Reliability and relationship of colposcopical , cytological and hystopathological findings in the diagnostic process

Background/Aim. The question about the accuracy of cytology and colposcopy is more and more asked due to false positive and negative findings on the basis of which the decision on biopsy is made. The aim of this study was to examine reliability of biopsies based only on abnormal colposcopical findings, before receiving the results of Papanicolaou (PA) smear, by comparing findings of colposcopical, cytological and histopathological (HP) examinations as well as determining validity of these diagnostic methods. Methods. The study involved all patients who had their regular colposcopical and cytological examinations in the outpatient department during a two-year period (2009–2010) in the Clinic for Gynecology and Obstetrics, Clinical Center of Serbia, Belgrade. The material for HP examination was obtained by colposcopically directed biopsy, due to abnormal colposcopical findings and without waiting for PA smear results. The data obtained by these methods were statistically analyzed and compared. Furthermore, validity of colposcopical and cytological examinations was assessed. Results. Out of 127 patients highly significantly more patients had more malignant cervical changes on colposcopical exam compared to HP (p = 0.000), and cytological exam (p = 0.000). Highly significantly more patients had more malignant cervical changes on PA smear than HP exam (p = 0.000), unless when findings were assessed in the widest sense of benign and malignant changes when there were no significant differences in these findings (p = 0,450). Sensitivity of colposcopy as a diagnostic method was 87.5%, specificity 24.14%, positive predictive value (+PV) was 34.65% and negative predictive value (–PV) 80.77%. Sensitivity of PA smear as a diagnostic method was 62.5%, specificity 87.36%, +PV was 69.44%, and –PV 83.52%. Conclusion. Regarding the results of our study it is best to make a decision on treatment according to findings of all the three methods. Cytological analysis is more reliable than colposcopical examination. Therefore, it is advisable that following abnormal colposcopical findings, PA smear should always be taken and only after receiving the results further diagnostics can be planned (biopsy and HP). A final decision on the therapy has to be made based on HP findings which are the only method that can give the ultimate reliable diagnosis of cervical changes.

The aim of this study was to examine reliability of biopsies based only on abnormal colposcopical findings, before receiving the results of Papanicolaou (PA) smear, by comparing findings of colposcopical, cytological and HP examinations, as well as to determine validity of these diagnostic methods.

Methods
The study involved all patients who had their regular colposcopical and cytological examinations in the outpatient department during a two-year period (2009-2010) in the Clinic for Gynecology and Obstetrics, Clinical Center of Serbia, Belgrade.The material for the HP examination was obtained by colposcopically directed biopsy, due to abnormal colposcopical findings and without waiting for PA smear results.Data obtained by these methods were statistically analyzed and compared.Descriptive [mean; median; standard deviation -(SD) coefficient of variation (CV%)] and analytical (Kolmogorov-Smirnov test -KSZ; Friedman test -FR 2 ; Wilcoxon test -Z) statistical methods were used.Furthermore, sensitivity, specificity, positive and negative predictive values (+PV and -PV respectively) of colposcopical and cytological examinations were assessed.For these calculations standard formulas were used: sensitivity = [(true positive / true positive + false negative) 100], specificity = [(true negative / true negative + false positive) x 100], +PV = (true positive / true positive + false positive) x 100] and -PV = [(true negative / true negative + false negative) x 100)].

Results
The study involved 127 consecutive patients.Their age span was from 17 to 79 years (mean = 35.11;median = 32.00;SD = 11.131).These values were heterogeneous (CV = 31.703%)and they were not normally distributed (KSZ = 1.602; p = 0.012).The mayority of patients (n = 32) were from 30 to 34 years old and somewhat less patients (n = 26) were 25 to 29 years old.Still, 89.9% of the patients were younger than 50 years and therefore the distribution is asymmetrical to the left, ie towards the younger age.

Findings
On colposcopical examination the most frequent was the finding of AW epithelium and/or mosaic ( 2 = 85.268; df = 5; p = 0.000).By HP analysis the most frequently diagnosed were ectopia and/or cervicitis ( 2 = 44.780;df = 7; p = 0.000).Cytological evaluation showed that PA II and III were highly significantly more frequent while there were less PA IV and V findings ( 2 = 155.866;df = 3; p = 0.000).
Categories of the diagnoses obtained by these three methods were highly statistically different (FR 2 = 88.487;df = 2; p = 0.000).The diagnoses obtained by colposcopy and HP were highly statistically different (Z = 3.655; p = 0.000).There were significantly more (n = 73) patients with more malignant changes on the colposcopical than on the HP examination.The diagnoses obtained by colposcopy and cytology were also highly statistically different (Z = 8.212; p = 0.000).There were significantly more (n = 53) patients with more malignant changes on colposcopical examination.The diagnoses obtained by PA and HP were highly statistically different (Z = 4.706; p = 0.000).There were significantly more (n = 108) patients with more malignant changes on PA than on HP examination.
For sensitivity, specificity, positive and negative predictive values evaluation of colposcopy and cytology, normal and other colposcopical findings, PA II as well as histopathologically asserted benignant lesions were regarded as benignant cervical changes.On the other hand, abnormal and suspicious colposcopical findings, PA III, IV and V, LSIL, HSIL and invasive cervical carcinoma formed a group of malignant cervical changes.Colposcopical and cytological results were compared to HP findings, considering that histopathology is the only method able to give the precise diagnosis of the condition.
The diagnoses assessed only regarding the benign and malignant lesions were also highly statistically different regarding the diagnostic methods used in our study (FR 2 = 93.671;df = 2; p = 0.000).There were significantly more (n = 66) patients with a more malignant changes on colposcopical than on HP examination (Z = 7.239; p = 0.000).There were also significantly more (n = 68) patients with more malignant changes on colposcopical than on cytological examination (Z = 7.714; p = 0.000).Regarding this wider categorisation of findings diagnoses made by PA and HP do not differ significantly (Z = 0.756; p = 0.450).
Colposcopical examination registered 21 true negative ie benignant findings and 35 true positive ie malignant findings.There were 66 false positive and 5 false negative cases.According to these results sensitivity of colposcopy as a di-  agnostic method was 87.5%, specificity 24.14%, +PV was 34.65% and -PV was 80.77%.Cytological examination registered 76 true negative ie benign findings, and 25 true positive ie malignant findings.There were 11 false positive and 15 false negative cases.According to these results sensitivity of PA smear as a diagnostic method was 62.5%, specificity 87.36%, +PV was 69.44% and -PV was 83.52%.
When all evaluated diagnostic methods were assessed together, 20 patients had benignant findings, while 22 had malignant findings on each of the three examinations.However, in 85 cases findings were of different categories on different examinations.

Discussion
Squamous cervical carcinoma develops from noninvasive forms and therefore the detection of abnormal pathological changes of the cervical epithelium and their removal is crucial for preventing a possibly fatal invasive carcinoma.Of major significance is the fact that early detected CIN is completely curable condition.The aim of successful treatment is not only a complete removal of the lesion but also sustaining the physiological cervical function as well as the reproductive female functioning 1,2 .
It is considered that CIN occurs mostly in early thirties with the prevalence of 2.6%, although lately it is more common in women in their twenties or even younger.After the age of 50 years the prevalence is decreasing to 0.9% 1,3 .Patients in the examined population were also mostly in their thirties, while only 10% were older than 50 years.Unfortunately, in our patients HSIL was more frequent than LSIL.
Classification of the lesions used in our study was according to BETHESDA system and in concordance with categorisations used by other authors in the studies from the available literature 4 .
Expholiative cytology is of great importance for diagnosing CIN as pathological findings can direct toward looking for and discovering cervical lesions.Even though cytology has undoubted benefits in cervical carcinoma prevention, it is accepted nowadays that some significant lesions can be omitted or underdiagnosed by PA testing due to the limitations of cytology concerning its sensitivity, specificity and predictive values 3,5,6 .There are different explanations in the literature for this less satisfactory PA diagnostic assessment in certain cases.One of them is the presence of only few abnormal, ie malignant cells in smear which can be easily omitted during examination due to their low frequency.This can usually happen in case of small cell lesions high in the endocervical canal 7 .Moreover, inflammation and bleeding can thwart the assessment 3 .Cohesive tissue fragments of HSIL can sometimes be incorrectly diagnosed as immature metaplastic cells 3 .Furthermore, technical problems occurring while smear is dried are reported 3 .Therefore, it can be recommended that correct smear taking and its optimal handling and cultivation must always be assured.However, patients should also be sent on colposcopical examination.
Colposcopical evaluation and directed biopsy are an essential diagnostic step in treatment of abnormal cytological findings 4 .Colposcopy is a high-quality and sensitive test for detection of CIN and can be considered as integrative and standard basic parts of screening for CIN.Colposcopical evaluation and directed biopsy remain crucial diagnostic procedures for squamous intraepithelial lesions and for identification of those cases that need treatment 8 .However, it must be taken into consideration that performing colposcopy and its occuracy depend in large part on training, experience and skills of the colposcopist.It is usually considered that the limitations of most studies, just like the one we conducted, come from the fact that different colposcopists perform examinations.Nevertheless, an experienced colposcopist's findings can be regarded as precise 4 .Considering that almost all accessible data are based on that principle, we also accepted this standpoint and regarded the findings of our colposcopists as valid.Detected abnormal colposcopical findings, followed by a positive cytological testing, demand for a guided biopsy, under the control of the colposcope in order to avoid obtaining incorrect samples and therefore false negative results.
The main objective of cervical screening is to identify women with HSIL.These women require prompt treatment because of possible disease progression, while those with CIN I can only be monitored as that condition can spontaneously remise.In the literature almost 20% of LSIL was overdiagnosed and between 20% and 30% of HSIL was underdiagnosed [9][10][11] .In our study such cases were also recorded, which presents a hazard.Synchronized application of colposcopy and cytology gives better chances for detection of those lesions that need treatment.Colposcopy, as the only diagnostic method, is not precise enough, but is useful for evaluation of the lesion stage 12,13 .In order to choose on adequate treatment biopsy is necessary 12,13 .Treatment must be based on HP findings due to a possible misinterpretation of both colposcopical and cytological examinations 11 .There are even cases in which mistakes in HP evaluation were made and when CIN was not appropriately dyagnosed 11 .
Using the similar methodology as in our studies, the analysis of validity and diagnostic precision of cytological findings of LSIL and HSIL, researchers concluded that PA and HP findings show statistically significant correlation 13,14 .In the examined population these results were achieved only when a wider categorisation is assessed.Otherwise on both PA and colposcopical examination more malignant changes in comparison with final HP diagnose were obtained.Therefore it can be said that both methods overestimate the stage of the lesion.
The available literature data state that sensitivity is higher than specificity of colposcopy 4,12 , which was also demonstrated in our study.While conducting screening low +PV of colposcopy may result in unnecessary treatment 12 .In our study +PV is also somewhat lower.Colposcopy has higher sensitivity and cytology higher specificity, so their findings are not doubling but supplementing themselves in providing better reliability.Besides, it was also registered that colposcopical findings were more malignant in compari-son with final HP diagnose and therefore other diagnostic methods should be undertaken.
High sensitivity of colposcopy in the investigated population implies that this method will in most cases recognise malignant changes.On the other hand, cytological analysis has high specificity and -PV, so this method will in most cases disregard benignant changes.It can be observed that our study showed higher values for all the examined parameters (sensitivity, specificity, + and -PV) and therefore is a more reliable diagnostic method.Moreover, as compared with diagnostic categories obtained from different examined methods it can be concluded that PA discovered all malignant lesions, but also had the most false positive findings.Colposcopy was the least reliable and as well had numerous false positive findings.If only benignant and malignant conditions are evaluated, in the widest sense, colposcopical findings were once more the least valid and statistically significantly different from HP diagnoses.Contrary, according to the fact that benignant and malignant PA ad HP findings do not differ significantly, it can be concluded that cytological analysis is a reliable method for differentiation between benignant and malignant findings.For more detailed and reliable diagnostics biopsy of the cervical change with HP analysis of the obtained tissue is still recommended 15 .
The roles of cytology, colposcopy and guided biopsy in prevention of cervical cancer are still developing.Regarding the some literature data it can be advised to base the decision on adequate treatment only on the abnormal colposcopical findings 14,16 .But according to the results of our study it can be concluded that the decision on treatment is best made when based on the findings of all the three methods.

Conclusion
Although both colposcopy and cytology are methods with high sensitivity, specificity, + and -PV, individually they are not absolutely valid.Based on the results of our study colposcopy directed biopsy cannot be recommended due to numerous false results and therefore unnecessary procedures.Cytological analysis is a more reliable method and as such is a necessary diagnostic tool.Therefore, it can be advised that following abnormal colposcopical findings, PA smear should always be taken and only after receiving the cytological results further diagnostics can be planned (biopsy and HP).A final decision on the therapy must be made in accordance with HP findings which are the only method that can lead to the ultimate reliable diagnosis of cervical changes.