Detection of Adenocarcinoma In Situ of the Cervix in Papanicolaou Tests: Comparison of Diagnostic Accuracy With Other High-Grade Lesions Open Access

Use of the interpretive category of adenocarcinoma in situ of the cervix is becoming increasingly common in cytologic material. There are several reasons for this. First, the incidence of the disease appears to be increasing.1–4 Second, better endocervical sampling devices are used commonly.5 Third, the morphologic criteria for the entity in cytologic material have been defined and increasingly refined.6–15 These studies have culminated in the recent recommendation by the International Academy of Cytology Task Force that cases of adenocarcinoma in situ be specifically diagnosed when possible16 and, most recently, with the inclusion of adenocarcinoma in situ as a separate interpretive category in the Bethesda 2001 classification system.17 However, several studies show no clear evidence that Papanicolaou test screening is an effective method of detecting either adenocarcinoma or adenocarcinoma in situ or that screening has reduced the mortality from adenocarcinoma.18–20 

Although specific morphologic criteria have been published, there is significant variability in the interpretation of glandular lesions.12,21 The likelihood of the recognition and precise categorization of this entity in the United States is not well known. To address this, we reviewed the results of the College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology (PAP Program) with conventional preparations for adenocarcinoma in situ during the years 2001–2002 and compared them with the results for other significant lesions.

The PAP Program is a quarterly mailed, glass-slide quality improvement program. The College of American Pathologists Laboratory Accreditation Program requires that all laboratories evaluating cervicovaginal cytology enroll in the PAP Program or an equivalent glass-slide program. As the single largest glass-slide educational cervicovaginal cytology interlaboratory comparison program, the PAP Program served 1169 laboratories in 2001, including 3831 pathologists and 3407 cytotechnologists. Cytology laboratories of all types participate, with the largest number (approximately 60%) being hospital laboratories. In addition, independent laboratories, federal and government laboratories, university laboratories, and others such as those associated with a group practice or physician's office also participate. The number of participants did not change significantly in 2002.

Participants and committee members generously contribute slides to the program. After receipt and accessioning into the program, the slides are reviewed by at least 3 experienced cytopathologists from the College of American Pathologists Cytopathology Resource Committee. Before acceptance into the program, each slide must be judged to be of good technical quality and an excellent example of the reference interpretation. All 3 reviewers must agree on the exact target interpretation, which must agree with the submitted interpretation. Further, slides submitted with an interpretation of low-grade squamous intraepithelial lesion or higher must be biopsy confirmed prior to acceptance into the program.

The PAP Program consists of 5 glass slides of cervicovaginal material mailed 4 times a year. The coded answer sheets have diagnostic menus that use terminology modified from the Bethesda system (Table 1). Referenced slides are placed into 1 of 3 selection series: the 000 series for unsatisfactory slides; the 100 series for normal, infectious, and reparative conditions; and the 200 series for epithelial cell abnormalities and carcinoma.

In the year 2001, cases with a reference interpretation of adenocarcinoma in situ were introduced into the program. As a result, few cases were available that had met the program's validation criteria, one of which requires at least 20 reviews by participants. Cases that have not yet met the validation criteria are placed into educational sets for circulation. Therefore, cases of adenocarcinoma in situ were only compared with cases of adenocarcinoma, high-grade squamous intraepithelial lesion, and squamous cell carcinoma that were in the educational portion of the program and had also not yet met the criteria for validation.

In this analysis, any interpretation in the 200 series (including low-grade squamous intraepithelial lesion, high-grade squamous intraepithelial lesion, squamous cell carcinoma, adenocarcinoma, adenocarcinoma in situ, and high-grade squamous intraepithelial lesion/carcinoma, not otherwise specified [NOS]) is counted as a true-positive result. Only cases interpreted as unsatisfactory, as negative, or as only containing organisms were counted as false-negative results. The percentages of review interpretations in exact agreement with the reference interpretation and the percentage of cases with a review interpretation of glandular abnormality (adenocarcinoma in situ or adenocarcinoma) among those with a reference interpretation of adenocarcinoma in situ or adenocarcinoma were calculated. In addition, the percentage of cases for which all review interpretations were concordant with the reference interpretation was calculated. All cases were conventional Papanicolaou tests. Categoric results were compared using a 2-tailed chi-square test unless specifically noted.

During the 2 years studied, a total of 26 separate cases of adenocarcinoma in situ were available for review. There were 213 reviews of these 26 Papanicolaou tests categorized as adenocarcinoma in situ, compared with 2821 reviews of adenocarcinoma, 7535 reviews of high-grade squamous intraepithelial lesion, and 1886 reviews of squamous cell carcinoma.

The false-negative rate for adenocarcinoma in situ (11.7%) was significantly greater than that for high-grade squamous intraepithelial lesion (4.6%, P < .001) and squamous cell carcinoma (3.3%, P < .001). The false-negative rate of adenocarcinoma (8.9%, P = .16) was not statistically significantly different from that of adenocarcinoma in situ.

The percentage of adenocarcinoma in situ cases that were recognized as glandular (either adenocarcinoma in situ or adenocarcinoma, NOS) was compared with the percentage of adenocarcinoma cases that were recognized as glandular. Of the 213 reviews of adenocarcinoma in situ, 146 (68.5%) were recognized as glandular, compared with 2025 (71.8%) of 2821 cases of adenocarcinoma. These percentages were not statistically significantly different (P = .31).

Adenocarcinoma in situ was specifically recognized in only 99 (46.5%) of 213 reviews, compared with 1873 (66.4%) of 2821 reviews of adenocarcinoma that were specifically recognized as adenocarcinoma (P < .001) and 2036 (72.2%) of 2821 reviews of adenocarcinoma that were specifically recognized as either adenocarcinoma or high-grade squamous intraepithelial lesion/carcinoma, NOS (P < .001).

Of the 26 cases of adenocarcinoma in situ, 20 had at least 5 reviews. Of these 20, no cases were recognized as adenocarcinoma in situ in all reviews, compared with 63 (22.9%) of 275 (P = .01) of adenocarcinoma cases with at least 5 responses that were specifically recognized as adenocarcinoma and 73 (26.5%) of 275 of adenocarcinoma cases with at least 5 responses that were specifically recognized as either adenocarcinoma or high-grade squamous intraepithelial lesion/carcinoma, NOS, in all reviews (P = .005, 2-tailed Fisher exact test). Of squamous cell carcinoma and high-grade squamous intraepithelial lesion cases with at least 5 reviews, 41 (21.7%) of 189 squamous cell carcinomas and 115 (15.1%) of 763 high-grade squamous intraepithelial lesions had exact review interpretations in all reviews. When high-grade squamous intraepithelial lesion/carcinoma, NOS, is considered an acceptable response, the corresponding proportions are 68 (36.0%) of 189 and 196 (25.7%) of 763, respectively. The comparison of adenocarcinoma in situ to high-grade squamous intraepithelial lesion when high-grade squamous intraepithelial lesion/carcinoma, NOS, is not considered an acceptable response approaches statistical significance (P = .06). All other comparisons of the percentage of cases with exact review interpretations in all reviews showed statistical significance.

The false-negative rates, the percentage of reviews in which the reference interpretation was specifically recognized, and the percentage of cases in which the reference interpretation was specifically recognized in all reviews for adenocarcinoma in situ, adenocarcinoma, squamous cell carcinoma, and high-grade squamous intraepithelial lesion are presented in Table 2.

Of the 213 adenocarcinoma in situ review interpretations, 25 were false negatives. Twenty-one of these were interpreted as negative for intraepithelial lesions or as containing organisms. Three were interpreted as unsatisfactory, and one was interpreted as atrophic vaginitis. Frequencies of interpretations of adenocarcinoma in situ cases are shown in Table 3.

The histology of adenocarcinoma in situ was first described in the early 1950s as glands with features resembling invasive adenocarcinoma but lacking stromal invasion. Changes similar to adenocarcinoma in situ were noted at the periphery of invasive adenocarcinoma, and adenocarcinoma in situ was thought to be a precursor of invasive adenocarcinoma. Common cytologic features of adenocarcinoma in situ include tightly crowded sheets of malignant cells that are architecturally disorganized and that may have short strips of pseudostratified columnar cells protruding off the edges. Nuclei at the periphery or in the strips may be partially denuded, which gives a frayed or feathered appearance. Isolated strips and rosettes may be seen. More poorly differentiated cases may not be as tightly crowded. Nuclei are enlarged, usually oval, and moderately to markedly hyperchromatic. Poorly differentiated cases may have round or bizarrely shaped nuclei. The chromatin is finely to moderately granular. Nucleoli are present in approximately half of the cases. Notably, tumor diathesis is absent.22,23 The cytologic features are illustrated in Figures 1 and 2.

Defining the difficulty of making the exact interpretation for a particular diagnostic entity is challenging, since the results of any study will depend heavily on the cases that are included in that study. We believe that adenocarcinoma in situ is relatively difficult to both recognize and precisely subcategorize. To support this hypothesis, we have chosen to compare the difficulty of interpreting adenocarcinoma in situ with the difficulties of interpreting adenocarcinoma, high-grade squamous intraepithelial lesion, and squamous cell carcinoma. For all the cases included in this study, the process for selecting the slides is similar in that each case is reviewed by 3 cytopathologists, and each case has a biopsy follow-up. The uniformity of this selection process supports the idea that these cases can be appropriately compared.

However, there are differences among these groups. First, there are many more cases of adenocarcinoma, high-grade squamous intraepithelial lesion, and squamous cell carcinoma than of adenocarcinoma in situ. This is both because of the relative incidence of the disease and because cases of adenocarcinoma in situ have been specifically categorized and used in the program only within the past 2 years. Most cases suggestive of adenocarcinoma in situ have been classified with atypical glandular cells in the past, and previous Papanicolaou surveys show that atypical glandular cells have a median laboratory reporting rate of 0.3%.24 It is possible that the performance we have demonstrated will improve with the inclusion of additional cases of adenocarcinoma in situ. Second, adenocarcinoma in situ is a relatively new diagnostic option in the program that has been introduced only in the past few years; the other diagnoses have been used for more than a decade. It is not unreasonable to assume that there is a learning curve associated with the introduction of each new interpretation and that the performance will improve as participants get accustomed to making that interpretation. It has been shown that with specific directed educational efforts, the recognition of adenocarcinoma in situ can be improved.8 

Nevertheless, with these caveats in mind, the results of this study are quite striking. In most respects that we can measure, adenocarcinoma in situ is more difficult to diagnose than adenocarcinoma, high-grade squamous intraepithelial lesion, or squamous cell carcinoma. Cases of adenocarcinoma in situ are more frequently missed than are cases of high-grade squamous intraepithelial lesion and squamous cell carcinoma and are less frequently specifically recognized than are cases of adenocarcinoma, high-grade squamous intraepithelial lesion, and squamous cell carcinoma. Adenocarcinoma in situ and adenocarcinoma are recognized as glandular lesions at similar rates, however. In addition, there were no individual cases in which all participants agreed that the interpretation was adenocarcinoma in situ. This is significantly less than the percentage of cases of high-grade squamous intraepithelial lesion, adenocarcinoma, and squamous cell carcinoma.

Reasons for the increased misclassification rate observed in this study are not known precisely, but it is possible that the abnormal architecture of the groups of endocervical cells was not appreciated. In the false-negative cases, the groups were likely overlooked or interpreted as reactive. Twenty-four (11.3%) of the review interpretations were high-grade squamous intraepithelial lesions or squamous cell carcinomas. This observation illustrates one of the difficulties of recognizing adenocarcinoma in situ. High-grade squamous intraepithelial lesions with glandular extension may have an appearance that is very similar to the crowded sheets of adenocarcinoma in situ, and high-grade squamous intraepithelial lesions are much more common entities. In these cases, the presence of a significant lesion would be recognized, and the correct diagnosis would probably be made on a follow-up histologic specimen.

The results of this study suggest that adenocarcinoma in situ cannot be routinely recognized or precisely subcategorized in routine practice. The Papanicolaou test was designed to detect squamous abnormalities. Despite the increased interest in and recognition of glandular lesions, there is no evidence that this test is an effective screening tool for these lesions. A major reason for this is because of sampling. However, our results suggest that simply recognizing and categorizing these lesions remain a problem as well. At present, there is little evidence that most cases of adenocarcinoma in situ can always be recognized as such in routine clinical practice. According to Bethesda 2001, Papanicolaou tests with specific criteria for adenocarcinoma in situ can be classified as such.17 However, this study suggests that laboratories and clinicians cannot expect all biopsy-proven adenocarcinoma in situ cases to be specifically identified. Cytopathology professionals may need to educate providers about this problem. Also, missed cases of adenocarcinoma in situ constitute a significant proportion of litigated Papanicolaou test cases in which a false-negative interpretation of the slide in question is alleged.25 This study indicates that it may not be practical to assert that a reasonable practitioner standard of care requires the detection of adenocarcinoma in situ in all cases. Whether or not there are specific morphologic characteristics that are associated with an increased likelihood of a false-negative interpretation is unknown at this time.