The Atypical Squamous Cells of Undetermined Significance–Low-Grade Squamous Intraepithelial Lesion (ASCUS–LSIL) Triage Study (ALTS) was launched to provide empirical evidence for the best management of women with minimally abnormal cervical tests. This randomized controlled trial showed that among women with atypical squamous cells of undetermined significance (ASCUS), human papillomavirus (HPV) testing for carcinogenic genotypes (“HPV triage”) was at least as sensitive for identifying women with cervical intraepithelial neoplasia grade 3 and worse lesions (CIN3 or worse) and referred half as many women to colposcopy, compared with universal immediate colposcopy1. As a corollary, Castle et al2 showed that in ALTS, among carcinogenic HPV-negative ASCUS women there was 3.0% and 1.4% absolute risk for CIN2 or more severe (CIN2 or worse) and CIN3 or worse diagnoses, respectively, over 2 years of follow-up, indicating the low risk of precancer and cancer among women negative for carcinogenic HPV. Moreover, in screening cohorts with longer follow-up, the absolute risk of CIN3 or worse associated with a negative HPV test has been shown to be very low (0.9%) compared with 6.9% over a 10-year follow-up period for women with a positive HPV test.3,4
Given the low risk of incipient CIN3 or worse, the optimal management of women with HPV-negative ASCUS is unclear. The problem is substantial in absolute numbers. In the United States, between 2.5 and 3 million (about 5%) of the 50–60 million Pap tests performed annually are reported as ASCUS.5,6 Triage by HPV testing is increasingly common, and of the women with ASCUS, approximately half, or more than 1 million women, per year would test HPV negative if all were triaged.7 The current clinical management guideline for women with HPV-negative ASCUS is to repeat cytology at 12 months.8 Using the ALTS data, we investigated the relative performances of cytology or HPV DNA testing or both at the 12-month follow-up visit for management of ASCUS carcinogenic HPV-negative women, with reference to risk of subsequent detection of cervical precancer.
MATERIALS AND METHODS
The data to address the aims of this analysis are from ALTS, described in detail elsewhere.9 Briefly, ALTS was a randomized clinical trial conducted at four clinical centers with enrollment of 5,060 women during 1996–1998. The study was conducted with the approval by the local and National Cancer Institute Institutional Review Boards. Nonpregnant women, aged 18 years or older, with cytologic diagnosis of ASCUS or low-grade squamous intraepithelial lesion (LSIL) within 6 months of the study enrollment, with no prior hysterectomy, no known history of ablative or excisional therapy to the cervix, and able to provide informed consent were eligible to participate in the ALTS trial. The primary objective of ALTS was to compare three management strategies for women with ASCUS (n=3,488) or LSIL (n=1,572): first, in the immediate colposcopy arm women were referred for colposcopy regardless of enrollment cytology or HPV result; second, HPV triage with referral to colposcopy if the enrollment HPV result was positive or missing, or cytology was high-grade squamous intraepithelial lesion (HSIL); and last, conservative management with referral to colposcopy if cytology at enrollment was HSIL. Women were followed up every 6 months for 2 years, with cytology and masked HPV testing at each visit. High-grade squamous intraepithelial lesion led to colposcopic referral, and women with histologic CIN2 or worse were offered treatment by loop electrosurgical excision procedure. At the exit visit, colposcopy was performed on all women who came to that visit regardless of randomization arm or prior procedures. At exit, the treatment threshold was lowered to include women with cytologic or histologic evidence of a persistent low-grade lesion. Clinical center pathologists’ interpretations were used for clinical management. An expert Pathology Quality Control Group, masked to other test results, provided independent reviews of histology outcomes.
At referral into the ALTS study and at subsequent follow-up visits, a PreservCyt (Cytyc Corp., Marlborough, MA) cytology specimen was collected on all women and prepared as previously described.9 Cytologic interpretations of the ThinPrep (Cytyc Corp.) by clinical centers were categorized according to the 1991 Bethesda System as negative, ASCUS, LSIL, or HSIL, independent of HPV results and clinical data.
Human papillomavirus testing was performed with Hybrid Capture 2 (Digene Corporation, Gaithersburg, MD) using probe set B on residual PreservCyt (Cytyc) aliquots. Hybrid Capture 2 is an FDA-approved, commercially available HPV test that collectively targets 13 carcinogenic HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68) without distinguishing specific types. At enrollment, adequate Hybrid Capture 2 results were available from 3,324 of 3,488 (95%) women with ASCUS.9
First, general descriptive statistics were performed to compare all ASCUS women with and without carcinogenic HPV at enrollment (referred to henceforth as HPV positivity). For categorical variables, two-way tabulation was used to calculate Pearson's χ2 test. For continuous variables, nonparametric Wilcoxon signed rank test for differences between medians was used as a test of statistical significance. We determined independent correlates of HPV positivity using multivariable logistic regression. We also investigated age-stratified, 2-year cumulative CIN2 or worse as diagnosed by the clinical center pathologists and a more stringent outcome of CIN3 or worse as diagnosed by the Pathology Quality Control group, among HPV-positive and -negative women with ASCUS.
Because the current clinical management guideline of women with HPV-negative ASCUS is repeat cytology at 12 months, we further estimated the absolute risk of developing CIN2 or worse, or CIN3 or worse lesions based on either a single Hybrid Capture 2 test for carcinogenic HPV, or cytology at an ASCUS threshold, or simultaneous testing with both at 12 months. An outcome of subsequently detected CIN3 or worse from the 12-month to the 24-month (exit) visit was ascertained. Using standard methods, we then evaluated sensitivity, specificity, positive and negative predictive values (PPV and NPV, respectively) for each test, and referral rates for predicting CIN3 or worse (diagnosed by Pathology Quality Control) during the 12–24 month follow-up period. McNemar's test was used to evaluate differences in the sensitivity and specificity (exact test was used for evaluating sensitivity). Differences between tests in PPV and NPV were tested using the generalized estimating equation-based method by Leisenring et al,10 which is an analog to McNemar's test that conditions on the basis of test outcomes rather than disease outcome. A χ2 test was used to evaluate differences between referral rates. Further, we present data using the less stringent outcome of clinical center diagnosis of CIN2 or worse to reflect the clinical threshold for treatment.
Of the 3,488 women with a referral ASCUS diagnosis in ALTS, 162 (5%) had inadequate Hybrid Capture 2 tests at the enrollment visit. Of the remaining 3,326, 47% (1,559) were HPV negative by the Hybrid Capture 2 test.
Table 1 presents distribution of selected factors comparing HPV-positive ASCUS to HPV-negative ASCUS participants. Human papillomavirus–negative women were older (median of 31 compared with 23 years, Wilcoxon P<.001), older at sexual debut (median of 17 compared with 16 years, Wilcoxon P<.001), had fewer lifetime sexual partners (P<.001), and fewer recent sexual partners (P<.001) compared with HPV-positive women. They were also less likely to have used oral contraceptive in the past 2 years (48% compared with 65%, P<.001), less likely to be smokers (27% compared with 36%, P<.001), and more likely to be enrolled at the Pittsburgh clinical site (P<.001). In multivariable analysis—after adjusting for clinical center—being older, initiating sex at an older age, and fewer numbers of lifetime and recent sexual partners were independently associated with being HPV-negative (Table 1).
Over the 2 years of ALTS among all women with HPV-negative ASCUS, 1.4% (22 of 1,559) were diagnosed with CIN3, compared with 15.2% (269 of 1,767) of women with HPV-positive ASCUS (P<.001). There were no cancers diagnosed among HPV-negative ASCUS women, whereas there were two cancers diagnosed among HPV-positive ASCUS women. Similarly, far fewer women with HPV-negative ASCUS had CIN2 or worse over their follow-up compared with HPV-positive ASCUS (3.0% compared with 26.2%, respectively, P<.001). Stratifying by age, both CIN3 or worse and CIN2 or worse diagnoses were less common for all age categories among women with HPV-negative ASCUS compared with HPV-positive ASCUS (Table 2).
We then investigated the theoretical performance (sensitivity, specificity, PPV, and NPV) of triage tests conducted at the 12 months follow-up of women with HPV-negative ASCUS for identifying women who will be diagnosed with CIN3 or worse. We limited our analytic group to women in the conservative management and HPV triage arms who did not have colposcopy before their 12-month visit to best simulate current management guidelines. Further, 238 participants with a missing Hybrid Capture 2 test or cytology at the 12-month visit were also excluded (Fig. 1). Compared with the 795 women in the analytic sample, excluded women were younger, reported more recent (past year) number of sexual partners, and were more likely to be current smokers. They were similar in their report of lifetime sexual partners, oral contraceptive use, age of sexual initiation, and subsequent CIN2 or worse or CIN3 or worse diagnosis.
There were no cases of cancer, 0.9% (7 of 795) cases of CIN3 and 2.3% (18 of 795) CIN2 or worse diagnosed between the 12- and 24-month visits. Test characteristics for the different management strategies at 12 months for subsequently diagnosed CIN3 during follow-up among 795 participants are presented in Table 3. Based on a threshold of ASCUS, cytology was 86% (6 of 7 cases) sensitive and 71% (560 of 788 noncases) specific for detecting subsequently diagnosed CIN3. By comparison, Hybrid Capture 2 was 71.4% sensitive (5 of 7 cases) and 84% (661 of 788 noncases) specific, more specific than cytology (P<.001). The absolute risk (equal in this instance to the clinical term of “positive predictive value”) for CIN3 was extremely low for both cytology and Hybrid Capture 2 positive test results (2.6% or 6 of 234 cases, and 3.8% or 5 of 132 cases, respectively, P=.2), because these women were HPV negative at study entry, leading to a generally low risk of CIN3 or worse regardless of the 12-month result. Cotesting with a combination of cytology and Hybrid Capture 2 at 12 months was 100% (7 of 7 cases) sensitive for detection of CIN3, but cotesting resulted in low specificity (61% or 480 of 788 noncases) and would have referred a high percentage of women to colposcopy (40% or 315 of 795). The absolute risk (PPV) of subsequent CIN3 was only 2.2% (7 of 315) among women who were either Hybrid Capture 2-positive or ASCUS-positive.
Test characteristics of different management strategies at the 12-month follow-up visit for subsequent detection of CIN2 or worse are also shown in Table 3. Hybrid Capture 2 or cytology at an ASCUS threshold had the same sensitivity (67% respectively, P=1.0), but Hybrid Capture 2 had significantly higher specificity than cytology (85% compared with 71%, respectively, P<.001) with lower referral rates (17% compared with 29%, respectively, P<.001). The absolute risk (PPV) of subsequent CIN2 or worse in the next 12 months following an ASCUS or worse cytology (5.1%) or Hybrid Capture 2 positive test (9.1%, P=.007) was low.
Comparing the performance of 12-month test strategies for subsequent detection of CIN3 and CIN2 or worse, we observed that sensitivity declined for CIN2 or worse compared with the more stringent outcome of CIN3 for all management modalities, and PPV increased. This would be expected based on the higher prevalence of CIN2 or worse compared with CIN3, rather than a reflection of the management strategies.
This analysis showed very low cumulative incidence of CIN3 (1.4%) and CIN2 or worse (3.0%) and no cancers among women with HPV-negative ASCUS over the 2-year study period. This is similar to other studies that have also shown low precancer risk among women with HPV-negative ASCUS despite methodologic differences.11–14 Given the low risk of precancer and cancer it might be appropriate to return women with HPV-negative ASCUS to routine screening intervals, which may be longer than 1 year depending on age and previous screening history, rather than triggering more intensive follow-up.15
However, even though the risk of subsequent precancer is very low among HPV-negative ASCUS women, the knowledge of an “abnormal” cytology may compel clinicians to more intensive follow-up. The question is then what are the best strategies for screening after HPV-negative ASCUS. This analysis showed that testing for carcinogenic HPV had significantly higher specificity compared with cytology alone (using an ASCUS threshold) for detecting subsequent CIN3, or CIN2 or worse lesions, with significantly lower referral rates. In our analysis, although the overall sample size was fairly large, because risk of disease is very low among HPV-negative ASCUS, we had insufficient number of outcomes (CIN3) to confidently assess differences in sensitivity among 12-month testing strategies. Because there may not be a large enough cohort with adequate follow-up to address differences in sensitivities at this time, it may be appropriate for the research and clinical community to pool data across studies to address this question.
Cotesting with both cytology and HPV testing was attractive in that all subsequent cases of CIN3 were detected, but the substantial disadvantage of this approach was lower specificity and a higher referral rate compared with a single test with either Hybrid Capture 2 or cytology. Women positive by either Hybrid Capture 2 or cytology (ASCUS threshold) had only a 2% and 5% positive predictive value or absolute risk of developing subsequent CIN3 or CIN2 or worse lesions, respectively, reflecting the low prevalence of precancer among women with HPV-negative ASCUS. In addition, CIN3 lesions detected in ALTS were relatively small,16 presumably due to the intensity of screening in the trial and rigorous disease ascertainment, and may not have invaded for many years if ever. Assuming 1 million HPV-negative ASCUS interpretations per year,6 based on these findings, follow-up using cotesting with cytology and Hybrid Capture 2 at 1 year would result in detection of 8,800 cases of CIN3 or worse at the cost of referring 400,000 women to colposcopy. Using just Hybrid Capture 2 testing at 1 year would drastically reduce the number of colposcopic referrals by 230,000 to 170,000 while detecting 6,460 CIN3 or worse cases. For a single cytology at an ASCUS threshold, there would be 290,000 referrals and 7,540 cases of CIN3 or worse. The increased sensitivity of diagnosis of CIN3 at 1 year, rather than in subsequent screening, that would be achieved by combining HPV testing and cytology needs to be weighed against the greatly increased health care costs, burden on the health care system, and the psychological effect of a positive test despite absence of high-grade disease for most affected women.
There were substantial differences between ASCUS HPV-negative compared with ASCUS HPV-positive women in that the HPV-negative women had a lower risk profile for cervical disease; they were older and had fewer numbers of lifetime and recent sexual partners. An earlier article from ALTS showed low agreement between Quality Control and Clinical Center pathologists in interpreting ASCUS, which the Quality Control group interpreted as negative in 38.6% of the cases.17 ASCUS is not a cytologic abnormality; rather, it is cytologic ambiguity that is clarified by HPV testing, with HPV-positive ASCUS representing true cytologic abnormality with an equivalent risk of cervical precancer as LSIL, and HPV-negative ASCUS representing benign look-alikes.
Last, in an effort to determine misclassification of carcinogenic HPV status, we compared baseline Hybrid Capture 2 results with polymerase chain reaction (PCR) results. We grouped PCR data to include the 13 HPV types in the Hybrid Capture 2 probe and HPV 66, which is shown to be highly cross-reactive with Hybrid Capture 218 and is carcinogenic. We found that 5% (n=174) of the Hybrid Capture 2-negative samples were positive by PCR for at least one of the HPV types in the Hybrid Capture 2 probe. Together these data demonstrate that although small, there is some degree of misclassification of HPV status using any assay, which may partially explain the low but nonzero incidence of CIN3 among women with HPV-negative ASCUS.
In ALTS, colposcopically directed biopsy was used to identify precancer,1,5,7 with the known limitation that colposcopy is subject to variability and imperfect sensitivity for detecting CIN3 or worse. The combination of intensive follow-up with cervical sampling every 6 months and exit colposcopy for all women likely resulted in excellent cumulative disease ascertainment. However, if some cases were missed at exit, leading to an overestimation of sensitivities for all management strategies, it is unlikely to bias our findings, because the same exit colposcopy was used on all women independent of the baseline HPV status.
In conclusion, these data extend previous findings of the very low absolute risk of subsequent detection of disease among carcinogenic HPV-negative women with ASCUS, suggesting they might return to a routine screening interval, which may be longer than 1 year depending on circumstances such as age and prior screening history. However, if increased surveillance is chosen with a 12-month follow-up visit, these data show that HPV testing had higher specificity and lower referral compared with cytology at a threshold of ASCUS. There were too few cases of CIN3 to distinguish differences in sensitivity among the testing strategies. Combining both HPV testing and cytology at 12 months results in extremely high sensitivity of detection of CIN3 or worse, but with probably unacceptably low specificity, high referral rates, and low positive predictive value.
1. Schiffman M, Solomon D. Findings to date from the ASCUS-LSIL Triage Study (ALTS). Arch Pathol Lab Med 2003;127:946–9.
2. Castle PE, Solomon D, Schiffman M, Wheeler CM. Human papillomavirus type 16 infections and 2-year absolute risk of cervical precancer in women with equivocal or mild cytologic abnormalities. J Natl Cancer Inst 2005;97:1066–71.
3. Sherman ME, Lorincz AT, Scott DR, Wacholder S, Castle PE, Glass AG, et al. Baseline cytology, human papillomavirus testing, and risk for cervical neoplasia: a 10-year cohort analysis. J Natl Cancer Inst 2003;95:46–52.
4. Kjaer S, Hogdall E, Frederiksen K, Munk C, van den Brule A, Svare E, et al. The absolute risk of cervical abnormalities in high-risk human papillomavirus-positive, cytologically normal women over a 10-year period. Cancer Res 2006;66:10630–6.
5. Solomon D, Schiffman M, Tarone R. 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;93:293–9.
6. Davey DD, Neal MH, Wilbur DC, Colgan TJ, Styer PE, Mody DR. Bethesda 2001 implementation and reporting rates: 2003 practices of participants in the College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology. Arch Pathol Lab Med 2004;128:1224–9.
7. ASCUS-LSIL Triage Study (ALTS) Group. Results of a randomized trial on the management of cytology interpretations of atypical squamous cells of undetermined significance. Am J Obstet Gynecol 2003;188:1383–92.
8. Wright TC Jr, Cox JT, Massad LS, Twiggs LB, Wilkinson EJ, 2001 ASCCP-sponsored Consensus Conference. 2001 Consensus Guidelines for the Management of Women with Cervical Cytological Abnormalities. J Low Genit Tract Dis 2002;6:127–43.
9. Schiffman M, Adrianza ME. ASCUS-LSIL Triage Study. Design, methods and characteristics of trial participants. Acta Cytol 2000;44:726–42.
10. Leisenring W, Alonzo T, Pepe MS. Comparisons of predictive values of binary medical diagnostic tests for paired designs. Biometrics 2000;56:345–51.
11. Carozzi FM, Confortini M, Cecchini S, Bisanzi S, Cariaggi MP, Pontenani G, et al. Triage with human papillomavirus testing of women with cytologic abnormalities prompting referral for colposcopy assessment. Cancer 2005;105:2–7.
12. Dalla Palma P, Pojer A, Girlando S. HPV triage of women with atypical squamous cells of undetermined significance: a 3-year experience in an Italian organized programme. Cytopathology 2005;16:22–6.
13. Manos MM, Kinney WK, Hurley LB, Sherman ME, Shieh-Ngai J, Kurman RJ, et al. Identifying women with cervical neoplasia: using human papillomavirus DNA testing for equivocal Papanicolaou results. JAMA 1999;281:1605–10.
14. Bergeron C, Jeannel D, Poveda J, Cassonnet P, Orth G. Human papillomavirus testing in women with mild cytologic atypia. Obstet Gynecol 2000;95:821–7.
15. Wright TC Jr, Cox JT, Massad LS, Twiggs LB, Wilkinson EJ. 2001 Consensus Guidelines for the management of women with cervical cytological abnormalities. JAMA 2002;287:2120–9.
16. Sherman ME, Wang SS, Tarone R, Rich L, Schiffman M. Histopathologic extent of cervical intraepithelial neoplasia 3 lesions in the atypical squamous cells of undetermined significance low-grade squamous intraepithelial lesion triage study: implications for subject safety and lead-time bias. Cancer Epidemiol Biomarkers Prev 2003;12:372–9.
17. Stoler MH, Schiffman M, Atypical Squamous Cells of Undetermined Significance-Low-grade Squamous Intraepithelial Lesion Triage Study (ALTS) Group. Interobserver reproducibility of cervical cytologic and histologic interpretations: realistic estimates from the ASCUS-LSIL Triage Study. JAMA 2001;285:1500–5.
18. Schiffman M, Wheeler CM, Dasgupta A, Solomon D, Castle PE. A comparison of a prototype PCR assay and hybrid capture 2 for detection of carcinogenic human papillomavirus DNA in women with equivocal or mildly abnormal papanicolaou smears. Am J Clin Pathol 2005;124:722–32.