This study was designed to determine whether analysis of the tumor markers CYFRA 21–1, CEA, and SCC in cytological fluid could improve the performance of NAB in the diagnosis of NSCLC. Our external validation study showed that NAB with additional evaluation of cytological tumor marker of CYFRA 21–1 can improve sensitivity and accuracy in the diagnosis for NSCLC.
Early diagnosis of lung cancer is essential for increased survival, but this is difficult because the symptoms are non-specific and are frequently found in the risk group of smoker patients. In clinical practice, lung cancer diagnosis still depends primarily on imaging techniques, such as x-ray and CT, and if a suspicious lesion is present, biopsy with histopathological examination is used to confirm the diagnosis.
Currently, transthoracic NAB is often performed to obtain a definitive diagnosis and is a useful procedure for diagnosing pulmonary nodules that are highly likely to be malignant, especially in patients who are not candidates for surgery . Although NAB of the lung is a relatively safe and accurate method for diagnosing lung lesions, the results of transthoracic needle biopsy of lung lesions often results in false negative in the diagnosis of malignancy, and non-specific results are common in NAB . Previous reports found that, transthoracic needle biopsy of lung lesions has a false negative rate of up to 29% for malignancy diagnosis [4, 5], and 27% of the non-specific results are reported to later turn out to be malignant . Therefore, patients with suspected lung malignancy but inconclusive results on initial lung biopsy often require a second transthoracic lung biopsy or surgical biopsy including video-assisted thoracoscopic surgery (VATS).
Currently, several tumor markers in the serum have been extensively studied in lung cancer, but none is specific for diagnosis of NSCLC. Tumor markers may aid in clinical diagnosis as well as prognosis and follow-up. Tumor marker determination is a simple, inexpensive test, available in most centers where lung cancer is diagnosed. However, the results to date do not support the widespread use of tumor markers for diagnosis.
We used cytological fluid obtained by biopsy as a new type of sample for tumor marker analysis. We hypothesized that cytological fluid obtained from NAB has the potential to be an effective sample, as it is obtained directly from tumor tissue, and because many biomarkers candidates will exist in high concentrations . Although lung cancer tissue biopsy is invasive, biopsy and resection are currently the gold standards for confirmative diagnosis, and thus are generally performed in cases of high suspicion of malignancy. Furthermore, performing the extra step of measuring concentrations of tumor markers in the fluid that is aspirated does not require an additional puncture, takes little extra time, and is easy.
Results from our previous study indicated that additional evaluation of tumor markers in the cytological fluid can improve the diagnostic performance of CT-guided NAB in NSCLC patients. Our previous results showed a significant increase in sensitivity and accuracy for NAB combined with CYFRA 21–1 compared with NAB alone (100% versus 85.7%, p = 0.001; 97.8% versus 89%, p = 0.0209, respectively) . This was supported by the results of this validation study. When we combined the results of serum tumor markers, sensitivity and accuracy were not significantly different between NAB combined with serum tumor markers and NAB alone (for all three serum tumor markers, p > 0.05). However, sensitivity increased significantly for NAB combined with any cytological tumor marker compared to NAB alone (95% for CYFRA 21–1, p < 0.001; 92.1% for CEA p = 0.002; and 91.4% for SCC, p = 0.003). The accuracy improved significantly for NAB combined with cytological CYFRA 21–1 compared with NAB alone (95.9% versus 88.1%, p < 0.001). When we compared the AUC between NAB combined with tumor markers and NAB alone, the AUC of NAB with cytological CYFRA was significantly larger than the AUC of NAB alone (0.966 versus 0.917, p = 0.009). Furthermore, in subgroup analysis, the sensitivity also increased significantly for NAB combined with any cytological tumor markers compared with NAB alone in early stage lung cancer. This indicated that this cytological tumor marker had additional value in the diagnosis of NSCLC. According to our validation study, NAB combined with cytological CYFRA 21–1 had the best diagnostic performance, similar to findings of previous studies [6, 10]. CYFRA is the most sensitive tumor marker for NSCLC. Wieskopf et al.  reported that serum CYFRA 21–1 was a sensitive and specific tumor marker for NSCLC diagnosis, appearing more sensitive and more specific than other tumor markers such as CEA and SCC.
In the results of subgroup analysis according to histological cell types, the sensitivity and accuracy of NAB combined with cytological CYFRA 21–1 was significantly higher than NAB alone for diagnosing adenocarcinoma. Although CYFRA is one of the most sensitive tumor markers available, its relationship with specific histology is controversial. Some studies reported that CYFRA has no clear relationship to different histological cell subtypes in NSCLC, whereas, others have reported that CYFRA is a more sensitive and specific tumor marker especially for the squamous cell subtype [6, 10, 20]. Previous studies using immunohistochemical analysis for resected adenocarcinoma and squamous cell carcinoma, showed that cytokeratin-19, the marker used for CYFRA21-1, stained both adenocarcinoma and squamous cell carcinoma strongly and indiscriminately . Although NAB combined with cytological CYFRA 21–1 had tendency to increase sensitivity and accuracy for diagnosing squamous cell carcinoma subtype, the sensitivity and accuracy were not significantly different between NAB alone and NAB combined with cytological CYFRA 21–1. This may be explained in that the sample size for squamous cell carcinoma subtype was small. In our study, more than half of the cases were adenocarcinomas, and only 20% were squamous cell carcinomas.
Based on our validation study, cytological fluid could be an effective sample for tumor markers and may be clinically useful in lung cancer diagnosis. This is because while lung biopsy using needle aspiration is a confirmative method for lung cancer diagnosis, 29% of NAB results can be non-diagnostic [18, 19]. Taking the extra step of measuring tumor markers concentrations in the aspirated fluid does not require an additional puncture, takes little extra time, and is easy to perform. Therefore, we believe that in cases of suspicious malignant nodules or masses showing a negative or inconclusive cytological result, determining tumor markers in the cytological fluid may be a helpful complementary tool for lung cancer diagnosis.
Our study has some limitations. First, the major two histologic types were not represented equally. More than half of the cases were adenocarcinomas, and only 20% were squamous cell carcinomas. Furthermore, in subgroup analysis, the sensitivity increased significantly for NAB combined with cytological tumor markers (CYFRA 21–1 and CEA) compared with NAB alone in the diagnosis of adenocarcinoma, whereas, the sensitivity and accuracy were not significantly different between NAB alone and NAB combined with any serum or cytological tumor markers for diagnosing squamous cell carcinoma. Therefore, the value of cytological tumor markers determined in this study might be limited for cell types other than adenocarcinoma. Second, the results may be influenced by the method used to choose the cut-off point. Although serum tumor markers have normal reference values, no reference normal values are available for cytological fluid levels of the tumor markers. In this study, we used the cut-off values for cytological tumor markers determined in our previous study to validate the cut-off values selected for tumor markers as an independent series. In our previous study, ROC curves were constructed using tumor marker values in the cytological fluid and a cut-off value was determined using the maximum Youden index for differentiation between malignant and benign lesions. Third, while most lesions had histopathologically confirmed diagnoses, 29 lesions required follow-up imaging studies and clinical examinations; the follow-up time used to classify lesions as benign was at least 12 months using imaging finding for lesion regression.