Clinical findings in this study are similar to those of other Carraguard studies in HIV-negative women8-10 and to other microbicide safety studies in HIV-positive women and men.17-20 In a multicountry safety study of PRO 2000 gel, genital findings were more common (27% of women had at least 1 cervical abnormality during 2 weeks of product use), but incident BV was slightly less common (26%) than in our study.17 As in our study, the PRO 2000 study showed fewer AEs in sexually active than sexually abstinent women.
Genital tract inflammation and epithelial disruption are thought to play important roles in the sexual transmission of HIV.21,22 Laboratory-based markers of inflammation and epithelial disruption are increasingly being used in microbicide studies to improve on the subjective nature of traditional safety endpoints. The clinical significance of these markers, and their utility for microbicide safety assessments, requires further study.
A few limitations of this study, which are common to phase 1 safety studies and microbicide trials in general, should be noted. Results should be interpreted as exploratory due to the small sample size. Participants were selected carefully according to stringent criteria and are therefore not representative of the general population in Durban. Several key data points, such as adherence and sexual activity, were assessed by self-report and were therefore prone to social desirability bias. Furthermore, the potential for bias in ascertainment of clinical outcomes cannot be ruled out, particularly in the no-product groups for whom treatment assignment could not be blinded. Finally, the circumcision status of male participants, which could influence male safety findings due to trapping of gel under the foreskin, was not recorded.
In conclusion, our findings support the continued development of Carraguard as a vaginal microbicide.
We are grateful to the study participants for their dedication to the study. We thank Beverly Winikoff, Kelly Blanchard, Robin Maguire, and other colleagues at the Population Council offices in Johannesburg and New York. We also thank John Karon, Catherine McLean, and staff involved in data collection at the MRC HIV Prevention Unit in Durban; the Lancet and Bouwer laboratories in Durban; the Cytology lab at Albert Luthuli Hospital in Durban; the University of Limpopo, Medunsa campus; the Department of Medical Microbiology in Pretoria; and the US CDC (Dr. Clyde Hart's laboratory) in Atlanta.
1. van de Wijgert J, Coggins C. Microbicides to prevent heterosexual transmission of HIV: ten years down the road. BETA. 2002;15:23-28.
2. Pearce-Pratt R, Phillips DM. Sulphated polysaccharides inhibit lymphocyte-to-epithelial transmission of human immunodeficiency virus-1. Biol Reprod. 1996;54:173-182.
3. Maguire RA, Bergman N, Phillips DM. Comparison of microbicides for efficacy in protecting mice against vaginal challenge with herpes simplex virus type 2, cytotoxicity, antibacterial properties, and sperm immobilization. Sex Transm Dis. 2001;28:259-265.
4. Nahmias AJ, Kibrick S, Bernfeld P. Effect of synthetic and biological polyanions on herpes simplex virus. Proc Soc Exp Biol Med. 1964;115:993-996.
5. Zacharopoulos VR, Phillips DM. Vaginal formulations of carrageenan protect mice from herpes simplex virus infection. Clin Diagn Lab Immunol. 1997;4:465-468.
6. Maguire RA, Zacharopoulos VR, Phillips DM. Carrageenan-based nonoxynol-9 spermicides for prevention of sexually transmitted infections. Sex Transm Dis. 1998;25:494-500.
7. Elias CJ, Coggins C, Alvarez F, et al. Colposcopic evaluation of a vaginal gel formulation of iota-carrageenan. Contraception. 1997;57:387-389.
8. Coggins C, Blanchard K, Alvarez F, et al. Preliminary safety and acceptability of a carrageenan gel for possible use as a vaginal microbicide. Sex Transm Infect. 2000;76:480-483.
9. Kilmarx PH, van de Wijgert JH, Chaikummao S, et al. Safety and acceptability of the candidate microbicide Carraguard in Thai Women: findings from a Phase II clinical trial. J Acquir Immune Defic Syndr. 2006;43:327-334.
10. Pistorius AG, van de Wijgert JH, Sebola M, et al. Microbicide trials for preventing HIV/AIDS in South Africa: phase II trial participants' experiences and psychological needs. SAHARA J. 2004;1:78-86.
11. World Health Organization. Manual for the standardization of colposcopy for the evaluation of vaginally administered products: an update. Geneva: WHO; 2000.
12. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol. 1991;29:297-301.
13. Moench T, Mehrazar K, Cone R, et al. Sensitive methods to detect epithelial disruption: tests for microhemorrhage in cervicovaginal lavages. J Acquir Immune Defic Syndr. 2004;37(Suppl 3):S194-S200.
14. World Health Organization Collaborating Centre for International Drug Monitoring. WHO Adverse Reaction Terminology. Uppsala, Sweden: WHO; 1991/1993.
15. World Medical Association. Declaration of Helsinki: recommendations guiding medical doctors in biomedical research involving human subjects. Cedex, France: World Medical Association; 1989. Available at: http://www.fda.gov/oc/health/helsinki89.html
. Accessed September 27, 2007.
16. International Conference on Harmonization. Guidance for industry: E6 good clinical practice: consolidated guidance. Geneva: International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use; 1996. Available at: http://www.fda.gov/cder/guidance/959fnl.pdf
. Accessed September 27, 2007.
17. Mayer KH, Karim SA, Kelly C, et al. Safety and tolerability of vaginal PRO 2000 gel in sexually active HIV-uninfected and abstinent HIV-infected women. AIDS. 2003;17:321-329.
18. El-Sadr WM, Mayer KH, Maslankowski L, et al. Safety and acceptability of cellulose sulfate as a vaginal microbicide in HIV-infected women. AIDS. 2006;20:1109-1116.
19. Van Damme L, Jespers V, Van Dyck E, et al. Penile application of dextrin sulphate gel (Emmelle). Contraception. 2002;66:133-136.
20. Tabet SR, Callahan MM, Mauck CK, et al. Safety and acceptability of penile application of 2 candidate topical microbicides: BufferGel and PRO 2000 gel: 3 randomized trials in healthy low-risk men and HIV-positive men. J Acquir Immune Defic Syndr. 2003;33:476-483.
21. Anderson DJ, Politch JA, Tucker LD, et al. Quantitation of mediators of inflammation and immunity in genital tract secretions and their relevance to HIV type 1 transmission. AIDS Res Hum Retroviruses. 1998;14(Suppl 1):S43-S49.
22. Coombs RW, Reichelderfer PS, Landay AL. Recent observations on HIV type-1 infection in the genital tract of men and women. AIDS. 2003;17:455-480.
© 2007 Lippincott Williams & Wilkins, Inc.