| |
Performance of anal cytology in a clinical setting when measured against histology and high-resolution anoscopy findings
|
| |
| |
AIDS:
28 January 2010 - Volume 24 - Issue 3 - p 373-379
Nathan, Mayura; Singh, Naveena; Garrett, Nigel; Hickey, Nicola; Prevost, Teresa; Sheaff, Michael
aDepartment of Sexual Health, Homerton University Hospital NHS Foundation Trust, UK
bBarts Sexual Health Centre, UK
cDepartment of Pathology, Barts and the London NHS Trust, London, UK
dDepartment of Public Health and Primary Care, University of Cambridge, Cambridge, UK.
"In this study, we assess liquid-based anal cytology as a possible screening tool in a HRA clinic set-up to treat anal neoplasia. Compared with histology, anal cytology has a sensitivity of 47-93% and with a corresponding specificity ranging from 32 to 92% in published literature......Despite the overall lower sensitivity in follow-up patients, anal cytology performs equally well as in new patients when two or more quadrant disease is present. Thus, the reduced sensitivity in follow-up patients is likely to be dependent on the higher frequency of one-quadrant disease following treatment and the lesion size. In this study, we have taken any abnormality in the anal pap test to indicate tissue abnormality in the anal canal......our findings suggest this enhanced sensitivity to be dependent on CD4 cell counts, as supported by no difference between HIV-negative group and HIV-positive group with a CD4 cell count of over 400 cells/µl......there was an association of large area disease with lower CD4 cell counts. Third, there was a higher incidence of high-grade disease in those with lower CD4 cell counts. This observation may have clinical implications. Anal cytology when performed soon after HIV diagnosis or when CD4 cell counts are less than 400 cells/µl may be a more sensitive tool.....given the late diagnosis of HIV in a proportion of people and further increased risk of anal cancer in this group, early screening will be prudent......The above observations may explain the variable sensitivity reported in the screening studies. Anal pap test necessarily needs to be followed up by HRA and biopsy for the detection of AIN 2/3"
Abstract
Objectives: The success of cervical cytology in screening for cervical neoplasia has led to the concept of anal cytology screening for anal neoplasia. Our objective is to study the performance of anal cytology as a screening tool.
Design: We assessed anal cytology against histology and high-resolution anoscopy in a clinical setting.
Methods: Anal pap test was obtained prior to high-resolution anoscopy examinations and biopsies. The results were analysed against a number of patient variables.
Results: From 395 individuals (93% men), 584 anal pap tests were obtained. HIV status was positive in 212 (54%) and negative in 156 (39%) individuals. On the basis of 288 histology results, the sensitivity of anal cytology to detect disease was 70% [95% confidence interval (CI) 64-75], whereas the specificity was 67% (95% CI 38-88). For high-grade disease (anal intraepithelial neoplasia 2/3), sensitivity of anal cytology was 81% (95% CI 70-90), and the negative predictive value was 85% (95% CI 76-92). Sensitivity was dependent on the area of disease (86% for two or more quadrants vs. 69% for one or more quadrants, P = 0.002) and HIV positivity (76% in HIV positive vs. 59% in HIV negative, P = 0.009). Amongst HIV-positive patients, the sensitivity was 90% when CD4 cell count was 400 cells/µl or less compared with 67% when CD4 cell count was above 400 cells/µl (P = 0.005).
Conclusion: Anal cytology performs similar to cervical cytology in a clinical setting. Sensitivity of anal smear is dependent on the area (quadrants) of disease present. Sensitivity of anal cytology is enhanced when CD4 cell count is less than 400 cells/µl in HIV-positive men. Our results may explain the variable sensitivity reported in the literature.
Introduction
A dramatic decline in age-standardized incidence rates (per 100 000) of cervical cancer was observed in south-east UK, from 1987 [12.68, 95% confidence interval (CI) 12.30-13.06] to 2002 (6.66, 95% CI 6.40-6.92), and has been ascribed to the introduction of National Health Service's organized cervical screening programme in 1988 [1]. During the same period, a consistent increase in anal cancer incidence was noted in both men (from 0.88 to 1.06) and women (from 0.81 to 1.18) [1]. In specific groups, such as HIV-positive patients, the risk of anal cancer is 60 per 100 000 patient-years (mean value, 1984-2003) and increased from 35 in the pre-HAART era (1984-1995) to 92 in the post-HAART era (1996-2003) [2]. From 1992 to 2003, similar increases in anal cancer in both the general population-based rates and in rates amongst HIV-infected patients have been noted in the United States [3]. Several studies [4-6] now demonstrate the evolution of invasive anal cancer from untreated or partially treated high-grade anal intraepithelial neoplasia (AIN3). We and other groups have previously published that effective treatment is feasible for high-grade AIN [7-9]. Identification and treatment of high-grade cervical neoplasia through cervical screening programmes have led to the dramatic decline in cervical cancer rates. Anal cytology has been looked at by several groups as a screening tool to detect those at risk of high-grade anal neoplasia, thus paving the way for treatment of anal neoplasia [10,11]. We set out to assess anal cytology against high-resolution anoscopy (HRA) and histology in two clinical units, conducted by the same specialist service, in the UK.
Discussion
The increasing incidence of anal squamous carcinoma in populations in Europe and USA has been well documented [1-3,13]. But, we still lack an authenticated screening tool for anal cancer precursor lesions. In one study [14], screening 10 049 women over 5-7 years, it showed that six of the 17 patients of invasive cervical cancer were missed by cervical cytology. Further, colposcopically directed biopsy, the gold standard for diagnosis of cervical neoplasia, has also its inaccuracies [15]. Despite the shortcomings in cervical cytology as a screening tool, its introduction to screen at-risk women has led to dramatic decline in cervical carcinoma rates both across Europe and USA [1,3]. In this study, we assess liquid-based anal cytology as a possible screening tool in a HRA clinic set-up to treat anal neoplasia. Compared with histology, anal cytology has a sensitivity of 47-93% and with a corresponding specificity ranging from 32 to 92% in published literature [10,11,16]. Our findings of sensitivity of 70% to identify any anal neoplasia with 81% sensitivity to detect high-grade lesion are in agreement with previous studies on screening with anal cytology [10,17]. Further, our specificity of 67% for any lesion compares well with previous publications. Our study, however, differs in a number of aspects. This is a real-life, clinical practice-based study. It compares, thus far, the largest number of histology diagnosis with anal cytology data. Interobserver variation is kept to a minimum by one individual undertaking all HRA examination and directed biopsies. This study also includes a large number of HIV-negative individuals and heterosexual individuals, enabling a comparative assessment in these groups against HIV-positive MSM group. Although anal intercourse is a well recognized risk factor for anal cancer precursor disease, presumably through the introduction of HPV, the presence of HPV in the anal canal has been recognized in the absence of anal intercourse [18].
One drawback of population-based screening studies is that they do not generate many histology results for comparison. The large histology sample, currently regarded as the gold standard for anal neoplasia diagnosis, enables us to draw a number of important conclusions from this study. In anal cytology validation, we demonstrate for the first time, the significance of quadrants of disease. Further work will be needed to increase the sensitivity of screening methods to identify smaller lesions. It is interesting to note that Pretorius et al. [19] recently found a similar difference in the sensitivity of cervical cytology in relation to the quadrants of cervical disease. One of the clinical dilemmas in assessing disease in colposcopy and HRA is the entity of 'nonspecific acetowhite change'. Mild or indistinct acetowhite change on acetic acid application is often thought to be HPV-related or due to inflammatory changes or due to injury. Thus, all 'nonspecific acetowhite changes' are not routinely biopsied by colposcopists. This study provides evidence that the sensitivity of anal cytology is not affected by these areas. Several population-based studies have used anal cytology for screening and have restricted HRA examination, and therefore AIN 2/3 diagnosis, to those with abnormal cytology [20,21]. Unlike our study in which all patients underwent HRA examination, some population-based studies have not included the false-negative smear cases with AIN 2/3 abnormality in their result analysis.
Similar to the previous authors, we demonstrate a higher sensitivity in HIV-positive men with anal cytology. However, our findings suggest this enhanced sensitivity to be dependent on CD4 cell counts, as supported by no difference between HIV-negative group and HIV-positive group with a CD4 cell count of over 400 cells/µl. There may be a number of explanations for this. First, the presence of HPV on anal epithelium is dependent on CD4 cell counts, increasing as the CD4 cell counts drop [22]. All cytology changes, including evidence of HPV, have been interpreted as abnormal. Second, there was an association of large area disease with lower CD4 cell counts. Third, there was a higher incidence of high-grade disease in those with lower CD4 cell counts. This observation may have clinical implications. Anal cytology when performed soon after HIV diagnosis or when CD4 cell counts are less than 400 cells/µl may be a more sensitive tool. With ART, CD4 cell counts increase and CD4 cell count nadir occurs early in the diagnosis of HIV. In addition, given the late diagnosis of HIV in a proportion of people and further increased risk of anal cancer in this group, early screening will be prudent [23]. The above observations may explain the variable sensitivity reported in the screening studies. Anal pap test necessarily needs to be followed up by HRA and biopsy for the detection of AIN 2/3 [24].
Our study further demonstrates the usefulness of anal cytology in patient follow-up after treatment of anal neoplasia. Despite the overall lower sensitivity in follow-up patients, anal cytology performs equally well as in new patients when two or more quadrant disease is present. Thus, the reduced sensitivity in follow-up patients is likely to be dependent on the higher frequency of one-quadrant disease following treatment and the lesion size. In this study, we have taken any abnormality in the anal pap test to indicate tissue abnormality in the anal canal. Similarly, several previous authors did not find a concordance between the grades of anal cytology and the grades of biopsy histology [10,11,24].
In this study, we did a retrospective analysis of prospectively collected data, which had several limitations. The number of smears on women and some ethnic groups was small, and therefore, the results for these groups lacked power. In this study, the number of people over 50 years of age was small, thus making interpretation in this group difficult. There is some evidence to suggest that there is an increasing incidence of anal cancer in older age groups, in both men and women, and hence, this area needs further study [1]. Patients attending our HRA clinics were referred by clinicians who suspected anal disease which would have introduced a degree of selection bias and may have reduced the generalizability of the study. Observer bias was reduced by pathologists being unaware of clinical study data. Previous studies have shown that variability in histology reporting occurs and that this may affect results when comparing the two groups [25]. Our large histology sample in this study may go some way to alleviate this problem. We did not observe a difference in sensitivity of anal cytology between those on ART and those without. However, our data did not capture the duration of ART prior to anal cytology testing, and as such, we cannot exclude a cumulative effect of ART on sensitivity.
Through this study, we demonstrate the usefulness of anal pap test in a clinical setting. Given the inaccuracies that exist with any method of evaluation, we need to have a number of assessment methods for the diagnosis of anal neoplasia, as is the case with cervical neoplasia. This study supports the introduction of an earlier screening in HIV-positive patients, given their higher burden of anal disease and increased sensitivity with lower CD4 cell count levels. Additionally, we demonstrate the use of anal cytology after treatment of anal neoplasia for screening for recurrent disease.
Results
During the study period, a total of 604 anal cytology samples were collected, of which four were inadequate whereas 16 had no results, leaving 584 smear results for evaluation. The 16 samples with no results were lost in transit to the laboratory. The group consisted of 367 men (93%) and 28 women. The median age of the study group was 36 years (interquartile range 30-43 years). One hundred and seventy-nine (45.3%) patients were of white British origin, 47 (11.9%) were of other white European origin and 34 (8.6%) were of black ethnicity. Other ethnicities varied from 0.3 to 3.5%. In terms of HIV serostatus, 212 (54%) patients were positive, 156 (39%) were negative and in 27, the status was not known. Of the patients, 297 (75.2%) were MSM, 65 (16.5%) were heterosexual and sexuality was not evident in 33 (8.4%).
Overall sensitivity of liquid-based anal cytology (anal pap test) was 70% in comparison to histology outcomes of biopsies. A similar sensitivity of 69% was measured against HRA examination findings. Further, a PPV of 97% against histology of any grade was observed. Compared with HRA results, anal cytology (anal pap test) had an NPV of 64% (Table 1).
Sixty-one patients had histology samples obtained from areas of 'nonspecific acetowhite change' of the mucosa noticed on HRA examination, with simultaneous cytology sampling. Sensitivity of anal pap test was 68% (255/376) when these changes were included in the analysis compared with 69% (218/315) when only substantive disease was considered (Table 2). Sensitivity of anal cytology was not affected by ethnicity, history of smoking, date of examination nor by the institution of examination (data not shown). Sensitivity of anal cytology was 71% in MSM compared with 58% in heterosexuals (P = 0.15) for histology of any grade. For HRA examination, sensitivity was 71% in MSM and 58% in heterosexuals (P = 0.16). In a multivariate analysis, using logistic regression with adjustment for HIV status, the difference in sensitivity between MSM and heterosexuals was lessened (P = 0.35 for HRA, P = 0.62 for histology of any grade). With high-grade disease, sensitivity of anal cytology was 80% in MSM and 81% amongst heterosexuals (P = 0.99). Sensitivity of anal cytology was not affected generally by age. However, there was a suggestion of reduced sensitivity in the individuals above 50 years of age (P = 0.06 for HRA, P = 0.17 for histology of any grade), but the number of cases was small (data not shown).
There was a significant difference (P = 0.002) in the ability of the anal pap test to predict disease between one or more quadrant disease and two or more quadrant disease (large area disease). Such difference was present both in new patients (74 vs. 85%, P = 0.033) and in follow-up patients (63 vs. 88%, P = 0.009). Similarly, the NPV of anal cytology was dependent on the volume of disease, being much reduced in cases with one or more quadrant disease compared to two or more quadrant disease (64 vs. 92%).
Sensitivity of anal cytology was much greater in HIV-positive patients compared with HIV-negative individuals. This difference was present both for HRA findings (75 and 61%, respectively, P = 0.015) and against histology of any grade (76 and 59%, respectively, P = 0.009). There was no difference in sensitivity for detection of high-grade disease by histology between HIV-positive and HIV-negative individuals (80%, 93%, P = 0.40). However, the number of patients in each group with high-grade disease was small. In HIV-positive patients, sensitivity of anal cytology was dependent on the CD4 cell counts. Sensitivity was less in those with a CD4 cell count over 400 cells/µl compared with those with a CD4 cell count of 200-400 cells/µl and less than 200 cells/µl (Table 3). High-grade disease was more common in the HIV-positive group (31%, 52/169) compared with the HIV-negative group (14%, 14/99). There was a significantly higher incidence (P = 0.001) of large area disease (two or more quadrant disease) in those with a CD4 cell count less than 200 cells/µl and between 200 and 400 cells/µl (67%, 46/69) compared with those with a CD4 cell count of over 400 cells/µl (43%, 53/122). However, using logistic regression analysis, significant difference in sensitivity (P = 0.004 for HRA, P = 0.004 for histology of high grade) in anal cytology between the HIV-positive and HIV-negative groups remained even after adjusting for high-grade disease and large volume disease. Comparison of HIV-negative group and HIV-positive group with a CD4 cell count of over 400 cells/µl did not show any difference (P = 0.26 for HRA, P = 0.29 for histology of any grade, P = 0.07 for histology of high grade). The trend towards a difference with high-grade disease will need a larger sample to verify. In contrast to CD4 cell count values, sensitivity of anal pap test was not affected by the 'current use' of antiretroviral treatment (ART). Against histology of any grade, sensitivity of anal cytology was 77% (86/111) in those taking ART and 73% (30/41) in those not on ART (P = 0.64). Similarly, against HRA examination findings, anal cytology had a sensitivity of 77% (101/132) in those taking ART compared with 73% (38/52) sensitivity in those on no treatment (P = 0.67). The number of patients with high-grade disease (AIN 2/3) was small and showed an 85% (33/39) sensitivity when ART was taken compared with 60% (6/10) when not taken (P = 0.15).
Using a multivariate analysis on sensitivity adjusting for quadrant, grade and CD4 cell count in HIV-positive cases, a significant difference (P < 0.001) remained between one or more quadrant group compared with those with two or more quadrant disease. A significant difference (P = 0.048) was also noted in sensitivity between the group with a CD4 cell count above 400 cells/µl compared with those with a CD4 cell count of 400 cells/µl or less. There was no difference (P = 0.46) between the group with low-grade disease compared with the group with high-grade disease. However, the number of cases with low and high-grade disease was small.
Sensitivity was higher in new patients compared with patients having follow-up visits, although the difference was marginal for high-grade disease (Table 4). This difference in sensitivity was most marked when one-quadrant disease (14.6% of new patients; 40.9% of follow-up patients) was considered (74 vs. 63%, P = 0.040) and was lost when two or more quadrant disease was compared (85 vs. 88%, P = 0.75).
We further compared the anal cytology grades with that of the histology grades (Table 5). There was poor concordance between histology grades and cytology grades (κ = 0.14). In this population, we looked at the factors that may lead to an increased risk of a positive smear, using multivariate logistic regression, considering sex, sexuality, new/follow-up status, CD4 cell count, ART, ethnicity, smoking, age and HIV status. The significant factors were HIV status (P < 0.001) and new/follow-up status (P < 0.001). The odds ratio (OR) for HIV-positive patients relative to HIV-negative individuals was 2.28, whereas the OR for new patients relative to follow-up patients was 2.62.
Methods
Liquid-based anal cytology (anal pap test) was introduced in September 2005, as a part of the work-up of patients attending HRA clinics. Study population consisted of all patients attending our services between September 2005 and June 2009. As a tertiary referral unit, patients referred consisted of those in whom a clinician suspected anal canal disease prior to referral. The primary objective of the study was to evaluate the performance of anal cytology against histological diagnosis. During each patient visit, HRA was done, but biopsies were not performed during visits for treatment or when a patient declined biopsy. As a second objective, we assessed the performance of anal cytology to detect presence of disease (any grade) against HRA examination findings. Third, we looked at factors that may impact on the performance of anal cytology when used in a clinical setting. As one of the factors, we included the 'nonspecific acetowhite changes' that were detected on HRA examination when 5% acetic acid application was made. These changes usually reflect a 'mixed bag' of human papilloma virus (HPV)-related changes, recent injury or inflammation on the mucosa as in the cervical disease.
All patients were interviewed, and their demographics and histories were obtained. Their ethnicities and sexuality were recorded as identified by patients using standard criteria used in the National Health Service. A dacron swab (cotton swabs used in one institution) moistened in water was introduced blindly into the anal canal up to 5 cm and gradually withdrawn with a rotationary movement while applying gentle pressure on the walls of the anal canal [10]. The swab was introduced into a liquid-based cytology transport medium (PreservCyt solution; Cytyc Ltd., West Sussex, UK). Examination was then continued using a high-resolution anoscope, followed by 5% acetic acid application. Any area suspicious of disease was biopsied under local anaesthetic injection using a Tischler biopsy forceps. After the examination, a diagrammatic representation of the areas of disease was drawn to indicate the number of quadrants involved and the disease sites as defined by a 'clock-face'. Data related to HIV were obtained from referral details, from patient notes and, in rare instances, from patient history. All data pertaining to date of birth, sex, sexuality, ethnicity, HIV serostatus, number of quadrants of disease, attendance type (new or follow-up), history of smoking, CD4 cell counts and HIV treatment status, if applicable, as well as histology and cytology results were entered on to the HRA database. All anal cytology samples were taken by one person from one institution and by two of the authors in the second institution. All HRA examinations were carried out by one of the authors, who had several years of experience with HRA examination.
Anal cytology was read by experienced cytopathologists, and all histology was read similarly in the same institution. All histology was interpreted using a system derived from National Health Service cervical screening programme (NHSCSP) publication 10, April 1999, which provides guidelines for reporting histopathology, including cervical intraepithelial neoplasia (CIN) in cervical screening in the UK, and classified as AIN1, AIN2 and AIN3. For analysis, AIN2 and AIN3 were pooled together as high-grade disease based on the better reproducibility using the two-tier system [12]. Anal cytology was classified using the guidelines published for reporting cervical cytopathology in the UK (NHSCSP publication 1, 2nd ed., May 2000) and consisted of negative (normal), borderline changes, koilocytosis, mild dyskaryosis, moderate dyskaryosis and severe dyskaryosis. Moderate and severe dyskaryosis were pooled together as high-grade smear, whereas all other categories, except normal smear, were included in the low-grade category.
The performance of anal cytology was assessed by comparing against HRA and histology. For comparison with histology, two separate definitions were used. The first defined a positive biopsy to be any abnormalities detected on biopsy (including koilocytes and HPV). The second defined a positive biopsy result to be only high-grade disease. Sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) are presented along with their associated 95% CIs. Differences in sensitivity between factors of interest were compared using chi-squared tests assessed at the 5% level of significance. The data were analysed using SPSS version 12.0.1 (SPSS Inc., Chicago, Illinois, USA).
| |
| |
| |
|
|
|
