Oskari Heikinheimo 1,3, Päivi Lehtovirta 1, Jukka Suni 2 and Jorma Paavonen 1

Human Reproduction Vol.21, No.11 pp. 2857 2861, 26 Advance Access publication July 31, 26. doi:1.193/humrep/del264 The levonorgestrel-releasing intrauterine system (LNG-IUS) in HIV-infected women effects on bleeding patterns, ovarian function and genital shedding of HIV Oskari Heikinheimo 1,3, Päivi Lehtovirta 1, Jukka Suni 2 and Jorma Paavonen 1 1 Department of Obstetrics and Gynaecology, University of Helsinki and 2 HUSLAB, Section of Virology, Helsinki University Central Hospital, Helsinki, Finland 3 To whom correspondence should be addressed at: Department of Obstetrics and Gynaecology, University of Helsinki, PO Box 14, SF-29, HUS, Helsinki, Finland. E-mail: oskari.heikinheimo@helsinki.fi BACKGROUND: Safe and effective contraceptives are needed for human immunodeficiency virus (HIV)-infected women. The levonorgestrel-releasing intrauterine system (LNG-IUS) is a highly effective contraceptive with additional health benefits. The objective of this study was to evaluate the effects of the LNG-IUS among HIV-infected women. METHODS: Twelve systematically managed HIV-infected women were studied prospectively. Following a 2-month run-in period, the subjects had an LNG-IUS inserted and were followed up for 1 year. Patterns of bleeding, blood haemoglobin and CD4-lymphocyte content, plasma HIV RNA, serum levels of LNG, of estradiol (E 2 ) and of ferritin and genital shedding of HIV RNA were monitored. RESULTS: Menstrual bleeding was reduced significantly during the use of the LNG-IUS; this was associated with slight increases in serum haemoglobin and ferritin levels. Serum E 2 concentrations remained in the follicular range in all subjects. Among subjects using antiretroviral medication, the proportion of cervicovaginal lavage specimens with detectable HIV RNA was 1% before and after the insertion of the LNG-IUS. CONCLUSIONS: The effects of the LNG-IUS on bleeding patterns, body iron stores and ovarian function were similar to those seen in healthy women. Genital shedding of HIV RNA was not affected by the LNG-IUS. These data encourage further studies on the effects of the LNG-IUS on reproductive health among HIVinfected women. Key words: contraception/haart/hiv/ovarian function/uterine bleeding Introduction At the end of 25, 18 million women worldwide are infected with human immunodeficiency virus (HIV). The great majority of these women are of fertile age, and an estimated 7 HIVinfected children are born annually (www.who.int/hiv). In an attempt to prevent horizontal transmission of HIV, condoms currently represent the only effective contraceptive method available (Mitchell and Stephens, 24). However, given the large number of HIV-infected children born annually, additional contraceptive methods for HIV-infected women are needed. Based on international recommendations, there should be no restrictions on the use of hormonal contraception for women at risk of HIV infection or for HIV-infected women (www.who.int/ reproductive-health/stis/hc_hiv/nairobi_statement. html). However, the use of copper-releasing intrauterine devices (Cu-IUDs) in HIV-infected women has been viewed with caution (www.fhi. org/en/rh/faqs/contrahiv_faq.htm). Yet, the available data suggest that the use of Cu-IUD is safe also in HIV-infected women (Morrison et al., 21). Regardless of the method, dual protection by means of condom use should be encouraged. The levonorgestrel-releasing intrauterine system (LNG-IUS) is a highly effective long-term contraceptive method, with a Pearl index of.1 (Luukkainen and Toivonen, 1995). Unlike Cu-IUDs, the LNG-IUS reduces menstrual bleeding (Milsom et al., 1991). In addition, the LNG-IUS may reduce the risk of pelvic inflammatory disease (PID) (Toivonen et al., 1991). Thus, the LNG-IUS has several features that can make it an ideal contraceptive method for HIV-infected women. The reduction of menstrual bleeding improves body iron stores and is likely to reduce exposure to infected blood. In addition, protection from PID might be valuable in a population prone to infectious morbidity. Hence, it is not surprising that the use of the LNG-IUS has been advocated for HIV-infected women (Guillebaud, 24). However, published literature on the use of the LNG-IUS in HIV-infected women is limited to one case report (Cooling, 1999). In this prospective study, we assessed the effects of the LNG- IUS on bleeding patterns, body iron stores and ovarian function, as well as on genital shedding of HIV during 1 year of follow-up among systematically managed HIV-infected women. The Author 26. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. 2857 For Permissions, please email: journals.permissions@oxfordjournals.org

O.Heikinheimo et al. Patients and methods Twelve HIV-infected women attending a gynaecological HIV-outpatient clinic (Lehtovirta et al., in press) were recruited. They were regularly menstruating subjects aged 18 45 years who used condoms as contraceptives. The demographic characteristics of the subjects are summarized in Table I. Each woman signed an informed consent document. The study protocol was approved by the Institutional Review Board of Helsinki University Central Hospital and the Finnish National Agency for Medicines. Exclusion criteria were CD4-lymphocyte count <.35 1 9 cells/ml, untreated squamous intraepithelial lesion (SIL) in a Pap smear, acute PID or history of PID during the year preceding this study, use of illicit drugs, active or chronic hepatitis, known Müllerian anomaly, use of oral contraceptives or progestins and pregnancy. This study was prospective. Following recruitment to the study, breast and gynaecological examinations, and vaginal ultrasonography were performed. A cervicovaginal lavage specimen was collected by flushing the vagina and cervix repeatedly (i.e. 3 4 times) with 1 ml of physiological saline. The LNG-IUS (MIRENA, Schering Oy, Turku, Finland) was inserted 2 months (range 1 4) after recruitment, between cycle days 1 and 7. A lavage specimen from the vagina and cervix and a blood sample were collected before the insertion of the LNG-IUS. Following insertion, correct location of the device was confirmed using vaginal ultrasonography. Gynaecological examination, vaginal ultrasonography, taking of cervicovaginal lavage specimens and blood sampling were performed at 1 week, 3 months, 6 months and 12 months following the insertion of the LNG-IUS. Serum and lavage specimens were frozen and stored at 8 C. Pap smears were collected at enrolment and at 6 and 12 months. The subjects kept diaries of bleeding (requiring protection) and spotting (requiring no protection or using panty liners only) during a 3-day period preceding the insertion of the LNG-IUS, as well as at 5 6 and 11 12 months following the insertion of the LNG-IUS. Management of HIV infection was carried out at the Department of Medicine, Helsinki University Hospital. Individual visits were scheduled at 3-month intervals. Laboratory records were consulted for CD4-lymphocyte counts, plasma HIV RNA measurements and blood haemoglobin values. The most recent values were used for calculations of mean CD4-lymphocyte and haemoglobin values. Laboratory assays Serum levels of estradiol (E 2 ) were measured by time-resolved fluoroimmunoassay, using commercial kits (Delfia, PerkinElmer Life Sciences, Turku, Finland). The levels of LNG were measured using radioimmunoassay (Weiner and Johansson, 1976; Suhonen et al., 1995). Serum levels of ferritin were measured in samples collected at the time of LNG-IUS insertion and at 12 months, using chemiluminescence immunoassays (Architect immunoanalyzer, Abbott Diagnostic Division, Abbott Park, IL, USA). Protein content of the lavage Table I. Demographic characteristics of the study subjects Age (years) 35.8 ± 4.6 Nulliparous 5/12 (42%) Duration of HIV infection (years) a 8.6 ± 2.7 Use of HAART b 1/12 (83%) CD4-lymphocyte level ( 1 9 /l).63 ±.18 The data are presented as mean ± SD. a Calculated from the first positive HIV antibody test. b Highly active antiretroviral therapy (HAART) (i.e. nine combinations of reverse transcriptase inhibitors and protease inhibitors). 2858 specimens was measured using Protein assay dye reagent (Bio-Rad Laboratoires Inc., Espoo, Finland). The HIV RNA load was measured using a COBAS TaqMan 48 HPS real-time assay (CTM48 HPS; Roche Molecular Systems, Pleasenton, CA, USA) with a sensitivity of <4 copies/ml. The quantification of HIV RNA in lavage specimens was carried out using the CTM48 HPS system as follows:.5 ml of the lavage sample was passed through a High Pure System (HPS) column to isolate RNA. The obtained RNA was run in a CTM48 analyzer together with internal controls. Control samples were created using lavage specimens from HIV-negative women, with known amounts of HIV-positive plasma added. Protein content of the lavage specimens (.5 ml) was measured thereafter, and the results are expressed as HIV RNA copies/mg of protein. Statistical analysis Chi-square and paired sign tests were used as appropriate. A two-tailed P-value <.5 was considered statistically significant. The statistical software used was StatView (SAS Institute Inc., Cary, NC, USA). Results The use of the LNG-IUS was well tolerated, and all subjects completed the 12 months of follow-up. No pregnancies occurred during the follow-up period. Eleven women chose to continue the use of the LNG-IUS following the completion of the study period. One LNG-IUS was removed at 12 months because of metrorrhagia resulting from the partial expulsion of the LNG-IUS. The patterns of bleeding and spotting in 3-day periods are shown in Figure 1. The number (mean ± SD) of days of bleeding decreased from 4.5 (±1.8) reported before the insertion of the LNG-IUS to 2.4 (±2.6) (P =.1) and 2.4 (±3.5) (P =.1) at 6 and 12 months, respectively. Seven of the 12 subjects reported spotting only or no bleeding at 12 months. The mean number of days of spotting was increased at 6 months (P =.2), thereafter decreasing to pretreatment levels. Reduction in bleeding was associated with slight increases in blood haemoglobin (135. ± 8.3 versus 138.6 ± 14.1 g/l; P = NS) and serum ferritin (24.7 ± 18.1 versus 26.5 ± 15.1 μg/l; P = NS) levels at 1 year of follow-up. Figures 2 and 3 show the circulating levels of LNG and E 2, respectively. The levels of LNG were similar in the subjects with and without antiretroviral medication, and therefore, the data were combined. As expected, a slight decrease was noted in circulating levels of LNG over the study period. The levels of E 2 remained in the follicular-phase range (i.e. >7 pmol/l) in all subjects throughout the study. The percentages of vaginal lavage samples with detectable HIV RNA are shown in Figure 4. Results from subjects with and without antiretroviral medication are shown separately. In women with antiretroviral medication, HIV RNA was detectable in 1% of the lavage specimens collected both before and after the insertion of the LNG-IUS. In samples with detectable HIV RNA, the viral load (mean ± SD) was 638 ± 822 copies/mg protein before and 655 ± 527 copies/mg protein after insertion of the LNG-IUS (P = NS). Among subjects using antiretroviral medication, all six vaginal lavage specimens were negative for HIV RNA in 6 of the

LNG-IUS in HIV-infected women A Bleeding/3 days B Spotting/3 days 7 6 5 4 3 2 1 8 7 6 5 4 3 2 1 Before 5-6 mo 11-12 mo Figure 1. Numbers of days of bleeding (A) and spotting (B) (mean + SD) in 3-day periods before and after the insertion of the LNG-IUS. Statistically significant differences in comparison with pre-treatment values are indicated (*P <.5, **P <.2). 1 subjects. Two subjects had detectable HIV RNA in only one and two subjects in two of the lavage specimens. Of the latter subjects, one had detectable HIV RNA in the specimens collected at recruitment and at 12 months (122 and 257 copies/mg protein, respectively), whereas in the subject with the partial expulsion of the LNG-IUS, HIV RNA was detectable in the samples collected at 1 week and at 6 months (1336 and 22 copies/mg protein, respectively). Among patients using antiretroviral medication, the plasma level of HIV RNA was below the detection limit of the assay (4 copies/ml) in 49 of the 5 samples collected during the study period (±1 month). Thus, only one subject had a single ** Before 5-6 mo 11-12 mo Figure 2. Circulating levels of levonorgestrel at 1 week and 3, 6 and 12 months after LNG-IUS insertion. * Figure 3. Circulating levels of estradiol before, during and after the insertion of the LNG-IUS. HIV detectable (%) 1 8 6 4 2 Pre Ins 1wk 3mo 6mo 12mo TIME Figure 4. Percentage of vaginal lavage specimens with detectable HIV RNA. Samples collected from subjects using antiretroviral medication (grey bars) and not using antiretroviral medication (black bars) are shown separately. measurable plasma HI viral load; all her vaginal lavage specimens were negative for HIV RNA. Among the two subjects who did not use antiretroviral medication, the plasma levels of HIV RNA varied between 135 and 23 6 copies/ml. In one, HIV RNA was detectable in the lavage specimens collected at recruitment, before the insertion of the LNG-IUS and at 3 months (353, 795 and 12 copies/mg protein, respectively), whereas in the other, HIV RNA was detectable in the samples collected at 1 week and 3 and 6 months (1693, 1238 and 1 167 copies/mg protein, respectively). The levels of CD4 lymphocytes (mean ± SD) were.63 ±.18 1 9 /l at the beginning of this study,.58 ±.17 1 9 /l after 6 months and.58 ±.15 1 9 /l after 12 months (P = NS). Papanicolau smear findings were all normal in 7 of the 12 (58%) subjects. Atypical squamous cells of undetermined significance (ASC-US) were present in two (17%), four (33%) and three (25%) of the Pap smears collected before LNG-IUS insertion and after 6 and 12 months (P = NS), respectively. No subject had cytologic changes consistent with squamous intraepithelial lesion (SIL). Discussion In this prospective study, the use of the LNG-IUS resulted in reduced vaginal bleeding, and at the end of this study, most of 2859

O.Heikinheimo et al. the women reported only occasional spotting. The reduced bleeding was associated with slight, albeit not statistically significant, increases in blood haemoglobin and ferritin values. Genital shedding of HIV remained constant before and during the use of the LNG-IUS. The LNG-IUS may not be a realistic option for women living in the areas hit hardest by the global HIV/AIDS pandemic. In addition, our study subjects were highly compliant. Thus, the present results are likely to be relevant only to HIVinfected women with access to high-quality health care services. It may also be argued that a randomized study design would have been more valuable. However, as the suppression of menstrual bleeding is a unique feature of LNG-IUS, the selection of a proper comparator would be difficult. It has been suggested that antiretroviral medication alters the metabolism of contraceptive steroids (Mitchell and Stephens, 24). However, the circulating levels of LNG among our subjects were in the same range as has been reported in healthy subjects (Bilian et al., 199; Suhonen et al., 1995), suggesting that the effect of the LNG-IUS is no different in women using antiretroviral medication. However, given the various combinations of antiretroviral medications used, the potential effect of a specific combination could not be assessed. The use of hormonal contraception has been associated with modest increases in genital shedding of HIV in some (Mostad et al., 1997; Wang et al., 24), but not all (Kovacs et al., 21), studies. It has been proposed that the suppression of ovarian activity and low circulating levels of E 2 because of hormonal contraception result in thinning of the vaginal mucosa and increased cervicovaginal shedding of HIV (Li and Short, 22). The suppression of endometrial epithelium by LNG is thought to be the main contraceptive mechanism of the LNG-IUS (Luukkainen and Toivonen, 1995). Most women display normal follicular development and ovulation during the use of LNG-IUS (Barbosa et al., 199). Similarly, in our study, the circulating levels of E 2 remained in the follicular phase range in all subjects. One of the key findings was that genital shedding of HIV RNA did not change after the insertion of the LNG-IUS. Similarly, the use of Cu-IUD does not increase genital shedding of HIV DNA (Richardson et al., 1999). This suggests that continued follicular activity associated with LNG-IUS or Cu-IUD use maintains vaginal epithelium and does not change genital shedding of HIV RNA. The circulating HIV load is the most important determinant of cervicovaginal shedding of HIV RNA, even among women using antiretroviral medication (Kovacs et al., 21; Benki et al., 24). In this study, occasional cervicovaginal shedding of HIV was detected in 4% of the subjects using highly active antiretroviral therapy (HAART). This is in line with the results of previous studies reporting 25 33% rates of detectable genital HIV RNA among women using antiretroviral medication and having an undetectable circulating viral load (Kovacs et al., 21; Fiore et al., 23). HIV RNA was detected in half of the lavage specimens of the two subjects not receiving antiretroviral medication. The rate of detectable genital shedding of HIV RNA has been reported to exceed 8% 286 among women not receiving antiretroviral medication (Kovacs et al., 21; Wang et al., 24; Nagot et al., 25). Thus, regardless of the circulating HIV load and possible additional contraceptives used, barrier methods of contraception are needed to reduce the risk of horizontal transmission of HIV. In summary, the use of the LNG-IUS among HIV-infected women resulted in the reduction of menstrual bleeding. Follicular activity and circulating E 2 levels did not change. The use of the LNG-IUS did not change the rate of genital shedding of HIV RNA. Thus, the use of the LNG-IUS appears to be safe among systematically managed HIV-infected women. These results encourage further studies on the use of the LNG-IUS to improve reproductive health and to prevent unintended pregnancy among HIV-infected women. Acknowledgements We thank Ms Teija Karkkulainen for her never-failing handling of the patients. Ms Nina Hedkrok, Ms Eira Meskanen and Ms Marjatta Tevilin are thanked for their expert technical help. Financial support from Helsinki University Central Hospital Research Funds and the Finnish Medical Foundation is gratefully acknowledged. References Barbosa I, Bakos O, Olsson S-E, Odlind V and Johansson E (199) Ovarian function during the use of a levonorgestrel-releasing IUD. Contraception 42,51 66. Benki S, Mostad S, Richardson B, Madaliya K, Kreiss J and Overbaugh J (24) Cyclic shedding of HIV-1 RNA in cervical secretions during the menstrual cycle. J Infect Dis 189,2192 221. Bilian X, Liying Z, Xuling Z, Mengchun J, Luukkainen T and Allonen H (199) Pharmacokinetic and pharmacodynamic studies of levonorgestrelreleasing intrauterine device. Contraception 41,353 362. Cooling H (1999) Successful use of levonorgestrel intrauterine system in a HIV positive woman. Br J Fam Plann 25,25 26. Fiore J, Suligoi B, Saracino A, Di Stefano M, Bugarini R, Lepera A, Favia A, Monno L, Angarano G and Pastore G (23) Correlates of HIV-1 shedding in cervicovaginal secretions and effects of antiretroviral therapies. AIDS 17,2169 2176. Guillebaud J (24) Contraception Your Questions Answered, 4th edn. Churchill Livingstone, London and Philadelphia, pp. 46 47. Kovacs A, Wasserman S, Burns D, Wright D, Cohn J, Landay A, Weber K, Cohen M, Levine A, Minkoff H et al. (21) Determinants of HIV-1 shedding in the genital tract of women. Lancet 358,1593 161. Lehtovirta P, Finne P, Nieminen P, Skogberg K, Savonius H, Paavonen J and Heikinheimo O (26) Prevalence and risk factors of squamous intraepithelial lesions of the cervix among human immunodeficiency virus (HIV) infected women a long term follow-up study among a low prevalence population. Int J STD AIDS, in press. Li M and Short R (22) How estrogen or progesterone might change a woman s susceptibility to HIV-1 infection. Aust N Z J Obstet Gynecol 42,472 475. Luukkainen T and Toivonen J (1995) Levonorgestrel-releasing IUD as a method of contraception with therapeutic properties. Contraception 52,269 276. Milsom I, Andersson K, Andersch B and Rybo G (1991) A comparison of flubriprofen, tranexamic acid, and a levonorgestrel-releasing intrauterine contraceptive device in the treatment of idiopathic menorrhagia. Am J Obstet Gynecol 164,879 883. Mitchell H and Stephens E (24) Contraceptive choice for HIV positive women. Sex Transm Infect 8,167 173. Morrison C, Sekadde-Kigondu C, Sinei S, Weiner D, Kwok C and Kokonya D (21) Is the intrauterine device appropriate contraception for HIV-1-infected women? BJOG 18,784 79. Mostad S, Overbaugh J, DeVange D, Welch M, Chohan B, Mandaliya K, Nyange P, Martin H Jr, Ndinya-Achola J, Bwayo J et al. (1997) Hormonal contraception, vitamin A deficiency, and other risk factors for shedding of HIV-1 infected cells from the cervix and vagina. Lancet 35,922 927.

LNG-IUS in HIV-infected women Nagot N, Foulongne V, Becquart P, Mayaud P, Konate I, Ouedraogo A, Defer M-C, Weiss H, Van de Perre P and Segondy M (25) Longitudinal assessment of HIV-1 and HSV-2 shedding in the genital tract of West African women. J Acquir Immune Defic Syndr 39,632 634. Richardson B, Morrison C, Sekadde-Kigondu C, Sinei S, Overbaugh J, DeVange Panteleeff D, Weiner D and Kreiss J (1999) Effect of intrauterine device use on cervical shedding of HIV-1 DNA. AIDS 13,291 297. Suhonen S, Holmström T, Allonen H and Lähteenmäki P (1995) Intrauterine and subdermal progestin administration in postmenopausal hormone replacement therapy. Fertil Steril 63,336 342. Toivonen J, Luukkainen T and Allonen H (1991) Protective effect of intrauterine release of levonorgestrel on pelvic infections: three years comparative experience of levonorgestrel- and copper-releasing intrauterine devices. Obstet Gynecol 77,261 264. Wang C, McClelland S, Overbaugh J, Reilly M, Panteleeff D, Mandaliya K, Chohan B, Lavreys L, Ndinya-Achola J and Kreiss J (24) The effect of hormonal contraception on genital shedding of HIV-1. AIDS 18,25 29. Weiner E and Johansson EDB (1976) Plasma levels of d-norgestrel, estradiol and progesterone during treatment with Silastic implants containing d- norgestrel. Contraception 14,81 92. Submitted on March 8, 26; resubmitted on May 3, 26; accepted on June 7, 26 2861