Background
Lung cancer has become the first cause of cancer-associated death in the world [
1]. This is a consistent opinion that early diagnosis and individualized therapy are conducive to improve the prognosis of lung cancer [
2]. Many studies have demonstrated that abnormal protein expressions and gene mutations are correlated with the ontogenesis and progression of lung cancer [
2], and reliable biomarkers derived from these abnormal molecules are more likely to help make the medical decision for individualized therapy [
3]. We also know that the high mortality of lung cancer is mainly due to early metastasis and progression, and early diagnosis of lung cancer can increase the 5-year survival rate from 15 to 80 % [
4]. Thus, new technology on early diagnosis and therapies are greatly required.
Recently, chronic inflammation has been showed to be associated with tumor progression, and many inflammatory factors could serve as diagnostic and prognostic biomarkers for special tumors [
5]. There is common view that inflammation can become chronic processes that may promote angiogenesis and proliferation of cells, thus it may play a clear role in carcinogenesis and pathogenesis [
6]. Serum amyloid A (SAA), a kind of cytokine-induced, acute inflammatory response proteins, has been known to be likely involved in cancers [
7]. Research shows that liver is mainly workplace for producing SAA protein which can stimulates the production of various cytokines, and SAA plays an important role in acute immune response [
8]. SAA protein in blood of patients with cancer often rises at its early stage, which have been identified both by immunochemistry and by proteomics methods in different cancers, such as lung, ovarian, renal, uterine, nasopharyngeal cancer and in melanoma [
7].
Up to now, lung cancer, a very common malignant tumor, has been considered as an inflammatory disease, and the development of lung cancer correlates various cell factors and inflammatory mediators. Previous studies have specially investigated the relationship between SAA and lung cancer. These studies suggest that higher SAA can distinguish lung cancer patients from healthy individuals as well as predict the prognosis of lung cancer [
9], which may be a potentially non-invasive biomarker for lung cancer. Here, we reviewed the medical literature as completely as possible, and conducted a meta-analysis to show the relationship between the expression of SAA and lung cancer and evaluate its value for lung cancer diagnosis.
Methods
Literature searching
The databases that we searched studies on SAA and lung cancer included Medline, Embase, Cochrance Library and Web of Science. The time scope that we defined was from the start of each database up to June 2016. The key words that we used for searching literature included: “lung cancer,” “lung malignancy,” “lung malignant tumor,” “lung neoplasms,” “serum amyloid A,” and “SAA.” We also conducted secondary searches for additional studies that regarding the SAA and lung cancer from the reference lists of included studies.
Inclusion and exclusion criteria of literature
The inclusion criteria: (1) patients in study must be histologically diagnosed with lung cancer; (2) must be case–control or cohort association studies; (3) detection method of CAA must be able to show the continuous variables; (4) studies must have reported sufficient quantitative data; and (5) the methods of data collection and analysis must be statistically acceptable. The exclusion criteria: (1) non-original reports (such as abstracts, letters, editorials and expert opinions and case reports); (2) did not report clearly serum level of SAA with continuous variables; (3) did not contain distinctively normal control; (4) patients had been given the chemotherapy and surgery before taking blood samples; and (5) non-human studies.
The extracted data included: (1) authors, countries and publication date; (2) study design and case number of different groups; (3) gender and age of patients; (4) tumor node metastases (TNM) classification of lung cancer patients; (5) histological classification; (6) detection method of SAA; (7) SAA level; (8) the number of true positives, true negatives, false positives, and false negatives.
Methodological quality assessment
We adopted the guidelines of the QUADAS-2 [
10,
11] (maximum score 14) tool to assess the methodological quality of included studies, in which appraisal is performed by empirical evidence, expert opinion, and formal consensus on assessing the quality of primary studies of diagnostic accuracy [
11]. In order to reduce the bias and improve the reliability, two authors independently assessed and reached a consensus. If there were a discrepancy, we would invite another expert to discuss it and reach a consistent opinion.
Statistical analysis
We performed the statistical analysis according to the following research idea. The standardized mean difference (SMD) and their 95% confidence intervals (CI) of lung cancer associated with the SAA was calculated directly from the data given in the eligible studies using two different meta-analysis approaches (fixed effect method and random effect method). The heterogeneity test between studies was assessed by the Chi-square test and I2. In the absence of heterogeneity, we used the fixed effects method, otherwise the random effect method was used. The overall effect of meta-analysis was tested using Z-scores with a significance of being set at p <0.05. We also ran a sensitivity analysis to determine whether the overall effect was affected by individual study. The publication bias was evaluated using Begg’s and Egger’s test respectively. Moreover, we drew a summary receiver operating characteristic (SROC) curve to determine the joint distribution of sensitivity and specificity. Statistical analysis was performed using SPSS (Version 22.0, Chicago, USA), RevMan 4.2 (Cochrane Collaboration), Meta DiSc statistical software (Version 1.4, Madrid, Spain), and Stata version 12.0 (TX, USA). All the tests were two-sided and the significant level was 0.05.
Discussion
Now, lung cancer has become the leading cause of malignancy-related deaths in the world [
17,
19], the 5-year survival rate for lung cancer is only slightly better than 10%. Lung cancer exhibits the highest mortality of all cancers mainly because most patients have developed into the advanced stage when the diagnosis of disease is confirmed [
4]. People believed that stable biomarkers which can be routinely measured in easily accessible samples effectively help make early-stage diagnosis for lung cancer [
20]. Blood is an easily accessible and rich body fluid. Research shows that blood plasma and serum contain specific proteins that provide potential circulating biomarkers [
21]. For example, the level of acute-phase SAA often increases in cancer patients, even at its early stage. This fact was registered in different common cancers, such as lung, ovarian, renal, uterine, and nasopharyngeal cancer and in melanoma [
7].
In this study, we reviewed the relevant studies comparing the expression of SAA between lung cancer and healthy individuals and found that patients with lung cancer showed a higher SAA level than those of healthy group. This result indicated that a higher SAA level certainly correlated with occurrence and development of lung cancer and that SAA could be an indicator of lung cancer. We noticed that there was methodology heterogeneity that existed between included studies, but we found that included studies had a very good clinical homogeneity. For instance, no biases of age and diagnosis were observed in these studies. Moreover, patients included in these studies were from East Asia, Europe and America, which embodied the globalization and thus eliminated the ethnic bias. In order to strength the reliability of results, we made a comparison of SAA positive rate and showed that SAA positive rate of patients with lung cancer was higher than that of healthy individuals.
We also found that most of studies had a moderate to higher quality assessed by using the QUADAS-2 scoring system. Subsequent analysis of sensitivity further showed that the exclusion of studies on an individual basis did not substantially modify the overall effect of meta-analysis. Bias evaluation [
11] in our analysis suggested that there was not a significant publication bias. Together, the results of this meta- analysis should be more stable. Previous studies have found that SAA can distinguish lung cancer patients from healthy controls as well as predict prognosis of lung cancer [
8,
12,
15,
17]. SAA is secreted during the acute phase of inflammation, including invertebrates and vertebrates, suggests that SAA has an essential role in all animals including humans [
15]. Study point out that overexpression of SAA is always correlated with inflammation and acute-phase responses [
16]. Further, investigation on cancers reveals that chronic inflammation is associated with development and progression of malignant tumors, and inflammatory factors can be applied as diagnostic and prognostic indicators for some malignant tumors. SAA is a kind of inflammatory factor, adding our findings, thus showing that there is strong relationship between chronic inflammation and incidence of lung cancer.
It is likely that SAA in pulmonary inflammation may be temporarily elevated and recovered soon after the elimination of infection, but not the same in cancers, which may represent a primary difference between benign and malignant diseases of lung [
17]. In our analysis, we were excited to find that LSCC displayed a much higher SAA level than LAC and SCLC, which gave us a very significant clue that we might specially use SAA for discerning LSCC from others. The results also confirmed by subsequent evidence that overexpression of SAA even was detected western blot analysis in LSCC, but not in others [
4]. It is widely known that there has still no efficient biomarker for LSCC diagnosis so far. As an indicator of the potential usefulness of SAA in the diagnosis of lung cancer, in particular in LSCC, we ought to investigate deeply the role of SAA in LSCC in the future.
It is unassailable, as a diagnostic marker, a good sensitivity and specificity are very important. In this meta-analysis of diagnostic test we found that the increase of SAA has a higher specificity (0.92; 95% CI: 0.88-0.95) for discerning lung cancer. However, the pooled sensitivity was only 0.56 (95% CI: 0.54-0.63), which suggested that SAA has a better role for distinguishing lung cancer but not for screening. Thus, when biopsy of tumor tissue is absent or insufficient in clinic, we may use the SAA as an indicator to discern lung cancer. However, the absence of increased SAA should not mean the impossibility of lung cancer. The DOR always indicate the test accuracy of a biomarker that bind the compromise of sensitivity and specificity to a quantitative data. People believed that a higher DOR values suggest a higher accuracy of diagnosis. In our analysis, the pooled DOR was 27.52, supporting that the SAA assay could be advantageous in the diagnosis of lung cancer. The definitive diagnosis of lung cancer usually requires tissue biopsies of adequate size. However, sometimes the tissues for pathology biopsy were insufficient, and then a test with SAA would help improve the differential diagnosis. The SROC curve has been recommended to represent the performance of a diagnostic test [
11]. Our analysis showed that the AUC of SAA was 0.9066, which indicated that the SAA has good value in terms of the discerning diagnosis of lung cancer. From the present data, we think that every patient with suspected lung cancer should undergo the test of SAA. Patients with positive SAA level should undergo further invasive procedures biopsies, and produce a final diagnosis.
The limitations of this study are as follows: first, some studies had small size; second, some studies had relatively low quality in clinical and statistical designs; third, detection methods of SAA were different in these studies. In the future, it is very crucial to compare the SAA status in different histology classification of lung cancer with large samples, multiple clinical centers. Although some deficiencies existed, the study still drew a conclusion that the SAA assay could be advantageous in the diagnosis of lung cancer, especially for LSCC.
Acknowledgements
We appreciate the great help of Mr. P H, and Miss G J as interviewers.