SEARCH

SEARCH BY CITATION

Keywords:

  • HIV;
  • cervical;
  • pathology;
  • screening;
  • diagnosis

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The objectives of this prospective multicentre international cohort study are to describe the characteristics of a cohort of HIV-1 positive women and determine the best management system by comparing cervical pathology according to results of cytology, colposcopy and human papillomavirus (HPV) testing at baseline and throughout follow-up. A. Cohorts of known HIV-positive women were recruited from 6 hospital-based European centres and a community-based South African centre. Following registration, women were reviewed every 6 months to undergo cervical surveillance including cytology, colposcopy, histopathology and HPV testing, using the HPV hybrid capture assay. Independent risk factors for the incidence of cytological abnormality and acquisition/clearance of HPV infection during follow up were identified. A total of 1,534 women were recruited, 400 of which were from South Africa. At baseline, among European women, 66% had normal cytology and half were HPV negative and among South African women, 45% had normal cytology and one third (32%) were HPV negative. The sensitivity of cytology (≥ASCUS) matched with that of colposcopy to detect CIN2+. Rate of detection of high grade CIN at 2 years was similar in European and South African women (11 and 9.3%, respectively). Cytology and HPV testing alone were each sufficiently sensitive as a screening test at 2 yearly intervals. Our data confirm the high prevalence of low-grade cytological abnormalities and high-risk HPV infection. Cytology appears to be sufficient for cervical surveillance, with HPV testing being less specific with poor positive predictive value. There appears to be no additional benefit from routine colposcopy. © 2007 Wiley-Liss, Inc.

It is now widely acknowledged that human papillomavirus (HPV) infection is an essential factor in the development of cervical intraepithelial neoplasia (CIN) and cancer of the cervix. It has also become clear that HIV-positive women carry an increased risk of persistent genital HPV infection and therefore have an increased risk of HPV-associated lower genital tract neoplasia, particularly CIN. Higher incidence and prevalence of CIN have been reported in HIV-1 positive women.1, 2, 3, 4, 5

The HIV/AIDS pandemic continues to spread unabated in many parts of the world and there is now an increasing number of HIV-infected women in European countries largely due to increased heterosexual transmission (UNAIDS report, Dec. 2004, www.unaids.org). Significant progress has been made in the development of antiretroviral therapy (HAART), with 23 at last count licensed antiretroviral drugs now in use that target HIV replication. With the advent and widespread use of combination HAART, AIDS-associated mortality and morbidity have fallen dramatically in industrialized countries.6, 7 In developed healthcare systems where combination HAART is available, the prospect for long-term survival from HIV/AIDS has been demonstrated and consequently preventive health care strategies become relevant. For HIV infected women, this means consideration of the risk to the cervix. In resource-poor healthcare settings the nonavailability of combination HAART means that women are at risk of dying early from AIDS and, although the risk of cervical cancer is significantly increased, death from opportunistic infections is likely to precede cervical pathology.8

At present, there is no clear consensus as how best to prevent and manage cervical disease in HIV-positive women. This is in part due to the lack of reliable information regarding the clinical effectiveness of cytology, colposcopy and HPV testing in HIV infected women.9, 10 Furthermore much of the work that has been done to date precedes the widespread use of combination HAART and the impact of this on the natural history of cervical disease in HIV infected women is still debated.11 There are also wide differences in HIV-infected cohorts in their ethnicity and behavioural characteristics that may influence outcomes in published studies. This international prospective study sought to determine the risks to the cervix in HIV infected women, in a European cohort with access to combination HAART as well as a cohort from South Africa without access to combination HAART. The overall aim of the project was to develop an evidence base with which clinical guidelines regarding cervical surveillance and management could be developed.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Clinical protocol

Seven centres involving specialist clinics responsible for the gynaecological management of women with HIV formed a collaborative group; Management of Abnormal Cytology in HIV-1 (MACH-1) infected women. HIV infected women either already under surveillance or newly diagnosed were recruited from 6 European centres between April 2000 and April 2003. These included Dublin, Edinburgh, London, Milan, Paris and Warsaw. An additional cohort was recruited from Cape Town in South Africa. Women gave informed consent to take part in the study, each centre having obtained local ethical approval. Following recruitment, women were interviewed using a standardised questionnaire on ethnic characteristics, age, marital status, tobacco use, route of HIV infection, age at first intercourse and total number of sexual partners. If heterosexually active, the number of male partners was recorded with frequency of condom use. Any history of previous cervical surveillance or treatment was also recorded. The prospective protocol involved 6 monthly cervical cytology, colposcopy with biopsy if clinically indicated and HPV testing. Cytology and histology were reported at each centre. Cytology was classified according to the Bethesda reporting system: normal, atypical squamous cells of undetermined significance (ASCUS), low-grade squamous intraepithelial lesions (SIL), high-grade SIL and cancer. Classification for colposcopy was normal, low grade and high grade. Histology was reported as normal, CIN1, CIN2–3 or invasive cancer. The most recent CD4+ cell counts and HIV RNA plasma viral load values within 6 months of the gynaecological examinations were obtained from clinical records. Data were collected into standardised report forms locally and were registered electronically at a central site supervised by a data manager in the data centre in Manchester. HPV testing was performed using the HPV hybrid capture II (HC2) assay (Digene Corporation, Gaithersburg, MD 20878).

Samples for HPV testing

HPV testing was performed in a single HPV laboratory for European women and in Cape Town for South African women. Both conventional and liquid-based cytology (LBC) specimens were used for cytology in different centres. Where conventional smears were standard practice, a second cervix brush sample was taken, rinsed in PreservCyt® medium which was sent for HPV testing. Where LBC samples were taken, residual sample after cytology was used for HPV testing. Samples were packaged in biocontainers and sent to Edinburgh by courier at room temperature approximately once per month. All transport regulations appropriate for each centre were closely followed.

Before HC2 testing, LBC samples required processing using the Digene Sample Conversion Kit. A 4 ml volume of cell suspension in PreservCyt fluid was mixed with 0.4 ml of sample conversion buffer, centrifuged and resuspended in a mix of specimen transport medium/denaturation reagent, transferred to a sterile vial and denatured at 65°C for 45 min. Denatured samples were stored at −20°C until batch testing was performed by HC2, starting at the hybridisation stage.

Hybrid capture assay

The hybrid capture assay system is a solution-based molecular hybridisation assay for qualitative detection of high-risk (HR) and low-risk HPVs using 2 separate RNA probe cocktails to distinguish oncogenic and nononcogenic types. Target DNA hybridised to the RNA probe is captured and amplified by a chemiluminescence detection system. The emission of light is measured as relative light units (RLUs). The emission is proportional to the amount of target DNA present and samples were considered screen positive when they gave an RLU Index > 1 when measured against a 1 pg/ml HPV16 DNA positive control. The first generation test (HC1) was used for some early samples and involved hybridisation in tubes. The second generation test (HC2) uses microtitre plate wells. Most tests were carried out in a single centre using the second version for HR types according to manufacturers' instructions with the HR probe cocktail which contains 13 HPV types (HPV 16,18,31,33,35,39,45,51,52,56,58,59,68).

Statistical methods

Data from the baseline questionnaire and all test results were entered into a specially designed study database. Baseline data have been tabulated and row or column percentages have been calculated to facilitate comparisons between groups. Formal comparison, at baseline, between women with high-grade histology (CIN2+) and various demographic and HIV-related factors used logistic regression. Imputation was used to adjust for incomplete assessment of women at each test. First, we used logistic regression models to predict missing test results based on histology and the other findings available for that individual and adjusting for centre. Next we used the actual and imputed test result to predict the probability of high-grade disease when histology was not available. Sensitivity and specificity could then be calculated adjusting for incomplete use of all tests at some visits. Confidence intervals were calculated either using bootstrapping techniques or the exact binomial method, as appropriate. To incorporate the follow-up of our patients, we estimated, as earlier, the probability of having high-grade disease at each visit. We then used these estimates to perform a weighted Cox analysis and produce corresponding Kaplan–Meier plots. Bootstrapping was used to calculate confidence intervals for survival at yearly intervals.

Analysis was performed in Stata (Stata Corp., 2003, Statistical Software: Release 8.0, College Station, TX: Stata Corporation).

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Patient characteristics at baseline

Between April 2000 and May 2004 a total of 1,534 HIV-positive women were recruited to the study. The median age overall was 33 years, 35 for Europe and 29 for South Africa (range between centres: 29–37). Forty nine percent of the women were sub-Saharan African, including all women from the South African centre and between 0 and 74% from the European centres. Acquisition of HIV in Europe was self-reported to be by injecting drug use in 34%, by blood products in 1%, and heterosexual contact in 62%. In South Africa, acquisition was reported as heterosexual contact by 99% of individuals.

Overall, 71% of the women (range between centres 40–78%) reported being currently heterosexually active; 28% (0–64%) of them used a condom less than 20% of the time and 50% (6–86%) less than 90% of the time. The corresponding figures for Europe and South Africa were 25 and 34%, respectively, for less than 20% of the time and 55 and 39% for less than 90% of the time. HAART was not used in South Africa, whereas between 56 and 79% of women in the European centres were on monotherapy or combination HAART. CD4+ cell counts were considerably lower in South Africa: 39% of women had a count of less than 200 cells/l, compared with between 11 and 27% in the European centres. The majority of South African women (61%) had HIV RNA plasma loads in excess of 10,000 copies/ml compared with only 12% of European women on HAART and 40% of European women not on HAART.

Baseline cytology and colposcopy

Overall 43% of women self-reported a previous cervical cytological result, 33% in South Africa and 47% in Europe and 5% a previous diagnosis of CIN2–3 on histology.

Of the 1,425 women, 1,026 in Europe and 399 in South Africa, with an evaluable cytology at baseline, Pap smears were normal in 66% of European women and 45% of South-African women (p < 0.001). The proportion of ASCUS smears was 7% for both Europe and South Africa. The proportions for LSIL were 17 and 34% in Europe and South Africa, respectively (p < 0.001), reflecting the high prevalence of HPV infection. The rate of high-grade cytology, similar in Europe (10%) and in South Africa (13%), is around 5 times the expected rate in HIV-negative women.

Eight hundred and forty six women had a negative cytological diagnosis, 670 in Europe and 176 in South Africa. Overall 175 (20.7%) had an abnormality at colposcopy, 118 (17.6%) in Europe and 57 (32%) in South Africa although none of these women who went on to have a biopsy were found to have high-grade disease at histology.

Out of 1,397 women who had a colposcopic examination at baseline, 589 (42%) had a positive result, of which low grade was the more frequently seen (27% in Europe and 37% in South Africa).

HPV infection at baseline

Samples for HPV DNA were available in 1,130 women. Initially there were some logistical difficulties in getting HPV testing organised in 2 of the European centres for baseline samples. Among the 1,130 women, 56% had HR-HPV detected, 49% in Europe and 68% in South Africa. In Europe, cervical HPV infection was more prevalent in women not on combination HAART compared with those on HAART: 60% versus 46%, respectively (p = 0.001). The corresponding prevalence for the South-African group (no HAART) was 68%.

The prevalence of HPV infection by age, CD4 count, viral load and antiretroviral therapy, where available, are shown in Table I. Unlike immunocompetent women, a high prevalence of HPV in excess of 40% is maintained at all ages, with the highest prevalence of 75% amongst the younger age group of women less than 25 years of age. Additionally, there is a clear association between low CD4 counts and increasing prevalence of HPV (p < 0.001).

Table I. HPV, Cytology and Colposcopy Results at Baseline According to Age, CD4 Cell Count, Plasma HIV Load, Antiretroviral Treatment, Histology, Centre and HIV Diagnosis
 HPV resultCytology resultColposcopy result
Positive, n = 631≥ASCUS, n = 560≥Low grade, n = 589
  • Number positive (% of those tested who were positive).

  • 1

    Values in parentheses are percentages.

  • 2

    Length of HIV infection is taken as the length between the first positive HIV test and the first visit.

Age group
 <25104 (75)176 (43)84 (51)
 25–29153 (54)121 (40)141 (47)
 30–34173 (57)153 (41)164 (44)
 35–39125 (48)125 (37)134 (40)
 40–4442 (40)56 (40)45 (33)
 ≥4533 (53)28 (38)18 (24)
 Unknown1 (20)1 (20)3 (50)
 Total631 (56)560 (40)589 (42)
CD4 count
 <200204 (74)185 (56)180 (55)
 200–400211 (59)199 (45)196 (45)
 ≥400192 (44)160 (28)195 (35)
 Unknown24 (42)16 (24)18 (25)
 Total631 (56)560 (40)589 (42)
Viral load
 <500162 (42)156 (30)181 (37)
 500–10,000161 (60)139 (41)143 (42)
 >10,000243 (70)219 (55)209 (52)
 Unknown65 (50)46 (31)56 (36)
 Total631 (56)560 (40)589 (42)
HAART
 No390 (65)310 (46)329 (49)
 Yes228 (46)239 (34)249 (37)
 Unknown13 (39)11 (33)11 (31)
 Total631 (56)560 (40)589 (42)
Histology result
 <High grade63 (95)82 (100)93 (98)
 High grade240 (75)215 (59)295 (78)
 No biopsy (normal colp)206 (38)146 (21)0 (0)
 No biopsy (abnormal or no colp)122 (61)117 (42)201 (100)
 Total631 (56)560 (40)589 (42)
Centre
 Dublin52 (62)38 (36)50 (76)
 Edinburgh32 (41)15 (21)6 (8)
 London55 (54)95 (57)109 (52)
 Milan126 (45)84 (30)82 (29)
 Paris51 (46)94 (41)93 (43)
 South Africa269 (68)216 (55)204 (51)
 Warsaw46 (58)18 (11)45 (31)
 Total631 (56)560 (40)589 (42)
Length of HIV infection2
 <1 year164 (62)136 (46)145 (49)
 1–5 years225 (62)194 (43)228 (52)
 ≥5 years224 (48)214 (34)197 (32)
 Unknown18 (50)16 (41)19 (46)
 Total631 (56)560 (40)589 (42)
 In Europe
Positive, n = 362≥ASCUS, n = 344≥Low grade, n = 385
Age group
 <2534 (71)23 (27)37 (52)
 25–2960 (56)41 (27)61 (42)
 30–34110 (50)108 (37)117 (40)
 35–3996 (45)96 (33)109 (37)
 40–4436 (39)53 (41)44 (35)
 ≥4525 (49)22 (35)14 (22)
 Unknown1 (50)1 (50)3 (100)
 Total362 (49)344 (34)385 (39)
CD4 count
 <20078 (64)78 (45)86 (50)
 200–400118 (55)124 (40)125 (42)
 ≥400143 (42)127 (27)156 (35)
 Unknown23 (41)15 (23)18 (25)
 Total362 (49)344 (34)385 (39)
Viral load
 <500148 (43)144 (30)170 (38)
 500–10,00095 (56)91 (38)94 (40)
 >10,00079 (62)79 (44)83 (46)
 Unknown40 (42)30 (26)38 (31)
 Total362 (49)344 (34)385 (39)
HAART
 No121 (60)94 (34)125 (45)
 Yes228 (46)239 (34)249 (37)
 Unknown13 (39)11 (33)11 (31)
 Total362 (49)344 (34)385 (39)
Histology result
 <High grade46 (94)65 (100)77 (99)
 High grade55 (79)66 (57)114 (93)
 No biopsy (normal colp)147 (35)102 (18)0 (0)
 No biopsy (abnormal or no colp)114 (60)111 (41)194 (100)
 Total362 (49)344 (34)385 (39)
Centre
 Dublin52 (62)38 (36)50 (76)
 Edinburgh32 (41)15 (21)6 (8)
 London55 (54)95 (57)109 (52)
 Milan126 (45)84 (30)82 (29)
 Paris51 (46)94 (41)93 (43)
 Warsaw46 (58)18 (11)45 (31)
 Total362 (49)344 (34)385 (39)
Length of HIV infection2
 <1 year50 (61)46 (39)57 (49)
 1–5 years86 (52)83 (33)118 (48)
 ≥5 years208 (46)199 (33)191 (32)
 Unknown18 (50)16 (41)19 (46)
 Total362 (49)344 (34)385 (39)
 In South Africa
Positive, n = 269≥ASCUS, n = 216≥Low grade, n = 204
Age group
 <2570 (77)53 (58)47 (51)
 25–2993 (62)80 (53)80 (53)
 30–3463 (75)45 (56)47 (56)
 35–3929 (63)29 (63)25 (54)
 40–446 (55)3 (27)1 (9)
 ≥458 (73)6 (55)4 (36)
 Unknown0 (0)0 (0)0 (0)
 Total269 (68)216 (55)204 (51)
CD4 count
 <200126 (81)107 (69)94 (60)
 200–40093 (66)75 (56)71 (51)
 ≥40049 (49)33 (33)39 (38)
 Unknown1 (100)1 (100)0 (0)
 Total269 (68)216 (55)204 (51)
Viral load
 <50014 (33)12 (29)11 (26)
 500–10,00066 (66)48 (50)49 (49)
 >10,000164 (75)140 (64)126 (57)
 Unknown25 (71)16 (46)18 (51)
 Total269 (68)216 (55)204 (51)
HAART
 No269 (68)216 (55)504 (51)
 Yes0 (0)0 (0)0 (0)
 Unknown0 (0)0 (0)0 (0)
 Total269 (68)216 (55)204 (51)
Histology result
 <High grade17 (100)17 (100)16 (94)
 High grade185 (73)149 (60)181 (71)
 No biopsy (normal colp)59 (49)44 (37)0 (0)
 No biopsy (abnormal or no colp)8 (100)6 (75)7 (100)
 Total269 (68)216 (55)204 (51)
Length of HIV infection2
 <1 year114 (63)90 (51)88 (48)
 1–5 years139 (71)111 (58)110 (56)
 ≥5 years16 (76)15 (71)6 (29)
 Unknown0 (0)0 (0)0 (0)
 Total269 (68)216 (55)204 (51)

The rate of HPV positive in cytologically negative women was 35% overall, 32% in Europe and 41% in South Africa. These figures are around 2–4 times higher than would generally be expected in HIV-negative women over a similar age range.12, 13

Diagnostic accuracy of tests

Estimates of the diagnostic accuracy of cytology, colposcopy and HPV testing are shown with separate data for Europe and South Africa in Table II. Overall baseline, colposcopy, cytology and HPV results were available in 90.6, 91.7 and 73.7% of women, respectively. The corresponding figures for follow-up visits were 92.0, 88.7 and 90.9%. If cytology was classified as abnormal with a threshold of ASCUS or worse, the sensitivity and specificity to detect high grade disease at baseline of cytology was 100.0 and 65.4%, respectively, with a positive predictive value of 23.1%. Colposcopy had similar sensitivity and specificity of 98.0 and 62.9%, respectively, and a similar positive predictive value of 21.6%. HPV testing also had high sensitivity (91.3%) but had considerably worse specificity than both tests (47.7%) and poorer positive predictive value (15.4%).

Table II. Diagnostic Accuracy of Cytology, Colposcopy and HPV (HC2) Testing for Detecting CIN2+ at Baseline
TestObserved sensitivity1Imputed sensitivity2Imputed specificity2Imputed PPV2
  • Estimated performance (95% confidence interval).

  • 1

    Observed number with high-grade disease who test positive/observed number with high-grade disease tested.

  • 2

    Based on 144.5 cases (96 observed + 48.5 imputed) of high-grade disease.

  • 3

    Exact binomial confidence interval (one-sided) calculated for observed numbers as sensitivity is 100%.

  • 4

    Exact binomial of observed numbers calculated for upper confidence limit as 100.0%.

Cytology ≥ ASCUS
 Overall82/82100.0 (95.6, 100.0)365.4 (61.9, 68.1)23.1 (16.2, 27.3)
 Europe65/65100.0 (94.5, 100.0)372.5 (68.6, 75.8)32.1 (22.1, 39.2)
 South Africa17/17100.0 (80.5, 100.0)346.9 (41.9, 51.4)8.3 (4.5, 12.2)
Cytology ≥ LSIL
 Overall80/8293.3 (85.0, 99.7)472.8 (69.7, 75.2)26.3 (19.1, 30.9)
 Europe63/6590.3 (80.4, 99.6)479.6 (76.2, 82.4)36.5 (26.6, 43.9)
 South Africa17/17100.0 (80.5, 100.0)354.7 (49.5, 59.3)9.5 (5.2, 13.8)
Colposcopy ≥ low-grade
 Overall93/9598.0 (95.5, 99.7)462.9 (60.4, 67.7)21.6 (16.9, 25.9)
 Europe77/7898.5 (96.0, 100.0)467.9 (64.3, 74.4)28.5 (22.2, 35.8)
 South Africa16/1794.4 (82.0, 99.9)451.1 (46.0, 56.2)8.5 (4.5, 12.7)
HPV (HC2+ve)
 Overall63/6691.3 (82.9, 99.1)447.7 (44.2, 51.4)15.4 (11.5, 18.3)
 Europe46/4988.4 (79.0, 98.7)452.9 (49.1, 57.7)19.6 (14.2, 24.5)
 South Africa17/17100.0 (80.5, 100.0)333.4 (28.7, 38.0)6.7 (3.7, 9.9)
HPV (HC2+ve) and cytology ≥ ASCUS
 Overall59/6191.3 (82.9, 99.6)471.8 (68.1, 74.3)25.2 (18.3, 29.7)
 Europe42/4488.4 (79.0, 99.4)478.9 (74.8, 82.2)35.3 (25.7, 42.9)
 South Africa17/17100.0 (80.5, 100.0)353.0 (48.1, 57.7)9.2 (5.1, 13.5)
HPV (HC2+ve) and Cytology ≥ LSIL
 Overall58/6186.8 (76.0, 98.9)76.7 (73.6, 79.2)27.7 (21.0, 32.5)
 Europe41/4481.2 (69.7, 96.7)83.1 (79.7, 85.8)38.4 (28.8, 46.7)
 South Africa17/17100.0 (80.5, 100.0)359.4 (54.5, 64.0)10.5 (5.8, 15.3)

Other risk factors

In logistic regression analysis (Table III), there was no evidence of CD4 count or HIV RNA plasma load being associated with an increased relative risk of high-grade disease when adjusted for smoking and whether an individual was on combination HAART as well as baseline cytology, colposcopy and HPV results.

Table III. The Effect of Baseline CD4 Cell Count, Plasma HIV Load, Smoking and Haart on Rate of High Grade Disease During Follow-Up
Risk factorsUnivariable analysisMultivariable analysis1
Rate ratio95% CIp-value2Rate ratio95% CIp-value2
  • 1

    Adjusted for baseline cytology, colposcopy, HPV, HAART and smoking.

  • 2

    p-value: Wald test for trend if more than one group.

CD4 count (cells/ml)
 Overall
  <2001.00.061.00.34
  200–4000.860.50, 1.46 1.000.51, 1.96 
  ≥4000.590.34, 1.03 0.680.31, 1.50 
 Europe      
  <2001.00.131.00.22
  200–4000.840.43, 1.64 0.860.32, 2.30 
  ≥4000.610.31, 1.19 0.550.20, 1.53 
 South Africa
  <2001.00.2261.00.99
  200–4000.870.35, 2.18 1.150.44, 2.99 
  ≥4000.460.14, 1.55 0.930.26, 3.33 
Plasma HIV load (copies/ml)
 Overall      
  <5001.00.481.00.79
  500–10,0001.580.92, 2.74 1.290.55, 3.01 
  ≥10,0001.470.82, 2.63 1.070.42, 2.76 
 Europe      
  <5001.00.421.00.57
  500–10,0001.470.80, 2.72 1.030.37, 2.87 
  ≥10,0001.460.74, 2.86 1.400.40, 4.89 
 South Africa      
  <5001.00.9681.00.36
  500–10,0003.030.46, 20.14 2.060.30, 14.15 
  ≥10,0002.330.37, 14.62 1.150.18, 7.45 
HAART
 Overall      
  No1.01.0
  Yes1.090.63, 1.890.761.260.53, 2.980.60
 Europe      
  No1.01.0
  Yes1.120.63, 1.980.701.260.51, 3.110.62
 South Africa      
  No
  Yes
Smoking
 Overall      
  No1.01.0
  Yes1.260.77, 2.060.351.140.55, 2.370.72
 Europe      
  No1.01.0
  Yes1.420.82, 2.460.221.260.52, 3.100.61
 South Africa      
  No1.01.0
  Yes0.570.10, 3.360.5320.710.12, 4.250.71

Follow up

Overall 711 (62.7%) of European women and 341 (85%) of South-African women underwent at least one follow-up visit. Follow-up in South Africa was generally more thorough with a median number of follow-up visits of 5 (range 1–7), with a median length between visits of 6.0 months (IQR: 5.7–6.7 months), and median length of follow-up time from the first visit to the last of 30.1 months (IQR: 23.2–35.9 months). In Europe, the median number of follow-up visits was 2 (range 1–12) over a median total length of follow-up of 23.2 months (IQR 12.9–36.1 months). The median length between visits was 8.5 months (IQR 6.0–12.5 months) in Europe.

Cytology, colposcopy and histology

The number of women with normal baseline cytology who acquired an abnormal cytology result during follow-up is shown separately for Europe and South Africa in Table IV. The key finding is ∼3-fold greater incidence of abnormal cytology in South Africa compared with Europe.

Table IV. Detection of Abnormal Cytology during Follow up of Those Initially Cytology Negative with Some Adequate Cytological Follow-Up
ResultEuropeSouth Africa
  • 1

    Values in parentheses are percentages.

Initially cytology negative at baseline401151
Throughout follow-up
 ≥ASCUS72 (18.0)178 (51.7)
 ≥LSIL54 (13.5)63 (41.7)
Within 12 months follow-up
 ≥ASCUS28 (7.0)33 (21.9)
 ≥LSIL18 (4.5)21 (13.9)

Overall, of the 572 women without high grade disease at baseline who were initially tested as cytologically negative and had further follow-up, 150 (26.2%) showed evidence of ASCUS+ throughout follow-up on at least one smear. Similarly, of the 534 women who were initially tested as colposcopically negative and high-grade negative with subsequent follow-up, 163 (30.5%) showed evidence of low-grade disease or higher throughout follow-up on at least one colposcopy.

HPV results

Of 419 women who were initially HPV positive but did not have high grade disease and who had some form of subsequent follow-up, 157 (37.5%) became HPV negative. Of these 419 women, 29 (7%) developed high grade disease and all of these were still HPV positive at the time of being diagnosed with high grade disease. Of the 348 women who were HPV negative at baseline with no high-grade disease, who were subsequently tested, 129 (37.1%) became HPV positive but only 3 (0.85%) developed high grade disease (all of which were HPV positive at the time of high grade disease). The HPV status at baseline exerted a highly significant effect on the incidence of abnormal cytology at a later stage. Additionally, of these HPV negative women who had subsequent follow-up, 77 (22.1%) developed abnormal cytology compared with 293 out of 419 HPV positive women (69.9%).

Hazard ratios for developing high grade disease following baseline HPV positive test, cytology (ASCUS+) and an abnormal colposcopy are shown in Table V. These hazard ratios are of a similar magnitude for the 3 diagnostic tests and also between Europe and South Africa. The cumulative risk of developing high-grade disease according to the baseline HPV and cytology result is shown by Kaplan–Meier plots in Figures 1 and 2. If the initial result was negative the estimated risk of high-grade disease at 1 year was extremely low for the HPV, cytology and colposcopy; 1.42, 1.50 and 1.42%, respectively. The corresponding risks for 2 and 3 years, respectively, are 2.34 and 2.90% for HPV; 3.88 and 5.59% for cytology; 3.66 and 6.31% for colposcopy. These figures, taken from the Kaplan–Meier estimates are somewhat higher than the observed figures that do not take into account loss-to-follow-up and incomplete ascertainment. These data suggest that screening every 2 years by cytology would be safe and cytology has a definite advantage over HPV as a stand alone test in terms of specificity.

thumbnail image

Figure 1. Kaplan-Meier failure function (of high grade disease) according to baseline test result in Europe.

Download figure to PowerPoint

thumbnail image

Figure 2. Kaplan-Meier failure function (of high grade disease) according to baseline test result in South Africa.

Download figure to PowerPoint

Table V. Hazard Ratio of High Grade Disease During Follow-Up (Excluding Baseline) According to Baseline Test Results
Test resultRate ratioAdjusted rate ratio1Adjusted rate ratio2
  • 1

    Adjusted for CD4 count, HIV viral load, HAART use and smoking.

  • 2

    Adjusted for Other two test results of HPV, colposcopy and cytology, CD4 count, HIV viral load, HAART use and smoking and restricted to those for nonmissing values for all variables.

  • 3

    Values in parentheses are (95% CI).

HPV (HR positive)
 Overall5.60 (2.60, 12.07)34.57 (2.03, 10.30)1.98 (0.74, 5.32)
 Europe6.59 (2.47, 17.62)5.59 (1.98, 15.77)2.48 (0.69, 8.96)
 South Africa5.09 (1.27, 20.41)3.77 (0.88, 16.20)1.34 (0.26, 6.97)
Cytology (>ASCUS) or ASCUS+
 Overall6.47 (3.78, 11.06)6.03 (3.33, 10.91)3.06 (1.31, 7.15)
 Europe7.55 (4.06, 14.04)6.84 (3.45, 13.59)3.04 (1.02, 9.05)
 South Africa5.67 (1.79, 18.01)5.95 (1.50, 23.66)3.67 (0.80, 16.77)
Colposcopy (any abnormality)
 Overall4.06 (2.41, 6.83)4.69 (2.66, 8.29)2.87 (1.40, 5.92)
 Europe4.65 (2.47, 8.74)5.00 (2.55, 9.79)2.81 (1.10, 7.21)
 South Africa3.13 (1.20, 8.15)4.33 (1.33, 14.12)3.06 (0.90, 10.40)

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

This is one of the largest reported cohort studies of cervical surveillance in HIV-infected women and is unique by virtue of routine colposcopic diagnosis in addition to routine cytology and HPV testing. The multinational make up of the cohort includes both women living in Europe who may well have previously been screened for cervical cancer and had access to combined antiretroviral therapy, and women from South African who have not been screened for cervical cancer before their HIV diagnosis, with no access to antiviral treatment. This provides a current and comprehensive panel of women living with HIV/AIDS.

The high abnormal cytology rates and HPV detection rates are very much in keeping with previous reports. The South-African group has a higher proportion of women with a CD4 count of <200 which marks them out as higher risk in relation to the development of HPV-associated neoplasia. Interestingly, the rate of high grade cytology was similar in South African and European women. The cervical outcomes provide a number of clinically relevant insights. The normal cytology rates of the whole cohort is only 60% with almost 10% having high-grade cytological abnormalities and almost a quarter having low-grade abnormalities. A strength which makes the cervical data more robust is the inclusion of routine colposcopic examination which enables the diagnostic accuracy of cytology to be more accurately assessed. Routine cervical biopsy in every case would have provided a better diagnostic gold standard, but in the presence of no abnormality the site of a biopsy cannot usefully be determined. Using colposcopy combined with both HPV and cytology testing we were able to estimate as best as possible the levels of high grade CIN. Previous literature has suggested that in HIV infected women, cervical cytology will be less sensitive for the detection of CIN.14 There were no cases in which histologically confirmed high grade disease at baseline was associated with a normal cytology result. Even when considering those with incomplete results a sensitivity of 100.0% was estimated. Additionally, in almost 80% of cases of proven high grade lesions histologically at baseline the corresponding cytology reveals high grade change.

We found a higher sensitivity for cytology than reported in previous studies among HIV-positive women, while the specificity was lower.15, 16, 17 The reported sensitivity and specificity, based on women in whom histological verification was achieved, appear unduly high and low respectively, for cytology. The explanation for this probably lies in the unusual nature of the group from which they were derived. These women have a high prevalence of HPV infection resulting in high levels of koilocytosis which in turn will produce a lot of background atypia both cytological and colposcopically. Hence, when the threshold for cytological abnormalityis set at ASCUS, and that for positive histology set at CIN2+, there will be a large number of women with false positive cytology. This is not such a problem for a group of women known to be at increased risk in whom the required diagnostic performance differs from that of a normal screened population. The colposcopic performance is within reported ranges18 as is the HPV testing.19

The high HPV infection rates overall using HC2 are due to the higher rate amongst the cytologically normal women. The 70% rate of HPV positivity in ASCUS suggests that HPV testing would not be cost efficient as a triage, in contrast to immunocompetent women.15

The follow up data indicate that neither the CD4 counts nor HIV viral load were independent predictors of increased incidence of high grade disease when HAART and smoking were controlled for. In terms of which screening test should be used, the Kaplan–Meier plots indicate that cytology and HPV each perform equally well, but with the advantage of better specificity in HIV-positive women, cytology would appear the better screening test, with HPV being reserved for triage purposes if it is used at all. Because of the high frequency of ASCUS cytology, HPV testing would reduce the need for colposcopy although only 30% of ASCUS at baseline were HR HPV negative. The rate of incidence of high-grade lesions following initial screening is sufficiently low to be confident about not screening at greater than 2 yearly intervals.

In conclusion, this prospective study has confirmed the high prevalence of HR HPV infection and low-grade cytological abnormalities in HIV-positive women. HPV testing was as sensitive as cytology and colposcopy but was significantly less specific with poor positive predictive value. Low CD4 count and high HIV viral load were associated with higher rates of HPV infection and abnormal cytology, but were not independent predictors of high grade disease. Cytology performed every 2 years with colposcopy for >ASCUS would appear to be the most useful clinical protocol.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The collaborative work of Sonata Walraven, Charles Lacey, Simon Barton, Annette Rochford, Asmaa El Midaoui, Monica Valieri and Anne-Marie Williamson is gratefully acknowledged.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  • 1
    Massad LS,Riester KA,Anastos KM,Fruchter RG,Palefsky JM,Burk RD,Burns D,Greenblatt RM,Muderspach LI,Miotti P. Prevalence and predictors of squamous cell abnormalities in Papanicolaou smears from women infected with HIV-1. Women's Interagency HIV Study Group. J Acquir Immune Defic Syndr 1999; 21: 3341.
  • 2
    Ellerbrock TV,Chiasson MA,Bush TJ,Sun XW,Sawo D,Brudney K,Wright TC,Jr. Incidence of cervical squamous intraepithelial lesions in HIV-infected women. JAMA 2000; 283: 10317.
  • 3
    La Ruche G,Ramon R,Mensah-Ado I,Bergeron C,Diomande M,Sylla-Koko F,Ehouman A,Toure-Coulibaly K,Welffens-Ekra C,Dabis F. Squamous intraepithelial lesions of the cervix, invasive cervical carcinoma, and immunosuppression induced by human immunodeficiency virus in Africa. Dyscer-CI Group. Cancer 1998; 82: 24018.
  • 4
    Ahdieh L,Klein RS,Burk R,Cu-Uvin S,Schuman P,Duerr A,Safaeian M,Astemborski J,Daniel R,Shah K. Prevalence, incidence, and type-specific persistence of human papillomavirus in human immunodeficiency virus (HIV)-positive and HIV-negative women. J Infect Dis 2001; 184: 68290.
  • 5
    Schuman P,Ohmit SE,Klein RS,Duerr A,Cu-Uvin S,Jamieson DJ,Anderson J,Shah KV. Longitudinal study of cervical squamous intraepithelial lesions in human immunodeficiency virus (HIV)-seropositive and at-risk HIV-seronegative women. J Infect Dis 2003; 188: 12836.
  • 6
    Palella FJ,Jr,Delaney KM,Moorman AC,Loveless MO,Fuhrer J,Satten GA,Aschman DJ,Holmberg SD. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med 1998; 338: 85360.
  • 7
    Mocroft A,Ledergerber B,Katlama C,Kirk O,Reiss P,d'Arminio Monforte A,Knysz B,Dietrich M,Phillips AN,Lundgren JD. Decline in the AIDS and death rates in the EuroSIDA study: an observational study. Lancet 2003; 362: 229.
  • 8
    Blacker J. The impact of AIDS on adult mortality: evidence from national and regional statistics. Aids 2004; 18( Suppl 2): S19S26.
  • 9
    Cohn JA,Gagnon S,Spence MR,Harrison DD,Kluzak TR,Langenberg P,Brinson C,Stein A,Hellinger J. The role of human papillomavirus deoxyribonucleic acid assay and repeated cervical cytologic examination in the detection of cervical intraepithelial neoplasia among human immunodeficiency virus-infected women. Cervical Disease Study Group of the American Foundation for AIDS Research Community Based Clinical Trials Network. Am J Obstet Gynecol 2001; 184: 32230.
  • 10
    Massad LS,Evans CT,Minkoff H,Watts DH,Strickler HD,Darragh T,Levine A,Anastos K,Moxley M,Passaro DJ. Natural history of grade 1 cervical intraepithelial neoplasia in women with human immunodeficiency virus. Obstet Gynecol 2004; 104: 107785.
  • 11
    Heard I,Palefsky JM,Kazatchkine MD. The impact of HIV antiviral therapy on human papillomavirus (HPV) infections and HPV-related diseases. Antivir Ther 2004; 9: 1322.
  • 12
    Clifford GM,Gallus S,Herrero R,Munoz N,Snijders PJ,Vaccarella S,Anh PT,Ferreccio C,Hieu NT,Matos E,Molano M,Rajkumar R, et al. Worldwide distribution of human papillomavirus types in cytologically normal women in the International Agency for Research on Cancer HPV prevalence surveys: a pooled analysis. Lancet 2005; 366: 9918.
  • 13
    Kitchener HC,Almonte M,Wheeler P,Desai M,Gilham C,Bailey A,Sargent A,Peto J. HPV testing in routine cervical screening: cross sectional data from the ARTISTIC trial. Br J Cancer 2006; 95: 5661.
  • 14
    Maiman M,Tarricone N,Vieira J,Suarez J,Serur E,Boyce JG. Colposcopic evaluation of human immunodeficiency virus-seropositive women. Obstet Gynecol 1991; 78: 848.
  • 15
    Korn AP,Autry M,DeRemer PA,Tan W. Sensitivity of the Papanicolaou smear in human immunodeficiency virus-infected women. Obstet Gynecol 1994; 83: 4014.
  • 16
    Fink MJ,Fruchter RG,Maiman M,Kelly P,Sedlis A,Webber CA,Chen P. The adequacy of cytology and colposcopy in diagnosing cervical neoplasia in HIV-seropositive women. Gynecol Oncol 1994; 55: 1337.
  • 17
    Wright TC,Jr,Ellerbrock TV,Chiasson MA,Van Devanter N,Sun XW. Cervical intraepithelial neoplasia in women infected with human immunodeficiency virus: prevalence, risk factors, and validity of Papanicolaou smears. New York Cervical Disease Study. Obstet Gynecol 1994; 84: 5917.
  • 18
    Mitchell MF,Schottenfeld D,Tortolero-Luna G,Cantor SB,Richards-Kortum R. Colposcopy for the diagnosis of squamous intraepithelial lesions: a meta-analysis. Obstet Gynecol 1998; 91: 62631.
  • 19
    Cuzick J,Szarewski A,Cubie H,Hulman G,Kitchener H,Luesley D,McGoogan E,Menon U,Terry G,Edwards R,Brooks C,Desai M, etal. Management of women who test positive for high-risk types of human papillomavirus: the HART study. Lancet 2003; 362: 18716.