|Home | About | Journals | Submit | Contact Us | Français|
To estimate the fraction of cervical intraepithelial neoplasia-2 (CIN-2) that might regress if untreated using data from the atypical squamous cells of undetermined significance (ASCUS) and low-grade squamous intraepithelial lesions (LSIL) triage study (ALTS).
We compared the cumulative occurrence of CIN-2 (n = 397) and CIN-3 or worse (n = 542) diagnosed by the Pathology Quality Control Group in three trial arms— immediate colposcopy, human papillomavirus (HPV) triage, and conservative management— over the 2-year duration of the ALTS trial. A non-parametric test of trend was used to test for differences in the number of CIN-2 cases relative to number of CIN-3 or worse cases across study arms, with an increasing percentage of women referred to colposcopy at baseline.
There were no significant differences in the cumulative 2-year cumulative CIN-3 or worse diagnoses by study arm (10.9%, conservative management; 10.3%, HPV; 10.9%, immediate colposcopy) (ptrend = 0.8) but there was a significant increase in CIN-2 diagnoses (5.8%, conservative management; 7.8%, HPV triage; 9.9%, immediate colposcopy) (ptrend < 0.001) in the study arms with increasing number of women referred to colposcopy at baseline. The relative differences in CIN-2 by study arm among women who tested HPV16 positive at baseline were less pronounced (ptrend = 0.1) than women who tested positive for other high-risk HPV genotypes (ptrend = 0.01).
There was evidence that approximately 40% of undiagnosed CIN-2 will regress over 2 years but CIN-2 caused by HPV16 may be less likely to regress than CIN-2 caused by other high-risk HPV genotypes.
In cervical cancer screening programs, women are sent to colposcopy for a diagnostic evaluation and biopsy of evident lesions if they screen positive by cytology or, increasingly, by various combinations of cytology and human papillomavirus (HPV) test results (1). Women diagnosed with a high-grade cervical lesion, defined in the U.S. as cervical intraepithelial neoplasia grade 2 (CIN2) or worse, are then treated, primarily by excision. However, there is increasing awareness that not all CIN3 (2) and especially not all CIN2 is “precancer” (3). In fact, a large proportion of CIN2 lesions may resolve without treatment, leading to recommendations not to treat CIN2 immediately in young women (4). The natural history of CIN2 is not clear, partly because most recent studies that have examined regression of histologically-confirmed high-grade lesions did not differentiate between CIN2 and CIN3 diagnoses, had sample sizes of ≤100 cases (5), and did not employ rigorous pathology review to rule out misclassification of CIN1 and CIN2.
To examine the specific issue of CIN2 regression, data from the atypical squamous cells of undetermined significance (ASCUS) and low-grade squamous intraepithelial lesion (LSIL) triage study (ALTS) were used. Different algorithms for referring women to colposcopy at baseline in ALTS were used analytically to compare how the aggressiveness of referral to colposcopy influenced the cumulative 2-year incidence of CIN2.
ALTS was a multi-site, randomized trial comparing three management strategies for women referred for ASCUS (n = 3,488) or LSIL (n = 1,572) conventional cytology (6–10). The arms were: 1) immediate colposcopy (IC arm) (referral to colposcopy regardless of enrollment test results); 2) HPV triage (HPV arm) (referral to colposcopy if the enrollment HPV result was positive by Hybrid Capture 2 (hc2; Qiagen Corporation, Gaithersburg, MD) or missing, or if the enrollment cytology was high-grade squamous intraepithelial lesion [HSIL]); or 3) conservative management (CM arm) (referral to colposcopy only if the enrollment cytology was HSIL). Among women with LSIL randomized to the CM arm (n = 675), seven women (1.0%) during enrollment and 136 women (20.1%) during follow-up were sent to colposcopy on the basis of a protocol modification initiated as a safety intervention (8); if this happened to be the 18-month follow-up visit and colposcopy was performed, the 18-month visit became the participant’s “exit” visit (see below).
At enrollment, all women received a pelvic examination with collection of two cervical specimens; the first specimen in PreservCyt for ThinPrep cytology (Cytyc Corporation, Marlborough, MA, USA) and hc2 pooled testing (Qiagen Corporation) and the second in specimen transport medium (STM; Qiagen Corporation) for HPV genotyping. Women in the three arms of the study were re-evaluated by cytology every 6 months for 2 years of follow-up and sent to colposcopy if cytology was HSIL. A CIN2 or worse (≥CIN2) diagnosis by the clinical center pathologists was the treatment threshold. An exit examination with colposcopy was scheduled for all women, with loop electrosurgical excision offered for ≥CIN2 or persistent low-grade lesions. The National Cancer Institute and local institutional review boards approved the study and all participants provided written informed consent.
Clinical management was based on the clinical center pathologists’ cytologic interpretations and histologic diagnoses as previously described (6–10). Referral smears, ThinPreps, and histology slides were also sent to the Pathology Quality Control Group (QC Pathology) based at the Johns Hopkins Hospital for review, including computer-assisted review, and secondary diagnoses as previously described (6–10).
Two HPV DNA tests were performed on clinical specimens collected at enrollment. Hybrid Capture 2 using probe set B, a pooled probe DNA test for one or more carcinogenic or high-risk (HR)-HPV genotypes (HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68), was performed on PreservCyt specimens (10). A positive test does not identify which HR HPV genotype(s) are present. Hybrid Capture 2 is also well-known to cross-react with untargeted HPV genotypes (11;12), including HPV66, recently classified as a HR-HPV genotype(13). A L1 consensus primer-based PCR assay (14;15) was performed on the STM specimen to test for 27 or 38 HPV genotypes.
HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66and 68 were considered primary HR HPV genotypes. An a priori HPV risk grouping was established according to cervical cancer risk as follows: 1) positive for HPV16; 2) HPV16 negative, but positive for other HR HPV genotypes by PCR or by hc2 if PCR was negative or missing; 3) negative for all HR genotypes but positive for non-HR HPV genotypes (n.b., specimens that tested positive for non-HR HPV by PCR and positive by hc2 were conservatively called non-HR HPV positive because hc2 sometimes cross-reacts with non-HR HPV genotypes (11;12); or 4) PCR negative.
We used the worst histological diagnosis by the Pathology Quality Control Group (QC Pathology) over the 2-year duration of ALTS for these analyses. In order to measure the regression of CIN2 (n = 397), we relied on the fact that different study arms of ALTS were less or more aggressive in colposcopic referral and detection of CIN2 or worse (≥CIN2): The CM arm sent 188 of 1839 women (10.2%), the HPV arm sent 768 of 1385 women (55.5%), and the IC arm sent 1317 of 1836 women (99.0%) women to colposcopy at enrollment (8;9). Forty women in the HPV arm, 1 with CIN2 and 6 with CIN3, were referred to colposcopy because of missing hc2 results. We used CIN3 or worse (≥CIN3) (n = 542; 535 CIN3 and 7 cancers) diagnoses as a reference group under the assumptions that 1) CIN3 was much less likely to regress than CIN2 and 2) the majority of ≥CIN3 diagnosed, regardless of the timing of the diagnosis, was already present at the time of enrollment.
An extension of the Wilcoxon rank-sum test was used as a non-parametric test for trend across study arms (16). Binomial exact confidence intervals were calculated for proportions. A P value < 0.05 was considered statistically significant. Stata version 8.2 (Stata Corporation, College Station, TX) was used for all statistical analyses.
We included in our analysis the 5,052 of 5,060 women (99.8%) for whom we had HPV data (hc2 and/or PCR). Table 1 shows the demographics and risk factors for women by their worst 2-year histologic diagnosis (<CIN2, CIN2, or ≥CIN3) as rendered by QC Pathology. As previously noted, smokers (vs. non-smokers)(17) and the less educated (18) were more likely to be diagnosed with ≥CIN3.
Table 2 shows that there was little difference (ptrend = 0.8) in the proportion of women diagnosed with ≥CIN3 over the two year duration of ALTS in the populations randomized to each study arm (10.9%, 10.3%, and 10.9% in the CM, HPV, and IC arms, respectively). However, the proportion of women diagnosed with CIN2 differed by study arm: 5.8% (95%CI = 4.7%–6.9%) in the CM arm, 7.8% (95%CI = 6.5%–9.4%) in the HPV arm, and 9.9% (95%CI = 8.6%–11.4%) in the IC arm. That is, there were more diagnoses of CIN2 with strategies referring larger proportions of women to baseline colposcopy in the trial (CM<HPV<IC), with 41.6% (95%CI = 24.3%–59.0%) fewer cases of CIN2 diagnosed in the CM arm compared to the IC arm. Given that follow-up and exit screening and management did not differ by study arm, the differences in proportion of women diagnosed with CIN2 by the study arm were due to the differences in the number of women sent to colposcopy at baseline in each study arm. There was a significant trend (ptrend < 0.001) of fewer two-year cumulative CIN2 histologic diagnoses by study arm relative to two-year cumulative ≥CIN3. Exclusion of women in the CM, referred for a LSIL Pap, and sent for an early colposcopic evaluation for safety reasons did not alter our findings (data not shown)(8).
Similar trends for CIN2 were observed in the ASCUS referral population (4.7% in the CM arm and 7.9% in the IC arm), the ASCUS referral population restricted to those who tested hc2 positive (8.2% in the CM arm and 12.8% in the IC arm), and the LSIL referral population (7.6% in the CM arm and 13.8% in the IC arm). The differences in proportion of CIN2 diagnosed in the CM versus the IC arms were less pronounce in women 30 and older (4.2% in the CM arm and 5.5% in the IC arm) than in women under the age of 30 (6.4% in the CM arm and 11.7% in the IC arm) but this age-specific effect did not reach statistical significance (p = 0.3).
Stratifying on the HPV risk group, this trend (of relative increase in number of CIN2 diagnoses by fraction of women sent to colposcopy within a study arm) was also observed among women with enrollment HR HPV infections (excluding HPV16) infections (ptrend = 0.01); the trend (ptrend = 0.1) was non-significant among women with HPV16, although women in the IC arm had relatively more CIN2 than in the other study arms. A similar pattern was observed for the clinical center histologic diagnoses although the clinical center pathologists tended to call more CIN2 and less ≥CIN3 in general than QC Pathology (Table 4).
The timing of the diagnosis of CIN2 and ≥CIN3 by study arm is shown in Figure 1. Of note, there were significantly fewer cases of both CIN2 and ≥CIN3 diagnosed at enrollment in the CM arm of the trial. There was a significantly fewer two-year cumulative CIN2 histologic diagnoses by study arm (Conservative Management Arm [CM] < HPV Triage Arm [HPV] < Immediate Colposcopy Arm [IC]) relative to two-year cumulative ≥CIN3 at enrollment but not during follow-up (ptrend = 0.5) and exit (ptrend = 0.3). The CIN2:≥CIN3 in the CM arm at enrollment was 0.40, a 3-fold lower than the ratio in the IC arm (ptrend < 0.001 by study arm), further evidence that the aggressiveness of the study arm in the colposcopy referral at enrollment influenced the overall number of CIN2 diagnosed over the 2 years within a study arm. There were no observed differences in the ratio of CIN2:≥CIN3 by study arm during follow-up (ptrend = 0.5) and exit (ptrend = 0.3), i.e., similar increases in cumulative incidence CIN2 and CIN3 and the relatively stable CIN2:≥CIN3 ratio in each arm, reflecting that women in all arms were managed similarly after enrollment
CIN2 diagnoses by QC Pathology that were called less than CIN2 or CIN2 by the clinical center pathologists were the most likely to be under-represented in the CM arm compared to the IC arm (Table 3). For example, the ratio of the number of QC pathology diagnosed CIN2 in the CM arm versus the IC arm among those diagnosed as less than CIN2 by clinical center pathologists was 0.40. By comparison, the ratio was 0.74 among diagnosed as CIN3 or more severe by the clinical center pathologists. Overall, there was a marginally significant trend of increasing differences in the number of CIN2 between study arms by less severe diagnosis by the clinical center pathologists (ptrend = 0.07). A similar pattern was observed when restricted to only HR HPV (excluding HPV16)-positive CIN2 (ptrend = 0.07).
We previously have shown that a portion of CIN2 diagnoses are also due to misclassification of disease (3;19) and to HPV infections by non-HR HPV genotypes (3). Here, we present data from ALTS showing that a significant portion of CIN2 diagnosed by central panel review, even caused by HR HPV genotypes, may represent only regressive lesions. Similar patterns were observed in the ASCUS referral population, in the ASCUS referrals that tested hc2 positive, and in the LSIL referral population. We estimated that approximately 40% of rigorously-defined CIN2 diagnoses in primarily young women (mean age = 24.8 years, median age = 23 years, range = 18–62 years) were destined to resolve on their own by using a “subtractive” analytic approach that compared the fraction found in the least sensitive (CM) and most sensitive (IC) study arm (by virtue of the proportion of women in arm referred to colposcopy) for detection of cervical abnormalities. The differences in CIN2 detection are due to the delay in the colposcopic evaluation of patients. Importantly, the time difference in diagnostic evaluation was assigned by the randomization, which created comparable populations. By comparison, 27% (20) and 19% (21) of histologically-confirmed CIN2 regressed over approximately 2 years of follow-up in the placebo arms of two chemoprevention trials.
We used ≥CIN3 as a reference to show that the study arms were roughly equivalent in cancer risk, and the differences in the criteria for enrollment colposcopy primarily affected the fraction of women diagnosed with CIN2. However, we note that, in addition to some CIN2 regressing, some HPV infections may have caused incident CIN2 while some CIN2 may have “progressed” to CIN3. Thus CIN2 regression cannot be precisely determined although reasonable estimations can be derived.
It is also worth mentioning that CIN2 is the least reproducible of all cervical diagnoses (19;22), and it is possible that the regressive nature of CIN2 will depend somewhat on the individual pathologist diagnosing it. As shown in table 2, this lack of agreement is directly related to the regression of CIN2: the likelihood of regression correlated with the severity of the second (original) diagnosis by the clinical center. For scientific rigor, we used the histopathologic diagnoses rendered by the Pathology Quality Control Group review in our analysis. Yet, we observed a similar difference in the fraction of CIN2 as diagnosed by the clinical center pathologists between the CM and IC arms.
In real-world settings, some clinical pathologists equivocate on the use of a CIN2 diagnosis, relying on non-standard diagnoses such as "CIN1-2” or “CIN2-3”. Their use may reflect further the uncertain nature of CIN2. It is plausible that each diagnosis will represent different likelihoods of regression although it is unclear how reliable such distinctions are, given how generally unreliable a CIN2 diagnosis is (19;22). The inclusion of lesions of an even more ambiguous nature in the definition of CIN2 would be expected to increase the overall regressive potential of CIN2. The use of equivocations like CIN1/2 should probably be discouraged given the lack of evidence of utility, unproven reliability, and the possible increase in the number of women receiving unnecessary therapy.
We point out that the CIN2 diagnosed at enrollment followed an ASCUS or LSIL Pap smear, which led to referral into the study. That is, women were not referred into ALTS because of a HSIL Pap smear. In ALTS, we previously found that women with biopsy-diagnosed CIN2 following the HSIL enrollment cytology (n.b., women enrolled into ALTS had cytology repeated at the enrollment visit) were more apt to have an underlying CIN3 than those with a less severe enrollment cytology (3). We therefore hypothesize that CIN2 following a HSIL cytology may be on average less regressive than that observed for CIN2 diagnosed in the ALTS population.
A CIN2 diagnosis on biopsy is the clinical threshold for treatment in the U.S., its treatment provides a margin of safety against cancer risk. However, a greater recognition of the true equivocal nature of CIN2 is needed. Aggressive clinical management of young women (median age = 25 years; mean age = 27.5 years in ALTS) who have ASCUS or LSIL cytology will lead to over-diagnosis of CIN2 and over-treatment of a significant number of regressive lesions. This is an important clinical issue since most CIN2 is diagnosed in women following ASCUS or LSIL cytology (23).
From these data, we suggest that strategies that send more women to colposcopy, such as immediate colposcopy for LSIL or ASC-US or HPV triage of ASC-US, offer the benefit of early detection of CIN3 at the cost of increased detection of CIN2, some proportion of which is destined to resolve and its treatment translates into patient morbidity without benefit. Repeat cytology at an ASCUS threshold as recommended, rather than a HSIL threshold as used in CM arm of ALTS, would be expected to yield different results, with more women going to colposcopy and higher proportion of CIN2 i.e., more akin to the IC and HPV arms of ALTS. However, this is only speculative since we did not evaluate this clinical algorithm in ALTS.
HPV16-positive CIN2 appears less likely to regress, perhaps as the result of its greater tendency to persist and its greater oncogenic potential to progress to precancerous lesions than other HPV genotypes (24). When HPV genotyping from validated tests becomes routinely available, detection of HPV16 may be a useful stratifier of risk (25) for deciding the clinical management of CIN2 diagnoses: HPV16-negative CIN2 diagnosed in young women with ASCUS or LSIL cytology could be managed less aggressively through increased surveillance rather than immediate treatment (1;4) while HPV16-positive CIN2, the most strongly linked with CIN3 (3) and least regressive, probably warrants inclusion with CIN3 for management.
More generally, a clinical trial to determine the best management strategies for CIN2, akin to ALTS for ASCUS and LSIL Pap smears, is needed. Several biomarkers, such as HPV genotyping and p16INK4A immunostaining (26–28), warrant evaluation for determining which women with CIN2 need immediate treatment versus surveillance. Key outcomes for any triage strategy for CIN2 would be the timely detection of CIN3 (misclassified as CIN2 (3) or progression of a carcinogenic HPV infection, diagnosed as CIN2, to CIN3), safety against cancer, and the reduction in the number of women with CIN2 treated as compared to the current standard-of-care (4) of excising treatment for all women diagnosed with CIN2.
The authors thank the ALTS Group Investigators for their help in planning and conducting the trial and Information Management Services, Inc. (Rockville, MD) for data management and programming support.
Financial Support: ALTS was supported by the National Cancer Institute, National Institutes of Health Department of Health and Human Services contracts CN-55153, CN-55154, CN-55155, CN-55156, CN-55157, CN-55158, CN-55159 and CN-55105. This research was supported [in part] by the Intramural Research Program of the NIH, National Cancer Institute. Some of the equipment and supplies used in these studies were donated or provided at reduced cost by Qiagen Corporation, Gaithersburg, MD; Cytyc Corporation, Marlborough, MA; National Testing Laboratories, Fenton, MO; DenVu, Tucson, AZ; TriPath Imaging, Inc., Burlington, NC; and Roche Molecular Systems Inc., Alameda, CA.
Financial Disclosure: Dr. Wheeler has received funding through the University of New Mexico to conduct HPV vaccine studies for Merck (Whitehouse Station, NJ) and GlaxoSmithKline (Philadelphia, PA). She has also received reagents for HPV genotyping from Roche Molecular Systems. The other authors have no potential conflicts of interest to disclose.