記住我
Abstract
Background: The management of patients with “Atypical Squamous Cells” (ASC) in conventional papanicolaou smears (CPS) is based on the risk of high-grade squamous intraepithelial lesion (HSIL). The efficacy of liquid-based cytology (LBC) to detect this premalignant lesion is variable, with little evidence of its performance in Colombian patients. Aims: The aim of this study is to determine the performance of LBC in the detection of premalignant lesions, in patients with ASC in CPS. Materials and Methods: Were obtained patients who attended colposcopy clinic due the result of ASC in CPS. An LBC was taken, which was interpreted by two pathologists without access to other results. The performance of LBC to detect HSIL, was determined, considering as a gold standard: histopathological study/negative-satisfactory colposcopy. Results: Were included 114 patients, with a mean age of 38.4 years (SD ± 13.3). LBC had abnormal results in 40.36% (n = 46), with a slightly higher proportion of low-grade squamous intraepithelial lesion (LSIL) than HSIL. The total of abnormal diagnoses by colposcopy and/or biopsy was 51.75% (n = 59), with a predominance of LSIL (36.84%). The sensitivity of the liquid-based cytology to detect premalignant lesions was 76.5%, specificity: 66.0%, positive predictive value: 28.3% and negative predictive value: 94.1%; The Cohen's kappa index of LBC for detecting HSIL was 0.2492 for the total population and 0.2907 for ≥30 years. Discussion: Although LBC decreases abnormal cytology and increases the detection of HSIL, which improves diagnostic accuracy; sensitivity and predictive values for detecting HSIL are not significantly different between CPS and LBC.
Keywords: Atypical squamous cells of the cervix, cytodiagnosis, cytological techniques, cytology, early detection of cancer, squamous intraepithelial lesions
How to cite this article: Introduction 
Cervical cancer is the fourth most common cancer in women globally and predominantly affects middle-aged women in underdeveloped countries.[1],[2],[3] In Latin America, it continues to have a significant impact despite the implementation several decades ago of screening programs based on conventional Pap smear More Details (CPS).[4],[5] In Colombia, the screening program for precancerous cervical lesions recommends using high-risk human papillomavirus (HR-HPV) DNA detection tests, and CPS or liquid-based cytology (LBC) for the diagnostic classification of HR-HPV-positive women.[6] A critical component of uterine cancer prevention programs is the appropriate management of patients with cytological reports of atypical squamous cells (ASC), including ASC of unknown significance (ASC-US), and atypia in squamous cells. It is not possible to rule out high-grade lesion (ASC-H). The fundamental premise to treat or follow them is based on the risk of the existence of a high-grade squamous intraepithelial lesion (HSIL), which is minimal in ASC-US and greater in ASC-H.[7]
Since the implementation of LBC, multiple studies have evaluated its diagnostic efficacy, obtaining variable results,[8],[9],[10],[11],[12],[13],[14],[15],[16] with scarce evidence of the performance of this test in Colombian patients.[17] The evaluation of the LBC performance in the proper categorization of cases with “ASC” in CPS, would provide evidence to support public health authorities in their decisions about the best algorithms for screening programs, and would help reduce the number of patients unnecessarily referred to colposcopy. The aim of this study is to evaluate the performance of LBC in the detection of HSIL in Colombian patients with ASC in CPS.
Subjects and Methods A prospective, observational study was conducted. Patients who attended for colposcopy with abnormal CPS reported as ASC, who agreed to participate in the study and signed an informed consent, were included. A cervical sample for LBC was obtained from these patients using alternately SurePath® or ThinPrep® technique, and colposcopic/histological diagnosis were considered as the gold standard [Figure 1]. Patients with any previous cervical surgical procedures, squamous intraepithelial lesion (SIL) or uterine cancer diagnosis were excluded. The study was approved by the Ethics Committees of the participating institutions.
Figure 1: Study design. Patients enrolled attended for colposcopy due to an ASC result in CPS. LBC sampling was performed with SurePath® or ThinPreP®. LBC slides were blindly evaluated by cytologists/pathologists. ASC: Atypical squamous cells; CPS: Conventional PAP smear; LBC: Liquid-based cytology; CPL: Cytopathology laboratory; CPU: Cervical pathology unit. Source: OriginalTechnique and interpretation of liquid-based cytology
The material for LBC was obtained by the gynecologist before colposcopy, following the recommendations of each manufacturer: SurePath® and ThinPrep®; a cytobrush was introduced into the cervical canal, slowly turning 1/4 to 1/2 turn in one direction without exhaustively removing the mucus to optimize the sample. Once obtained, the sample was processed according to the indications of the supplier of each LBC technology. The LBC were interpreted by two pathologists who had no knowledge of the CPS result, nor access to the results of the colposcopy or histopathological study. A cytological report and evaluation of the smear quality were made according to the Bethesda System 2014. Cases with abnormal results were re-evaluated by a second pathologist, and dissenting cases were discussed to reach a consensus.
Colposcopic and histological diagnosis (gold standard)
For the colposcopy report, the following criteria were used: satisfactory, unsatisfactory, negative or normal, positive or abnormal, with a lesion suggestive of SIL or more. Colposcopic abnormalities underwent biopsy and/or endocervical curettage (ECC) when the gynecologist and colposcopist deemed it necessary. The histological diagnosis was made using the nomenclature of cervical intraepithelial neoplasia (CIN) assimilated to Bethesda: CIN1 and/or HPV infection/low-grade squamous intraepithelial lesion (LSIL), and CIN2-CIN3/HSIL.
At the end of the cytological, colposcopic, and histological analysis phase, the gold standards to determine the performance of LBC were defined as follows:
Normal/negative for SIL or malignancy, if: (a) all screening tests were negative, (b) LBC was abnormal, but colposcopy revealed no lesion and no biopsies were taken, or, (c) LBC and colposcopy were abnormal, but biopsy revealed no lesion.Abnormal/positive when the biopsy diagnosis was ≥ CIN1. If the colposcopy did not show a lesion and ECC was done, it was considered a gold standard. The clinical endpoint for the study was a histological diagnosis of CIN2+.Colposcopy was considered as a gold standard when it was negative at the first appointment and continued negative at control (six months later).Unnecessary references to colposcopy were defined as the proportion of the total number of patients referred to colposcopy who did not have CIN2+.Statistical analysis
To determine the performance of LBC in the detection of ≥ CIN2, the parameters mentioned above were established. In addition, a dichotomous division of the LBC results was made: 1) No HSIL: negative cytology, ASC-US, Atypia in glandular cells (AGC), or LSIL, and 2) HSIL: cytology with ASC-H, HSIL, or higher. Positive (PPV) and negative (NPV) predictive values were established, a 95% confidence interval was considered.
The diagnostic concordance was determined using the Kappa index between LBC and the histological study. For this, the diagnoses were divided into two groups: negative/ASC-US/AGC/LSIL and ASC-H/HSIL. To measure the agreement, the categorization of values created by Landis and Koch was used.[18] A Fisher's test was used to determine differences between the number of true positives and true negatives in both age groups (<30 and ≥30 years) and both LBC techniques. A P value < 0.05 was considered statistically significant. The data were processed using the SPSS v16 statistical software (IBM®, Armonk, NY).
Results In a period of 14 months, 121 patients attended to colposcopy after an ASC in CPS; six were excluded due to technical difficulties, and one was excluded from the analysis because of an unsatisfactory result in the LBC (percentage of unsatisfactory samples 0.87%, n = 1). Finally, 114 patients were included with previous report of ASC-US in CPS, (there were no reports of ASC-H in CPS). The mean age was 38.4 years (SD ± 13.3) with 37 (32.5%) patients under 30 years. Most of the LBC were processed by the SurePath technique® and 22.8% (n = 26) were processed by the ThinPrep technique®[Table 1].
Table 1: Patients with ASC result in conventional cytology included in the study, results of CBL, colposcopic, and histological studies (n=114)There were 46 (40.36%) cases with abnormal results in LBCs, with a predominance of LSIL (11.4%, n = 13) over HSIL (7.0%, n = 8). In 59 cases (51.75%), a final abnormal diagnosis was obtained by colposcopy or biopsy; similarly, with a predominance of LSIL over HSIL (36.84% and 14.91%, respectively) [Table 1].
[Table 2] shows the correlation between the results of LBC and those of colposcopy or histological study (final result). LBC was negative in 59.64% (n = 68) of the cases, 35 of them also had a negative final result, while, in 42.6% (n = 29) of them the final result was LSIL, and in 4 (5.8%) of the cases was HSIL/CIN2+. Six of the cases with HSIL (75%) in the LBC had the same diagnosis in the biopsy, and 30.7% (n = 4) of the cases with LSIL in the LBC were diagnosed as HSIL/CIN2 + in the biopsy [Table 2].
Table 2: Correlation of the LBC result with the colposcopic or histological studyWhen analyzing the results of the LBC compared to those of the gold standard, it was found that the sensitivity of the LBC to detect HSIL in the whole population was 76.5% (95%CI = 53.37 - 99.58). It was lower (66.7%, 95%CI = 20.61 - 100) in the group <30 years, than in the group ≥30 years (81.8%, 95%CI = 54.48 - 100); with the SurePath technique® sensitivity was 73.3% (95%CI = 47.62 - 99.05) and with the ThinPrep technique®was 100% (95%CI = 75.00 - 100). The specificity of the LBC to detect HSIL for all ages was 66.0% (95%CI = 56.04-75.92), for the group <30 years it was 61.3% (95%CI = 42.53-80.05), and 68.2% (95%CI = 56.19-80.18) for the group ≥30 years; with the SurePath technique® specificity was 61.6% (95% CI = 49.8-73.48) and 79.2% (95%CI = 60.84-97.5) with the ThinPrep technique®[Table 3].
Regarding the predictive values of LBC to detect HSIL, the PPV was 28.3% (95% CI = 14.16-42.36) for the entire population, 25.0% (95%CI = 0.66-49.34) for <30 years, and 30.0% (95%CI = 11.94-48.06) for ≥30 years; with the SurePath® technique PPV was 28.2% (95%CI = 12.8-43.61) and with the ThinPrep® technique 28.6% (95%CI = 0.00-69). The NPV was 94.1% (95%CI = 87.79-100.0) for the entire population, 90.5% (95%CI = 75.54-100) for <30 years, and 95.7% (95%CI = 88.91-100) for ≥30 years; with the SurePath® technique NPV was 91.8% (95%CI = 83.15-100) and 100% (95%CI = 97.37-100) with the ThinPrep® technique [Table 3]. The Cohen's kappa index of LBC for detecting HSIL was 0.2492 (95%CI = 0.0939-0.4045) for the total population, 0.1672 (95% CI = -0.0984–0.4328) for <30 years age group, and 0.2907 (95% CI = 0.0995-0.482) for ≥30 years [Table 3].
A result of HSIL in the LBC showed a PPV to detect HSIL of 69.0% (95%CI = 40.3-98.17) in the entire population, it was higher (72.7%, 95% CI = 41.86-100) in ≥30 years age group, than in the population <30 years (50.0%, 95%CI = 0.00-100). The NPV was 92% (95%CI = 86.32-97.84) for all ages, and likewise, it was higher (95.5%, 95%CI = 89.67-100) in ≥30 years and lower (85.7%, 95% CI = 72.69-98.74), in <30 years [Table 4].
There were not statistically significant differences in the number of true positives (p = 0.58, Fisher test), or true negatives (p = 0.50, Fisher's test) between both age groups (< 30 and ≥30 years); nor in true positives (p = 1.0, Fisher's test), or true negatives (p = 0.14, Fisher's test) between the SurePath® and the ThinPrep® techniques. The calculation of unnecessary references to colposcopy, defined as the proportion of the total number of patients who would have been referred to colposcopy by the result of the LBC and did not have CIN2+, yielded a value of 71.73%.
Discussion LBC was implemented to overcome the deficiencies of CPS, however, despite having replaced it in cervical cancer screening in many industrialized countries,[11] it has not yet been established whether it is more effective in detecting this neoplasm and its premalignant lesions, and there is concern about the conflicting results in studies that have compared both tests.[9],[11],[12],[13],[14],[19]
Various authors reported lower sensitivity of LBC than CPS in the detection of CIN2+,[9],[19],[20],[21],[22] consistent with this research findings, in which, with the same evaluation parameters (detection capacity of CIN2+ using as a criterion of positivity ASC-US in cytology), values of sensitivity and global specificity for LBC were lower than those described by Klug,[10] and by several Asian studies, such as the study of Shanmugapriya,[23] in India with 200 patients from 23- to 70-years of age which reported sensitivity of 89.5% and specificity of 77.1%, and the study of Nishio with a sensitivity of 100%.[16] Our sensitivity values are even lower that those reported in the literature in the age group of <30 years.[20],[21],[22],[23],[24],[25] This may be due to the less experience in our country with the CBL technique, which only a few laboratories offer and have been around for a relatively short time.
However, the sensitivity found in this work is higher than the LBC sensitivity value of 25% reported by Longatto-Filho, who only found greater sensitivity of LBC compared to CPS when the cut-off point was LSIL.[24] It is also higher than the sensitivity of 26.1%, described by Phaliwong, who also reported specificity of 100%, PPV of 75% and NPV of 100%.[25] Our results are more consistent with those described in the meta-analysis led by Koliopoulos, in which, the detection capacity of NIC2+ was evaluated using an ASC-US+ result on cytology as a criterion for positivity, obtaining a cumulative sensitivity of 72.9% and a cumulative specificity of 90.3%, but without changes in the results between the groups of women <30 or ≥30 years of age.[20]
The specificity of LBC in all the groups evaluated does not exceed 68%, being higher in patients 30 years or above. This means that around 30% of women sent to colposcopy do not present SIL, in contrast to these findings, other studies found a specificity of 91.5%[26] and 75%[27] for the LBC.
When comparing between LBC and CPS, the results reported by Fregnani et al.,[28] show that LBC detected significantly more lesions (ASC-US+) than CPS, but, when dividing the diagnostic categories into two groups: negative/ASCUS/LSIL and ASC-H/AGC/HSIL, they did not find an statistically significant difference in the detection rates between CPS and LBC. Meanwhile, in the study of Hashmi et al., the detection rate of SIL using LBC was significantly higher than that of CPS.[15]
In the present study, the total number of abnormal cytology decreased with the use of LBC, and there was a higher proportion of diagnosis of both LSIL (11.4% with respect to approximately 0.5-3.4% expected in CPS), and HSIL (7.0% with respect to approximately 0.1-1.0% expected in CPS), decreasing the ASC-US which is the most common cytological abnormality in all the screened populations.[29] When correlating the results of LBC with those of the biopsy, there was a complete concordance with the histological diagnosis in the cases with HSIL in LBC SurePath®, findings that coincide with those previously reported by Longatto-Filho.[30]
With regard to diagnostic validity, multiple studies have been conducted to determine the efficacy indices of LBC, however, the results have been variable and contradictory.[13],[14] Jin-Kyoung et al.,[20] performed a meta-analysis of Korean studies and obtained a cumulative sensitivity and specificity of 92% and 79%, respectively, to detect NIC2+, using the ASC-US result as a criterion for considering a cytology as positive. Jeyakumar et al.,[3] conducted a study with 120 patients over 34 years of age, finding values of 100% for the four efficacy rates described. Qin-Jing et al.,[23] performed a meta-analysis that covered 25,830 women in 13 studies, in which, they reported a sensitivity of 81.0% and a specificity of 95.4%, PPV of 38.3%, and NPV of 99.3%, to detect NIC2+ with cut-off points of ASC-H and LSIL. The finding of values higher than those of the present study is striking, even despite stricter positivity criteria.
Regarding the NPV to detect HSIL obtained in the present study, the lower limit of the confidence interval indicates an excellent ability to distinguish patients without HSIL and those who most likely only require observation in case of presenting LSIL. On the other hand, the PPV to diagnose HSIL reports that only in a little more than a quarter of cases with abnormal LBC was there an underlying HSIL. However, when the cytological result is dichotomized into HSIL and NO-HSIL, this PPV increases, being higher in patients ≥30 years of age, but it is lower than that found by Longatto et al.,[24] of 50% in LBC when the cut-off point of the cytological diagnosis was HSIL. In contrast, the NPV of this study is similar to the 99.1% that they reported.
The concordance of LBC to detect cases of underlying HSIL/CIN2+, calculated by the kappa index, was in no case higher than discrete, with slightly higher values in the group of patients aged 30 years or above. Meanwhile, a high concordance of HR-HPV detection molecular tests to detect HSIL/CIN2 + has been reported in cases with cytological results of ASC-US.[29] However, it should be remembered that cytology, whether CPS or LBC, is a screening method whose function is to detect patients who need to be evaluated by colposcopy, and is not considered a definitive diagnostic test.
In relation to the evaluation of the two LBC techniques (ThinPrep® and SurePath®), the results calculated for the ThinPrep® method were not considered valuable or relevant due to the low number of cases with HSIL in the histological diagnosis. The results obtained with SurePath® were very similar to those calculated in a general way in the total population, being the sensitivity lower than that demonstrated by Fang-Hui Zhao.[31]
One of the limitations of this study is that the population evaluated had a previous result of ASC-US in CPS, which increases the prevalence of SIL, thus influencing the calculated predictive values. Due to this change in prevalence, it is not possible to adequately determine whether LBC results in an increase in the percentage of ASC-US and LSIL with respect to conventional smears. In addition, colposcopy, which is an operator-dependent study, had to be considered as a gold standard when it was satisfactory/negative because not all patients had a histological study.
One of the advantages of using LBC over its conventional counterpart, in which multiple studies coincide,[14],[32],[33] including this one, is the significantly lower percentage of unsatisfactory samples, being, for example, 1 of 1.61 in the study by Pankaj et al.,[31] The LBC leaves low cellularity as the only cause of sample unsatisfactory[32] as in the only case reported in this work. Other advantages are: lower percentage of unsatisfactory samples, availability of residual samples for HPV detection,[33],[34],[35],[36] possibility of using computer-assisted screening and sample preservation to search for biomarker expression.[37],[38],[39],[40] However, the divergent results when compared with CPS and the reports of very similar detection rates of epithelial abnormalities between both techniques, added to the cost of its implementation, have led CPS to be a better option in underdeveloped countries for screening than CBL.[26]
To our knowledge, this is the first prospective study conducted in Colombian patients with abnormal ASC cervical cytology to assess the diagnostic performance of LBC and CPS in detecting cervical premalignant lesions compared to histopathology. It can be concluded that although LBC decreases the total number of abnormal cytology and increases the detection of HSIL, improving diagnostic precision and decreasing the number of ASC-US, its concordance with the gold standard is discreet, being higher to detect HSIL, especially in patients aged 30 years or above.
Acknowledgments
To the Universidad de Cartagena, Cartagena, Colombia, for financing this study, and to the Clinica Maternidad Rafael Calvo, Cartagena, Colombia, for allowing us to carry it out.
Financial support and sponsorship
This study was part of the project “Evaluation of the diagnostic performance of p16/Ki67 in the detection of high-grade squamous intraepithelial lesions of the cervix in patients with cytology with atypia in squamous cells (ASC)” financed by Universidad de Cartagena, Cartagena, Colombia.
Conflicts of interest
There are no conflicts of interest.
References 
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/joc.joc_22_22
留言 (0)