Customize your JAMA Network experience by selecting one or more topics from the list below.
Harris TG, Burk RD, Palefsky JM, Massad LS, Bang JY, Anastos K, Minkoff H, Hall CB, Bacon MC, Levine AM, Watts DH, Silverberg MJ, Xue X, Melnick SL, Strickler HD. Incidence of Cervical Squamous Intraepithelial Lesions Associated With HIV Serostatus, CD4 Cell Counts, and Human Papillomavirus Test Results. JAMA. 2005;293(12):1471–1476. doi:10.1001/jama.293.12.1471
Context Recent cervical cancer screening guidelines state that the interval
between screenings can be safely extended to 3 years in healthy women 30 years
or older who have normal cytology results and have negative test results for
oncogenic human papillomavirus (HPV) DNA.
Objective To determine the incidence of squamous intraepithelial lesions (SILs)
in HIV-seropositive women with normal cytology results, by baseline HPV DNA
Design, Setting, and Patients Participants were HIV-seropositive (n = 855; mean age, 36
years) and HIV-seronegative (n = 343; mean age, 34 years) US women
with normal baseline cervical cytology who were enrolled in the Women’s
Interagency HIV Study (WIHS), a large, multi-institutional prospective cohort
study. Since their recruitment during 1994-1995, WIHS participants have been
followed up semi-annually with repeated Pap smears for a median of 7 years.
Main Outcome Measure The cumulative incidence of any SIL and high-grade SIL or cancer (HSIL+)
was estimated according to baseline HPV DNA results, stratified by HIV serostatus
and CD4 T-cell count.
Results Development of any SIL in women with negative HPV results (both oncogenic
and nononcogenic) at 2 years was as follows: in HIV-seropositive women with
CD4 counts less than 200/μL, 9% (95% CI, 1%-18%); with CD4 counts between
200/μL and 500/μL, 9% (95% CI, 4%-13%); and with CD4 counts greater
than 500/μL, 4% (95% CI, 1%-7%). The CIs for these estimates overlapped
with those for HIV-seronegative women with normal baseline cytology who were
HPV-negative (3%; 95% CI, 1%-5%), indicating that at 2 years, there were no
large absolute differences in the cumulative incidence of any SIL between
groups. Furthermore, no HPV-negative participants in any group developed HSIL+
lesions within 3 years. Multivariate Cox models showed that on a relative
scale, the incidence of any SIL among HIV-seropositive women with CD4 counts
greater than 500/μL (hazard ratio [HR], 1.2; 95% CI, 0.5-3.0), but not
those with CD4 counts less than or equal to 500/μL (HR, 2.9; 95% CI, 1.2-7.1),
was similar to that in HIV-seronegative women.
Conclusion The similar low cumulative incidence of any SIL among HIV-seronegative
and HIV-seropositive women with CD4 counts greater than 500/μL and who
had normal cervical cytology and HPV-negative test results suggests that similar
cervical cancer screening practices may be applicable to both groups, although
this strategy warrants evaluation in an appropriate clinical trial.
Cervical cancer screening recommendations in the United States have
been recently updated and now advise using an interval of 3 years between
screenings in healthy women 30 years or older who have normal cytology results
and who test negative for oncogenic (cancer-associated) human papillomavirus
(HPV) DNA.1,2 The recommended
interval is 6 to 12 months for women with normal cytology results and detectable
oncogenic HPV. If no HPV test is conducted, 3 consecutive normal annual Papanicolaou
(Pap) smear results are required before the Pap smear frequency is changed
to once every 2 or 3 years. Support for these recommendations comes from several
large observational studies.1-5
However, guidelines for human immunodeficiency virus (HIV)–seropositive
women have not been revised since 19951,2,6,7 and
state that HIV-seropositive women should obtain 2 Pap smears 6 months apart
after the initial HIV diagnosis and, if results of both are normal, should
undergo annual cytologic screening. Human papillomavirus test results are
not considered,1,2,6 even
though economic models have suggested that HPV testing in HIV-seropositive
women might be cost-effective.8 In this study,
we determined the cumulative incidence of cervical squamous intraepithelial
lesions (SILs) among HIV-seropositive and HIV-seronegative women according
to baseline HPV results in a large prospective cohort. We sought to determine,
in keeping with recommendations for HIV-seronegative women, whether a single
initial HPV test result can be used to determine the appropriate cervical
cancer screening interval in an HIV-seropositive woman with normal cervical
Participants were HIV-seropositive (n = 855) and HIV-seronegative
(n = 343) women who were enrolled in the Women’s Interagency
HIV Study (WIHS) and had normal cervical cytology at baseline. The cohort
has been described in detail elsewhere.9-11 Briefly,
2059 HIV-seropositive women and 569 HIV-seronegative women were enrolled during
1994-1995 from similar clinical and outreach sources in New York City, Chicago,
Ill, Los Angeles and San Francisco, Calif, and the District of Columbia. All
participants provided written informed consent, and the study protocol was
approved by each local institutional review board. The HIV-seropositive cohort
has been previously shown to have demographic characteristics and risk behaviors
similar to those of US women with AIDS nationwide.9
Participants were not included in the current analysis if (1) their
baseline Pap smear was abnormal or missing (834 HIV-seropositive, 106 HIV-seronegative);
(2) they were missing information on baseline HPV status (126 HIV-seropositive,
33 HIV-seronegative) or CD4 T-cell count (35 HIV-seropositive); (3) they did
not have an intact cervix at baseline (113 HIV-seropositive, 31 HIV-seronegative);
(4) follow-up data were unavailable (96 HIV-seropositive, 43 HIV-seronegative);
or (5) they HIV-seroconverted during follow-up (13 HIV-seronegative). Race
and ethnicity information was obtained via questionnaire. Study participants
were asked whether they were Hispanic and to identify their race from the
following choices: black, white, Asian/Pacific Islander, Native American/Alaskan
Native, or other. We then categorized the women further as Hispanic, black
(non-Hispanic), white (non-Hispanic), and other. Race and ethnicity were included
in the analysis because they may be related to HIV serostatus and the risk
At baseline and semi-annual clinical visits, WIHS participants underwent
a pelvic examination with collection of a cervicovaginal lavage for HPV testing,
followed by a Pap smear using a wooden Ayres spatula and cytologic brush.
All Pap smears were interpreted centrally with the 1991 Bethesda System12 criteria. Two independent cytotechnologists examined
each Pap smear, and all smears identified as abnormal by either technologist,
as well as 10% of all negative smears, were evaluated by a cytopathologist.
The WIHS protocol calls for colposcopy for all women with an abnormal Pap
smear result, but compliance with colposcopy has been incomplete (approximately
70%) in this high-risk population, and there is no central review of histologic
results. For these reasons, we used cytology to define end points in our analysis.
Cytology is primarily subject to false-negative results (ie, it is insensitive),13-16 and
occasional low-grade SILs are known to contain high-grade lesions when evaluated
by biopsy.17-20 Therefore,
we used “any SIL” as our main end point and high-grade SIL or
cancer (HSIL+) as a secondary end point.
HPV DNA testing was conducted using a well-established MY09/MY11 polymerase
chain reaction assay.10,11 HPV
types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, and 73 were considered
oncogenic. All other HPVs, including types 6, 11, 13, 26, 32, 34, 40, 42,
53, 54, 55, 57, 61, 62, 64, 66, 67, 69, 70, 71 (AE8), 72, 81 (AE7), 82 (W13B
and AE2), 83 (PAP291), 84 (PAP155),85 (AE5) 89 (AE6), AE9, and AE10, as well
as all those HPVs that hybridized only with the consensus probe, were considered
Standard life-table methods were used to estimate the cumulative incidence
of any SIL and HSIL+ according to baseline HPV DNA test results, stratified
by HIV serostatus and CD4 T-cell count.21 Cox
models were used to conduct multivariate analyses after demonstration that
the proportional hazard assumption applied to this data set (data not shown).22 Participants who had a hysterectomy (n = 14)
after their baseline visit were censored at the visit before the procedure.
In the analysis of HSIL+, women were additionally censored if they reported
cervical treatment. Treatment of low-grade lesions is at the discretion of
the physician in the WIHS, but high-grade lesions must be treated. For analytic
purposes, we considered as treated any woman who answered affirmatively to
the question, “Since your last study visit, were you treated for any
cervical or other abnormality?” Although self-report is not entirely
reliable, we used this broad definition to be conservative; we could not exclude
the possibility that some women who received treatment privately did so without
this information being captured by our medical record review. In the footnotes
to the tables, we report not only the number censored because of a self-report
of treatment but also the number of these women who had an abnormal biopsy
result (ie, those most likely to have actually received treatment).
P<.05 was set for significance. Analyses
were conducted with SAS version 8.2 (SAS Institute Inc, Cary, NC) and StatXact
version 6.0 (Cytel Software, Cambridge, Mass).
Table 1 shows the baseline characteristics
of the HIV-seropositive and HIV-seronegative women in the WIHS who had normal
cytology at baseline and were included in the current analysis. Mean age was
36 years (range 18-66 years) and 34 years (range 17-55 years) among the HIV-seropositive
and HIV-seronegative women, respectively. Approximately 50% of the participants
categorized themselves as black and 25% to 30% as Hispanic. HIV-seronegative
women were somewhat more likely to be currently sexually active than HIV-seropositive
women. Participants in both groups were followed up for a median of 7 years.
The main analyses were truncated at 5 years, however, because a longer interval
was not thought to be relevant, given the 3-year interval recommended for
Pap-smear screening in HIV-seronegative women.
Among the HIV-seropositive women who had normal cytology results and
were HPV-negative, 9% (95% confidence interval [CI], 1%-18%) with CD4 counts
less than 200/μL, 9% (CI, 4%-13%) with CD4 counts of 200/μL
to 500/μL, and 4% (CI, 1%-7%) with CD4 counts greater than 500/μL developed
any SIL within 2 years (Table 2). The
CIs for these estimates overlapped with that in HIV-seronegative participants
who were HPV-negative (3%; CI, 1%-5%), indicating that after 2 years there
were no large or significant absolute differences in the cumulative incidence
of any SIL between groups (all P>.25).
At 3 years, however, those with less than 200/μL and 200/μL to
500/μL CD4 T cells at baseline had a cumulative incidence of any SIL of
29% (95% CI, 15%-44%) and 14% (95% CI, 8%-20%), respectively. HIV-seropositive/HPV-negative
women with CD4 T cell counts greater than 500/μL continued to have a low
rate of any SIL through 3 years of follow-up (6%; 95% CI, 2%-10%), a cumulative
incidence comparable to that in HIV-seronegative women (5%; 95% CI, 2%-9%).
There were no cases of HSIL or cancer during the first 3 years of follow-up
in any HPV-negative subgroup. In fact, there were no cases of cancer through
the entire 7-year follow-up period.
The risk of any SIL was greater for participants positive for a nononcogenic
HPV type at baseline (Table 3) compared
with that for the HPV-negative women. At 2 years of follow-up, HIV-seropositive
women with a nononcogenic HPV infection and CD4 counts less than 200/μL
had a cumulative incidence rate of 31% (95% CI, 19%-42%) and those with 200/μL
to 500/μL had a rate of 24% (95% CI, 17%-32%). The rate was much lower
among HIV-seropositive women with greater than 500/μL CD4 T cells (6%;
95% CI, 1%-11%) and in HIV-seronegative women (8%; 95% CI, 1%-15%). Few incident
HSILs and no cancers were observed throughout follow-up in the nononcogenic
Using the commercially available test for HPV, HPV test results are
clinically considered negative if no oncogenic HPV type is detected (ie, women
who are HPV DNA negative and those with a nononcogenic HPV type are grouped
together). HIV-seropositive women with CD4 counts less than 200/μL and
200/μL to 500/μL who were negative for an oncogenic HPV had a 2-year
cumulative incidence of any SIL of 21% (95% CI, 13%-29%) and 15% (95% CI,
11%-20%), respectively. HIV-seropositive women with CD4 counts greater than
500/μL had a 2-year rate (5%; 95% CI, 2%-7%) similar to the low rate in
HIV-seronegative women (4%; 95% CI, 2%-7%), and even at 3 years the cumulative
incidence of any SIL was similar in HIV-seropositive women with CD4 counts
greater than 500/μL (9%; 95% CI, 5%-13%) and HIV-seronegative women (6%;
95% CI, 3%-9%). HIV-seropositive women positive for an oncogenic HPV type
at baseline had the highest cumulative incidence of any SIL, and even those
with CD4 counts greater than 500/μL had a markedly higher rate than HIV-seronegative
women (Table 4).
Multivariate Cox models that controlled for age and race/ethnicity largely
corroborated the above results: through 3 years of follow-up, the incidence
of any SIL was similar in HIV-seronegative and HIV-seropositive women with
CD4 counts greater than 500/μL who had negative results for oncogenic HPV
(hazard ratio [HR], 1.4; 95% CI, 0.7-2.7) or all HPV (HR, 1.2; 95% CI, 0.5-3.0).
HIV-seropositive women with CD4 counts of 500/μL or less, however, had
a greater incidence of any SIL relative to HIV-seronegative participants after
just 2 years, even among women with negative test results for all HPVs (HR,
2.9; CI, 1.2-7.1). Last, we measured the strength of association of incident
SIL with nononcogenic (HR, 1.6; 95% CI, 1.2-2.1) and oncogenic (HR, 2.7; 95%
CI, 2.0-3.7) HPV infection at baseline, controlling for HIV serostatus, CD4
T-cell count, age, and race/ethnicity and using all 5 years of data in this
study. For HSIL+, the corresponding HRs were 1.7 (95% CI, 0.3-8.9) for nononcogenic
and 10.2 (95% CI, 2.3-44.5) for oncogenic HPV.
In this observational cohort study, HIV-seronegative and HIV-seropositive
women who had normal cytology results with CD4 counts greater than 500/μL
and who had negative test results for HPV at baseline had a similar low cumulative
incidence of any SIL for 3 years or more. If these findings are confirmed,
comparable risk of cervical lesions would suggest that comparable cancer-screening
practices may be applicable to both groups. Current cervical cancer screening
recommendations for HIV-seronegative women 30 years of age or older advise
using an interval of 3 years between screenings among those who have normal
cytology results and negative test results for oncogenic HPV.1,2 Most
women in our study were aged 30 years or older. However, only an adequately
powered clinical trial can determine whether the cervical cancer screening
strategy in HIV-seronegative women can be safely used in HIV-seropositive
women with CD4 counts greater than 500/μL.
Among women with negative test results for all HPVs, the absolute difference
in the cumulative incidence of any SIL in HIV-seropositive women with CD4
counts of 500/μL or less (9% cumulative incidence) compared with that in
HIV-seronegative women (3% cumulative incidence) was only 6%. There were also
no cases of HSIL+ in any HPV-negative women for 3 or more years. Whether HPV
DNA testing might be useful in determining the appropriate frequency of cervical
cancer screening among HIV-seropositive women with normal cytology results
and CD4 counts of 500/μL or less may also warrant formal evaluation in
a clinical trial.
There has, to our knowledge, been little published about the risk of
SIL among HIV-seropositive women with normal cytology results as stratified
by HPV test results. Ellerbrock et al23 reported
a low cumulative incidence of SIL (approximately 5% at 24 months) among HIV-seropositive
women who were HPV negative at baseline, consistent with our results. Conversely,
Delmas et al24 reported that at 18 months,
23% of HIV-seropositive women who were HPV negative at baseline had developed
a cervical lesion. Younger age (eg, <35 years) is an established risk factor
for incident HPV infection and SIL, and the women in the study by Delmas et
al24 were younger than those in either the
study by Ellerbrock et al23 or our study; the
median age in the study by Ellerbrock et al was approximately 35 years, and
it was 37 years in our study, whereas only a third of women in the study by
Delmas et al were 34 years or older. Nonetheless, given the conflicting reports,
it is appropriate to be cautious in interpreting the findings of our study.
Furthermore, our study has important limitations. First, it is possible
that we have underestimated the rate of neoplasia in the cohort because our
end points of SIL and HSIL+ were determined by using cytology without histologic
confirmation, and there was no end-of-study colposcopy or blind biopsy to
seek occult lesions. However, in a population in which low-grade SIL is uncommon
and there is little or no HSIL detected, we believe that we can be reasonably
certain that few cases of high-grade neoplasia have been missed. Given that
participants were also evaluated every 6 months for more than 5 years, it
would seem unlikely that much significant disease went undetected. A second
major concern is that the polymerase chain reaction assay we used, albeit
well established and widely used, is not a commercial, clinically approved
test.10,11 In fact, in cervical
cancer screening, the HPV DNA test currently approved by the US Food and Drug
Administration is approved only for the detection of oncogenic HPV types.
Therefore, unlike in our study, women who are infected with nononcogenic HPV
would be considered as testing negative with the approved test as it is generally
used. With reagents available from the manufacturer, however, this HPV DNA
test can detect a wide range of oncogenic and nononcogenic HPV types, which
is of interest because our data showed that in HIV-seropositive women with
CD4 counts of 500/μL or less, only those who were negative for both oncogenic
and nononcogenic HPV had low rates of any SIL.
Another possible limitation of our investigation is that we did not
consider changes in HPV and immune status over time in our analyses. The similarity,
though, in the 3-year cumulative incidence rate of any SIL in HIV-seronegative
and HIV-seropositive women with CD4 cell counts greater than 500/μL who
were oncogenic HPV-negative at baseline suggests that this was not a major
concern, at least among HIV-seropositive women who were fairly immunocompetent
at the time of evaluation. In any case, changes in HPV and immune status over
time were not highly relevant to our research question, whether a single initial
HPV test result can be used to determine the appropriate interval between
Pap smears in HIV-seropositive women.
Finally, the most important limitation to our study was its observational
design. Consideration will also need to be given to the psychosocial costs
of a positive HPV test in HIV-seropositive women, many of whom will not develop
SIL. However, according to the results of our study, we believe that the use
of HPV testing in HIV-seropositive women warrants evaluation in a formal clinical
Corresponding Author: Tiffany G. Harris,
PhD, Department of Epidemiology and Population Health, Albert Einstein College
of Medicine, 1300 Morris Park Ave, Belfer 1308A, Bronx, NY 10461 (email@example.com).
Author Contributions: Dr Harris had full access
to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Harris, Burk, Palefsky,
Acquisition of data: Burk, Palefsky, Massad,
Anastos, Minkoff, Bacon, Levine, Strickler.
Analysis and interpretation of data: Harris,
Burk, Palefsky, Massad, Bang, Hall, Bacon, Watts, Silverberg, Xue, Melnick,
Drafting of the manuscript: Harris, Burk, Palefsky,
Bang, Xue, Strickler.
Critical revision of the manuscript for important
intellectual content: Harris, Burk, Palefsky, Massad, Anastos, Minkoff,
Hall, Bacon, Levine, Watts, Silverberg, Xue, Melnick, Strickler.
Statistical analysis: Harris, Burk, Palefsky,
Bang, Hall, Silverberg, Xue, Strickler.
Obtained funding: Burk, Palefsky, Anastos,
Minkoff, Levine, Watts, Strickler.
Administrative, technical, or material support:
Burk, Palefsky, Anastos, Bacon, Levine, Watts, Strickler.
Study supervision: Burk, Palefsky, Watts, Strickler.
Financial Disclosures: None reported.
Funding/Support: HPV DNA testing is funded
through R01-CA-085178. All specimens and other data in this study were collected
by the Women’s Interagency HIV Study (WIHS) Collaborative Study Group
with centers (principal investigators) at New York City/Bronx Consortium (Kathryn
Anastos, MD); Brooklyn, NY (Howard Minkoff, MD); Washington, DC Metropolitan
Consortium (Mary Young, MD); The Connie Wofsy Study Consortium of Northern
California (Ruth Greenblatt, MD); Los Angeles County/Southern California Consortium
(Alexandra Levine, MD); Chicago Consortium (Mardge Cohen, MD); Data Coordinating
Center (Stephen Gange, MD). The WIHS is funded by the National Institute of
Allergy and Infectious Diseases with supplemental funding from the National
Cancer Institute and the National Institute on Drug Abuse (U01-AI-35004, U01-AI-31834,
U01-AI-34994, U01-AI-34989, U01-AI-34993, and U01-AI-42590). Funding is also
provided by the National Institute of Child Health and Human Development (U01-CH-32632)
and the National Center for Research Resources (M01-RR-00071, M01-RR-00079,
Role of the Sponsors: The WIHS is an NIH-funded
multicenter cohort study, and the funding sources had a role in the WIHS study
design; in the collection, analysis, and interpretation of data; and in the
preparation, review, and approval of the manuscript.
Previous Presentation: Presented in part at
the 8th International Conference on Malignancies in AIDS and Other Immunodeficiencies
(ICMAOI): Basic, Epidemiologic and Clinical Research; April 29-30, 2004; Bethesda,