Comparison of Human Papillomavirus DNA Prevalence in Atypical Squamous Cells of Undetermined Significance Subcategories as Defined by the Original Bethesda 1991 and the New Bethesda 2001 Systems

In 1991, the National Cancer Institute sponsored a workshop during which The Bethesda System (TBS) for reporting cervical cytology was established. Ten years later, a new workshop was held to review and revise the existing terminology. The original Bethesda System (TBS 1991)1 established a diagnostic category of atypical squamous cells of undetermined significance (ASCUS), which was further divided into the following groups: ASCUS, favor a reactive process (ASCUS-R); ASCUS, not otherwise specified (ASCUS-NOS); and ASCUS, favor a low- or a high-grade squamous intraepithelial lesion (ASCUS-L or ASCUS-H, respectively). The revised TBS 20012 divides ASCUS into 2 categories: atypical squamous cells of undetermined significance (ASC-US) and atypical squamous cells, cannot exclude high-grade squamous intraepithelial lesion (ASC-H). The Bethesda System 2001 eliminated the ASCUS-R category with a recommendation that ASCUS-R cases be judiciously downgraded to “negative for intraepithelial lesion or malignancy” (negative). The rationale for the elimination of this subcategory was that the clinical follow-up in patients with a diagnosis of ASCUS-R was benign.2 Qualification of ASCUS, however, has poor interobserver and intraobserver reproducibility, because it is based on subjective criteria and encompasses a spectrum of different morphologic cell types. The reported κ value for the interobserver agreement for diagnosis of ASCUS-R is only 0.23.3 Different laboratories have different thresholds for diagnosis of ASCUS-R, and currently there may be different thresholds for downgrading ASCUS-R to the negative category. There is a justified concern that the sensitivity of the Papanicolaou (Pap) test may be compromised with the new changes in ASCUS classification.

The primary goal of this study was to determine the impact of the new ASCUS subclassification on the accuracy of Pap test diagnosis by examining the prevalence of HPV DNA in different ASCUS subcategories, as defined by TBS 1991 and TBS 2001. Human papillomavirus testing has been suggested as a useful tool for quality control and for standardizing diagnoses between different cytology laboratories. For that reason, in this study we used HPV testing to evaluate the accuracy of cytologic diagnoses.

The Bethesda System 1991 describes several different morphologic subtypes of ASCUS based on cellular morphology. These subtypes include atypical mature squamous cells (ASCUS-M), atypical immature squamous cells (ASCUS-IM), atypical parakeratotic cells (APK), atypical repair (AREP), and atypical atrophic squamous cells (AATR). This system of subclassification of ASCUS has better interobserver and intraobserver reproducibility; however, the clinical significance of this subclassification has not been established. The second goal of the study was to examine whether any of these morphologic subcategories are more frequently associated with HPV positivity. In addition, we were interested in whether specific individual morphologic features of atypical squamous cells are predictive of HPV DNA detection.

Consecutive cases of ThinPrep cervical Pap tests with original diagnoses of ASCUS (n = 109; mean age, 36 years; range, 15–80 years) were collected between 1998 and 1999. In addition, consecutive cases of low-grade squamous intraepithelial lesion (LSIL) and high-grade squamous intraepithelial lesion (HSIL) (n = 52; mean age, 34 years; age range, 15–75 years) were collected for a positive control group. Age-matched cases with original diagnoses of “negative for intraepithelial lesion or malignancy” (negative) and no known history of cervical dysplasia were selected for the negative control group (n = 48; mean age, 32 years; age range, 20–50 years).

The ThinPrep slides from each case were evaluated independently by 2 cytologists and 1 cytotechnologist blinded to both the original diagnosis and other pathologist review. The diagnostic groups included (a) negative, (b) ASCUS, and (c) LSIL/HSIL. A consensus diagnosis was defined as an agreement between at least 2 reviewers. In rare cases, the diagnosis was split 3 ways between negative, ASCUS, and squamous intraepithelial lesion (SIL); these cases were classified ASCUS. All ASCUS cases were further evaluated in detail. The cases were first subclassified according to TBS 1991 into the following groups: ASCUS-R, ASCUS-;NOS, ASCUS-L, or ASCUS-H. In a second review, the same cases were classified according to TBS 2001, which includes only 2 ASCUS categories, ASC-US and ASC-H. Furthermore, all ASCUS cases were subclassified according to morphology into atypical mature cells, atypical immature cells, atypical parakeratotic cells, atypical repair, and atypical atrophic cells. Individual cytologic features of the atypical cells were recorded and scored as absent/minimal/mild (=0) or moderate/marked (=1). The recorded features included nuclear enlargement, increased nuclear- cytoplasmic ratio, hyperchromasia, irregular or coarse chromatin, irregular nuclear membrane, variation of the nuclear size and shape in the group of cells, and multinucleation or perinuclear halo.

One milliliter of ThinPrep material was spun down for 2 minutes at 13 000 rpm. The supernatant was discarded and the pellet was dried overnight. The pellet was incubated with 250 μL proteinase K (1 mg/mL) in 50mM Tris–hydrogen chloride, pH 8.0, and 0.5% Tween 20 for 18 hours at 56°C. Following heat inactivation at 95°C for 10 minutes, 10 μL of the supernatant was used for polymerase chain reaction (PCR). The entire preparation was carried out in a specially dedicated laboratory to avoid PCR product carryover. Adequate DNA quality was established by PCR amplification of β-globin gene, resulting in a 96-base pair (bp) product.4 

Broad-spectrum HPV DNA amplification was performed using the short PCR fragment (SPF 10) primer set, as described previously.5 SPF 10 PCR amplifies a 65-bp fragment from the L1 region of the HPV genome. Briefly, HPV DNA amplification was performed in a total volume of 50 μL containing 10 μL of isolated DNA, 10mM Tris–hydrogen chloride (pH 9.0), 50mM potassium chloride, 2.0mM magnesium chloride, 0.1% Triton X-100, 0.01% gelatin, 200μM of each deoxynucleoside triphosphate, 15 pmol of each of the forward and reverse primers, and 1.5 U of AmpliTaq gold (Perkin-Elmer, Norwalk, Conn). AmpliTaq Gold was activated by incubation at 94°C for 9 minutes. Human papillomavirus DNA was amplified in 40 cycles of 30 seconds at 94°C, 45 seconds at 52°C, 45 seconds at 72°C, and a final extension of 5 minutes at 72°C. Each experiment was performed with separate positive (HPV DNA) and negative (water) controls. Confirmation of the presence of amplified HPV-specific sequences was performed with HPV DNA enzyme immunoassay, a microtiter plate–based hybridization assay. The exact HPV DNA enzyme immunoassay conditions were described previously.5 

Samples identified as positive for HPV DNA were genotyped with the Line Probe Assay (LiPA) (Innogenetics Inc, Alpharetta, Ga). Twenty-five individual HPV genotypes (high-risk HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68, and 70; and low-risk HPV types 6, 11, 34, 40, 42–44, 53, 54, and 74) can be identified simultaneously in a single assay. The exact assay conditions have been described previously.6 Briefly, 10 μL of denatured HPV PCR product was hybridized for 60 minutes at 50°C to genotype-specific probes immobilized as parallel lines on a nitrocellulose strip. Following the washing step, the products of hybridization were visualized in a color reaction using alkaline phosphatase–streptavidin conjugate, 5-bromo-4-chloro-3-indolylphosphate, and nitroblue tetrazolium.

After the consensus review of 109 ASCUS cases, 79 (72%) of the cases were confirmed as ASCUS, 15 (14%) cases were reclassified as SIL, and 15 (14%) cases were reclassified as negative. Fifty-two SIL cases were reviewed, and the diagnosis was confirmed in 42 (81%) cases; the remaining 10 (18%) cases were downgraded to ASCUS. In the 48 cases in the negative group, the diagnosis was confirmed in 44 (91%) cases and upgraded to ASCUS in the remaining 4 (8%) cases.

The reason for classifying a case as ASCUS was qualitative in 86% of cases (cytologic features fell short of SIL diagnosis), quantitative in 13% of cases (<5 dysplastic cells on a slide), and due to downgrading because of severe inflammation in 1% of cases.

Prevalence of HPV DNA in the main diagnostic groups classified according to TBS 1991 is shown in Table 1. The overall HPV prevalence in all ASCUS cases (49.4%) was significantly higher than in negative cases (32.2%) with P = .03 by χ² test. HPV DNA prevalence was also significantly higher in LSIL/HSIL cases (95.8/90.0%) than in all ASCUS cases (49.4%) with P = .005 by χ² test. The prevalence of high-risk HPV types, as well as the prevalence of multiple HPV–type infections, increased as the level of cytologic abnormality increased and was the highest in HSIL cases. High-risk HPVs were identified in 11.8% of negative cases, 30.1% of all ASCUS cases, 66.6% of LSIL cases, and 90.0% of HSIL cases.

The ASCUS-R category represented the largest category under TBS 1991, accounting for 45% of ASCUS cases, followed by ASCUS-L (29%), ASCUS-NOS (18%), and ASCUS-H (7.5%) (Table 1). Prevalence of HPV DNA in ASCUS-R cases (33.3%) was not significantly different from that in the negative group (32.2%) with P = .90. The percentage of high-risk HPV types was almost twice as high in ASCUS-R versus negative cases (21.4% vs 11.8%); however, the difference was not statistically significant (P = .19). Human papillomavirus positivity was significantly higher in the remaining ASCUS subcategories and ranged from 47% in the ASCUS-NOS group to 85.7% in the ASCUS-H group. The percentage of high-risk HPV types and multiple HPV–type infections was the highest in ASCUS- H cases, as compared to all other ASCUS subcategories.

Following the second consensus review of ASCUS cases using the guidelines of TBS 2001, 45% (42/93) of ASCUS cases were downgraded to the negative category (Table 2). The overall prevalence of HPV and prevalence of high-risk HPV types in these downgraded cases was not significantly different from that of negative cases diagnosed under TBS 1991 (P = .53). The overall HPV prevalence and prevalence of high-risk HPV types in ASCUS diagnosed according to TBS 2001 was, however, significantly higher than in the downgraded cases (P = .04).

Table 3 shows the correlation between ASCUS diagnoses rendered according to TBS 1991 versus TBS 2001. Fifty-seven percent (24/42) of ASCUS-R, 47% (8/17) of ASCUS-NOS, and 25% (7/27) of ASCUS-L cases were downgraded to negative under TBS 2001. In addition, 3 ASCUS-;H cases were downgraded to negative; 1 of these 3 cases was positive for high-risk HPV types. Interestingly, 3 ASCUS-R cases were upgraded to ASC-H; 2 were positive for high-risk HPV types. Only 1 of 7 cases diagnosed as ASCUS-H in the first review (TBS 1991) was diagnosed as ASC-H on the second review (TBS 2001).

Four different morphologic categories of ASCUS were identified in the collected cases: atypical mature squamous cells (53%), atypical immature squamous cells (18%), atypical parakeratotic cells (25%), and atypical atrophic squamous cells (3%). Cases of atypical repair were not identified in this study. There was no statistically significant difference in HPV DNA prevalence between the different morphologic ASCUS subtypes (Table 4) (P between .49 and .86). The percentage of high-risk HPV types in the ASCUS-IM group (47%) was higher than in the remaining categories; however, the difference was not statistically significant (P = .14). No high-risk HPV types were detected among 3 cases with atypical atrophic squamous cells.

We were interested in whether there are specific individual morphologic features of atypical squamous cells that are more frequently associated with HPV DNA detection. Eight different morphologic features of atypical squamous cells were scored by each reviewer as either present (1) or absent (0). There was poor agreement between the scores of different reviewers evaluating the same case (average κ < 0.3).

Table 5 shows the comparison between the frequencies of the individual morphologic features (after combining the scores of all reviewers) in HPV-positive and HPV-negative ASCUS cases. We found no statistically significant difference in the frequency of any specific feature between the HPV-positive and HPV-negative groups. Nuclear enlargement, irregular nuclear membrane, and coarse chromatin were the most common findings and were identified with similar frequency in HPV-positive and HPV-negative cases. Presence of coarse chromatin was slightly more common in the HPV-positive group, but the difference was not statistically significant.

The goal of our study was to evaluate the impact of the new guidelines for classification of ASCUS (TBS 2001) on the accuracy of Pap test diagnosis. We used HPV testing for verification of diagnostic accuracy. Validation of the Pap diagnosis is most commonly performed by either correlation with the histologic diagnosis or by correlation with the results of HPV DNA testing. Positive correlation between the histologic findings and the Pap test result is the best confirmation of diagnostic accuracy; however, in many instances, patients with mild cytologic abnormalities are not evaluated colposcopically, and thus histologic follow-up is not available. Furthermore, biopsies performed during colposcopic examinations are subject to sampling errors with a notable false-negative rate (10%– 32%).7 In addition, interobserver reproducibility of the biopsy diagnosis is moderate to poor, particularly in cases of LSIL.8 

Human papillomavirus testing, on the other hand, may be easily performed on all cervical Pap tests obtained for liquid-based cytology. Sampling error is thought to be negligible since the entire circumference of the cervix is being sampled, and it is an objective test. Human papillomavirus testing, however, has its own shortcomings. The most significant is the high rate of HPV positivity (up to 39%) in negative Pap tests in the population of women younger than 40 years.9 

In this study, consecutive ASCUS cases were retrospectively reviewed and subclassified according to the criteria described in TBS 1991 and then re-reviewed and subclassified according to TBS 2001 guidelines. On the second review, after elimination of the ASCUS-R category and downgrading of a proportion of cases to the negative category, the number of cases diagnosed as ASCUS decreased by 45%. This finding suggests that the new guidelines may result in a significant reduction in the number of ASCUS diagnoses and therefore may improve the specificity of the abnormal Pap test diagnosis. The second important finding was that HPV positivity in downgraded cases was not significantly different from HPV positivity in negative cases diagnosed according to TBS 1991. This result suggests that elimination of the ASCUS-R category and downgrading of selected cases are not likely to decrease the sensitivity of the Pap test.

To our knowledge, this is the first study to compare HPV prevalence in the diagnostic groups as defined by the new TBS 2001. Previously reported results of HPV detection in Pap tests are based on the TBS 1991 classification. The average prevalence of HPV DNA in ASCUS is approximately 50% for any HPV type3,10–13 and ranges from 30% to 49% for high-risk HPV types.12,14 The results of our study are consistent with these previous reports; HPV DNA was identified in 49.4% of ASCUS cases defined according to TBS 1991, with high-risk HPV types identified in 30.1% of these cases. The reported HPV prevalence in negative Pap tests ranges from 11.5%3 to 40.0%9 for any HPV type and from 13%12 to 31%14 for high-risk HPV types. This wide range of HPV prevalence in negative Pap tests may be due to a selection of high-risk versus low-risk patient populations and different age stratifications in the tested groups. The patient population at New York Presbyterian Hospital, Weill Medical College of Cornell University is a high-risk population, which is reflected by a high prevalence of HPV DNA in the negative group (32.2% for any HPV and 11.8% for high-risk HPV types). The reported HPV positivity in LSIL and HSIL is usually greater than 90%,10–15 and our results are consistent with previous studies.

Few studies have compared HPV prevalence in different subcategories of ASCUS. In the most recent study, Hughes et al12 detected high-risk HPV types in 18.5% of ASCUS-;R, 27.7% of ASCUS-NOS, 40.9% of ASCUS-L, 56.0% of ASCUS-H, and 15.2% of negative cases. In another study, Zuna et al15 reported detection of high-risk HPV types in 31.4% of ASCUS-R cases and 52.0% of ASCUS, favor dysplasia cases, and in 17% of negative cases. Furthermore, Crum et al3 reported the prevalence of high-risk HPV types as 8.8% in ASCUS-R cases; 17.4% in ASCUS-NOS cases; 47.8% in ASCUS, favor dysplasia cases; and 5.6% in negative cases. Finally, Sherman et al16,17 detected high- risk HPV types in ASCUS-L (63.2%) and ASCUS-H cases (85.6%). Our results most closely parallel the results of Hughes et al.12 Similar to Hughes et al12 and Crum et al,3 we did not find significant differences in total HPV prevalence or high-risk HPV prevalence between negative and ASCUS-R cases.

An interesting finding of our study is the correlation between multiple HPV infection and cytologic changes. We observed that the rate of multiple HPV–type infection increased as the degree of cytologic abnormality of a Pap test increased, with the highest level in HSIL cases (33%). Multiple HPV–type infection may be due to either synchronous acquisition of multiple viral types or, alternatively, ineffective elimination of consecutive infections, and as such may be a marker of an inadequate immune response to HPV. Other investigators observed a higher rate of multiple HPVs in women with LSIL (23%) as compared to HSIL (7%).18 In addition, it has been reported that the rate of multiple HPV infection has a positive correlation with the high number of recent sexual partners and a negative correlation with the increasing age of the subjects.18,19 These results suggest that multiple-type infection may be related to the higher exposure to multiple HPVs and to the onset of sexual activity, when the immune response is not yet established.19 Further research is required to clarify the clinical significance of infection with multiple HPV types.

The ASCUS diagnosis is highly subjective and has poor interobserver and intraobserver reproducibility. In an attempt to narrow the diagnostic criteria, we tried to identify the morphologic features of ASCUS cells that were more frequently associated with HPV DNA detection; however, no specific morphologic features of atypical cells were found to be more commonly associated with HPV positivity when these features were analyzed individually. In addition, we did not find significant differences in overall HPV prevalence between the morphologic subtypes of ASCUS; nonetheless, ASCUS-IM showed slightly higher prevalence of high-risk HPV types. No other reports of HPV detection in morphologic subtypes of ASCUS are available. Several studies, however, reported histologic follow-up for ASCUS cases subclassified according to morphology. Sheils et al20 reported detection of SIL in follow-up biopsies of 10% of patients with mature ASCUS, 24% of metaplastic ASCUS, and 42% of immature metaplastic ASCUS. Voytek et al21 reported that 50% of patients with atypical parakeratotic cells had SIL on a follow-up Pap test or biopsy. In contrast, ASCUS associated with atrophy rarely correlates with the presence of SIL. Abati et al22 and Symmans et al23 both reported SIL in a mere 12% of patients with atypical atrophic smears. Ettler et al24 analyzed individual morphologic features of ASCUS cells and correlated their presence with detection of dysplasia in the biopsies. Of all the examined features (increased nuclear size, irregular chromatin pattern, nuclear membrane irregularity, presence of nucleoli, multinucleation, perinuclear halo, and metaplastic morphology), only nuclear membrane irregularity was independently predictive of dysplasia.

In conclusion, the results of our study indicate that elimination of the ASCUS-R category in TBS 2001 may result in a significant decrease in the number of ASCUS diagnoses. Furthermore, cases downgraded to the negative category had a relatively low prevalence of HPV DNA, and it is therefore expected that the new TBS 2001 will increase the specificity of the Pap test without compromising its sensitivity.