Methylated genes are explored as circulating diagnostic biomarkers of OvCa. Ibanez et al. screened ovarian tumors and matched preoperative sera/plasma and peritoneal fluids for RASSF1A and BRAC1 methylation status [7]. Hypermethylation of one or both genes was demonstrated in 68 % of tumors. They then expanded coverage by including the following genes: APC, CDKN2A, DAPK, and CDKN2A/ARF, which enabled the remaining tumors to be identified. Identical methylation pattern was present in 82 % of serum/plasma samples (including 76.5 % of stage I disease samples). Peritoneal fluid was positive in 93.3 % of the samples, including samples from patients with tumors that were scored conventionally as atypia or negative by cytology. Microarray-based screening of 58 genes, and using the most differentially methylated genes in naïve Bayesian analysis enabled 10 genes to be identified for further study. The sensitivity and specificity in tissue samples were 69 % and 70 %, respectively. A 5-gene panel analyzed in plasma samples achieved a sensitivity of 85 % and specificity of 61 %. This assay demonstrates diagnostic potential [8]. In a follow-up study, Levenson’s group expanded their proof-of-principle study in an attempt to differentiate women with healthy ovaries from those with OvCa and benign ovarian diseases, and to possibly differentiate benign ovarian diseases from cancer using methylation of a number of gene sets in ccfDNA. Methylation of RASSF1A, CALCA, and EP300 differentiated cancer from healthy controls at a sensitivity of 90% and specificity of 86.7 %. Similarly, methylation of BRAC1, CALCA, and CDKN1C differentiated women with benign ovarian disease from healthy controls at a sensitivity of 90 % and specificity of 76.7 %. Finally, RASSF1A and PGR-PROX methylation in ccfDNA differentiated cancer from benign ovarian disease [9]. Bondurant et al. used a sensitive assay that enabled detection of 1 methylated among 100,000 unmethylated alleles [10]. Using this assay, RASSF1A was methylated in 51 % of invasive serous OvCa samples, and was concordant in all available matched preoperative serum samples. Serial serum sampling enabled therapy response monitoring that was detectable as fluctuations in RASSF1A methylation in some patients, reflective of disease status.

In addition to tumor grade and histologic cell type, the cytologic examination of peritoneal fluid is part of the staging strategies of OvCa. Women with positive peritoneal fluid cytology have poor prognosis regardless of FIGO stage. The inherent issues with cytology call for better prognostic biomarkers or complementary assays to cytology. As a proof-of-principle study to identify sensitive prognostic methylation biomarkers, peritoneal fluids were collected at surgery from women with OvCa, who also received platinum-based chemotherapy after surgery. Fifteen genes were selected for study. Women with fewer methylated genes had poor or shorter overall survival evidenced by univariate and multivariate analysis, independent of age, FIGO stage, or grade. Thus, methylation in genes from this series may confer sensitivity to platinum-based chemotherapy [11].

MiRNA deregulation is associated with OvCa. In tissues, and also in body fluids, the aberrant expression of let-7 and miR-200 family members appears to be associated with various types of OvCa. Several miRNAs, possibly with tumor suppressor functions, are downregulated in OvCa tissues and cell lines. These include let-7d, miR-125b-1, miR-127, miR-140, miR-145, and miR-199a. Specifically, in serous OvCa, let-7b, miR-10b, miR-26a, miR-29a, miR-99a, miR-100, miR-125a, miR-125b, miR-143, miR-145, miR-199a, and miR-214 are downregulated. Similarly, a number of possible oncomirs are upregulated in OvCa samples, and some are subtype associated. OvCa subtype-specific upregulated miRNAs include miR-30a-5p and miR-30a-3p in clear cell OvCa and miR-192 and miR-194 in mucinous OvCa, while serous OvCa is associated with increases in miR-16, miR-20a, miR-21, miR-23a, miR-23b, miR-27, miR-93, miR-141, miR-200a, miR-200b, and miR-200c. Some miRNAs deregulated in epithelial OvCa tissue samples also appear to correlate with various prognostic parameters such as DFS, PFS, Recurrence-FS, and OS. Those associated with good prognosis include miR-100, miR-150, miR-187, miR-200a, miR-200c, miR-335, miR-410, miR-645, and the ratio of miR-221/miR-222, while deregulated miR-21, miR-25, miR-29, miR-203, and miR-221 confer poor prognosis.

Several OvCa-associated miRNAs have defined functions in tumor initiation and progression and serve as therapeutic targets as well. Some miRNAs are implicated in EMT, cancer stem cell maintenance, angiogenesis, and extracellular matrix remodeling in OvCa. As examples, the tumor suppressor miR-138 represses genes involved in EMT and hence is downregulated in metastatic OvCa cells. Similarly, miR-125a inhibits EMT through AT-rich interactive domain 3B (ARID3B). EGFR signaling through the ETS family of transcription factor (PEA3) represses miR-125a expression in metastatic OvCa. MiR-200a and miR-200c are also downregulated in CD117-positive/CD44-positive OvCa stem cells compared to CD117-negative/CD44-negative cells. Increased expression of these miRNAs decreased ZEB1 and vimentin levels in association with E-cadherin expression, which impaired migration, invasion, and colony formation. MiR-125b, miR-145, and miR-199a target HIF-1α and VEGF to suppress tumor angiogenesis, and all are deregulated in OvCa. Finally, miR-335 is downregulated in OvCa, because it inhibits metastasis by targeting the glycoprotein tenascin C (TNC) that is a negative regulator of cell–extracellular matrix interaction.

Circulating miRNA deregulation with diagnostic and prognostic potential are demonstrated in samples from OvCa patients. While mostly not validated, their potential role in the clinic is expected and should make a difference in the diagnosis and management of this disease.

In 2008, Taylor et al. first reported on differential levels of exosomal miRNA between women with OvCa and those with benign ovarian adenomas [16]. They selected eight known OvCa-associated miRNAs (miR-21, miR-141, miR-200a, miR-200b, miR-200c, miR-203, miR-205, and miR-214) and demonstrated that these were significantly enriched in exosomes from cancer patients compared to controls. Of interest, these circulating exosomal miRNA levels were not correlated with tumor stage or grade, suggesting their potential as early OvCa detection biomarkers. The elevated levels of the miR-200 family members (miR-200a, miR-200b, and miR-200c) in sera from women with serous OvCa were confirmed. Resnick et al. reported a similar study, whereby 21 differentially expressed miRNAs in OvCa tissue samples were selected a priori, and tested on serum samples from patients and controls [17]. Of the 21 miRNAs, five (miR21, miR-29a, miR-92, miR-93, and miR-126) were elevated, while the levels of three (miR-99a, miR-127, and miR-155) were decreased in sera from women with OvCa compared to healthy controls. Using whole blood samples from patients and controls, eight miRNAs (miR-16, miR-29a, miR-30c-1, miR-106b, miR-155, miR-146a, miR-191, and miR-383) were demonstrated to be deregulated in cancer patients, and this finding was consistent with those in OvCa tissue samples. miR-30c-1-3p was upregulated, while downregulated were miR-181a-3p, miR-342-3p, and miR-450b-5p [18]. A microarray profiling of sera, ascites, and tissue samples from women with serous OvCa uncovered four miRNAs (let-7b, miR26a, miR-132, and miR-145) that were significantly decreased in sera from patients compared to controls [19]. Plasma let-7f and miR-205 have also been identified as early detection biomarkers (stage I) of OvCa [20].

Endometriosis is a precursor lesion to some types of OvCa. The ability to differentiate endometriosis from OvCa is clinically relevant. The work by Suryawanski et al. sheds some light on the potential of miRNA in this regard [21]. Global profiling of circulating miRNA in women with endometriosis-associated OvCa enabled the identification of miR-16, miR-191, and miR-195 to be elevated in women with endometriosis. This miRNA signature had a detection sensitivity of 88 % and specificity of 60 %. Three other miRNAs (miR-16, miR-21, and miR-191) were able to differentiate women with endometriosis-associated OvCa from healthy women at a sensitivity of 86 % and specificity of 85 %. Importantly, miR-21, miR-362-5p, and miR-1274a performed at a sensitivity of 57 % and a specificity of 91 % in separating between women with endometriosis and those with endometriosis-associated OvCa. Moreover, miR-21, miR-191, and miR-1975 could stratify women with sOvCa and endometriosis-associated OvCa at a sensitivity of 86 % and specificity of 79 %. A separate set of miRNAs (miR-362-5p, miR-628-3p, and miR-1915) performed at a sensitivity of 90 % and specificity of 73 % in differentiating women with endometriosis from those with sOvCa. Finally, for separating between women with sOvCa and healthy controls, miR-16, miR-191, and miR-4284 achieved a sensitivity and specificity of 90 % and 55 %, respectively.

Several circulating miRNAs that are deregulated in OvCa have been associated with clinical outcomes. Decreases in let-7f levels correlate with poor prognosis, and elevated miR-221 in sera from women with eOvCa is associated with FIGO stage and tumor grade. In multivariate analysis, high circulating miR-221 was an independent predictor of poor prognosis [22]. Circulating levels of miR-92 correlate with lymph node metastasis and clinical stage of OvCa, while advanced FIGO stage, high tumor grade, and poor OS were associated with elevated miR-21 [23]. Five preoperative plasma miRNAs significantly predicted shorter OS. However, after multiple testing, only miR-1290 emerged as a robust prognostic biomarker, with AUROCC of 0.87, p = 0.05 [24].

Several discovery studies have explored the use of serum m/z spectral peak signatures for OvCa detection. The seminal work of Petricoin et al. is noteworthy [25]. This group used serum proteomic SELDI TOF MS spectra applied to an iterative search algorithm to identify a discriminatory pattern for cancer detection. This “cluster -proteomic -pattern” biomarker achieved a sensitivity of 100 %, a specificity of 95 %, and a PPV of 94 % for OvCa. Since then, several groups have applied proteomic spectral patterns as cancer biomarkers.

Four peptide peaks at m/z 6195, 6311, 6366, and 11498 achieved a sensitivity of 87 % and a specificity of 95 % in a validation sample set. A peak at m/z 4475 only appeared after chemotherapy [26]. Lin et al. observed four peaks (m/z 5147.06, 6190.48, 11522.6, 11537.7 Da) in plasma from cancer patients but not controls, and 2 peaks (5295.5, 8780.48) in controls but not in cancer patients [27]. The sensitivity and specificity of these peaks were 96.3 % and 100 % for OvCa. Wang et al. also observed seven discriminatory peaks at m/z 3940, 4099, 4144, 4479, and 5488 (upregulated) and 8588 and 13783 (downregulated) in advanced stage OvCa that had sensitivity of 93.94 % and specificity of 93.55 % [28]. These biomarkers also detected 90 % of early stage OvCas. Of 98 peaks that discriminated malignant from benign tumors, 46 were finally identified from average linkage clustering to be more relevant [29]. Similarly, in women with normal CA125 levels, 25 peaks discriminated malignant from the benign group [30]. Mass spectrometric peaks of sera could differentiate stage I/II disease from controls at a sensitivity of 80% with AUROCC of 0.82, and stage III/IV at a sensitivity of 93 % and AUROCC of 0.92. Note that the majority of these studies involved small samples sizes and hence are only proof-of-concept discovery findings.

Proteomics approaches have enabled the identification of several proteins and peptides with differential levels in circulation of women with OvCa. Among these numerous proteins are haptoglobin, apolipoprotein A1, transthyretin, amyloid AI, transferrin, β-hemoglobin, and β2-microglobulin that show potential as OvCa biomarkers.

Serum amyloid A1 and hemoglobin are among the early protein biomarkers identified by proteomic studies for OvCa. Moshkovskii et al. observed invariant peaks at m/z 11.7 and 11.5 kDa that were present in 55.6 % of OvCa but at low intensities in only 5.8 % of controls [31]. These peaks were identified as amyloid A1 (11.68 kDa) and its N-terminal arginine-truncated form (11.52 kDa). Woong-Shick et al. also identified two peaks with significant differential expression between cancer and controls as hemoglobin-α (15.1 kDa) and hemoglobin-β (15.8 kDa) chains [32]. The sensitivity for intact hemoglobin was 77 % in sera from cancer compared to controls.

Several investigators have demonstrated the clinical performance of serum apolipoprotein A1 in OvCa. Kozak et al. used SELDI TOF MS to identify 14 differentially expressed peaks categorized into three protein panels designated as screening panel (five candidates), validation panel I (five candidates), and validation panel II (four candidates) [33]. Independently, they all performed very well in differentiating between benign and malignant ovarian neoplasia (AUROCC of ≥0.90). The three panels correctly classified 93 % of blinded samples comprising of various ovarian tumors and normal patient samples. In a follow-up study, 5 of the original 14 peaks were identified as transthyretin (TT, 2 peaks with m/z 13.9 kDa and 12.9 kDa), β-hemoglobin (Hb, m/z 15.9 kDa), apolipoprotein A1 (Apo-A1, m/z 28 kDa), and transferrin (TF, m/z 79 kDa) [34]. These biomarkers improved early-stage OvCa detection with AUROCC of 0.933 (compared to 0.833 for CA125). However, CA125 incorporation increased the performance to 0.959. For mucinous type OvCa, the five biomarkers were more discriminatory than CA125 alone (AUROCC of 0.959 vs. 0.613), and CA125 incorporation failed to improve detection of mucinous type OvCa. Nosov et al. extended this study to include the detection of serous and endometrioid type OvCa by analysis of Apo-A1, TT, TF, and CA125 levels [35]. This study, which included 358 serum samples from patients with benign adnexal masses, early and late stage OvCa, as well as healthy controls achieved a sensitivity of 96 % for OvCa detection (a sensitivity of 98 % for detection of early-stage endometrioid type cancer). A multicenter case–control study to validate serum biomarkers focused on 3 proteins (Apo-A1, TT, and cleavage fragment of inter-alpha trypsin inhibitor heavy chain 4) for which immunoassays were available [36]. Independent cross-validation for detection of early-stage OvCa was undertaken. These biomarkers improved detection of OvCa by CA125. The 4-biomarker combination had a sensitivity of 74 % at a fixed specificity of 97 %. Clarke et al. examined a panel of seven markers (apolipoprotein A1, truncated transthyretin, transferrin, hepcidin, b-2-microglobulin, connective tissue activating protein III, and inter-alpha-trypsin inhibitor heavy-chain 4) for OvCa detection [37]. However, in a training set, only three of these (Apo-A1, TT, and CTAPIII) performed the best with a sensitivity of 54 % at a specificity of 98 %, while CA125 in this series had a sensitivity of 68 %, and the combination increased the sensitivity to 88 %. In a validation assay, 84 % sensitivity was achieved at the set specificity of 98 %. This seven panel biomarker, however, failed to improve sensitivity beyond CA125 alone for preclinical detection of OvCa [38].

Haptoglobin 1 (HAP1) is another potential serum biomarker for OvCa. The work by Ye et al. using SELDI TOF MS and LC MS/MS identified a peak differentially expressed at a significant level between cancer and controls as alpha chain of haptoglobin [39]. This alone performed at a sensitivity of 64 % and specificity of 90 %, and together with CA125, the panel had a high sensitivity of 91 % and specificity of 95 % for detection of OvCa. Six peaks significantly upregulated in all groups of OvCa patients were identified as isoforms of haptoglobin 1 precursor [40]. This group subsequently identified a downregulated protein, transferrin, as well as the previously upregulated HAP1 in grade 3 OvCas [41]. The changes were also detected in peritoneal fluids from patients. Haptoglobin levels decreased (transferrin remained unchanged) following six cycles of taxol/carboplatin chemotherapy.

Several other potential serum proteins are uncovered for OvCa. Havrilesky et al. evaluated a panel of 9 biomarkers (HE4, glycodelin, MMP7, SLPI, Plau-R, MUC1, Inhibin A, PAI-1, and CA125) for early OvCa detection and 4 biomarkers (HE4, Glycodelin, MMP7, and CA125) for recurrence monitoring of OvCa [42]. For OvCa detection, the highest sensitivity and specificity achieved with various biomarker combinations were 80.5 % and 96.5 % for early-stage disease and 89.2 % and 97.2 % for late-stage disease. Recurrence was predicted in 100 % of the women with recurrent disease (compared to 96 % for CA125 alone). At least one biomarker was elevated earlier (6–69 weeks) than CA125, and before clinical detection of recurrence in 52 % of the patients. Of 6 proteins identified in another study, 2 (CCL18 and CXCL1) were selected by multivariate predictive model and validated [43]. Immunoassays were developed and used to screen 535 serum specimens comprising of people with OvCa, benign pelvic masses, other non-gynecologic cancers, and healthy controls. As a panel, the two biomarkers had a sensitivity of 92 % and a specificity of 97 % for OvCa detection. In combination with CA125, the sensitivity improved to 99% with healthy women, and 94 % for women with benign pelvic masses as controls, at a specificity of 92 %. Sera collected several months before cancer diagnosis from 295 women and 585 controls were profiled by MALDI MS for early detection biomarkers [44]. Two peaks, CTAPIII and platelet factor 4 (PF4), in combination with CA125 were discriminatory between cancer and controls months before cancer diagnosis, and this was earlier than predictions by CA125 alone.

Other multi-analyte serum biomarkers have proven clinically useful for OvCa. Zhang et al. tested the utility of 4-biomarker panel consisting of CA125II, CA72-4, CA15-3, and M-CSF in an artificial neural network-derived composite index for detection of early-stage OvCa [45]. The panel performed much better than CA125II alone in the detection of early-stage OvCa with a sensitivity of 71 % at a set specificity of 98 %. Hogdall et al. similarly examined 7-marker panel for utility in triaging women with pelvic masses for subsequent evaluation [46]. In multivariate logistic regression analysis, CA125, age, and the 7-proteomic biomarkers had independent predictive abilities of OvCa. The combination of these metrics (the 7-marker panel, CA125, and age) constituted the Danish Index (DK-Index). For the detection of EOC, the DK-Index had a sensitivity of 95 % and a specificity of 81 %, which was much superior to CA125 alone. Phase II multi-analyte biomarker (CA125, CRP, serum amyloid A, IL-6, and IL-8) trial for OvCa diagnosis included sera from 150 cases and 212 controls, and was validated in an independent cohort of 183 women. The 5-panel biomarker performance was significantly greater than CA125 alone in the validation cohort, as well as for stage I/II disease. Sensitivity and specificity were 94.1 % and 91.3 % for the validation cohort, and 92.3 % and 91.3 % for early-stage disease [47]. In a subsequent large-scale study, the 5 serum biomarkers were tested on sera from 222 women with EOC, 223 with benign disease, 53 with borderline OvCa, and 244 healthy controls [48]. The biomarkers were significantly elevated in sera from OvCa women compared to all the control groups. Assay performance was significantly much better than CA125 for discriminating borderline EOC from benign and healthy controls (AUROCC was 0.884 vs. 0.843 for CA125). At a specificity of 95%, the in vitro diagnostic multivariate index assay (IVDMIA) achieved a sensitivity of 69.5 % (compared to 62.5 % for CA125 alone).

Bast and colleagues identified CA125 in 1981, when developing the monoclonal antibody OC125 against the serous OvCa-derived cell line, OVCA 433 [49]. It is a high molecular weight glycoprotein encoded by MUC16. The protein comprises of a tandem repeat of 156 amino acids at its N-terminal, a phosphorylation site at the C-terminal, and a possible transmembrane region.

The commercial immunoassay originally marketed in 1983 used the OC125 antibody. A second iteration of the test included another antibody, M11, with distinct epitope from OC125. The FDA recommends CA125 for limited intended use in the management of OvCa. However, studies show it has value in screening, differential diagnosis of a pelvic mass, treatment response and recurrence monitoring, as well as prognostic prediction.

While useful, CA125 alone faces numerous confounding factors. CA125 levels are influenced by factors such as age and race. Aging is associated with decreasing levels, and race-associated variations are observed in women after menopause. A recorded 20–50 % variation is recorded, with African-American women having lower concentrations than white women. Moreover, menstrual cycle affects serum levels, and levels increase in 1–2 % of healthy people, in 5 % of women with other gynecological conditions, and in up to 28 % of non-gynecological malignancies.

The CA125 assay is affected by analytical variables such as sample handling. Samples for CA125 analysis are therefore recommended to be assayed as soon as serum is separated; otherwise, they can be refrigerated at 4 ^°C for up to 5 days or frozen at −20 ^°C for short term (2 weeks to 3 months) or deep frozen at −70 ^°C for long term storage.