The incidence and mortality of lung cancer remain obstinately high with the environment deterioration and the population aging [1, 2]. Malignant pleural effusion (MPE), a common complication of advanced lung cancer, severely impedes respiration and circulation, affecting patients’ life quality [3] with greatly shortened survival period [4]. Currently, diagnosis accuracy and treatment efficacy for MPE are still to be improved [5], and individual tumor markers such as carcinoembryonic antigen (CEA) and cytokeratin 19 fragments (CYFRA 21–1) in making diagnosis is also unsatisfactory [6, 7]. Therefore, further study on MPE is necessary for reliable diagnostic methods.

Secreted phosphoprotein 1 (SPP1), also known as osteopontin, is a secreted arginine-glycine-aspartic acid (RGD)-containing phosphorylated glycoprotein with a molecular weight of about 325,000 [8]. It was first isolated as an extracellular matrix (ECM) protein from bone matrix [9]. The human SPP1 gene locates on chromosome 4 (4q13) with 7 exons and 6 introns [10]. The protein can be secreted by various cells, including osteoclasts, macrophages, epithelial cells, endothelial cells [11, 12]. In addition, tumor cells can also secrete soluble SPP1, which is different from the matrix protein [13].

The expression of SPP1 is strongly associated with tumorigenesis and tumor metastasis. It has been noted that the expression of SPP1 is high in numerous tumors, including lung cancer [14, 15]. Ongoing research has discovered that SPP1 in MPE due to lung cancer has numerous effects: it progresses the tumor both in its growing and metastasis, and prolongs the survival of cancer cells [16]. Moreover, SPP1 increases vascular permeability. Cui et al. found that SPP1 could increase pleural vascular permeability by inducing vascular endothelial growth factor (VEGF) expression [17]. Additional studies have confirmed that SPP1 has a non-VEGF-dependent effect on vascular permeability as well. Also, SPP1 promotes intrapleural cancer dissemination [18]. The invasion of ECM by tumor cells may lead to conditions favorable for metastasis [19]. In an in vitro study, Hu et al. found that SPP1 promotes invasion of ECM by NSCLC cells, and its effect may be blocked by SPP1 antibodies [20]. Therefore, SPP1 is essential in the formation and development of MPE.

To the best of our knowledge, however, there have not been any studies on the relationship between SPP1 in MPE and the diagnosis and prognosis of lung cancer. So this study is performed to investigate whether the expression of SPP1 in MPE associated with NSCLC are useful for NSCLC in making diagnosis and prognosis.

In this study, the concentration of SPP1 in pleural effusion samples from 109 patients was determined by ELISA. SPP1 concentrations were significantly higher in the MPE patients than those in the control patients. Based on the cutoff value, the higher the SPP1 levels in pleural effusion specimens were, the lower the false diagnosis rate of MPE would be. In addition, SPP1 levels in MPE specimens were positively correlated with the number of extrapulmonary metastases, and the prognosis of patients was significantly better in the low-SPP1-expressing patients than in the high-expressing patients. SPP1 in MPE was an independent prognostic factor of NSCLC-induced MPE, indicating that SPP1 in pleural effusion can be used for the auxiliary diagnosis of NSCLC-induced MPE, and it may be a prognostic indicator for patient survival and risk of extrapulmonary metastases.

Few investigators have studied the diagnostic and prognostic value of SPP1 in MPE associated with lung cancer, but data on SPP1 on cancer progression is increasing in recent years. Neal et al. found that patients with colon cancer, breast cancer, prostate cancer, and lung cancer had increased SPP1 levels in serum compared with healthy controls; it was almost 2 times higher in patients with lung cancers. Schneider et al. determined the expression levels of SPP1 and osteonectin (ON) in lung cancer tissues using quantitative reverse-transcription polymerase chain reaction and found that it was the increased expression of SPP1 rather than ON was associated with poor prognosis [21]. By testing SPP1 levels in stage I-IIIA NSCLC tissues using immunohistochemistry, and follow-ups for patient outcomes, Italian investigators found that Stage I patients with high expression of SPP1 had poorer prognosis than those with low expression of SPP1 [22]. Mack et al. assessed plasma SPP1 levels of 56 patients with advanced NSCLC who received paclitaxel + carboplatin chemotherapy in the clinical SWOG Study S0003 and continuously monitored plasma SPP1 levels and the treatment response of patients during cycles of chemotherapy. They found that plasma SPP1 levels were significantly associated with treatment response, PFS and OS. Patients with low plasma SPP1 levels had better OS and PFS than those with high levels [23]. Our results are in consistent with these previous studies; however, those studies did not analyze the association between SPP1 level and different clinical features, they and investigated samples of tumor tissue or blood instead of pleural effusion.

Some of the findings on SPP1 concentrations in MPE have been studied. Morsi et al. determined the SPP1 concentrations of serum and pleural effusion from 80 patients (20 with transudative effusion; 30 with MPE, including 14 patients with NSCLC; 30 with tuberculous effusion) using ELISA. Their findings suggested that there was minimal difference between the serum SPP1 levels and effusion/serum SPP1 ratios of the patients with MPE and those with tuberculous effusion [24]. However, Moschos et al. measured serum and pleural effusion levels of SPP1 from 109 patients (42 with MPE, including 19 patients with NSCLC; and 67 with benign diseases) using ELISA, and found that patients with MPE had higher pleural effusion levels and lower serum levels of SPP1 than those with benign disease [25]. Our study showed that SPP1 levels in MPE were significantly higher in the study group than those of the controls. We hypothesize that the findings of Morsi et al. might be attributed to the heterogeneity in malignant effusion as well as less patients were with MPE. The tumor burden and metastases of the sort of group was smaller than the remaining cases. The rest 12 cases are not lung cancer patients, which could lead to larger secretory differentiation and less patients with MPE.

Although novel technologies are continually being introduced, pleural fluid analysis is still inadequate for the diagnosis of MPE. The diagnostic characteristics of single tumor marker in pleural effusion has been reported as follows: CEA (sensitivity 45.9%; specificity 97.0%), CA125 (sensitivity 48%; specificity 85%), CA15-3 (sensitivity 51%; specificity 96%), CA19-9 (sensitivity 25%; specificity 96%) and CYFRA21-1 (sensitivity 55%; specificity 91%) [6, 7]. In this study, we have shown that SPP1 in pleural effusion has diagnostic characteristics for MPE similar to those of CEA in pleural fluid.

In addition, our results suggested that the level of SPP1 in MPE was an independent prognostic factor for NSCLC. Both PFS and OS were significantly better in the low-SPP1-expressing patients than in the high-expressing patients. Chi-square analysis showed that SPP1 levels were positively correlated with the number of extrapulmonary metastases, indicating that patients with high levels of SPP1 expression had higher risk of extrapulmonary metastasis. This finding is consistent with the results from a previous study showing that SPP1 can promote tumor growth and metastasis [11]. As previously mentioned, high SPP1 concentrations can inhibit apoptosis of pleural tumor cells and promote intrapleural tumor dissemination. We also conjecture that high levels of SPP1 may prompts tumor cells invasion of the ECM into parietal pleura, proceeding to intravasation of tumor cells into the blood stream and subsequent extravasation assisted with urokinase plasminogen activator (uPA) and matrix metalloproteases (MMPs), leading to migration of neoplastic cells to form metastasis at secondary sites. In conclusion, high concentrations of SPP1 in pleural fluid (especially ≥ 1247.9 ng/ml) are more dangerous and aggressive.

In this study, we evaluated the value of SPP1 as an indicator for the auxiliary diagnosis of MPE and for the prognosis of patients with NSCLC. The cutoff value determined in this study is specific but not sensitive enough, which might have resulted in false-negative results in patients with low SPP1 levels. The reasons for low sensitivity may be the following: a) Interindividual differences may be enhanced when the sample amount is small; b) SPP1 is also related to the formation of lung granulomas because of its role in cellular immunity [26], as research has shown that circulating SPP1 levels in patients with pulmonary tuberculosis were significantly higher than in healthy controls, while the levels might reflect the activity of tuberculosis [27]; and c) SPP1 levels in exudative pleural effusion (e.g., tuberculous pleural effusion) were significantly higher than those in transudative hydrothorax [24, 25]. There is a high prevalence of tuberculous pleurisy in China, and patients with tuberculous pleurisy in our study accounted for as many as 75% of together with the control patients, while patients with transudative hydrothorax only accounted for 8.3%. The control group was composed of patients whose disease were not well-controlled, either inflammation originated from pulmonary or granulomatous diseases could be disseminated to pleural space in non-malignant patients and then followed the pleural fluid from the histopathological lesion. This may have accounted for a high level of SPP1 in the control patients, and the value for the controls overlapped the value for the study patients with low expression levels of SPP1. This would lead to reduced sensitivity of the cutoff value.

Our study has inevitable limitations because of various cancers that induce MPE are dramatically different in cellular morphology and biological behaviors. Therefore, further studies with larger sample sizes and optimized pathological types of cancer are needed to verify our findings. We could not measure the corresponding serum samples because the serum samples were not adequate to match all of pleural fluid in the study. This needs valid research in future studies.

In summary, our study suggests that quantitative determination of SPP1 concentrations in pleural effusions may provide an auxiliary diagnostic modality for differentiating between benign and malignant pleural effusion. An increased level (≥1247.90 ng/ml) in MPE may be an indicator of poor prognosis and risky for distant metastasis. This approach is simple and feasible. Further study is needed because the pathogenesis of MPE is complicated and the pathological mechanisms of association between SPP1 and MPE remain unclear.