Characteristics of HPV infection over time in European women who are HIV-1 positive

Authors

  • I Heard,

    1. HPV National Reference Centre, Pasteur Institute, Paris, France
    2. Groupe Hospitalier Pitié-Salpétrière, Université Pierre et Marie Curie, Paris, France
    3. INSERM U943, Paris, France
    Search for more papers by this author
  • HA Cubie,

    1. Scottish HPV Reference Laboratory, Royal Infirmary of Edinburgh, Edinburgh, UK
    Search for more papers by this author
  • D Mesher,

    1. Cancer Research UK Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Queen Mary University of London, Barts & The London School of Medicine and Dentistry, London, UK
    Search for more papers by this author
  • P Sasieni,

    1. Cancer Research UK Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Queen Mary University of London, Barts & The London School of Medicine and Dentistry, London, UK
    Search for more papers by this author
  • for the MACH-1 Study Group


Dr I Heard, HPV National Reference Centre, Pasteur Institute, 25 rue du Dr Roux, 75015, Paris. Email isabelle.heard@pasteur.fr

Abstract

Objective  The objectives of the study were to investigate high-risk human papillomavirus (HR-HPV) infection and type distribution in women infected with HIV-1, and to determine the relevance of HR-HPV positivity and persistence/loss to the development of high-grade cervical disease.

Design  A total of 518 European women infected with HIV attending for routine gynaecological care consented to 6-monthly follow-up visits over 3 years, with surveillance of cytology, colposcopy and histopathology, where relevant, and longer follow-up, where possible.

Setting  European women infected with HIV attending for routine gynaecological care.

Population or sample  518 European women infected with HIV attending for gynaecological care in 6 hospital-based European centres - Dublin, Edinburgh, London, Milan, Paris, and Warsaw.

Methods  Cervical screening was achieved by liquid-based cytology (LBC) of brush samples in PreservCyt® medium. The HPV testing of residual samples was performed by Hybrid-Capture II, with genotyping of positives using the HPV Line Blot Assay. Histology results were accessed where available.

Main outcome measures  Description of high risk human papillomavirus (HR-HPV) infection and type distribution in HIV-1 infected women.

Results  The estimated prevalence at baseline of any HR-HPV type was 49.5% (46.3–52.8%): 10.2% for HPV 16 and 4.3% for HPV 18. The prevalence increased with increasing immunosuppression. Multiple infections were detected in 26.8%. HR-HPV genotypes were detected in 34.9% of cases with normal cytology, in 77.2% of cases with atypical squamous cells of undetermined significance/low-grade squamous intraepithelial lesion (ASCUS/LSIL) and in 90.8% of cases with high-grade SIL (HSIL). The prevalence of HPV 16 in HSIL was 38.5%, with the three most common types thereafter having prevalence rates of 19.2% (HPV 58), 19.2% (HPV 53) and 16.6% (HPV 52). The overall persistence of any high-risk type was 55.8%. We found that 6 months persistence of HPV 16 occurred in 24 women. Seven cases of high-grade cervical disease developed, and all were associated with initial/persistent HR-HPV positivity.

Conclusions  A wide diversity of HPV types was evident, and multiple infections were common. Detection or persistence of any HR-HPV was associated with a very low incidence of subsequent high-grade disease.

Introduction

Cervical infection by certain (so-called ‘high-risk’) human papillomaviruses may cause cervical cancer and precancerous lesions. Among women infected with HIV-1, it has been shown that there is an increased risk of cervical cancer and its precursors.1,2 As in the general population, both HPV type and persistence of HPV infection are considered important factors determining the strength of this association between HPV and cervical abnormalities.3 In addition, it has been shown that HIV viral load and CD4+ cell count have a strong association with the prevalence of HPV infection.4 However, the increased availability, uptake and adherence to anti-retroviral therapy regimes have reduced the level of immunosuppression and plasma HIV load in infected patients.5 This might reduce the incidence of precancerous lesions and cervical cancers. Conversely, if HPV persistence remains high, longer survival in adequately treated women could lead to an increase in high-grade cervical disease and, if untreated, cancers in the future.

Our earlier report on the MACH-1 (management of abnormal cytology in HIV-1 infected women) study group showed that annual cytology was adequate for surveillance in women infected with HIV.6 There was also the suggestion that 3-yearly HPV testing might be as good as annual cytology. Independently, a number of studies have shown both a broader range of HPV types and a higher frequency of multiple infections in women infected with HIV.3 It is therefore important to study HIV-infected cohorts longitudinally to assess the impact of HPV type diversity and persistence on the development of high-grade cervical disease. The HPV type-specific results from the South African cohort of MACH-1 were recently published.7

The present study aimed to evaluate the hypothesis that high-risk human papillomavirus (HR-HPV) persistence is frequent in the setting of HIV, and generates a high risk of progression to cancer. We present HR-HPV results from the follow-up of the European cohorts of MACH-1 for up to 6 years.

Methods

The MACH-1 collaborative group included six hospital-based European centres: Dublin, Edinburgh, London, Milan, Paris and Warsaw. Women who gave informed consent were invited for 6-monthly follow-up visits for 3 years, with surveillance of cytology, colposcopy and histopathology, where relevant. Each centre followed local protocols for the follow-up of patients. Cytology was classified according to the Bethesda reporting system: normal, atypical squamous cells of undetermined significance (ASCUS), low-grade squamous intraepithelial lesion (LSIL), high-grade SIL (HSIL) and cancer. Histology was reported as normal, cervical intraepithelial neoplasia 1 (CIN1), CIN2, CIN3 or invasive cancer. If no biopsy was taken, histology was considered to be normal, provided colposcopy was normal. In general, LSIL was not treated. HPV test results were not used to influence patient management. Follow-up beyond the initial 3 years of MACH-1 was achieved within the Dublin, Edinburgh and Paris cohorts.

Samples

Over 1100 women positive for HIV-1 were recruited from European centres to participate in a prospective study on cervical disease and HPV infection. Baseline data on the cohort has already been published.6 Where conventional cervical smears were standard practice, a second cervix brush sample was taken in PreservCyt medium, whereas in centres using liquid-based cytology (LBC), residual cell suspension was saved. We attempted to follow-up all women for 3 years with 6-monthly visits to detect high-grade cervical disease. Of the total cohort, 933 women had at least one HR-HPV test; there were 518 women that had at least two samples taken, and 270 of these women had at least one HPV-positive/genotype result.

HPV tests

The HPV screening was performed using Digene HR-HPV Hybrid Capture II (HC-II). LBC samples were processed using Sample Conversion Buffer (also Digene), denatured at 65°C for 45 minutes and stored at −80°C. HC-II was carried out according to the manufacturer’s instructions. HC-II is a sandwich capture hybridisation assay with signal amplification to detect HR-HPV using a pool of oligonucleotides covering 13 HR-HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68). Results are expressed as an index by comparing the chemiluminescent signal in relative light units (RLUs) with a standard cut-off. Samples with an RLU/CO value >1 were considered to be screen positive.

Samples that were HC-II positive were further tested using an HPV Line Blot Assay (LBA) to identify specific HPV genotypes. DNA was extracted from residual material in PreservCyt medium using silica columns (either manually or using a Qiagen Biorobot 9604 (Qiagen, Crawley, England)). LBA was a ‘research use only’ (RUO) genotyping assay, the forerunner of the CE-marked HPV Linear Array (LA; Roche Molecular Systems). Both LBA and LA are PCR based, and use PYGMY primers to amplify a 450-bp fragment of HPV L1 (8), with detection by reverse hybridization, onto which multiple HPV probes are bound. Hybridisation was carried out at 53°C according to the manufacturer’s instructions. The strips allow detection of HR-HPV types 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 67, 68, CP6970, 73 and 82, as well as low-risk (LR) HPV types 6, 11, 40, 42, 54, 55, 57, 61, 62, 64, 71, 72, 81, 83 and 84.

Statistical analysis

Crude estimates of HPV type-specific prevalence would be misleading, because typing was performed only on 270 of the 933 women, and more frequently in samples that were HC-II positive than in samples found to be HC-II negative. In order to estimate the prevalence of individual HPV types in the study cohort, we stratified women according to their baseline and subsequent HC-II results, and obtained a weighted estimate of the overall prevalence. Explicitly, we created four strata defined by whether the baseline HC-II result was positive or negative, and whether any subsequent samples were HC-II positive (or were all negative). Individuals with all negative HC-II results were assumed to have no HR-HPV types. Within the remaining strata, a weighted estimate was calculated as sum(pini)/sum(ni), where ni is the number of samples in stratum i, and pi is the observed prevalence of the given HPV type among typed samples. These were calculated separately within groups of baseline factors (age, CD4+ cell count and cytology), and prevalence of infection was tabulated according to the same factors.

Stratified tests for trend were performed using a weighted logistic regression. For age and CD4+ cell count, we considered a linear increase. For the test for trend for cytology, we coded as 0, 1 or 10 for the logistic regression for cytology normal, ASCUS/LSIL and HSIL/cancer, respectively, in order to place more importance on the difference between ASCUS/LSIL and HSIL, and less importance on the difference between normal and ASCUS/LSIL.

The acquisition of HPV types over time is illustrated using Kaplan–Meier plots of time from the first test (negative for the specific type) to first positive test (individuals were censored at the time of the last test). Similarly for loss of HPV type, we considered time from first positive test to first negative test for that type.

To calculate the cumulative incidence of CIN2+, histological events were backdated to the time of the preceding abnormal cytology if there was one within 6 months, and follow-up time at risk was taken from the date of the initial genotyping test (or first negative HPV test, for those with no genotyping) until the date of the last test recorded for that individual.

Results

Characteristics of cohorts

Of the 933 women with HPV results, 518 (55.5%) provided at least two samples >6 months apart for HPV testing. These women were aged between 17 and 64 years, with a median of 35 years, and this was similar between centres. Of the 518 women with at least two HPV results, 480 had a known CD4+ cell count: 29.4% (141) had a CD4+ cell count between 200 and 400 cells/mm3, whereas 16.7% (80) had a CD4+ cell count of <200 cells/mm3. Altogether, 69.7% of patients were on combination antiretroviral therapy (cART) at the time of the first genotyping, including 80% of those with a CD4+ cell count of <200 cells/mm3. At the time of the first HPV test, cytology was normal in 331 women, ASCUS/LSIL in 100 women, HSIL or worse in 41 women, inadequate in four women and missing for 42 women. The mean follow-up period after the initial HPV test was 29.1 months. Two hundred and seventy (28.9%) women had at least one HPV-positive sample available for genotyping, and no HPV DNA was detected in any of the samples from the remaining 248 women. A flow chart has been included to summarise recruitment and number of samples received and genotyped throughout the study (Figure 1). The mean number of genotyping visits was 2.7. There were no cases of cervical cancer at baseline or during follow-up.

Figure 1.

 Flowchart to summarise recruitment and number of samples received and genotyped during the study.

HPV at baseline

Baseline HPV prevalence was calculated from genotyping the first samples from 270 women, weighted to the whole cohort that provided samples (933), giving an estimated prevalence of at least one HR-HPV type of 49.5% (46.3–52.8%). HR-HPV prevalence increased from 44.6 to 52.0 to 62.9% with decreasing CD4 cell count (CD4 > 400, 400–200 and <200 cells/mm3, respectively; = 0.01). The prevalence of HR-HPV for those on cART at the time of first genotyping was 47.8%, compared with 56.5% for women not receiving cART. HR-HPV prevalence also increased with the presence and severity of abnormalities seen on cytology: 34.9% for normal cytology, 77.2% for ASCUS/LSIL and 90.8% for HSIL.

HPV type distribution

HPV 16 and 18 together represented the highest prevalence overall (13.8%), and within each age group (16.4 and 12.1%, respectively; Table 1). HPV 16, 53, 52 and 58 were the most prevalent individual types (10.2, 10.2, 8.8 and 7.5%, respectively). In the older age group (>30 years), however, HPV 16 and 53 were the most frequent types, with the same estimated prevalence of 9.7%. Eight HPV types had a prevalence of 5% or more. In women under 30 years of age, 57.9% were positive for high-risk HPV compared with 48.2% of women aged 30 years or older. HPV 16 and 18 were together found significantly less frequently in women with CD4 > 400 than in women who were further immunocompromised (CD4 < 200 cells/mm3) (10.6 and 16.6%, respectively; = 0.093). HPV 53 and 58 were the most frequently detected types in women with CD4 < 200 cells/mm3. HPV 58, 66 and 82 showed a significantly increasing prevalence as the CD4 count decreased (< 0.001, < 0.001 and = 0.007, respectively).

Table 1. HR-HPV types according to age, CD4 count and cytology
HR-HPV typeOverall
Number (weighted percentage)
AgeCD4 CountCytology
<30 years≥30 years<200 cells/mm3200–400 cells/mm3≥400 cells/mm3NormalASCUS/LSILHSIL/Cancer
Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)
HR 16/1861 (13.8)11 (15.6)50 (13.6)10 (16.6)22 (17.6)23 (10.6)18 (7.4)20 (19.7)18 (50.4)
HR 1645 (10.2)10 (13.9)35 (9.7)8 (13.3)14 (11.0)18 (8.5)13 (5.2)14 (14.2)13 (38.5)
HR 5344 (10.2)9 (13.7)35 (9.7)10 (18.4)14 (9.7)19 (9.3)16 (6.4)18 (19.6)7 (19.2)
HR 5240 (8.8)9 (12.8)31 (8.2)5 (8.3)13 (9.6)20 (9.0)15 (6.0)17 (16.5)7 (16.6)
HR 5833 (7.5)4 (5.5)29 (7.9)11 (18.2)9 (7.1)9 (4.0)10 (4.2)13 (12.8)7 (19.2)
HR 6626 (5.7)3 (4.2)23 (6.1)6 (10.8)15 (10.4)3 (1.3)8 (3.2)14 (13.8)2 (7.3)
HR 5626 (5.4)4 (4.0)24 (5.7)3 (5.0)11 (7.2)12 (5.2)9 (3.4)12 (11.4)2 (4.7)
HR 5124 (5.2)1 (1.4)24 (5.8)6 (9.8)6 (4.3)11 (4.4)6 (2.2)11 (10.4)5 (11.8)
HR 4522 (5.3)5 (7.5)17 (5.0)3 (4.9)9 (7.1)9 (4.6)7 (2.7)8 (8.6)3 (12.3)
HR 3520 (4.4)2 (2.8)18 (4.8)2 (3.3)6 (3.9)10 (4.5)3 (1.3)10 (10.6)4 (9.5)
HR 6820 (4.5)5 (7.0)15 (4.1)4 (6.6)7 (5.4)6 (2.6)12 (4.9)5 (4.5)2 (4.7)
HR 1819 (4.3)1 (1.7)18 (4.7)3 (5.0)9 (7.4)5 (2.1)6 (2.6)7 (6.5)6 (14.2)
HR 3119 (4.3)4 (5.5)15 (4.1)2 (3.3)9 (6.9)8 (3.7)9 (3.6)7 (7.3)1 (2.4)
HR 3917 (3.7)4 (5.9)13 (3.4)2 (3.3)8 (6.0)6 (2.7)5 (1.8)7 (6.8)2 (4.7)
HR 7016 (3.9)1 (1.4)15 (4.4)2 (3.3)3 (2.8)8 (4.0)11 (4.6)2 (2.3)2 (4.7)
HR 3315 (3.4)4 (5.9)11 (2.9)5 (8.2)3 (2.2)6 (2.6)3 (1.3)7 (6.9)5 (11.8)
HR 5915 (3.4)5 (7.5)10 (2.7)4 (6.6)4 (2.9)6 (3.0)8 (3.3)3 (3.0)4 (9.5)
HR 739 (2.1)3 (4.5)6 (1.7)1 (1.7)3 (2.2)5 (2.5)3 (1.3)1 (1.0)4 (12.1)
HR 829 (1.9)3 (4.2)6 (1.5)3 (4.9)5 (3.2)1 (0.4)2 (0.8)3 (3.0)1 (2.4)
HR IS39 4 (0.9)0 (0.0)4 (1.1)2 (3.3)1 (0.7)1 (0.4)3 (1.2)1 (1.0)0 (0.0)
HR 694 (0.8)1 (1.4)3 (0.7)0 (0.0)1 (0.7)3 (1.1)0 (0.0)3 (2.5)1 (2.4)
HR 673 (0.7)0 (0.0)3 (0.9)0 (0.0)2 (1.8)0 (0.0)1 (0.5)1 (1.0)1 (2.4)
HR 261 (0.2)0 (0.0)1 (0.3)1 (1.7)0 (0.0)0 (0.0)0 (0.0)1 (1.0)0 (0.0)
Any HR type216 (49.5)41 (57.9)174 (48.2)37 (62.9)68 (52.0)96 (44.6)87 (34.9)75 (77.2)34 (90.8)
No HR type54 (50.5)8 (42.1)45 (51.8)5 (37.1)19 (48.0)25 (55.4)35 (65.1)9 (22.8)2 (9.2)
Total270 (100.0)49 (100.0)219 (100.0)42 (100.0)87 (100.0)121 (100.0)122 (100.0)84 (100.0)36 (100.0)

HPV 16/18 genotypes were detected in 7.4% of normal cytology, 19.7% of ASCUS/LSIL and 50.4% of HSIL (Table 1). Three HPV types (16, 52 and 53) showed a level of >5% in women with normal cytology. HPV types 16, 58, 53 and 52 were the most frequently detected types among women with HSIL (38.5, 19.2, 19.2 and 16.6%, respectively). Types 16, 18, 33, 45, 51, 52, 58 and 73 showed a significant increase in prevalence as cytology increased from normal to ASCUS/LSIL to HSIL.

Infection with multiple HPV types

Overall, the prevalence of multiple infections (two or more HPV types) was 26.8%, corresponding to 54.3% of women with a high-risk infection (Table 2). It was similar in women under 30 years of age, and in those of 30 years of age or older, and increased with lower CD4 counts. Multiple infections were detected at a higher percentage in HSIL than in normal cytology (59.9 and 17.0%, respectively), but the difference in proportions of multiple infections was not as great when considered as a proportion of women with an infection (48.7 and 65.9%, respectively). However, the prevalence of four or more types increased markedly with increasing levels of cytological abnormality (P value for trend < 0.0001).

Table 2. Rate of multiple HPV infections according to age, CD4 count and cytology
Number of HR typesOverall
Number (weighted percentage)
AgeCD4 CountCytology
<30 years≥30 years<200 cells/mm3200–400 cells/mm3≥400 cells/mm3NormalASCUS/LSILHSIL/Cancer
Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)Number (weighted percentage)
054 (50.5)8 (42.2)45 (51.8)5 (37.2)19 (48.0)25 (55.5)35 (65.1)9 (22.8)2 (9.2)
198 (22.6)19 (26.5)78 (21.9)12 (20.7)24 (18.4)54 (25.3)45 (17.9)31 (33.3)12 (31.0)
267 (15.8)11 (15.6)56 (16.0)12 (20.6)25 (20.7)26 (12.0)25 (10.4)22 (22.8)12 (33.6)
326 (5.7)8 (11.7)18 (4.7)7 (11.6)9 (6.1)8 (3.8)13 (5.1)10 (9.7)3 (7.1)
≥425 (5.3)3 (4.2)22 (5.6)6 (10.0)10 (6.8)8 (3.5)4 (1.5)12 (11.4)7 (19.2)
Total270 (100.0)49 (100.0)219 (100.0)42 (100)87 (100)121 (100)122 (100)84 (100)36 (100)

HPV detection during follow-up

The loss of any HR-HPV type is shown in Figure 2. Seventy-seven percent of any HPV types were cleared within 2 years. HPV 18 was cleared more quickly than other types. Similarly, the gain of HR-HPV types is shown in Figure 3.

Figure 2.

 The proportion of women initially positive for an HR-HPV who have loss of HR-HPV type positivity over time (in years). Line produced using Kaplan–Meier estimates, considering time from first positive test to first negative test for that type, with individuals censored at the time of the last test.

Figure 3.

 The proportion of women initially negative for an HR-HPV who acquire HR-HPV-type positivity over time (in years). Line produced using Kaplan–Meier estimates, considering time from first negative test to first positive test for that type, with individuals censored at the time of the last test.

There were 231 women who had at least two HPV-positive and genotyped samples more than 6 months apart. The overall persistence of any high-risk type was 55.8% (129/231). Of these, 24 women (10.4%) had persistent HPV 16, eight women (3.5%) had persistent HPV 18 and 114 women (49.4%) had persistence of at least one other HR-HPV during their follow-up period.

HPV detection related to the development of cervical disease

The result of the last cytology was compared with the baseline HPV result. Only 12.3% (38/310) of women who were HPV positive at baseline presented with abnormal cytology at any stage through the follow-up (15 ASCUS, 22 LSIL and 1 HSIL). Abnormal cytology occurred in 9.7% (25/257) of women who were persistently HPV negative (7 ASCUS, 17 LSIL and 1 HSIL).

The cumulative incidence of histologically-confirmed CIN2+ was calculated for the 518 women with adequate follow-up after an initial HPV test, and was analysed according to initial and subsequent HPV testing (Table 3). Of the 518 women, 261 had at least one positive HC2 test and also a further valid genotyping test. The remaining women had only negative HPV tests and no genotyping was performed. None of the women who were HPV negative throughout developed CIN2+. The cumulative incidence of histologically confirmed CIN2+ was also calculated for women with at least two tests showing HPV-type-specific persistence. Only three cases of CIN2+ were found in the cohort during the original MACH-1 study, all of which showed persistent infection: one individual with persistent HPV 66, one individual with persistent HPV 31 and 45, and one individual with persistent HPV 33, 51, 52 and 82. No other women developed disease after the second HPV test during the equivalent of 179 woman years of follow-up. Furthermore, none of the 24 women with persistent HPV 16 developed abnormalities during the period of follow-up (Table 3). Attempts were made to obtain further cytology and histology information from women who had persistent HPV 16 or persistent infection with any other HR-HPV. It was possible to obtain data from 65 women from the cohorts in three sites (Dublin, Edinburgh and Paris). This revealed three additional cases of CIN2+ in women who had previously had persistent HR-HPV (one with HPV 16 and 33 persistence, one with HPV 31 and 45 persistence, and one with HPV 35 persistence). The safety of this protocol is evident from the absence of interval cancers in our cohort.

Table 3. Rates of CIN2 +  acquisition according to HPV test results
 Rate of CIN2+ (95% CI)
  1. *HR other excludes HR 16.

Initial HPV result (total n = 518)
Negative (n = 310)0%
Positive for HR 16 (n = 44)0%
Positive for other HR-HPV (n = 164)*3.6% (1.7–%)
Initial/next HPV result (total n = 518)
Negative/negative (n = 293)0%
HR 16/HR 16 (n = 24)0%
HR 16/HR other (n = 6)*0%
HR other/HR 16 (n = 4)*68.1% (17.0–272.5%)
HR other/HR other (n = 103)*3.2% (1.2–8.6%)
Negative/HR any (n = 17)0%
HR any/negative (n = 71)1.3% (0.1–9.3%)

Discussion

HPV prevalence

The women in our study who were positive for HIV had a high prevalence of infection with oncogenic HPV genotypes (49.5%), as well as multiple types (26.8%). These findings are fairly consistent with other studies of women infected with HIV living in Europe that showed a range of HPV prevalence, from 40.3 to 91% (reviewed by De Vuyst et al.2). As one would expect, a lower overall HPV prevalence of around 36–41% for any HPV has been demonstrated in women positive for HIV with no cytological abnormalities, compared with those with abnormal cytology.3,9 In the HIV-free population, the rate of HPV infection is highest in women below 30 years of age and decreases at older ages. Such a decrease in prevalence in older women was not observed in our study population of women positive for HIV, with around 50% of women being infected with HR-HPV. This might be related to the persistence of HPV infection across the entire genital tract or the reactivation of latent viral infection, even in the absence of sexual activity.10 Such a high rate of infection with HR-HPV over the age of 30 years might decrease the benefit of combined HPV and Pap tests for cervical cancer screening in this population.

Several studies have consistently shown a higher prevalence of HPV DNA in women with CD4 counts lower than 200 cells/mm3.2 We similarly observed a significant increase in the prevalence of HR types from 44.6% in women with a CD4 count >400 cells/mm3 to 62.9% in severely immunocompromised women (= 0.01). Such a high rate of HR-HPV infection in women with low CD4 counts suggests that these patients may benefit from sustained cervical cancer screening. As in the general population, an increasing severity of cervical cytology was associated with a higher prevalence of HR-HPV. Nevertheless, asymptomatic HR-HPV infection was frequent, as HR-HPVs were detected in one-third of women with normal cytology. This is of relevance when considering methods of cervical screening for women positive for HIV, because a high prevalence of HPV infection may mean that the specificity of HPV testing for detection of cervical high-grade disease and cancer is too low to be clinically useful.11 However, as women positive for HIV are recommended to have annual screening, the high long-term sensitivity might be sufficient to offer 2- or even 3-yearly screening to those who are HPV negative, and restrict annual surveillance to women who are HPV positive.

HPV type distribution

Recent reports suggest that HPV 16 and 18 are not as common in individuals infected with HIV as for other high-risk types.12 Although HPV 16 was not the most common type associated with normal cytology in our study, it was by far the most frequently associated type in HSIL/cancer (Table 1). HPV 18 ranks 12th in estimated percentage in cases of normal cytology, and fifth in cases of HSIL. As observed in women from the Women’s Interagency HIV Study (WIHS),13 HPV 16 prevalence increased with lower CD4 cell count as well as with increasing cytological severity. Overall, among women with HSIL, the prevalence of HPV 16/18 was 50.4% (Table 1), a percentage somewhat similar to that observed in HSIL in the general population.14 This suggests a similar benefit of the available prophylactic HPV vaccines against HPV-16 and -18 precancerous lesions, regardless of HIV status, provided vaccine-related protection is sustained in the population infected with HIV.

In the MACH-1 cohort, HPV 53, 52 and 58 were the most prevalent after HPV 16, infecting 10.2, 8.8 and 7.5% of women, respectively. HPV 52 and 53 were common in samples with both normal or ASCUS/LSIL cytology and, although HPV 58, 53 and 52 occupied second, third and fourth positions in cases of HSIL/cancer, the high frequency in those without high-grade disease suggests that they are not such important risk factors for high-grade disease, possibly because of lower integration rates for non-HPV 16/18 type DNA, as postulated by McKenzie et al.12 The differences in oncogenic potential for these different HPV types also reflects their relative positions in HPV phylogeny, with a recent classification report considering HPV 52 and 58 as ‘high risk’ for cervical cancer, with sufficient supporting evidence, and HPV 53 as ‘probable high risk’, with limited evidence to support a role in cervical cancer.15

The predominance of HPV 50–59 has been noted elsewhere, including the South African cohort of MACH-1.7 There are several reports of HPV 52 as the most common type in HIV-positive study groups in Europe,16,17 and in Canada,18 but the association with grade of dysplasia is not well documented. In their meta-analysis of 20 studies of HPV in women infected with HIV, Clifford et al. noted that HPV 16 was the most common type detected, followed by HPV 58 and then HPV 18.3 HPV 53 was not one of the top six types.

Studies carried out a few years ago such as this one, the Clifford meta-analysis and the recent publication from Anastos et al.19 used the best technology at the time. Whereas LBA was a prototype genotyping kit, we are content that the detection of the wide range of HPV types provided a reasonable and reproducible output. However, there will always be slight differences in the distribution of types when studies using different technologies are compared. It remains a possibility, therefore, that newer assays using different technologies, such as luminex or pyrosequencing, might give a slightly different distribution of types.

Incidence of CIN2+

In order to study the incidence of CIN2+ over time, clinical follow-up data from women who showed persistent HR-HPV during the MACH-1 study period has been obtained in the three cohorts in Dublin, Edinburgh and Paris. Only three ‘late’ cases of CIN2+ were identified (3/65; 4.6%). The low incidence of high-grade CIN in this study might be as a result of extensive treatment of patients prior to enrolment in the study. If treatment confers long-term protection against recurrence, this might explain the low incidence of high-grade CIN in women infected with both HIV and HPV in this study.20 This might also be related to an effect of cART on HPV infection and/or cervical lesions.21,22

This large study of HPV infection and cervical neoplasia in women positive for HIV-1 in the cART era has confirmed the very high rates of HPV infection, very high rates of multiple HPV infections and significant HR-HPV persistence. Nevertheless, we showed relatively low rates of HSIL and an absence of cervical cancer. Therefore, we can conclude that in women with well-controlled HIV, the risk of progression of HPV-related lesions remains low. This study supports the suggestion of 3-yearly cervical screening for women with well-controlled HIV infection.

Disclosure of interests

I.H. has no conflicts of interest to disclose. H.A.C. has received occasional consultancy payments for commercial advisory boards/testing of new HPV products, and expenses/hospitality when asked to speak at meetings for the same companies, namely GSK, SPMSD, Abbott Diagnostics, GenProbe, Hologic, NorChip, Qiagen and Roche Molecular Systems. D.M. has no conflicts of interest to disclose. P.S. has been a member of an advisory group on HPV testing for Gen-Probe and for Roche, and has had his expenses paid to attend HPV conferences by GSK.

Contribution to authorship

A writing committee consisting of the four listed authors was formed from the MACH-1 study group. I.H. and H.A.C. contributed to the design and execution of the study, whereas D.M. and P.S. were responsible for data collation and analysis. All four contributed significantly to the final article.

Details of ethical approval

Each centre was responsible for obtaining the research ethical approval relevant for their country. This included permission to send residual samples to Edinburgh for HPV testing.

Funding

We are grateful to the Parthenon Trust and to 3M Pharmaceuticals for an unrestricted educational grant, which together supported the European cohorts in the MACH-1 study group. P.S. and D.M. were supported by CRUK grant number EMS C8162/A10406.

Acknowledgements

MACH-1 study group: Alberto Agarossi, Investigator, University of Milan, Italy; Gerry Beattie, Investigator, St John’s Hospital, Livingstone, UK; Caroline Bradbeer, Investigator, St Thomas’ Hospital, London, UK; Elena Casolati, Investigator, University of Milan, Italy; Heather Cubie, Clinical Virology, Royal Infirmary of Edinburgh, UK; Lynette Denny, Investigator, University of Cape Town, South Africa; Isabelle Heard, Investigator, Pasteur Institute, Paris; Henry Kitchener, Principal Investigator, University of Manchester, UK; Fiona Lyons, Investigator, Saint James Hospital, Dublin, Ireland; David Mesher, Statistician, Queen Mary University of London, UK; Catherine Moore, Clinical Virology, Royal Infirmary of Edinburgh, UK; Linsey Nelson, Data Manager, University of Manchester, UK;

Tomasz Niemicz, Investigator, Research Institute of Mother and Child, Warsaw, Poland; Peter Sasieni, Epidemiologist, Queen Mary University of London, UK. We particularly thank our colleagues in Edinburgh (Dr Gerry Beattie) and Dublin (Drs Fiona Mulcahy and Aisling Loy) for accessing the clinical follow-up data of women who had been included in the MACH-1 cohort.

Ancillary