The aim of the present study was to analyze the prospect of cyst fluid as a source for biomarker discovery. We detected numerous differently expressed protein peaks in our samples, showing that proteomic analysis of ovarian cyst fluids may serve as a valuable tool in the ongoing quest for new ovarian cancer biomarkers. Biomarkers produced by the tumor are most likely to be found close to the tumor in the initial phase, before excretion into the blood . The existence of a protein in cyst fluid can, however, also depend on diffusion from the blood stream. The peaks with highest ROC AUC separating benign from malignant ovarian cyst corresponded to proteins that are also present in blood . Even if some of these proteins are not directly expressed by the tumors, they may reflect reactions that are closely related to carcinogenesis, such as increased thrombotic activity and inflammation, and may thus still be candidates as markers for tumor detection.
MS techniques allow rapid and high throughput analysis of the entire proteome in a small biological sample. One disadvantage with large scale screening methods using serum samples is the dominance of highly abundant proteins. Ninety-five percent of the serum consists of only 20 abundant proteins i.e. albumin, immunoglobulins, transferrin and macroglobulins . By using ovarian cystic fluid we made the assumption that we would increase the chance for detection of rare, more tumor-specific proteins. However, there seem to be a high resemblance between ovarian cyst fluid and the serum proteome since the abundant large proteins were the most prominent ones also in the present study. Similar resemblances between serum and fluids from other compartments such as pancreatic cysts, ascites and breast duct fluids have also been described [8, 11, 12]. Use of careful and selective depletion of high abundance proteins from the ovarian cyst fluid might increase access to other specific biomarkers. One major limitation with the method used in the present study is the lack of exact protein identification of the mass-peaks. Fortunately the abundant peaks observed as differentially expressed between benign and malignant ovarian cystic fluid, have been purified and confidently identified in several earlier studies [19–27]. Especially the peaks for ApoC-I and TTR have been correctly identified after several different preanalytic conditions . To further validate ovarian cyst fluids as a source for biomarker discovery of rare but specific proteins, depletion must be performed and a MS with accurate protein identification i.e isobaric tags for relative and absolute quantification (iTRAQ) could be used.
We compared the protein peak profiles in cysts from benign and malignant ovarian tumors and detected peaks corresponding to five potential biomarkers. In combination with serum CA125, two mass peaks were found to be predictors of malignancy (ROC AUC 0.94). The first peak with the suggested protein, ApoC-III, is a very low-density lipoprotein, a carrier protein for diverse proteins, mainly synthesized in the liver and small intestines. ApoC-III inhibits lipoprotein lipase and hepatic lipase and plays an important role in regulating the triglyceride rich lipoproteins and thereof regulates many cellular functions. Our study is the first to describe increased ApoC-III in ovarian cyst fluids. In the analysis of an eleven protein panel in serum from patients with ovarian cancer, increased Apo-CIII was one of the markers that could distinguish women with cancer from those with benign conditions . The second peak with the suggested protein, PCI (SerpinA5), is a component of the anticoagulant protein C pathway, originally identified in human plasma as an inhibitor of activated protein C. PCI also inhibits urinary plasminogen activator (uPA), which is a mediatior of tumor cell invasion . PCI is expressed in various human fluids and tissues but has not previously been described in ovarian cyst fluids. Gene and protein expression analysis revealed SERPINA5, the gene encoding PCI, as down-regulated in serous ovarian carcinomas compared with serous borderline tumors, which is consistent with our MS results [30, 31]. SERPINA5 expression is decreased in renal and prostate cancer and elevated levels have been related to better survival in breast cancer [29, 32, 33]. However, although we found a significant decrease of PCI levels in the malignant compared to the benign cohort using SELDI-TOF MS we could not confirm our MS results with an independent method, ELISA. This suggests that the peak we identified as PCI were incorrect identified. However, due to the various functions and versatility of PCI it is challenging to investigate PCI protein expression in body fluids . In blood PCI is found in various forms including inactive, cleaved, and complex-bound. PCI is also modified through post-translational events and has been reported as microheterogeneous as a result of differences in N-glycan structures .
The other three peaks with ROC AUC > 70 identified with mass matching from the literature in this study was TTR, SAA4 and ApoC-I. TTR formerly called prealbumin, has previously been identified as a potential early diagnostic marker for ovarian cancer and known as a marker for acute-phase inflammatory response and nutritional status [22, 24]. A study of ovarian tumors showed a direct correlation between serum levels, intracellular expression of SAA4 and malignant state of the epithelial cells . In colon cancer SAA4 levels increases gradually from epithelial dysplasia to cancer in tissue samples and are highest in metastatic cancer, which supports the possibility of its role in carcinogenesis . In our study, SAA4 was up-regulated in ovarian cyst fluid from patients with malignant tumors, which is consistent with validated SELDI-TOF MS in serum from ovarian cancer patients . The peak for ApoC-I was also increased in the malignant ovarian cysts in our study. Takano et al.  have shown that ApoC-I is abundantly expressed in pancreatic neoplastic epithelium. It was also detected in the culture medium of a pancreatic cancer cell line, which suggests that cancer cells secrete ApoC-I. The elevated levels of ApoC-I in our analysis may also depend on secretion of the protein from tumor cells.
The ultimate goal is to find superior biomarkers to those already known to detect early EOC. Our group and others evaluated CA125 and HE4 in blood from patients with pelvic tumors. In the study by Moore et al. the ROC AUC for CA125 was 0.83 compared to our 0.87 and in combination CA125 with HE4 0.91 and 0.85 respectively [3, 4]. Bearing in mind that the selected peaks in this study has not been analyzed for exact peptide sequence we still choose to perform a comparative estimation of ROC AUC between the benign and malignant samples. None of the protein peaks detected in our study showed a better specificity than CA125 when considered in isolation. However, a combination of the peaks for ApoC-III, PCI, and CA125 improved the ROC AUC significantly. The currently presented marker combination of serum CA125 and two cystic fluid mass-peaks (ROC AUC 0.94) was comparable to the above-mentioned algorithms [3, 4]. Despite the drawbacks of the method used in our present study, our results strengthen our hypothesis that it is possible to find new biomarkers in the ovarian cystic fluid that can diagnose early ovarian cancer.