In 2020, cervical cancer (CC) was the fourth most common cancer in women worldwide, both in incidence (with >600,000 new cases) and mortality (>340,000 deaths) (https://gco.iarc.fr/, accessed on 22 June 2021). It is well-known that an infection with high-risk types of human papillomavirus (HR HPV) is a major etiological factor of CC, responsible for the majority of cases. The relatively easy accessibility of the cervix enables clinicians to search for precancerous lesions directly in samples collected with a cervical brush. For a long time, the cytology-based Pap test (or Pap smear) was the only method of screening, which undoubtedly helped to reduce overall incidence and mortality, especially in the Western world. However, its high rate of sampling errors, presence of obscuring material (such as blood or mucus), and interpretation mistakes led to the development of its more practical, faster, less obscured, and more accurate alternative, called liquid-based cytology (LBC) ^[1][41]. Compared to conventional cytology, LBC is shown to have a higher sensitivity for detecting low-grade squamous intraepithelial lesions (LSILs), but not for high-grade squamous intraepithelial lesions (HSILs) ^[2][42]. On the other hand, LBC offers another practical advantage over conventional cytology, i.e., the sample in the liquid medium can be used in parallel for HPV testing (described below) without the need for further sample collection.

Despite the success of cytology-based methods, they offer only moderate sensitivities for the diagnosis of HSILs. A large clinical trial, ATHENA ^[3][43], showed that molecular-based HPV DNA testing was a more sensitive alternative, demonstrating that one in four women who were HPV 16-positive would have cervical disease within three years and that nearly one in seven with normal Pap cytology who were HPV 16-positive had a high-grade cervical lesion that was missed by cytology. Based on those results, in 2014, the FDA approved the use of the Roche Cobas^® HPV test instead of Pap cytology for first-line primary screening in women 25 and older. Numerous HPV DNA tests are currently available (Table 1).

Another commonly used method in triage testing is the dual immunohistochemical (IHC) staining of p16INK4a and Ki-67. It is now believed that a positive staining for both is reliable evidence that a cell has been transformed due to a persistent HR HPV infection, increasing diagnostic certainty and enabling risk stratification ^[4][5][6][44,45,46]. On the other hand, it should be noted that the use of p16/Ki-67 IHC in younger women (<35 years) suffering from LSIL was significantly less effective than in older women ^[7][47].

Due to the viral origin of CC, prophylactic HPV L1 virus-like particle vaccines were introduced in 2009; namely, the bivalent Cervarix vaccine against HPV 16 and 18 and the quadrivalent Gardasil vaccine against HPV 6, 11, 16, and 18. Later, the nonavalent vaccine Gardasil 9 (HPV types 6, 11, 16, 18, 31, 33, 45, 52, and 58) was developed, which is now the only vaccine used in the U.S. These vaccines are most useful when applied to both girls and boys at an early age (11 or 12 years) but it is recommended for up to 26 years of age ^[8][48]. Older age groups have lower benefits due to an active sexual life and, thus, a higher probability of prior virus exposure. Clinical trials have demonstrated that HPV vaccines are highly effective in preventing HPV infection, but only before the first exposure to the virus ^[9][49]. Moreover, HPV vaccines have been found to reduce infections in anal ^[10][50] and oral regions ^[11][51]. The trials for the approval of the second-generation vaccine Gardasil 9 were found to be safe and almost 100% effective in preventing cervical, vulvar, and vaginal infections and precancers caused by all targeted HPV types ^[12][52]. Vaccinations still represent the only preventive tool in cervical cancer elimination, yet it faces challenges such as high costs, the requirement for multiple doses, a lack of access in developing countries, and a lack of community engagement.

The standard management of CC includes a sequence of cytology and HPV triage (although HPV test implementation differs across countries), followed by a colposcopy (a microscope-aided inspection of the surface of the cervix) ^[13][53], biopsy, and histological examination to confirm the diagnosis and help treatment decisions. When HSIL is confirmed, the WHO guidelines recommend various treatment options, but conization by a large loop excision of the transformation zone is currently preferred ^[14][54]. These WHO recommendations apply to women > 30 years of age but may extend to younger women with a high risk of HSIL. The WHO gives priority to the screening of women aged 30–49 years, rather than maximizing the number of screening tests in a woman’s lifetime. The early recognition of the disease is very important, enabling treatment in the precancerous stage when treatment is minimally burdensome with an excellent prognosis and can be done on an outpatient basis. The combination of early screening with vaccinations could drastically reduce the number of CC cases and deaths each year, making CC a preventable disease.

Despite the usefulness of the above-mentioned tests, novel biomarkers have been investigated that would reduce the colposcopic referral rate for those women with HR HPV infections that are unlikely to progress to invasive CC. Probably the most promising biomarker seems to be the detection of the HPV E6 and/or E7 mRNAs, elevated levels of which, in cervical samples, indicate increased viral activity and thus the potential presence of a transforming infection. For instance, a Korean study from 2014 demonstrated the sensitivity and specificity of the RT-qPCR assay for mRNA detection to be 91% and 98.6%, respectively, when analyzing HSILs ^[15][55].

A more recent class of biomarkers that is not currently applied in clinical practices include various types of short or long non-coding RNAs, altered DNA methylation patterns, or the gain of chromosome 3q. For example, mounting evidence suggests that miRNAs show specific expression profiles at various stages of cervical pathology. However, there is currently no established algorithm that would incorporate the alterations in miRNA expressions to CC screening, as discussed by Pisarska et al. ^[16][56]. In the same review, the authors comprehensibly described the relation of miRNA profiles with HPV infection. Another class of non-coding RNAs, lncRNAs, have emerged as potential biomarkers with interesting applications, including the detection of minimal residual diseases, auxiliary staging, real-time drug resistance monitoring, predicting the risk of metastatic relapse, and patient prognosis ^[17][18][19][57,58,59]. Most of the literature describes two circulating lncRNAs in blood with potential diagnostic and prognostic applications in CC (PVT1 and HOTAIR), but neither has shown significant upregulation patterns for them to be considered as reliable biomarkers ^[19][59].

DNA methylation is another epigenetic mechanism that has been extensively investigated in the process of cervical carcinogenesis. Most studies used LBC samples to examine the methylation profiles of promoters of tumor suppressor genes in CC ^[20][21][60,61], as well as LSILs and HSILs ^[22][23][62,63]. Methylation rates usually correlated with the disease stage (being highest in invasive carcinomas). In precancerous lesions, frequently observed differences in the rate of methylation between LSILs and HSILs were reported, e.g., for the promoters of the SFRP gene family ^[24][64], namely CDKN2A ^[25][65], HS3ST2 ^[26][66], CADM1, MAL ^[27][28][67,68], DAPK ^[29][69], and SOX1 ^[30][70]. Besides human genes, the methylation of HPV DNA has also been proposed as a novel biomarker for the triage of HPV-positive women. Higher HPV methylation was associated with increased disease severity, especially for the HPV L1 gene (reviewed in ^[31][71]). Interestingly, we observed the opposite trend for the E6 gene promoter where gradual demethylation correlated with the progression of precancerous lesions ^[32][72]. This is not surprising given that the E6 protein acts as an oncogene and not as a tumor suppressor. Despite numerous studies, a reliable and widely accepted methylation panel of genes that would improve the clinical performance is yet to be established.

The amplification of chromosome 3q has been consistently observed in both HSILs and invasive cervical squamous cell carcinomas. Chromosome arm 3q contains a human telomerase gene in region 3q26, and the overexpression of this gene, especially the catalytic subunit of human telomerase reverse transcriptase (hTERT), is one of the crucial steps for malignant transformation. Several studies have shown that the gain of chromosome 3q26 is the most consistent genetic abnormality in HSIL ^{[32][33][34][35][36][37]}[72,73,74,75,76,77], and the frequency of the 3q26 gain has been shown to increase with the severity of the disease ^[38][78]. A study by Heitmann et al. demonstrated a sensitivity of 80%, a specificity of 90%, a negative predictive value (NPV) of 98%, and a positive predictive value (PPV) of 44% for the automated scanning for the 3q26 gain among women with LSIL cytology at colposcopy ^[33][73]. More prospective studies are underway to evaluate this biomarker for LSIL detection.

Studies that detect circulating diagnostic biomarkers of CC are less frequent, which is not surprising given that the cervix is easily accessible and thus a blood analysis seems unnecessary. In fact, most authors focus on the role of HPV-circulating DNA as a prognostic biomarker in the blood of patients with primary tumors to monitor the advanced stages or possible metastases ^{[39][40][41][42][43]}[79,80,81,82,83], but not to analyze precancerous lesions. It was reported that non-metastatic CC patients with circulating HPV DNA in plasma had a tendency towards poor clinical outcomes and the development of recurrent distant metastases ^[40][80]. A recent meta-analysis confirmed that circulating HPV DNA in patients with CC could be used as a noninvasive dynamic biomarker of this type of tumor, with high specificity but only moderate sensitivity ^[44][84].

One of the few studies analyzing circulating HPV DNA in precancerous lesions is by Cocuzza et al. who evaluated whether circulating HPV DNA could be detected in the plasma samples of 120 women with regressed lesions, ASCUS, LSILs, and HSILs ^[45][85]. The authors found 53/120 samples to be positive for one of seven HR HPV types (HPV 16, 18, 31, 33, 45, 51, and 52), but only 41/120 samples were HPV-positive when analyzing the plasma of the same patients. The concomitant detection of the seven HR HPV types investigated in both cervical and plasma samples increased with the severity of the disease, ranging from 15.4% (4/26) in women with normal cytology/regressed lesions to 38.9% (7/18) in women with HSIL. Although the authors concluded that circulating HPV DNA in plasma samples could represent an interesting diagnostic biomarker for precancerous lesions, further studies are required to confirm their findings.

Another option to non-invasively monitor cancer markers is to analyze tumor-specific mutations in the ctDNA shed by a primary tumor into the bloodstream, which is an approach often used for other tumor types, but not in CC, where other options are available (as described above). Nevertheless, a study by Lee et al. reported a large screening of the mutations in CC from ctDNA using NGS on a panel of 24 genes ^[46][86]. Results showed that 18 of the 24 genes in the NGS panel had mutations across 24 CC patients, including somatic alterations of the mutated genes (ZFHX3 in 83%, KMT2C in 79%, KMT2D in 79%, NSD1 in 67%, ATM in 38%, and RNF213 in 27%). Moreover, the authors concluded that the RNF213 mutation could be useful for monitoring responses to chemotherapy and radiotherapy. Furthermore, in a study by Charo et al. NGS was applied for the analysis of ctDNA in the plasma of patients with gynecological malignancies of the cervix, ovary, or uterus ^[47][87]. The plasma of thirteen CC patients exhibited mutations in PIK3CA (N = 8, 61.5%), TP53 (N = 5, 38.5%), FBXW7 (N = 3, 23.1%), ERBB2 (N = 2, 15.4%), and PTEN (N = 2, 15.4%). However, these studies did not include women with precancerous lesions (LSILs or HSILs), and it is therefore difficult to say which gene mutations are potentially useful predictors of precancer progression.

Endometrial cancer (EC) is the most common cancer of the female genital tract in developed countries and is the sixth most common cancer in women worldwide, with more than 400,000 new cases diagnosed in 2020 ^[48][88]. Both its incidence and its associated mortality are increasing ^[49][89]. In routine clinical practices, EC is classified into type I or type II, based, in particular, on the histology and grade. Type I (endometrioid) is more common (80–85% of cases) and includes low-grade, diploid, hormone receptor–positive endometrial tumors with a good prognosis. Common molecular alterations identified in type I tumors are microsatellite instability, as well as PTEN, KRAS, and CTNNB1 mutations ^[50][90]. Type II (non-endometrioid, serous) represents high-grade, usually aneuploid, hormone receptor–negative endometrial tumors frequently associated with a poor prognosis and an increased risk of metastasis development. In this type we often encounter TP53 mutations, Her-2/neu amplifications, negative or reduced E-cadherin expressions, and the inactivation of p16 by mutation or hypermethylation. However, the most important limitation of the classification of endometrioid and serous carcinomas is that the categorizations and behaviors often differ from the theory in specific cases ^[51][91]. Recent molecular studies found that EC comprises a range of diseases with distinct genetic and molecular features. Four novel EC categories have recently been proposed by The Cancer Genome Atlas Research Network (TCGA), such as polymerase epsilon (POLE), ultra-mutated, microsatellite unstable (MSI), copy-number low, and copy-number high ^[52][92]. This new genomic-based characterization evoked the reclassification of EC, which has recently led to the updating of the European guidelines for diagnosis and treatment ^[53][93]. However, significant clinical issues, such as the more detailed stratification of the non-specific molecular profile of EC, still remain to be solved. Thus, further studies should be focused on the integration of molecular and clinicopathological features ^[54][94].

The human endometrium is a highly regenerative tissue, adopting multiple different physiological states during life. During the reproductive years, the endometrium undergoes monthly cycles of growth and regression in response to oscillating levels of estrogen and progesterone sustained by stem/progenitor cells ^[55][95]. Thus, an increased rate of mutations in this tissue can be expected, and it mirrors the fact that EC is one the most common gynecological tumors. Accordingly, normal endometrial glands (over 50%) frequently carry ‘driver’ mutations in cancer genes such as KRAS, PIK3CA, FGFR2, and/or PTEN loss, the burden of which increases with age ^[56][96]. Whole-genome sequencing showed that normal human endometrial glands are clonal cell populations with total mutation burdens that increase at about 29 base substitutions per year; however, these are many-fold lower than those of EC ^[57][97]. Interestingly, the extremely high mutation loads attributed to the DNA mismatch repair deficiency and POLE mutations, as well as structural and copy number alterations, are specific to EC, not to normal epithelial cells ^[58][98]. Comprehensive examination of the timing of pathogenic somatic POLE mutations in sporadic endometrial tumors by whole genome sequencing confirmed that pathogenic somatic POLE mutations occurred early, and are possibly initiating events in endometrial and colorectal tumorigenesis ^[59][99]. It was also shown that the acquisition of a POLE mutation caused a distinct pattern of mutations in cancer driver genes, a substantially increased mutation burden, and an enhanced immune response, detectable even in precancerous lesions.

In a recently published study, Aguilar et al. identified the presence of driver mutations (e.g., in KRAS, PTEN, CTNNB1, etc.) by high-throughput sequencing of serial endometrial biopsies taken several years before the onset of EC, even without previously diagnosing atypical hyperplasia or endometrioid intraepithelial neoplasia (EIN). It is important to note that these mutations were confirmed in the invasive cancer. This research provided unique insights into precancer initiation and progression and clearly demonstrated the existence of endometrial premalignant lesions with definitive mutations not readily identifiable by histology ^[60][100]. These findings are also supported by a case report in which tumor-specific mutations were identified in an asymptomatic individual without clinical or pathologic evidence of cancer nearly one year before symptoms developed, i.e., postmenopausal bleeding and a single microscopic focus of EC diagnosed at the time of hysteroscopy ^[61][101].

From the histological point of view, endometrial hyperplasia represents a spectrum of irregular morphological alterations, whereby the abnormal proliferation of the endometrial glands results in an increase in the gland-to-stroma ratio compared to the endometrium from the proliferative phase of the cycle. Nevertheless, different types of EC derive from different precursor lesions. Type I EC typically develops from atypical hyperplasia or EIN, depending on the classification system ^[62][102]. It was shown that the risk of progression to carcinoma in women with non-atypical endometrial hyperplasia was <5%, while almost 30% of women with atypical endometrial hyperplasia were diagnosed with EC ^[63][103]. Moreover, up to 50% of women with atypical hyperplasia on an endometrial biopsy have EC in the resection specimen ^[64][104]. Type II EC usually develops on the atrophic endometrium, often on an endometrial polyp. This lesion is composed of cytologically malignant cells, like those seen in uterine serous carcinoma, lining the surface of the endometrium or endometrial glands without the invasion of the endometrial stroma, myometrium, or lymphovascular spaces ^[65][105]. Although technically non-invasive in appearance, these tumors have been associated with extrauterine disease, reflecting their aggressive biology ^[66][106]. However, there are many unanswered questions, and it is not clear if cancers obey these model paradigms in all cases.

Taken together, it can be assumed that endometrial precancerous lesions preceding the development of invasive ECs are molecularly different from the normal endometrium, frequently show a monoclonal growth pattern, and share some, but not necessarily all, features of a malignant endometrium ^[67][107]. Indeed, these molecular alterations, predominantly single hotspot cancer driver mutations in the context of suspicious histopathologic features, should be detectable already in precursor lesions, especially those with an increased risk of progression to EC and could have sufficient sensitivity and specificity.

Although the detection of precancerous lesions, and the patient risk stratification in general, are increasingly important for early diagnoses and the prevention of cancer, the screening of EC and precancerous dysplasia in the non-symptomatic population basically does not exist (Table 1). In most cases, EC is diagnosed in symptomatic women. There are several lines of evidence that a diagnosis of endometrial hyperplasia may precede the development of endometrioid EC, as they share common environmental predisposing risk factors, such as a postmenopausal status, family history, nulliparity, obesity, diabetes mellitus, long-term Tamoxifen therapy, elevated estrogen levels, smoking, etc. ^[68][69][108,109]. Known genetic risk factors for endometrial cancer are germline mutations in DNA, mismatch repair genes associated with Lynch syndrome ^[70][110], and germline PTEN mutations responsible for Cowden syndrome ^[71][111]. Hereditary EC makes up approximately 2% to 5% of all cases ^[72][112]. For women who tested negative for hereditary mutations, the concept of a polygenic risk score (PRS) based on genetic variants determined by genome-wide association studies would be of potential interest, since it predicts women who are at high risk of developing EC ^[73][113]. Moreover, the integration of an EC PRS with other known EC risk factors (e.g. obesity) should improve the risk stratification accuracy and could provide opportunities for population-based screening ^[74][114].

In current practices, annually performed clinical examinations and transvaginal ultrasounds are insufficient. On the other hand, an additional endometrial biopsy and outpatient hysteroscopy could improve screening results but are not well tolerated. Pipelle sampling can be limited in cases with a non-representative biopsy specimen, or cervical stenosis ^[75][4]. This has made EC and its precancerous lesions a major issue for health care investigations. Already, G. Papanicolaou has been interested in the possibility of the diagnostic value of cervical smears. Due to the anatomical continuity of the uterine cavity within the cervix, material from routine cervical Pap brush samples represent a unique opportunity to detect the signs of disease shed from the upper genital tract ^[76][2]. However, a systematic review by Frias-Gomez et al. showed that only approximately 40–50% of ECs can be detected with a morphological evaluation of cervical smears ^[77][115]. A recent retrospective study, focused on the sensitivity of cervical cytology in EC detection, showed a sensitivity of only 25.6% ^[78][116]. Cervical Pap brush samples were also subjected to PapSEEK, the targeted sequencing of specific PCR amplicons generated from the specific loci of 18 genes, which resulted in the detection of 81% EC-positive patients and 78% early-stage EC cases ^[79][8].

A recently published pilot cytological analysis of self-collected voided urine samples and vaginal samples collected with a Delphi screener before routine clinical procedures reported that the combination of non-invasive urogenital sampling with cytology distinguished malignant from non-malignant causes of postmenopausal bleeding, with approximately 90% accuracy ^[80][14]. Following these promising results, the multicenter prospective validation study (DETECT study, ISRCTN58863784) has been launched to support the utilization of this non-invasive test in clinical practice ^[81][117].

Proteins, rather than genes or RNAs, perhaps reflect the properties of living tissue most accurately. Thus, proteomics has emerged as the technique of choice for biomarker discovery. A number of blood-based biomarker candidates for EC detection have been reported, belonging to hormones, cancer-associated antigens, adipokines, complement factors, plasma glycoproteins, plasma lipoproteins, enzymes and their inhibitors, growth factors, etc. (for detailed information, see the review by Njoku et al. ^[82][118]). Urine represents an attractive biofluid for biomarker discovery and, indeed, elevated levels of the zinc-alpha-2 glycoprotein, alpha-1 acid glycoprotein, and CD59 indicating the presence of EC were identified by proteomics ^[83][119]. Several IHC biomarkers have been also investigated in combination with hematoxylin and eosin (H&E) staining to improve the diagnostics of endometrial precancers and to predict the risk of the transition from hyperplasia to EC. There are several prominent IHC candidates, such as PTEN, p53, PAX2, beta-catenin, E-cadherin, and proteins involved in the DNA mismatch repair pathways (MLH1, MLH2, and MSH6) that may be useful in predicting malignant progression ^[84][120].

Nowadays, genomic analyses offer an excellent opportunity to stratify the risk of EC progression; however, despite the considerable worldwide incidence of EC, there is currently no blood-based biomarker in routine use for EC patients ^[85][121]. Circulating tumor DNA has been recently shown as an important source of genetic information that may enable the detection of both early- and late-stage EC. An NGS analysis of ctDNA from peripheral blood plasma revealed that a mutation in at least one of the four genes, PTEN, PIK3CA, KRAS, and CTTNB1, can be detected in more than 90% of EC patients ^[85][121]. Notably, several studies have determined ctDNA in different body fluids and have confirmed its clinical utility for EC patients. For instance, an NGS analysis of uterine aspirates, in combination with an analysis of ctDNA and CTC obtained from blood samples, clearly indicate the potential clinical applicability ^[86][122].

Efforts focused on the investigation of circulating free miRNAs as potential biomarkers of early EC are also growing, since these miRNAs have been described as potential biomarkers for these malignancies ^[87][123]. Interestingly, several studies have identified miRNAs in extracellular vesicles from different body fluids, e.g., urine ^[88][124], peritoneal lavage ^[89][125], and blood serum ^[90][126]. Recently, a large plasma-derived exosomal miRNA study identified miR-15a-5p as a valuable diagnostic biomarker for the early detection of EC ^[91][127]. Another study identified four miRNAs associated with EC (oncogenic miRNAs miR-135b and miR-205, as well as tumor suppressor miRNAs miR-30a-3p and miR-21) ^[87][123].

Metabolites represent another promising source for the detection of early-stage endometrial cancer and can be detected in endometrial tissue, brush and lavage specimens, blood samples, and urine ^[76][2]. Blood metabolites are of great interest since they are easily accessible, although they have a limited yield. Several blood-based metabolites have been suggested as potential EC biomarkers and are mostly by-products of lipids and amino acids ^[92][128]. Interestingly, the most commonly reported dysregulated metabolic pathways responsible for the presence of biomarkers in the serum of EC patients are the lipid- and glycolysis-related pathways ^[93][94][129,130]. One of the most promising is phosphocholine, whose elevated levels (approximately a 70% increase) have been identified in EC patients ^[95][131]. Recently, Njoku et al. showed that the determination of specific lipid metabolites in blood, such as phospholipids and sphingolipids, could enable the early detection of EC ^[96][132].