Biomarkers in Medicine

By Pınar Atukeren and Hafize Uzun

Biomarkers in Medicine is a comprehensive guide to understanding the current and future status of biomarkers. The book features 27 chapters focusing on disease biomarkers for diseases such as cancer, neurodegenerative diseases, cardiac diseases, metabolic conditions and much more.

This book supplies readers with the unique insight of experts in multiple specialties in medicine and life sciences who have extensive experience in diagnostics and clinical laboratories. The book includes case studies and practical examples from different classes of biomarkers on different platforms, including new data for biomarkers in different therapeutic indications. In addition to presenting biomarker information, each chapter covers the relevant pathology and also emphasizes on preclinical and clinical manifestation of the disease process.

Clinicians managing patients or clinical trials, clinical researchers, clinical laboratories, diagnostic companies, regulatory agencies, medical school graduate students, academic students, and the general public involved in healthcare delivery will all benefit from information presented in this book.

• Language: English
• Publisher: Bentham Science Publishers
• Release date: Sep 1, 2022
• ISBN: 9789815040463

Book preview

New Biomarkers and Immunotherapy Decision

Abdurrahman Yigit¹, Berkay Kuscu², Ali Kirik², Ruhsen Ozcaglayan², Cigdem Usul Afsar³, *

¹ Department of Internal Medicine and Medical Oncology, Canakkale 18 Mart University Medical Faculty, Canakkale, Turkey

² Department of Internal Medicine, Balikesir University Medical Faculty, Balıkesir, Turkey

³ Department of Internal Medicine and Medical Oncology, Istinye University Medical Faculty, İstinye Üniversitesi Topkapı Kampüsü, Maltepe Mah., Teyyareci Sami Sk., No.3 Zeytinburnu, Istanbul 34010, Turkey

Abstract

As immune checkpoint blockade and other immune-based therapy approaches lead to broad treatment advances among patients with advanced cancer, an important consideration is how to best select patients whose tumors will respond to these therapies. As a consequence predictive and prognostic markers are needed. There are genomic features, such as tumour mutation burden (TMB), microsatellite instability (MSI), and immune phenotype features, such as programmed death-ligand 1 (PD-L1), CTLA-4 and tumour infiltrating lymphocytes (TILs), to predict response to immunotherapies (ITs). Several studies show the correlation between TMB and predicted neoantigen load across multiple cancer types. Response to immune checkpoint inhibitors is higher in tumours with high TMB. The candidate biomarker that has been studied mostly other than TMB is PD-L1 expression in trials utilizing programmed cell death-1 (PD-1) blockade. PD-L1 and PD-1 expression are dynamic markers that change in relation to local cytokines and other factors, and the thresholds that separate positive and negative PD-L1 expressions remain under debate. PD-L1 expression is now a routine diagnostic marker for patients with newly diagnosed NSCLC. The potential applicability of PD-L1 in other disease settings is still uncertain. Microsatellite instability is characterised by high rates of alterations to repetitive DNA sequences caused by impaired mismatch repair (MMR); MSI was the biomarker was approved according to tumor's initial location. Combining TMB with specific genomic alterations is crucial. Moreover, new biomarkers are being investigated.

Keywords: Checkpoint inhibitor, Immunotherapy, MSI, PD-L1, PD1, Predictive , Prognostic , TIL, TMB.

---

* Corresponding author Cigdem Usul Afsar: Department of Internal Medicine and Medical Oncology, Istinye University Medical Faculty, İstinye Üniversitesi Topkapı Kampüsü, Maltepe Mah., Teyyareci Sami Sk., No.3 Zeytinburnu, Istanbul 34010, Turkey; Tel: [phone]; E-mails: [e-mail], [e-mail]

NEW BIOMARKERS AND IMMUNOTHERAPY DECISION

Principles of Cancer İmmunotherapy

Immunotherapy (IT) is a type of biological therapy. Biological therapy is a type of treatment that uses cytokines made from living organisms to treat cancer. What is expected is that the immune system detects and destroys the abnormal cells and likely inhibits the growth of many cancers. For example, immune cells are sometimes found in and around tumors. These cells, called tumor-infiltrating lymphocytes or TILs, are a sign that the immune system is responding to the tumor. People whose tumors contain TIL generally do better than people who do not have tumors [1].

Even though the immune system can prevent or slow cancer growth, cancer cells have ways to avoid destruction mediated by the immune system. Cancer cells may have genetic changes that make them less visible to the immune system; they also have proteins on their surface that turn off immune cells and change the normal cells around the tumor, thus interfering with how the immune system responds to the cancer cells [2]. Immunotherapy helps the immune system to act better against cancer.

Tumor Immunology

Cell types involved in tumor recognition and rejection — An effective and specific cytotoxic immune response against a tumor requires a complex, serially evolving interaction between various immune cell types in the adaptive and innate immune systems. The Th1/Th2 subclasses of CD8+ lymphocytes and CD4+ T lymphocytes are conventionally referred to as cytotoxic T cells and helper T cells. CD8+ and CD4+ lymphocytes initiate the distinction between self and non-self antigens through recognition at the immune synapse. Natural killer (NK) cells do not require antigen presentation by the major histocompatibility complex (MHC) for cytotoxic activity. In fact, NK cells target cells with low MHC class 1 expression for degradation. Like T cells, NK cells express numerous inhibitory molecules, notably the very different lethal immunoglobulin-like receptor (KIR) subtypes [3].

Additional cell types, such as FoxP3+, CD25+, CD4+, T regulatory (Treg), and myeloid-derived suppressor cells (MDSCs), largely inhibit cytotoxic T lymphocyte activity [4, 5]. Th17 cells, subsets of CD4+ T cells that secrete interleukin (IL)-17, are implicated in autoimmunity and cancer [6]. Macrophages differentiate into at least two different phenotypes: M1 macrophages, which release interferon (IFN) gamma and are responsible for phagocytosis, and M2 macrophages, which release cytokines, such as IL-4, IL-10, transforming growth factor-beta (TGF-beta), and dampen inflammatory responses and foster tolerance [7]. The immune synapse, the most widely studied phenomenon in immunologic surveillance, is the ability of T lymphocytes to distinguish self versus non-self antigens, which are presented by antigen-presenting cells (APCs) such as dendritic cells. Overall, the cytotoxic activity of a CD8+ T cell is regulated by the presence and spatial orientation of a set of stimulatory and inhibitory receptors whose expression is regulated by a myriad of cytokines. Together, this configuration is often referred to as the immune synapse.

Therapeutic Approaches

A number of therapeutic approaches are being studied to unleash the immune system and control malignancy. These approaches include cytokines, T cells (checkpoint inhibitors, agonism of costimulatory receptors), manipulation of T cells, oncolytic viruses, therapies directed at other cell types, and vaccines.

Cytokines — Early approaches to immunotherapy exploited the numerous diverse effects of cytokines and other substances that affect immune cell activity. Examples include Interleukin (IL)-2, which was originally discovered as a T cell growth factor. IL-2 has pleiotropic effects on both cytotoxic T cell function and T-regulatory (Treg) cell maintenance. Effects vary in part with the dose and timing of IL-2 administration [8]. Although IL-2 use has been largely supplanted by the use of checkpoint inhibitors, bolus high-dose IL-2 achieved durable objective responses in a minority of patients with melanoma and renal cell carcinoma (RCC), serving as proof of principle that the immune system could eliminate cancer cells [9, 10].

Immune Checkpoint Inhibitors (ICI)

1) PD-1 and PD Ligand 1/2: Programmed cell death 1 (PD-1) is a transmembrane protein expressed on T cells, B cells, and NK cells. It is an inhibitory molecule that binds to the PD-1 ligand (PD-L1; also known as B7-H1) and PD-L2 (B7-H2). PD-L1 is expressed on the surface of multiple tissue types, including many tumor cells, as well as hematopoietic cells; PD-L2 is more restricted to hematopoietic cells. The PD-1:PD-L1/2 interaction directly inhibits apoptosis of the tumor cell, promotes peripheral T effector cell exhaustion, and promotes the conversion of T effector cells to Treg cells [11, 12]. Based upon prolonged overall survival in phase III trials and durable responses in phase II studies, antibodies inhibiting PD-1 (pembrolizumab, nivolumab) and PD-L1 (atezolizumab, avelumab, durvalumab) have been approved for a number of clinical indications and are being evaluated in multiple other malignancies; these are discussed in specific disease-related topics.

2) CTLA-4: CTLA-4 exerts its effect when it is present on the cell surface of CD4+ and CD8+ T lymphocytes, where it has a higher affinity for the costimulatory receptors CD80 and CD86 (B7-1 and B7-2) on antigen-presenting cells (APCs) than the T cell costimulatory receptor CD28 [13]. The expression of CTLA-4 is upregulated by the degree of T cell receptor (TCR) activation and cytokines such as IL-12 and IFN gamma, forming a feedback inhibition loop on activated T effector cells. As a result, CTLA-4 can be broadly considered a physiologic brake on the CD4+ and CD8+ T cell activation that is triggered by APCs. The anti-CTLA-4 antibody ipilimumab was the first immune checkpoint inhibitor to be approved based upon its ability to prolong survival in patients with metastatic melanoma [14]. Ipilimumab has also been approved as adjuvant therapy for high-risk melanoma as an alternative to IFN.

Vaccines — There is a long history of using adaptive immune recognition of a cancer-associated antigen to influence antitumor responses. Vaccination methods are very diverse, and a full review is beyond the scope of this article. The only vaccine-based therapy currently approved for advanced cancer is sipuleucel-T, an autologous dendritic cell preparation targeted to prostatic acid phosphatase (PAP), which has shown an overall survival benefit in men with castrate-resistant prostate adenocarcinoma [15].

THE PREDICTIVE AND PROGNOSTIC MARKERS OF IMMUNOTHERAPY

As immune checkpoint blockade and other immune-based therapy approaches lead to broad treatment advances among patients with advanced cancer, an important consideration is how to best select patients whose tumors will respond to these therapies. There are needs for predictive and prognostic markers. There are genomic features such as tumour mutation burden (TMB) and microsatellite instability (MSI) and immune phenotype features such as programmed death ligand 1 (PD-L1), CTLA-4 and tumour infiltrating lymphocytes (TILs) to predict response to immunotherapies (ITs). The most important factors of response prediction to ICIs are summarized in Table 1.

Table 1 Response prediction to ICIs.

(TMB: tumor mutation burden; PD-L1: programmed death ligand 1; HLA: human leucocyte antigen; TGF: transforming growth factor; ICI: immune checkpoint inhibitor).

TMB: Tumours with a high number of mutations are more likely to produce new, abnormal proteins, and present them as so-called neoantigens, which can potentially be recognised and targeted by the immune system [16, 17].

Across many cancers, higher TMB is associated with higher levels of predicted neoantigens [16, 18]. Several studies show the correlation between TMB and predicted neoantigen load across multiple cancer types [19-22]. Response to immune checkpoint inhibitors is higher in tumours with high TMB. In other words, tumors with high levels of TMB (e.g., sun-exposed cutaneous melanoma, NSCLC, bladder cancer, and microsatellite-unstable colorectal carcinomas) can benefit more from immune checkpoint blockade [16, 17, 20, 23, 24]. Higher TMB across cancer types was associated with increased mutagen exposure (eg, lung cancer, melanoma), alterations in specific genes, including those involved in DNA mismatch repair/replication and increased age [25, 26].

Approximately 16.4% of cancers had a TMB greater than 10 mutations/Mb, and 7.3% had a TMB greater than 20 mutations/Mb [27]. TMB are comparable to prior estimates of TMB in other cancer cohorts, including whole-exome studies [27]. TMB stratification lacks standardisation [28, 29] and cut-off points determining high and low TMB vary between cancer types [30] and between studies of the same cancer type [29]. Variability in TMB cut-offs places importance on clinical validation for all TMB assays in each indication [28].

There is an evolution of TMB as an immunotherapy biomarker over the last several years. In CheckMate 026, 032, 038, 275, 227 trials and Keynote 012, 028 trials, it has been shown that higher TMB is associated with a better response to IT [26]. In a study evaluating the association between atezolizumab efficacy and TMB in NSCLC patients for 2nd line therapy, enroled in POPLAR, BIRCH or FIR trials showed that treatment benefit of atezolizumab (as measured by OS, PFS and ORR) was associated with high TMB [31]. This association occurred in both unselected and PD-L1-selected patients. PD-L1 expression and TMB were both independent predictors of treatment response [31].

In a retrospective study assessing association of TMB and response to various immunotherapies in 151 patients representing 21 different cancer types which were subgrouped into melanoma, NSCLC (58%) or other cancers (42%) revealed that high TMB was independently associated with better outcome parameters (PFS and RR), regardless of subgroup. Of patients with melanoma or NSCLC treated with anti-PD-1 / PD-L1 monotherapy, 44% with high TMB, compared to 5% with low TMB, experienced a complete or partial response to treatment (p = 0.0023) [32].

PD-L1: The candidate biomarker that has been studied mostly other than TMB is PD-L1 expression in trials utilizing programmed cell death-1 (PD-1) blockade. PD-L1 and PD-1 expression are dynamic markers that change in relation to local cytokines and other factors, and the thresholds that separate positive and negative PD-L1 expression remain under debate. Still, most trials with either retrospective or prospective assessments of PD-L1 status have shown trends for increased response rates to PD-1 blockade in PD-L1 positive tumors [33-36]. Most notably, in patients with newly diagnosed advanced non-small cell lung cancers (NSCLCs) with ≥50 percent PD-L1 expression who were randomized to pembrolizumab or chemotherapy, those randomized to pembrolizumab had a significantly improved objective response rate, progression-free survival, and overall survival [37]. On the basis of this trial, PD-L1 expression is now a routine diagnostic marker for patients with newly diagnosed NSCLC. The potential applicability of PD-L1 in other disease settings is still uncertain.

MSI: Microsatellite instability (MSI) is characterised by high rates of alterations to repetitive DNA sequences caused by impaired mismatch repair (MMR) and MSI was the first instance of the FDA approving a drug ‘based on a tumour's biomarker without regard to the tumour's original location [38, 39]. Thirty-one % of MSI-high mCRC patients who progressed on prior therapy achieved an objective response to nivolumab [40]. TMB measurement captures most MSI-high patients. For example; the vast majority (83%) of MSI-high samples were also TMB-high however the opposite was not true, with only 16% of TMB-high specimens also MSI-high [25, 41].

Therefore, combining PD-L1 with emerging biomarkers to address current limitations is spatial. PD-L1 protein expression by immunohistochemistry (IHC) can predict response to PD-(L)1 inhibitors in a variety of tumour types, but is not absolute. Factors potentially limiting sensitivity and specificity of available IHC assays are; assay performance and positivity threshold, spatial and temporal expression heterogeneity, adaptive vs constitutive expression, limited sampling and alternate PD-1 / PD-L1 ligand / receptor activity [42]. So, the emerging biomarkers to potentially address limitations are TILs; high TMB and gene expression signatures (GES) [43].

TMB correlates only weakly with PD-L1 tumour expression in patients with NSCLC [44]. PD-L1 status and TMB showed only a weak correlation, with 60% of PD-L1 negative samples exhibiting high TMB [44].

Studies support TMB and PD-L1 as independent biomarkers for response to immunotherapy in NSCLC [45, 46]. However, PD-L1 expression and TMB are not significantly correlated within most cancer subtypes. PD-L1 expression and TMB may each inform the use of immune checkpoint inhibitor therapy but treatment responsiveness may be regulated by different mechanisms between these biomarkers [27].

Combining all these data; we can conclude that TMB-high identifies a population independent of PD-L1 status and additionally TMB and PD-L1 are complementary biomarkers carrying independent information.

TMB can be integrated with TIL measurement to predict pre-treatment prognosis. TMB-high is associated with immune cell infiltration in many cancers [47]. Only in TMB-high patients, TILs are a prognostic marker in breast cancer. 100% of TMB-high/TIL-high patients survived 10 years, regardless of tumour subtype [48]. For urothelial carcinoma, TMB and TILs are biomarkers which correlate with disease outcome at least partially independently. Therefore, integration of TMB and TILs could aid in predicting diagnosis [49]. TILs can indicate an immunologically ‘hot’ tumour which may respond better to immunotherapy.

It is important that no patient with an advanced cancer and an established clinical rationale for use of an immune therapy agent should be refused immune therapy on the basis of lack of PD-L1 expression or any other investigational biomarker. Additional gene-expression-based signatures for immune response are also under active investigation and combining TMB with either TIL or PD-L1 measurement enriches for patients who may respond to immune checkpoint inhibitors [50, 51].

Combining TMB with specific genomic alterations is crucial. Development of tailored immunotherapy approaches may be facilitated by combining TMB and/or PD-L1 measurements with genomic alterations associated with response or resistance to immunotherapy. For instance; STK11 is associated with poor response to immunotherapies [46, 52, 53]. EGFR is a potential marker of resistance in immunotherapy; MET is linked with increased immune infiltration and activation phenotype [53] and BRAF is associated with prolonged time on immunotherapies [53]. STK11/LKB1 co-mutations are associated with inferior ORR with PD-1 blockade in KRAS-mutant, TMB-high lung cancer [52].

TMB is relevant for identifying patients to combine immune checkpoint inhibitors. High TMB has shown to correlate with efficacy of PD-1 plus CTLA-4 inhibition in lung cancers and their distinct mechanisms of action explains their synergistic effect [54-56]. TMB and PD-L1 are independent variables. Additionally benefit with combination immune checkpoint inhibitors was independent of PD-L1 expression in the high TMB patient population and TMB associated strongly with efficacy of immunotherapies [54, 55]. These data are further supported in a phase 3 non-small cell lung cancer trial (Checkmate 227) comparing first-line nivolumab plus ipilimumab with chemotherapy in NSCLC [57]. In the low TMB (< 10 mutations / Mb) patient population PFS does not differ significantly between treatment arms (low vs high PD-L1) [57].

In CheckMate-032 trial which evaluated combination immune checkpoint inhibition in small-cell lung cancer (SCLC) patients with high TMB; 1-year survival almost doubled in TMB-high patients treated with combination vs monotherapy [55]. PD-L1 was positive among 10-12% of patients and not predictive of response rate. However, TMB has a role as a biomarker in SCLC and may help identify patients most likely to benefit from combination immunotherapy [55]. In phase 3 CASPIAN and Keynote 604 studies which evaluated the addition of IT to chemotherapy in SCLC patients, the benefit of IT on survival was independent of the PD-L1 status [58, 59].

Immune checkpoint blockade (ICB) provides clinical benefits to a subset of patients with cancer. ICB using inhibitors of programmed death-1 or its ligand (anti-PD-1/PD-L1) has emerged as an effective therapy for many cancer types. However, a minority of patients (<20%) respond to treatment or have durable clinical benefits.

Circulating tumor DNA (ctDNA) within peripheral blood plasma provides noninvasive access to cancer-specific somatic mutations [60]. ctDNA as a biomarker, can be used to manage patients with advanced cancer by replacing tissue biopsy for genotyping of specific mutations which are linked with therapeutic response in a noninvasive manner. However, ctDNA has not yet been clinically implemented for patients treated with ICB [61].

There is a study that evaluated the performance of an amplicon-based bespoke ctDNA detection platform for prognostication and response monitoring in patients treated with the anti-PD-1 monoclonal antibody, pembrolizumab. They hypothesized that the baseline ctDNA levels would be prognostic and early changes in ctDNA levels would head the radiographic response. These findings could help advance the implementation of ctDNA-based testing in the context of ICB treatment across cancer types. Biomarker-directed use of ICB is an important frontier in precision medicine. Despite the dramatic improvement in clinical outcomes in some cancer types, only a minority of patients with solid cancer will derive sustained response or meaningful clinical benefit.

Hypothetical biomarkers could provide clinical utility in this setting by identifying patients most likely to benefit from ICB before or shortly after treatment initiation. Tumor characteristics, including TMB and PD-L1 have shown variable predictive value depending on cancer type. MSI-H phenotype identifies patients with favorable outcomes to ICB treatment but is an uncommon feature in many cancer types, including those enrolled in the INSPIRE trial [62]. In this study, the clinical validity of ctDNA before and during ICB treatment as a prognostic, predictive and pharmacodynamic tool was investigated. Although low baseline ctDNA levels were associated with favorable clinical outcomes in this multi-cohort study, the effect size was modest and partly confounded by tumor type. Patients with malignant melanoma, who may have a higher response rate to pembrolizumab than the other patients enrolled in INSPIRE, displayed the lowest average ctDNA levels. Of note, patients with TNBC were among those with the highest baseline ctDNA levels, which is consistent with other findings that ctDNA levels are higher in TNBC (triple negatif breast cancer) than other breast cancer subtypes [63].

Immunotherapy Toxicity

There are predictive markers of IT toxicity. Immunologic biomarkers are being studied as a way to predict the risk of immunotherapy-related adverse events (irAEs) and as an aid in the early identification of such complications. Examples include interleukin 17 (IL-17) [64], eosinophilia [65, 66], and combined toxicity scores based on gene expression profiling of immunologically predictive cytokines [67, 68]. The optimal predictive biomarker remains to be defined. Some autoantibodies as predictors for survival and immune-related adverse events in checkpoint-inhibition therapy of metastasized melanoma have been discovered recently [69]. Machine learning and artificial intelligence strategies are under investigation to predict the benefit of IT and prognosis.

CONCLUSION

As the principles of cancer, IT and tumor immunology is better understood, there is an urgent need for prognostic and predictive markers. Genomic features such as TMB and MSI and immune features such as PD-L1, CTLA-4 and TILs are predictive markers. Several studies show the correlation between TMB and predicted neoantigen load across multiple cancer types. Response to immune checkpoint inhibitors is higher in tumours with high TMB. Another candidate biomarker that has been studied mostly other than TMB is PD-L1 expression in trials utilizing programmed cell death-1 (PD-1) blockade. PD-L1 protein expression is determined by using Combined Positive Score (CPS) in some types of cancers. PD-L1 expression or CPS is now a routine diagnostic marker for patients with newly diagnosed NSCLC, bladder cancer, and gastrointestinal cancer. MSI was the first instance of the FDA approving a drug ‘based on a tumour's biomarker without regard to the tumour's original location. Combining TMB with specific genomic alterations is crucial. Immunologic biomarkers are being studied as a way to predict the risk of irAEs and as an aid in the early identification of such complications. Morever, new biomarkers are being investigated for both therapy benefit and toxicity prediction.

CONSENT FOR PUBLICATION

Not applicable.

CONFLICT OF INTEREST

The authors declare no conflict of interest, financial or otherwise.

ACKNOWLEDGEMENT

Declared none.

REFERENCES

Biomarkers in Gynecologic Tumors

Selim Afsar¹, *

¹ Department of Obstetrics and Gynecology, Balıkesir University Medical Faculty, Balikesir, Turkey

Abstract

Gynecologic malignancies are one of the most frequent cancers amongst women. Biomarkers are crucial for the differential diagnosis of adnexal masses; however, their potential for diagnosis is limited. In the era of difficulty in ovarian cancer screening, novel biomarkers are defined, but CA125 still remains the most valuable one. Circulating tumor DNAs, DNA hypermethylation, metabolites, microRNAs, and kallikreins have recently turned out as ovarian cancer biomarkers and are being applied to clinical practice. For uterine cancer, genomic classification has now been described, it will be used as a prognostic tool. In this chapter, we describe ovarian, endometrial, and cervical cancer biomarkers in detail.

Keywords: Biomarker, CA125, Cervical cancer, cfDNA, HE4, High-copy number, miRNA, MSI, MSS, Ovarian cancer, POLE-mutated, ROMA, SCC-Ag, Uterine cancer.

---

* Corresponding author Selim Afsar: Department of Obstetrics and Gynecology, Balıkesir University Medical Faculty, Balikesir, Turkey; Tel: [phone]; E-mail: [e-mail]

INTRODUCTION

The lifetime risk of developing ovarian cancer (OC) is approximately 1.3% in women, and OC is the most common cause of gynecological cancer deaths [1]. OC is a generalized term for tumors that involve the ovary, and it can be classified into three different subgroups: epithelial, germ cell, and specialized stromal cell tumors. The most common form of ovarian cancer is epithelial ovarian cancer (EOC), which consists of four main histologic subtypes: serous, endometrioid, clear cell, and mucinous cancer. Serous ovarian carcinoma is the most common form [2].

Ovarian and Fallopian Tube Cancers

Carbohydrate Antigen 125 (CA125)

CA125 is a mucin-type glycoprotein that is associated with the cell membrane,

and it is recognized by the OC125 murine monoclonal antibody. CA125 is not only expressed in the tubal, endometrial, and endocervical epithelium but also mesothelial cells of the pleura, pericardium, and peritoneum [3-5]. Hence, CA125 is not a specific biomarker for OC.

The cutoff value of 35 U/mL is used in routine clinical practice in premenopausal women, but levels of CA125 tend to be lower in postmenopausal women, so a cutoff value of 26 U/mL has been suggested [4, 6]. Elevated serum levels of CA125 can be detected in approximately 85% of patients with EOC, especially in serous type, but also in benign and physiological states such as fibroids, endometriosis, menstruation, and pregnancy as well as in other malignancies such as pancreas, liver, colorectal and breast cancer (Table 1) [7-9].

Screening

Screening modalities for OC are primarily focused on CA125 and the use of transvaginal ultrasonography (TVU). Early detection of OC that requires a screening test with high sensitivity (>75%) and ultrahigh specificity (99.7%) could improve overall survival [10]. False positivity for CA125 has been shown approximately in 1% of the healthy population and 5% of patients with benign disease that limits its use as a stand-alone test [4, 11].

Table 1 CA125 elevations in gynecologic diseases and non-gynecologic malignancies.

aMoore RG, Miller MC, Steinhoff MM, et al. Serum HE4 levels are less frequently elevated than CA125 in women with benign gynecologic disorders. Am J Obstet Gynecol 2012; 206(4):351. bJacobs I, Bast RC Jr. The CA 125 tumour-associated antigen: a review of the literature. Hum Reprod 1989; 4:1-12. cMuyldermans M, Cornillie FJ, Koninckx PR. CA125 and endometriosis. Hum Reprod Update 1999; 1(2):173-87. dBast RC Jr, Klug TL, St John E, et al. A radioimmunoassay using a monoclonal antibody to monitor the course of epithelial ovarian cancer. N Engl J Med 1983; 309:883-7.

The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial enrolled 78,216 women aged 55-74 years, with 39,105 women randomized annual screening. Women were screened annually with CA-125 for 6 years and TVU for 4 years. Neither at a median follow-up of 12.4 years nor at an extended follow-up of 14,7 years, the PLCO trial reported reduction in mortality rates [12, 13]. On the contrary, the false positivity in the PLCO trial was approximately 5%, resulting in complicated surgeries [12].

In the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS), 202,638 postmenopausal women aged 50-74 were randomized to either control or annually multimodal screening group (MMS) or ultrasonography screening (USS) group in a 2:1:1 ratio [14]. Although the trial reported an insignificant mortality reduction on primary analysis (15% in the MMS group, 11% in the USS group), further analysis revealed that a declined mortality was seen after seven years of follow-up (23% in the MMS group, 21% in the USS group) [14].

Hereditary breast-ovarian cancer (HBOC) syndrome, which is related to mutations in BRCA1 and BRCA2 genes, accounts for 14% of EOC. Other identified genes (Lynch syndrome, RAD51C, RAD51D, and BRIP1) account for a minor percentage of EOC [15]. The average risk of developing OC is around 44-45% for BRCA1-mutated women and 12-17% for BRCA2-mutated women [16, 17]. Serous ovarian cancer is the most common histologic subtype for both BRCA mutation carriers [18].

The lifetime risk of endometrial cancer in Lynch syndrome (LS) carriers is 40% due to mutations in mismatch repair genes (MLH1, MSH2, MSH6, and PMS2) [19]. LS is also associated with an increased risk of OC that is generally diagnosed at 45 years of age [20]. Furthermore mutations in BRIP1, RAD51C, and RAD51D have been associated with moderate or high risks of OC [21].

Even if routine OC screening with TVU or CA125 is not recommended in women with BRCA mutations, it may be considered for short-term surveillance in women at greater risk of OC. OC screening should start at age 30–35 years until the time of risk-reducing bilateral salpingo-oophorectomy (RRSO) [22]. In the United Kingdom Familial Ovarian Cancer Screening Study (UK-FOCSS), women were screened with CA125 every 4-months and with TVU annually. RRSO was encouraged throughout the study that resulted in the diagnosis of a significant number of early staged OC [23].

Prediction Models

A variety of combined prediction models have been proposed to estimate the risk of OC in adnexal masses prior to surgery [24]. The risk of malignancy index (RMI) is a scoring system based on menopausal status, ultrasound findings, and the serum CA125 level. The last version of RMI is called RMI4, which includes tumor size score additionally [25]. In a meta-analysis, the diagnostic performance of RMI was reported with a sensitivity of 75% and specificity of 92% [26].

Subjective evaluation of adnexal masses prior to surgery by an experienced ultrasonographer is generally considered the best strategy for differential diagnosis. The handicap of this approach is it requires an expert ultrasonographer [26]. To overcome this obstacle, the International Ovarian Tumor Analysis (IOTA) group created different prediction models based on patients’ characteristics and TVU findings using a logistic regression (LR) analysis (LR1 and LR2) [27]. LR1, the main IOTA logistic regression model, had the best performance with an area under the receiver operating characteristic curve (AUC) of 0.96, according to an external validation study [28]. The Assessment of Different Neoplasias in the adneXa (ADNEX) model combines 3 clinical variables with 6 TVU features. The ADNEX model was validated with the AUC of 0.94 for the basic discrimination between benign and malignant tumors [29]. LR2 (AUC:0.92) had a better diagnostic performance compared to RMI (AUC:0.88), and all IOTA strategies had a superior test performance compared with RMI [30].

Prognosis

Stage I EOC patients with higher preoperative CA125 levels (≥65U/mL) had a significantly poorer 5-year survival compared to lower CA125 levels (<65U/mL) [31]. In accordance with this report, an Australian multicenter study revealed that stage I EOC patients with CA125 levels ≤30U/mL had an outstanding 5-year overall survival (OS), and this was an independent prognostic factor [32]. On the contrary, CA125 is not an independent prognostic factor in advanced stage OC patients [33].

It was reported that a decline in serum CA125 levels during the initial 2 cycles of platinum-based chemotherapy was a strong independent prognostic factor for survival [34]. Patients in complete clinical remission after standard primary chemotherapy who had a very low CA125 level (less than 5 kU/L) had a significantly longer progression-free survival (PFS) and OS than patients with higher CA125 levels [35]. In a French multicentric study, it was concluded that a prolonged half-life of CA125 was an indicator of both persistent CA125 production and poor response to chemotherapy [36].

Recurrence

Among patients with EOC who are on complete clinical remission, a progressive low-level increase in serum CA-125 levels, even if within the normal range, is strongly related to disease recurrence, but it should be always supported by CT or PET scan imaging [37]. The MRC OVO5/EORTC55955 trial registered 1442 patients with OC in complete remission after first-line platinum-based chemotherapy and a normal CA125 concentration. Clinical examinations and CA125 measurements were done every 3 months. If CA125 concentration exceeded twice the upper limit of normal, patients were randomly allocated (1:1) to the early treatment group or delayed treatment group to whom CA125 measurements were continued. This trial showed no evidence of survival benefit with early treatment of relapse on the basis of a raised CA125 concentration alone [38]. Notwithstanding that, the European Society of Gynecologic Oncologists (ESGO) announced against the withdrawal of CA125 monitoring during routine follow-up. Consequently, CA125 monitoring should be advised in specific patients after who (i) are eligible for future clinical trials on second-line chemotherapy or already are treated as part of a clinical trial after complete response to first-line chemotherapy (iii) are not going routine follow-up (iv) are eligible for secondary surgery at recurrence [39].

Human Epididymis Protein 4 (HE4)

HE4, is a novel serum biomarker and has been reported as the most promising one in EOC. HE4 is primarily expressed from the reproductive and respiratory tracts, and increased serum levels can be detected in OC [40]. The HE4 serum levels in healthy women have been reported to range from 60 pmol/L to 150 pmol/L due to the relationship between increasing HE4 serum levels with age [41]. The serum concentration of HE4 shows no significant variations during the menstrual cycle or during hormonal treatment, but HE4 levels are lower in premenopausal women than in postmenopausal women [42]. HE4 levels are decreased in pregnancy and women with endometriosis [8].

It has been proposed that HE4 improves the diagnostic specificity of CA-125 with a similar sensitivity [43]. There is no clear evidence that HE4 is a better biomarker than CA125 in terms of screening.

Diagnosis

In a metaanalysis it has been shown that the pooled sensitivity and specificity for HE4 among patients with adnexal masses were 74% (95% CI, 69-78%) and 90% (95% CI, 87-92%), respectively [44]. It has been reported that HE4 was not superior to CA125 for differential diagnosis, on the contrary, another meta-analysis revealed that HE4 was better than CA125 for the diagnosis of ovarian cancer in terms of sensitivity and specificity [45, 46]. Consequently, there is no agreement on the diagnostic performance of HE4 compared to CA125.

Prognosis

Elevated HE4 levels are a powerful and independent factor of poorer prognosis in EOC compared to CA125 [47] and are also significantly correlated with higher tumor grade, serous histology, peritoneal involvement, nodal invasion tumor stage, operation time, and residual tumor size [48].

Risk of Ovarian Malignancy Algorithm (ROMA)

The Risk of Ovarian Malignancy Algorithm (ROMA) is a logistic regression algorithm that uses serum HE4 and CA125 levels with menopausal status to assess the risk of OC in women with adnexal masses [49]. The sensitivity and specificity of ROMA algorithm were detected between 76-86% and 74-95%, respectively, using different types of measurement methods [50]. In a meta-analysis, the ROMA algorithm was demonstrated as having a lesser specificity than HE4 (84% vs 94%), but a higher correlation than CA125 (84% vs 78%) [45]. It was reported that the ROMA algorithm was more sensitive than HE4 (96.7% vs 73.3%) but less specific than HE4 (80% vs 98.6%), in addition to similar AUC (0.97 vs 0.96) [51]. The dual measurement of CA125 and HE4 is apparently the best diagnostic tool over and above ROMA algorithms [50].

Circulating Cell-Free DNA (cfDNA)

Plenty of studies investigated cfDNA as a potential biomarker in the clinical management of ovarian cancer patients in terms of diagnosis and prognosis. It was concluded that the preoperative plasma total cfDNA levels are significantly elevated in patients with EOC, and this is an independent prognostic factor for death from OC [52].

It was reported that the sensitivity and specificity cfDNA (88.9% and 89.5%) was higher than CA125 (75% and 52.6%) and HE4 (80.6% and 68.4%) in OC detection [53].

The most common mutation in OC was TP53 which accounts for approximately 96% of all somatic mutations [54]. The mutations of TP53 in circulating DNA of advanced OC patients were identified at allele frequencies as low as 2%, with sensitivity and specificity of >97% by means of a new method called tagged-amplicon deep sequencing (TAm-Seq) [55].

It was demonstrated that a significant correlation between serum DNA levels and residual tumor load of >1 cm after primary surgery (p=0.0001), and both parameters were associated with a higher risk of relapse (p=0.0001 and p=0.020, respectively) and a poorer OS (p=0.021 and p=0.010, respectively) [56].

DNA Methylation

It was reported that the 6-gene DNA hypermethylation analysis showed 82% sensitivity and 100% specificity in the serum of OC patients [57]. It was concluded that the 7-gene DNA hypermethylation analysis showed higher sensitivity (85.3% vs 56.1%) and specificity (90.5% vs 64.15%) for cfDNA compared to CA125 in stage I OC [58]. The 3-gene DNA hypermethylation analysis revealed that cfDNA could detect OC much earlier than the exact diagnosis [59].

Metabolites

Metabolites are the end products of cellular processes and their alterations can be regarded as a response to genetic or environmental changes. Using gas chromatography/time-of-flight mass spectrometry (GC-TOF MS), 66 invasive OC and 9 borderline tumors of the ovary were analysed, and this analysis found that 51 metabolites were significantly different between borderline tumors and OC (p< 0.01) [60].

Metabolite profiling of 35-paired plasma samples from 35 EOC patients before and after surgery using rapid resolution liquid chromatography-mass spectrometry was revealed that identified 10 common biomarkers that were predictive for EOC recurrence. The AUC values in pre- and post-operative plasma were 0.815 and 0.909, respectively; the combined-AUC value reached 0.964 [61].

MicroRNAs

MicroRNAs (miRNAs) are small non-coding RNAs that regulate gene expression within cells [62]. In EOC tumors compared to normal ovarian tissues, upregulation of miR-181a, miR-616 and miR-590-3p, and downregulation of miR-125b, miR-148a-3p, and miR-375 levels have been reported [63-68]. It was found that some miRNAs were up- or down-regulated among different subtypes of EOCs using microarray analyses to compare serous, endometrioid, and clear cell tumors with normal ovarian tissues [69]. For example; miR-510 expression was higher in low-grade serous carcinoma and clear cell carcinoma subtypes but lower in high-grade serous carcinoma compared to normal ovarian tissues [70]. The potential role of miR-93, miR-325, and miR-492 was reported in the malign transformation of endometriosis to EOC [71].

There was a correlation between upregulation of miR-520h with increased ascites volume and poorer survival of EOC patients. In contrast, downregulation of miR-26b was inversely correlated with tumour stage, grade, and ascites volume [72, 73]. It was demonstrated that there was a correlation between the downregulation of miR-216 with lymphovascular invasion, upregulation of miR-133a-2, miR-145, and miR-126 with uterine invasion, and upregulation of miR-302c with pelvic peritoneum invasion [69].

Carcinoembryonic Antigen (CEA)

CEA is still the widely used tumor marker in the management of colorectal cancer [74]. In a Danish ovarian cancer study, it was revealed that 18% of borderline tumors and 4% of OCs were positive for CEA expression, which was especially higher in mucinous subtype compared with other histological subtypes (p<0.00001). In a Cox survival analysis, the CEA expression was found to be a prognostic factor (HR = 2.12, 95%CI 1.11-4.05) [75].

Alpha-Fetoprotein (AFP)

Alpha-Fetoprotein (AFP) is an oncofetal protein produced by the fetal yolk sac, liver, and upper gastrointestinal tract. AFP rarely increases in OC patients but is increased in patients with the germ cell tumors (yolk sac tumor, immature teratoma, dysgerminoma) presented in young women and adolescents [76]. AFP can be used as a reliable biomarker for the prediction of tumor recurrence during the follow-up period of women with endodermal sinus tumors [77].

Inhibin B

Inhibin B is released from granulosa cells and its serum levels are higher in the early follicular phase [78]. Circulating inhibin B concentrations are undetectable after menopause. Inhibin B is superior to inhibin A, with reported sensitivities between 88-100% for inhibin B and 67-77% for inhibin A, for adult-type granulosa cell tumor (AGCT) patients [79-81]. Inhibin B levels fluctuate during the menstrual cycle, and it can also be raised in EOC, especially in the mucinous subtype [80]. Additionally, inhibin B concentrations are in normal range in 10-15% of AGCTs [79, 81].

Human Chorionic Gonadotropin (hCG)

hCG is a glycoprotein hormone that is released at very high concentrations by the placental trophoblasts. Placental and other trophoblastic tumors ordinally release hCG and it is a very sensitive marker. In addition, many patients with nontrophoblastic tumors release the free β-subunit of hCG (hCGβ), which is a powerful marker of poorer prognosis [82].

Dysgerminomas generally release hCG and cause precocious puberty in young adults [83, 84]. Other germ cell tumors, especially embryonal carcinomas, may also release hCG. Yolk sac tumors principally produce alpha-fetoprotein and occasionally hCG [85, 86].

It was reported that hCGβ was the only independent prognostic factor correlated with stage and grade in OC [87]. The combined use of urine hCGβcf and serum CA125 was found to improve the diagnostic power in OC [88].

Kallikreins

The major members of the kallikrein-related peptidases (KLKs) have been shown to be promising biomarkers for OC [89]. It was found that OC patients with higher tumor tissue KLK7 ELISA-levels had a two-fold decreased risk of mortality or recurrence compared to lower levels [90]. It has been reported that KLK13 expression in OC tumor tissue correlated with early-stage disease and favorable OS [91]. The OVSCORE, is an algorithm to predict surgical outcome in OC patients based on tumor grade and volume of ascites with KLKs, KLK6 and KLK13. In the multivariate Cox regression analysis for OS, only KLK7 was a significant marker of OS [92].

Other Markers

Ova1 is a biomarker panel assay that consists of 5-serum protein biomarkers (CA125-II, transferrin, beta-2 microglobulin, apolipoprotein A-1, and transthyretin) and it is used for the triage of patients with adnexal masses with 96% sensitivity and 35% specificity [93, 94]. The next generation of Ova1, displayed 91% sensitivity and 69% specificity that combined CA125-II, HE4, apolipoprotein A-1, follicle stimulating hormone and transferrin [94, 95].

The Copenhagen Index (CPH–I) is similar to ROMA and RMI in terms of the combination of HE4, CA125 and age; however, without considering TVU and menopausal status [96].

The concentrations of CA125, HE4, prolactin, IL-2R, CA 15-3, CA19-9, CA72-4, MIF, Cyfra 21-1, TNFR1, TNFR2, IL-6, IL-7, IL-10, IGFBP1, TSH, TNF-α, GH, TIMP-1, ACTH, and osteopontin were reported to be significantly (P < 0.001) higher in the serum of patients with early-stage ovarian cancer compared to healthy women, whilst serum levels of HE4, IL-2R, prolactin, CA15-3, CA19-9, CA72-4, Cyfra 21-1, TNFR1, TNFR2, IL-6, IL-7, IL-10, TNF-α, TSH, IGFBP1, MMP-7, VCAM-1, eotaxin-1, FSH, LH, ErbB2, ApoA1,TTR, adiponectin, and CD40L differed significantly (P < 0.01) between patients with early-stage (stage I and II) and late-stage (stage III and IV) OC [97]. However, serum biomarkers other than CA125 are not currently used as a detection tool for the early stage detection due to their lower sensitivity or specificity [98].

Uterine Cancer

Endometrial cancer (EC) is the most common gynecologic tumor and the fourth most frequent cancer in women worldwide [99]. The incidence and mortality of EC have increased in the last few decades due to the higher overall prevalence of obesity and metabolic syndrome [100]. To date, there is no established serum marker in the role of the clinical management of EC.

The most important prognostic factors of EC patients include tumor grade, histological subtype, depth of myometrial invasion, cervical involvement, tumor size, lymphovascular space invasion (LVSI) and lymph node status [101].

CA125

It has been found that CA125 level was elevated in 24.6% of EC patients, and only 10% of EC patients with stage I and II disease have elevated CA125 levels [102, 103]. It was reported that CA125 displayed 20.8% sensitivity and 95% specificity for patients with stage I disease and 32.9% sensitivity and 95% specificity for patients with stage II–IV disease [102].

HE4

HE4 expression is significantly increased in EC tissue and the serum of EC patients so that increased HE4 levels can be detected in approximately 88% of EC patients [104]. HE4 has a higher sensitivity and specificity compared to CA125, especially in the early stages of EC. It was shown that serum HE4 had a relatively lower sensitivity (0.65, 95%CI: 0.63–0.67) and a higher specificity (0.93, 95%CI: 0.92–0.95) with the AUC value of 0.7499, proposing a moderate performance in the diagnosis of EC. HE4 is a promising biomarker for the diagnosis of EC, although its sensitivity requires improvement [105].

Genomic Classification

According to The Cancer Genome Atlas (TCGA) Research Network, EC is classified into four categories: POLE ultramutated, microsatellite stability unstable (MSI) hypermutated, copy-number low (microsatellite stable, MSS), and copy-number high (serous-like) [106]. The POLE ultramutated group contains endometrioid EC (EEC) with a few serous EC (SEC) and has an excellent prognosis. The MSI hypermutated and MSS groups are composed of endometrioid histology and they have a moderate prognosis. The copy-number high (serous-like) group includes SEC and high-grade EEC and has the worst prognosis [107].

This classification of EC has been confirmed with the ProMisE (Proactive Molecular Risk Classifier for Endometrial Cancer) study using less costly methods [108]. The TCGA molecular classification has improved the current risk stratification system by providing additional prognostic information [108-110]. It has been reported that copy-number high EEC-G1 patients would be ranked as a poor prognostic group, on the contrary all EC patients with POLE-hypermutated would be accepted as an excellent prognostic group even if they are in the higher histological grade or advanced FIGO stage [106].

Other Markers

It has been shown that upregulation of miR-944 and miR-301 were associated with more aggressive disease and lower survival rates [111, 112]. On the contrary, higher levels of miR-205 are related to <50% myometrial invasion and early-stage EECs with improved OS [113].

Three hyper-methylated genes (BHLHE22, CDO1, and CELF4) were identified in cervical scrapings and showed 83-96% sensitivity and 78-96% specificity for EC. The combination of any two genes increased sensitivity to 91.8%, specificity to 95.5% [114].

It has been shown that higher levels of estrogen receptor (ER) or progesterone receptor (PR) were related to an improved OS. In contrast, higher HER2 levels were associated with decreased OS. Especially, HER2 was a powerful predictor of survival in serous EC [115].

Cervical Cancer

The screening of cervical cancer is based on liquid-based cytology and high risk human papilloma virus (HR-HPV) DNA detection in cervical specimens. As yet, there has been no identified biomarker available for screening purposes even though a vast number of serum markers have been inquired in cervical cancer.

Squamous Cell Carcinoma Antigen (SCC-Ag)

Elevated serum levels of SCC-Ag have been detected in 28-88% of cervical squamous cell carcinomas that is correlated with the stage and size of cancer, and depth of the stromal invasion [116, 117]. It was shown that increased serum levels of SCC-Ag were predictive of either prognosis or response to treatment and recurrence [118-120].

Other Markers

Increased levels of serum fragments of cytokeratin (CYFRA) have been detected in 42-52% of patients with cervical squamous cell carcinoma [121]. It was reported that elevated CYFRA levels were not related to prognosis [118]. Serum CA125 levels increased in %75 of patients with cervical adenocarcinoma however, only %26 of patients with squamous cell carcinoma of the cervix have high CA 125 levels [122]. Immunosuppressive acidic protein (IAP) levels are elevated in 43-51% of cervical carcinoma that was associated with the stage of the disease and lymph node involvement therefore it may be used for prognosis and evaluation of treatment efficacy [123]. COX-2 was associated with a poorer response to chemoradiotherapy and cancer-related death in patients with stage IIB cervical adenocarcinoma [124].

CONCLUSION

It has utmost importance to differentiate between the benign and malignant adnexal masses prior to the surgery. For that purpose, various prognostic models have been developed and validated. The genomic classification of uterine cancers would be put into use as a risk stratification method for adjuvant treatment decisions.

CONSENT FOR PUBLICATION

Not applicable.

CONFLICT OF INTEREST

The authors declare no conflict of interest, financial or otherwise

ACKNOWLEDGEMENT

I gratefully thank Prof. Dr. Hafize UZUN for giving me the opportunity to be a part of this successful book Biomarkers in Medicine.

REFERENCES