Fax: (212) 639-6318
Anorectal cytology as a screening tool for anal squamous lesions
Cytologic, anoscopic, and histologic correlation
Article first published online: 9 DEC 2003
Copyright © 2003 American Cancer Society
Volume 102, Issue 1, pages 19–26, 25 February 2004
How to Cite
Friedlander, M. A., Stier, E. and Lin, O. (2004), Anorectal cytology as a screening tool for anal squamous lesions. Cancer, 102: 19–26. doi: 10.1002/cncr.11888
- Issue published online: 11 FEB 2004
- Article first published online: 9 DEC 2003
- Manuscript Accepted: 30 SEP 2003
- Manuscript Revised: 27 AUG 2003
- Manuscript Received: 28 MAY 2003
Anorectal cytology has been increasingly used as a screening method for anal squamous lesions, particularly in high-risk, homosexual, patients with human immunodeficiency virus infection. The diagnostic cytologic, anoscopic, and histologic criteria bear some resemblance to the criteria used in cervicovaginal samples with few differences. It is important to recognize these differences because they can lead to an erroneous diagnosis of dysplasia and unnecessary procedures.
Seventy-eight anorectal cytology specimens from 51 patients were reviewed blindly. Of the 51 patients, 33 were HIV positive. The cytology specimens consisted of 75 ThinPrep (Cytyc, Boxborough, MA) and 3 conventional Papanicolaou-stained smear specimens. The revised diagnosis was compared with the original diagnosis, corresponding histology specimens, and anoscopic results, when available.
Six specimens were unsatisfactory for review. The revised diagnosis was negative in 15 patients, atypical squamous cells of undetermined significance in 3 patients, low-grade squamous intraepithelial lesions in 24 patients, high-grade squamous intraepithelial lesions in 28 patients, and squamous cell carcinoma (SQC) in 2 patients. Five patients with an original diagnosis of SQC had the diagnosis revised upon review of their specimens. It is noteworthy that these five specimens showed the presence of atypical parakeratotic cells. Thirty-two patients had anoscopic evaluation and 30 patients had histologic correlation. Twenty-seven patients with abnormal anoscopic findings had confirmed abnormal histologic findings. Twenty- five of the 32 (78%) patients had abnormal cytology that correlated with abnormal anoscopic findings.
Anorectal cytology is an accurate method for screening patients for anal squamous lesions. Atypical parakeratotic cells represent a potential pitfall. Anoscopy is important in confirming the presence of a lesion, but only a biopsy can accurately determine the grade of a lesion. Cancer (Cancer Cytopathol) 2004;102:19–26. © 2003 American Cancer Society.
Anorectal disease is common among patients with human immunodeficiency virus (HIV) infection.1–7 Anal squamous cell carcinoma (SQC) is a rare tumor but it is the fourth most common reported malignancy among men with HIV infection.8 The estimated incidence of anal SQC in homosexual men before the onset of the HIV epidemic was approximately 35 per 100,000. This incidence is close to the incidence of SQC in women before the use of cervical cytology screening programs. It is noteworthy that the incidence of SQC in HIV-positive men is twice that of HIV-negative homosexual men.
The exact pathogenesis of anal SQC remains unknown, although it probably arises and behaves the same way as cervicovaginal lesions do in women. Anal SQC has comparable histologic features with cervical SQC. It is frequently associated with squamous intraepithelial neoplasia and has a strong association with human papillomavirus (HPV).9 Moreover, both lesions arise frequently in the squamocolumnar epithelium junction.10 Cytology has been proposed as a potential screening tool in the evaluation of anorectal specimens for anorectal disease because of the morphologic similarities of anal and cervical SQC and intraepithelial neoplasias.10, 11 In addition, cytologic screening is recognized as a cost-effective and simple method. Recently, the 2001 Bethesda guidelines for cervical cytology included an appendix for anal cytology.12
The principles applied in anal cytology are supposedly the same as cervical cytology. However, there are few studies addressing morphologic findings in anal cytology to support this assumption. Scholefield et al.13 reported some guidelines regarding anal cytology but included only 30 anal preparations without histologic correlation or follow-up. Sherman et al.14 described some morphologic features encountered in smears and ThinPrep (Cytyc, Boxborough, MA) preparations from anal specimens, but no histologic or anoscopic correlation was provided. The objectives of the current study include the evaluation of anorectal cytology as a screening tool, the correlation of anorectal cytology with anoscopic and histologic findings, and the identification of potential cytologic diagnostic pitfalls.
MATERIALS AND METHODS
Seventy-eight consecutive anorectal specimens, representing 51 patients, were collected from the files of the cytology service at Memorial Sloan-Kettering Cancer Center (New York, NY) over a period of 3 years. Thirty-three of the 51 patients were HIV positive (27 males and 6 females). All HIV-positive females also had a history of gynecologic (vulvar, vaginal, or cervical) intraepithelial neoplasia. The mean age of the patients was 43 years (range, 26–74 years). The 75 ThinPrep specimens and 3 conventional smears were stained with the Papanicolaou stain. Of the 51 patients, 32 underwent anoscopy (5 received standard anoscopy and 27 received high-resolution anoscopy [HRA]). The anoscopic impression was classified as benign, condyloma (including anal intraepithelial neoplasia [AIN] grade I) and anal dysplasia (including AIN II or higher-grade lesions). Thirty of these 32 patients had concurrent or subsequent anorectal biopsy specimens. All cytology slides were screened blindly by a cytotechnologist who recorded a revised diagnosis and the presence of glandular cells, squamous metaplastic cells, anucleate squames, microorganisms, bacteria, inflammation, viral effect, fecal material, parakeratosis, atypical parakeratosis, koilocytes, dysplastic squamous cells, and neoplastic cells. Terminology, criteria, and guidelines for the evaluation of anorectal specimens paralleled those used for gynecologic cytology as suggested in the 2001 Bethesda guidelines. The cytology specimens were classified as negative for malignancy, atypical squamous cells of undetermined significance (ASCUS), low-grade squamous intraepithelial lesion (LSIL), high-grade squamous intraepithelial lesion (HSIL), and SQC. The histologic diagnoses were classified as benign, AIN, and SQC. Grades I, II, and III were assigned to all cases of AIN. All cytology and histology specimens were reviewed by a cytopathologist (OL) for a final diagnosis. The cytologic and anoscopic findings were correlated with the histologic findings.
Review of cytology specimens demonstrated that six specimens were unsatisfactory for evaluation because of the paucity of squamous cells. The ‘revised’ diagnoses were negative for malignancy (n = 15), ASCUS (n = 3), LSIL (n = 24), HSIL (n = 28), and SQC (n = 2). The revised diagnosis was concordant with the histologic diagnosis in 26 specimens representing 20 patients (Table 1). Two specimens diagnosed cytologically as LSIL and HSIL had concurrent benign biopsies. Nine histologically diagnosed HSIL were undercalled cytologically as unsatisfactory for evaluation (n = 1), negative for malignancy (n = 1), and LSIL (n = 7). With histology as the gold standard, the sensitivity of cytology to distinguish benign from dysplastic or malignant lesions was 92%, but the specificity was only 50% (Table 1).
Of the 32 patients who underwent anoscopy (both standard and HRA), 25 (78%) had abnormal cytology that correlated with anoscopic findings of condyloma, anal or perianal dysplasia. One patient was believed to have HSIL on cytology, but no lesion was found on anoscopy. This patient had a previous history of perianal HSIL. In three patients with abnormal anoscopic findings on HRA, no abnormal cells were identified in the cytology specimens. Two specimens considered to be cytologically benign correlated with normal anoscopic findings. One specimen was considered unsatisfactory for cytologic evaluation and the anoscopic examination revealed the presence of dysplasia.
Using histology as the gold standard, HRA was more accurate in predicting the grade of the lesion than standard anoscopy. Twenty-one of 25 (84%) specimens evaluated by HRA corresponded exactly to the histologic findings. Standard anoscopy, in contrast, correlated with the histologic findings in only 2 of 5 cases (40%). The anoscopic and histologic findings are summarized in Table 2.
|Histology||Standard anoscopy||High-resolution anoscopy|
|Benign||Condyloma or AIN I||AIN II or higher||Benign||Condyloma or AIN I||AIN II or higher|
|AIN II or higher||3||2||17|
Complete agreement between the initial and revised cytologic diagnosis occurred in 51(65%) specimens (Table 3). Five specimens originally diagnosed on cytology as SQC had the diagnosis revised to HSIL (one specimen), LSIL (two specimens), ASCUS (one specimen), and negative (one specimen). The histologic diagnosis in the four abnormal specimens was dysplasia. The specimen with the revised diagnosis of negative for malignancy was confirmed histologically. Eight specimens initially reported as ASCUS or LSIL were revised as HSIL. Six of these eight specimens were confirmed histologically.
|Change in diagnosis||No. of cases|
|Negative to unsatisfactory specimen||4|
|Negative to LSIL||1|
|ASCUS to ≥ SIL||6|
|ASCUS to LSIL||4|
|ASCUS to HSIL||1|
|ASCUS to SQC||1|
|LSIL to HSIL||7|
|HSIL to other diagnosis||4|
|HSIL to negative||1|
|HSIL to LSIL||3|
|SQC to other diagnosis||5|
|SQC to negative||1|
|SQC to ASCUS||1|
|SQC to LSIL||2|
|SQC to HSIL||1|
The cytomorphologic features are summarized in Table 4. A transformation zone component of glandular cells and/or squamous metaplastic cells was identified in 70 (90%) specimens, of which 69 specimens were ThinPrep preparations. Fifty-three of the 57 (93%) abnormal specimens (ASCUS/LSIL/HSIL/SQC) contained a t-zone component. Four abnormal specimens (one ASCUS, two LSIL, and one SQC) did not contain a t-zone component. Two of the three conventional preparations did not contain a t-zone component, although one of these specimens contained malignancy.
|Revised cytologic diagnosis||T-zone component||Anucleated squames||Coarse chromatin pattern||Irregular nuclear membrane||Increased N/C ratio||Parakeratosis||Atypical parakeratosis||Koilocytes||Inflammation||Bacteria||Fecal material|
|Glandular cells||SQM||> 50%||< 50%|
|UNS (n = 6)||2||0||1||4||3||1||2||3|
|Negative (n = 15)||11||13||9||6||1||10||6||11||11|
|ASCUS (n = 3)||1||2||3||1||3||3||1||2||3||3|
|LSIL (n = 24)||16||20||12||11||13||8||21||22||16||10||3||15||13|
|HSIL (n = 28)||17||27||8||9||18||15||31||26||22||3||1||14||13|
|SQC (n = 2)||1||1||1||2||2||2||2||2||1||0||0|
Parakeratosis was present in 66 (84%) specimens. Atypical parakeratotic cells and koilocytes were present in 48 (62%) and 16 (21%) specimens, respectively. Two specimens exhibited changes suggestive of Herpes virus infection. The changes in one of them masqueraded an SQC as observed in the biopsy specimen.
The use of anorectal cytology as a potential screening tool for anal lesions in high-risk populations, such as HIV-positive men, is an area of increasing interest in the medical community. Many published studies advocate the use of cytology in the diagnosis of anal lesions because of similarities to cervical disease and its cost-effectiveness.9, 10 Fewer studies, however, have described the cytomorphologic features and diagnostic limitations associated with this new specimen type.13–15 The current study evaluated the use of anorectal cytology in a high-risk population and the cytomorphologic features associated with a variety of anorectal lesions.
Detection of abnormalities in ThinPrep slides is more effective than conventional preparations.14 Sherman et al.14 showed that anal ThinPrep specimens contained glandular cells twice as often as conventional smears. Our findings are similar despite the low number of conventional smears for comparison. In our series, 69 of 75 (92%) ThinPrep specimens contained a glandular component in contrast to 1 of 3 (33%) conventional smears. Sherman et al. also demonstrated that ThinPrep specimens detect almost eight times as many squamous intraepithelial lesions (SIL) compared with conventional smears, and SIL is diagnosed significantly more often in ThinPrep specimens containing t-zone elements compared with those that do not. Darragh et al.15 demonstrated that ThinPrep preparations containing a t-zone component detected AIN five times more frequently than those that completely lacked a t-zone element. These findings also appear to be true in our study. We found that 90% of the specimens contained a t-zone component and 93% of the abnormal specimens contained a t-zone component. Nonetheless, specimens containing cellular, well preserved, and adequately fixed material may be as important for determining adequacy than the presence of a t-zone component.14 More studies need to be performed to evaluate adequacy guidelines for anal cytology.
The presence of fecal material, inflammation, excessive bacteria, poor preservation, and excessive air-drying artifact may contribute to poor visualization of abnormal cells in anorectal samples. These factors tend to be more prevalent in conventional smears and the use of thin-layer, fluid-based preparations can reduce the presence of these factors, thereby contributing to fewer false-negative diagnoses.14, 15 The majority of cases in the current study were ThinPrep preparations (75 of 78), and such factors did not hinder evaluation of the specimens.
In the current study, the cytologic features of AIN lesions were similar to those observed in cervical neoplastic lesions. LSIL lesions contained cells reminiscent of mature squamous cells. They usually had eosinophilic or orangeophilic cytoplasm with enlarged round to ovoid nuclei, two-to-eight times the size of an intermediate nucleus. The nuclei were hyperchromatic and had evenly distributed, granular chromatin and smooth-to-irregular nuclear membranes (Fig. 1). HSIL lesions contained cells reminiscent of metaplastic squamous cells with basophilic and/or eosinophilic cytoplasm. The nuclei were hyperchromatic with unevenly distributed, coarsely granular chromatin and irregular nuclear contours. The nuclear-to-cytoplasmic (N/C) ratios were high, and nucleoli were not appreciated (Fig. 2). The two cases of SQC contained sheets and single pleomorphic tumor cells with variable amounts of basophilic and/or orangeophilic cytoplasm. Hyperchromatic nuclei were round, oval, and varied in shape, and they contained coarsely granular, unevenly distributed chromatin and irregular nuclear contours. Nucleoli also were present in some cells. Tumor diathesis was not readily appreciated in these two cases.
Concordance between the initial and revised cytologic diagnoses occurred in 51 of 78 (65%) specimens. Fourteen cases were undercalled, reflecting the diagnostic difficulties associated with evaluating anorectal specimens. Eight cases of HSIL were undercalled as ASCUS or LSIL. Failure to detect rare, small HSIL cells as well as interpretive problems associated with distinguishing between HSIL and atypical squamous metaplasia accounted for the misdiagnosis.
The ubiquitous presence of atypical keratinized squamous cells (Fig. 3) was a common finding in the current study, as well as in two previous studies,14, 16 and is unique to anorectal cytology. Atypical keratinized cells are usually associated with a high suspicion for an abnormal keratinized lesion or SQC in cervicovaginal specimens. However, in anal specimens, such cells should be interpreted with care. The appearance of these keratinized cells can vary from benign to markedly atypical and a false-positive diagnosis of SQC can easily be made. In the current study, atypical parakeratotic cells were prevalent and found in 62% of the specimens. Five specimens originally diagnosed as SQC and downgraded in the review contained atypical, keratinized squamous cells. These atypical parakeratotic cells contributed to the original overdiagnosis. Furthermore, these five specimens were among the first cases observed in our series, so the misdiagnosis occurred early in our learning process of evaluating these specimens. We agree with Sherman et al.14 that the diagnosis of AIN should be based on the presence of nonkeratinizing cells with altered N/C ratios and nuclear abnormalities such as hyperchromasia, coarse chromatin, and irregular nuclear contours.
Another potential pitfall that we encountered was a case of SQC, which contained nuclear changes consistent with Herpes infection (Fig. 4). The accompanying reactive epithelial changes masqueraded the SQC. Reactive epithelial changes should be approached with care to avoid a misdiagnosis of malignancy, especially in HIV-positive patients, who are prone to multiple infectious diseases.
Seven histologically confirmed HSIL cases were undercalled as LSIL. Other published studies confirmed the diagnostic difficulties associated with accurately grading anorectal neoplasia.10, 11 Palefsky et al.11 showed that the grade of disease on anal cytology did not always correspond to the histologic grade. However, Goldstone et al.10 reported that the diagnosis of HSIL in cytology specimens corresponded to a high probability that HSIL would be found at anoscopy. The cytologic diagnosis of LSIL or HSIL with a negative biopsy should not dismiss the absence of severe AIN disease. Palefsky et al. noted that HSIL occurred more commonly at the squamocolumnar junction and treatment of only visible perianal lesions may lead to undertreatment of more severe disease possibly present in the anorectal canal. In addition, it has been reported that these lesions are better visualized by HRA.10
In the current study, anoscopy was a very sensitive method to detect AIN. For determining the grade of lesion, HRA demonstrated an accuracy of 84% compared with 40% for standard anoscopy. The higher accuracy can be explained by the finding that HRA uses the same principles as colposcopy. HRA consists of the application of acetic acid and visualization of the anal mucosa under magnification (Fig. 5) as opposed to the simple visualization of the anal canal used in standard anoscopy. Our findings emphasize the importance of HRA in the evaluation of anal lesions as recommended by several authors.5, 10
Similar to cervical carcinoma, HPV infection plays a significant role in the development of anal condyloma, AIN, and anal carcinoma.2, 3, 9, 17–20 It is noteworthy that koilocytes, a common finding in cervicovaginal cytology, were less frequent and were observed in only 16 (21%) specimens. Other studies have reported the absence of koilocytosis in anorectal smears despite the presence of AIN.15 Limited testing on cytology samples for HPV hindered our ability to investigate the relation of HPV types with anal AIN. Three of four male patients whose anorectal samples were submitted for HPV DNA assays tested positive for high-risk viral HPV DNA strains. Two of these three patients demonstrated the presence of AIN II in the anal biopsies.
The incidence of anal SQC among HIV-positive women is not known. Few studies show that the incidence of AIN among HIV-positive women is greater than the incidence among HIV-negative women.17, 18 In the current study, 6 of 19 female patients were HIV positive. All 6 (100%) had evidence of anal condyloma and/or AIN. Of 13 HIV-negative women, only 4 (31%) had evidence of AIN. A history of genital warts and infection with Chlamydia trachomatis or Herpes simplex type 2 are also associated with an increased risk for the development of anal carcinoma.20 In our study, 14 of 19 women had a history of dysplasia in the vulva, vagina, or cervix. Of these 14 women, 4 (29%) were diagnosed with AIN disease. These findings suggest that HIV infection and a history of sexually transmitted disease represent risk factors for the development of AIN and anal SQC.
The high sensitivity obtained in the current series (92%) suggests that the use of anorectal cytology in the evaluation of anal dysplasia can be useful, particularly in high-risk individuals, such as those who are HIV positive. Many cytologic similarities exist between squamous intraepithelial disease in anorectal and cervical sites. However, diagnostic pitfalls exist. The cytologic diagnosis of dysplasia should be based on nuclear alterations and N/C ratios rather than on cytoplasmic features. In addition, the presence of opportunistic infections such as Herpes simplex virus and marked reactive/reparative cellular changes should not preclude the diagnosis of malignancy. The current study demonstrates that anorectal cytology is useful in the early detection of neoplastic processes; however, special attention must be given to potential pitfalls. Anoscopy is important in confirming the presence of a lesion, particularly HRA. Nonetheless, only a biopsy can accurately determine the grade of a lesion.
- 12National Cancer Institute. Bethesda Forum: specimen adequacy post-workshop recommendations-anal rectal cytology (appendix). Available from URL: http://bethesda2001.cancer.gov/postwrkshp_recs.html[accessed April 2002].