	Ovarian cancer is a relatively rare but deadly disease. Biomarker panels for the detection of ovarian cancer are usually combined with screening for CA125, a known marker of ovarian cancer, and transabdominal or transvaginal ultrasonography. Combining these tests increases the sensitivity, specificity, and positive predictive value. However, there are currently no early detection methods suitable for population-wide screening. Women with ovarian cancer are usually treated with a combination of chemotherapy drugs, including cisplatin or carboplatin and paclitaxel. Biomarkers to identify mechanisms of drug resistance in relapsed ovarian cancer could be used to direct the choice of salvage chemotherapy and improve response rates and overall survival. This article discusses the inverse relationship between cisplatin and paclitaxel resistance and the roles of BRCA1, ERCC1 and TLR-4/ MyD88 signalling as potential predictive biomarkers of chemotherapy-resistant ovarian cancer.

	Worldwide, there are ~204,000 new cases of ovarian cancer each year, accounting for 4% of cancers diagnosed in women.

 According to the National Cancer Institute (NCI), the incidence of ovarian cancer is 13.1 cases for every 100,000 women.

 While this is much lower than the incidence of breast cancer (128.3 cases for every 100,000 women), 5-year overall survival (OS) rates are much worse for ovarian cancer than for breast cancer: 49.5% compared with 89.1%.

	This is partly due to a lack of early detection methods for ovarian cancer, resulting in more than half of ovarian cancer diagnoses occurring after the disease has metastasized. NCI stage distribution statistics show that 15% of ovarian cancer cases are diagnosed while the cancer is still confined to the primary site (localized stage), 17% are diagnosed after the cancer has spread to regional lymph nodes or directly beyond the primary site, 62% are diagnosed after the cancer has already metastasized (distant stage), and staging information is unknown for the remaining 7% of cases.

 The corresponding 5-year relative survival rates are 93.8% for localized, 72.8% for regional, 28.2% for distant, and 27.3% for unstaged.

	The low incidence of ovarian cancer means that the average general practitioner sees relatively few cases. Women with ovarian cancer often experience symptoms of the disease before they are diagnosed, but these symptoms (persistent pelvic and stomach pain, difficulty eating, early satiety, and increased stomach size or persistent bloating) overlap with the symptomatology of many less serious conditions, making it difficult to make a differential diagnosis. The only well-defined risk factors for ovarian cancer include being postmenopausal, having a family history of breast or ovarian cancer, or having a BRCA mutation.

 Due to the low incidence rate, limited known risk factors, and nonspecific symptoms, the disease is easily overlooked. Treating ovarian cancer while it is still in the early stages improves patient survival, making the discovery of new biomarkers to facilitate early detection paramount.

 addressed the challenges general practitioners face regarding ovarian cancer screening.

 Clarke-Pearson presented the case of an asymptomatic 56-year-old woman with no family history of ovarian or breast cancer who asked to be screened for ovarian cancer after a friend died of the disease. A physical examination of the patient found no abnormalities. In discussing whether the practitioner should screen the patient for ovarian cancer, Clarke-Pearson notes that no professional society recommends routine ovarian cancer screening of the general population.

	Making a definitive diagnosis of ovarian cancer requires surgical removal of the abnormal ovary and the related fallopian tube. The author concluded that “the potential harms outweigh the potential benefits,” because the risk of false positives with today’s inexact screening methods could lead to needless surgery if patients were routinely screened for ovarian cancer.

	The prevalence of ovarian cancer is very low in the general population, so any test—whether it is an imaging modality or a biomarker assay&mdash;must have very high specificity and sensitivity to yield a positive predictive value (PPV) sufficiently high to justify widespread screening (

). These metrics must be high enough to identify cancer accurately in women reporting symptoms suggestive of the disease and in asymptomatic women at greater risk for the disease.

	In addition, our understanding of the biological progression of ovarian cancer is fairly limited, and we do not know whether early-stage disease evolves into late-stage disease or if advanced disease arises independently.

 If early- and late-stage disease are biologically diverse, an effective tool for ovarian cancer screening would need to allow clinicians to distinguish between the two. This suggests that developing an accurate method of screening for ovarian cancer is likely to involve biomarkers. The discovery of additional biomarkers for ovarian cancer would increase the specificity and sensitivity (

) of a screening test without the need for more invasive procedures.

	Screening With Cancer Antigen 125 (CA125)

	One biomarker already in use is CA125, also known as MUC16. This is a mullerian duct differentiation antigen that is overexpressed in epithelial ovarian cancer cells and secreted into the blood.

 This membrane-bound protein is a member of the mucin family and thought to play a role in signal transduction and metastasis.

 CA125 is not a suitable biomarker for routine ovarian cancer screening in the general population because other malignancies and some benign gynecologic diseases are also associated with an increase in serum CA125 levels (

). In addition, <50% of patients with stage I ovarian cancer are found to have elevated levels of CA125.

 Women with high levels of CA125 are usually presumed to have ovarian cancer unless a combination of ultrasonography, laparotomy, and histopathology point to another differential diagnosis. Combining these diagnostic tools increases the specificity and sensitivity of the diagnosis.

	A 2005 study by Moss and associates examined the use of CA125 in clinical practice and found that CA125 cancer in only 20% of the 221 women suspected of having a malignancy.

 The authors determined that CA125 screening for ovarian cancer in this population had a sensitivity of 88.6%, but a specificity of only 72.0%. False-positive results were most often attributable to another malignancy (26%), benign ovarian disease (14%), and other benign gynecological conditions (9%).

	Menon and colleagues determined that the utility of CA125 is improved by taking serial measurements over time and applying the risk of ovarian cancer (ROC) algorithm (

 Women were randomized between a control group and a screening group (n = 6682). The women with normal results (ROC <1 in 2000; n = 5213) resumed annual screening. CA125 testing was repeated for women with intermediate results (ROC 1 in 2000 to 1 in 500; n = 1228). As a result of this repeat testing, another 53 women were classified as having elevated risk. These patients plus the 91 with elevated ROC levels (>1 in 500) on initial testing underwent level II screening involving transvaginal ultrasonography. Sixteen women underwent surgery: eleven had benign pathology; one had ovarian recurrence of breast cancer; one had borderline papillary serous ovarian carcinoma; and three had primary invasive epithelial ovarian cancer (EOC). The specificity of this approach for detecting EOC was 99.8% (95% confidence interval [CI], 99.7%-99.9%), and the PPV was 19% (95% CI, 4.1%-45.6%).

 A larger follow-up study is underway that is applying the ROC algorithm to >202,638 women at average risk for ovarian cancer. Results are not expected until 2011.

	Studies have investigated transabdominal and transvaginal ultrasonography as strategies for detecting changes in the ovaries that (1) precede the onset of symptoms and (2) are undetectable by pelvic examination.

 Results have been inconsistent, and there is scant evidence to support the use of ultrasonography alone in this setting.

	In a trial involving asymptomatic women self-referred for ultrasonography, those with an abnormal ovary underwent surgery; the PPV for this study was 1.5%.

 Another study of asymptomatic women included only those considered high risk, with a family history of ovarian cancer. In this trial, ultrasonography had a PPV of 14%.

 The marked disparity in PPV between these trials is likely attributable to the differences in their respective patient populations, with high-risk patients yielding a higher PPV.

	The most promising study involved a multimodal approach, combining transvaginal ultrasonography with testing for ovarian cancer biomarkers such as CA125.

 Women were randomized to receive testing with both methods or with transvaginal ultrasonography alone. The PPV for the multimodal arm was a relatively high 43.3% compared with 5.3% for the cohort that received only ultrasonography. These results are particularly encouraging because the investigators excluded women with a family history of ovarian cancer from the study. Inclusion of this category of subjects has contributed to higher PPVs in other studies.

	None of the published trials on ultrasonography screening for ovarian cancer included a control group of women who did not receive screening. Therefore, it is not possible to determine whether using the modality contributed to improved OS in these studies.

	Researchers have used various proteomic techniques in ovarian cancer cell lines and patient samples to identify biomarkers that appear to correspond with early-stage ovarian cancer. Combining the power of multiple biomarkers into one biomarker panel for ovarian cancer screening might produce a test that is specific and sensitive enough to be highly beneficial and cost effective. The use of biomarker panels usually relies on the interpretation of the pattern of the different markers in relationship to one another rather than relying on the absolute level of each marker.

	Although there have been many retrospective studies of biomarker panels in ovarian cancer, they have yet to be used widely to diagnose ovarian cancer in the general population. Studies of biomarker panels generally show a 5% to 10% increase in sensitivity compared with CA125 alone, but the panels have also been associated with decreased specificity.

	In a study using a panel incorporating CA125, leptin, prolactin, osteopontin, insulin-like growth factor II, and macrophage migration inhibitory factor, the presence or absence of disease was correctly determined in 221 out of 224 samples (98.7%).

 While this particular panel of biomarkers proved very successful, controversy arose over the study’s methodological approaches in calculating specificity, sensitivity, and PPV.

 The study used a training set of samples taken from healthy individuals or women with ovarian cancer to define the biomarkers, which was followed by a testing set of blinded samples to validate the identified biomarkers. Data obtained from both sets were included in the analyses, rather than data only from the testing set.12 In calculating the PPV, the investigators used a population prevalence of ~50%, rather than 0.04%. This resulted in a claimed PPV of 99.2%, when 6.5% was a more realistic calculation. A PPV of 6.5% means that only 1 in 15 women with a positive test result would actually have ovarian cancer (the remaining 14 would be false positives).

	If this test were used in practice, the vast majority of women tested would suffer needless anxiety and undergo unnecessary diagnostic procedures, potentially resulting in the removal of one or both ovaries and fallopian tubes.

 There are clearly many challenges in translating biomarkers from ovarian cell lines and clinical samples into a clinical test that can be used to detect ovarian cancer in the general population or in patients reporting symptoms.

	The detection of relapsed ovarian cancer is a simpler proposition, because there is less likelihood of benign disease acting as a confounding factor. We already know the patient had ovarian cancer; the question is whether it has recurred. Serum CA125 levels are commonly used to monitor patients for disease progression or relapse. A spike in CA125 levels can potentially indicate recurrence prior to the patient experiencing any symptoms.

	Researchers compared the accuracy of CA125 to a biomarker panel of human epididymis protein 4, glycodelin, and matrix metalloproteinase 7 in detecting recurrence.

 They evaluated 260 samples from 30 patients with ovarian cancer who were monitored longitudinally after diagnosis. The biomarker panel had 100% sensitivity in predicting recurrence compared with 96% for CA125. At least one of the biomarkers in the panel was found to increase earlier (range, 6-69 wk) than CA125, and this increase occurred prior to clinical evidence of recurrence in 14 out of 27 patients (52%).

	It is unclear whether the early detection of relapsed ovarian cancer leads to improved survival or increases the patient’s quality of life. A study presented by Rustin and associates at the American Society of Clinical Oncology Annual Meeting in 2009 reported that patients in remission after first-line platinum-based therapy did not benefit from earlier second- and third-line treatments compared with patients for whom treatment was delayed until symptoms occurred. CA125 levels were measured for all patients every 3 months. Patients whose levels exceeded twice the upper limit of normal were randomized to immediate second-line treatment or to continuing with blinded CA125 measurements, commencing treatment only after experiencing symptoms of recurrence. Rustin and colleagues said there was no difference in OS or quality of life between the immediate- and delayed-treatment arms (

 = .091). Further, patients who responded well to second-line therapy but subsequently relapsed received third-line chemotherapy a median of 4.6 months earlier, impairing quality of life without providing a survival benefit.

	Using Prognostic Markers to Guide Therapy

	Platinum chemotherapy or chemotherapy regimens containing a platinum and a taxane are standard first-line therapies for ovarian carcinomas. Initial response to a platinum-containing regimen is high, but up to 80% of patients eventually relapse and become platinum and/or taxane-resistant. Using Markman’s criteria, clinical platinum resistance is defined as disease progression after a platinum-free interval of <6 months.

 Paclitaxel resistance is not clearly defined in the literature, but some have described it in similar terms to platinum resistance: disease progression following a taxane-free interval of <6 months.

	Due to the different mechanisms of action for platinums and taxanes, drugs from these two classes are often combined in cancer therapy. Work in cell lines, however, suggests alternating rather than concurrent administration of the two classes might be beneficial. In a systematic review published in 2007, we examined cross-resistance between cisplatin and paclitaxel in cell models of acquired drug resistance.18 The majority of cell models demonstrated an inverse-resistance relationship between cisplatin and paclitaxel. When cisplatin resistance occurred, 68% of cell lines were not resistant to paclitaxel and some were even more sensitive to paclitaxel. The reverse was true of paclitaxel-resistant cell lines, with 66% having no resistance or increased sensitivity to cisplatin. This suggests that nearly two-thirds of cancer patients might benefit from a treatment plan that alternates cisplatin and paclitaxel. The challenge is how to identify which patients will respond well to alternating therapy between cisplatin and paclitaxel. This is because while two-thirds of cancer patients may respond well to this treatment strategy, the other third would respond poorly and would need to be treated differently.

	We have also analyzed whether the inverse relationship we observed between cisplatin and paclitaxel resistance in cell lines was apparent in the clinic. Using a systematic literature search, we compared studies of paclitaxel versus oxaliplatin salvage therapy in platinum-resistant ovarian carcinoma. The cisplatin-resistant patients had a higher response rate when treated with single-agent paclitaxel than with oxaliplatin; 22% of the 1918 cisplatin-resistant patients treated with paclitaxel responded, whereas only 8% of the 91 cisplatin-resistant patients treated with oxaliplatin responded (

 The better response rates seen with the non—cross-resistant agent paclitaxel versus the cross-resistant  agent oxaliplatin correlate with the in vitro data. This suggests that the inverse-resistance relationship seen in cell models may be present in the clinic, as there is a significantly higher percentage of patients with platinum-resistant ovarian cancer who respond when treated with paclitaxel.

	Surprisingly, the data also showed that platinum-resistant patients who had received paclitaxel previously (n = 232) responded better to salvage therapy with single-agent paclitaxel (relative risk [RR], 35.3%) than the paclitaxel-naïve patients (n = 1918; RR, 22.7%; 

 Patients in both cohorts were of similar age, had similar performance status and FIGO stage, and received a similar number of prior chemotherapy cycles. There were differences in histology, but they did not account for the difference in response rates to paclitaxel salvage therapy. Usually, if patients have received a drug and experienced disease progression, they are less likely to respond to therapy with subsequent exposure to the same drug. Although one must be cautious in interpreting these summary findings due to the potential for biases in pooling of patients across studies, if the findings do reflect the true clinical response to these agents, they suggest that initial co-treatment with platinum and paclitaxel may improve the outcome of paclitaxel salvage therapy. We are currently preparing a Cochrane review on this topic.

	To find genetic modifications that might be responsible for the inverse-resistance relationship between cisplatin and paclitaxel, we performed another review of the cell lines, with the goal of identifying biomarkers potentially indicative of this phenotype.

 Our review found many targets, but the gene with the most published data linking it to the inverse-resistance pattern was BRCA1. When BRCA1 expression is artificially increased by transfection, cisplatin resistance and paclitaxel sensitivity occur

; conversely, when BRCA1 expression is decreased by small interfering RNA (siRNA), this leads to paclitaxel resistance and cisplatin sensitivity.

	A germ line mutation in one BRCA1/2 allele is associated with a high risk of developing breast and ovarian cancer. Cells carrying heterozygous loss-of-function BRCA mutations can lose the remaining wild type allele, resulting in deficient homologous-recombination DNA repair. This leads to the genetic aberrations that drive carcinogenesis.

	If there is any positive side to a BRCA1 mutation, it is that patients who have it often have better response to DNA-damaging chemotherapy agents such as cisplatin because the capacity of their tumor cells to repair DNA damage is diminished. This contributes to higher survival rates. Ovarian cancer patients with BRCA1 mutations, treated with at least 2 courses of cisplatin-based chemotherapy, have a higher 5-year survival rate compared with age- and treatment-matched patients with sporadic ovarian cancer (78.6% vs 30.3%, respectively).

 Other studies of patients with ovarian cancer also demonstrate this pattern of increased OS in BRCA1 mutation carriers.

 Although ovarian cancers with mutated BRCA1 are initially more sensitive to platinum compounds, they will eventually develop platinum resistance. One of the possible mechanisms for this resistance is the occurrence of additional mutations in the BRCA1 gene that restore the reading frame and functionality of the protein.

	Swisher and associates found that of the 6 recurrent platinum-resistant ovarian tumors in their study, 4 had developed secondary mutations in BRCA1 that restored the reading frame. None of the 3 platinum-sensitive recurrent tumors developed BRCA1 sequence alterations.

 A similar study involving the BRCA2-mutated pancreatic cancer cell line Capan-1 showed that restoration of the mutation to wild type mediated platinum-resistance

	Evidence is increasing of an important role for altered protein expression of BRCA1 as a prognostic marker in the sporadic forms of both ovarian and breast cancer. The level of BRCA1 expression plays a predictive role in the response to chemotherapy.

 Carser and colleagues found that patients with BRCA1 immunohistochemistry (IHC)—negative tumors had a tendency toward improved OS compared with patients whose tumors were BRCA1 IHC positive (41.5 mo vs 31.2 mo, respectively; 

 = .160). It was also discovered that patients with BRCA1-negative tumors had similar outcomes with platinum chemotherapy and combination platinum/taxane chemotherapy. In contrast, patients with BRCA1 IHC—positive tumors had better OS with combined platinum/taxane chemotherapy (n = 118) than with platinum chemotherapy alone (n = 110; OS, 41.7 mo vs 19.8 mo, respectively; 

 = .0004). BRCA1 mutation status and expression levels therefore have the potential to be used as predictive markers for response to primary and salvage chemotherapy with platinum and taxane chemotherapy.

	Platinum chemotherapy causes cytotoxicity by binding to DNA, forming adducts that become “roadblocks” for the progression of RNA and DNA polymerases. This causes cell cycle arrest and apoptosis. Excision repair cross-complementation group 1 (ERCC1) is a DNA repair protein involved in the nucleotide excision repair removal of platinum adducts from DNA.

 Increased expression of ERCC1 has been associated with cisplatin resistance in ovarian cancer cell lines.

	Using IHC staining, Steffensen and colleagues found that 13.9% of tumor samples from 101 women with newly diagnosed ovarian cancer overexpressed ERCC1. Of the 12 patients who were platinum resistant, 9 (75%) overexpressed ERCC1, compared with only 16 of 60 (26.7%) platinum-sensitive patients (

 = .0013). They also found a significant difference in progression-free survival (PFS) favoring the ERCC1-negative patients (

 Other studies have found no significant correlation between ERCC1 expression and platinum resistance in ovarian cancer. In one study, Stadlmann and associates found that 5 out of 18 (27.8%) platinum-resistant

	patients overexpressed ERCC1 compared with 7 out of the 41 (17.8%) platinum-sensitive patients (

	This illustrates the challenge of identifying biomarkers for platinum resistance, as there are many mechanisms of platinum resistance that have been characterized in cellular models,

 only one of which is an increase in DNA repair, which can be mediated by many proteins, not only ERCC1. In fact, work in our laboratory in H69 small cell lung cancer cells showed decreased protein expression of ERCC1 in response to cisplatin and oxaliplatin treatment. This was in association with the formation of a lower molecular weight splice variant that has been hypothesized to be associated with decreased DNA repair activity.

 Transfection of a splice variant of ERCC1 missing exon 8 has also been shown not to induce cisplatin resistance in ovarian cancer cells, whereas transfection of wild-type full length ERCC1 does induce resistance.

	Polymorphisms in ERCC1 have been shown to influence patient outcomes to platinum/taxane combination chemotherapy in ovarian cancer. The C8092A polymorphism was found to be an independent predictor of PFS and OS in women with optimally resected EOC; patients with a C/C genotype had significantly longer PFS and OS than those with a C/A or A/A genotype.

 The ERCC1 codon 118 polymorphism was found to be predictive of initial response to platinum/taxane chemotherapy in ovarian cancer, but not disease progression or OS.

	A combined analysis of ERCC1 microRNA expression and codon 118 polymorphism could potentially be used to determine which patients are likely to respond to combination chemotherapy with a platinum and a taxane or single-agent platinum chemotherapy. Ovarian cancer patients with high ERCC1 expression or the C/C genotype at codon 118 may benefit from the combination of platinum and paclitaxel, while those with low ERCC1

	expression or the C/T or T/T genotype may respond well to a platinum agent without paclitaxel.

	The conflicting data on ERCC1 illustrate the challenge of identifying biomarkers for platinum resistance. ERCC1 is potentially a useful biomarker for chemoresistance in ovarian cancer, but it needs to be analyzed at DNA, RNA, and protein levels to get a full picture of the functionality of the protein and its potential role in resistance.

	The TLR-4/MyD88 signaling pathway has recently been proposed as a risk factor for both carcinogenesis and chemoresistance in ovarian cancer 

 Cancer cells appear capable of producing a microenvironment that promotes cell proliferation through activation of the TLR-4/MyD88 pathway. Nuclear factor kB activation results in secretion of molecules that promote inflammation and proliferation, such as tumor necrosis factor-alpha, interleukins, vascular endothelial growth factor, and other factors.

	While TLR-4 expression is ubiquitous in EOC cells, a subgroup of these cells differentially expressing MyD88 has demonstrated enhanced cytokine/chemokine production and cellular proliferation upon activation of TLR-4.

 This differential MyD88 expression is the basis of the proposal by Chen and associates to classify EOC cells as type 1 and type 2.44 Type 1/MyD88-positive EOC cells have a functioning TLR-4/MyD88 pathway and are possibly indicative of an ovarian cancer stem cell that is highly resistant to pro-apoptotic signaling. In contrast, type 2/MyD88-negative EOC cells lack this functioning pathway and might suggest more differentiated tumors that are more responsive to apoptosis and therefore less biologically aggressive. Recent work has also demonstrated that expression of MyD88 in EOC is associated with significantly shorter PFS.

	TLR-4/MyD88 signaling also appears to be a significant factor in chemoresistance. It is known that paclitaxel is a ligand of TLR-4 and, as such, can induce activation of the TLR-4/MyD88 pathway within type 1 EOC cells.

 When type 2 EOC cells are transfected with MyD88, making them “type 1” cells, resistance to paclitaxel is induced.

 The reverse effect has also been shown, with siRNA and vector-mediated knockdown of TLR-4 in ovarian cancer cell lines causing paclitaxel sensitivity.

 Interestingly, resistance to carboplatin was not induced by increased expression of MyD88.

 This seems to indicate that the type 1/type 2 hypothesis is also consistent with the inverse resistance between platinum and taxane chemotherapy described above.

	It appears likely that although paclitaxel treatment causes apoptosis in more differentiated, type 2/MyD88-negative epithelial ovarian cancer cells, the therapy might be detrimental in patients with type 1/MyD88-positive cells. It is also possible that recurrent disease represents a pool of cancer stem cells that have been “selected out” by paclitaxel treatment. The evidence suggests that TLR-4/MyD88 signaling has the potential to be used as a predictive marker for salvage chemotherapy with paclitaxel.

	Ovarian cancer is often not detected until it is advanced, leading to a high rate of mortality for this disease. Clinical experience shows that treating ovarian cancer improves OS, but currently available screening methods lack the specificity, sensitivity, and PPV to be applied to the general population. They have an unacceptably high rate of false positives, leading to unnecessary surgical procedures. As a result, no professional organization has issued guidelines recommending any type of routine screening for the general population. New biomarker discoveries in ovarian cancer appear to offer the best hope for developing an effective system for screening women before they become symptomatic. Ensuring high levels of specificity, sensitivity, and PPV will likely require developing a biomarker panel that is combined with CA125 screening and transvaginal ultrasonography.

	Biomarkers also show potential for use to screen for ovarian cancer and as predictors of chemoresistance. Evidence suggests they might have a future role in guiding initial and salvage chemotherapy. Analyses of mutation and polymorphism status, as well as gene and protein expression, are needed to help us understand how a given gene mediates chemoresistance. Two categorization systems have already been proposed (type 1 and type 2) based on the activity of the TLR-4/MyD88 signaling pathway and the inverse resistance/cross resistance between platinum and taxane chemotherapy, which may involve alterations in BRCA1. Understanding the interaction and overlap between these two hypotheses will allow us to define subcategories of drug-resistant ovarian cancer and allow a shift toward individualized chemotherapy for relapsed patients.

	This research was supported by a Marie Curie International Incoming Fellowship (Dr Britta Stordal) from the European Union FP7 program.

	Britta Stordal, PhD, is a postdoctoral fellow with the National Institute for Cellular Biology, Dublin City University, Ireland. John O’Leary, MD, PhD, is head of the Department of Histopathology, at The Coombe Women’s Hospital affiliated with Trinity College Dublin, Ireland.

	Dr. Britta Stordal, National Institute for Cellular Biotechnology, Dublin City University, Glasnevin, Dublin 9, Ireland, or e-mail 

Articles

Worldwide, there are ~204,000 new cases of ovarian cancer each year, accounting for 4% of cancers diagnosed in women.

Britta Stordal, PhD

Articles

Biomarkers in Ovarian Cancer: Early Detection and Chemoresistance