Risk of gynaecological malignancies in cytologically atypical glandular cells: follow-up study of a nationwide screening population
C-A Chen, Department of Obstetrics and Gynaecology, College of Medicine, National Taiwan University, No.7, Chung-Shan South Road, Taipei 100, Taiwan. Email email@example.com
Please cite this paper as: Cheng W-F, Chen Y-L, You S-L, Chen C-J, Chen Y-C, Hsieh C-Y, Chen C-A. Risk of gynaecological malignancies in cytologically atypical glandular cells: follow-up study of a nationwide screening population. BJOG 2011;118:34–41.
Objective To investigate the relationship between screening status, clinical characteristics and risk of gynaecological malignancies in women with a cytological diagnosis of atypical glandular cells (AGC).
Design Prospective study of a screened population.
Population Case series from nationwide screening population.
Methods The 8281 women who were diagnosed with cytological AGC for the first time were divided into screened (5386 women) and unscreened (2895 cases) groups according to their screening status. Follow-up histological reports were analysed.
Main outcome measures Diagnosis of cervical, uterine, or ovarian cancers.
Results Of the 323 women who developed gynaecological malignancies, 271 had invasive cervical cancers, 40 had uterine cancers and 12 had ovarian cancers, with a mean follow up of 1.9 years and 50 740 person-years. Previous screening status was a strong risk predictor of gynaecological malignancies (hazard ratio 1.69, P = 0.0027). Compared with the general screening population, women with a first diagnosis of cytological AGC had significantly increased ratios of developing gynaecological malignancies (17.85-fold for cervical cancer, 5.68-fold for uterine cancer, and 2.04-fold for ovarian cancer, P < 0.05). When compared with women aged <35 years, those in other age groups had a significantly higher risk of developing gynaecological cancers (age ≥60 years, hazard ratio 1.99, 95% CI 1.20–2.37, P = 0.016).
Conclusions Comprehensive evaluation for women with cytological AGC, including pelvic examination, ultrasonography, colposcopy, endocervical curettage, cervical biopsy and endometrial biopsy needs to be considered, especially for those with risk factors (i.e. >60 years old, lower educational status, previous Papanicolaou smear interval longer than 2 years, or no previous Papanicolaou smear).
Cytological evaluation of cells obtained from the cervix and vagina is used for detecting cervical cancer and its precursors. Expanded cytological criteria were introduced in the 1980s to improve detection of precursor endocervical lesions.1 Atypical cells were divided into atypical squamous or glandular cells of undetermined significance (ASC-US or AGC-US, respectively) in The Bethesda System (TBS) 1988.2 Later, atypical glandular abnormalities were categorised into three groups in TBS 2001: (i) atypical glandular cells (AGC) either endocervical, endometrial, or ‘glandular cells’ not otherwise specified; (ii) AGC ‘favour neoplasia’; and (iii) endocervical adenocarcinoma in situ.
An AGC is defined as a glandular cell exhibiting changes beyond reactive or reparative changes but lacking the unequivocal features of invasive adenocarcinoma.3 The reported prevalence has ranged from 0.08 to 5.96% since the AGC terminology came into use.4,5 It is not uncommon for AGC to be associated with a significant underlying neoplastic condition including adenocarcinomas of the cervix, endometrium, ovary and fallopian tube.5–9 Intensive evaluation with colposcopy, cervical biopsy, endocervical curettage, human papillomavirus (HPV) testing, and even endometrial biopsy and gynaecological sonography have been proposed for patients with AGC cytology for initial evaluation and to rule out possible neoplasia.6,7,10,11 Because of the high incidence of neoplasia, the poor sensitivity of all test modalities, and the lack of cost-effective management methods, knowledge of the incidence of the development of the various gynaecological malignancies in women with cytological AGC and the ability to identify risk factors of developing gynaecological malignancies of the these women will be important in clinical practice. Most investigations focus on the underlying pathology among patients with AGC identified during cervical Papanicolaou (Pap) smear.12 The long-term impact of cytological AGC in a cohort study also warrants investigation.
In the present nationwide population-based study, the database of cervical Pap smears and the histopathological results were retrieved from the Taiwan Cervical Cancer Screening Registration System. Women who were diagnosed as having AGC for the first time and the biological gradient and the screening status of the risk for gynaecological malignancies were investigated. The study aimed to identify risk factors of developing gynaecological malignancies for these women and to make the appropriate recommendations for these women.
The annual cervical screening programme using the Pap smear test was launched for women aged ≥30 years in 1995 in Taiwan. The results of women diagnosed with AGC were retrieved from the database of the Taiwan Cervical Cancer Screening Registration System for screenings performed from 1 January 1995 to 31 December 2004. The institutional review boards approved the study protocol.
Conventional cervical Pap smears were performed nationwide by gynaecologists, family physicians, or trained public-health nurses. For the cytological examination, around 100 cytological laboratories with cyto-technicians or cytologists classified the Pap smears. Certified cytopathologists re-checked, revised and confirmed the results. In Taiwan, reports of Pap smears taken between 1995 and 1997 were classified using TBS 1991 13 whereas those taken between 1998 and 2004 were classified using TBS 2001.5 The laboratories then reported the smear results to the respective hospitals/clinics and to the central information system for registration, which reported the abnormal Pap smear results to the Health Promotion Bureau. The Health Promotion Bureau then asked the local public nurses to inform and follow up the women with abnormal Pap smears, and report back on their management. The reported cytological abnormalities included atypical squamous cells of undetermined significance (ASC-US), atypical squamous cells that did not exclude high-grade squamous intraepithelial lesions (ASC-H), atypical glandular cells of undetermined significance (AGC), low-grade squamous intraepithelial lesions (LSILs), high-grade squamous intraepithelial lesions (HSILs), and malignant diseases.
To investigate women newly diagnosed with AGC and to avoid misclassifying follow-up Pap smears after treatment, inclusion and exclusion criteria for the study were set up. Only women with the cytological diagnosis of AGC for the first time were enrolled. The exclusion criteria were women with: (i) any previous epithelial cell changes on Pap smear (including ASC-US or AGC); (ii) previous cervical histology of LSIL/HSIL/invasive cancer; (iii) previous hysterectomy; (iv) previous pelvic irradiation; or (v) any kind of gynaecological malignancies within 6 months of AGC cytology. Women with a time from Pap smear (AGC) to diagnosis of invasive cancer longer than 6 months were included. If the interval was <6 months between Pap smear (AGC or AGC-favour neoplasia and adenocarcinoma in situ) and diagnosis was invasive gynaecological malignancy, then the case was regarded as an undetermined invasive gynaecological malignancy and was excluded for analysis.
The women with a diagnosis of cytological AGC for the first time were then divided into two groups: the unscreened group comprised women without previous Pap screening and the screened group comprised women whose previous Pap smears reported no epithelial abnormalities. The databases of the registration system (TCXR) were searched from 1 January 1995 to 31 December 2004 for the follow-up data on these patients. All cases of histopathological cervical, uterine or ovarian cancer among these women with cytological AGC diagnosed between July 1995 and December 2006 were obtained from the computerised profiles of the National Cancer Registry System of the Department of Health.
The follow up of each subject (in person-years) was calculated from the date of enrollment to the date of gynaecological cancer diagnosis, date of death, or last date of linked data available from Taiwan Cancer Registry or Death Certification Profile, whichever came first until 31 December 2006. Data for subjects whose gynaecological cancer was not diagnosed were censored on the date of death or the last date of follow up until 31 December 2006. Incidence rates were calculated by dividing the number of incident gynaecological cancer cases by the number of person-years of follow-up.
A Cox’s proportional-hazards model estimated the association between the history of Pap smear and the risk of gynaecological cancer, with adjustments to other risk factors, including age, participation year, previous screening status, educational status and residence. The 95% CI for the hazard ratios (HR) were also calculated. Statistical significance levels were determined by two-tailed test and a P value <0.05 was considered statistically significant. Statistical analysis was performed with the SAS software version 8.12 (SAS Inc, Cary, NC, USA). To compare the gynaecological cancer rates in Taiwan and the cytological AGC population, standardised incidence ratios (SIR) from gynaecological malignancies in the cytological AGC population were also calculated using the section of the general female population that was >30 years in Taiwan as the standard population (SIR = 100), The 95% CI for SIR were obtained by the exact method.14
Based on the inclusion and exclusion criteria, 8281 women were diagnosed with AGC for the first time. Their mean age was 45.1 years and their basic characteristics are shown in Table 1. Among them, 30.0% were aged ≥50 years and 65.0% had at least one normal Pap smear result. The 2895 (35%) women who had never had a Pap smear in their lives were defined as the ‘unscreened group’ and the 5386 women who had undergone previous Pap smears but without any abnormal epithelial changes were defined as the ‘screened group’. Of the screened group, 72.3% had received a Pap smear within the last 2 years. There were 114 and 86 women in the screened and unscreened groups excluded because the interval between cytology and diagnosis was <6 months.
Table 1. Baseline characteristics of women with cytological AGC according to Pap smear screening (n = 8281)
|Screened period (year)|| || ||<0.0001|| |
|1985–1997||1237 (42.7)||398 (7.4)|| ||1635 (19.7)|
|1998–2004||1658 (57.3)||4988 (92.6)||6646 (80.3)|
|Age (years)|| || ||<0.0001|| |
|<35||393 (13.6)||499 (9.3)|| ||892 (10.7)|
|35–39||467 (16.1)||950 (17.6)||1417 (17.1)|
|40–44||560 (19.3)||1206 (22.4)||1766 (21.3)|
|45–49||545 (18.8)||1185 (22.0)||1730 (20.9)|
|50–54||349 (12.1)||841 (15.6)||1190 (14.4)|
|55–59||202 (7.0)||308 (5.7)||510 (6.1)|
|≥60||379 (13.1)||397 (7.4)||776 (9.5)|
|Education status|| || ||<0.0001|| |
|Elementary School||1414 (48.8)||2180 (40.5)|| ||3594 (43.4)|
|Junior high School||360 (12.4)||730 (13.5)||1090 (13.2)|
|Senior high School||755 (26.1)||1541 (28.6)||2296 (22.7)|
|Graduate School||366 (12.7)||935 (17.4)||1301 (15.7)|
|Residential region|| || ||<0.0001|| |
|Metropolitan||645 (22.3)||1339 (24.9)|| ||1984 (24.0)|
|Cities||306 (10.6)||710 (13.2)||1016 (12.2)|
|Townships||1158 (40.0)||2069 (38.4)||3227 (39.0)|
|County||786 (27.1)||1268 (23.5)||2054 (24.8)|
|Previous screened interval (years)|| || ||<0.0001|| |
|<1||–||1621 (30.1)|| ||1621 (19.6)|
|1–2||–||2273 (42.2)||2273 (27.4)|
|>2||–||1492 (27.7)||1492 (18.0)|
|Never||2895 (100)||–||2895 (35.0)|
|Total||2895 (35.0)||5386||8281 (100)|
There were 50 739.18 person-years of follow up, with an average follow up of 6.13 years. The mean ages of screened and unscreened groups were 44.8 and 45.4 years. The mean diagnostic intervals from cytology to diagnosis in the screened and unscreened groups were 1127 ± 707 and 1161 ± 802 days. The incidences of gynaecological malignancies analysed by screening status as well as other risk factors upon enrollment are shown in Table 2. When compared with women aged <35 years, those in other age groups had significantly higher risk of developing gynaecological cancers (Table 2) (age ≥60 years, HR 1.99, 95% CI 1.20–2.37, P = 0.016). The risk of developing gynaecological cancers was significantly higher in the unscreened group compared with the screened group, regardless of previous Pap smear intervals (HR 1.69, 95% CI 1.20–2.37, P = 0.0027). However, women with higher educational status had less incidence of developing gynaecological cancers than those with lower educational status (reference, elementary school, for senior high school HR 0.73, 95% CI 0.54–0.99, P = 0.040; for college or graduate school HR 0.49, 95% CI 0.31–0.76, P = 0.0016) (Table 2). Women screened in different areas did not show any statistical difference in their development of gynaecological cancers.
Table 2. Risk factors associated with gynaecological cancers in patients with cytological AGC (n = 8281) by Cox regression analysis
|1995–1997||1635||15 855.0||106||668.6||1|| |
|1998–2004||6646||34 884.2||217||662.1||0.88 (0.68–1.16)||0.37|
|40–44||1766||10 911.7||63||577.4||1.53 (0.92–2.55)||0.10|
|45–49||1730||10 361.9||81||781.7||1.96 (1.19–3.25)||0.0086|
|Previous screened interval (years)|
|1–2||2273||12 448.3||55||441.8||0.91 (0.62–1.36)||0.65|
|Never||2895||22 036.0||171||776.0||1.69 (1.20–2.37)||0.0027|
|Elementary School||3594||22 809.7||183||802.3||1|| |
|Junior high School||1090||6696.7||45||672.0||0.91 (0.65–1.28)||0.60|
|Senior high School||2296||13 597.2||70||514.8||0.73 (0.54–0.99)||0.040|
|Colleges or graduate school||1301||7635.6||25||327.4||0.49 (0.31–0.76)||0.0016|
|Metropolitan||1984||12 540.9||66||526.3||1|| |
|Townships||3227||20 067.9||129||642.8||1.06 (0.79–1.44)||0.69|
|County||2054||12 594.3||97||770.2||1.18 (0.86–1.63)||0.31|
The 5386 women with a diagnosis of cytological AGC in the screened group were analysed further. Women >60 years had the highest risk of developing gynaecological malignancies (reference, age 30–34 years, HR 2.83, 95% CI 1.30–6.15, P = 0.0085) (Table 3). Women whose previous Pap smear was longer ago than 2 years (reference, previous Pap smear <1 year ago, HR 1.59, 95% CI 1.06–2.40, P = 0.025) also had a significantly increased risk of developing gynaecological malignancies (Table 3).
Table 3. Risk factors associated with gynaecological cancers in previously screened patients with cytological AGC (n = 5386) by Cox regression analysis
|1998–2004||4988||24 942.6||133||533.2||0.84 (0.50–1.41)||0.51|
|Previous screened interval (years)|
|1–2||2273||12 448.3||55||441.8||0.96 (0.64–1.42)||0.83|
|Elementary School||2180||11 980.6||80||667.8||1|| |
|Junior high School||730||3874.6||21||542.0||0.95 (0.58–1.56)||0.83|
|Senior high School||1541||8011.8||32||399.4||0.72 (0.46–1.13)||0.16|
|Colleges or graduate school||935||4836.1||19||392.9||0.72 (0.42–1.23)||0.23|
|Townships||2069||11 224.4||58||516.7||0.95 (0.62–1.45)||0.80|
In the 2895 women in the unscreened group, those categorised in other age groups (≥35 years) compared with women <35 years old had an increasing risk for developing gynaecological cancers (HR 2.42, 95% CI 1.19–4.93, P = 0.015 for women aged 35–39 years; HR 1.91, 95% CI 0.93–3.93, P = 0.078 for women aged 40–44 years; HR 2.30, 95% CI 1.13–4.72, P = 0.022 for women aged 45–49 years; HR 1.38, 95% CI 0.61–3.12, P = 0.44 for women aged 50–54 years, HR 2.54, 95% CI 1.13–5.71, P = 0.024 for women aged 55–59 years; and HR 1.51, 95% CI 0.68–3.37, P = 0.31 for women aged 60–64 years) (Table 4). Moreover, lower education level was also linked to an increasing risk of developing gynaecological cancers (reference, elementary school, senior high school HR 0.75, 95% CI 0.50–1.13, P = 0.164; college or graduate school HR 0.26, 95% CI 0.11–0.60, P = 0.0016) (Table 4).
Table 4. Risk factors associated with gynaecological cancers in unscreened women of cytological AGC (n = 2895) by Cox regression analysis
|1995–1997||1237||12 099.4||87||719.0||1|| |
|Elementary School||1414||10 829.1||103||951.1||1|| |
|Junior high School||360||2822.0||24||850.5||0.91 (0.57–1.44)||0.68|
|Senior high School||755||5585.4||38||680.3||0.75 (0.50–1.13)||0.16|
|Colleges or graduate school||366||2799.5||6||214.3||0.26 (0.11–0.60)||0.0016|
The relative risk of various gynaecological malignancies in women with a diagnosis of cytological AGC was further evaluated and compared with that of women in the normal population. Of the 323 cases of gynaecological cancers identified, 271 had cervical cancer, 40 had uterine cancer and 12 had ovarian cancer during follow up. The histological types of the 271 women with invasive cervical cancers were 141 (52.0%) squamous cell carcinomas, 102 (37.6%) adenocarcinomas, 17 (6.3%) adenosquamous cell carcinomas and 11 (4.1%) other histological types.
The SIR of various gynaecological cancers, such as cervical cancer, uterine cancer and ovarian cancer, between women with cytological AGC and women in the normal population are shown in Table 5. The incidence of cervical cancer in the general population in Taiwan was approximately 30/100 000 person-years (Cancer registration system, Taiwan, 2007). However, the incidence of developing cervical cancer in women with cytological AGC was approximately 534/100 000 person-years in this study. Women with a diagnosis of cytological AGC had a 17.85-fold higher risk of developing invasive cervical cancer compared with the normal population (Table 5). In addition, the incidences of uterine and ovarian cancers in the general population of Taiwan were around 13 and 12/100 000 person-years (Cancer registration system, Taiwan, 2007), respectively. However, the incidences of developing uterine and ovarian cancers in the women diagnosed with cytological AGC were around 79 and 24/100 000 person-year in this study. Women with cytological AGC had 5.69-fold and 2.04-fold higher chances of developing uterine and ovarian cancers, respectively, compared with the normal population (Table 5).
Table 5. Standardised incidence ratios of gynaecological malignancies between cytological AGC and normal populations in Taiwan
The different revisions of TBS of AGC show no differences in development of gynaecological malignancies. The TBS was first introduced in 1988 for reporting cervical/vaginal cytology findings.15 Revisions were made in 1991 13 and 2001 16 to improve its sensitivity and specificity. Comparing TBS 2001 and TBS 1991 in terms of categories of atypical glandular abnormalities on Pap smear, TBS 2001 reporting system is better for detecting underlying gynaecological lesions, including pre-malignant and malignant changes after follow up.9 However, when focusing on detecting incidences of gynaecological cancers in AGC, our study revealed that there are no obvious differences between TBS 1991 and TBS 2001 (crude HR 0.88, 95% CI 0.68–1.16, P = 0.37) (Table 2).
Patient age is closely related to the risk of developing gynaecological cancers in cytological AGC. Studies have documented that older patients with cytological AGC have a greater possibility of developing preinvasive diseases or carcinomas.17 As shown in the results here, women, who are >60 years old (crude HR 1.99, 95% CI 1.14–3.47, P = 0.016) have a higher possibility of developing gynaecological cancers after follow-up compared with those <35 years old (Table 2). Moreover, even women in the screened group aged ≥60 years (crude HR 2.83, 95% CI 1.30–6.15, P = 0.0085) still have a higher possibility of developing gynaecological cancers after follow up (Table 3). Those categorised in the other age groups (≥35 years old) have an elevated trend of developing gynaecological cancers compared with women <35 years old in the unscreened group (Table 4).
Educational status was correlated with the risk of developing gynaecological malignancies in women with cytological AGC. In addition to older age, women with cytological AGC with lower educational status also had an increased risk of developing gynaecological cancers (Table 2). This trend was observed in both screened and unscreened groups, especially in the unscreened group (women of college or graduate school with crude HR 0.26, 95% CI 0.11–0.60, P = 0.0016) (Table 4). A possible explanation is that nearly half of the women (47.7%) in the unscreened group were of lower educational status (elementary school). Another possibility is that women with lower educational status may not have good compliance in obtaining regular Pap smears and in further management after detecting cytological AGC for the early detection of gynaecological malignancies.18
Women with cytological AGC still need to be carefully evaluated for their cervical squamous lesions rather than glandular lesions. Focusing on the distribution of malignant lesions in women with cytological AGC, many reports demonstrate endometrium-originating cancers as the most common findings, followed by cervix-originating carcinomas.7,19 However, most women with cytological AGC who developed gynaecological cancers had cervical carcinomas in this survey, with 271 invasive cervical carcinomas found out of 323 (83.9%) women diagnosed as having gynaecological carcinomas after long-term follow up. Fifty-two percent (141/271) of these were squamous cell carcinomas. Only 37.6% (102/271) and 6.3% (17/271) were adenocarcinomas and adenosquamous carcinomas, respectively. In addition, only 12.4% (40/323) and 3.7% (12/323) of the malignancies occurred at the uterus and ovaries, respectively. Zhao et al.20 also report that the majority of women with cytological AGC have more cervical squamous lesions than cervical glandular lesions in these patients. The probable reason may be that squamous lesions with glandular involvement may yield round cell clusters with smooth peripheral contours and nuclear pseudostratifications that mimic endocervical glandular lesions, that may be categorised as AGC.
A diagnosis of AGC on Pap smear is also an indication of significant gynaecological pathology other than that of cervical origin.4 Over half of patients with AGC-US smears have either no recognisable histological lesion or a benign lesion, such as endocervical polyp, micro-glandular hyperplasia, or endocervical tubal metaplasia.21 A diagnosis of AGC-US is associated with other malignancies in rare but significant instances, including malignancies of the upper genital tract, like endometrial19,22 and ovarian cancers.20,23 Our survey also revealed that women with cytological AGC have a significantly higher risk of developing gynaecological malignancies other than cervical carcinomas, about six-fold in endometrial carcinoma, and about two-fold in ovarian cancer (Table 5). Therefore, women with cytological AGC have pathologies located beyond the female genital tract other than cervical cancer.
Testing for HPV can differentiate between women with cytological AGC for the possibility of cervical and non-cervical diseases. Previous investigators reported that HPV testing could be considered as triage of patients with AGC because of its higher positive predictive value for significant disease than ASC-US.11,24 However, they still recommend that non-cervical or other HPV-negative glandular neoplasia be considered in all patients with AGC, particularly in older patients.11 Schnatz et al.25 reported that the rate of HPV-associated disease among women with positive HPV is significantly higher (40.0%) compared with HPV-negative women (4.0%). Women at risk for endometrial or extrauterine malignancies still need to undergo appropriate evaluation regardless of their HPV status.
All women with AGC should undergo a comprehensive initial examination. Colposcopy with endocervical sampling is recommended for women with cytological AGC regardless of HPV status. Endometrial sampling is also recommended for women >35 years or with clinical indications suggesting that they may be at risk for endometrial pathologies because AGC on Pap smear represents a significant potential risk of developing malignancies of the whole female reproductive tract.26 Women of known HPV status with cytological AGC need repeat cytological testing combined with HPV DNA testing at 6 or 12 months. If both tests are negative, then women can return to routine annual cytological testing.26 Women with a diagnosis of cytological AGC but with unknown HPV status after coloscopic examination need repeat cytological tests at 6-month intervals.26 Our study shows that the women with cytological AGC also have higher risks of developing ovarian cancer compared with normal populations (Table 5). We would recommend comprehensive evaluation for these patients at 6-month intervals, including pelvic examination, ultrasonography and measurement of serum CA125 levels to detect the ovarian malignancy based on our results.
The age, education and screening history are related to the development of gynaecological malignancies in women with cytological AGC. Our data revealed that older women, women without previous screened history, and those with lower educational status had higher incidences of developing gynaecological malignancies. We hypothesised that younger women, women with regular screening, and those with higher education found it easier to have access to medical care or advice. So the risks of them developing gynaecological cancers were lower than of older women without previous screening or with lower education.
In conclusion, our study revealed that women with cytological AGC had a higher risk of developing gynaecological cancers, including those with cervical, uterine and ovarian origins. Therefore, comprehensive evaluation for these women, including pelvic examination, ultrasonography, colposcopy, endocervical curettage, cervical biopsy and endometrial biopsy, may be considered, especially for those with risk factors (i.e. older than 60 years old, lower educational status (elementary school), previous Pap smear interval longer than 2 years, or no previous Pap smear).
Disclosure of interests
The authors declare that they have no conflicts of interest with regard to any of the material presented in this paper.
Contribution to authorship
We confirm that each author has contributed materially to the paper, and that no individuals qualified for authorship have been omitted. The order of authorship is related to the relative individual contributions.
Details of ethics approval
The institutional review boards approved the study protocol (NTUH200910043R) and the National Institutes of Health registration number is NCT01015079.
This work was supported by the grant from the Bureau of Health Promotion of the Department of Health, Taiwan.
The authors express their sincere thanks to the Taiwan Cancer Registration System and the Taiwan Cervical Cancer Screening Task Force for providing the data.