Clinical usefulness of serum cytokeratin 19 fragment as a tumor marker for lung cancer

Yoshinobu Maeda, Yoshihiko Segawa, Nagio Takigawa, Ichiro Takata, Nobukazu Fujimoto

Research output: Contribution to journalArticlepeer-review

12 Citations (Scopus)


Serum soluble cytokeratin 19 fragment (CYFRA) levels were measured in 251 patients with lung cancer and 139 patients with benign lung diseases to determine the clinical usefulness of CYFRA level determination in the diagnosis and monitoring of lung cancer. Serum levels of CYFRA were measured by a 2-step sandwich ELISA method. When the cut-off value was defined as 3.5 ng/ml, which was associated with a specificity of 95% for benign lung diseases, CYFRA had a high sensitivity (53%) in all patients with lung cancer. Both the serum level of CYFRA and its sensitivity increased significantly with the increase in clinical stage. A comparison of areas under receiver operating characteristic curves showed that CYFRA had the most power of discrimination in the diagnosis of lung cancer among markers including carcinoembryonic antigen, squamous cell carcinoma antigen, carbohydrate antigen 19-9, and neuron-specific enolase. A good correlation was found between serial changes in serum CYFRA levels during therapy and clinical responses for 18 patients who underwent chemotherapy and/or radiotherapy. Our findings suggest that CYFRA may be a marker of choice for screening and monitoring of lung cancer, particularly squamous cell carcinoma.

Original languageEnglish
Pages (from-to)764-771
Number of pages8
JournalInternal Medicine
Issue number10
Publication statusPublished - Oct 1996
Externally publishedYes


  • Carbohydrate antigen 19-9
  • Carcinoembryonic antigen
  • Neuron-specific enolase
  • Non-small cell lung cancer
  • Squamous cell carcinoma
  • Squamous cell carcinoma antigen

ASJC Scopus subject areas

  • Internal Medicine


Dive into the research topics of 'Clinical usefulness of serum cytokeratin 19 fragment as a tumor marker for lung cancer'. Together they form a unique fingerprint.

Cite this