Background
Urinary bladder cancer (BCa), which ranks first in the list of the most life-threatening urinary malignancies, has become a worldwide issue of public health [
1]. According to the latest report, almost 20 thousand people die of bladder cancer every year in the USA, not to mention over 79,000 new cases are diagnosed [
2]. Among the BCa patients, over 70% suffer from non-muscle-invasive bladder cancer (NMIBC), which can be treated by transurethral resection of bladder tumors (TURBT) and have a much better prognosis than muscle-invasive bladder cancer (MIBC) [
3]. However, the tumor can quickly invade the muscle layer and progress frequently to a lethal condition with limited treatment options [
4]. Hence, the early diagnosis and intervention of bladder cancer is the key of treating the disease and improving outcomes.
Despite the great progress made in molecular and genetic diagnostics, the screening of bladder cancer is still trapped by the weakness of the current diagnostic methods such as cystoscopy and urine cytology. Cystoscopy is still the standard way for the detection and diagnosis of bladder cancer. But this invasive examining method is costly and may lead to trauma and infection of the urinary system [
5]. Urine cytology (Cyto), although non-invasive and specific, has a rather low sensitivity (approximately 35%) with an increased risk of missed diagnosis [
6,
7]. To get rid of this predicament, scientists have spent over 20 years looking for a sensitive, specific, and non-invasive biomarker for the detection of bladder cancer. A variety of urinary markers, due to their non-invasiveness and simplicity, have been developed currently [
8]. Among them, U.S. Food and Drug Administration has approved bladder tumor antigen (BTA), fibrin/fibrinogen degradation product (FDP), and nuclear matrix protein 22 (NMP 22) for clinically detecting patients with bladder cancer [
9].
Fibronectin (Fn) is a large dimeric structural glycoprotein which basically has two biological types, termed plasma and cellular fibronectin [
10]. Plasma Fn is synthesized by hepatocytes and released to the circulation while the cellular fibronectin participates in constituting the extracellular matrix and can be found in most of the tissues [
11]. In the urinary tract, when tumor exists, fibronectin can be present due to the degradation of the extracellular matrix caused by proteases as well as the leakage from the blood [
12‐
14]. Therefore, the urine Fn may become a potential biomarker for detecting bladder cancer. In 1993, Shen et al. for the first time revealed that the level of urine Fn in bladder cancer patients was significantly higher than that in patients with benign urothelial diseases and the health groups and urine Fn could be utilized as a valuable biomarker for diagnosing bladder cancer [
15]. This finding was quickly confirmed in the same year by Malmstrom, who further proved the follow-up value of urine Fn in bladder cancer patients [
16]. Since then, the diagnostic value of this urine molecule in detecting bladder cancer has been discussed by various researches and some satisfactory results were reported [
17‐
24]. Recent study showed that urine Fn has a sensitivity of 91.4% and a specificity of 87.8% in detecting residual bladder tumor after TURBT [
21]. Moreover, a significantly higher level of urine Fn was found in MIBC patients by some investigators [
19,
20]. Although the extensive analyses have been carried out, due to the limited clinical trials, different types of patients, insufficient study populations, and heterogeneous cut-off values, the application of urine Fn in the diagnosis of bladder cancer still needs to be verified, and a detailed evaluation of its diagnostic value would be an essential step before the biomarker’s popularization.
In order to fully analyze the diagnostic performance of urine Fn in bladder cancer patients, we conduct a systematic review with meta-analysis based on eight original researches, which will allow us to sum the relevant researches up and provide more precise estimates of the diagnostic value of urine Fn. Moreover, we also examined whether the combination of urine Fn and urine cytology (Fn+Cyto) can remedy the rather low sensitivity of urine cytology and enhance its diagnostic performance in bladder cancer.
Discussion
Bladder cancer is a frequent and life-threatening tumor with huge metastasis rate and mortality [
34]. However, a non-invasive, timely and accurate diagnosis of BCa remains lacking in our clinical practice. Cystoscopy along with biopsy is still the gold standard for diagnosing BCa, but this invasive testing method has various complications and the results to some extent depend on the samples’ quality and urologists’ as well as pathologists’ experiences. Urine cytology, as a useful adjunct for cystoscopy, often suffers from quite low sensitivity and has atypical results [
35,
36]. In order to find a non-invasive and accurate way of detecting BCa and improve the low sensitivity of cytology, many voided urine molecules such as NMP22, BTA, and urine Fn have been developed as non-invasive diagnostic biomarkers for this malignancy. Among those biomarkers, the urine Fn test performance has been well studied since its introduction by Shen and Malmstrom in 1993 [
15,
16]. Due to the limitation of the previous studies, the detailed diagnostic power of urine Fn in the detection of BCa still needs to be investigated. For this reason, we conducted the meta-analysis to pool all the eligible studies to fully evaluate the real and detailed diagnostic performance of urine Fn for bladder cancer.
In the present meta-analysis, we included eight individual studies, containing 744 bladder cancer patients. Our meta-analysis showed that urine Fn had a relatively high diagnostic value, with an AUC of 0.83, a Q* index of 0.798 and a rather restricted confidence region and prediction region. The pooled sensitivity and specificity of urine Fn were 0.80 and 0.79, respectively, which were nearly the same. Besides, the overall DOR was 15.18. Although the PLR and NLR of this biomarker seemed not to be so outstanding, based on the above-mentioned good results, we still concluded that urine Fn could be a powerful biomarker for detecting BCa. Besides, a certain degree of heterogeneity was observed in this meta-analysis, especially in Spe and DOR. Accordingly, the meta-regression showed that measurement units, pathological types, assay methods, and the proportion of NMIBC cases were the main source of heterogeneity.
In consideration of the influences of confounding factors on diagnostic accuracy, subgroup analysis was carried out on the basis of some common covariates and the results were complicated. First, some investigators tried to use urine creatinine or some other calibrators to adjust the urine Fn level in order to correct the differences caused by urine concentrations [
19,
21‐
23]. Thus, in some researches, the urine Fn/Cr ratio was widely used to replace the urine Fn concentration (μg/L). The subgroup analysis, however, did not show any improvement of the diagnostic accuracy in the “non-μg/L” subgroup. Instead, the group using the original Fn concentration as the measurement unit had a much better diagnostic performance than the other group, with an AUC of 0.88 (0.80–0.94) versus 0.80 (0.76–0.83). We all know that the urine creatinine to a great extent depends on the renal function of different patients, so choosing it to adjust the urine Fn concentration might lead to more complicated and vague outcomes rather than more precise results. Therefore, the original urine Fn concentration was recommended for the future researches for the sake of more reliable results. Moreover, the studies based on both BTCC and SCC had a larger AUC (0.89) than that of the studies based on BTCC alone (0.86), indicating that urine Fn also had potential diagnostic value for SCC and might be popularized in the high incidence area of bladder SCC, such as Egypt. In addition, to our knowledge, the urine Fn level might be significantly higher in MIBC patients than that in NMIBC patients [
19,
20]. Interestingly, the pooled analysis of the “NMIBC > 50%” group exhibited better diagnostic performances. This result indicated that the urine Fn could be used as a biomarker for both NMIBC and MIBC cases. More researches are needed so as to compare the diagnostic performance of urine Fn in NMIBC patients.
Furthermore, we also collected data of a combined method (urine Fn combined with urine cytology). The pooled sensitivity, specificity, DOR, diagnostic score, PLR, NLR, and AUC was 0.86 (95%CI = 0.82–0.90), 0.77 (95%CI = 0.70–0.84), 21.20 (95%CI = 14.30–31.44), 3.05 (95%CI = 2.66–3.45), 3.82 (95%CI = 2.87–5.08), 0.18 (95%CI = 0.13–0.24), and 0.89 (95%CI = 0.86–0.92), respectively, and the Z tests showed that the AUC and the sensitivity of the combined methods were significantly higher. These results suggested that the performance of urine Fn to detect bladder cancer could be significantly improved in combination with urine cytology. Regarding the widely reported “Achilles’ heel” of urine cytology—the rather low sensitivity, this combined method (Fn+Cyto) appeared to enhance the sensitivity for both Fn alone and cytology alone. In addition, the application of the two tests in combination would become a potential alternative choice to partly displace the invasive cystoscopic evaluations.
Besides, the limitations of this meta-analysis were as follows. First, a not small heterogeneity was observed in this meta-analysis, especially in the Spe and DOR. Although we tried to explore the sources of the heterogeneity via performing Spearman rank correlation test, univariable meta-regression and subgroup analysis, the heterogeneity of DOR was still obvious. We noticed that the I2 of the DOR were slightly decreased in “BTCC & SCC” group, so we think the heterogeneity of DOR might come from the different pathological types. Second, the included bladder cancer cases were less than 1000 patients and the limited sample size could influence the analytical power. Third, due to the various units of measurement and the diverse values in different studies, we could not give a pooled cut-off value of urine Fn in detecting bladder cancer. The specific cut-off value for clinical use can vary from country to country and from hospital to hospital, which should be confirmed by clinical practice based on a large population. Finally, the diagnostic biomarker of urine Fn has not been widely applied in clinical practice. Future studies should focus on how to improve the diagnostic power of urine Fn and how to test urine Fn more efficiently and easily. More and more large, multi-center and prospective studies are needed in order to validate the diagnostic power of urine Fn.