Pleural effusion is a common symptom of cancers in the pleural cavities. Accumulation of pleural fluid is caused partly by an increase in vascular permeability and tissue leakage and can limit lung expansion and impede breathing. The effusion is withdrawn, and malignant cells from pleural effusions are often available for cytopathological diagnosis long before biopsies are obtained [
The cytological diagnosis of malignant disease in the pleura is supported by ancillary methods such as immunocytochemistry, fluorescent in situ hybridization (FISH), electron microscopy, and biomarker analyses [
Syndecan-1 (CD138), a cell surface proteoglycan, has been proposed as a cellular marker for distinguishing adenocarcinoma from mesothelioma [
This study aimed to evaluate the diagnostic and prognostic performance of soluble syndecan-1 in patients’ serum and pleural effusions.
In this study, 256 pleural effusions (74 carcinoma, 89 malignant mesothelioma, and 93 benign effusions) and 231 serum samples (74 carcinoma, 91 malignant mesothelioma, and 66 benign conditions) were analysed for syndecan-1 and compared to their osteopontin levels. Pleural effusions were prospectively and consecutively collected in different time periods at the Department of Pathology and Cytology, Karolinska University Hospital in Huddinge, Sweden, between the years 2005 and 2011. Late time periods only included patients with lung cancer or malignant mesothelioma. Serum samples were prospectively and consecutively collected at the Chest Diseases Department of Eskisehir Osmangazi University in Eskisehir, Turkey, between the years 2002 and 2004. Serum samples were collected as part of a parallel study that evaluates the diagnostic effect of seven different biomarkers (including syndecan-1) and their combined diagnostic value for a malignant pleural mesothelioma. All samples were collected before any treatment was given. Samples were centrifuged for 1,700 g for 10 minutes without delay. Acellular supernatants were stored without additives at −20°C (effusions) or −80°C (sera). For patient demographics, see Table
Demographic data.
Pleural effusions | Number of patients | Male (%) | Female (%) | Age, median (IQR) |
---|---|---|---|---|
Benign | 93 | 63 (68) | 30 (32) | 68 (54–80) |
Malignant | 74 | 23 (31) | 51 (69) | 68 (62–78) |
Lung cancer |
|
|
|
|
Breast cancer |
|
|
|
|
Ovarian and fallopian cancers |
|
|
|
|
Other malignancies |
|
|
|
|
Cancer of unknown primary |
|
|
|
|
Malignant mesothelioma | 89 | 79 (89) | 10 (11) | 70 (63–78) |
|
||||
Sera | Number of patients | Male (%) | Female (%) | Age, median (IQR) |
|
||||
Benign | 66 | 52 (79) | 14 (21) | 59 (48–71) |
Benign asbestos pleuritis |
|
|
|
|
Malignant | 74 | 44 (59) | 30 (41) | 61 (54–69) |
Lung cancer |
|
|
|
|
Breast cancer |
|
|
|
|
Ovarian and fallopian cancers |
|
|
|
|
Other malignancies |
|
|
|
|
Cancer of unknown primary |
|
|
|
|
Malignant mesothelioma | 91 | 34 (37) | 57 (63) | 65 (56–69) |
Epithelioid |
|
|
|
|
Biphasic |
|
|
|
|
Sarcomatoid |
|
|
|
|
Undetermined |
|
|
|
|
Age (IQR: interquartile range) and patient subgrouping in the two analysed materials. The high proportion of female mesothelioma patients in the serum material is most likely due to environmental asbestos and erionite exposure, which relates to geographical distribution and has also been reported by others [
Inclusion criteria have been previously described [
Metastatic pleural disease correlates with an advanced stage, while information of stage for malignant mesothelioma patients was not available for this study. For this reason we measured vascular endothelial growth factor (VEGF) in a subset of pleural effusions from patients with malignant mesothelioma (
In order to evaluate the relation between cell-bound and shed syndecan-1, a separate cohort of additional 18 pleural effusions, containing well-preserved malignant cells, was collected from 2012 to 2013 at Karolinska University Hospital in Huddinge. These samples were compared simultaneously for their pleural effusion concentrations of soluble syndecan-1 and their expression of membrane-bound syndecan-1 on tumour cells.
Syndecan-1 and VEGF were measured using ELISA: Human CD138 (syndecan-1) from Gen-Probe Diaclone, France (cat. number 950.640.192) and Human VEGF Quantikine ELISA from R&D Systems, UK (cat. number DVE00), respectively. ELISAs were performed according to the manufacturer’s instructions. Effusions were diluted 1 : 3 for syndecan-1 analysis and 1 : 5, 1 : 10, or 1 : 25 for VEGF analysis using kit-dilution buffers as blanks. Optical densities were determined using a spectrophotometer (BioTek’s PowerWave HT, Winooski, VT, USA) at 450 nm. Patient samples were analysed in duplicates by investigators blind to patients’ diagnoses and survival times.
To investigate the relationship between soluble and cell-bound syndecan-1, the latter was assessed by immunocytochemistry on tumour cells from the pleural effusion paired with their ELISA readout of shed syndecan-1 levels in the corresponding effusion supernatant. For immunocytochemistry, the pleural effusions were centrifuged for 10 min at 8,000 g and if necessary, erythrocytes were lysed using ammonium chloride (BD Pharm Lyse, BD Biosciences, CA, USA). Cells were immobilized on SuperFrost Plus Slides (Thermo Fisher Scientific Inc., Waltham, MA, USA), using cytospin preparations. Immunostaining was performed using a Lecia BOND-III automated IHC. Epitope retrieval was done by pretreating the slides in a citrate buffer, pH 6.0 (Bond Epitope Retrieval Solution 1, Leica Microsystems GmbH). Endogenous peroxidase activity was blocked with 3% H2O2 followed by incubation with syndecan-1 primary antibody (CD138, clone MI15, diluted 1 : 100, IgG1, DakoCytomation, CA, USA). Bound antibodies were demonstrated with the Bond Polymer Refined Detection kit (Leica DS 9800), and the cells were counterstained with haematoxylin.
Two experienced cytopathologists (KD and AH) evaluated all slides independently and were blinded to clinical diagnosis and levels of soluble syndecan-1. Cell-bound syndecan-1 expression was assessed by semiquantitative scoring which includes (i) the percentage of syndecan-1 positive tumour cells (0–100%) and (ii) the signal intensity (4-point scale). The scoring for signal intensity corresponded to 0, negative; 1, weak staining; 2, moderate positive; and 3, strong positive staining. The semiquantitative immunocytochemical (ICC) score for the cell-bound syndecan-1 expression level was provided by the multiplication of the percentage (0–100%) of syndecan-1 positive staining by the factor (1–4) corresponding to the staining intensity of the tumour cells.
Levels of soluble syndecan-1 and osteopontin were compared between patients with cancer and those without, using the Mann-Whitney test calculating two-tailed exact
Logistic regression was used to create a predictive model for each biomarker, with cancer or without cancer as outcome, coded as 1 and 0, respectively. Univariate odds ratios were calculated. Model calibration is recommended by several studies [
ROC plots with areas under the curves (AUC) and their 95% confidence intervals were made with the GraphPad Prism software. The calculations of sensitivity, specificity, positive predictive value, and negative predictive value were calculated with cutoffs from maximum sensitivity × specificity from ROC curves.
Discrimination slopes are recommended by several studies [
Survival analysis was performed with cutoff values based on the highest and most significant hazard ratio [
Spearman correlation analysis was used to assess the relationship between soluble syndecan-1 in effusions and corresponding cell-bound syndecan-1 in patients suffering from various pleura malignancies. Analyses were performed and graphs were created using the GraphPad Prism software.
This study was approved by the ethical review board of Stockholm, Sweden (2009/1138-31/3), and the ethical review board of Eskisehir University, Turkey. All patients had given informed consent.
Syndecan-1 and osteopontin levels were both significantly higher in malignant pleural effusions than in those with benign conditions, the difference being more pronounced with syndecan-1. However, neither soluble syndecan-1 nor osteopontin differentiated patient groups in sera (Figure
Levels of syndecan-1 and osteopontin are higher in pleural effusions from malignant tumours than from benign conditions. Syndecan-1 and osteopontin expression levels in benign and malignant pleural effusions, as measured by ELISA.
Soluble syndecan-1 levels in various malignant tumours. Significantly elevated soluble syndecan-1 levels were measured in several malignant tumours compared to benign disease.
Syndecan-1 levels in pleural effusions significantly predicted malignant disease (odds ratio 8.59, 95% CI 3.67 to 20.09). Area under the ROC curve was 0.71 (95% CI 0.65, 0.78), model calibration was acceptable (Nagelkerke’s
Cut-offs were derived from the maximum sensitivity × specificity in respective receiver operator characteristic curve.
Biomarker | Fluid | Cut-off (ng/mL) | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) |
---|---|---|---|---|---|---|
Syndecan-1 | Pl. effs | 65.7 | 74.9 | 61.3 | 0.77 | 0.57 |
Syndecan-1 | Sera | 151.6 | 46.1 | 59.1 | 0.70 | 0.27 |
Osteopontin | Pl. effs | 2034.0 | 65.5 | 61.1 | 0.60 | 0.65 |
Osteopontin | Sera | 217.3 | 52.7 | 60.6 | 0.77 | 0.34 |
Pl. effs: pleural effusions; PPV: positive predictive value; NPV: negative predictive value.
Soluble syndecan-1 has a diagnostic value for malignancy in pleural effusions. ROC plots for syndecan-1 in pleural effusions and serum (a) and osteopontin in pleural effusions and serum (b). Figures (c) to (f) only pertain to pleural effusions. Calibration curves evaluate the relation between the predicted and actual probabilities for soluble syndecan-1 and osteopontin, panels (c) and (d), respectively. Nagelkerke’s
Osteopontin in effusions also significantly predicted a malignant disease (odds ratio 1.39, 95% CI 1.03, 1.88). Area under the ROC curve was 0.65 (95% CI 0.57 to 0.73), the model calibration was poor (Nagelkerke’s
Cutoff values derived from highest and most significant hazard ratios for pleural effusions are reported in Supplementary Figure S2. Median survival time of patients with pleural metastases and an effusion syndecan-1 level <235.1 ng/mL was 12.9 months, compared to only 1.7 months in patients with higher syndecan-1 levels (hazard ratio 2.38, 95% CI 1.56 to 5.44). Stratifying malignant mesothelioma patients in “low” and “high” syndecan-1 level using a cutoff of 100.2 ng/mL resulted in median survival times of 17.0 and 7.8 months, respectively (hazard ratio 2.77, 95% CI 1.35 to 5.68; Figure
Both soluble syndecan-1 and osteopontin have prognostic roles for patients with malignant mesothelioma or metastatic pleural disease. Cutoffs for “high” and “low” syndecan-1 expression were identified by the online web application Cutoff Finder [
Among mesothelioma patients with serum syndecan-1 levels higher than the median (144 ng/mL), the median survival time was 9.0 months, while in those with lower serum levels, it was 11.0 months; however, this difference was not statistically significant (hazard ratio 1.43, 95% CI 0.63 to 3.98; Figure
To assess the relationship between cell-bound and soluble syndecan-1, we performed immunocytochemistry on cells in pleural effusion samples paired with their ELISA readout of soluble syndecan-1 levels (
Strong cellular immune-reactivity for syndecan-1 on malignant cells correlates with higher levels of soluble syndecan-1 in pleural effusions. Spearman correlation was used to test the goodness of fit between levels of soluble syndecan-1 and cell bound syndecan-1 intensity times the percentage of positive tumour cells (semiquantitative ICC score;
The importance of the syndecan proteoglycan family has been implicated in several aspects of cancer [
Stratifying the patients into disease subgroups, the highest levels were seen in effusions due to metastases from lung cancer, breast cancer, and gastric cancer, possibly reflecting their epithelial origin. Syndecan-1 levels were also higher in effusions from patients with mesothelioma. This suggests a more general pathophysiological role of syndecan-1 in cancers.
Effusion and serum levels of syndecan-1 correlated moderately, but the diagnostic value of syndecan-1 could not be demonstrated in sera. Here the differences between analysed patient groups are smaller, as is the dynamic range
We further show that soluble syndecan-1 in pleural effusions carry strong prognostic value for patients with pleural malignancies. Low levels of shed syndecan-1 predict a more favourable prognosis, showing more than 11 months increased median survival for patients with pleural metastases and 9.2 months for malignant mesothelioma; both are diseases with average median survival times of around, or less than, one year. This trend is also seen in the two main subgroups of metastatic tumours—that is, lung and breast adenocarcinoma—although both these subgroups are too small to show statistical significance. These findings are in concordance with earlier studies on multiple myeloma [
Structurally, the extracellular domain of syndecan-1 carries attachment sites for heparan sulfate and chondroitin sulfate chains. Since heparan sulfate chains act as binding sites for several ligands such as growth factors and chemokines, proteolytic shedding of the extracellular domain is crucial in regulating various signaling pathways [
In our cohorts, higher levels of soluble syndecan-1 corresponded to stronger immunocytochemical staining of syndecan-1. It seems that the soluble fraction increased proportionally to the expression of syndecan-1 on these cells. We cannot, however, exclude a concomitant increase in protease activity and syndecan-1 ectodomain shedding in the same cells. Such shedding has been associated with cancer progression, risk for recurrence, and prognosis [
Several studies suggest elevated pretreatment soluble serum syndecan-1 level as a predictor of poor prognosis and impaired effect of chemotherapy [
In this study we show a possible diagnostic and prognostic role for soluble syndecan-1 in effusion cytology. The use of this proteoglycan as an effusion biomarker could be helpful in the evaluation of effusions. Similar to most established biomarkers—immunological or immunocytochemical—the discriminatory power of this analysis is insufficient to use as a sole diagnostic marker in the individual case. The combination of this parameter with other biomarkers in logistic models remains to be studied. This, however, should then warrant further studies on the mechanisms behind the release of syndecan-1 as well as possible predictive effects and effects on epithelial mesenchymal transition in tumours.
In summary, this study describes the clinical value of analysing soluble syndecan-1 in serum and pleural effusions. Syndecan-1 separates malignant and benign conditions when measured in pleural effusion supernatants. Furthermore, we show a striking correlation between syndecan-1 levels and patients’ survival, which is of interest for future predictive translational research. Hence, the inclusion of measuring syndecan-1 could be a valuable clinical tool, whether on its own or in a panel of soluble biomarkers.
The authors declare that there is no conflict of interests regarding the publication of this paper.
This work received financial support from AFA Insurance, the Swedish Research Council (VR), the Swedish Society of Medicine, and the Swedish Cancer Society. AFA is a nonprofit insurance company that is jointly owned by the Swedish Labour Organization and the Swedish Employers Organization.