Assessing the malignant potential of ovarian inclusion cysts in postmenopausal women within the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS): a prospective cohort study

Authors


Professor U. Menon, Gynaecological Cancer Research Centre, UCL EGA Institute for Women’s Health, Maple House, 149 Tottenham Court Road, London W1T 7DN, UK. Email u.menon@ucl.ac.uk

Abstract

Please cite this paper as: Sharma A, Gentry-Maharaj A, Burnell M, Fourkala E, Campbell S, Amso N, Seif M, Ryan A, Parmar M, Jacobs I, Menon U, for the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS). Assessing the malignant potential of ovarian inclusion cysts in postmenopausal women within the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS): a prospective cohort study. BJOG 2012;119:207–219.

Objective  To evaluate the malignant potential of ultrasound-detected ovarian inclusion cysts in the development of ovarian cancer (OC) in postmenopausal women.

Design  Prospective cohort study.

Setting  UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS).

Population  Postmenopausal women.

Methods  In UKCTOCS, women in the ultrasound group have annual scans. Women with inclusion cysts (single/multiple anechoic ≤10-mm ovarian cysts) and normal ovaries (both uniform hypoechogenicity) on their first scan were identified and followed up through cancer registry/questionnaires.

Main outcome measures  Relative risk (RR) of developing OC, invasive epithelial ovarian cancer (iEOC), breast cancer (BC) and endometrial cancer (EC) in women with inclusion cysts relative to those with normal ovaries. The incidence was compared with UK age-adjusted expected rates (Office for National Statistics, 2005).

Results  Postmenopausal women (n = 48 230) attended the year 1 (11 June 2001–6 December 2006) screen; 1234 (2.5%) had inclusion cysts alone and 22 914 had normal scans. By 1 November 2009 (median follow-up, 6.13 years; interquartile range, 4.96–6.98 years), four, three (one Type II), seven and 22 women with inclusion cysts and 32, 29 (20 Type II), 90 and 397 women with normal ovaries were diagnosed with OC, iEOC, EC and BC, respectively. The RR values for the respective cancers (OC [RR, 2.32; confidence interval [CI], 0.86–6.28], iEOC [RR, 1.92; CI, 0.62–5.92], EC [RR, 1.44; CI, 0.68–3.05], BC [RR, 1.12; CI, 0.73–1.73]) were not increased. There was no difference between the observed versus expected incidence rates for these cancers in women with inclusion cysts.

Conclusions  Postmenopausal women with ultrasound-detected inclusion cysts do not seem to be at increased risk of ovarian or breast/endometrial (hormone-dependent) cancers.

Introduction

Our understanding of the natural history of ovarian cancer (OC) is limited and, despite numerous efforts, a precursor lesion has not yet been found. Time-honoured concepts that OC originates from inclusion cysts (ICs) of the ovary that undergo malignant transformation1 are being increasingly questioned. Recent studies have shown that OC is a heterogeneous disease and can be broadly classified into two types. Type I tumours consist of slow-growing cancers, usually diagnosed at an early stage. These are the low-grade serous/endometrioid, clear cell, mucinous and transitional tumours. The bulk (75%) of primary invasive epithelial ovarian cancers (iEOCs) consists of Type II tumours which are highly aggressive, usually diagnosed at an advanced stage and are responsible for 90% of cancer-specific mortality. These include high-grade, serous, undifferentiated and mixed Mullerian (carcinosarcoma) OCs. There is growing evidence that a significant proportion of the latter arise from the fallopian tube or other structures derived from the Mullerian tract and involve the ovary secondarily.2

ICs are defined as cysts in the ovarian cortex that are lined by one layer of cells resembling the surface epithelium.3 One view of their origin is that they result from entrapment of the surface epithelium during ovulation.4,5 Those detected on histological examination are reported to be significantly more common in ovaries contralateral to those containing epithelial cancers, are associated with p53 mutations and are present at a higher incidence in ovaries of women at familial risk of OC.6–8 Transvaginal ultrasound (TVS) allows the detailed study of the morphological features of ovaries and is able to detect ICs. As a result, the general population OC screening trial, the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS), in which women in the ultrasound group have annual scans coupled with systematic follow-up, provides an opportunity to assess prospectively the correlation between ICs in postmenopausal women and OC.

ICs in postmenopausal ovaries may imply irregular follicular activity.9 The report also explores whether women with ICs are at increased risk of developing oestrogen-dependent cancers, such as breast and endometrial malignancies.

Methods

Subjects

UKCTOCS is a multicentre randomised controlled trial of 202 638 postmenopausal women, aged 50–74 years, allocated in a 2:1:1 ratio to: (1) a control group (no screening); (2) a multimodal group (annual screening with CA125 as a primary test and ultrasound as a secondary test); and (3) an ultrasound group (annual screening with ultrasound as primary and secondary test). Women were recruited through 13 trial centres located in NHS hospitals in England, Wales and Northern Ireland.10 Eligibility criteria included: (1) age ≥50 and <75 years; and (2) postmenopausal status defined as >12 months of amenorrhoea following a natural or surgical menopause or >12 months of hormone replacement therapy commenced for menopausal symptoms. The exclusion criteria were: (1) a history of bilateral oophorectomy; (2) active malignancy (women with a past history of malignancy were eligible if they had no documented persistent or recurrent disease); (3) increased risk of OC because of familial predisposition; and (4) a previous history of OC. Following confirmation of eligibility, 50 639 women were randomised to the ultrasound group. At recruitment, women completed a baseline questionnaire which included height, weight, body mass index, use of contraceptive pill, parity, fertility, previous hysterectomy, length of hormone replacement therapy use, age at menopause and years from menopause.

In the ultrasound group, women undergo annual TVS at one of the trial centres.10 Transabdominal scans are undertaken when women find TVS unacceptable or painful. Ovarian morphology and dimensions are assessed and volumes are determined using the formula for an ovoid (d1 x d2 x d3 x 0.532). The first scan (Level 1) and any repeat scans, following an unsatisfactory first scan, are performed by Type 1 sonographers who are, at minimum, certified sonographers, midwives or doctors who undertake gynaecological scanning in the NHS.11 All undergo additional training as part of the trial for the assessment of postmenopausal ovaries. If abnormalities are detected on Level 1 scans, women undergo a repeat scan (Level 2).11 These are performed by Type 2 sonographers who are senior sonographers (mostly superintendent level), experienced gynaecologists or radiologists with established expertise in gynaecological scanning.11 The vast majority of scans (2001–2007) were performed using a dedicated Kretz SA9900 ultrasound machine (Medison, Seoul, South Korea). The scan data are entered on a central web-based trial management system. The scan images are transferred from the trial centres using magneto-optical disks for central archiving on a weekly basis.11

Follow-up

All participants were followed up through the NHS Information Centre for Health and Social Care (formerly Office for National Statistics, ONS) in England and Wales and the Central Services Agency and Cancer Registry in Northern Ireland, who provide regular notification of any cancer registrations or deaths. For the purpose of this analysis, up-to-date cancer registration data were obtained from the agencies on 1 November 2009. Further information was obtained from women who had been on the trial for 3.5 years following randomisation through their postal follow-up questionnaires. In addition, women continued to attend for subsequent annual screens.

Confirmation of diagnosis

Medical records, discharge summaries, multidisciplinary team review notes, operative notes and histopathology/cytology reports were obtained for all women reported to have ovarian, tubal, peritoneal endometrial cancer or disseminated cancer of unknown origin (International Classification of Diseases, 10th edition [ICD-10] codes C56, C57.0, C48.2, C54.1 and C80) during trial surgery or via cancer registry flagging, follow-up questionnaire or directly from the participants. The final diagnosis, including the primary site, stage, grade and cancer morphology, was made by an independent Outcomes Committee. Information regarding breast cancer (BC) (ICD-10 code C50) was also obtained from the sources detailed above. Treating physicians were sent a BC questionnaire which included questions on histology.

Data analysis

For the purpose of this analysis, ovarian scans were excluded from the analysis if one or both ovaries were not visualised, irrespective of whether a good or poor view of the pelvis was obtained. When both ovaries were visualised, ovarian morphology was classified as follows:

  • 1 Normal if both ovaries were visualised and had uniform echogenicity.
  • 2 IC if there was a unilocular (single or multiple), thin-walled, anechoic cyst with no papillary projections, measuring ≤10 mm in maximum diameter.
  • 3 Simple cyst if the ovary had a single, thin-walled, anechoic cyst with no septa or papillary projections measuring >10 mm.
  • 4 Complex if the ovary had nonuniform echogenicity other than simple cyst.

The scan data held in the central trial management system were censored for analysis on 30 June 2007. To obtain complete information on ICs, all data recorded in the following fields were searched: incidental findings (text field), notes (text field) and ovarian cyst dimensions (numerical field). The search terms used were: *Inc*, Ger*, Grm*, *cyst*, *simple*. All ICs as per definition were included. Data from annual screens completed in years 1–5 were analysed to determine the incidence and persistence of ICs.

The relative risks (RR) of developing OC, iEOC, BC and EC in women with ICs alone in one or both ovaries, ICs and simple cysts on the same/contralateral ovary, or ICs and complex cysts on the same/contralateral ovary on the first year scan, compared with the entire cohort of women found to have normal ovaries on their first year scan, were calculated. The observed cancer rates in women with ICs were also compared with the expected rates based on national cancer incidence rates (ONS, 2005), and the standardised incidence ratios (SIRs, ratio of observed to expected cancer incidence × 100) were calculated. Age-adjusted expected incidence rates were estimated for each woman, for each year or partial year on trial, and then summed over all women in the cohort to produce an overall expected rate. The expected rate was also adjusted dynamically to reflect the aging woman, and was censored at 1 November 2009 or by death if that was sooner. All confidence intervals for SIRs were based on an assumed Poisson distribution for the observed cancers. Incidence rates for primary iEOC were not available from ONS, and therefore subgroup analysis of iEOC was not possible.

Logistic regression on the IC data using baseline data, such as height, weight, body mass index, use of contraceptive pill, parity, fertility, hysterectomy, length of hormone replacement therapy use, age at randomisation, age at scan, age at menopause and years from menopause, was performed to determine whether there was any statistical link between IC prevalence and the available epidemiological data.

Results

Of the 50 639 women randomised to the ultrasound arm, 48 230 (median age, 60.6 years; interquartile range [IQR], 56.0–66.1 years) completed the first year screen. The reasons for withdrawal in the ultrasound arm are detailed elsewhere.11 Of the 48 230 women who completed the scan, 22 914 had a normal scan, 21 943 had one or both ovaries not visualised, 197 had only simple cysts, 197 had ICs with simple cysts and 1745 had complex features on scan. The latter included 111 women with ICs and complex cysts on the same or contralateral ovaries. ICs alone (with no accompanying simple cyst or complex lesions on the same or contralateral ovaries) were noted in 1234 women (median age, 60.0 years; IQR, 55.6–65.0 years). In addition, 197 women were found to have ICs with a simple cyst.

On 30 June 2007, when scan data were censored for analysis, 91.5% of the women (1131/1234) with ICs alone had completed two and 72.6% (897/1234) had completed three annual screens. Overall, ICs persisted unchanged in 23.2% of women (208/897) over three annual screens and resolved spontaneously in 40.2% (361/897) (Figure 1). Figure 2 details the appearance of new ICs in years 2–5 in the cohort of women (22 914) with normal scans in the first year screen. In women who continued to have normal scans, new ICs developed in 2–3% annually.

Figure 1.

 Outcome in postmenopausal women detected to have inclusion cysts (ICs) and no other abnormalities (ICs alone). CI, confidence interval. *Seven had surgery. **Nine had surgery.

Figure 2.

 Incidence of new inclusion cysts (ICs) with no other abnormalities (ICs alone) in women with persistent normal ovarian ultrasound scans during 5-year follow-up. CI, confidence interval; OC, ovarian cancer.

On 1 November 2009, the women had undergone a median follow-up of 6.13 years (IQR, 4.96–6.98 years). In addition to registry data, other follow-up information (questionnaire and screen attendance) was available in 1206 (97.7%) of the women with ICs and in 20 514 (89.5%) of the normal women. In the 1234 women with ICs on the first year scan, four OCs (one borderline, three epithelial), seven ECs and 22 BCs had been diagnosed. Further OC details in this group of women with ICs are described in Table 1. Only one of the three iEOCs developed in the same ovary in which IC was previously observed. In women with normal scans, 32 were diagnosed with OC (three borderline, 29 epithelial), 90 with EC and 397 with BC. Further OC details in women with normal year 1 scans are described in Table 2. The RR values for OC (RR, 2.32; CI, 0.86–6.28), iEOC (RR, 1.92; CI, 0.62–5.92), EC (RR, 1.44; CI, 0.68–3.05) and BC (RR, 1.12; CI, 0.73–1.73) were not increased in postmenopausal women with IC alone when compared with women with normal ovaries (Table 3). None of the 32 women with normal year 1 scans, who were later diagnosed with OC, had ICs on ultrasound scans performed in subsequent years prior to diagnosis. When compared with ONS 2005, there was no significant difference between the observed versus expected incidence rates for primary ovarian neoplasms in the 1234 women with IC (OC [SIR, 1.09; CI, 0.30–2.82], BC [SIR, 0.84; CI, 0.53–1.28] or EC [SIR, 1.58; CI, 0.64–3.26]).

Table 1.   Details of ovarian cancers detected in the cohort of women with inclusion cysts in year 1
CaseFinal diagnosis, stage, grade, typeSide of ICsUltrasound findings at first scanUltrasound findings prior to cancer diagnosisTime from first scan on trial to diagnosis of OCCT/MRI findings prior to surgery (if performed)Findings at surgery/histology
  1. CT, computed tomography; IC, inclusion cyst; MRI, magnetic resonance imaging; OC, ovarian cancer.

1Right serous cystadenocarcinoma, Stage 1A, grade 1, Type IRight ovary3 ICs on right ovary measuring 7.9, 6.5 and 5.8 mm in maximum diameter. Left ovary normal2 cysts on right ovary with septae, papillations, solid areas and positive colour Doppler. Left ovary not seen4 years 7 monthsMRI: the right ovary was enlarged and had a complex appearance, possibly as a result of cystadenofibroma. However, the appearances were not histologically specific and a malignant lesion could not be completely ruled outRight ovary contained a 6 × 5 × 4 cm3 mixed solid and cystic tumour. Left ovary was normal
2Left clear cell carcinoma, Stage 1C, grade 3, Type IRight ovary1 IC on right ovary measuring 7 mm. Left ovary normalNormal right ovary. Left ovary showed complex cyst measuring 21.7 mm with irregular walls9 monthsNo other imaging details availableLeft-sided mobile 4-cm ovarian mass with no capsular disease
3Ovarian adenocarcinoma, Stage 3C, grade 3, Type IIRight ovary1 IC on right ovary measuring 5 mm. Left ovary normalNo further scans8 monthsCT: moderate abdominal ascites with extensive nodular peritoneal deposits. There was omental cake and small bowel mesenteric node disease. The bulk of the peritoneal disease appeared centred over the right adnexal region, which is the likely sourceRight adnexae showed extensive disease with spread to bowel and omentum
4Right borderline serous papillary ovarian cancer, Type IRight ovary1 IC on right ovary measuring 9.2 mm. Left ovary normalNormal left ovary. Right ovary showed complex cyst measuring 35.6 mm with irregular walls, appearances suggestive of an endometriotic cyst4 years 8 monthsMRI: gradually enlarging (over 4 years) right ovarian cyst with no overtly malignant features. A benign aetiology seems most likely, although malignancy cannot be ruled outRight ovary contained a 6 × 4 × 5 cm3 mixed solid and cystic tumour. Left ovary was normal
Table 2.   Details of ovarian cancers detected in the cohort of women with normal scans in year 1
CaseFinal diagnosisTypeICs described on any scans preceding OC diagnosisLast annual screen preceding OC diagnosis
  1. IC, inclusion cyst; OC, ovarian cancer.

 1Bilateral ovarian serous papillary carcinomaIINoneAnnual 6
 2Metastatic ovarian adenocarcinomaIINoneAnnual 6
 3Bilateral ovarian serous cystadenocarcinomaIINoneAnnual 5
 4Bilateral ovarian serous papillary carcinomaIINoneAnnual 4
 5Bilateral ovarian serous papillary carcinomaIINoneAnnual 4
 6Right ovarian serous adenocarcinomaIINoneAnnual 4
 7Advanced metastatic papillary ovarian adenocarcinoma with bilateral pleural effusionsIINoneAnnual 4
 8Bilateral ovarian serous papillary cystadenocarcinomaIINoneAnnual 4
 9Right ovarian serous cystadenomaIINoneAnnual 4
10Bilateral ovarian serous papillary carcinomaIINoneAnnual 3
11Bilateral ovarian serous papillary carcinomaIINoneAnnual 3
12Bilateral ovarian serous papillary adenocarcinomaIINoneAnnual 3
13Bilateral ovarian serous adenocarcinomaIINoneAnnual 3
14Bilateral ovarian serous carcinomaIINoneAnnual 3
15Right ovarian adenocarcinomaIINoneAnnual 2
16Bilateral ovarian serous papillary carcinomaIINoneAnnual 2
17Bilateral ovarian serous cystadenocarcinomaIINoneAnnual 1
18Ovarian carcinosarcomaIINoneAnnual 1
19Bilateral ovarian serous cystadenocarcinomaIINoneAnnual 1
20Bilateral ovarian serous papillary adenocarcinomaIINoneAnnual 1
21Right ovarian mucinous adenocarcinomaINoneAnnual 7
22Right ovarian mucinous cystadenocarcinomaINoneAnnual 5
23Left ovarian endometrioid adenocarcinomaINoneAnnual 4
24Bilateral ovarian serous adenocarcinomaINoneAnnual 4
25Ovarian endometrioid adenocarcinomaINoneAnnual 4
26Left ovarian serous cystadenocarcinomaINoneAnnual 1
27Right ovarian clear cell adenocarcinomaINoneAnnual 3
28Bilateral ovarian serous papillary cystadenocarcinomaINoneAnnual 3
29Bilateral ovarian serous papillary adenocarcinomaINoneAnnual 2
30Left ovarian borderline serous cystadenomaINoneAnnual 1
31Right ovarian borderline serous tumourINoneAnnual 1
32Right ovarian borderline serous cystadenofibromaINoneAnnual 3
Table 3.   Relative risks of developing gynaecological cancers in women with inclusion cysts
Findings on 1st scanNumberObserved OCIncidence rate/100 000Relative riskLower 95% CIUpper 95% CI
  1. CI, confidence interval; IC, inclusion cyst; OC, ovarian cancer; SC, simple cyst.

Primary malignant neoplasm (C56)
Normal2291432139.65   
IC alone12344324.152.320.866.28
IC and SC19731522.8410.913.5633
Primary invasive epithelial ovarian cancer
Normal2291429126.56   
IC alone12343243.111.920.625.92
IC and SC1971507.614.010.6922.92
Primary endometrial cancer
Normal2291490392.77   
IC alone12347567.261.440.683.05
IC and SC19700004.88
Primary breast cancer
Normal229143971732.57   
IC alone1234221782.821.030.731.73
IC and SC19721015.230.590.162.1

Logistic regression on baseline data (height, weight, body mass index, use of contraceptive pill, parity, fertility, hysterectomy, length of hormone replacement therapy use, age at randomisation, age at menopause and years from menopause) showed three variables to be significantly related to IC prevalence. These were hysterectomy (odds ratio [OR], 1.34), age at randomisation (OR, 0.97; each increase in year decreasing the odds of ICs by 3%) and weight (OR, 0.99; every 1 kg decreasing the odds of ICs by 1%).

Discussion

To our knowledge, this is the only prospective study on the long-term outcomes in women with ovarian ICs detected on ultrasound examination. It shows that there is no increased incidence of primary ovarian or, in particular, primary iEOCs in postmenopausal women with ultrasound-detected ICs alone on a median follow-up of 6 years. New ICs appear in the ovaries of postmenopausal women with an annual incidence of 2.6%, with 40% of ICs resolving spontaneously.

Our findings in this prospective cohort of postmenopausal women from the general population lend further support to the more recent hypothesis that a proportion of primary invasive epithelial ovarian malignancies, especially Type II OCs, may not develop from ovarian ICs. The literature suggests instead that these cancers might originate in Mullerian epithelial structures, especially the distal fallopian tube.2,12 Early serous carcinomas of the fallopian tubes and tubal intraepithelial carcinoma have been identified in women with BRCA gene mutations at risk-reducing salpingo-oophorectomies.13–15 In addition, recent studies have shown that over 70% of sporadic ovarian and peritoneal cancers have mucosal tubal involvement, with the majority of serous tubal intraepithelial carcinomas overexpressing p53 similar to high-grade serous OCs.16 All of these observations add to the growing belief that OCs originate in other pelvic organs (tube, endometrium) and involve the ovary secondarily. Ultrasound studies of simple adnexal cysts, which are not specific to ICs and include larger simple cysts up to 10 cm in size, have also not found a significantly increased risk of OC in these women.9,17–22

There are conflicting results with regard to the association of histologically detected ICs and an increased risk of OC (Table 4). Although both Mittal et al.23 and Kaur et al.24 reported significantly higher proportions of ICs in the contralateral ovary of women who underwent surgery for unilateral OC, when compared with women who underwent surgery for benign conditions, Westhoff et al.25 and Tressera et al.,26 in similar studies, failed to demonstrate an increase. Similar contradictory findings have been reported on the incidence of ICs in normal ovaries removed during surgery in both BRCA mutation carriers and other high-risk groups.8,27–31 Some of this is related to differences in interpretation of the data between studies. In our study too, the risk of OC appeared to increase in women who had ovaries with both ICs and simple/complex cysts (Table 3) as opposed to ICs alone. This highlights the importance of clearly defining the cohort. It is important in this context to highlight that there was no increased risk of iEOC associated with any of the cohorts.

Table 4.   Studies on inclusion cysts
ReferenceMethod of diagnosis of ICPopulationAge (years)IC incidenceConclusionSupports hypothesis that increased number of ICs is a risk factor for OC
  1. BRCA, breast cancer mutation gene; BSO, bilateral salpingo-oophorectomy; IC, inclusion cyst; OC, ovarian cancer; USS, ultrasound.

General population studies
Tok et al.33Histological184 women (operated on for benign gynaecological conditions)18–85 (mean, 48.78), only eight women were under 4040.8% (75/184)ICs, cortical invaginations, stromal hyperplasia, surface papillomatosis, epithelial pseudostratification are more likely to correlate with women’s reproductive features. Women with ICs were older, had lower age at menarche and had higher menarche-to-pregnancy and menarche-to-operation timesNo
Valentin et al.32In vitro USS/histology (autopsy study)52 women (104 adnexa)64–96 (mean, 79)33% (17/52) on USS, histologically 52% (27/52) left ovary and 48% (25/52) right ovaryAdnexal cysts, especially simple cysts, are common in postmenopausal women and could be regarded as normalNo
High-risk population studies
Qi Cai et al.34Histological52 high risk, 66 controls<45 to >7050% (59/118, 24 in high risk, 35 in general population)No statistically significant difference in histopathological features of those in high-risk groups with BRCA1/BRCA2 mutations. Frequency of ICs, papillomatosis, cortical invaginations, epithelial stratifications may be age related. ICs were noted more frequently in the 45–70-year age groupNo
Barakat et al.27Histological18 high risk, 20 controlsHigh risk, 51 (range, 34–68); controls, 53 (range, 39–65)39% (7/18) in cases, 35% (7/18) in controlsNo significant difference between the two groupsNo
Casey et al.28Histological33 BRCA mutation carriers, 27 women from high-risk families but mutation negativeMutation carriers, 39.7 (range, 31–70); mutation negative, 40.9 (range, 26–64)Not statedNo increase in number of ICs noted. Ovarian surface micropapillae were significantly higher in the mutation carriersNo
Stratton et al.30Histological37 (11 BRCA mutation carriers, 26 not carriers)Mutation carriers, 33–66 (median, 45); mutation negative, 33–63 (median, 44)Mutation carriers, 54.5% (6/11); mutation negative, 42.3% (11/26)No significant difference between the groups. In the entire group, only one mutation-positive woman had a microinvasive serous adenocarcinoma of ovaryNo
Sherman et al.29HistologicalCases, 61: 26 benign ovaries removed during staging for contralateral OC; 35 prophylactic oophorectomy for familial risk. Controls, 121: oophorectomy for reasons other than ovarian diseaseMean age: contralateral OC group, 57 ± 15; family history group, 41 ± 9; controls, 53 ± 13Cases, 31 (51%); controls, 80 (66%)Average number of ICs higher among controls than cases (not statistically significant). Average number of cysts per woman not statistically significant between cases and controlsNo
Salazar et al.8Histological (not blinded)Cases: 20 prophylactically removed ovaries in high-risk women. Controls: 20 ovaries removed for other reasonsMean age: high risk, 42.1; general population, 44.7Cases, 70%; controls, 25% (P = 0.006%)Cancer-prone ovaries are histologically different from normal ovaries; 75% of high-risk women will have a combination of two or more histological features, such as ICs, pseudostratification of surface epithelium, papillomatosis, cortical invaginations, stromal hyperplasia and increased follicular activity when compared with 30% of the controls.Yes
Women with unilateral OC
Kaur et al.24Histological29 with unilateral OC, 19 controls with benign ovarian pathology, 39 controls with bilateral normal ovaries58.9 (SD, 10.7), 57.3 (SD, 17.4) and 54.9 (SD, 10.9), respectively86% (25/29) in women with contralateral OC and 56% (23/39) in normal ovariesICs present in more of the cases and in greater numbers than in the controls. Significant differences in the contralateral normal ovary, such as stratification and nuclear atypia in the surface epithelium of the ovary, epithelial clefts and ICsYes
Tresserra et al.26HistologicalContralateral ovaries of 20 women with unilateral OC, 20 women without OC, 12 women with normal ovaries and contralateral benign ovarian pathology. Total 72 nonpathological ovariesNormal, 33–62 (median, 50); benign, 32–61 (median, 49); OC, 27–72 (median, 49)Normal, 75% (15/20); benign, 58% (7/12); cancer, 70% (14/20)Cortical invaginations may play important role in OC development. No statistically significant difference in number of ICs on comparison of the three groupsNo
Westhoff et al.25Histological148 incidental oophorectomy, 37 with contralateral OCCases: median, 55 (31–78); controls, matched within 2 years of casesMean number of IC for OC cases, 2.7; mean number of IC for controls, 3.6 (93% of ovaries examined in preliminary study showed IC)ICs are not associated with OCNo
Resta et al.6Histological200 hysterectomy with BSO surgical specimens—4 groups. Group 1, 50 nonpregnant women with no hyperplastic or neoplastic changes; Group 2, 50 women with epithelial tumours of contralateral ovary; Group 3, 50 women with adenocarcinoma of endometrium but no concomitant ovarian disease; Group 4, 50 polycystic ovariesGroup 1, mean age, 54.2 (37–84); Group 2, 47.5 (17–75), 29 were postmenopausal; Group 3, 60.8 (43–79), 48 were postmenopausal; Group 4, 43.7 (32–52)IC present in greater than 50% in Groups 2 and 3Hyperplastic and metaplastic changes in surface epithelium of the ovary and ICs may be considered as precursor lesions of epithelial OC. The possibility of a hormonal role in the development of epithelial cancers is suggestedYes
Mittal et al.23Histological42 cases, 42 controlsMedian age 55 (38–79) for cases, 54.5 (37–80) for controlsCases, 79% (33/42); controls, 62% (26/42)Significantly higher ICs in cases than controls (P < 0.01)Yes

In our study, the prevalence of ICs alone at the first screen was 2.5% (1234/48 230), with new ICs appearing in 2.8% (623/21 776) in the second year and 2.9% (326/11 104) in the third year of the trial in women with persistent normal ovarian scans in preceding years (Figure 2). The prevalence rates are lower than those reported in the other general population studies on ICs. In an autopsy study of 52 postmenopausal women, ICs were reported in 33% (17/52) on ultrasound examination of the removed ovaries. Histological comparison revealed a higher incidence of ICs (48% in the right ovaries and 52% in the left ovaries), suggesting that very small cysts were probably not detected on ultrasound.32 This study (despite unusual methodology) is, to our knowledge, the only investigation in which ultrasound-detected ICs have been correlated with histological findings. All others currently reported in the literature involve histological examination alone. The incidence of ICs detected in histological studies varies from 33 to 93% (4).6,8,23–33 Part of the higher incidence reflects differences in study populations—women undergoing surgery and the inclusion of premenopausal women.

ICs persisted in 23.2% of women and resolved spontaneously in 40.2% of women in our study. Although there are no studies specific to the persistence of ICs, our findings are in keeping with previous cohort studies of simple (unilocular) cysts of all sizes.19,20 Valentin and Akrawi19 followed 160 postmenopausal women, aged 47–87 years, with ovarian cysts (76% unilocular, 16% <10 mm) ultrasonically for a median of 3 years. The cysts persisted unchanged in 49%, and 29% showed spontaneous regression. Cysts that disappeared were smaller and occurred in younger women.19

Hysterectomy increased the odds of ICs, as did the age and weight of the women. It is well known that the visualisation of ovaries is harder in women who have had a previous hysterectomy, as it is difficult to define the landmarks. The presence of a cyst improves the visualisation of the ovary, leading to a false elevation of incidence rates. The decline in ICs at increasing age is probably in keeping with ovarian atrophy with age and the associated decline in visualisation. Ovarian visualisation on scan has been shown to peak in the second decade, to decline over the following decades with a significant decline in the seventh decade.34,35 It is also likely that some of the younger women who had undergone a hysterectomy with ovarian conservation were not truly menopausal and the ICs detected were the result of persistent follicular activity.

It has been suggested that the presence of ICs may imply irregular follicular activity in postmenopausal women. There is some concern that this activity may increase the risk of these women developing oestrogen-dependent tumours, such as BC and EC. In our study, the presence of ICs in postmenopausal women was not associated with an increase in the incidence of BC or EC.

The strengths of our study are as follows: its size; the prospective cohort design with postmenopausal women from the general population randomly invited to participate; the exclusion of women who may be at increased risk of familial OC; standardised data collection with all ultrasonographers having additional training in the scanning of postmenopausal ovaries; systematic follow-up of cancers through both ‘flagging’ via national cancer registries and postal questionnaires, with additional information gained at attendance at annual screening; confirmation of OC diagnosis by an independent Outcomes Committee. The additional follow-up is important as there can be a time lag in national cancer registration. The main limitations, as in any ultrasound study, are the resolution of imaging (it is likely that a proportion of ICs were not detected because of their size), the subjective nature of scanning and the exclusion of women from this analysis in whom one or both ovaries were not visualised. The analysis defined IC as an anechoic cyst in the ovary measuring ≤10 mm, and it is possible that some ICs included were not within the ovarian stroma, but on the surface. In addition, a consequence of a single ovarian IC being considered as a normal finding in UKCTOCS could be that, in some women, the presence of a single IC was not recorded. The constant incidence over 5 years of new ICs at around 2.5%, however, suggests that this was not a major issue.

Conclusions

Our data show that ultrasound-detected ICs in postmenopausal women do not seem to be associated with an increased incidence of primary invasive ovarian or hormone-dependent cancers, such as breast and endometrium. A longer follow-up is required to definitively confirm these findings.

Disclosure of interests

Ian Jacobs has consultancy arrangements with Becton Dickinson, who have an interest in tumour markers and OC. They have provided consulting fees, funds for research and staff, but are not directly related to this study. Usha Menon has a financial interest through UCL Business and Abcodia Ltd. in the third party exploitation of clinical trials biobanks, which have been developed through the research at UCL. No other financial disclosures.

Contribution to authorship

Usha Menon and Aarti Sharma were involved in design, data analysis and interpretation, drafting of the paper and editing of the final version. Matthew Burnell was involved in analysis and interpretation, and editing of the final version of the paper. Aleksandra Gentry-Maharaj, Evangelia-Ourania Fourkala and Andy Ryan contributed to data collection and editing of the final version of the paper. Stuart Campbell, Nazar Amso, Mourad W. Seif, Mahesh Parmar and Ian Jacobs were involved in design and editing the final version of the paper.

Details of ethics approval

The UKCTOCS study was approved by the UK North West Multicentre Research Ethics Committees (North West MREC 00/8/34). It is registered as an International Standard Randomised Controlled Trial number ISRCTN22488978.

Funding

The UKCTOCS trial was core funded by the Medical Research Council, Cancer Research UK and the Department of Health, with additional support from the Eve Appeal, Special Trustees of Bart’s and the London, and Special Trustees of University College Hospital London (UCLH). A major portion of this work was carried out at UCLH/UCL within the ‘women’s health theme’ of the NIHR UCLH/UCL Comprehensive Biomedical Research Centre supported by the Department of Health. The researchers are independent from the funders.

Acknowledgements

We are very grateful to the many volunteers throughout the UK who participated in the trial, and the entire medical, nursing and administrative staff who work on UKCTOCS. The collaborators to the UKCTOCS study not included in the authors’ list are as follows: Centre Leads: Keith Godfrey, Northern Gynaecological Oncology Centre, Queen Elizabeth Hospital, Gateshead; David Oram, Department of Gynaecological Oncology, St. Bartholomew’s Hospital, London; Jonathan Herod, Department of Gynaecology, Liverpool Women’s Hospital, Liverpool; Karin Williamson, Department of Gynaecological Oncology, Nottingham City Hospital, Nottingham; Ian Scott, Department of Gynaecological Oncology, Derby City Hospital, Derby; Tim Mould, Department of Gynaecology, Royal Free Hospital, London; Robert Woolas, Department of Gynaecological Oncology, St. Mary’s Hospital, Portsmouth; John Murdoch, Department of Gynaecological Oncology, St. Michael’s Hospital, Bristol; Stephen Dobbs, Department of Gynaecological Oncology, Belfast City Hospital, Belfast; Simon Leeson, Department of Gynaecological Oncology, Llandudno Hospital, Llandudno; Derek Cruickshank, Department of Gynaecological Oncology, James Cook University Hospital, Middlesbrough. Ultrasound subcommittee members not on the authors’ list include: Gwendoline Fletcher, University College London, London; Gill Turner, Derby City Hospital, Derby; Carol Brunell, University College Hospital London, London; Rani Rangar, Queen Elizabeth Hospital, Gateshead; Kathy Ford, Liverpool Women’s Hospital, Liverpool.

Ancillary