Papanicoloau test (Pap-smear) based on exfoliative cytology is the primary test in many countries [12, 13]. It has a wide range of specificity (14–97 %) and sensitivity (11–99 %) [14]. Due to issues related to sampling and interpretation errors, liquid based cytology (LBC) was introduced in the mid-1990s to 2000s [15]. LBC was recommended as the primary test in the NHS cervical screening program in England and Wales [16] with a changeover to this technology in 2008 [12]. Whether LBC is a better test than Pap-smear remains to be determined, as recent meta-analysis demonstrated similar performance of both screening tests for histologically confirmed CIN 2 or worse [17].

Cytology-based screening is expensive and resource intensive and needs a well-organized infrastructure for repeat testing and for these reasons is not feasible in low resource setting. Alternative screening strategies have been developed [18, 19]. These mainly rely on visual inspection of the cervix, either after application of 3–5 % acetic acid (VIA), visual inspection with a magnifying glass (VIAM) or visual inspection after Lugol’s iodine (VILI). Although a subjective test, VIA still has similar sensitivity and specificity of 14–95 % and 14–98 % respectively as Pap-smear [19]. Single round of VIA, followed by immediate colposcopy and treatment in a cluster randomized controlled trial (RCT) in India resulted in significant reduction (35 %) in mortality at 7 years [20].

As virtually all cervical cancer cases (99.7 %) are due to HPV, the attention has focused on detection of viral DNA [21]. Several assays for HPV have been developed; either DNA hybridization or PCR based assays which determine the presence of high risk genotypes. HPV testing is incorporated in the screening strategies in a number of countries, including the UK [12, 13, 22]. These assays have mainly been used to triage women who have equivocal smears or for checking samples for proof of cure.

To further refine identification of lesions most likely to progress, more recent assays focus on detecting the HPV viral oncoproteins E6 and E7, which are closely associated with the process of malignant transformation in cervical cancer. Compared to cytology, HPV tests are more reproducible, stable over a range of ages, have a higher NPV and better sensitivity but specificity is lower [23–26]. Combination with cytology yields higher sensitivities [27].

The performance of various screening tests is outlined in Table 26.2 [14, 17, 19, 28–30].

As HPV testing has a high NPV, it may allow extension of the screening interval in cervical cancer. Recent data suggests that this can be as much as up to 6 years [31]. For these reasons, HPV has been suggested as a primary test in cervical cancer screening. A pilot of HPV testing with reflex cytology has recently been announced in the UK [32]. To avoid unnecessary repeat testing and treatment in younger women where lesions are often due to transient infections, it has been suggested that HPV testing should begin at 29 years of age (POBASCAM trial) [33] or even later (NTCC trial) [34].

In low resource settings, a single screening episode using primary HPV screening [35] can achieve a significant reduction in advanced cancers and 50 % mortality reduction at 7 years. Superior performance of HPV compared to VIA in detecting CIN2–3 lesions has also been reported [36].

As existing HPV DNA assays are expensive and results are only available after several hours, alternative HPV tests are being developed for LMIC settings where a ‘screen and treat’ approach is considered as the best strategy. A low cost assay, careHPV (Quiagen) is in late development for use in rural or remote communities. This assay has demonstrated higher sensitivity (90 %) for identifying moderate or severe cervical disease (CIN 2+) compared to either VIA (41 %) or LBC (85 %). The test requires no electricity or running water and results are available in 2.5 hours [28].

HPV testing, in addition, lends itself to self-sampling allowing access to cervical cancer screening to a greater number of women. Self-collected samples perform well in assays and the performance of the test is comparable to those collected by healthcare professionals [37, 38].

Cervical cancer is currently the only cancer for which vaccination is advocated by the WHO [39]. Currently there are two HPV vaccines against HPV16 and HPV18, Gardasil (Merck, USA) and Cervarix (GSK, UK), which are highly effective in preventing high grade abnormalities [40] and in modeling studies have been suggested to confer protection up to 20 years [41]. HPV vaccination programs target adolescent girls, mainly aged 12–17 years, with reduction in CIN3+ incidence in young women <18 years already reported. Screening should still continue in those that have not been vaccinated.

Transvaginal ultrasound scanning (TVS) to assess ovarian size and morphology is used to screen for ovarian cancer [49, 50] as it gives a superior view of the pelvic organs and is acceptable to the women [51]. However in postmenopausal women, ovaries can be difficult to visualize although this can be overcome through quality assurance and monitoring [52]. In parallel with the models developed to distinguish benign from malignant adnexal masses in patients presenting with symptoms [53], algorithms have been developed in asymptomatic women undergoing screening [54, 55]. Both are based on the fact that certain ovarian features (papillary projections, complex ovarian cysts with wall abnormalities or solid areas) are strongly associated with the presence of malignancy compared to others (septal thickness, unilocular ovarian cysts <10 cm in diameter and inclusions cysts) [56–58]. Women with complex ovarian cysts at primary screening have repeat scans in 6 weeks to decrease false positive rates as many lesions can resolve spontaneously.

So far, two large studies/trials have investigated the performance of ultrasound-based strategy in ovarian cancer screening.

The University of Kentucky Ovarian Cancer Screening Trial is a single-arm single center ultrasound screening study involving 25,327 women who underwent annual screening between 1987 and 2005. The study has reported encouraging sensitivity/specificity with 9.3 operations carried out per case detected. A stage shift was observed with 82 % of the primary ovarian cancers being early stage (I/II) [54]. Recently, the study reported increased survival in those that were screened [59]. However, as this was not an RCT, it is very likely that a healthy volunteer affect and other biases contributed to the apparent impact of screening [60].

The performance of a TVS-only strategy has been investigated in an RCT. In the ultrasound arm of the general population trial, the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS), where 202,638 post-menopausal women aged 50–74 years were randomized to either control or annual screening with ultrasound or a multimodal strategy in a 2:1:1 fashion [55, 61] (www.​ukctocs.​org.​uk). In addition to the effect of ovarian cancer on mortality, the trial is also investigating the acceptability, compliance, costs, and performance characteristics of the two screening strategies, ultrasound and CA125-based, and the physical and psychological morbidity of screening. The results of the prevalence screen suggests inferior performance of ultrasound screening alone (19 surgeries per case of primary ovarian/FT cancer) compared to the multimodal strategy (3 surgeries per case detected) [55]. Mortality results are expected in 2015.

The discovery that CA125 was raised 5 years in advance of ovarian cancer [62] led to it being investigated as a screening test. 2.9 % of healthy postmenopausal women have elevated CA125 levels, limiting its use as a stand-alone test [63]. This has been overcome by using TVS as a second-line test in a multimodal screening strategy [64, 65] with TVS interpretation based on ovarian morphology [66, 67]. The interpretation of CA125 has been further improved by incorporating age, menopausal status and the rate of change of CA125 values over time into a statistical algorithm (Risk of Ovarian Cancer, ROC) [68–70]. In a RCT of 13,582 postmenopausal women, aged over 50, the ROC algorithm demonstrated high specificity (99.8 %) and PPV (19 %) for primary invasive EOC [70].

The only ovarian cancer screening trial in the general population to use the ROC algorithm is UKCTOCS [55, 61]. In the multimodal arm, CA125 is interpreted using the ROC algorithm to triage the women into low, intermediate and elevated risk. Those at intermediate risk have a repeat CA125 in 12 weeks, whereas those with elevated risk are referred for a transvaginal scan and repeat CA125 in 6 weeks. On the prevalence screen, this strategy had very encouraging sensitivity (89 %) [55] which was maintained at the incidence screens [71].

In the women at high risk (due to a family history of ovarian/breast cancer or mutations in genes such as BRCA1/2) annual screening is not regarded as effective [72, 73]. As most of the cancers detected in this group of women are high grade serous EOCs which progress rapidly, a shorter screening interval has been suggested. A screening strategy incorporating CA125 interpreted using the ROC algorithm and annual TVS has been investigated in the UK Familial Ovarian Cancer Screening Study (UKFOCSS), which is a prospective screening study involving 5,732 women [74]. The only results available from this trial are those from Phase I where 3,563 women at high risk underwent annual screening with CA125 using a cut-off and TVS. Although sensitivity for detection of incident OC/FT was encouraging (81.3 % if occult cancers were classified as false negatives and 87.5 % classified as true positives), only 4 (30.8 %) of 13 incident screen-detected OC/FTCs were detected at stage I/II. Advanced stage OC (>IIIC) was more likely detected in those diagnosed over a year from the last screen compared to those diagnosed within a year of screening (85.7 % v 26.1 %; P = 0.009). UKFOCSS Phase I results further confirm that annual screening does not lead to detection of early-stage disease [74]. Results of the performance of four-monthly screening with ROC are expected in 2015 from UKFOCSS Phase II and the US trials, the GOG-199 study and the US Cancer Genetics Network trial [75]. Current recommendation remains that women at high risk should undergo genetic counseling with a view to risk reducing surgery once they have completed their family.

To further improve performance characteristics, a number of marker panels alone or in combination with CA125 have been evaluated with little success. The exception is Human Epididymis protein 4 (HE4) which is elevated in EOC but not in benign gynecologic conditions [76]. HE4 as an additional screening marker to CA125 is currently being investigated in the Novel Markers Trial [77].

TVS and CA125 have been has been used together in primary screening in order to increase sensitivity. The downside is an increase in false positive screens. Two large trials undertaken over the last three decades have evaluated this strategy.

The Japanese Shizuoka Cohort Study of Ovarian Cancer Screening [78] was an RCT of 82,487 low-risk postmenopausal women who were screened using an annual ultrasound and CA125 using a cut-off. Encouraging sensitivity of 77.1 % and specificity of 99.9 % were reported [78]. The women in the screened arm were more likely to be detected at early stage (63 %) compared to the control arm (38 %). The mortality impact has however not been reported.

The US Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial enrolled 78,237 women aged 55–74 years, with 34,202 women randomized to ovarian cancer screening. Women were screened using a combination of serum CA125 (using a 35 kU/L cut-off), and transvaginal ultrasound for 3 years followed by CA125 alone for a further 2 years. During four annual screens in 34,261 postmenopausal women, 89 invasive ovarian/ peritoneal cancers were detected, of which 60 were screen detected. Overall, 19.5 surgeries were performed per screen-detected cancer [79]. No mortality benefit was found at a median follow up of 12.4 years [80]. Furthermore, in those who had surgery and were not found to have cancer, the complication rate was 15 % [80]. The results need to be interpreted with caution, as 40.6 % of the women were diagnosed after screening ended. Furthermore, CA125 was interpreted using an absolute cut-off (rather than a time series algorithm), and management of screen positives was at the discretion of the treating clinician (rather than via a well-defined protocol) [81].

The details of the four large ovarian cancer screening trials in the general population are outlined in Table 26.3 [54, 55, 59, 78–80].

Following the review of the PLCO mortality data, the US Preventive Services Task Force (USPSTF) has reaffirmed their previous recommendation that ovarian cancer screening is not advocated [82]. However, they state that the impact of a time series algorithm-based strategy on mortality from UKCTOCS is eagerly awaited.

Symptoms for ovarian cancer are non-specific until the disease is in advanced stages. Most commonly reported symptoms experienced by the women 3–6 months prior to diagnosis are abdominal (77 %), gastrointestinal (70 %), pain (58 %), constitutional (50 %), urinary (34 %), and pelvic (26 %) with gynecologic symptoms being the least common [83]. The frequency and severity of symptoms are higher in women with ovarian cancer than in those with other conditions and appear not to be related to stage of the cancer [84, 85]. Goff et al. have developed a symptom index [86] that may help in identifying women with ovarian cancer but there is concern about whether it has adequate performance characteristics (sensitivity of 64 %, specificity of 88 %) as a stand-alone screen. On using the symptom index to select women to undergo CA125 and HE4 testing, the specificity improved to 98.5 % at a decrease in sensitivity to 58 % [87]. More recently, the data form the DoVE pilot project, where women who presented with symptoms underwent screening with serum CA125 and TVS, suggested that this approach can pick up more cancers than that reported from screening studies [88]. However, we feel that only large trials using this approach will be able to answer confirm these preliminary findings.

Increasingly, there are efforts to institute national ovarian cancer symptom awareness campaigns in women and primary care physicians [89–92].

Recent progress which has immediate implication for ovarian cancer screening is the increasing evidence that ovarian cancer is a heterogeneous disease with well-defined genetic and phenotypic subtypes. Some authors group the latter into Type I (low-grade serous, low-grade endometrioid, clear cell, mucinous and transitional/Brenner) carcinomas which are slow growing and have good prognosis and more aggressive Type II (high grade serous, high grade endometrioid, undifferentiated tumours and carcinosarcomas) [93]. In the future, the focus of screening is likely to be invasive EOC with different strategies for detecting Type I (TVS based) versus Type II (marker time profile based) cancers. Unlike cervical cancer, the view over the past few decades has been that a pre-malignant lesion for ovarian cancer does not exist and therefore all efforts were focused on detecting early-stage disease [94, 95]. However, most researchers now accept that the origins of a majority of ovarian cancers may not be in the ovary but elsewhere in the Mullerian tract [96]. Crum et al. have identified premalignant serous tubal intraepithelial cancer (STIC) lesions in the distal fallopian tube and suggested a model of ‘fimbrial-ovarian’ serous neoplasia, with a proportion of serous ovarian cancers starting as STIC lesions and spreading to the ovary [97]. This development raises the possibility of primary prevention. Work is already underway to identify these lesions in vitro [98].

As most (95 %) of the women who develop endometrial cancer present with abnormal vaginal bleeding and are diagnosed with early stage disease, screening is not currently recommended. However, the presence of a precursor lesion, atypical endometrial hyperplasia (AEH), raises the possibility of primary prevention as in cervical cancer. Screening is currently only recommended in women with Lynch Syndrome (LS) or Hereditary Nonpolyposis Colorectal Cancer (HPNCC) [102, 103] whose risk of endometrial cancer by age 70 ranges from 54 % in MLH1 mutation carriers, 21 % in MSH2 to 16 % in MSH6 [104]. Women with LS also have an 8–10 % lifetime risk of ovarian cancer. Although the efficacy of such endometrial screening remains unproven, women are offered screening with annual TVS and endometrial biopsy from the age of 35 [105–107]. There is lack of consensus on an appropriate cut-off value for endometrial thickness (ET) on TVS screening in asymptomatic premenopausal women, and interval cancers are known to occur [108, 109]. The superior performance of annual outpatient hysteroscopy and endometrial sampling over TVS was reported in a prospective observational cohort study of 41 LS women attending a tertiary high-risk familial gynecological cancer clinic [110] with four cases of endometrial cancer/AEH detected using this approach compared to two by TVS. It confirms the guidance on the need for endometrial sampling in these women.

There has been limited enthusiasm to explore screening for endometrial cancer in the general population in view of its good prognosis. However, when diagnosed at late stage, the survival rates are very similar to those of ovarian cancer where major efforts have been made in the last few decades to detect the disease early through screening [55, 80]. In 2001, Fleischer et al. reported detecting one case of endometrial cancer and four cases of AEH on TVS screening of 1,926 asymptomatic postmenopausal women from the general population [111]. More recent data from 37,038 women in the ultrasound arm of UKCTOCS demonstrated that in the general population, ET cut-off of 5 and 10 mm had a sensitivity of 80.5 and 54.1 %, at a specificity 85.7 and 97.2 %, respectively. A 5 mm cut-off would result in 56, whereas 10 mm in 17 diagnostic interventions per case detected [112]. A logistic regression model incorporating epidemiological data (oral contraceptive pill use, age at menarche, number of pregnancies, weight, age, and history of cancer), was able to stratify women according to risk of endometrial cancer. The quarter at most risk included 40 % of endometrial cancers or AEH cases [112]. Work is underway to improve this risk stratification strategy by adding hormonal factors/novel biomarkers so that screening can be offered only to those at highest risk.

In general population screening, endometrial sampling is limited to those with increased ET. It can be performed using either Pipelle endometrial biopsy or hysteroscopy. The former, though well-established and easily performed as an outpatient procedure, has a 10 % procedure failure rate and inadequate tissue yield [113] especially in post-menopausal women. Furthermore, cancers have been reported to be missed on Pipelle alone. Outpatient hysteroscopy is increasingly the gold-standard and is tolerated as well as endometrial sampling [114]. Hysteroscopy has an advantage over TVS and Pipelle as it can detect pathology missed by both tests [115, 116]. Although accepted as the gold standard, it still has an 8–11 % failure rate [117, 118].