Ovarian cancer is a major cause of cancer death among women. A woman’s survival often hinges on doctors detecting the tumor before it has spread beyond the ovary. Unfortunately, most women with ovarian cancer are not diagnosed until they have symptoms – such as pelvic pain, bloating, swelling of the abdomen or appetite loss. By then, the disease has usually spread and is difficult to treat. There is currently no reliable test to diagnose ovarian cancer before symptoms emerge. Some tests measure proteins in the blood or use ultrasound images to identify ovary tumors. These tests usually still identify the disease too late. Sometimes they produce “false positive” results, which may cause women without cancer to undergo unnecessary surgery.

Many ovarian cancers have defects in small pieces of genetic information called microRNAs. These microRNAs impact the tumor in multiple ways, and cells release microRNAs into the blood. Testing a seemingly healthy women’s blood for the same pattern of altered microRNAs found in women with ovarian cancer might be one way to detect the disease earlier.

Now, Elias et al. have identified a pattern of seven microRNAs in the blood that appears to predict ovarian cancer. In the experiments, a computer program searched for microRNA patterns in women with ovarian cancer. The program sifted through the microRNAs in blood from women with and without ovarian cancer. Over time, the computer program “learned” to identify a pattern of microRNAs found only in women with ovarian cancer. It then created a formula for identifying ovarian cancer based on seven of the microRNAs.

Elias et al. then verified that the formula accurately detected ovarian cancer by testing it on blood samples from more women with and without cancer. They also found the seven microRNAs in tiny ovarian cancer tumors collected from women. This suggests the formula might be able to detect even the smallest tumors. More studies are needed to determine when this cancer-linked pattern first emerges and confirm that this ovarian cancer-detection formula works. If the test is validated, it might be used to screen women who are at high risk for ovarian cancer because of mutations in the BRCA1 and BRCA2 genes.

Invasive epithelial ovarian cancer (EOC) is the leading cause of death from gynecologic cancer among women in developed countries (Siegel et al., 2016). Most women with EOC present with advanced stage disease, where 5 year survival rates average 25–30%, highlighting the need for an effective screening strategy. Unfortunately, two large-scale randomized clinical trials involving ultrasound and CA125, including the Prostate, Lung, Colorectal, and Ovarian Cancer (PLCO) trial and the United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) trial did not demonstrate a meaningful impact on overall survival from EOC (Zhu et al., 2011; Jacobs et al., 2016). These and other non-experimental longitudinal studies reaffirm CA125 can detect advanced disease but with poorer sensitivity for early stage and non-serous cancers. In addition, CA125 has limited specificity, with the majority of abnormal CA125 values being the result of non-gynecologic malignancies or benign gynecologic conditions (Moss et al., 2005). The hope that adding more biomarkers to CA125 would improve screening was not realized in a re-analysis of the PLCO data as well as a recent longitudinal study from the European Prospective Investigation of Nutrition and Cancer (Zhu et al., 2011; Terry et al., 2016). In a separate strategy to improve EOC outcome, several panels (which have CA125 as part of them) have received FDA approval to be used in the differential diagnosis of EOC to encourage referral of EOC cases to centers with greater expertise in cancer surgery and chemotherapeutic treatment (Karst and Drapkin, 2010). However, these have not been effective for early diagnosis.

Among the alternatives to serum proteins for the diagnosis or early detection of EOC, circulating microRNAs (miRNAs) have shown great potential (Nakamura et al., 2016). miRNAs are short (18–24 nucleotide) non-coding RNAs that regulate gene expression through post-transcriptional modification of mRNA transcripts. miRNAs have several advantages over protein measures: (1) PCR amplifies detection of rare transcripts in blood; (2) all miRNAs use the same units of measure, easing incorporation into multiplexed panels; and (3) miRNAs play a critical role in ovarian cancer biology, whereas the function of CA125 is unknown (Deb et al., 2017; Katz et al., 2015). Moreover, non-invasive sampling of circulating miRNAs has a clear advantage over analytes obtained through biopsy (Wang et al., 2016).

Preliminary studies have suggested that circulating miRNAs profiles are altered in women with ovarian cancer (Nakamura et al., 2016; Chung et al., 2013; Langhe et al., 2015; Resnick et al., 2009; Zuberi et al., 2015; Samuel and Carter, 2016). In addition, miRNAs have prognostic significance for EOC survival (Merritt et al., 2008; Bagnoli et al., 2016; Cramer and Elias, 2016). However, efforts to develop a diagnostic signature based on circulating miRNAs have been hampered by issues regarding the best statistical approach to develop a model, reproducibility of miRNA measurement across technology platforms (e.g. qPCR, next generation sequencing, microarray), and the biologic heterogeneity of EOC (Nakamura et al., 2016). In this study, our objective was to develop a serum-based miRNA model for the diagnosis of ovarian cancer that could address these concerns and demonstrate the biologic and clinical relevance of this diagnostic tool.

To produce our diagnostic circulating miRNA signature from human sera, we constructed a study population of pre-treatment (prior to either surgery or chemotherapy) subjects comprising 179 women selected from three independent prospective studies (ERASMOS, PMP, and NECC) (Table 1). ERASMOS contributed consecutive cases presenting for evaluation of an adnexal mass, while PMP allowed enrichment of the population for specific histopathologic diagnoses. NECC added healthy controls age-matched to PMP. After completing small RNA sequencing on the sera, subjects were randomly assigned into model training and testing sets (Figure 1). After the randomization, the training and testing sets were demographically similar, and there were no differences in the distribution of histopathological diagnoses between the sets (Table 2).

Figure 1

Download asset Open asset

We then deployed a series of statistical tools, including machine-learning approaches to analyze the miRNA-seq data to create an algorithm with the best performance for discriminating cases of ovarian cancer from either benign tumors, non-invasive (‘borderline’) tumors, or healthy controls. This began by using three different potential strategies for selecting miRNA variable inputs to build the models: significance-based (by t-test), correlation-based feature subset, or expression fold change (Table 3). Each miRNA variable list method was entered into one of 11 different models, which were compared both by AUC (Table 4) as well as sensitivity and specificity (Figure 2).

Figure 2

Download asset Open asset

Although many of the models performed well, the neural network model employing miRNA expression fold changes was the only model to meet our pre-specified statistical objective with an AUC of 0.90 (95% CI: 0.81–0.99; p=0.03 over a theoretical AUC of 0.75). The network consisted of 14 individual miRNAs with seven neurons in the hidden layer (Source code 1). As the network relied on complex interactions between miRNA levels we tested whether its performance was not biased by batch adjustment performed at the initial step of the analysis. The neural network worked equally well on the adjusted and unadjusted raw datasets with an AUC of 0.93 (95%CI: 0.89–0.98) on the training and 0.90 (95%CI 0.80–0.99) on the testing set (Figure 3; Supplementary file 1A [by model] and 1B [by sample]). In post-hoc secondary analyses, the neural network worked equally well for older and younger patients, serous and non-serous histologies, and early and advanced stage disease (Supplementary file 2A-C).

Figure 3

Download asset Open asset

Serum CA125 data were available for 120 subjects (Supplementary file 1B and 3A). Among these, the neural network (AUC 0.93; 95% CI 0.88–0.97) significantly outperformed CA125 (AUC 0.74; 95% CI 0.65–0.83; p=0.001; Figure 4). The primary advantage of the neural network over CA125 was avoiding false positives (8/43 for the neural network versus 23/43 for CA125; p=0.002) (Supplementary file 2A). Notably, the neural network and CA125 levels were independent of one another (Figure 4—figure supplement 1; Supplementary file 3B). We tested using the neural network and CA125 in a tiered testing strategy, subjecting all negative neural network algorithm results to a second review with CA125, but found this would increase the probability of a false positive test result from 4.2% (5/120) to 19.2% (23/120) and a false negative rate from 5.8% (7/120) to 13.3% (16/120) (Figure 4—figure supplement 2). The alternative of initial screening with CA125 followed by neural network yielded only three additional correctly diagnosed cases of invasive cancer at the expense of 19 additional false positive results.

Figure 4 with 2 supplements see all

Download asset Open asset

The specificity of the neural network algorithm for the diagnosis of ovarian cancer was tested using an external, independent, dataset previously published by Keller, et al (Keller et al., 2011). These data were generated via a third technology platform, probe-based microarray, which fortunately contained all 14 miRNAs from our original signature, allowing for 1:1 mapping without exclusions (Supplementary file 4A and Supplementary file 6). The neural network perfectly classified patients in the training set (AUC 1.00, 95% CI 1.00–1.00) and provided very good discriminatory power on the testing set (AUC 0.93, 95% CI 0.81–1.00), with an overall sensitivity of 75% and specificity of 100%. The signature was specific to ovarian cancer compared to all other diagnoses, as it did not show any clinically-efficient diagnostic capabilities for any of the 12 other morbidities analysed in the set and showed good performance in distinguishing ovarian cancer samples against all other diagnoses combined (AUC 0.92, 95% CI 0.82–1.00) (Figure 5).

Figure 5

Download asset Open asset

Having established our miRNAs of interest using next generation sequencing, we next sought to validate the sequencing data across technology platforms by measuring the miRNAs from the neural network using qPCR. While small RNA sequencing is a more robust technology for miRNA discovery, qPCR is a more time efficient and cost-effective diagnostic tool. For this we used 120 samples from PMP and NECC for which we had excess RNA. We internally validated the 14 miRNAs in the neural network (plus an additional nine potential reference miRNAs derived from the sequencing data) by qPCR and recalibrated the algorithm to accept qPCR inputs (Supplementary file 6). We then performed a global sensitivity analysis on the best neural network for qPCR data and iteratively removed the variables which did the least in terms of improving the classifier’s performance. This reduced the neural network to only seven miRNAs (miR-29a-3p, miR-92a-3p, miR-200c-3p, miR-320c, miR-335–5 p, miR-450b-5p, and miR-1307–5 p) plus four normalizers (miR-423–3 p, miR-191–5 p, miR-221–3 p, and miR-103a-3p). To increase the statistical power of this qPCR-based classifier and create a fully locked-down model for clinical application, we added 205 more samples from PMP and NECC, including more than 100 additional healthy controls, to create a 325 subject population for qPCR model development (Table 5).

These samples were randomized 3:1 into training and testing sets to create a neural network. The resulting network performed well with an AUC 0.89 on the training set and AUC 0.80 on the testing set.

We then tested the clinical performance of the final, locked-down diagnostic test on a completely independent external sample set collected from 51 preoperative patients treated in Lodz, Poland (Table 6). In this population, the neural network had a positive predictive value of 91.3% (95% CI: 73.3–97.6%) and a negative predictive value of 78.6% (95% CI: 64.2–88.2%) with an AUC of 0.85 (Figure 6).

Figure 6

Download asset Open asset

Ideally, a serum biomarker should have biologic relevance to the clinical disease. To this end, we returned to the ERASMOS patient set to examine if the expression levels of the miRNAs changed in the cancer patients after surgical cytoreduction. Among the patients with ovarian cancer in the study, 27 had both preoperative and postoperative serum miRNAs profiled. These included 4/7 target miRNAs in the qPCR neural network model. Circulating levels of all three miRNAs decreased within 72 hr of tumor removal, with significant changes for miR-200a-3p and miR-200c-3p (Figure 7A–D).

Figure 7

Download asset Open asset

We also wanted to test if the miRNAs were in fact coming from the earliest lesions of this disease. For this, we assembled paraffin-embedded tissue sections from independent sets of 15 cases of serous tubal intraepithelial carcinomas and 15 Stage I high grade (serous or Grade three endometrioid) epithelial ovarian cancers. Immunohistochemistry was performed on sequential sections for TP53 and Ki67 to highlight the lesions. We then performed in situ hybridization for three of the miRNAs in our neural network; mir-200c-3p, mir-335–5 p, and mir-92a-3p (Figure 8). In 100% of the samples, there was complete overlap between lesional cells and the miRNAs crucial for neural network performance, suggesting that the miRNAs detected in the serum are present even in early lesions in the fallopian tube epithelium and raising the possibility of detection of pre-metastatic disease.

Figure 8

Download asset Open asset

Finally, we have constructed a web calculator (http://biostat.umed.pl/ovaries) to demonstrate how to use these models. The calculator accepts various inputs describing on the method of circulating miRNA quantification (sequencing, qPCR, or microarray) and returns the estimated probability of ovarian cancer for a given patient.

We have described the development of a diagnostic model for ovarian cancer using sequencing of circulating miRNA. This is the first study in ovarian cancer to combine next generation sequencing technology for serum miRNA with machine learning techniques. Not only does sequencing provide greater sensitivity for miRNA detection than other methods, but expression levels of various miRNAs are not linearly related and relationships among miRNAs tend to be obscured by more basic statistical approaches. The neural network as presented has several advantages over a traditional biomarker like CA125. The neural network recognized more Stage I/II ovarian cancers and had significantly fewer false positives. This likely reflects an ability to discriminate relevant biology more than to quantify tumor burden. For example, the neural network correctly classified 35/43 (81%) borderline tumors as being non-invasive neoplasms, compared to just 20/43 (47%; p=0.002) for CA125. An additional strength of our study is the incorporation of multiple independent datasets. The ERASMOS specimens were obtained from cases enrolled sequentially, reflecting the natural frequency of different ovarian tumor subtypes in the clinical population, including the fact that most women with invasive ovarian cancer present with advanced stage disease. The Pelvic Mass Protocol samples allowed us to enrich the study population for less common clinical cases that would be expected to confound a conventional screening algorithm, including benign complex ovarian masses, borderline tumors, early stage cancers, and non-serous histologic subtypes. NECC provided age-matched healthy controls. The specificity of our model was tested using a publicly available dataset from Keller, et al where we showed that the neural network performed well across disease stages, histologic subtypes, and diagnostic platforms. This ability to specifically identify ovarian cancers and discriminate ovarian cancer from other diagnoses sets the current work apart from prior miRNA studies (Nakamura et al., 2016; Chung et al., 2013; Resnick et al., 2009; Zuberi et al., 2015; Häusler et al., 2010; Zheng et al., 2013). Finally, we tested our signature using a completely external, independent set of samples from Poland, showing that in a clinical sample set the test performed well without additional modifications.

There appears to be biologic relevance to the serum miRNAs in the neural network. The rapid change in circulating levels after surgical cytoreduction for mir-200a and mir-200c suggests these are being produced actively by tumors. Although other miRNAs did not have as great of a decrease, this may reflect differing half-lives for different miRNA species. In future work, it would be interesting to measure changes over a longer time frame than 72 hr, but that was the endpoint for ERASMOS, which is an anesthesia-focused study. We also demonstrated expression of several miRNAs from the neural network in pre-metastatic lesions. This both confirms prior work suggesting that these miRNAs are detectable in advanced ovarian cancers specimens and adds the new finding that these miRNAs are expressed in very early stage and even pre-invasive lesions (Bagnoli et al., 2016). Future work will examine the kinetics of these miRNA changes in tumor pathogenesis.

The phase II specimens used in this study are like those used to support development of assay panels subsequently approved for the differential diagnosis of ovarian cancer vs. a benign pelvic mass. The first panel, named OVA1, was approved by the FDA in 2009 and consisted of 5 analytes including CA125 (Zhang et al., 2004; Ueland et al., 2011). While those authors emphasized the assay’s negative predictive value of 95% (when combined with physician assessment), the assay had an AUC of only 0.80 (95% CI: 0.73–0.88) for pre-menopausal women and 0.82 (95% CI: 0.77–0.87) for post-menopausal women. The second panel was approved in 2011 and consisted of just two markers, CA125 and HE4, combined with menopausal status (Moore et al., 2010). While the ROMA algorithm had an overall AUC for discriminating cancer from benign tumors of 0.91 (95% CI: 0.88–0.94), this was in the setting of including borderline tumors as malignancies. Moreover, the positive predictive value of the test for distinguishing benign masses from Stage I/II EOC was only 0.27. In 2016, the FDA approved an updated version of the OVA1 test which retained CA125 but replaced 2 of the markers with HE4 and FSH (Coleman et al., 2016). This improved the overall AUC to 0.92 (95% CI: 0.89–0.96) for the assay alone and 0.94 (95% CI: 0.91–0.97) when combined with physician assessment, although 80% of the tumors in this study were benign. Although the above panels included some clinical information and therefore are not equivalent to our panel, we point out that the AUC of our panel to distinguish a malignant from benign pelvic mass was similar, while not including borderline tumors as positive results and agnostic to clinical or imaging information. As timely referral to a gynecologic oncologist is a strong predictor of ovarian cancer survival, we believe that there is a role for a test based on blood markers alone (Earle et al., 2006).

FDA approval of the various panels for use in the differential diagnosis of pelvic masses did not extend to their use in the general population. Based upon the results of the PLCO and UKCTOCS randomized clinical trials (or so called ‘phase 4’) the US Preventive Services Task Force and the Society of Gynecologic Oncology (SGO) have not recommended routine screening for ovarian cancer (Zhu et al., 2011; Skates et al., 2001). However, screening with CA125 and transvaginal ultrasound is recommended by the National Comprehensive Cancer Network guidelines and the SGO for women with known hereditary syndromes of ovarian cancer (such as women with germline BRCA1/2 mutations), even though there is currently no evidence that this screening strategy improves survival in elevated risk populations (Schorge et al., 2010).

Recent studies (Chung et al., 2013; Langhe et al., 2015; Resnick et al., 2009; Zuberi et al., 2015) have identified circulating (serum/plasma) miRNAs that are altered in ovarian carcinomas, and there is limited overlap with miRNAs that emerged from our analysis. One possible cause of this difference is the limited number of samples examined in these studies. For example, in Langhe et al, a training set of 5 serous ovarian carcinomas and five benign serous cystadenomas were selected for the initial experiments. The validation set was 20 serous ovarian carcinomas and 20 benign serous cystadenomas. In Resnick et al, 28 ovarian carcinoma patients and 15 healthy controls were used to identify to differential expression of circulating miRNAs. Such limited numbers diminish the statistical robustness of the results. Another possible cause for the differences is the miRNA expression profiling platform. Recently a study (Mestdagh et al., 2014) systematically compared 12 different miRNA expression platforms. Specifically, for serum miRNAs there was a 12-fold difference between the highest and lowest number of detected miRNA when identical samples were profiled by different platforms. According to this report the LNA-based platform from Exiqon has the highest specificity but maybe limited for sensitivity thereby for detection. To circumvent both these concerns we started with next-generation sequencing of 179 samples which captures all small RNAs and addresses any issues of detection or specificity due to limitations of platforms. Next, we did validation using the Exiqon qRT-PCR platform on 325 local samples, and a further validation using an additional cohort of 51 samples from Poland. The large number of samples along with the methodologies used for identification and validation of the circulating miRNAs in our study provides strong support for our conclusions and distinguishes our work from prior reports.

Our study does have several limitations. Whether our miRNA panel will prove useful in the differential diagnosis of early detection will require further study in the following areas. First, additional study is necessary to determine whether integrating clinical risk factors could further improve its performance. Second, confirmation in other phase II data sets are necessary to validate our study results and demonstrate its generalizability. Third, specimens collected and stored months or years prior to a clinical diagnosis (so called phase III specimens) are necessary to demonstrate the model’s potential in the early detection of EOC in a general population or elevated risk setting. For the former, we have access to PLCO specimens; and for the latter, we plan to apply for specimens to the National Clinical Trials Network. Fourth, a logical extension of our work is to determine whether our current miRNA panel (or a new one) would be useful in predicting survival after EOC. A tissue-based MiROvaR signature involving 35 miRNAs for predicting EOC prognosis has recently been described (Bagnoli et al., 2016). Although several miRNAs appear in the tissue signature and our model (Supplementary file 4B), full concordance is unlikely since the tissue model was built to predict prognosis whereas our model was built to predict diagnosis. In addition, about two-thirds of the miRNAs in the tissue signature are not reliably detectable in circulation, which can be attributed to the fact that relatively few miRNAs circulate in serum (Mestdagh et al., 2014; Dinh et al., 2016). Our serum panel is reliant on a smaller number of miRNAs simply because the neural model prioritizes ones that provide novel information. If miRNAs are correlated (for example within the same chromosomal cluster), they will be invariant and knowing one will convey sufficient information about the rest for them to be excluded from model building. Finally, more is to be learned about the basic biology of serum miRNA. Are they all coming from cancer cells or also other cells in the tumor microenvironment? (Likely, both are included in the signature). It is noteworthy that two of the miRNAs are members of the mir-200 family, confirming prior reports identifying these miRNAs as overexpressed in ovarian cancer (Zuberi et al., 2015; Pecot et al., 2013). Some of the miRNAs incorporated into the neural network algorithm have connections to other disease types. For example, miR-1246 has been identified in the serum of ovarian cancer, lung cancer, prostate cancer, and stroke patients (Todeschini et al., 2017; Zhang et al., 2016; Alhasan et al., 2016; Li et al., 2015). However, as noted in Figure 5, the network as a whole was specific to ovarian cancer, again emphasizing the importance of multimarker panels.

In conclusion, serum miRNA adds to the toolbox of options to diagnose ovarian cancer. We plan several future studies to characterize the miRNA neural network. Whether serum miRNA offers a lead time advantage over other putative biomarkers remains to be proven. We need to study the performance characteristics of the miRNA neural network in high risk and low risk populations. Finally, we are performing laboratory investigations to elucidate the biologic function of these miRNAs and to understand the kinetics of miRNA expression in ovarian cancer pathogenesis. With our improved understanding of miRNA analytic approaches, we can develop better models for this and other diseases.

1. 1. Elias KM
   2. et al

   (2017) Serum microRNA sequencing for diagnosis of invasive ovarian cancer

   Publicly available at the NCBI Gene Expression Omnibus (accession no: GSE94533).

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE94533

1. 1. Keller A
   2. Leidinger P
   3. Bauer A
   4. Elsharawy A
   5. Haas J
   6. Backes C

   (2011) The human Whole miRNOme project version 1

   Publicly available at the NCBI Gene Expression Omnibus (accession no: GSE31568).

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE31568

1. 1. Alhasan AH
   2. Scott AW
   3. Wu JJ
   4. Feng G
   5. Meeks JJ
   6. Thaxton CS
   7. Mirkin CA

   (2016) Circulating microRNA signature for the diagnosis of very high-risk prostate cancer

   PNAS 113:10655–10660.

   https://doi.org/10.1073/pnas.1611596113

   • PubMed
   • Google Scholar
2. 1. Andersen CL
   2. Jensen JL
   3. Ørntoft TF

   (2004) Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets

   Cancer Research 64:5245–5250.

   https://doi.org/10.1158/0008-5472.CAN-04-0496

   • PubMed
   • Google Scholar
3. 1. Bagnoli M
   2. Canevari S
   3. Califano D
   4. Losito S
   5. Maio MD
   6. Raspagliesi F
   7. Carcangiu ML
   8. Toffoli G
   9. Cecchin E
   10. Sorio R
   11. Canzonieri V
   12. Russo D
   13. Scognamiglio G
   14. Chiappetta G
   15. Baldassarre G
   16. Lorusso D
   17. Scambia G
   18. Zannoni GF
   19. Savarese A
   20. Carosi M
   21. Scollo P
   22. Breda E
   23. Murgia V
   24. Perrone F
   25. Pignata S
   26. De Cecco L
   27. Mezzanzanica D
   28. Multicentre Italian Trials in Ovarian cancer (MITO) translational group

   (2016) Development and validation of a microRNA-based signature (MiROvaR) to predict early relapse or progression of epithelial ovarian cancer: a cohort study

   The Lancet Oncology 17:1137–1146.

   https://doi.org/10.1016/S1470-2045(16)30108-5

   • PubMed
   • Google Scholar
4. 1. Broyden CG

   (1970) The convergence of a class of double-rank minimization algorithms 1. general considerations

   IMA Journal of Applied Mathematics 6:76–90.

   https://doi.org/10.1093/imamat/6.1.76

   • Google Scholar
5. 1. Bustin SA
   2. Benes V
   3. Garson JA
   4. Hellemans J
   5. Huggett J
   6. Kubista M
   7. Mueller R
   8. Nolan T
   9. Pfaffl MW
   10. Shipley GL
   11. Vandesompele J
   12. Wittwer CT

   (2009) The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments

   Clinical Chemistry 55:611–622.

   https://doi.org/10.1373/clinchem.2008.112797

   • PubMed
   • Google Scholar
6. 1. Chung YW
   2. Bae HS
   3. Song JY
   4. Lee JK
   5. Lee NW
   6. Kim T
   7. Lee KW

   (2013) Detection of microRNA as novel biomarkers of epithelial ovarian cancer from the serum of ovarian cancer patients

   International Journal of Gynecological Cancer 23:673–679.

   https://doi.org/10.1097/IGC.0b013e31828c166d

   • PubMed
   • Google Scholar
7. 1. Coleman RL
   2. Herzog TJ
   3. Chan DW
   4. Munroe DG
   5. Pappas TC
   6. Smith A
   7. Zhang Z
   8. Wolf J

   (2016) Validation of a second-generation multivariate index assay for malignancy risk of adnexal masses

   American Journal of Obstetrics and Gynecology 215:82.e1–8282.

   https://doi.org/10.1016/j.ajog.2016.03.003

   • PubMed
   • Google Scholar
8. 1. Collins GS
   2. Reitsma JB
   3. Altman DG
   4. Moons KG

   (2015) Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement

   Annals of Internal Medicine 162:55–63.

   https://doi.org/10.7326/M14-0697

   • PubMed
   • Google Scholar
9. 1. Cramer DW
   2. Vitonis AF
   3. Welch WR
   4. Terry KL
   5. Goodman A
   6. Rueda BR
   7. Berkowitz RS

   (2010) Correlates of the preoperative level of CA125 at presentation of ovarian cancer

   Gynecologic Oncology 119:462–468.

   https://doi.org/10.1016/j.ygyno.2010.08.028

   • PubMed
   • Google Scholar
10. 1. Cramer DW
    2. Elias KM

    (2016) A prognostically relevant miRNA signature for epithelial ovarian cancer

    The Lancet Oncology 17:1032–1033.

    https://doi.org/10.1016/S1470-2045(16)30149-8

    • PubMed
    • Google Scholar
11. 1. Deb B
    2. Uddin A
    3. Chakraborty S

    (2017) miRNAs and ovarian cancer: An overview

    Journal of Cellular Physiology.

    https://doi.org/10.1002/jcp.26095

    • PubMed
    • Google Scholar
12. 1. Dinh TK
    2. Fendler W
    3. Chałubińska-Fendler J
    4. Acharya SS
    5. O'Leary C
    6. Deraska PV
    7. D'Andrea AD
    8. Chowdhury D
    9. Kozono D

    (2016) Circulating miR-29a and miR-150 correlate with delivered dose during thoracic radiation therapy for non-small cell lung cancer

    Radiation Oncology 11:61.

    https://doi.org/10.1186/s13014-016-0636-4

    • PubMed
    • Google Scholar
13. 1. Earle CC
    2. Schrag D
    3. Neville BA
    4. Yabroff KR
    5. Topor M
    6. Fahey A
    7. Trimble EL
    8. Bodurka DC
    9. Bristow RE
    10. Carney M
    11. Warren JL

    (2006) Effect of surgeon specialty on processes of care and outcomes for ovarian cancer patients

    JNCI: Journal of the National Cancer Institute 98:172–180.

    https://doi.org/10.1093/jnci/djj019

    • PubMed
    • Google Scholar
14. 1. Elias KM
    2. Fusco A
    3. Kang S
    4. Michaud D
    5. Berkowitz RS
    6. Horowitz NS
    7. Frendl G

    (2015) A prospective phase 0 study on the effects of anesthetic selection on serum miRNA profiles during primary cytoreductive surgery for suspected ovarian cancer

    Gynecologic Oncology 137:1.

    https://doi.org/10.1016/j.ygyno.2015.01.341

    • Google Scholar
15. 1. Fletcher R

    (1970) A new approach to variable metric algorithms

    The Computer Journal 13:317–322.

    https://doi.org/10.1093/comjnl/13.3.317

    • Google Scholar
16. 1. Goldfarb D

    (1970) A family of variable-metric methods derived by variational means

    Mathematics of Computation 24:23–26.

    https://doi.org/10.1090/S0025-5718-1970-0258249-6

    • Google Scholar
17. Book

    1. Hall MA

    (1998)

    Correlation-Based Feature Subset Selection for Machine Learning

    Hamilton, New Zealand: University of Waikato.

    • Google Scholar
18. 1. Hanley JA
    2. McNeil BJ

    (1982) The meaning and use of the area under a receiver operating characteristic (ROC) curve

    Radiology 143:29–36.

    https://doi.org/10.1148/radiology.143.1.7063747

    • PubMed
    • Google Scholar
19. 1. Häusler SF
    2. Keller A
    3. Chandran PA
    4. Ziegler K
    5. Zipp K
    6. Heuer S
    7. Krockenberger M
    8. Engel JB
    9. Hönig A
    10. Scheffler M
    11. Dietl J
    12. Wischhusen J

    (2010) Whole blood-derived miRNA profiles as potential new tools for ovarian cancer screening

    British Journal of Cancer 103:693–700.

    https://doi.org/10.1038/sj.bjc.6605833

    • PubMed
    • Google Scholar
20. 1. Jacobs IJ
    2. Menon U
    3. Ryan A
    4. Gentry-Maharaj A
    5. Burnell M
    6. Kalsi JK
    7. Amso NN
    8. Apostolidou S
    9. Benjamin E
    10. Cruickshank D
    11. Crump DN
    12. Davies SK
    13. Dawnay A
    14. Dobbs S
    15. Fletcher G
    16. Ford J
    17. Godfrey K
    18. Gunu R
    19. Habib M
    20. Hallett R
    21. Herod J
    22. Jenkins H
    23. Karpinskyj C
    24. Leeson S
    25. Lewis SJ
    26. Liston WR
    27. Lopes A
    28. Mould T
    29. Murdoch J
    30. Oram D
    31. Rabideau DJ
    32. Reynolds K
    33. Scott I
    34. Seif MW
    35. Sharma A
    36. Singh N
    37. Taylor J
    38. Warburton F
    39. Widschwendter M
    40. Williamson K
    41. Woolas R
    42. Fallowfield L
    43. McGuire AJ
    44. Campbell S
    45. Parmar M
    46. Skates SJ

    (2016) Ovarian cancer screening and mortality in the UK collaborative trial of ovarian cancer screening (UKCTOCS): a randomised controlled trial

    The Lancet 387:945–956.

    https://doi.org/10.1016/S0140-6736(15)01224-6

    • PubMed
    • Google Scholar
21. 1. Johnson WE
    2. Li C
    3. Rabinovic A

    (2007) Adjusting batch effects in microarray expression data using empirical Bayes methods

    Biostatistics 8:118–127.

    https://doi.org/10.1093/biostatistics/kxj037

    • PubMed
    • Google Scholar
22. 1. Karst AM
    2. Drapkin R

    (2010) Ovarian cancer pathogenesis: a model in evolution

    Journal of Oncology 2010:932371.

    https://doi.org/10.1155/2010/932371

    • PubMed
    • Google Scholar
23. 1. Katz B
    2. Tropé CG
    3. Reich R
    4. Davidson B

    (2015) MicroRNAs in ovarian cancer

    Human Pathology 46:1245–1256.

    https://doi.org/10.1016/j.humpath.2015.06.013

    • PubMed
    • Google Scholar
24. 1. Keller A
    2. Leidinger P
    3. Bauer A
    4. Elsharawy A
    5. Haas J
    6. Backes C
    7. Wendschlag A
    8. Giese N
    9. Tjaden C
    10. Ott K
    11. Werner J
    12. Hackert T
    13. Ruprecht K
    14. Huwer H
    15. Huebers J
    16. Jacobs G
    17. Rosenstiel P
    18. Dommisch H
    19. Schaefer A
    20. Müller-Quernheim J
    21. Wullich B
    22. Keck B
    23. Graf N
    24. Reichrath J
    25. Vogel B
    26. Nebel A
    27. Jager SU
    28. Staehler P
    29. Amarantos I
    30. Boisguerin V
    31. Staehler C
    32. Beier M
    33. Scheffler M
    34. Büchler MW
    35. Wischhusen J
    36. Haeusler SF
    37. Dietl J
    38. Hofmann S
    39. Lenhof HP
    40. Schreiber S
    41. Katus HA
    42. Rottbauer W
    43. Meder B
    44. Hoheisel JD
    45. Franke A
    46. Meese E

    (2011) Toward the blood-borne miRNome of human diseases

    Nature Methods 8:841–843.

    https://doi.org/10.1038/nmeth.1682

    • PubMed
    • Google Scholar
25. 1. Langhe R
    2. Norris L
    3. Saadeh FA
    4. Blackshields G
    5. Varley R
    6. Harrison A
    7. Gleeson N
    8. Spillane C
    9. Martin C
    10. O'Donnell DM
    11. D'Arcy T
    12. O'Leary J
    13. O'Toole S

    (2015) A novel serum microRNA panel to discriminate benign from malignant ovarian disease

    Cancer Letters 356:628–636.

    https://doi.org/10.1016/j.canlet.2014.10.010

    • PubMed
    • Google Scholar
26. 1. Li P
    2. Teng F
    3. Gao F
    4. Zhang M
    5. Wu J
    6. Zhang C

    (2015) Identification of circulating microRNAs as potential biomarkers for detecting acute ischemic stroke

    Cellular and Molecular Neurobiology 35:433–447.

    https://doi.org/10.1007/s10571-014-0139-5

    • PubMed
    • Google Scholar
27. 1. Merritt WM
    2. Lin YG
    3. Han LY
    4. Kamat AA
    5. Spannuth WA
    6. Schmandt R
    7. Urbauer D
    8. Pennacchio LA
    9. Cheng JF
    10. Nick AM
    11. Deavers MT
    12. Mourad-Zeidan A
    13. Wang H
    14. Mueller P
    15. Lenburg ME
    16. Gray JW
    17. Mok S
    18. Birrer MJ
    19. Lopez-Berestein G
    20. Coleman RL
    21. Bar-Eli M
    22. Sood AK

    (2008) Dicer, Drosha, and outcomes in patients with ovarian cancer

    New England Journal of Medicine 359:2641–2650.

    https://doi.org/10.1056/NEJMoa0803785

    • PubMed
    • Google Scholar
28. 1. Mestdagh P
    2. Hartmann N
    3. Baeriswyl L
    4. Andreasen D
    5. Bernard N
    6. Chen C
    7. Cheo D
    8. D'Andrade P
    9. DeMayo M
    10. Dennis L
    11. Derveaux S
    12. Feng Y
    13. Fulmer-Smentek S
    14. Gerstmayer B
    15. Gouffon J
    16. Grimley C
    17. Lader E
    18. Lee KY
    19. Luo S
    20. Mouritzen P
    21. Narayanan A
    22. Patel S
    23. Peiffer S
    24. Rüberg S
    25. Schroth G
    26. Schuster D
    27. Shaffer JM
    28. Shelton EJ
    29. Silveria S
    30. Ulmanella U
    31. Veeramachaneni V
    32. Staedtler F
    33. Peters T
    34. Guettouche T
    35. Wong L
    36. Vandesompele J

    (2014) Evaluation of quantitative miRNA expression platforms in the microRNA quality control (miRQC) study

    Nature Methods 11:809–815.

    https://doi.org/10.1038/nmeth.3014

    • PubMed
    • Google Scholar
29. 1. Moore RG
    2. Jabre-Raughley M
    3. Brown AK
    4. Robison KM
    5. Miller MC
    6. Allard WJ
    7. Kurman RJ
    8. Bast RC
    9. Skates SJ

    (2010) Comparison of a novel multiple marker assay vs the Risk of Malignancy Index for the prediction of epithelial ovarian cancer in patients with a pelvic mass

    American Journal of Obstetrics and Gynecology 203:228.e1–22228.

    https://doi.org/10.1016/j.ajog.2010.03.043

    • PubMed
    • Google Scholar
30. 1. Moss EL
    2. Hollingworth J
    3. Reynolds TM

    (2005) The role of CA125 in clinical practice

    Journal of Clinical Pathology 58:308–312.

    https://doi.org/10.1136/jcp.2004.018077

    • PubMed
    • Google Scholar
31. 1. Nakamura K
    2. Sawada K
    3. Yoshimura A
    4. Kinose Y
    5. Nakatsuka E
    6. Kimura T

    (2016) Clinical relevance of circulating cell-free microRNAs in ovarian cancer

    Molecular Cancer 15:48.

    https://doi.org/10.1186/s12943-016-0536-0

    • PubMed
    • Google Scholar
32. 1. Pecot CV
    2. Rupaimoole R
    3. Yang D
    4. Akbani R
    5. Ivan C
    6. Lu C
    7. Wu S
    8. Han HD
    9. Shah MY
    10. Rodriguez-Aguayo C
    11. Bottsford-Miller J
    12. Liu Y
    13. Kim SB
    14. Unruh A
    15. Gonzalez-Villasana V
    16. Huang L
    17. Zand B
    18. Moreno-Smith M
    19. Mangala LS
    20. Taylor M
    21. Dalton HJ
    22. Sehgal V
    23. Wen Y
    24. Kang Y
    25. Baggerly KA
    26. Lee JS
    27. Ram PT
    28. Ravoori MK
    29. Kundra V
    30. Zhang X
    31. Ali-Fehmi R
    32. Gonzalez-Angulo AM
    33. Massion PP
    34. Calin GA
    35. Lopez-Berestein G
    36. Zhang W
    37. Sood AK

    (2013) Tumour angiogenesis regulation by the miR-200 family

    Nature Communications 4:2427.

    https://doi.org/10.1038/ncomms3427

    • PubMed
    • Google Scholar
33. 1. Perets R
    2. Wyant GA
    3. Muto KW
    4. Bijron JG
    5. Poole BB
    6. Chin KT
    7. Chen JY
    8. Ohman AW
    9. Stepule CD
    10. Kwak S
    11. Karst AM
    12. Hirsch MS
    13. Setlur SR
    14. Crum CP
    15. Dinulescu DM
    16. Drapkin R

    (2013) Transformation of the fallopian tube secretory epithelium leads to high-grade serous ovarian cancer in Brca;Tp53;Pten models

    Cancer Cell 24:751–765.

    https://doi.org/10.1016/j.ccr.2013.10.013

    • PubMed
    • Google Scholar
34. 1. Resnick KE
    2. Alder H
    3. Hagan JP
    4. Richardson DL
    5. Croce CM
    6. Cohn DE

    (2009) The detection of differentially expressed microRNAs from the serum of ovarian cancer patients using a novel real-time PCR platform

    Gynecologic Oncology 112:55–59.

    https://doi.org/10.1016/j.ygyno.2008.08.036

    • PubMed
    • Google Scholar
35. 1. Rice MS
    2. Murphy MA
    3. Vitonis AF
    4. Cramer DW
    5. Titus LJ
    6. Tworoger SS
    7. Terry KL

    (2013) Tubal ligation, hysterectomy and epithelial ovarian cancer in the New England Case-Control Study

    International Journal of Cancer 133:2415–2421.

    https://doi.org/10.1002/ijc.28249

    • PubMed
    • Google Scholar
36. 1. Samuel P
    2. Carter DR

    (2016) The diagnostic and prognostic potential of micrornas in epithelial ovarian carcinoma

    Molecular Diagnosis & Therapy 21:59–73.

    https://doi.org/10.1007/s40291-016-0242-z

    • Google Scholar
37. 1. Schorge JO
    2. Modesitt SC
    3. Coleman RL
    4. Cohn DE
    5. Kauff ND
    6. Duska LR
    7. Herzog TJ

    (2010) SGO white paper on ovarian cancer: etiology, screening and surveillance

    Gynecologic Oncology 119:7–17.

    https://doi.org/10.1016/j.ygyno.2010.06.003

    • PubMed
    • Google Scholar
38. 1. Shanno DF

    (1970) Conditioning of quasi-Newton methods for function minimization

    Mathematics of Computation 24:647–656.

    https://doi.org/10.1090/S0025-5718-1970-0274029-X

    • Google Scholar
39. 1. Shanno DF
    2. Kettler PC

    (1970) Optimal conditioning of quasi-Newton methods

    Mathematics of Computation 24:657–664.

    https://doi.org/10.1090/S0025-5718-1970-0274030-6

    • Google Scholar
40. 1. Siegel RL
    2. Miller KD
    3. Jemal A

    (2016) Cancer statistics, 2016

    CA: A Cancer Journal for Clinicians 66:7–30.

    https://doi.org/10.3322/caac.21332

    • PubMed
    • Google Scholar
41. 1. Skates SJ
    2. Jacobs IJ
    3. Knapp RC

    (2001) Tumor markers in screening for ovarian cancer

    Methods in Molecular Medicine 39:61–73.

    https://doi.org/10.1385/1-59259-071-3:61

    • PubMed
    • Google Scholar
42. 1. Terry KL
    2. Schock H
    3. Fortner RT
    4. Hüsing A
    5. Fichorova RN
    6. Yamamoto HS
    7. Vitonis AF
    8. Johnson T
    9. Overvad K
    10. Tjønneland A
    11. Boutron-Ruault MC
    12. Mesrine S
    13. Severi G
    14. Dossus L
    15. Rinaldi S
    16. Boeing H
    17. Benetou V
    18. Lagiou P
    19. Trichopoulou A
    20. Krogh V
    21. Kuhn E
    22. Panico S
    23. Bueno-de-Mesquita HB
    24. Onland-Moret NC
    25. Peeters PH
    26. Gram IT
    27. Weiderpass E
    28. Duell EJ
    29. Sanchez MJ
    30. Ardanaz E
    31. Etxezarreta N
    32. Navarro C
    33. Idahl A
    34. Lundin E
    35. Jirström K
    36. Manjer J
    37. Wareham NJ
    38. Khaw KT
    39. Byrne KS
    40. Travis RC
    41. Gunter MJ
    42. Merritt MA
    43. Riboli E
    44. Cramer DW
    45. Kaaks R

    (2016) A prospective evaluation of early detection biomarkers for ovarian cancer in the european EPIC cohort

    Clinical Cancer Research 22:4664–4675.

    https://doi.org/10.1158/1078-0432.CCR-16-0316

    • PubMed
    • Google Scholar
43. 1. Todeschini P
    2. Salviato E
    3. Paracchini L
    4. Ferracin M
    5. Petrillo M
    6. Zanotti L
    7. Tognon G
    8. Gambino A
    9. Calura E
    10. Caratti G
    11. Martini P
    12. Beltrame L
    13. Maragoni L
    14. Gallo D
    15. Odicino FE
    16. Sartori E
    17. Scambia G
    18. Negrini M
    19. Ravaggi A
    20. D'Incalci M
    21. Marchini S
    22. Bignotti E
    23. Romualdi C

    (2017) Circulating miRNA landscape identifies miR-1246 as promising diagnostic biomarker in high-grade serous ovarian carcinoma: A validation across two independent cohorts

    Cancer Letters 388:320–327.

    https://doi.org/10.1016/j.canlet.2016.12.017

    • PubMed
    • Google Scholar
44. 1. Ueland FR
    2. Desimone CP
    3. Seamon LG
    4. Miller RA
    5. Goodrich S
    6. Podzielinski I
    7. Sokoll L
    8. Smith A
    9. van Nagell JR
    10. Zhang Z

    (2011) Effectiveness of a multivariate index assay in the preoperative assessment of ovarian tumors

    Obstetrics & Gynecology 117:1289–1297.

    https://doi.org/10.1097/AOG.0b013e31821b5118

    • PubMed
    • Google Scholar
45. 1. Vandesompele J
    2. De Preter K
    3. Pattyn F
    4. Poppe B
    5. Van Roy N
    6. De Paepe A
    7. Speleman F

    (2002)

    Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes

    Genome Biology 3:RESEARCH0034.

    • PubMed
    • Google Scholar
46. 1. Vorwerk S
    2. Ganter K
    3. Cheng Y
    4. Hoheisel J
    5. Stähler PF
    6. Beier M

    (2008) Microfluidic-based enzymatic on-chip labeling of miRNAs

    New Biotechnology 25:142–149.

    https://doi.org/10.1016/j.nbt.2008.08.005

    • PubMed
    • Google Scholar
47. 1. Wang Y
    2. Sundfeldt K
    3. Mateoiu C
    4. Shih I
    5. Kurman RJ
    6. Schaefer J
    7. Silliman N
    8. Kinde I
    9. Springer S
    10. Foote M
    11. Kristjansdottir B
    12. James N
    13. Kinzler KW
    14. Papadopoulos N
    15. Diaz LA
    16. Vogelstein B

    (2016) Diagnostic potential of tumor DNA from ovarian cyst fluid

    eLife 5:e15175.

    https://doi.org/10.7554/eLife.15175

    • PubMed
    • Google Scholar
48. 1. Williams KA
    2. Labidi-Galy SI
    3. Terry KL
    4. Vitonis AF
    5. Welch WR
    6. Goodman A
    7. Cramer DW

    (2014) Prognostic significance and predictors of the neutrophil-to-lymphocyte ratio in ovarian cancer

    Gynecologic Oncology 132:542–550.

    https://doi.org/10.1016/j.ygyno.2014.01.026

    • PubMed
    • Google Scholar
49. Book

    1. Witten I

    (2016)

    Data Mining: Practical Machine Learning Tools and Techniques

    Boston, MA: Elsevier.

    • Google Scholar
50. 1. Wu Y
    2. Wei B
    3. Liu H
    4. Li T
    5. Rayner S

    (2011) MiRPara: a SVM-based software tool for prediction of most probable microRNA coding regions in genome scale sequences

    BMC Bioinformatics 12:107.

    https://doi.org/10.1186/1471-2105-12-107

    • PubMed
    • Google Scholar
51. 1. Youden WJ

    (1950) Index for rating diagnostic tests

    Cancer 3:32–35.

    https://doi.org/10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>3.0.CO;2-3

    • PubMed
    • Google Scholar
52. 1. Zhang Z
    2. Bast RC
    3. Yu Y
    4. Li J
    5. Sokoll LJ
    6. Rai AJ
    7. Rosenzweig JM
    8. Cameron B
    9. Wang YY
    10. Meng XY
    11. Berchuck A
    12. Van Haaften-Day C
    13. Hacker NF
    14. de Bruijn HW
    15. van der Zee AG
    16. Jacobs IJ
    17. Fung ET
    18. Chan DW

    (2004) Three biomarkers identified from serum proteomic analysis for the detection of early stage ovarian cancer

    Cancer Research 64:5882–5890.

    https://doi.org/10.1158/0008-5472.CAN-04-0746

    • PubMed
    • Google Scholar
53. 1. Zhang WC
    2. Chin TM
    3. Yang H
    4. Nga ME
    5. Lunny DP
    6. Lim EK
    7. Sun LL
    8. Pang YH
    9. Leow YN
    10. Malusay SR
    11. Lim PX
    12. Lee JZ
    13. Tan BJ
    14. Shyh-Chang N
    15. Lim EH
    16. Lim WT
    17. Tan DS
    18. Tan EH
    19. Tai BC
    20. Soo RA
    21. Tam WL
    22. Lim B

    (2016) Tumour-initiating cell-specific miR-1246 and miR-1290 expression converge to promote non-small cell lung cancer progression

    Nature Communications 7:11702.

    https://doi.org/10.1038/ncomms11702

    • PubMed
    • Google Scholar
54. 1. Zheng H
    2. Zhang L
    3. Zhao Y
    4. Yang D
    5. Song F
    6. Wen Y
    7. Hao Q
    8. Hu Z
    9. Zhang W
    10. Chen K

    (2013) Plasma miRNAs as diagnostic and prognostic biomarkers for ovarian cancer

    PLoS One 8:e77853.

    https://doi.org/10.1371/journal.pone.0077853

    • PubMed
    • Google Scholar
55. 1. Zhu CS
    2. Pinsky PF
    3. Cramer DW
    4. Ransohoff DF
    5. Hartge P
    6. Pfeiffer RM
    7. Urban N
    8. Mor G
    9. Bast RC
    10. Moore LE
    11. Lokshin AE
    12. McIntosh MW
    13. Skates SJ
    14. Vitonis A
    15. Zhang Z
    16. Ward DC
    17. Symanowski JT
    18. Lomakin A
    19. Fung ET
    20. Sluss PM
    21. Scholler N
    22. Lu KH
    23. Marrangoni AM
    24. Patriotis C
    25. Srivastava S
    26. Buys SS
    27. Berg CD
    28. PLCO Project Team

    (2011) A framework for evaluating biomarkers for early detection: validation of biomarker panels for ovarian cancer

    Cancer Prevention Research 4:375–383.

    https://doi.org/10.1158/1940-6207.CAPR-10-0193

    • PubMed
    • Google Scholar
56. 1. Zuberi M
    2. Mir R
    3. Das J
    4. Ahmad I
    5. Javid J
    6. Yadav P
    7. Masroor M
    8. Ahmad S
    9. Ray PC
    10. Saxena A

    (2015) Expression of serum miR-200a, miR-200b, and miR-200c as candidate biomarkers in epithelial ovarian cancer and their association with clinicopathological features

    Clinical and Translational Oncology 17:779–787.

    https://doi.org/10.1007/s12094-015-1303-1

    • PubMed
    • Google Scholar

Author details

1. Kevin M Elias

   1. Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Boston, United States
   2. Harvard Medical School, Boston, United States
   3. Surgical ICU Translational Research Center, Brigham and Women’s Hospital, Boston, United States

   Contribution

   Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1502-5553

2. Wojciech Fendler

   1. Harvard Medical School, Boston, United States
   2. Department of Biostatistics and Translational Medicine, Medical University of Lodz, Lodz, Poland
   3. Department of Radiation Oncology, Dana-Farber Cancer Institute, Boston, United States

   Contribution

   Data curation, Software, Formal analysis, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-5083-9168

3. Konrad Stawiski

   Department of Biostatistics and Translational Medicine, Medical University of Lodz, Lodz, Poland

   Contribution

   Software, Formal analysis

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6550-3384

4. Stephen J Fiascone

   1. Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Boston, United States
   2. Harvard Medical School, Boston, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

5. Allison F Vitonis

   1. Harvard Medical School, Boston, United States
   2. Obstetrics and Gynecology Epidemiology Center, Department of Obstetrics and Gynecology, Brigham and Women’s Hospital, Boston, United States
   3. Department of Epidemiology, Harvard School of Public Health, Boston, United States

   Contribution

   Resources

   Competing interests

   No competing interests declared

6. Ross S Berkowitz

   1. Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Boston, United States
   2. Harvard Medical School, Boston, United States

   Contribution

   Resources, Writing—review and editing

   Competing interests

   No competing interests declared

7. Gyorgy Frendl

   1. Harvard Medical School, Boston, United States
   2. Surgical ICU Translational Research Center, Brigham and Women’s Hospital, Boston, United States
   3. Department of Anesthesiology, Perioperative and Pain Medicine, Brigham and Women’s Hospital, Boston, United States

   Contribution

   Resources

   Competing interests

   No competing interests declared

8. Panagiotis Konstantinopoulos

   1. Harvard Medical School, Boston, United States
   2. Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, United States

   Contribution

   Writing—review and editing

   Competing interests

   No competing interests declared

9. Christopher P Crum

   1. Harvard Medical School, Boston, United States
   2. Division of Women’s and Perinatal Pathology, Department of Pathology, Brigham and Women’s Hospital, Boston, United States

   Contribution

   Resources, Formal analysis

   Competing interests

   No competing interests declared

10. Magdalena Kedzierska

    Department of Clinical Oncology, Medical University of Lodz, Lodz, Poland

    Contribution

    Resources

    Competing interests

    No competing interests declared

11. Daniel W Cramer

    1. Harvard Medical School, Boston, United States
    2. Obstetrics and Gynecology Epidemiology Center, Department of Obstetrics and Gynecology, Brigham and Women’s Hospital, Boston, United States
    3. Department of Epidemiology, Harvard School of Public Health, Boston, United States

    Contribution

    Resources, Supervision, Methodology, Writing—review and editing

    Competing interests

    No competing interests declared

12. Dipanjan Chowdhury

    1. Harvard Medical School, Boston, United States
    2. Department of Radiation Oncology, Dana-Farber Cancer Institute, Boston, United States

    Contribution

    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing

    For correspondence

    dipanjan_chowdhury@dfci.harvard.edu

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-5645-3752

• 8,082

  views

• 1,131

  downloads

• 103

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.