Meta analysis of the value of CYFRA21-1 in early diagnosis of non small cell lung cancer

Navigation:Home > Respiratory Medicine > Lung Cancer > Meta analysis of the value of CYFRA21-1 in early diagnosis of non small cell lung cancer

Liu Yanfeng 1 Yang Weilin 2 Yang Shuanying 1 is 1 Li Cheng Xi'an 3 Zhang Wei 1 Zheng Huadong 2 Du Jie Huo 1 Shufen 1 Chapter 2 Lina 1 Lin Bu


Liu Yanfeng 1 Yang Weilin 2 Yang Shuanying 1 is 1 Li Cheng Xi'an 3 Zhang Wei 1 Zheng Huadong 2 Du Jie Huo 1 Shufen 1 Chapter 2 Lina 1 Lin Bu Tian 1 East rock should be 1

1 Department of respiration, the Second Affiliated Hospital of Xi'an Jiao Tong University Medical College, Xi'an 710004

The old Department of respiration of the Second Affiliated Hospital of Xi'an Jiao Tong University Medical College 2

Department of respiratory medicine, Tongchuan People's Hospital, Shaanxi 3, China

Correspondent: Yang Shuanying

Tel:13991392919; email:

Abstract: Objective To evaluate the value of CYFRA21-1 in early diagnosis of non-small cell lung cancer. Clinical study on CYFRA21-1 methods have been published in the diagnosis of non-small cell lung cancer, evaluate the quality of the original research according to the requirements of Meta analysis of the retrieved, Meta analysis of all the results meet the conditions. Results a total of 6 articles were included, of which 4 Chinese, English 2, sample of 398 cases, which were confirmed by pathology of stage a NSCLC was 140 cases, the homogeneity of the study better heterogeneity test showed that, through the summary and determine the model calculated the pooled sensitivity was 35% (24%-44%) summary, the specificity was 96% (92%-98%), summary the positive predictive value was 7.52 (4.07-13.90), summary the negative predictive value was 0.66 (0.49-0.89), the area under SROC curve was 0.94. Conclusion CYFRA21-1 can be used for early diagnosis of non-small cell lung cancer, and can be used as one of the most important indexes.

[Key words] CYFRA21-1 analysis of non small cell lung cancer Meta

Meta-analysis on Diagnostic of in Early Non-small Cell Stage Lung Value CYFRA21-1 Cancer

[Abstract] Objective To evaluate the diagnostic Value of CYFRA21-1 in early stage non-small cell lung cancer. Methods All published clinical studies which evaluated the diagnostic value of CYFRA21-1 in non-small cell lung cancer were reviewed. The retrieved original studies were evaluated according to the requirements of Meta-analysis. Meta-analysis was used in all eligible results. Results Six data including 4 in Chinese and 2 in English were collected. The sample volume was 398, of which 140 cases confirmed by pathology was in a period of NSCLC. I - III Heterogeneity test showed that the homogeneity of the study was good. By using deterministic models to analyze the data, the value of the weighted sensitivity was 35% (24%-44%), the Specificity was 96% (92%-98%), the positive likelihood ratio was 7.52 (4.07-13.90), negative likelihood ratio was 0.66 (0.49-0.89), and the SROC area under the curve (AUC) was 0.94. Conclusions CYFRA21-1 could be regarded as one of the valuable reference tests in early stage NSCLC but more, high quality, lager sized, prospective studies are required to evaluate its clinical value.

[Key Words] CYFRA21-1 cell lung cancer Meta-analysis Non-small

CYFRA21-1 is a soluble fragment of cytokeratin 19, which is mainly used as a tumor marker for the diagnosis of lung cancer. Cytokeratin (cytokeratins) is a subunit of intermediate filaments that form the structural proteins of epithelial cells. According to the molecular weight and the difference of the two dimensional electrophoresis, it can be divided into 20 types. They are divided into two subgroups: class I (acidic protein), class II (basic protein). Cytokeratin is different polymer composed of class I and class II keratin. Cytokeratin 19 (CYK-19) is a class of keratin (acidic protein) with a molecular weight of about 40000 Da, the smallest member of the keratin family. CYK-19 is widely distributed on the surface of normal tissue, such as stratified or squamous epithelium. In malignant epithelial cells, activation of protease accelerates cell degradation, making a large number of cytokeratin fragments released into the blood, the soluble fragments can be combined with the two monoclonal antibodies KS19.1 and BM19.21 specific, so called CYFRA21-1[1]. CYFRA21-1 is considered to be a mainly used for detection of lung cancer tumor markers, especially for non-small cell lung cancer (non-small cell lung cancer, NSCLC) has important value in the diagnosis of NSCLC, the positive rate was 50% ~ 85%. The serum level of CYFRA21-1 was positively correlated with the clinical stage of [2], and it could be used as an effective index for early recurrence after lung cancer surgery, radiotherapy and chemotherapy. High serum levels of CYFRA21-1 suggest that the disease is progressing and the prognosis is poor. The success of the treatment is the rapid decline in serum concentration of CYFRA21-1, while the other is not complete removal of the lesion. Serum levels decreased and then increased, suggesting that the disease recurrence. In recent years, the role of CYFRA21-1 in the diagnosis of NSCLC is studied, to make different degrees of recognition of its value in the early diagnosis of NSCLC, but the independent study of the sample is small, there are some limitations, cannot exclude the existence of negative CYFRA21-1 to lung cancer. Therefore, in order to correctly and scientifically evaluate the value of CYFRA21-1 in the early diagnosis of non-small cell lung cancer, we collected the published literature for Meta analysis.

1 materials and methods

1.1 inclusion and exclusion criteria

1.1.1 study of diagnostic accuracy for diagnostic accuracy; extraction of four grid data; limited to published in English and Chinese literature.

1.1.2 patients with lung cancer, lung benign lesions, all patients with lung cancer confirmed by pathology, clinical stage is limited to I, II, III a.

1.1.3 diagnostic test methods CYFRA21-1 detection application commercial reagents.

1.1.4 measurements were combined with sensitivity (SEN), combining specificity (SPE), positive likelihood ratio (PLR), negative likelihood ratio (NLR), and area under the SROC curve.

1.1.5 exclusion criteria: (1) abstracts, lectures, reviews and reviews. (2) the control group included the normal population. (3) all or part of NSCLC patients were not confirmed by pathology. (4) no TNM staging.

1.2 retrieval strategies: to "CYFRA21-1" and "non small cell lung cancer" theme words combined retrieval in 1990 - 2009 MEDLINE, EMbase, Chinese biomedical literature database, China biomedical journal literature database, China academic journal database, Chinese science and Technology Journal Full-text Database, manual retrieval related magazines, and from the related references in the search may meet the inclusion criteria of literature.

1.3 literature screening and data extraction according to the pre established inclusion and exclusion criteria were screened to extract the published information, people study demographic data, disease situation, detection method and detection data, and use the four table description.

1.4 the quality evaluation of literature: the quality classification of literature mainly refers to the quality standard of [2], which is evaluated independently by two reviewers. Combined with the specific circumstances are as follows: (1) whether the establishment of "gold standard" or reference standards; (2) whether the independent compared with the gold standard or reference standards; (3) whether the comparative analysis of the data using blind method; (4) whether into the research object, whether the patient including a variety of diagnostic tests for use. Clinical practice; (5) whether to use the correct method to determine the critical value; (6) is provided for calculation of the sensitivity, specificity and likelihood ratio data; (7) whether the diagnostic tests can be repeated; (8) to study whether the case to explain the exit. According to this standard, the quality of research will be divided into 5 grades: Grade A, comply with all quality standards; B, meet the first 4 criteria, the other does not fully comply with the standard; grade C, meet the first 3 criteria, the other does not fully comply with the standard; grade D, meet the first 2 criteria, other standards are not completely meet; E, does not comply with any 2.

1.5 statistical analysis: heterogeneity test: heterogeneity analysis using Meta -Disc software. Diagnostic test of heterogeneity was caused by threshold effect and non threshold effect, the test by calculating the sensitivity and logarithm (1- specificity) of the threshold effect test Spearman correlation coefficient of logarithm, by calculating the DOR (diagnostic odds ratio) Cochran-Q value test of non threshold effects. If P>, 0.05 had no statistical heterogeneity, fixed effect model was used. If P< and 0.05 have statistical heterogeneity, the random effects model. With the sensitivity calculation using the Meta Disc software [3,4] (SEN), and specificity (SPE), positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds (DOR) and 95% confidence interval (confidence, interval, CI), and the summary receiver operating characteristic (SROC) curve analysis and the estimation of the area under the SROC curve (area under, curve, AUC).

2 Results

2.1 research into basic features: there were 231 literatures, by reading the title and abstract, screened out 39 studies met the inclusion criteria, we included 6 studies, including 4 Chinese [5-8], English 2 [9,10], a total of 398 subjects, including the gold standard for the diagnosis of NSCLC I - III a stage 140 cases, benign lung disease (benign lung, diseases, BLD) in 258 cases. The basic characteristics of each study are shown in table 1.

2.2 the quality of the included studies was evaluated: all the studies were based on the pathological diagnosis as the gold standard, and compared with the pathological diagnosis. Patients in the control group were treated with benign pulmonary disease. 1 cases of [6] did not specify the critical value of the experimental results. All references were not mentioned in the results of the observation by the blind and the research object with continuity. Specific evaluation results are shown in table 1.

Table 1 Characteristics of the study




research design

Blind method


Case number









One thousand nine hundred and ninety-six

Diagnostic test




Three point three







Two thousand and one

Diagnostic test











Two thousand and three

Diagnostic test




Three point three






Hui Jie Duan

Two thousand and nine

Diagnostic test




Three point three






One thousand nine hundred and ninety-seven

Diagnostic test




Three point three






Nisman B

One thousand nine hundred and ninety-eight

Diagnostic test



One hundred and seventeen

Three point three





2.3 the results of Meta analysis showed that the Spearman correlation coefficient between the logarithm of sensitivity and the logarithm of 1- specificity was 0.232, P=0.658, indicating that there was no threshold effect. The Cochran-Q value of DOR is 5.95, P=0.311, suggesting that there is no threshold effect. Therefore, it is considered that the inclusion of the literature is homogeneous, and the fixed effect model is used. See figure 1. The sensitivity of the 6 studies was (24%-44%), and the specificity was 96% (92%-98%), as shown in figure 2. The positive predictive value was 7.52 (4.07-13.90), and the negative predictive value was (0.49-0.89), as shown in figure 3. The SROC curve is shown in Figure 4, AUC is 0.9441.

Note: the red dots in the diagram each line through independent research, dot for 95% confidence interval, the arrow area shows the data not shown area. The greater the dot, line distribution is narrow, the amount of sample that is greater, the greater the accuracy, the weights given in the consolidated greater; conversely, the accuracy is small, weight is also smaller. The red diamond point is a summary value.

Figure 1. Forest plot of odds ratio of CYFRA21-1 for early diagnosis of NSCLC

Note: the red dots in the diagram each line through independent research, dot 95% confidence interval. The greater the dot, line distribution is narrow, the amount of sample that is greater, the greater the accuracy, the weights given in the consolidated greater; conversely, the accuracy is small, weight is also smaller. The red diamond point is a summary value.

Figure 2 forest map of CYFRA21-1 sensitivity and specificity for early diagnosis of NSCLC

Note: PLR shows that the probability of correct diagnosis is positive and the probability of error is positive. The greater the odds ratio, the greater the probability that the positive is true. The smaller the NLR, the greater the likelihood that the test result is true.

Figure 3 CYFRA21-1 positive likelihood ratio and negative likelihood ratio forest map for early diagnosis of NSCLC

Note: the blue line in the middle is SROC, and the blue line on both sides is 95% confidence interval. The red dots in figure 6. The greater the dot, the study suggests that the samples quantity is bigger, the greater the accuracy, the weights given in the consolidated more; instead, the accuracy of the smaller, the smaller the weights given.

Figure 4 receiver operating characteristic curve of CYFRA21-1 for early diagnosis of NSCLC

3 discussion

Due to the early general lung cancer is often asymptomatic by imaging examination of lung cancer, often the cancer has spread, and lost the chance of early surgical treatment, so to find good specificity and high sensitivity of serum tumor markers is an important clinical issue. CYFRA 21-1 is the most valuable tumor marker for non-small cell lung cancer ([11]). For NSCLC, I, II surgery as the main treatment, the third stage a resection rate of about 20-30%, therefore, the study included I - III a phase NSCLC patients.

The results showed that the sensitivity of CYFRA21-1 for early diagnosis of was 35% (27%-44%), which indicated that the rate of missed diagnosis was higher than that of NSCLC. The specificity was 96% (92%-98%), which showed that the misdiagnosis rate was low, which was about 4%. At the same time reflect the likelihood ratio belongs to the composite index of sensitivity and specificity, when PLR is greater than 10 and less than 0.1 NLR, with a diagnosis of convincing effect, when PLR is greater than 5 and less than 0.2 NLR, with a strong diagnostic efficiency, this study summarizes PLR 7.52 (4.07-13.90), suggesting that through early detection of CYFRA21-1 patients in NSCLC, the positive result for the chance is 7.52 times BLD. Summary of NLR was 0.66 (0.49-0.89), suggesting that CYFRA21-1 detection of NSCLC error is the correct judgment of the negative judgment of the value of 66%, indicating that CYFRA21-1 has diagnostic value, but missed diagnosis rate is higher. Effect of SROC curve analysis method with the threshold level, further shows the size of diagnostic test accuracy level, by the area under the curve was calculated and the Q* values of AUC=0.94, Q*=0.88, are close to 1, indicating its high diagnostic efficiency.

In all we included in the literature have significant differences in CYFRA21-1 NSCLC detection sensitivity and specificity, sensitivity 11%-67%, specificity 93%-100%, the reason may be: people (such as disease severity and comorbidity), different test conditions (such as different technology, test, operator); control group patients are different; different kits; determine the thresholds of different methods etc..

This evaluation system for the literature, to evaluate, need further research in accordance with the requirements of clinical research, further prospective studies for early diagnosis of CYFRA21-1 for detection of NSCLC. The evaluation results from the included studies, because only in the literature, English literature, literature are included in the literature without bias; blinded measurement bias; database, positive results of literature was published, and the negative results of the literature tend not to be published, or not to be published in a higher level the magazine, so in the study often negative conclusion is less, the presence of publication bias in the literature; the control group with benign lung diseases including tuberculosis, pneumonia, although according to clinical criteria can be diagnosed, but without pathological diagnosis, so the existence of literature part verification bias and different verification bias; most of the literature on patients with no NSCLC TNM staging, a meta-analysis of multi center for early diagnosis of NSCLC can not. Therefore, the results of this study cannot be generalized to every laboratory. Future research should be: as far as possible the use of diagnostic test report standards (STARD statement), to improve the reporting quality of diagnostic tests; the operation process should be detailed description of the gold standard test and whether or not the blind method; should specify the selection criteria of the research object, a detailed description of the test method and quality control.

In summary, CYFRA21-1 of early NSCLC positive values indicate that the high probability of the disease, the misdiagnosis rate was only 4%, is suitable for early diagnosis of NSCLC, but not suitable for the screening of suspected cases, negative misdiagnosis rate is as high as 65%, false negative rate is higher, therefore, the clinical suspected early NSCLC and CYFRA21-1 is negative the diagnosis cannot be ruled out. The area under SROC curve was 0.94, which indicated that the diagnostic efficiency was high. In order to clear the value of early diagnosis of CYFRA21-1 NSCLC, provides a more scientific and objective basis for its clinical application, also need multi center cooperation, accumulated cases of sufficient sample size and disease stratification, set up control group to adopt a unified detection method with strict external quality control measures, to reduce the bias of guarantee the high quality, credibility, strong guidance results, early diagnosis more accurate evaluation of CYFRA21-1 NSCLC.


1 Takei Y, Minato K, Tsuchiya S, et al.CYFRA21-1: indicator survival and effect in lung cancer.Oncology, 1997,54 (1): 43-47 therapeutic (of)

2 Wang Jiyao. Evidence based medicine and clinical practice. Beijing: Science Press, 2006:14-22

3 Zhang Tiansong, the application of.Meta-Disc software to analysis the clock Fumiaki diagnostic test in Meta. Evidence based medicine, 2008,8 (2): 97-108

4 Liu key, Wu Taixiang. Introduction of -SROC curve method for Meta analysis of diagnostic test. Chinese Journal of evidence based medicine, 2003,3 (1): 41-44

5 pm Peishun.CYFRA21-1 in lung cancer. The diagnostic value of tumor, 1996,16 (2): 123-124

6 Xiong Ying, Fan Zhongyuan, Song Xiaoyu. Clinical application of several tumor markers in primary lung cancer. Chin J Med Med, 2001,4 (9): 721-722

7 Xia Yi, Zhao Bai Ying, should be cleared. The clinical significance of serum CYFRA2 1-1 and CEA detection in non-small cell lung cancer. Zhejiang medical journal, 2003,25 (3): 189-190

8 Duan Huijie, Huang Xiangzeng. The clinical value of serum tumor marker CYFRA21-1 in non-small cell lung cancer. Journal of Xinjiang Medical University, 2009,32 (2): 165-168

9 Br CHOT JM, Chevret S, Nataf J, et al.Diagnostic and prognostic value of CYFRA 21-1 compared with other tumour markers in patients with non-small cell lung cancer:a prospective study of 116 patients.Eur J Cancer, 1997:33385-391

10 Nisman B, Lafair J, Heching N, et al.Evaluation of tissue polypeptide specific antigen, CYFRA 21-1, and carcinoembryonic antigen in nonsmall cell lung carcinoma:does the combined use of cytokeratin markers give any additional information.Cancer, 1998,82:1850-1859?

11 Maeda, Segawa Y, Takigawa N, et al.Clinical of cytokeratin fragment as tumor marker for lung cancer.Intern Med, 1996,35 (3): 764 (usefulness), serum (a)

Cerebral Vascular Disease,Acne,Heart Disease,Deaf,Headache,Std,Condyloma Acuminatum,Fibroid,Pneumonia,Brain Trauma,。 Rehabilitation Blog 

Rehabilitation Blog @ 2018