Correlation between cytological and histopathological diagnosis of non-small cell lung cancer and accuracy of cytology in diagnosis of lung cancer

Background/Aim. Lung cancer is one of the most common cancer types worldwide.
 More than 70% of patients are diagnosed in advanced stages of disease, with
 limited therapeutic options based on cytological and histopathological
 material. The value of cytology in diagnosing and subtyping of non-small cell
 lung cancer is very important for modern personalized therapies. The aim of
 this study is to find out the concordance between cytological and
 histopathological diagnosis of non-small cell lung cancer, and the accuracy,
 sensitivity, specificity, positive and negative predictive value of cytology
 in diagnosing lung cancer. Methods. Two-year retrospective study included 169
 patients with cytological diagnosis of non-small cell lung cancer,
 histopathological small biopsy and surgical specimens for diagnosis
 confirmation, that were copmpared. Histopathological diagnosis on surgical
 specimens was golden standard for evaluation concordance to cytological
 diagnosis of non-small cell lung cancer and evaluation accuracy, specificity,
 sensitivity, positive and negative prognostic value of cytology as diagnostic
 method for detecting lung cancer. Results. This study included 76.3 %
 (129/169) male and 23.7% (40/169) female, age between 39 and 83, average
 62.53?7.6. There was no statistically significant difference between ages of
 different genders (p=0.207). The most frequent among cytological diagnosis
 was non-small cell lung cancer in 58.58% (99/169) patients. Concordance
 between cytological and histopathological diagnoses of surgical specimens was
 61.48%. There was no statistically significant difference between cytological
 diagnoses and histopathological diagnoses of small biopsies specimens
 (p=0.856). Sensitivity, specificity, positive and negative prognostic value
 and accuracy of cytology as diagnostic method of lung cancer were 94.98%,
 98.60%, 95.72%, 98.35% and 97.71%, respectively. Conclusion. Cytological
 diagnosis of non-small cell lung cancer is accurate, with high sensitivity,
 specificity, and benefits for patients. Most patients are diagnosed in
 advanced stage of cancer when there is no surgical therapy option and the
 only available diagnostic material is small biopsy sampled during
 bronchoscopy.

Based on WHO recommendations for diagnosing by examining cytological samples and small biopsies, lung adenocarcinoma is epithelial malignant tumor morphologically with glandular differentiation, vacuolated cytoplasm, mucin production, enlarged nuclei or specific immunohistochemical marker expression -napsin-A or thyroid transcription factor-1 (TTF-1) positivity after immunostaining. Squamous cell lung cancer is epithelial malignant tumor morphologically with keratinization, dense cytoplasm, intracellular bridges or specific immunohistochemical marker expression -p40 or p63 positivity after immunostaining. NSCLC, not otherwise specified (NOS) include cancers without either morphological characteristics specific for adenocarcinoma or squamous cell carcinoma or immunostaing positivity 4,6 .
Accuracy, sensitivity, specificity, positive and negative prognostic value of cytology in diagnosis and staging of NSCLC have been monitored since 1980. Plenty of cytological methods of sampling are in use, including exfoliative methods (sputum, bronchoalveolar lavage -BAL, bronchial aspiration and brush cytology) and aspiration methods (transbronchial needle aspiration -TBNA) 7 . Nowadays specificity of cytology is up to 100% and sensitivity between 60% and 90%, depending on sampling method.
Ultrasound guided TBNA has increased sensitivity of cytology and decreased number of false negative cytological diagnoses. The value of cytology in diagnosing of NSCLC and subtyping it to adenocarcinoma and squamous cell carcinoma is very important for modern personalized molecular therapies and immunotherapy while rapid diagnostic on small samples is preferred for patient benefit, less complications while sampling and appropriate therapy time 8 .
The aim of this study is to find out the concordance between cytological diagnosis of NSCLC and histopathological diagnosis, and the accuracy, sensitivity, specificity, positive and negative predictive value of cytology in diagnosing lung cancer.

Methods
Two-year retrospective study was done at the Department of Cytology of the  TBNA using needle 19G (for tumors not visualized in bronchial lumen),  brush cytology, bronchial content aspiration, sputum or BAL (for centrally located tumors in bronchial lumen)  and "tru cut" needle biopsy (for tumors localized on the periphery of the lung).
Among patients, 78.60% had material sampled using only one method, 20.20% of patients had material sampled using two and 1.20% using three methods. Cytological methods of sampling were presented on Graphic 1.
Cytological smears were made on microscopic slides from the material of each patient; also cytospins were made in cases of large amount of material. Microscopic slides were air-dyed and stained by May-Gruenwald-Giemsa (MGG) method. After microscopic evaluation, cytological diagnosis that were made were malignant -NSCLC, adenocarcinoma or squamous cell carcinoma.
Material for small biopsy histopathological evaluation was sampled simultaneously with cytological material during bronchoscopy using bronchial biopsy or TBNA with 19 G needle. It was fixed in buffered 4% formalin for 12 hours, dehydrated by increased alcohol concentration, cleared by chloroform, embedded in paraffin and cut by microtome (Leica) to slices measured 4µm. After that, it was deparaffinized and stained by hematoxylin and Concordance between cytological and histopathological diagnoses of surgical specimens was 61.48%. Unlike it, the concordance between histopathologic diagnoses of small biopsies specimen and surgical specimens was 95.2%.
There was neither statistically significant difference between cytological diagnosis and histopathological diagnoses of small biopsies specimens (p=0.856), nor between cytological diagnosis and histopathological diagnoses of surgical specimens (p=0.196). In addition, there was no statistically significant difference between histopathological diagnoses of small biopsies and surgical specimens (p=0.230). Discordance in cytological, small biopsies and histopathological diagnoses on surgical specimens were presented in Table 1.
There was statistically significant difference between diagnoses on cytological specimens, depending of method of sampling (p=0.001). There was statistically significant less discordance in cytological diagnoses on material sampled by TBNA with histopathological diagnoses. Discordance in cytological and histopathological diagnoses on surgical specimens, depending on cytological sampling method were presented in Table 1.

Discussion
Lung cancer is the most common cause of morbidity and mortality worldwide 10 . The highest incidence of lung cancer is in age between 65 i 74, average 70 11 . Patients in this study were slightly younger with average age 60 in female and 63 in male. The youngest was 39-year-old patient, similar to data in previous investigations 12 . Although the gender distribution of lung cancer patients is equal in developed countries, there were three times more male than female in this study, as it was in other developing countries 10,11,13,14 .
An adequate sample for cytological and histopathological analysis has been obtained during bronchoscopy. The sample is fundamental for evaluation, confirmation and in some cases staging of tumor visualized during bronchoscopy 15 . Accuracy of diagnostic methods depends on location of tumor, its dimensions, type and technical aspects including the level of bronchoscopist's and pathologist's experience. Cytological diagnosis during bronchoscopy is preferable in centrally localized tumors, unlike tumors localized at the periphery of the lung when transbronchial biopsy, TBNA or transthoracic biopsy should be done 16 .
As it was in other studies worldwide, 26.63% of the our patients had diagnosed squamous cell carcinoma based on cytomorphological criteria. 46.75% of the patients have been diagnosed squamous cell carcinoma on small biopsy material and all the diagnosis were confirmed on surgical specimens. The reason of fewer patients with cytologically diagnosed squamous cell carcinoma was poor differentiation of squamous cell carcinoma in approximately half of the patients. Those patients were diagnosed NSCLC cytologically and needed further immunostaing for more precise histopathological diagnosis 12,17,18 .
In spite of 45.56% lung adenocarcinoma histopathlogically diagnosed on small biopsies material and confirmed on surgical specimen, cytologically were diagnosed only 13.02% lung adenocarcinoma. Similar results, with a small number of cytologically diagnosed lung adenocarcinoma in patients, were obtained in other studies 18,19 . The majority of histopathologically diagnosed adenocarcinoma were cytologically diagnosed as NSCLC, because of the lack of cytomorphologically specific features significant for adenocarcinoma diagnosis according to the newest WHO 2015 criteria 4 .
Besides lack of cytomoprhological characteristics for cytological adenocarcinoma diagnosis, small number of viable cells, large amount of necrosis, tumor heterogeneity and artifacts could also cause misdiagnosis 19,20 . The precise diagnosis of adenocarcinoma on cytological specimen is very important because of novel diagnostic methods. More cell blocks with paraffin embedded cytological material and possibility further immunostaining are made from a part or a rest of cytological material. Also, necessity of cell's viability for novel diagnostic methods is another advantage for cytological diagnosis 2,8,22 . In our study, concordance of cytological and histopathological diagnosis was 61.48%, as it was reported in literature 12,19,21,22 . There was less statistically significant difference between cytological and histopathological diagnoses after sampling by TBNA method in both our and other researches. TBNA sampled material for cytological diagnosis is more abundant, better preserved, with more viable cells and less necrotic parts 12 .
Avoiding diagnostic mistakes is very important because false positive diagnosis could lead to disease and even death. False negative diagnosis could, on the other hand, postpone earlier diagnosis and therapy 23

Conclusion
Cytological diagnosis of lung cancer is accurate, with high sensitivity and specificity. Even there is some discordance between cytological and histopathologic diagnosis of non-small cell lung cancer, it was not statistically significant. The value of cytology is because less material and less time is needed for diagnosis, which is very important in advanced inoperable stages of diseases.