- Email: [email protected]
Prevalence of visible disruption of cervical epithelium and cervical ectopy in african women using depo-provera®
ORIGINAL RESEARCH ARTICLE
Prevalence of Visible Disruption of Cervical Epithelium and Cervical Ectopy in African Women Using Depo-Provera威 Louise Kuhn,* Lynette Denny,† Amy E. Pollack,‡ and Thomas C. Wright§
Associations between Depo-Provera威 (injectable, progesterone-only contraceptive) use and visible disruption of cervical epithelium and cervical ectopy were investigated using data collected as part of a cervical cancer screening study in periurban Cape Town, South Africa. Women were interviewed about their contraceptive use, and underwent a gynecologic examination that included two 35-mm photographs of the cervix after application of 5% acetic acid. Photographs of 723 subjects were reviewed (blind to clinical information and using systematic criteria developed before review) for evidence of atrophy and epithelial disruption, including inflammation and ulceration. The percentage of the cervix covered with columnar epithelium (ectopy) was also estimated from the photographs. A random sample of 85 photographs was reviewed again for reliability. A total of 121 current users of Depo-Provera were no more likely to have evidence of epithelial disruption (38%) than 574 nonusers (39%), odds ratio (OR) ⫽ 1.37, 95% CI: 0.89 –2.11 adjusting for age and parity. The prevalence of significant ectopy (columnar epithelium covering ⬎10% of the cervix) was also no different among current Depo-Provera users (OR ⫽ 1.22, 95% CI: 0.80 –1.86 adjusting for age and parity). Reliability of visual scoring of epithelial disruption and ectopy was excellent ( ⫽ 0.8). Although the underlying prevalence of visible disruption of cervical epithelium was very high, current use of DepoProvera was not associated with increased prevalence of visible disruption of the cervical epithelium or with ectopy in this sample of African women. CONTRACEPTION 1999; 59:363–367 © 1999 Elsevier Science Inc. All rights reserved.
*Gertrude H. Sergievsky Center, College of Physicians and Surgeons, Columbia University, New York, New York; †Department of Obstetrics & Gynaecology, University of Cape Town, Cape Town, South Africa; ‡AVSC International, New York, New York ; and §Department of Pathology, College of Physicians and Surgeons, Columbia University, New York, New York Name and address for correspondence: L Kuhn, Columbia University, Sergeivsky Center, 630 W 168th Street, New York, NY 10032; Tel.: (212) 305-2398; Fax: (212) 305-2426; e-mail: [email protected] Submitted for publication March 30, 1999 Revised June 7, 1999 Accepted for publication June 7, 1999
© 1999 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
KEY WORDS:
hormonal contraception, progesterone, cervical ectopy, cervical inflammation, heterosexual HIV transmission, Africa
Introduction
A
n experimental study of female rhesus macaques found that subcutaneous progesterone implants increased the efficiency of vaginal simian immunodeficiency virus (SIV) transmission 7.7-fold. When pretreated with progesterone implants, 14 of 18 macaques acquired infection after vaginal inoculation with SIV compared with one of 10 macaqes pretreated with placebo implants.1 Thickness of vaginal squamous epithelium was found to be reduced substantially among three treated macaques compared with three control macaques in a subsequent experiment.1 Based on these findings, the authors concluded that enhancement of vaginal transmission of SIV associated with exogenous progesterone is most likely due to thinning of vaginal epithelium. A thinner epithelium would be expected to be likely to show evidence of loss of mucosal integrity and to be more susceptible to trauma. Foci of ulceration might allow HIV easier passage to susceptible target cells such as Langerhans cells, dendritic cells, or lymphocytes within lower subepithelial levels. These findings have raised considerable concern, as long-acting injectable human contraceptives have synthetic forms of progesterone as their active ingredients. One such contraceptive is depot-medroxyprogesterone acetate, marketed as Depo-Provera威, which is widely used in parts of southern Africa where the HIV epidemic is at its worst. Although observational epidemiological studies have not consistently demonstrated increased susceptibility to HIV transmission with the use of hormonal contraceptives, few studies have investigated progesterone-only contraceptives specifically, and among these few, most have significant methodological limitations.2,3 ISSN 0010-7824/99/$20.00 PII S0010-7824(99)00049-9
364
Kuhn et al.
Contraception 1999;59:363–367
It is generally assumed, but has not been directly shown, that the use of Depo-Provera in women leads to thinning of the vaginal and cervical epithelium with possible secondary ulceration and erosion. In this study, we investigated whether or not women who used Depo-Provera are more likely to have visible evidence of epithelial thinning and trauma than women not using it. Our purpose was to complement studies of heterosexual HIV transmission.
Materials and Methods The data were collected as part of an ongoing cervical cancer screening study that is being conducted in a large periurban, informal settlement (Khayelitsha) outside of Cape Town, South Africa. Women aged 35– 65 years undergo a gynecologic examination that includes several cervical cancer screening tests, including two 35-mm color photographs of the cervix taken by a trained clinician after application of 5% acetic acid. A clinical history is also taken that includes questions about current and past use of birth control. The methods of the larger study are described in more detail elsewhere.4 All patients were recruited with the approval of the Human Subjects’ Institutional Review Board of Columbia University and Research Ethics Committee of the University of Cape Town. Cervical photographs from the first 723 participants were reviewed for evidence of cervical epithelial disruption. Specific criteria were developed before review modifying available preliminary guidelines for standardized colposcopy.5 All cervical photographs were reviewed blind to any clinical information. The slides were reviewed for macroscopically visible evidence of ulceration, inflammation or atrophy. Ulceration was defined as sharply demarcated areas of disrupted epithelium with bleeding. Inflammation was defined as erythematous and edematous changes involving endocervical type columnar epithelium (endocervicitis) or ectocervical type squamous epithelium (ectocervicitis). Ectocervicitis was further classified as either grade 1 (less severe), characterized by focal scattered vascular punctations, similar to the changes of “colpitis macularis” observed in women with trichomoniasis, or grade 2 (more severe), characterized by larger confluent regions of erythema that frequently appeared “punched out” or as shallow erosions (Figure 1). Atrophy was defined as apparent thinning of the cervical epithelium, often accompanied by petechial hemorrhages or diffuse whitening of the epithelium. The percentage of the visible portion of the cervix covered with columnar epithelium (cervical ectopy) was also estimated from each photograph as previously described.6 Ectopy covering 10%
Figure 1. Representative examples from the series of cervical photographs are displayed illustrating the four conditions visually scored. A. Atrophy. Cervical epithelium appears thin, petechial hemorrhages are observed, and there is diffuse whitening of the epithelium. B. Endocervical inflammation. Columnar epithelium appears erythematous and edematous, cell definition is poor, and there may be yellowing in color. C. Grade 1 ectocervical inflammation. Focal scattered vascular punctations are seen in the squamous epithelium. D. Grade 2 ectocervical inflammation. Larger confluent regions of erythema are seen; they appear “punched out” or as shallow erosions.
or more of the cervix was classified as significant ectopy. A random sample of 85 photographs was reviewed again to measure reliability. The prevalence of cervical changes among women currently using Depo-Provera were compared to those who had used Depo-Provera in the past but were not current users, and to those who had never used it. Differences were tested using 2 tests. Differences between the groups were adjusted for age, parity, current and past numbers of sexual partners, marital status, age at first intercourse, and cigarette smoking, using multiple logistic regression. Reliability of diagnosis on repeat review was described using the coefficient of agreement.
Results Photographs from 695 (96%) of the 723 study participants were acceptable for evaluation. Among the 695 women, 7 (1%) had areas of ulceration, 89 (13%) had visible endocervical inflammation, 170 (25%) visible ectocervical inflammation (110 [16%] grade 1 ectocervicitis and 58 [8%] grade 2 ectocervicitis) and 56 (8%) had atrophic changes. Overall, 268 (38%, 95% CI
Contraception 1999;59:363–367
Cervical Epithelium Integrity and Depo-Provera
365
Table 1. Prevalence of disruption of cervical integrity (visible ulceration, inflammation, or atrophy) among 694 women (aged 35– 65 years) in Khayelitsha, South Africa, by age (in years) and parity Percentage of women with visible disruption of cervical epithelium (number in each cell)
Parity None 1 or 2 3 or 4 ⱖ5 Total
35
36–39
40–44
45–49
50–65
Total
10 (10) 21 (75) 32 (69) 43 (7) 26 (161)
17 (12) 8 (46) 34 (82) 42 (36) 34 (176)
0 (2) 29 (21) 36 (61) 40 (67) 36 (151)
0 (3) 44 (9) 39 (28) 45 (51) 42 (91)
0 (0) 22 (9) 35 (23) 72 (83) 61 (115)
11 (27) 26 (160) 35 (263) 52 (244) 38 (694)
34 – 42) women had one or more of these conditions. Representative examples of photographs from women with each of these conditions are displayed in Figure 1. Disruption of cervical epithelium increased with age and parity (Table 1). The percentage of women diagnosed with visible epithelial disruption increased from 26% in women 35 years old to 61% in women ⬎50 years old; and from 11% in women with no children to 52% in women with five or more children (Table 1). In a logistic regression model including both age and parity, both characteristics were significant predictors of epithelial disruption (for each year of age, the odds ratio [OR] ⫽ 1.03, 95% confidence interval [CI]: 1.01–1.06, and for each additional child, OR ⫽ 1.21, 95% CI: 1.11–1.31). Presence of significant ectopy (involving ⱖ10% of the cervix) decreased with age but increased with parity (Table 2). Each year of age was associated with a decreased odds of ectopy (OR ⫽ 0.92, 95% CI: 0.89 – 0.94 adjusting for parity) and each additional child with an increased odds (OR ⫽ 1.21, 95% CI: 1.11–1.32 adjusting for age). The majority (83%) of women reported some past or current use of birth control: 76% reported using Depo-Provera, 38% oral contraceptives, 3% condoms, and 3% other methods of birth control in the past or present. Current use of Depo-Provera was reported by
17% of women, current use of oral contraceptives by 2%, and condoms by 0.6% (4 women). Visible ulceration, inflammation or atrophy was not more common among women reporting current use of Depo-Provera. Among 121 current DepoProvera users, 38% had at least one of these conditions, compared with 34% of 408 reporting only past use of Depo-Provera, and 49% of 166 women who had never used it. In addition, no consistent increases in epithelial disruption were identified in users of DepoProvera stratified by age or parity (Table 3). After adjusting for age and parity as continuous variables, there continued to be no differences in the odds of epithelial disruption (OR ⫽ 1.37 95% CI: 0.89 –2.11) between current users of Depo-Provera and nonusers. Further adjustment for current and past numbers of sexual partners, marital status, age at first intercourse, and cigarette smoking did not change the magnitude of these associations. The percentage of women with significant ectopy was higher in current and past users of Depo-Provera in unadjusted analyses, but no trends persisted after adjustment for age and parity as continuous variables (OR ⫽ 1.22 95% CI: 0.80 –1.86). Within the sample of 85 cervical photographs that were re-reviewed, reliability of the diagnosis of ectocervical inflammation ( coefficient of agreement [] ⫽ 0.9), atrophy ( ⫽ 0.8), and ectopy ( ⫽ 0.8) was
Table 2. Prevalence of significant cervical ectopy (ⱖ10% of the cervix) among 694 women (aged 35– 65 years) in Khayelitsha, South Africa, by age (in years) and parity Percentage of women with significant cervical ectopy (number in each cell)
Parity None 1 or 2 3 or 4 ⱖ5 Total
35
36–39
40–44
45–49
50–65
Total
0 (10) 37 (75) 42 (69) 29 (7) 23 (161)
8 (12) 41 (46) 43 (82) 53 (36) 25 (176)
0 (2) 10 (21) 40 (61) 48 (67) 21 (151)
0 (3) 22 (9) 29 (28) 33 (51) 13 (91)
0 (0) 22 (9) 17 (23) 21 (83) 16 (115)
4 (27) 33 (160) 38 (263) 36 (244) 35 (694)
366
Kuhn et al.
Contraception 1999;59:363–367
Table 3. Prevalence of disruption of cervical integrity (visible ulceration, inflammation, or atrophy) and significant cervical ectopy (ⱖ10% of the cervix) by Depo-Provera use among age- and parity-specific strata
Age-specific strata Age 35–39 years: Age 40–49 years: Age 50⫹ years: Parity-specific strata Parity 0–2: Parity 3⫹:
N
% With cervical disruption
% With significant ectopy
Current Depo-Provera users Past Depo-Provera users Never users Current Depo-Provera users Past Depo-Provera users Never users Current Depo-Provera users Past Depo-Provera users Never users
94 189 54 26 172 45 1 47 67
37.2 26.5 33.3 38.5 37.8 42.2 1/1 53.2 65.7
44.7 41.8 22.2 46.2 37.8 22.2 0/1 25.5 16.4
Current Depo-Provera users Past Depo-Provera users Never users Current Depo-Provera users Past Depo-Provera users Never users
29 99 59 92 308 107
27.6 20.2 28.8 41.3 38.6 59.8
31.0 37.4 13.6 48.9 38.3 23.4
excellent, but that of endocervical inflammation was only moderate ( ⫽ 0.5). Reliability of the diagnosis of any evidence of cervical disruption combined (either ulceration, inflammation or atrophy) was excellent ( ⫽ 0.8).
Discussion The underlying prevalence of visible disruption of the cervical epithelium was very high in this sample of African women. The high prevalence may be associated with frequent (and often untreated) lower genital tract infections in this population. We have, subsequent to this analysis, started testing all women recruited into the ongoing study for trichomoniasis, gonorrhea, and chlamydia. Culture-confirmed trichomonas is detectable in approximately 20% of women; chlamydia detectable by molecular methods in approximately 4%, and gonorrhea in 3% (data unpublished). Our scoring of macroscopically visible disruption of the cervical epithelium and cervical inflammation is not intended as a proxy measurement of infection; rather, it is intended as a measure of a phenomenon that is conceptually distinct, although related. Not all infections result in epithelial disruption severe enough to become clinically detectable, and disruption may have noninfectious causes such as trauma. Epithelial disruption may also be related to the use of intravaginal agents. The use of herbs and other agents to dry the vagina before sexual intercourse or to self-treat infections is a practice found to be common in other studies in southern Africa.7,8 The prevalence of epithelial disruption also in-
creased with age and parity in this analysis. The reasons for the association with parity are unclear but may reflect sexual activity and exposure to sexually transmitted agents. The association with age may reflect lower levels of gynecologic health care in older women or enhanced susceptibility of mildly atrophic epithelium to disruption by infectious agents. Older postmenopausal women were found to be at higher risk of HIV transmission in one early, discordant partner study.9 Against this high background rate of epithelial disruption, we did not observe any increase in the prevalence of visible disruption (ulceration, inflammation, or atrophy) of endocervical or ectocervical epithelium associated with current or past use of Depo-Provera. One possible explanation for the lack of an observed increase is that subtle effects of hormonal contraceptives may be difficult to discern against the background of a high prevalence of underlying epithelial lesions. In addition, the effects of Depo-Provera may not be apparent if duration of use is short. Data on duration of use was not collected for the women described here; however, it is likely to be of medium duration because, among subsequent women recruited from the same site, the mean duration of Depo-Provera use among current users was 50 months (66% reported use for ⬎3 years). Further investigation of women with longer use may be necessary to detect possible cervical changes associated with Depo-Provera. We also observed no effects of Depo-Provera on the detection of significant cervical ectopy. In the current study, ectopy was observed to decline with age and
Contraception 1999;59:363–367
increase with parity adjusted for age consistent with its expected biological maturation processes. Too few women were current users of estrogen-containing oral contraceptives in this analysis to examine this association with ectopy. Inasmuch as the columnar epithelium of cervical ectopy may be more easily traumatized than mature squamous epithelium, it has been suggested that ectopy may be a potential risk factor for heterosexual HIV transmission.10 Ectopy has also been proposed as a possible explanation of the apparent increased susceptibility of young women to acquire HIV infection through heterosexual contact. Disruption of the integrity of the cervical mucosa has been proposed as one of the mechanisms underlying increased HIV transmission with other sexually transmitted diseases, an association for which there is now unequivocal evidence. Therefore, the high underlying prevalence of visible disruption of the cervical epithelium in this sample of middle-aged African women is quite concerning. Although disruption of the integrity of cervical epithelium may not be required for HIV transmission to occur, it is likely to increase susceptibility to infection.11 It may also be associated with increased infectivity. Associations between cervical shedding of HIV and cervical inflammation have been observed in some studies.12 Our failure to observe associations between current and past use of Depo-Provera and disruption of the cervical epithelium should not be taken as evidence against an association between the use of DepoProvera and heterosexual transmission of HIV. Associations between Depo-Provera use and HIV transmission may be independent of any disruptive effects on the cervical epithelium. Instead, associations may be due to changes in the cellular composition of the epithelium or in mucosal immunology rather than to changes in epithelial thickness making it more susceptible to disruption. However, our findings do cast doubt on the notion that simple epithelial thinning may account for putative effects of Depo-Provera. Direct measurement of cervical integrity in studies of heterosexual HIV transmission may be informative to clarify mechanisms of transmission. Practical, valid, and reliable methods for measuring cervical epithelial integrity may also be important for studies evaluating the safety and efficacy of microbicides for prevention of HIV transmission.13 Use of 35-mm photographs to document the presence of cervical lesions is noninvasive and has excellent reliability.
Cervical Epithelium Integrity and Depo-Provera
367
The validity of this method needs to be further investigated.
Acknowledgments We thank Fred Kosteki of National Testing Laboratories, St. Louis, MO, who provided training and processed all photographs. This study was supported in part through grants from AVSC International and the Gates Foundation.
References 1. Marx PA, Spira AI, Gettie A, et al. Progesterone implants enhance SIV vaginal transmission and early virus load. Nature Med 1996;2:1084 –9. 2. Daly CC, Helling-Giese GE, Mati JK, Hunter DJ. Contraceptive methods and the transmission of HIV: implications for family planning. Genitourin Med 1994;70: 110 –7. 3. Stephenson JM. Systematic review of hormonal contraception and risk of HIV transmission: when to resist meta-analysis. AIDS 1998;12:545–53. 4. Denny L, Kuhn L, Risi L, et al. Two-stage cervical cancer screening: an alternative for resource-poor settings; submitted. 5. World Health Organization. Manual for the Standardization of Colposcopy for the Evaluation of Vaginally Administered Products. Geneva: World Health Organization, 1994. 6. Gilmour E, Ellerbrock TV, Koulos J, et al. Measuring cervical ectopy: direct visual assessment versus computerized planimetry. Am J Obstet Gynecol 1997;176: 108 –11. 7. Runganga A, Pitts M, McMaster J. The use of herbal and other agents to enhance sexual experience. Soc Sci Med 1992;35:1037– 42. 8. Dallabetta GA, Miotti PG, Chiphangwi JD, Liomba G, Canner JK, Saah AJ. Traditional vaginal agents: use and association with HIV infection in Malawian women. AIDS 1995;9:293–7. 9. Peterman TA, Stoneburner RL, Allen JR, Jaffe HW, Curran, JW. Risk of human immunodeficiency virus transmission from heterosexual adults with transfusion-associated infections. JAMA 1988;259:55– 8. 10. Plourde PJ, Pepin J, Agoki E, et al. Human immunodeficiency virus type 1 seroconversion in women with genital ulcers. J Infect Dis 1994;170:313–7. 11. Alexander NJ. Sexual transmission of human immunodeficiency virus: virus entry into the male and female genital tract. Fertil Steril 1990;54:1–18. 12. Kreiss J, Willerford DM, Hensel M, et al. Association between cervical inflammation and cervical shedding of human immunodeficiency virus DNA. J Infect Dis 1994;170:1597–1601. 13. Elias CJ, Heise LL. Challenges for the development of female-controlled vaginal microbicides. AIDS 1994;8: 1–9.
Prevalence of Visible Disruption of Cervical Epithelium and Cervical Ectopy in African Women Using Depo-Provera威 Louise Kuhn,* Lynette Denny,† Amy E. Pollack,‡ and Thomas C. Wright§
Associations between Depo-Provera威 (injectable, progesterone-only contraceptive) use and visible disruption of cervical epithelium and cervical ectopy were investigated using data collected as part of a cervical cancer screening study in periurban Cape Town, South Africa. Women were interviewed about their contraceptive use, and underwent a gynecologic examination that included two 35-mm photographs of the cervix after application of 5% acetic acid. Photographs of 723 subjects were reviewed (blind to clinical information and using systematic criteria developed before review) for evidence of atrophy and epithelial disruption, including inflammation and ulceration. The percentage of the cervix covered with columnar epithelium (ectopy) was also estimated from the photographs. A random sample of 85 photographs was reviewed again for reliability. A total of 121 current users of Depo-Provera were no more likely to have evidence of epithelial disruption (38%) than 574 nonusers (39%), odds ratio (OR) ⫽ 1.37, 95% CI: 0.89 –2.11 adjusting for age and parity. The prevalence of significant ectopy (columnar epithelium covering ⬎10% of the cervix) was also no different among current Depo-Provera users (OR ⫽ 1.22, 95% CI: 0.80 –1.86 adjusting for age and parity). Reliability of visual scoring of epithelial disruption and ectopy was excellent ( ⫽ 0.8). Although the underlying prevalence of visible disruption of cervical epithelium was very high, current use of DepoProvera was not associated with increased prevalence of visible disruption of the cervical epithelium or with ectopy in this sample of African women. CONTRACEPTION 1999; 59:363–367 © 1999 Elsevier Science Inc. All rights reserved.
*Gertrude H. Sergievsky Center, College of Physicians and Surgeons, Columbia University, New York, New York; †Department of Obstetrics & Gynaecology, University of Cape Town, Cape Town, South Africa; ‡AVSC International, New York, New York ; and §Department of Pathology, College of Physicians and Surgeons, Columbia University, New York, New York Name and address for correspondence: L Kuhn, Columbia University, Sergeivsky Center, 630 W 168th Street, New York, NY 10032; Tel.: (212) 305-2398; Fax: (212) 305-2426; e-mail: [email protected] Submitted for publication March 30, 1999 Revised June 7, 1999 Accepted for publication June 7, 1999
© 1999 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
KEY WORDS:
hormonal contraception, progesterone, cervical ectopy, cervical inflammation, heterosexual HIV transmission, Africa
Introduction
A
n experimental study of female rhesus macaques found that subcutaneous progesterone implants increased the efficiency of vaginal simian immunodeficiency virus (SIV) transmission 7.7-fold. When pretreated with progesterone implants, 14 of 18 macaques acquired infection after vaginal inoculation with SIV compared with one of 10 macaqes pretreated with placebo implants.1 Thickness of vaginal squamous epithelium was found to be reduced substantially among three treated macaques compared with three control macaques in a subsequent experiment.1 Based on these findings, the authors concluded that enhancement of vaginal transmission of SIV associated with exogenous progesterone is most likely due to thinning of vaginal epithelium. A thinner epithelium would be expected to be likely to show evidence of loss of mucosal integrity and to be more susceptible to trauma. Foci of ulceration might allow HIV easier passage to susceptible target cells such as Langerhans cells, dendritic cells, or lymphocytes within lower subepithelial levels. These findings have raised considerable concern, as long-acting injectable human contraceptives have synthetic forms of progesterone as their active ingredients. One such contraceptive is depot-medroxyprogesterone acetate, marketed as Depo-Provera威, which is widely used in parts of southern Africa where the HIV epidemic is at its worst. Although observational epidemiological studies have not consistently demonstrated increased susceptibility to HIV transmission with the use of hormonal contraceptives, few studies have investigated progesterone-only contraceptives specifically, and among these few, most have significant methodological limitations.2,3 ISSN 0010-7824/99/$20.00 PII S0010-7824(99)00049-9
364
Kuhn et al.
Contraception 1999;59:363–367
It is generally assumed, but has not been directly shown, that the use of Depo-Provera in women leads to thinning of the vaginal and cervical epithelium with possible secondary ulceration and erosion. In this study, we investigated whether or not women who used Depo-Provera are more likely to have visible evidence of epithelial thinning and trauma than women not using it. Our purpose was to complement studies of heterosexual HIV transmission.
Materials and Methods The data were collected as part of an ongoing cervical cancer screening study that is being conducted in a large periurban, informal settlement (Khayelitsha) outside of Cape Town, South Africa. Women aged 35– 65 years undergo a gynecologic examination that includes several cervical cancer screening tests, including two 35-mm color photographs of the cervix taken by a trained clinician after application of 5% acetic acid. A clinical history is also taken that includes questions about current and past use of birth control. The methods of the larger study are described in more detail elsewhere.4 All patients were recruited with the approval of the Human Subjects’ Institutional Review Board of Columbia University and Research Ethics Committee of the University of Cape Town. Cervical photographs from the first 723 participants were reviewed for evidence of cervical epithelial disruption. Specific criteria were developed before review modifying available preliminary guidelines for standardized colposcopy.5 All cervical photographs were reviewed blind to any clinical information. The slides were reviewed for macroscopically visible evidence of ulceration, inflammation or atrophy. Ulceration was defined as sharply demarcated areas of disrupted epithelium with bleeding. Inflammation was defined as erythematous and edematous changes involving endocervical type columnar epithelium (endocervicitis) or ectocervical type squamous epithelium (ectocervicitis). Ectocervicitis was further classified as either grade 1 (less severe), characterized by focal scattered vascular punctations, similar to the changes of “colpitis macularis” observed in women with trichomoniasis, or grade 2 (more severe), characterized by larger confluent regions of erythema that frequently appeared “punched out” or as shallow erosions (Figure 1). Atrophy was defined as apparent thinning of the cervical epithelium, often accompanied by petechial hemorrhages or diffuse whitening of the epithelium. The percentage of the visible portion of the cervix covered with columnar epithelium (cervical ectopy) was also estimated from each photograph as previously described.6 Ectopy covering 10%
Figure 1. Representative examples from the series of cervical photographs are displayed illustrating the four conditions visually scored. A. Atrophy. Cervical epithelium appears thin, petechial hemorrhages are observed, and there is diffuse whitening of the epithelium. B. Endocervical inflammation. Columnar epithelium appears erythematous and edematous, cell definition is poor, and there may be yellowing in color. C. Grade 1 ectocervical inflammation. Focal scattered vascular punctations are seen in the squamous epithelium. D. Grade 2 ectocervical inflammation. Larger confluent regions of erythema are seen; they appear “punched out” or as shallow erosions.
or more of the cervix was classified as significant ectopy. A random sample of 85 photographs was reviewed again to measure reliability. The prevalence of cervical changes among women currently using Depo-Provera were compared to those who had used Depo-Provera in the past but were not current users, and to those who had never used it. Differences were tested using 2 tests. Differences between the groups were adjusted for age, parity, current and past numbers of sexual partners, marital status, age at first intercourse, and cigarette smoking, using multiple logistic regression. Reliability of diagnosis on repeat review was described using the coefficient of agreement.
Results Photographs from 695 (96%) of the 723 study participants were acceptable for evaluation. Among the 695 women, 7 (1%) had areas of ulceration, 89 (13%) had visible endocervical inflammation, 170 (25%) visible ectocervical inflammation (110 [16%] grade 1 ectocervicitis and 58 [8%] grade 2 ectocervicitis) and 56 (8%) had atrophic changes. Overall, 268 (38%, 95% CI
Contraception 1999;59:363–367
Cervical Epithelium Integrity and Depo-Provera
365
Table 1. Prevalence of disruption of cervical integrity (visible ulceration, inflammation, or atrophy) among 694 women (aged 35– 65 years) in Khayelitsha, South Africa, by age (in years) and parity Percentage of women with visible disruption of cervical epithelium (number in each cell)
Parity None 1 or 2 3 or 4 ⱖ5 Total
35
36–39
40–44
45–49
50–65
Total
10 (10) 21 (75) 32 (69) 43 (7) 26 (161)
17 (12) 8 (46) 34 (82) 42 (36) 34 (176)
0 (2) 29 (21) 36 (61) 40 (67) 36 (151)
0 (3) 44 (9) 39 (28) 45 (51) 42 (91)
0 (0) 22 (9) 35 (23) 72 (83) 61 (115)
11 (27) 26 (160) 35 (263) 52 (244) 38 (694)
34 – 42) women had one or more of these conditions. Representative examples of photographs from women with each of these conditions are displayed in Figure 1. Disruption of cervical epithelium increased with age and parity (Table 1). The percentage of women diagnosed with visible epithelial disruption increased from 26% in women 35 years old to 61% in women ⬎50 years old; and from 11% in women with no children to 52% in women with five or more children (Table 1). In a logistic regression model including both age and parity, both characteristics were significant predictors of epithelial disruption (for each year of age, the odds ratio [OR] ⫽ 1.03, 95% confidence interval [CI]: 1.01–1.06, and for each additional child, OR ⫽ 1.21, 95% CI: 1.11–1.31). Presence of significant ectopy (involving ⱖ10% of the cervix) decreased with age but increased with parity (Table 2). Each year of age was associated with a decreased odds of ectopy (OR ⫽ 0.92, 95% CI: 0.89 – 0.94 adjusting for parity) and each additional child with an increased odds (OR ⫽ 1.21, 95% CI: 1.11–1.32 adjusting for age). The majority (83%) of women reported some past or current use of birth control: 76% reported using Depo-Provera, 38% oral contraceptives, 3% condoms, and 3% other methods of birth control in the past or present. Current use of Depo-Provera was reported by
17% of women, current use of oral contraceptives by 2%, and condoms by 0.6% (4 women). Visible ulceration, inflammation or atrophy was not more common among women reporting current use of Depo-Provera. Among 121 current DepoProvera users, 38% had at least one of these conditions, compared with 34% of 408 reporting only past use of Depo-Provera, and 49% of 166 women who had never used it. In addition, no consistent increases in epithelial disruption were identified in users of DepoProvera stratified by age or parity (Table 3). After adjusting for age and parity as continuous variables, there continued to be no differences in the odds of epithelial disruption (OR ⫽ 1.37 95% CI: 0.89 –2.11) between current users of Depo-Provera and nonusers. Further adjustment for current and past numbers of sexual partners, marital status, age at first intercourse, and cigarette smoking did not change the magnitude of these associations. The percentage of women with significant ectopy was higher in current and past users of Depo-Provera in unadjusted analyses, but no trends persisted after adjustment for age and parity as continuous variables (OR ⫽ 1.22 95% CI: 0.80 –1.86). Within the sample of 85 cervical photographs that were re-reviewed, reliability of the diagnosis of ectocervical inflammation ( coefficient of agreement [] ⫽ 0.9), atrophy ( ⫽ 0.8), and ectopy ( ⫽ 0.8) was
Table 2. Prevalence of significant cervical ectopy (ⱖ10% of the cervix) among 694 women (aged 35– 65 years) in Khayelitsha, South Africa, by age (in years) and parity Percentage of women with significant cervical ectopy (number in each cell)
Parity None 1 or 2 3 or 4 ⱖ5 Total
35
36–39
40–44
45–49
50–65
Total
0 (10) 37 (75) 42 (69) 29 (7) 23 (161)
8 (12) 41 (46) 43 (82) 53 (36) 25 (176)
0 (2) 10 (21) 40 (61) 48 (67) 21 (151)
0 (3) 22 (9) 29 (28) 33 (51) 13 (91)
0 (0) 22 (9) 17 (23) 21 (83) 16 (115)
4 (27) 33 (160) 38 (263) 36 (244) 35 (694)
366
Kuhn et al.
Contraception 1999;59:363–367
Table 3. Prevalence of disruption of cervical integrity (visible ulceration, inflammation, or atrophy) and significant cervical ectopy (ⱖ10% of the cervix) by Depo-Provera use among age- and parity-specific strata
Age-specific strata Age 35–39 years: Age 40–49 years: Age 50⫹ years: Parity-specific strata Parity 0–2: Parity 3⫹:
N
% With cervical disruption
% With significant ectopy
Current Depo-Provera users Past Depo-Provera users Never users Current Depo-Provera users Past Depo-Provera users Never users Current Depo-Provera users Past Depo-Provera users Never users
94 189 54 26 172 45 1 47 67
37.2 26.5 33.3 38.5 37.8 42.2 1/1 53.2 65.7
44.7 41.8 22.2 46.2 37.8 22.2 0/1 25.5 16.4
Current Depo-Provera users Past Depo-Provera users Never users Current Depo-Provera users Past Depo-Provera users Never users
29 99 59 92 308 107
27.6 20.2 28.8 41.3 38.6 59.8
31.0 37.4 13.6 48.9 38.3 23.4
excellent, but that of endocervical inflammation was only moderate ( ⫽ 0.5). Reliability of the diagnosis of any evidence of cervical disruption combined (either ulceration, inflammation or atrophy) was excellent ( ⫽ 0.8).
Discussion The underlying prevalence of visible disruption of the cervical epithelium was very high in this sample of African women. The high prevalence may be associated with frequent (and often untreated) lower genital tract infections in this population. We have, subsequent to this analysis, started testing all women recruited into the ongoing study for trichomoniasis, gonorrhea, and chlamydia. Culture-confirmed trichomonas is detectable in approximately 20% of women; chlamydia detectable by molecular methods in approximately 4%, and gonorrhea in 3% (data unpublished). Our scoring of macroscopically visible disruption of the cervical epithelium and cervical inflammation is not intended as a proxy measurement of infection; rather, it is intended as a measure of a phenomenon that is conceptually distinct, although related. Not all infections result in epithelial disruption severe enough to become clinically detectable, and disruption may have noninfectious causes such as trauma. Epithelial disruption may also be related to the use of intravaginal agents. The use of herbs and other agents to dry the vagina before sexual intercourse or to self-treat infections is a practice found to be common in other studies in southern Africa.7,8 The prevalence of epithelial disruption also in-
creased with age and parity in this analysis. The reasons for the association with parity are unclear but may reflect sexual activity and exposure to sexually transmitted agents. The association with age may reflect lower levels of gynecologic health care in older women or enhanced susceptibility of mildly atrophic epithelium to disruption by infectious agents. Older postmenopausal women were found to be at higher risk of HIV transmission in one early, discordant partner study.9 Against this high background rate of epithelial disruption, we did not observe any increase in the prevalence of visible disruption (ulceration, inflammation, or atrophy) of endocervical or ectocervical epithelium associated with current or past use of Depo-Provera. One possible explanation for the lack of an observed increase is that subtle effects of hormonal contraceptives may be difficult to discern against the background of a high prevalence of underlying epithelial lesions. In addition, the effects of Depo-Provera may not be apparent if duration of use is short. Data on duration of use was not collected for the women described here; however, it is likely to be of medium duration because, among subsequent women recruited from the same site, the mean duration of Depo-Provera use among current users was 50 months (66% reported use for ⬎3 years). Further investigation of women with longer use may be necessary to detect possible cervical changes associated with Depo-Provera. We also observed no effects of Depo-Provera on the detection of significant cervical ectopy. In the current study, ectopy was observed to decline with age and
Contraception 1999;59:363–367
increase with parity adjusted for age consistent with its expected biological maturation processes. Too few women were current users of estrogen-containing oral contraceptives in this analysis to examine this association with ectopy. Inasmuch as the columnar epithelium of cervical ectopy may be more easily traumatized than mature squamous epithelium, it has been suggested that ectopy may be a potential risk factor for heterosexual HIV transmission.10 Ectopy has also been proposed as a possible explanation of the apparent increased susceptibility of young women to acquire HIV infection through heterosexual contact. Disruption of the integrity of the cervical mucosa has been proposed as one of the mechanisms underlying increased HIV transmission with other sexually transmitted diseases, an association for which there is now unequivocal evidence. Therefore, the high underlying prevalence of visible disruption of the cervical epithelium in this sample of middle-aged African women is quite concerning. Although disruption of the integrity of cervical epithelium may not be required for HIV transmission to occur, it is likely to increase susceptibility to infection.11 It may also be associated with increased infectivity. Associations between cervical shedding of HIV and cervical inflammation have been observed in some studies.12 Our failure to observe associations between current and past use of Depo-Provera and disruption of the cervical epithelium should not be taken as evidence against an association between the use of DepoProvera and heterosexual transmission of HIV. Associations between Depo-Provera use and HIV transmission may be independent of any disruptive effects on the cervical epithelium. Instead, associations may be due to changes in the cellular composition of the epithelium or in mucosal immunology rather than to changes in epithelial thickness making it more susceptible to disruption. However, our findings do cast doubt on the notion that simple epithelial thinning may account for putative effects of Depo-Provera. Direct measurement of cervical integrity in studies of heterosexual HIV transmission may be informative to clarify mechanisms of transmission. Practical, valid, and reliable methods for measuring cervical epithelial integrity may also be important for studies evaluating the safety and efficacy of microbicides for prevention of HIV transmission.13 Use of 35-mm photographs to document the presence of cervical lesions is noninvasive and has excellent reliability.
Cervical Epithelium Integrity and Depo-Provera
367
The validity of this method needs to be further investigated.
Acknowledgments We thank Fred Kosteki of National Testing Laboratories, St. Louis, MO, who provided training and processed all photographs. This study was supported in part through grants from AVSC International and the Gates Foundation.
References 1. Marx PA, Spira AI, Gettie A, et al. Progesterone implants enhance SIV vaginal transmission and early virus load. Nature Med 1996;2:1084 –9. 2. Daly CC, Helling-Giese GE, Mati JK, Hunter DJ. Contraceptive methods and the transmission of HIV: implications for family planning. Genitourin Med 1994;70: 110 –7. 3. Stephenson JM. Systematic review of hormonal contraception and risk of HIV transmission: when to resist meta-analysis. AIDS 1998;12:545–53. 4. Denny L, Kuhn L, Risi L, et al. Two-stage cervical cancer screening: an alternative for resource-poor settings; submitted. 5. World Health Organization. Manual for the Standardization of Colposcopy for the Evaluation of Vaginally Administered Products. Geneva: World Health Organization, 1994. 6. Gilmour E, Ellerbrock TV, Koulos J, et al. Measuring cervical ectopy: direct visual assessment versus computerized planimetry. Am J Obstet Gynecol 1997;176: 108 –11. 7. Runganga A, Pitts M, McMaster J. The use of herbal and other agents to enhance sexual experience. Soc Sci Med 1992;35:1037– 42. 8. Dallabetta GA, Miotti PG, Chiphangwi JD, Liomba G, Canner JK, Saah AJ. Traditional vaginal agents: use and association with HIV infection in Malawian women. AIDS 1995;9:293–7. 9. Peterman TA, Stoneburner RL, Allen JR, Jaffe HW, Curran, JW. Risk of human immunodeficiency virus transmission from heterosexual adults with transfusion-associated infections. JAMA 1988;259:55– 8. 10. Plourde PJ, Pepin J, Agoki E, et al. Human immunodeficiency virus type 1 seroconversion in women with genital ulcers. J Infect Dis 1994;170:313–7. 11. Alexander NJ. Sexual transmission of human immunodeficiency virus: virus entry into the male and female genital tract. Fertil Steril 1990;54:1–18. 12. Kreiss J, Willerford DM, Hensel M, et al. Association between cervical inflammation and cervical shedding of human immunodeficiency virus DNA. J Infect Dis 1994;170:1597–1601. 13. Elias CJ, Heise LL. Challenges for the development of female-controlled vaginal microbicides. AIDS 1994;8: 1–9.