Volume 199, Issue 1 , Pages 26.e1-26.e8, July 2008
Evaluation of primary cervical cancer screening with an oncogenic human papillomavirus DNA test and cervical cytologic findings among women who attended family planning clinics in the United States
Article Outline
Objective
Our goal was to evaluate the performance of screening with (1) Papanicolaou and human papillomavirus (HPV) DNA testing and (2) Papanicolaou testing with reflex HPV testing of atypical squamous cells of undetermined significance for detecting cervical intraepithelial neoplasia grade 3 or more in clinics that serve low-income women in the United States.
Study Design
There were 4799 women who were recruited primarily from Planned Parenthood clinics and who were screened with liquid-based Papanicolaou testing and HPV DNA testing and referred for biopsy based on a positive test result for oncogenic HPV DNA or a Papanicolaou test that showed atypical squamous cells of undetermined significance or more.
Results
Among 931 women who were 30-50 years of age, the sensitivity of reflex HPV testing was 53.8% (range, 38.2%-72.3%). The sensitivity of HPV DNA and Papanicolaou testing was 91% (range, 74.6%-100%). The specificity of reflex HPV testing was 95.1% (range, 93.8%-96.3%). Generally, the specificity of HPV DNA and Papanicolaou testing was low.
Conclusion
Among US women who are ≥30 years old, HPV DNA and Papanicolaou testing is a reasonable cervical cancer screening strategy.
Key words: cervical cancer, human papillomavirus (HPV), screening
Cervical cancer is the second most common cancer in women world wide.1 The mainstay of cervical cancer screening programs has been the Papanicolaou test, which is recommended currently in the United States for women beginning at the age of 21 years or within 3 years of sexual debut.2, 3 The incidence of cervical cancer has decreased substantially in populations in which screening has been used. In the United States, a 70% reduction in cervical cancer incidence and mortality rate has been seen since Papanicolaou screening began.4 However, this screening test has been shown to have severe limitations that include poor sensitivity, positive predictive value, and reproducibility.5, 6
Because of the overwhelming evidence that suggests that human papillomavirus (HPV) infection is a primary cause of cervical cancer,7, 8, 9 many clinicians have advocated the use of HPV testing in screening for high-grade cervical intraepithelial neoplasia (CIN). Currently, the only Food and Drug Administration (FDA)–approved HPV DNA test is the Hybrid Capture 2 (hc2) test (Digene, Gaithersburg, MD), which is a signal amplification test that detects the presence of 13 oncogenic (high-risk) HPV types. The hc2 test is recommended for the triage of women with Papanicolaou diagnoses of atypical squamous cells of undetermined significance (ASC-US), which is a strategy called reflex HPV testing.10 With this framework, a woman is referred to colposcopy if she has cytologic evidence of at least a low-grade intraepithelial lesion (LSIL) or if she has both an ASC-US Papanicolaou test result and a positive hc2 test for oncogenic HPV types. The reflex testing strategy has been shown to have a higher specificity than the repeat Papanicolaou test with a 3- to 6-month interval, but the sensitivity of this test appears to be similar to that of the Papanicolaou test.11
In March 2003, the FDA approved the hc2 test for use in combination with the Papanicolaou test in primary screening of women ≥30 years old. In women younger than 30 years, this testing strategy is thought to have low specificity and low predictive value of a positive test (PVP), which is a phenomenon likely attributable to the high prevalence of HPV infection among women 18-29 years of age. Many published studies that evaluate the performance of screening strategies for cervical cancer have limitations that result from uncontrolled bias, such as verification bias. Specifically, many studies do not refer women with negative screening test results to colposcopy. When test characteristics are calculated with data from only those women with abnormal screening test results, sensitivity estimates are biased upward and specificity cannot be calculated.11, 12
The FDA approval of cotesting with Papanicolaou and HPV DNA points to the need for studies in the United States to evaluate its performance in different clinical settings. The present study was undertaken to evaluate the performance of cotesting with Papanicolaou and HPV DNA for improving screening accuracy for the detection of high-grade CIN or cancer in a population of women ≥30 years old who attended publicly financed clinics in the United States.
Materials and Methods
Study population
Between December 1997 and November 2001, 4799 women who were seen for annual examinations at 1 of 3 Planned Parenthood of Western Washington clinics or to Harborview Medical Center were enrolled in the study. Eligibility requirements for participation in the study included being between the ages of 18 and 50 years, not planning to continue a current pregnancy, and having no history of treatment for cervical neoplasia, hysterectomy, or chronic immune suppression. Women who met these criteria were invited to participate in the study and were enrolled if they provided written informed consent. The University of Washington Institutional Review Board approved all study protocols.
Clinical examinations
At the screening visit, women who were enrolled completed a brief questionnaire that elicited demographic, reproductive, and sexual history information. A pelvic examination was then performed, and screening test samples were taken for cytologic and HPV DNA tests. For cervical cytologic testing, a clinician used an Ayres spatula to collect cells from the transformation zone and a cytobrush to collect cells from the endocervical canal. Both the spatula and brush were rinsed directly in vials that contained 20 mL of cytologic fluid (PreservCyt; Cytyc Corp, Boxborough, MA) for processing and production of a thin-layer cytologic slide (ThinPrep; Cytyc Corp). A cervical specimen for HPV DNA testing was taken by rotating a Dacron-tipped swab in the endocervical os and swabbing it on the ectocervical epithelium. The swab was then placed in specimen transport medium (Digene, Gaithersburg, MD). All specimen transport medium specimens were tested for HPV DNA by polymerase chain reaction (PCR). After January 2000, the residual thin-layer cytologic liquid was used for HPV DNA testing by hc2 (n = 1816 tests). Before January 2000, the hc2 was performed only on screening visit samples taken from the 713 women who attended colposcopy.
Women were referred for colposcopy and biopsy on the basis of any of the following screening test results: (1) cytologic diagnoses of ASC-US, atypical glandular cells, LSIL, or high-grade intraepithelial lesion (HSIL) or higher, (2) a positive PCR test for oncogenic HPV types, or (3) a positive hc2 test result for oncogenic HPV types (starting January 2000). In addition, a random sample of women with normal Papanicolaou and negative HPV DNA test results was also invited to undergo colposcopy and biopsy. Of 2950 women with normal Papanicolaou and negative HPV DNA test results, 242 women (7.7%) had colposcopy performed. All women who were eligible for colposcopy were telephoned a minimum of 3 times and were sent a certified letter if telephone contact failed. The median time between the screening visit and the colposcopy visit was 60 days (range, 6-991 days).
At the colposcopy and biopsy visit, a detailed questionnaire was completed; cervical samples for cytologic and HPV DNA testing were taken according to the same collection protocols that were outlined for the screening visit. Colposcopies were performed by a nurse practitioner with extensive training in colposcopy or by a board-certified gynecologist, both of whom had at least 7 years of clinical practice in colposcopy. The computerized colposcopy system (DenVu, Tucson, AZ) was used to record information and pictures of the cervix. Colposcopically guided biopsies of the worst visible lesions were obtained, or if there was no visible lesion, a biopsy was taken at the 12-o'clock position on the cervix. Endocervical curettage was performed if the colposcopy was unsatisfactory, if the lesions that were seen did not correlate with cytologic results, or if the lesions extended into the endocervical canal. Women with histologic diagnoses of CIN2 or 3 were referred for loop electrosurgical excision procedures. Women with invasive cancer were referred to a gynecologic oncologist for appropriate staging and treatment.
Cytologic and histologic evaluation
Cytologic specimens were processed at Harborview Medical Center, University of Washington. The PreservCyt vial was loaded into the Cytyc ThinPrep Processor and processed according to the manufacturer's specifications. The thin-layer slide was then stained with the Papanicolaou stain, screened by a cytotechnologist (certified by Cytyc Corp), and reviewed by pathologists at Harborview Medical Center (Seattle, WA) who had no knowledge of other laboratory or clinical data. Tests were classified as unsatisfactory based on finding >60% of the slide target area without epithelial cells. Cellular changes were classified according to the Bethesda classification system as negative, ASC-US, atypical glandular cells, LSIL, HSIL, or suspicious of cancer.13 A random 10% sample of all slides that read out as normal was rescreened manually as mandated by federal law.14
Slides of biopsy specimens were reviewed by the pathologist without knowledge of other clinical or laboratory data. Specimens were classified as negative, atypia, CIN grades 1, 2 or 3, carcinoma in situ, adenocarcinoma in situ, microinvasive cancer, or invasive cancer.
HPV DNA testing
The hc2 test was configured to detect in a single assay ≥1 of the following oncogenic HPV types: HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, or 68. Testing of residual PreservCyt liquid was performed on all screening samples beginning in January 2000 and before January 2000 on all screening samples of women who attended colposcopy. A positive result was recorded for specimens with a relative light unit of ≥1, which corresponded to ≥5000 HPV DNA copies.
HPV DNA PCR amplification reactions were performed with 5′ biotinylated MY09, MY11, and HMB01 primers and Amplitaq Gold polymerase (Roche, Emeryville, CA). To prevent PCR product carryover, deoxythymidine triphosphate was replaced by deoxyuridine triphosphate, and uracil-N-glycosylase was added. The human β-globin gene was coamplified in the HPV reaction mix using 5′ biotinylated primers PC04 and GH20 to monitor specimen adequacy. Two microliters of each DNA specimen, which represented 8 μL of original specimen transport medium sample, was added to 100 μL of reaction mix. PCR amplification was carried out in a thermal cycler (Perkin Elmer TC 9600; GMI Inc, Ramsey, MN) with the following profile: 95°C for 9 minutes to activate the Amplitaq Gold; 40 cycles of 95°C for 1 minute, 55°C for 1 minute, and 72°C for 1 minute for each, and a 5-minute terminal extension at 72°C.
HPV DNA typing analysis was performed (according to the manufacturer's specifications) using the reverse-line strip test (Roche) to detect oncogenic HPV 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 55, 56, 58, 59, 68, 73, 82, and 84. All liquid detection reagents were from Amplicor strip detection reagent kits (Roche). The HPV PCR amplification reagents and detection strips manufactured and supplied by Roche, though not commercially available, were extensively quality control tested before use in our study.
Data analysis
Test performance estimates of sensitivity, specificity, the proportion of women who would be referred to colposcopy, PVP and predictive value of a negative test (PVN) were calculated for 4 different screening strategies. These strategies included (1) referral to colposcopy for women with a cytologic diagnosis of ASCUS or greater or a positive hc2 test result; (2) reflex HPV testing, at which time a woman would be referred to colposcopy if she had a cytologic diagnosis of LSIL or greater or if she had an ASC-US Papanicolaou and a positive hc2 test result; (3) the co-testing with Papanicolaou and HPV DNA, at which time a woman would be referred to colposcopy if she had a cytologic diagnosis of HSIL or greater or a positive hc2 test result; and (4) referral based solely on a positive hc2 test result.
To calculate the estimates of test performance, 2 × 2 tables were generated by cross-classifying screening test results and biopsy results. A woman was considered test-positive if she would be referred to colposcopy on the basis of her screening test result; this definition of test positivity varied by screening strategy. All estimates were calculated with either CIN2 or higher or CIN3 or higher as the criterion standard of disease status. For women with missing or inadequate Papanicolaou or HPV DNA test results, screening test values initially were coded as positive. A sensitivity analysis was conducted to determine how coding of missing and inadequate values as negative affected the estimates of test performance.
Verification bias occurs in screening studies when the probability of disease status verification (in our case, biopsy) is dependent on the screening test results. In our study, it appeared that the likelihood of attending the colposcopy visit was dependent on the perceived severity of the screening test result. For example, 88% of women who had an HSIL Papanicolaou and a positive HPV DNA test result for oncogenic HPV types attended the colposcopy visit, as compared with only 7% of women who had normal Papanicolaou and negative HPV DNA test results. To correct for verification bias, 6 screening strata were created based on 6 possible combinations of Papanicolaou and HPV DNA screening test results. The strata included (1) Papanicolaou normal, HPV negative test results; (2) Papanicolaou normal, HPV positive test results; (3) ASC-US/LSIL, HPV negative test results; (4) ASC-US/LSIL, HPV positive test results; (5) HSIL, HPV negative test results; and (6) HSIL, HPV positive test results. Each woman who attended the colposcopy-biopsy visit was assigned to 1 of the strata, based on her screening test results.
To correct for verification bias, estimates of sensitivity, specificity, the proportion of women who would be referred to colposcopy, PVPs, and PVNs were weighted by the probability of attending colposcopy within each stratum. Specifications regarding performance of these calculations have been published previously.11 Corrected estimates were obtained, which represent the performance characteristics of each screening strategy that would have been observed if all screened women had attended the colposcopy visit. An assumption of the correction for verification bias is that the distribution of CIN lesion grades is the same among women in the same stratum, independent of whether they attended the colposcopy visit. This assumption is plausible because no factors other than Papanicolaou and HPV DNA test results were associated independently with both disease status and attendance at the colposcopy visit.
Comparisons between proportions were made with the χ2 test, SEs, and 95% CIs for corrected estimates were calculated with the use of bootstrap resampling in which the sampling fractions were considered fixed.15 All analyses were conducted using Stata software (versions 7.0 and 8.0; Stata Corp, College Station, TX).
Results
Study population characteristics
Between December 1997 and November 2001, 4799 women were recruited in the study and had a median age of 23 years. The women were predominantly white (72%) and had a median lifetime number of sex partners of 6 (range, 0-500). Of enrolled women, 19% (n = 931) were between the ages of 30 and 50 years (median, 34 years). Population characteristics are reported in Table 1 by age. Compared with women under the age of 30 years, women aged ≥30 years reported a greater number of lifetime sex partners (median, 9 vs 5; χ2(5) = 171.9; P < .001) were more likely to be parous (χ2(7) = 388.7; P < .001), had higher monthly incomes (χ2(5) = 178; P < .001), and were more likely to have been educated beyond high school (χ2(1) = 27.1; P < .001). Older and younger women were similar in terms of race/ethnicity and current oral contraceptive use.
TABLE 1. Characteristics of the women who were screened (n = 4799)
| Characteristic | Age ≥30 y (n = 931) | Age <30 y (n = 3868) | ||
|---|---|---|---|---|
| n | % | n | % | |
| Race/ethnicity | ||||
 African American | 122 | 13.1 | 376 | 9.7 |
| American Indian | 14 | 1.5 | 50 | 1.3 |
| Asian | 17 | 1.8 | 139 | 3.6 |
| Hispanic | 33 | 3.5 | 143 | 3.7 |
| White | 686 | 73.7 | 2772 | 72.7 |
| Other | 59 | 6.3 | 388 | 10.0 |
| Lifetime sex partners (n) | ||||
| 0 | 3 | 0.3 | 5 | 0.1 |
| 1 | 45 | 4.9 | 516 | 13.4 |
| 2 | 34 | 3.7 | 381 | 9.9 |
| 3-5 | 220 | 23.8 | 1103 | 28.6 |
| 6-9 | 165 | 17.9 | 709 | 18.4 |
| ≥10 | 456 | 49.4 | 1150 | 29.8 |
| Parity (n) | ||||
| 0 | 360 | 38.7 | 2592 | 67.1 |
| 1 | 223 | 24.0 | 756 | 19.6 |
| 2 | 200 | 21.5 | 349 | 9.0 |
| ≥3 | 147 | 6.3 | 168 | 4.3 |
| Currently using oral contraceptives | ||||
| Yes | 351 | 37.7 | 1373 | 35.5 |
| No | 580 | 62.3 | 2491 | 64.5 |
| Education | ||||
| ≤High school | 302 | 32.5 | 2252 | 58.2 |
| >High school | 628 | 67.5 | 1616 | 41.8 |
| Monthly income ($) | ||||
| ≤400 | 153 | 16.6 | 1061 | 27.6 |
| 401-800 | 164 | 17.8 | 1068 | 27.8 |
| 801-1300 | 233 | 25.3 | 911 | 23.7 |
| >1300 | 371 | 40.3 | 803 | 20.9 |
Screening test results
The screening thin-layer Papanicolaou test results of 891 women were abnormal (18.6%), and oncogenic HPV types were detected by hc2 in 29.2% of women. Women aged ≥30 years had a lower prevalence of oncogenic HPV by hc2 than women under age 30 years (21% vs 30%) and were more likely to have normal Papanicolaou test results (86% vs 80%). Screening test results are presented by age in Table 2. Women with more severe cytologic diagnoses were more likely to have positive hc2 test results for oncogenic HPV DNA than women with less severe cytologic diagnoses. The proportions of women who were oncogenic HPV positive by hc2 for different cytologic diagnoses are presented in Table 3 by age.
TABLE 2. Screening test results by age
| Test | Age ≥30 y (n = 931) | Age <30 y (n = 3868) | ||
|---|---|---|---|---|
| n | % | n | % | |
| Thin-layer Papanicolaou test | ||||
| HSIL | 21 | 2.1 | 115 | 3 |
| LSIL | 9 | 1.0 | 180 | 4.7 |
| ASC-US | 78 | 8.4 | 389 | 10.1 |
| Atypical glandular cells | 2 | 0.2 | 2 | 0.1 |
| AIS | 0 | 0 | 1 | 0.03 |
| CIS | 2 | 0.21 | 2 | 0.05 |
| Normal | 798 | 85.6 | 3107 | 80.3 |
| Unsatisfactory | 21 | 2.3 | 72 | 1.9 |
| Test for HPV DNA by signal amplificationa | ||||
| Positive for high-risk types | 80 | 20.9 | 451 | 30.3 |
| Negative | 293 | 76.7 | 992 | 66.6 |
| Indeterminate ratio | 1 | 0.3 | 6 | 0.4 |
| Insufficient | 8 | 2.1 | 40 | 2.7 |
aBased on a sample of women who were screened by signal amplification beginning in January 2000. |
TABLE 3. Distribution of hc2 positive test results for oncogenic HPV DNAa by cytological result and age
| Age ≥30 y | Age <30 y | |||
|---|---|---|---|---|
| hc2+ | hc2+ | |||
| n | % | n | % | |
| Papanicolaou diagnosis | ||||
| Normal | 52 | 16.4 | 257 | 22.7 |
| ASC-US | 12 | 40 | 79 | 48.5 |
| Atypical glandular cells | 0 | 0 | 2 | 100 |
| LSIL | 3 | 75 | 55 | 93.2 |
| HSIL | 11 | 100 | 50 | 89.3 |
| Inadequate | 1 | 11 | 7 | 25 |
aBased on a sample of women that began in January 2000, when hc2 became part of the screening protocol. |
Histologic results
Histologic diagnoses were available for 1241 of 1251 women who attended colposcopy (6 women did not undergo biopsy, and 4 samples were considered unsatisfactory for evaluation). Of the women who had available histologic diagnoses, 231 women had results of CIN2 or higher, and 147 women had results of CIN3 or higher. Histologic results are reported by age in Table 4. The verification bias corrected prevalence of biopsy-confirmed CIN3 or higher was 4.5% in women ≥30 years old and 4.7% in women <30 years old. The corrected prevalence of CIN2 or higher was 7.6% in women ≥30 years old and 8.7% in women <30 years old.
TABLE 4. Histologic results for 1241 women who underwent colposcopy and biopsy and who had available diagnoses, by age
| Histologic result | Age ≥30 y | Age <30 y | ||
|---|---|---|---|---|
| n | % | n | % | |
| CIN1 | 10 | 4.5 | 163 | 16.0 |
| CIN2 | 12 | 5.4 | 72 | 7.1 |
| CIN3 | 28 | 12.6 | 119 | 11.6 |
| Atypia | 26 | 11.7 | 156 | 15.3 |
| Normal | 146 | 65.5 | 509 | 49.8 |
Screening strategy performance
For each of the 5 screening strategies that were evaluated, sensitivities tended to be lower and specificities tended to be higher for women aged ≥30 years than for women under age 30 years. The screening test performance characteristics for detection of biopsy-confirmed CIN3 are presented by age in the upper one-half of Table 5. Reflex HPV testing had the lowest sensitivity for detection of CIN3 of any of the 5 combined HPV DNA and Papanicolaou testing strategies. In women aged 30-50 years, the sensitivity of reflex HPV testing was 53.8% (95% CI, 38.2%-72.3%). However, the specificity of reflex HPV testing in women aged ≥30 years was 95.1% (95% CI, 93.8%-96.3%), which was the highest specificity for any of the screening strategies evaluated. The cotesting with Papanicolaou and HPV DNA had a sensitivity of 91% (95% CI, 74.6%-100%) in women aged ≥30 years, regardless of whether the cut off for a positive test was defined as (1) hc2+ or ASC-US or greater or (2) hc2+ or HSIL or greater. However, the specificity and PVP were higher and the proportion of women who would be referred to colposcopy was lower with the HSIL or greater cutoff for defining a positive cotesting with Papanicolaou and HPV DNA test result. The sensitivity of the hc2 test alone for the detection of CIN3 or higher was 88.2% (95% CI, 69.9%-97.3) in women aged ≥30 years. The specificity proportion of women who would be referred to colposcopy, PVP, and PVN of the hc2 test result alone were similar to that of the cotesting with Papanicolaou and HPV DNA test with the use of a cut off of HSIL or greater on cytologic testing or a positive hc2 test result as the definition of a positive test. Using CIN2 or higher as the criterion standard for disease status resulted in substantially lower sensitivities and marginally higher specificities for each of their 5 screening strategies. Performance of the screening strategies for detection of CIN2 or higher as the criterion standard of disease status is reported in the lower one-half of Table 5.
TABLE 5. Corrected sensitivity, specificity, percentage of referral to colposcopy, PVP, and PVN estimates for detection of CIN3 in the total population of women and by age ≥ 30 and age <30
| Screening strategy (definition of a test +) | Corrected sensitivity (%) | Corrected specificity (%) | Referral to colpobiopsy (%) | PVP (%) | PVN (%) |
|---|---|---|---|---|---|
| CIN3 or more | |||||
| Cotesting with Papanicolaou and HPV DNA: hc2+ or ≥ASC-US | |||||
| Age ≥30 y | 91.0(74.6-100) | 71.6(66.3-75.8) | 31.2 | 13.2 | 99.4 |
| Age <30 y | 95.8(87.7-98.7) | 62.0(58.9-64.5) | 40.7 | 11.1 | 99.7 |
| Cotesting with Papanicolaou and HPV DNA: hc2+ or ≥HSIL | |||||
| Age ≥30 y | 91.0(74.6-100) | 79.7(73.9-83.7) | 23.5 | 17.5 | 99.5 |
| Age <30 y | 92.2(83.6-96.5) | 69.0(65.9-71.5) | 33.9 | 12.8 | 99.4 |
| Reflex HPV testing: hc2+ and ASC-US or ≥LSIL | |||||
| Age ≥30 y | 53.8(38.2-72.3) | 95.1(93.8-96.3) | 7.1 | 34.3 | 97.8 |
| Age <30 y | 60.0(50.4-69.7) | 88.0(87.1-88.8) | 14.3 | 19.8 | 97.8 |
| Hc2 followed by Papanicolaou testing: hc2+ and ≥ASC-US | |||||
| Age ≥30 y | 51.1(35.2-69.6) | 96.0(94.8-96.9) | 6.1 | 37.5 | 97.6 |
| Age <30 y | 58.1(48.6-67.7) | 89.2(88.3-89.9) | 13.1 | 20.9 | 97.7 |
| Hybrid capture 2 hc2+ for ≥1 of 13 oncogenic HPV types | |||||
| Age ≥30 y | 88.2(69.9-97.3) | 79.9(74.2-83.9) | 23.1 | 17.2 | 99.3 |
| Age <30 y | 90.3(81.7-95.0) | 69.3(66.1-71.8) | 33.5 | 12.7 | 99.3 |
| CIN2 or more | |||||
| Cotesting with Papanicolaou and HPV DNA: hc2+ or ≥ASC-US | |||||
| Age ≥30 y | 74.1(43.0-92.7) | 72.4(67.3-76.7) | 31.2 | 18.1 | 97.1 |
| Age <30 y | 83.6(68.9-93.6) | 63.4(60.2-66.1) | 40.7 | 17.8 | 97.6 |
| Cotesting with Papanicolaou and HPV DNA: hc2+ or ≥HSIL | |||||
| Age ≥30 y | 71.5(41.9-90.4) | 80.5(75.0-84.8) | 23.5 | 23.3 | 97.2 |
| Age <30 y | 80.4(65.8-90.7) | 70.5(67.3-73.2) | 33.9 | 20.6 | 97.4 |
| Reflex HPV testing: hc2+ and ASC-US or ≥ LSIL | |||||
| Age ≥30 y | 42.9(26.6-59.0) | 95.9(94.6-97.2) | 7.1 | 46.3 | 95.3 |
| Age <30 y | 50.8(40.1-58.8) | 89.2(88.4-90.2) | 14.3 | 30.9 | 95.0 |
| hc2 followed by Papanicolaou testing: hc2+ and ≥ASC-US | |||||
| Age ≥30 y | 41.3(24.9-57.0) | 96.8(95.7-97.8) | 6.1 | 51.3 | 95.2 |
| Age <30 y | 48.7(38.8-57.2) | 90.3(89.5-91.2) | 13.1 | 32.4 | 94.9 |
| Hybrid capture 2: hc2+ for ≥1 of 13 oncogenic HPV types | |||||
| Age ≥30 y | 69.9(38.8-88.4) | 80.7(75.3-85.1) | 23.1 | 23.1 | 97.0 |
| Age <30 y | 78.3(64.3-88.6) | 70.7(67.5-73.4) | 33.5 | 20.3 | 97.2 |
Comment
This study is among the first evaluations of the performance of cotesting with Papanicolaou and HPV DNA tests in primary screening of US women. Although the test has been approved by the FDA for use in primary screening of women aged ≥30 years, approval was based largely on data from other countries, where screening algorithms and cytologic diagnoses are not the same as those used in US clinics. We evaluated the performance of cotesting with Papanicolaou and HPV DNA test using 2 different definitions of a positive test and compared cotesting with Papanicolaou and HPV DNA tests to the currently recommended use of the hc2 test in reflex HPV testing of ASC-US Papanicolaou smears.
Although the Papanicolaou smear has been the hallmark of cervical cancer screening for the past 50 years, it has some severe limitations. Approximately one-half of all cervical cancer cases occur in women who have been screened within 5 years of diagnosis.16 One of the main concerns with the Papanicolaou test is its low sensitivity. In a previous analysis of the women who were enrolled in our study, the thin-layer Papanicolaou test alone, with the use of a cut off for test positivity of ASC-US or greater, was found to have a sensitivity of 49.7% (range, 32.9%-71.5%) for detection of biopsy-confirmed CIN3 in women aged ≥30 years.11 Reflex HPV testing uses a Papanicolaou test as the primary screen and performs HPV DNA testing only in secondary triage of women who have ASC-US Papanicolaou test results. However, this screening strategy does not correct the problem of false negatives that plagues cytologic diagnosis. Our data suggest that the sensitivity of reflex HPV testing is 54%. Thus, almost one-half of all cases of CIN3 would be missed with the use of either the Papanicolaou test alone or reflex HPV testing of women with ASC-US Papanicolaou test results.
The performance of the evaluated testing strategies differs from the performance of the same tests as reported by Cuzick et al,17 who used data from the British study titled “HPV in Addition to Routine Testing (HART).” For example, our estimates of the sensitivity of the hc2 test are much lower than those found in the HART study (69.9% vs 97.1% sensitive for detection of CIN2 or greater). The differences in test performance could be attributable partly to differences in the British and US clinical populations of older women, the diagnostic criteria used for cytologic and histologic information, and overall study design. In the HART study, women were referred to colposcopy on the basis of a positive Papanicolaou test or a positive cervical hc2 test result. If a woman tested negative on both the hc2 test and cytologic examination, she was considered to be CIN2 negative. In our study, women were referred to colposcopically guided biopsy on the basis of an abnormal Papanicolaou test and a positive test result from at least 1 of ≥2 HPV DNA tests (hc2 from cervical or vaginal samples or PCR from cervical, vaginal, or urine samples). When our data were evaluated using biopsy data from women who were referred to colpobiopsy based solely on an abnormal Papanicolaou or positive cervical hc2 test result, the performance of the hc2 test improved. Specifically, the sensitivity of the hc2 test for detection of CIN2 or greater increased from 69.9%-77.6%, and the sensitivity of the hc2 test for detection of CIN3 or greater increased from 88.2%-96.9%. Another potentially important difference between the 2 studies was the protocol for biopsy. In the British study, biopsies were performed if any abnormality was seen on colposcopic examination. Our study protocol specified that biopsies of the worst visible lesions were to be obtained or, if there was no visible lesion, that a biopsy was to be taken at the 12 o'clock position on the cervix.
Our results suggest that, in women aged ≥30 years, cotesting with Papanicolaou smear and HPV DNA was more sensitive than reflex HPV testing for the detection of high-grade cervical lesions (91% vs 54%), provided women with a positive hc2 test and negative Papanicolaou test were referred to colposcopy and biopsy. The sensitivity of cotesting with Papanicolaou smear and HPV DNA was 91%, regardless of whether the definition of a positive test was (1) hc2+ or HSIL or greater or (2) hc2+ or ASC-US or greater. Although, theoretically, the sensitivity of the test that uses a cutoff of ASC-US should be higher than the use of the HSIL cut off, all women between the ages of 30 and 50 years in our population who had ASC-US/LSIL and CIN3 also had positive hc2 test results. Furthermore, cotesting with Papanicolaou smear and HPV DNA and referring women to colposcopy based on a positive hc2 test or at least an HSIL Papanicolaou test was more specific, had a higher PVP, and would refer fewer women to colposcopy than cotesting with Papanicolaou and HPV DNA that would refer women to colposcopy based on a positive hc2 test or at least an ASC-US Papanicolaou test.
The relatively high sensitivity of co-testing with Papanicolaou and HPV DNA tests is offset by the fact that the PVP was 17.5%, which suggests that 81.5% of women who test positive would be false-positive results. Furthermore, almost one-quarter of all women who were screened would be referred to colposcopy with the use of cotesting with the Papanicolaou and HPV DNA screening strategy. Our data indicate that only 7% of women would be referred for colposcopy based on screening with reflex HPV testing, which had a PVP of 34% and suggests that 56% of women who screened positive would be false positive. On the other hand, the PVN was 97.8% for reflex HPV testing, compared with a PVN of 99.5% for the cotesting with Papanicolaou and HPV DNA tests. This suggests that of 100,000 women who screen negative, 500 cases of CIN3 or higher would be missed by the cotesting with Papanicolaou and HPV DNA test; reflex HPV testing would miss 2200 cases of CIN3. So for every case of CIN3 or higher that would be missed by the co-testing with Papanicolaou and HPV DNA tests, 4.4 cases would be missed by reflex HPV testing.
Cotesting with Papanicolaou and HPV DNA tests requires more financial resources than the Papanicolaou smear or reflex HPV testing of those with ASC-US Papanicolaou test results. Referring a larger proportion of women to colposcopy will further increase costs. However, most women will test negative on both the HPV DNA test and the Papanicolaou test. Of our population of women (30-50 years old) who attended publicly financed clinics, approximately 70% of the women tested negative on both the Papanicolaou and hc2 tests. Extending the screening interval for these women to 3 years, as recommended,2 and follow-up cotesting in 6-12 months of women with positive HPV DNA test results and negative Papanicolaou test results with referral to colposcopy and biopsy of only those with repeated abnormal Papanicolaou test results or positive HPV DNA test results3 has been shown to offset any excess costs that are accrued by the combined test and the increase in colposcopic procedures.18
Although this may be the case among women in general screening populations, it might not hold true for any given clinic population, such as the family planning clinic population that was included in our study or in clinics that serve high-risk populations where the prevalence of HPV is higher than that of the general population.
The prevalence of oncogenic HPV in our family planning clinic population was 21%, which was higher than estimates from other screening populations in the United States.19 In such populations, in which a greater proportion of women would be referred for colposcopy and biopsy, it is not clear that primary HPV screening would be a cost-effective strategy, even with an extended screening interval. Alternatively, strategies that involve primary HPV DNA testing followed by Papanicolaou testing of positive results with immediate referral to colposcopy and biopsy of double positive results and repeat Papanicolaou testing in 12 months of those women with positive results on only the initial HPV DNA test might be most cost effective, especially with an extended screening interval.
References
- . Cancer incidence in five continents (vol VII). Lyon, France: IARC; 1997;
- American Cancer Society guideline for the early detection of cervical neoplasia and cancer. CA Cancer J Clin. 2002;52:342–362
- . Clinical management guidelines for obstetrician-gynecologists: ACOG Practice Bulletin No.: 61: human papillomavirus. Obstet Gynecol. 2005;105:905–918
- . Cancer statistics, 1999. CA Cancer J Clin. 1999;48:6–29
- . Comparison of three management strategies for patients with atypical squamous cells of undetermined significance: baseline results from a randomized trial. J Natl Cancer Inst. 2001;93:293–299
- . Interobserver reproducibility of cervical cytologic and histologic interpretations: realistic estimates from the ASCUS-LSIL Triage Study. JAMA. 2001;285:1500–1505
- Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol. 1999;189:12–19
- Prevalence of human papillomavirus in cervical cancer: a worldwide perspective: international biological study on cervical cancer (IBSCC) study group. J Natl Cancer Inst. 1995;87:796–802
- Epidemiologic evidence showing that human papillomavirus infection causes most cervical intraepithelial neoplasia. J Natl Cancer Inst. 1993;85:958–964
- 2001 Consensus guidelines for the management of women with cervical cytological abnormalities. JAMA. 2002;287:2120–2129
- Evaluation of human papillomavirus testing in primary screening for cervical abnormalities: comparison of sensitivity, specificity, and frequency of referral. JAMA. 2002;288:1749–1757
- Shanxi Province Cervical Cancer Screening Study: a cross-sectional comparative trial of multiple techniques to detect cervical neoplasia. Gynecol Oncol. 2001;83:439–444
- The 2001 Bethesda system: terminology for reporting results of cervical cytology. JAMA. 2002;287:2114–2119
- Center for devices and Radiological Health, US Food and Drug Administration. Clinical Laboratory Improvement Methods; 2002. Accessed January 14, 2008. Available at: http://wwwn.cdc.gov/clia/chronol/aspy
- An introduction to the bootstrap. New York: Chapman and Hall; 1993;436.
- . Estimating the efficacy of screening by auditing smear histories of women with and without cervical cancer: the National Co-ordinating Network for Cervical Screening Working Group. Br J Cancer. 1996;73:1001–1005
- Management of women who test positive for high-risk types of human papillomavirus: the HART study. Lancet. 2003;362:1871–1876
- . Cost-effectiveness of human papillomavirus DNA testing for cervical cancer screening in women aged 30 years or more. Obstet Gynecol. 2004;103:619–631
- Baseline cytology, human papillomavirus testing, and risk for cervical neoplasia: a 10-year cohort analysis. J Natl Cancer Inst. 2003;95:46–52
Cite this article as: Baseman JG, Kulasingam SL, Harris TG, et al. Evaluation of primary cervical cancer screening with an oncogenic human papillomavirus DNA test and cervical cytologic findings among women who attended family planning clinics in the United States. Am J Obstet Gynecol 2008;199:26.e1-26.e8.
PII: S0002-9378(07)02235-1
doi:10.1016/j.ajog.2007.11.051
© 2008 Mosby, Inc. All rights reserved.
Volume 199, Issue 1 , Pages 26.e1-26.e8, July 2008
