Does the assessment of serum markers in patients with lung cancer aid in the clinical decision making process?

This survey describes potential clinical applications of the tumour markers CYFRA 21-1, SCC antigen, NSE, and CEA in patients with lung cancer. Due to the rather low prevalence of bronchogenic carcinoma in the general public and the limited diagnostic accuracy, currently available tumour markers are unsuitable for the screening of asymptomatic individuals. All studies performed so far in patients with histologically confirmed NSCLC, agree that the best performance characteristics, in terms of sensitivity and specificity, were obtained with the CYFRA 21-1 test (sensitivity: 40-66%, specificity: 95% versus patients with benign pulmonary disorders) while NSE was found to be the marker of first choice in patients with SCLC (sensitivity: 77-85%). For diagnostic purpose, the value of tumour markers must be compared with the efficiency of standard clinical methods including imaging techniques and cytopathological examinations (detection rates: sputum cytology: 40-70%, biopsy at bronchoscopy in central tumours: 95-98%, biopsy at bronchoscopy + bronchial washing + thin needle aspiration in peripheral tumours: 85%). These figures show that the diagnostic yield of cytopathological examinations by far exceeds that of tumour markers. In addition, these investigations supply with histology and give informations on the T-stage (bronchoscopy). Tumour markers, however, may be used for diagnosis in advanced stages in which patients are very often not eligible for extensive investigations due to their performance status. In the differential diagnosis between NSCLC and SCLC a combination of CYFRA 21-1 and NSE was claimed to be helpful. It was demonstrated that 97% of patients could be correctly classified. NSE was shown to be useful to distinguish SCLC from malignant lymphoma, both the Hodgkin's (rate of false-positive elevations: 6.5%) and the non-Hodgkin's (rate of false positive elevations: 22.4%) types. By applying a cut-off point of NSE assays of 21.9 ng/ml corresponding to a 95% specificity versus the lymphoma group, SCLC is still indicated by elevated NSE levels with a sensitivity of 57.7%. Although a positive correlation of marker concentrations with increasing anatomical tumour extent could be demonstrated, the markers cannot be used for staging purposes due to a considerable overlap of marker levels between the individual stages. CYFRA 21-1 was shown to be unable to differentiate between operable (TNM I-IIIa) and inoperable (TNM IIIb/IV) NSCLC patients. The latter were identified with a detection rate of only 17% by the CYFRA 21-1 test (specificity 95% versus operable patients, cut-off point 20 ng/ml). Pretreatment-measured tumour markers, in particular CYFRA 21-1, were shown to provide prognostic information for the overall survival. The negative prognostic effect of CYFRA-21-1 was independent of classical prognostic markers such as performance status and tumour extent. There are several potential applications of serially-assessed tumour markers for disease monitoring of patients under therapy. In SCLC, increasing NSE levels within the remission phase were demonstrated to be strongly suggestive of tumour recurrence. This finding should give rise to further diagnostic procedures. NSE, however, was not able to differentiate between partial and complete remission since, in both cases, NSE levels dropped to the normal range; thus, NSE cannot replace clinical response evaluations. In NSCLC, it was found that curative surgery resulted in a significant drop of preoperatively elevated CYFRA 21-1 or SCC antigen levels down to the normal range. Although rising SCC antigen levels in the postoperative surveillance of patients with squamous cell carcinoma indicated very early tumour relapse, these results are of minor clinical utility due to the absence of curative therapy. Serial measurement of CYFRA 21-1 during chemotherapy in patients with inoperable squamous cell carcinoma has shown that there is a concordance of 74% between the course of the m