Liquid Based Cytology for the Detection of Cervical Intraepithelial Lesions among Women in Jimma Town, Ethiopia

Background: Cervical cancer is the second leading type of female cancer in Ethiopia. Screening is primarily conducted using visual inspection with 5% acetic acid (VIA). Liquid-based cytology (LBC) has not yet been used in Ethiopia. Method: Women aged 21-65 years were tested using LBC and VIA for the detection of cervical dysplasia. Logistic regression analysis was conducted to identify associated factors. Cohen’s K test was conducted to test agreement between LBC and VIA. Results: Forty-two percent (n=188) of 448 participants were 31 to 40 years of age and only two participants were above 60. Of the 448 participants, 419 (93.5%) were tested with LBC, 294 (65.6%) VIA and 272 (60.7%) with both LBC and VIA. Among women screened using LBC, 305(72.8%) were negative for intraepithelial lesion or malignancy (NILM), 97 (23.2%) had low grade squamous intraepithelial lesion (LSIL) and 17 (4.1%) high grade squamous intraepithelial lesion (HSIL). Presence of cervical lesions was generally lower in younger and older aged women. Majority, 39(40%) with LSIL and 10 (59%) of women with HSIL were 41-50 years old. Women aged 51-60 years were more likely to have abnormal intraepithelial lesions compared to women aged 21-30 (AOR= 20.9,95%CI=[7.2-60.9], p=0.00). Out of 47 (10.8%) HIV-patients,14(32.56%) had intraepithelial lesion of which 10(23.3%) and 4(9.3%) had LSIL and HSIL, respectively. Among women screened with VIA, 18 (6.1%) were positive. Among the 272 (60.7%) women screened using both LBC and VIA, 6 (2.2%) were positive on both LBC and VIA. The level of agreement between the two tests was weak and statistically signicant (kappa value=0.155, p=0.006). Conclusion:

Conclusion: LBC detected high cervical squamous intra-epithelial lesions in our setting. VIA was a less reliable predictor of cervical squamous intra-epithelial lesions than LBC. Evaluating diagnostic accuracy of both LBC and VIA against histological endpoint should be completed before adopting either screening modalities.

Background
Human papilloma virus (HPV) is the most common sexually transmitted infection in the world [1]. Human papilloma virus (HPV) causes a variety of malignancies, with cervical cancer being the most important and prevalent [2]. Cervical cancer is a leading public health challenge globally. According to worldwide estimates of cancer, 569, 847 women were diagnosed for cervical cancer as new cases and 311,365 women died in 2018 [3]. Majority, (85%) of deaths occurred in low-and middle-income countries [4]. In Africa, 119, 284 new cases of cervical cancer were diagnosed and 81, 687 women died in 2018.The highest rate cervical cancer was found in eastern and western region [3].
In Ethiopia, 5.8% of national mortality is attributable to cancer. Incidence is increasing because of the aging population. Cervical cancer is the second leading cause of female cancer in women aged 15 to 44 years in Ethiopia [5]. According to Global Cancer Observatory estimate 6,294 new cases were diagnosed and 4,884 women died from the disease in 2018 [6]. According to the Ethiopian Ministry of Health, approximately 80% of reported cases of cancer are diagnosed at advanced stages when very little can be done to treat the disease [5].
In 2016, Ethiopia introduced a national cancer control plan that included a strategy of using visual inspection with acetic acid (VIA) and corresponding treatment of positives. Visual inspection with acetic acid (VIA) continues to be the only cervical cancer screening modality in the country [5]. There is no organized cytology-based cervical lesion screening program in Ethiopia.
As per the researchers' knowledge, no cervical cancer screening has ever been conducted in Ethiopia that combines liquid-based cytology (LBC) and visual inspection with acetic acid (VIA). This study documents the burden of cervical lesions and predictors of abnormal cervical cytology as well as comparing LBC and VIA screening modalities.

Study Design and Setting
A cross sectional study was conducted in Jimma Town from February 2017 to May 2018. Jimma is located 350 kilometers southwest of Addis Ababa. A total of 448 non-pregnant women ages 21-65 who visited Jimma University Hospital, Marie Stopes international and Family Guidance Association of Ethiopia (FGAE) Clinics for VIA screening as part of the national cervical cancer screening program were enrolled consecutively. Women with complete hysterectomy, gross tumor on the cervix, prior surgical procedures involving the cervix, menstruating, and those with no history of sexual activity were excluded.
Written informed consent was obtained and the procedure of the test was explained to women.

Demographic and Risk Factors Data
Demographic information and risk factors for cervical cancer were collected using questionnaires prepared in Afan Oromo and Amharic languages. The collected data included occupation, educational status, age, parity, marital status, history of contraceptive use, age at rst sexual intercourse, smoking, number of lifetime sexual partners, family history of cervical cancer, sexually transmitted infections (STIs) and alcohol consumption.

Liquid Based Cytology (LBC)
An automated liquid-based cytology, SurePath™ liquid-based Pap test (BD, USA), was employed for cytological sample preparation. After removing obscuring mucus from the cervix with a cotton swab, endocervical and ectocervical cells were collected with cytobrush. This cytobrush immediately rinsed in a vial containing SurePath Preservative Fluid. Samples were transported at room temperature for analysis at International Clinical Laboratories (ICL) in Addis Ababa using BD PrepMate TM and PrepStain™ Slide Processor. Vials containing samples were labeled and placed into the BD PrepMate™ Slide Processor in which a liquid-based ltration process removed mucus and debris, preserving cell morphology, and making a smear of even distribution. All slides were stained with the BD SurePath Kit Cytology Stain and examined by two pathologists [7] who enrolled in the College of American Pathologists (CAP) pro ciency program and received stained LBC slides every three months and participating in the external quality assurance scheme. LBC test results were reported based on the Bethesda gynecologic cytology guideline [8].
Visual Inspection with Acetic Acid (VIA) Women visited health facilities in Jimma Town involved in national cervical cancer screening program were enrolled for VIA. Women with invisible transformation zone were excluded from the study. After obtaining informed consent, sterile plastic spatula was inserted into the vagina to visualize the cervix. Then after, 5% acetic acid was applied to the cervix for one minute. Positive test was interpreted as a ''sharp, distinct, well-de ned, dense (opaque, dull or oyster white) aceto-white area with or without raised margins" according to the standard guideline [9,10]. VIA examination was done by experienced clinical nurses who participating in the national cervical cancer screening program using VIA.

Data Analysis
Data was checked for completeness, coded and entered into Epi data version 3.1 and exported to STATA 14 for analysis. Descriptive statistics, frequency and proportion were used to describe demographic variables. Sub-group analysis was conducted for HIV patients. Logistic regression analysis was used to identify risk factors for abnormal cervical cytology on LBC test. A Kappa test was used to assess the percent agreement between LBC and VIA. p-values < 0.05 at 95% con dence was considered statistically signi cant at 95% con dence.

Predictors of Abnormal Cytology by LBC
Bivariate logistic regression analysis revealed that parity, age and condom use during sexual intercourse were candidates for multivariate regression at p=<0.25. Multivariate regression revealed that age was an independent predictor of LSIL or HSIL. Odds of being positive for cervical squamous-extraepithelial lesions were higher in women older than 31 years of age.

Discussion
In Ethiopia, 29 million women over 14 years of age are at risk of developing cervical cancer [5]. In 2018, 6,294 women were diagnosed as new cervical cancer cases and 4,884 women died from the disease [6].
Even though cervical cancer burden is high in Ethiopia, the national cancer screening program is based solely on VIA, which has high variability due to examiners' judgment [11]. Our study shares the results of the rst cervical cancer screening in Ethiopia using LBC. In our study, prevalence of abnormal squamous intraepithelial lesion was 114(27%), which is higher than the 17% shown in a study in China [12]. Prevalence of LSIL and HSIL were 23.2% and 4.1 %, respectively, much higher than the 1.9% and 0.6% prevalence, respectively, observed in Sao Paulo. [13]. A study in India reported lower rate of LSIL (7.5%), but higher HSIL (10.5%) [14]. Signi cantly, lower prevalence of LSIL and HSIL (2%) and (2.4%) were shown the Netherlands and Germany, respectively, [15,16]. Low prevalence rate of cervical squamous intra-epithelial lesions in developed nations may be due to the availability of the HPV vaccines [17] and the presence of organized cervical cancer screening [18], which is new to Ethiopia.
In our study, higher proportion of women aged 41-50 tested positive on LBC screening test whereas we observed lower prevalence of cervical lesions in younger and older women. Visibility of SCJ is the prerequisite for VIA examination and women with invisible SCJ are exempt for VIA examination [19]. In our study, women with invisible SCJ, underwent LBC testing and 60 (50%) had either HSIL or LSIL by LBC.
Logistic regression showed women aged 51-60 years had higher odds of having cervical squamous intraepithelial lesion compared to younger ages. LBC screening was better at detecting HSIL and cervical lesions in older ages, which is not true for VIA screening [20].
HIV infection is a risk factor for persistent HPV-infection, a necessary condition for the development of squamous interepithelial lesions and HIV infected women are disproportionately affected by cervical lesions [21]. In our study, 14(32.56%) HIV-patients had cervical squamous intraepithelial lesions, higher than prevalence in the total study population (27%). While the rate of LSIL (23.3%) among HIV-positive patients was similar to the full study cohort and prevalence of HSIL (9.3%) was nearly double. A study in South Africa recorded higher prevalence of LSIL (32.5%) and HSIL (23.3%) [22] than our study whereas a study in Nigeria showed LSIL and HSIL rates to be 14.3% and 4.3%, respectively, among HIV-positive women [23].
VIA detected 18(6.1%) cases of cervical lesions, which is similar to the 5% reported in West Shewa, Ethiopia [24], but much lower than the 12.9% reported in another study in Jimma Town, [25] as well as studies in Rwanda and China, where 14.7% [26] and 11.4% [12] of women had cervical lesions, respectively. Among women who were tested using both LBC and VIA in our study, a high proportion (83.3%) that tested positive with LBC, tested negative on VIA, meaning VIA screening missed a considerable proportion of women with abnormal cervical lesions. This nding is similar to a study in China that showed VIA missed the majority of CIN2+ in older women and was less sensitive than LBC [12]. As our study showed, there was no agreement between LBC and VIA screening tests and variability was statistically signi cant (kappa =0.155, P=0.006).
Organized cytology-based screening is the most e cient screening method for the detection of cervical lesions and has resulted in signi cant reduction in cervical cancer in developed countries [27]. Financial constraints and technical challenges hinder implementing cytology-based screening in low-and middleincome countries like Ethiopia.

Conclusions
Given that VIA screening missed most cervical lesions detected by LBC in our study, and that a high number of cervical epithelial lesions were detected by LBC, a larger study should be undertaken to determine the diagnostic accuracy of both LBC and VIA against a histological endpoint before adopting either or both as screening modalities.

Availability of Data and Materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.