Articles

Ovarian cancer population screening and mortality after

long-term follow-up in the UK Collaborative Trial of Ovarian
Cancer Screening (UKCTOCS): a randomised controlled trial
Usha Menon, Aleksandra Gentry-Maharaj, Matthew Burnell, Naveena Singh, Andy Ryan, Chloe Karpinskyj, Giulia Carlino, Julie Taylor,
Susan K Massingham, Maria Raikou, Jatinderpal K Kalsi, Robert Woolas, Ranjit Manchanda, Rupali Arora, Laura Casey, Anne Dawnay, Stephen Dobbs,
Simon Leeson, Tim Mould, Mourad W Seif, Aarti Sharma, Karin Williamson, Yiling Liu, Lesley Fallowfield, Alistair J McGuire, Stuart Campbell,
Steven J Skates, Ian J Jacobs, Mahesh Parmar

Summary
Lancet 2021; 397: 2182–93 Background Ovarian cancer continues to have a poor prognosis with the majority of women diagnosed with advanced
Published Online disease. Therefore, we undertook the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) to determine
May 12, 2021 if population screening can reduce deaths due to the disease. We report on ovarian cancer mortality after long-term
https://doi.org/10.1016/
follow-up in UKCTOCS.
S0140-6736(21)00731-5
See Comment page 2128
Methods In this randomised controlled trial, postmenopausal women aged 50–74 years were recruited from 13 centres
MRC Clinical Trials Unit,
Institute of Clinical Trials and
in National Health Service trusts in England, Wales, and Northern Ireland. Exclusion criteria were bilateral
Methodology oophorectomy, previous ovarian or active non-ovarian malignancy, or increased familial ovarian cancer risk. The trial
(Prof U Menon FRCOG, management system confirmed eligibility and randomly allocated participants in blocks of 32 using computer
A Gentry-Maharaj PhD, generated random numbers to annual multimodal screening (MMS), annual transvaginal ultrasound screening
M Burnell PhD,
A Ryan PhD, C Karpinskyj MSc,
(USS), or no screening, in a 1:1:2 ratio. Follow-up was through national registries. The primary outcome was death
G Carlino MSc, due to ovarian or tubal cancer (WHO 2014 criteria) by June 30, 2020. Analyses were by intention to screen, comparing
S K Massingham RGN, MMS and USS separately with no screening using the versatile test. Investigators and participants were aware of
Prof M Parmar DPhil), Clinical
screening type, whereas the outcomes review committee were masked to randomisation group. This study is
Epidemiology, Institute of
Health Informatics registered with ISRCTN, 22488978, and ClinicalTrials.gov, NCT00058032.
(J Taylor MSc), and Department
of Women’s Cancer, Institute Findings Between April 17, 2001, and Sept 29, 2005, of 1 243 282 women invited, 202 638 were recruited and randomly
for Women’s Health
assigned, and 202 562 were included in the analysis: 50 625 (25·0%) in the MMS group, 50 623 (25·0%) in the USS group,
(J K Kalsi PhD,
Prof I J Jacobs FRCOG), and 101 314 (50·0%) in the no screening group. At a median follow-up of 16·3 years (IQR 15·1–17·3), 2055 women were
University College London, diagnosed with tubal or ovarian cancer: 522 (1·0%) of 50 625 in the MMS group, 517 (1·0%) of 50 623 in the USS group,
London, UK; Department of and 1016 (1·0%) of 101 314 in the no screening group. Compared with no screening, there was a 47·2% (95% CI
Cellular Pathology
19·7 to 81·1) increase in stage I and 24·5% (–41·8 to –2·0) decrease in stage IV disease incidence in the MMS group.
(Prof N Singh FRCPath,
L Casey FRCPath), Department Overall the incidence of stage I or II disease was 39·2% (95% CI 16·1 to 66·9) higher in the MMS group than in the no
of Clinical Biochemistry screening group, whereas the incidence of stage III or IV disease was 10·2% (–21·3 to 2·4) lower. 1206 women died of
(A Dawnay PhD), and the disease: 296 (0·6%) of 50 625 in the MMS group, 291 (0·6%) of 50 623 in the USS group, and 619 (0·6%) of 101 314 in
Department of Gynaecological
the no screening group. No significant reduction in ovarian and tubal cancer deaths was observed in the MMS (p=0·58)
Oncology
(Prof R Manchanda PhD), Barts or USS (p=0·36) groups compared with the no screening group.
Health NHS Trust, London, UK;
Department of Health Policy, Interpretation The reduction in stage III or IV disease incidence in the MMS group was not sufficient to translate into
London School of Economics
and Political Science, London,
lives saved, illustrating the importance of specifying cancer mortality as the primary outcome in screening trials.
UK (M Raikou PhD, Given that screening did not significantly reduce ovarian and tubal cancer deaths, general population screening
Prof A J McGuire PhD); cannot be recommended.
Department of Economics,
University of Piraeus, Athens,
Greece (M Raikou); Department
Funding National Institute for Health Research, Cancer Research UK, and The Eve Appeal.
of Gynaecological Oncology,
Queen Alexandra Hospital, Copyright © 2021 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
Portsmouth, UK
(R Woolas FRCOG); Wolfson
Institute of Preventive
Introduction detection, spanning across four decades.2–6 All trials have
Medicine, CRUK Barts Cancer Ovarian cancer remains the most deadly of all used combinations of the biomarker CA125 and pelvic
Centre, Queen Mary University gynaecological cancers. Most patients (58%) are imaging using transvaginal ultrasound scans (TVS).
of London, London, UK diagnosed at an advanced stage (III or IV), which is Despite these extensive endeavours, to date there is no
(Prof R Manchanda PhD);
Department of Cellular
associated with poor survival (5-year survival is 27% for evidence that screening for ovarian cancer saves lives.7–9
Pathology (R Arora FRCPath) stage III and 13% for stage IV ovarian cancer).1 The In our multicentre randomised trial (UK Collabora­
and Department of greater than 90% survival rates in women detected tive Trial of Ovarian Cancer Screening [UKCTOCS]),
Gynaecological Oncology at stage I1 has spurred international efforts in early 202 638 women from the general population were

2182 www.thelancet.com Vol 397 June 5, 2021

 Articles

(T Mould FRCOG), University

Research in context College London Hospitals NHS
Trust, London, UK; Department
Evidence before this study benefit. Following extended follow-up (median 14·7 years), of Gynaecological Oncology,
We searched PubMed from Jan 1, 2015, to Dec 31, 2020, with the trial confirmed previous findings of no ovarian cancer Belfast City Hospital, Belfast,
no language restrictions for randomised controlled trials for mortality reduction between the screen and control arms. UK (S Dobbs FRCOG);
Department of Obstetrics and
ovarian cancer screening that reported mortality data.
Added value of this study Gynaecology, Ysbyty Gwynedd,
The following keywords were used to search the database: Bangor, UK (S Leeson FRCOG);
Long-term follow-up (median follow-up >16 years after
“ovarian cancer” AND “randomised controlled trial” AND Division of Gynaecology and
recruitment) in the largest ovarian cancer screening trial, Cancer Sciences, St Mary’s
“screening” AND “mortality”. We found two relevant
to our knowledge, provides definitive new evidence that Hospital, Manchester, UK
publications. In the UK Collaborative Trial of Ovarian Cancer (M W Seif FRCOG); Department
neither screening approaches used in UKCTOCS reduced
Screening (UKCTOCS; n=202 638), at a median follow-up of of Obstetrics and Gynaecology,
deaths from ovarian cancer, compared with no screening.
11·1 years, no significant reduction in deaths from ovarian University Hospital of Wales,
This result was despite a 47·2% increase in incidence of Cardiff, UK (A Sharma FRCOG);
cancer was seen in either of the screen groups (multimodal or
women with ovarian and tubal cancer diagnosed at stage I Department of Gynaecological
ultrasound) compared with the no screening group. Oncology, Nottingham City
and 24·5% decrease in those diagnosed with stage IV disease
A reduction in deaths was seen but was delayed and only Hospital, Nottingham, UK
in the multimodal group compared with the no screening
apparent after about 7 years. There was a suggestion that (K Williamson FRCOG); MGH
group. Importantly, however, there was only a 10·2% decrease Biostatistics, Department of
15% fewer women in the multimodal screening group and
in overall incidence of stage III or IV disease. Medicine, Massachusetts
11% fewer in the ultrasound screening group died from ovarian General Hospital, Boston, MA,
cancer compared with the no screening group. Additionally, Implications of all the available evidence USA (S J Skates PhD, Y Liu MS);
a significantly greater proportion (13%) of women with ovarian General population screening for ovarian and tubal cancer Department of Medicine,
Harvard Medical School,
cancers in the multimodal group but not in the ultrasound with either approach used in UKCTOCS cannot currently be
Boston, MA, USA (S J Skates);
group were found at an earlier stage (stage I and II) compared recommended. We need a screening strategy that can detect Sussex Health Outcomes
with the no screening group. As the data did not definitively ovarian and tubal cancer in asymptomatic women even Research and Education in
answer the question of whether screening saved lives, earlier in its course and in a larger proportion of women than Cancer, SHORE-C, Brighton and
Sussex Medical School,
follow-up was continued to gather more evidence. The Ovarian the tests used in the trial. Meanwhile, our results emphasise
University of Sussex, Brighton,
Cancer Screening arm of the Prostate Lung Colorectal Ovarian the importance of having ovarian and tubal cancer mortality UK (Prof L Fallowfield DPhil);
(PLCO) cancer trial in the USA is the only other large as the primary outcome in screening trials. Create Health, London, UK
randomised controlled trial (n=78 216) to explore mortality (Prof S Campbell FRCOG);
Department of Women’s
Health, University of
New South Wales, Sydney,
randomly assigned to two annual screening groups— We commissioned specialised software from the NHS NSW, Australia (Prof I J Jacobs)
multimodal screening (MMS; longitudinal CA125 and to randomly select women aged 50–74 years and then Correspondence to:
second line TVS) and ultrasound screening (USS; TVS flag them on primary care trusts’ registers and allow Prof Usha Menon, MRC Clinical
first and second-line test), and a no screening group. We electronic transfer of their personal and general practice Trials Unit, Institute of Clinical
Trials and Methodology,
reported previously (median follow-up of 11·1 years), that details. We then sent women personal invitations and
University College London,
an absolute proportion of 13% more women with ovarian, logged replies on the trial management system. Women London WC1V 6LJ, UK
tubal, and peritoneal cancer were diagnosed with stage I attended a recruitment clinic at the regional centre [email protected]
or II disease in the MMS group than in the no screening where they viewed an information video, completed a
group. There was no change in stage in the USS group. recruitment questionnaire, and provided written consent
There was no evidence of a reduction in disease-specific and a baseline serum sample. We scanned recruitment
deaths in either screened group compared with the no questionnaires at the coordinating centre into a bespoke
screening group using the Cox version of the log-rank trial management system.
test. The observed reduction in deaths was delayed and Inclusion criteria were 50–74 years of age and
the cumulative mortality curves appeared to be diverging postmenopausal status. Exclusion criteria were bilateral
at the time of previous reporting.9 Therefore, we aimed oophorectomy, previous ovarian or active non-ovarian
to continue follow-up and report here on the long- malignancy, or increased familial ovarian cancer risk.
term mortality effects of ovarian cancer screening in Ethical approval was provided by the UK North West
UKCTOCS. Multicentre Research Ethics Committees (00/8/34) on
June 23, 2000. All participants provided written informed
Methods consent. The trial design has been previously published
Study design and participants and the protocol is available online.9–12 For the protocol see http://
We did a randomised controlled trial (UKCTOCS) of ukctocs.mrcctu.ucl.ac.uk/
media/1066/ukctocs-protocol_
postmenopausal women aged 50–74 years from the Randomisation and masking v90_19feb2020.pdf
general population recruited through 13 centres in The trial management system confirmed eligibility and
National Health Service (NHS) Trusts in England, Wales, then randomly allocated women using the Visual Basic
and Northern Ireland with use of age-sex registers of randomisation statement and the RND function to
27 primary care trusts.10 annual screening using the MMS or USS strategy, or no

www.thelancet.com Vol 397 June 5, 2021 2183

 Articles

screening in a 1:1:2 ratio. The trial management system due to ovarian and tubal cancer based on disease
allocated a set of 32 random numbers to each regional progression (new or increases in size of previously
centre, of which eight were allocated to MMS, eight to documented lesions on imaging, clinical worsening, or
USS, and the remaining 16 to no screening. We randomly rising biomarker concentrations). In the WHO 2014
allocated each successive volunteer within the regional classification, the definition for primary peritoneal
centre to one of the numbers and subsequently randomly cancers was revised. The outcomes review committee
allocated them into a group. Investigators and partici­ chair (NS) reviewed all 41 cancers previously classified as
pants were aware of group allocation but members of the primary peritoneal as per WHO 2003 classification.9 The
outcomes committee were masked to randomisation outcomes review committee re-staged all ovarian and
group. tubal cancers using FIGO 2014 criteria (previously staged
using FIGO 2003) diagnosed in 2001–14.
Procedures
Annual screening in the MMS group used serum CA125 Outcomes
measurements, with the pattern over time interpreted The primary outcome was death due to ovarian
using the risk of ovarian cancer (ROCA) calculation,13 (ICD-10 C56) or tubal (ICD-10 C57·0) cancer. Ovarian
which identifies significant rises in CA125 concentration cancer includes primary non-epithelial ovarian cancer,
above baseline. On the basis of risk, women were border­line epithelial ovarian cancer, and invasive
triaged to normal (annual screening), intermediate epithelial ovarian cancer. As stated above, ovarian cancer
(repeat CA125 ROCA test in 3 months), and elevated was defined using the revised WHO 2014 definition.15,16
(repeat CA125 ROCA test and transvaginal USS as a The site assignment is in contrast to the previous
second-line test in 6 weeks) risk. Annual screening in mortality analysis (censorship Dec 31, 2014), which used
the USS group used TVS as the primary test, which was the WHO 2003 definition.17 The majority (40 of 41) of
classified as normal (annual screening), unsatisfactory previously classified primary peritoneal cancers using
(repeat in 3 months), or abnormal (scan with a senior WHO 2003 criteria were reclassified as ovarian or tubal
ultrasonographer within 6 weeks). In both groups, cancers. Secondary outcomes were ovarian and tubal
women with persistent abnormalities were assessed by cancer incidence and stage. For all outcomes, subgroup
a trial clinician and referred to the NHS where they analysis was undertaken for invasive epithelial ovarian
underwent further investigation or surgery. We deemed and tubal cancer. All outcome data were kept confidential
women who had surgery or a biopsy for suspected until unmasking. Case fatality rate by stage was a post-
ovarian cancer after clinical assessment as screen hoc outcome.
positive.
Women were linked using their NHS number to Statistical analysis
national cancer and death registration data and Hospital At previous analysis (censorship Dec 31, 2014), there
See Online for appendix Episode Statistics records (appendix p 4). An additional were 358 ovarian and tubal cancer deaths in the no
questionnaire was sent in June, 2020, to a subset of screening group.9 Compared with the no screening
participants who had either exited the national registries group, the mean estimated relative mortality reduction
or for whom it was not possible from Hospital Episode in deaths was 11% (Cox model p=0·24) in the MMS
Statistics data to ascertain if both ovaries had been group and 9% (Cox model p=0·32) in the USS group.
removed. Any mortality reduction was only apparent about 7 years
Throughout the trial, we interrogated the available after randomisation. 162 (45%) of 358 of the deaths in
sources to identify women diagnosed with any of the no screening group during 2001–14 occurred before
19 International Classification of Diseases (ICD)-10 codes 7 years. In 2015, for the no screening versus MMS or
for possible ovarian or tubal cancer and retrieved copies USS comparisons, we estimated that an additional
of medical notes.12,14 The only exception was women 233 no screening group events would give 80% power at
with malignant neoplasm without specification of site a two-sided 5% significance level for a difference in
(ICD-10 C80), who also had another non-ovarian, tubal, relative mortality of 25% during long-term (2015–20)
or peritoneal cancer registration. Medical notes, with follow-up, conditional on the observed mortality
any mention of randomisation group redacted, were reduction of 11%. This estimate translated to a target
reviewed by the outcomes review committee consisting sample size of 591 overall events in the no screening
of gynaecological pathologists and onco­ logists (NS, group: all 233 new and 73% (431 of 591) of total no
RM, RW, RA, LC, AS, and KW). The outcomes review screening group events would occur beyond 7 years.
committee assigned cancer site (ie, whether ovarian, No formal adjustment was made to the test for having
tubal, or other site using a previously audited prespecified previously analysed the data in 2015 or making
algorithm),14 Federation of Gynecology and Obstetrics two screen group comparisons. Instead, we decided to
(FIGO) 2014 stage, grade, morphology, ovarian cancer openly describe the multiplicity issues and acknowledge
type, and cause of death. We defined ovarian and tubal the unadjusted p values. As the number of events were
cancer using the WHO 2014 classi­fication15,16 and death less than anticipated on the planned censorship date

2184 www.thelancet.com Vol 397 June 5, 2021

 Articles

of Dec 31, 2018, follow-up was extended with a new rationale underpinning this important change are
censorship date of June 30, 2020. reported separately.18 In short, given the accumulating
Descriptive statistics regarding ovarian cancer death external evidence of delayed mortality effects in
and incidence were created including tabulations of screening trials, most experts supported the change in
histology, stage, and screen type by group. The primary primary analysis to a test that was sensitive to delayed
mortality analysis was changed from the 2016 report, in effects. We chose the versatile test that was agnostic to
which we used a Cox version of the log-rank test,9 the specific form of the screening effect. The versatile
which is most powerful under proportional hazards. For test, described in 2016,18 is a combination test of
the current analysis, we extensively discussed the best three log-rank test statistics (Z1, Z2, Z3), covering early,
approach within trial management and trial steering constant, and late effects respectively (appendix p 3).
committees, and consulted 12 independent international All analyses were by intention to screen. The primary
statistical, trial, and screening experts. The details and mortality analysis was an MMS versus no screening and

1 243 282 women assessed for eligibility

1 038 192 ineligible

78 225 self-reported as not meeting inclusion criteria
172 149 declined to participate
777 626 did not respond
10 192 not recruited as target reached

205 090 enrolled

2452 excluded
1402 did not meet inclusion criteria
1050 declined to participate

202 638 randomised

101 359 assigned no screening 50 639 assigned USS 50 640 assigned MMS

2389 were not screened 557 were not screened

28 died 2 died
27 bilateral salpingo- 5 bilateral salpingo-
oophorectomy oophorectomy
2334 declined intervention 550 declined intervention

45 excluded* 16 excluded* 15 excluded*

42 bilateral salpingo- 12 bilateral salpingo- 11 bilateral salpingo-
oophorectomy before oophorectomy before oophorectomy before
recruitment recruitment recruitment
3 exited registry before 2 ovarian cancer diagnosed 2 ovarian cancer diagnosed
randomisation before randomisation before randomisation
2 exited registry before 2 exited registry before
randomisation randomisation

101 314 included in primary analysis 50 623 included in primary analysis 50 625 included in primary analysis

5038 incomplete follow-up 2601 incomplete follow-up 2515 incomplete follow-up

5038 declined or exited registry 2601 declined or exited registry 2515 declined or exited registry
before June 30, 2020 before June 30, 2020 before June 30, 2020

96 276 complete follow-up 48 022 complete follow-up 48 110 complete follow-up

Figure 1: Trial profile

MMS=multimodal screening. USS=ultrasound screening. *Events occurred before recruitment, but were discovered after randomisation.

www.thelancet.com Vol 397 June 5, 2021 2185

 Articles

USS versus no screening analysis of the primary outcome using standard Kaplan-Meier methods, based on time
using the versatile test,19 with a Royston-Parmar model to from randomisation to diagnosis. Death from other
estimate survival differences. We defined survival time causes and bilateral salpingo-oophorectomy were
from date of randomisation (t0=0) to date of death due to censoring events. Administrative censorship was the
ovarian or tubal cancer or censorship, or sooner if the same as for the mortality analysis (June 30, 2020).
volunteer died of another cause or was lost to follow-up. Ovarian and tubal cancer incidence rates were explored
No allowances were made for screening non-compliance parametrically using a Royston-Parmar model that
(study groups) or contamination (no screening group). specifically allowed exploration of the underlying hazard
We describe potential time-dependent features of the functions for cancer incidence. For the secondary
screening effect by estimating the hazard ratio (HR) outcome of ovarian and tubal cancer incidence by stage,
and the absolute survival difference at the prespecified and the subgroup analysis of invasive epithelial ovarian
timepoints of 5, 10, 15, and 18 years (maximal follow-up and tubal cancers, we used incidence rate ratios with
was 19·3 years) using a flexible parametric Royston- 95% CIs to compare no screening versus MMS and
Parmar model (appendix p 3). USS groups separately. We also calculated stage-specific
We undertook two secondary analyses of the primary ovarian and tubal cancer case fatality rates.
outcome. We fitted a proportional hazards Cox model to We used Stata, version 16 (versatile test verswlr
the primary outcome data. To allow for a formal analysis function) for all statistical analyses. Results were
of the late effects of ovarian cancer, not subject to issues independently verified by a different statistician using R,
of data re-use and multiple testing, we also fitted a version 4.0.2 (versatile test logrank.maxtest of nph
Cox model to the new data acquired since Jan 1, 2015. package). This trial is registered with ISRCTN, 22488978,
Both the methods and results of sensitivity analyses are and ClinicalTrials.gov, NCT00058032.
detailed in the appendix (pp 8–9). Survival from diag­
nosis in women with ovarian and tubal cancer in the no Role of the funding source
screening group was also compared to national age and The funders of the study had no role in study design,
period adjusted survival rates at 1, 5, and 10 years. We data collection, data analysis, data interpretation, or
undertook a subgroup analysis using the versatile test writing of the report.
of invasive epithelial ovarian and tubal cancer death, in
which other ovarian cancers were censored at death. Results
For the secondary outcome, cumulative incidence of Between April 17, 2001, and Sept 29, 2005, we invited
ovarian and tubal cancer were presented graphically 1 243 282 women to participate and randomly assigned

Total Screen Cancers not detected by screening

positives
Screen negatives Screen negatives Never attended Diagnosed >1 year
≤1 year from last test >1 year after last test screening after end of
of screening episode of screening episode screening*
Multimodal screening (50 625 women, 789 129 women-years)
Ovarian and tubal cancer 522 (100%) 212 (41%) 41 (8%) 41 (8%) 3 (1%) 225 (43%)
Non-epithelial ovarian cancer 16 (100%) 7 (44%) 2 (13%) 2 (13%) 0 5 (31%)
Borderline epithelial ovarian cancer 54 (100%) 24 (44%) 10 (19%) 5 (9%) 0 15 (28%)
Invasive epithelial ovarian and tubal 452 (100%) 181 (40%) 29 (6%) 34 (8%) 3 (1%) 205 (45%)
cancer
Ultrasound screening (50 623 women, 790 231 women-years)
Ovarian and tubal cancer 517 (100%) 164 (32%) 63 (12%) 50 (10%) 19 (4%) 221 (43%)
Non-epithelial ovarian cancer 13 (100%) 11 (85%) 0 1 (8%) 0 1 (8%)
Borderline epithelial ovarian cancer 59 (100%) 48 (81%) 2 (3%) 1 (2%) 3 (5%) 5 (8%)
Invasive epithelial ovarian and tubal 445 (100%) 105 (24%) 61 (14%) 48 (11%) 16 (4%) 215 (48%)
cancer
No screening (101 314 women, 1 577 517 women-years)
Ovarian and tubal cancer 1016† (100%) ·· ·· 514 (51%) ·· 499 (49%)
Non-epithelial ovarian cancer 17 (100%) ·· ·· 7 (41%) ·· 10 (59%)
Borderline epithelial ovarian cancer 91 (100%) ·· ·· 50 (55%) ·· 41 (45%)
Invasive epithelial ovarian and tubal 905 (100%) ·· ·· 457 (50%) ·· 448 (50%)
cancer
Data are n (%). *Screening end Dec 31, 2011. †Includes one case in which histology was not available and two cases of neoplasm of uncertain or unknown behaviour.

Table 1: Ovarian and tubal cancers grouped by primary site and screening status

2186 www.thelancet.com Vol 397 June 5, 2021

 Articles

202 638 (16·3% of 1 243 282; figure 1).10 We followed-up 673 345 annual screens: 345 570 in the MMS group and
participants until June 30, 2020. The final cohort 327 775 in the USS group. Compliance with screening
eligible for analysis consisted of 202 562 (>99·9%) of was high (81% in the MMS group and 78% in the
202 638 women: 50 625 (>99·9%) in the MMS group, USS group) with women undergoing a median of
50 623 (>99·9%) in the USS group, and 101 314 (>99·9%) eight annual screens.9
in the no screening group. We excluded 76 (<0·1%) After the end of annual screening on Dec 31, 2011, all
women (15 [<0·1%] in the MMS group; 16 [<0·1%] in the women were followed up until the censorship date of
USS group; and 45 [<0·1%] in the no screening group; June 30, 2020. Complete follow-up until censorship
figure 1) as they had bilateral salpingo-oophorectomy, or death was possible in 192 478 (95·0%) women
ovarian cancer before joining the trial, or had exited the (48 110 [95·0%] in the MMS group; 48 022 [94·9%] in the
registry before randomisation. As previously reported, USS group; and 96 276 [95·0%] in the no screening
baseline characteristics were balanced between the group) resulting in 3·16 million women-years. Median
groups.9 Screening ended on Dec 31, 2011. We undertook follow-up was 16·3 years (IQR 15·1–17·3) for all groups.

A
1300 No screening group
1200 MMS group
USS group
1100
Cumulative incidence of ovarian and tubal cancer

1000
900
800
per 100 000 women

700
600
500
400
300
200 MMS vs no screening HR 1·04 (95% CI 0·93–1·15); p=0·51
100 USS vs no screening HR 1·04 (95% CI 0·94–1·16); p=0·43

0
0 2 4 6 8 10 12 14 16 18
Number at risk
(number censored)
No screening 101 297 (99) 100 339 (122) 98 911 (126) 97 202 (135) 95 311 (129) 92 414 (129) 88 732 (125) 85 151 (110) 56 370 (39) 9992
MMS 50 618 (82) 50 046 (63) 49 277 (58) 48 322 (70) 47 219 (67) 45 752 (60) 43 994 (68) 42 257 (39) 27 837 (15) 4947
USS 50 614 (93) 49 199 (51) 48 295 (59) 47 350 (66) 46 241 (49) 44 858 (58) 43 238 (71) 41 534 (49) 27 405 (20) 4912

B
1300
1200
Cumulative incidence of invasive epithelial ovarian

1100
1000
and tubal cancer per 100 000 women

900
800
700
600
500
400
300
200 MMS vs no screening HR 1·01 (95% CI 0·90–1·13); p=0·92
100 USS vs no screening HR 1·01 (95% CI 0·90–1·13); p=0·89

0
0 2 4 6 8 10 12 14 16 18
Time since randomisation (years)
Number at risk
(number censored)
No screening 101 297 (83) 100 339 (110) 98 911 (116) 97 202 (126) 95 311 (111) 92 414 (111) 88 732 (108) 85 151 (102) 56 370 (37) 9992
MMS 50 618 (71) 50 046 (49) 49 277 (48) 48 322 (57) 47 219 (62) 45 752 (53) 43 994 (61) 42 257 (36) 27 837 (15) 4947
USS 50 614 (57) 49 199 (41) 48 295 (52) 47 350 (58) 46 241 (45) 44 858 (56) 43 238 (68) 41 534 (47) 27 405 (20) 4912

Figure 2: Kaplan-Meier cumulative incidence per 100 000 women for all ovarian and tubal cancers (A) and for invasive epithelial ovarian and tubal cancers (B)
MMS=multimodal screening. USS=ultrasound screening. HR=hazard ratio.

www.thelancet.com Vol 397 June 5, 2021 2187

 Articles

We identified 4482 women with the 19 prespecified in the USS group was less pronounced, and this
ICD-10 codes for possible ovarian and tubal cancer observation is elucidated by the Royston-Parmar model
who were reviewed by the outcomes review committee hazard functions (appendix pp 19–20), in which the
(appendix p 5). Of them, 2055 (45·9%) were confirmed to rate of cancer incidence drops below the no screening
have ovarian or tubal cancer (table 1). The incidence group between years 4–14 approximately, before rising
of ovarian and tubal cancer per 100 000 women-years back above the no screening group rate.
was 67·7 (95% CI 61·9–73·5; 522 cancers; 770 967 women- Overall, 1805 (87·8%) of 2055 women (452 [0·9%] in
years) in the MMS group, 68·2 (62·4–74·1; 517 cancers; the MMS group; 445 [0·9%] in the USS group;
755 677 women-years) in the USS group, and 65·4 905 [0·9%] in the no screening group) had invasive
(61·4–69·4; 1016 cancers; 1 552 703 women-years) in the epithelial ovarian or tubal cancers. The proportion of
no screening group (appendix p 11). Figure 2A provides the most aggressive type II cancers (79·2% in the MMS
Kaplan-Meier cumulative cancer rates for all women and group; 82·2% in the USS group; and 76·4% in the no
figure 2B provides Kaplan-Meier cumulative cancer rates screening group) was similar across the groups
for invasive epithelial ovarian and tubal cancers. Both (appendix p 6). At 9·5 years after the end of screening,
plots show a greater number of cancer diagnoses in compared with the no screening group, there was a
the screening groups during the first screening year, 47·2% (95% CI 19·7 to 81·1) higher incidence of stage I
reflecting the lead time to diagnosis achieved by screening. disease and a 24·5% (–41·8 to –2·0) lower incidence of
The difference was largely maintained throughout the stage IV disease in the MMS group (table 2). Overall,
screening phase before apparent catch-up by the no there was a 39·2% (95% CI 16·1 to 66·9) higher
screening group during the extended period of follow-up incidence of stage I or II disease and 10·2% (–21·3 to 2·4)
after the end of screening. However, the pattern of catch-up lower incidence of stage III or IV disease in the MMS

FIGO 2014 stage Total

I II III IV Unable to stage
Ovarian and tubal cancers (WHO 2014 classification)*
No screening
Cases 212 (20·9%) 73 (7·2%) 510 (50·2%) 208 (20·5%) 13 (1·3%) 1016
Deaths 20 (9·4%) 24 (32·9%) 391 (76·7%) 174 (83·7%) 10 (76·9%) 609 (59·9%)
MMS
Cases 155 (29·7%) 42 (8·0%) 242 (46·4%) 78 (14·9%) 5 (1·0%) 522
Deaths 23 (14·8%) 16 (38·1%) 190 (78·5%) 62 (79·5%) 4 (80·0%) 291 (55·7%)
USS
Cases 121 (23·4%) 36 (7·0%) 253 (48·9%) 105 (20·3%) 2 (0·4%) 517
Deaths 8 (6·6%) 6 (16·7%) 188 (74·3%) 88 (83·8%) 2 (100·0%) 290 (56·1%)
Between group differences in cases compared with no screening at 9·5 years after end of screening†
MMS 47·2% (19·7 to 81·1) 15·9% (–20·7 to 69·4) –4·4% (–18·0 to 11·4) –24·5% (–41·8 to –2·0) ·· ··
USS 17·0% (–6·4 to 46·2) 1·1% (–32·2 to 50·6) 1·7% (–12·6 to 18·2) 3·4% (–18·2 to 30·8) ·· ··
Invasive epithelial ovarian and tubal cancers (WHO 2014 classification)*
No screening
Cases 116 (12·8%) 69 (7·6%) 501 (55·3%) 208 (23·0%) 12 (1·3%) 906
Deaths 18 (15·5%) 24 (34·8%) 391 (78·0%) 174 (83·7%) 10 (83·3%) 617 (68·1%)
MMS
Cases 91 (20·1%) 41 (9·1%) 237 (52·4%) 78 (17·3%) 5 (1·1%) 452
Deaths 22 (24·2%) 16 (39·0%) 190 (80·2%) 62 (79·5%) 4 (80·0%) 294 (65·0%)
USS
Cases 55 (12·4%) 35 (7·9%) 249 (56·0%) 104 (23·4%) 2 (0·4%) 445
Deaths 7 (12·7%) 6 (17·1%) 186 (74·7%) 87 (83·7%) 2 (100·0%) 288 (64·7%)
Between group differences in cases compared with no screening at 9·5 years after end of screening†
MMS 52·2% (16·8 to 98·4) 15·8% (–19·4 to 66·4) –4·8% (–18·3 to 10·9) –23·7% (–40·7 to –1·7) ·· ··
USS –8·0% (–32·7 to 25·7) –5·2% (–35·8 to 39·9) 0·5% (–13·6 to 16·8) –0·8% (–21·3 to 25·1) ·· ··
Data for cases are n (%); data for deaths are n (case fatality rate for stage); data for between group differences in cases are percentage (95% CI). FIGO=Federation of
Gynecology and Obstetrics. MMS=multimodal screening. USS=ultrasound screening. *Includes cases previously designated as primary peritoneal cancer as per WHO 2003
classification. †Between group differences from a poisson model with length of analysis time as exposure variable; percentage difference taken from the incidence rate ratio,
where percentage difference equals incidence rate ratio minus 1 multiplied by 100%.

Table 2: Summary of incidence and case fatality rate by FIGO 2014 stage for ovarian and tubal cancer

2188 www.thelancet.com Vol 397 June 5, 2021

 Articles

group compared with the no screening group. For the

Number of Deaths (n) Maximum χ² or mortality p value
subgroup analysis of invasive epithelial ovarian and women (n) reduction (hazard ratio)
tubal cancers, the changes in stage distribution in the
Primary analysis* (ovarian and tubal cancer deaths)
MMS group compared with the no screening group
MMS 50 625 296 0·41 0·58
persisted. There was no evidence of a change in
USS 50 622 291 1·23 0·36
incidence in any stage in the USS group compared with
No screening 101 314 619 ·· ··
the no screening group.
Subgroup outcome analysis* (invasive epithelial ovarian and tubal cancer)
At censorship, 1206 (0·6%) women had died of ovarian
MMS 50 625 295 0·41 0·60
cancer: 296 (0·6%) of 50 625 in the MMS group,
USS 50 622 287 1·37 0·33
291 (0·6%) of 50 623 in the USS group, and 619 (0·6%)
of 101 314 in the no screening group. Ovarian and tubal No screening 101 314 617 ·· ··

cancer deaths and incidence by year from randomisation Secondary analysis† (all data 2001–20)

is detailed in the appendix (p 7). The versatile test MMS 50 625 296 0·96 (0·83 to 1·10; 0·068) 0·52
(primary analysis) showed that there was no evidence of USS 50 622 291 0·94 (0·82 to 1·08; 0·067) 0·37
a reduction in ovarian and tubal cancer deaths in either No screening 101 314 619 ·· ··
the MMS (p=0·58) or USS (p=0·36) group compared Secondary analysis† (only data 2015–20)
with the no screening group (table 3). Figure 3 shows the MMS 45 999 136 1·05 (0·86 to 1·30; 0·112) 0·63
Kaplan-Meier cumulative death rates, with any diver­ USS 46 079 128 0·99 (0·80 to 1·22; 0·107) 0·91
gence between the screen and no screen groups being No screening 91 808 258 ·· ··
minimal. A sensitivity analysis that only considered data Data are maximum χ² for primary and subgroup analyses or mortality reduction (95% CI; SE) for secondary analyses.
obtained by equivalent means of electronic health records MMS=multimodal screening. USS=ultrasound screening. *Versatile test. †Cox model.
across all three groups also showed no evidence of a Table 3: Summary of analyses of relative reduction of ovarian and tubal cancer deaths
difference using the versatile test result for both the
MMS (p=0·60) and USS (p=0·37) group (appendix p 9).
This analysis and other sensitivity analyses are detailed (174 [83·7%] of 208 in the no screening group vs
in the appendix (pp 8–9). A Cox model (secondary 62 [79·5%] of 78 in the MMS group), which persisted
analysis) estimated a HR of 0·96 (95%CI 0·83–1·10) for on subgroup analysis of invasive epithelial ovarian
MMS versus no screening and 0·94 (0·82–1·08) for USS cancer. In the USS group, the stage-specific case fatality
versus no screening. A Cox model fitted only to data rates appear to be similar to the no screening group,
from 2015 onwards (secondary analysis) estimated a except for stage II, but numbers in this group are small
HR of 1·05 (95%CI 0·86–1·30) for MMS versus no (table 2).
screening and 0·99 (0·80–1·22) for USS versus no Survival from diagnosis in women with ovarian and
screening. tubal cancer in the no screening group was better in
For the secondary outcome of invasive epithelial comparison to national age and period adjusted survival
ovarian and tubal cancers, there were 295 (0·6%) deaths rates (1 year 77% vs 68%; 5 year 40% vs 37%; appendix
in the MMS group, 287 (0·6%) in the USS group, and p 18).
619 (0·6%) in the no screening group (table 3). The
cumulative death rates similarly showed no evidence of Discussion
an effect of screening (appendix p 11). The versatile test Our results from the largest ovarian cancer screening
for mortality reduction showed no evidence of dif­ trial to date show that on long-term follow-up (median
ference in both the MMS group (p=0·60) and the USS 16·3 years after randomisation), neither MMS or USS, as
group (p=0·33). The appendix shows the Royston- used in UKCTOCS, significantly reduced deaths from
Parmar model fit to the non-parametric Kaplan-Meier ovarian and tubal cancer. There was a 47·2% higher
curves (pp 12–13) and the associated hazard functions incidence of stage I cancer and 24·5% lower incidence of
for each group (pp 14–15). All hazard functions were stage IV cancer, resulting in an overall 39·2% higher
consistently increasing with only small differences incidence of stage I or II cancer and 10·2% lower
between the screening and no screening groups. At incidence of stage III or IV cancer in the MMS group
18 years after randomisation, the Royston-Parmar model than in the no screening group. General population
estimates of survival differences per 100 000 women screening for ovarian and tubal cancer with either of the
were 36·7 (95% CI –65·3 to 138·8) for MMS compared screening strategies cannot be recommended based on
with no screening and 52·9 (–48·2 to 153·9) for USS evidence to date. The changes in stage distribution in the
compared with no screening (figure 3, appendix p 10). MMS group did not translate into mortality reduction,
Compared with no screening, in the MMS group, we emphasising the importance of having disease-specific
observed a higher ovarian and tubal cancer case fatality mortality as the primary outcome in ovarian cancer
rate in patients with stage I disease (20 [9·4%] of 212 in screening trials.
the no screening group vs 23 [14·8%] of 155 in the All women were treated within the NHS where,
MMS group) and a lower rate with stage IV disease since 2004, patients are managed within a designated

www.thelancet.com Vol 397 June 5, 2021 2189

 Articles

800 No screening group

MMS group
700 USS group
Cumulative ovarian and tubal cancer mortality

600
per 100 000 women

500

400

300

200 Versatile test: MMS vs no screening p=0·58; USS vs no screening p=0·36

Number of deaths avoided (absolute survival difference) per 100 000 women
100 at 18 years after randomisation*: MMS vs no screening 36·7 (95% CI
–65·3 to 138·8); USS vs no screening 52·9 (–48·2 to 153·9)
0
0 2 4 6 8 10 12 14 16 18
Time since randomisation (years)
Number at risk
No screening 101 314 (18) 100 761 (47) 99 751 (71) 98 393 (66) 96 854 (100) 94 251 (92) 90 830 (86) 87 495 (97) 58 093 (40) 10 333
MMS 50 625 (10) 50 359 (23) 49 883 (31) 49 236 (31) 48 430 (50) 47 174 (33) 45 531 (59) 43 863 (50) 29 014 (8) 5194
USS 50 622 (10) 50 351 (22) 49 862 (30) 49 247 (38) 48 451 (48) 47 199 (30) 45 635 (49) 43 994 (46) 29 165 (16) 5255

Figure 3: Kaplan-Meier cumulative mortality for ovarian and tubal cancer per 100 000 women
MMS=multimodal screening. USS=ultrasound screening. *Royston-Parmar model based estimates of the effect of screening (appendix p 10).

cancer-specific multidisciplinary team setting and ovarian Our results have implications for ovarian cancer
cancer surgery is done only in gynaecological oncology symptom awareness campaigns as most women who
cancer centres by subspecialty trained and accredited were screen-detected had no high-alert symptoms22 and
gynaecological oncologists. Therefore, there is unlikely to were diagnosed earlier than would have been possible
be variability in quality of care between the randomisation with a symptom-based approach. Our findings suggest
groups. Detailed analysis of stage and histology specific that earlier diagnosis of invasive epithelial ovarian and
treatment is underway and will be the subject of a future tubal cancer in the symptomatic population is unlikely
report. to translate into reduced mortality. However, it is
Achieving a mortality reduction will require a screening important to note that there have been substantial
strategy that can detect ovarian and tubal cancer even advances in the treatment of advanced disease in the
earlier and in a larger proportion of women than we were past 10 years since the end of screening in Dec 31, 2011.
able to achieve. Our findings make it even more The advances in treatment, in combination with earlier
important that before general population screening is diagnosis could contribute to better quality of life
introduced, any new test is shown to reduce ovarian and and improved outcomes. In addition, achieving a rapid
tubal cancer deaths in a future randomised controlled diagnosis is of great importance to women and their
trial. These trials take many years to complete but the families.
high compliance with annual screening in UKCTOCS Our mortality results are similar to those reported
suggests that women are very motivated to join them. for ovarian cancer screening in the Prostate Lung
Given that such trials take considerable time, it is Colorectal Ovarian (PLCO) cancer screening trial, the
probable that population screening for ovarian cancer is only other large randomised controlled trial to report
more than a decade away. on the effects of ovarian cancer screening on mortality.7,8
It is difficult to extrapolate these results to ovarian In the PLCO trial, no evidence was seen of a reduction
cancer screening of women at high-risk, in which the in ovarian cancer deaths between the screen and no
strategy has involved MMS once every 3–4 months screen arms, either at median follow-up of 12·47 or
alongside risk-reducing surgery and resulted in a 14·78 years. However, in UKCTOCS, we found a higher
significant reduction in the proportion of women incidence of stage I and lower incidence of stage IV
diagnosed with advanced disease.20,21 Biological dif­ disease in the MMS group than in the no screening
ferences also exist between cancers in women with group. The PLCO trial found no evidence of a difference
BRCA gene mutations and the general population, which in stage dis­tribution between the screened and non-
result in improved treatment responses in carriers of screened groups.7,8 The use of a longitudinal CA125
BRCA mutations. Unfortunately, it is unlikely that the algorithm in the MMS group instead of a single CA125
true effect of screening on mortality will ever be assessed cutoff, as in the PLCO trial, might have contributed to
in this population as a randomised controlled trial is this difference. We have previously shown that a
challenging, and potentially very effective preventive longitudinal algorithm allows us to detect disease
measures such as risk reducing salpingectomy with earlier and with greater sensitivity than a single CA125
delayed oophorectomy are being evaluated. cutoff.23,24

2190 www.thelancet.com Vol 397 June 5, 2021

 Articles

Despite the 24·5% reduction in stage IV disease in not have the same effect. Stage-specific mortality and
the MMS group, the overall reduction in stage III or IV treatment will be the subject of further in-depth analyses,
incidence 9·5 years after end of screening was only 10%, which will be reported in a separate publication.
with little change in stage III incidence. Previous reports The key strengths of UKCTOCS have been previously
have highlighted the need for a large reduction in late- detailed9 and include scale; multicentre design;
stage incidence as a prerequisite for reducing cancer adherence to protocol through use of a bespoke web-
mortality. However, it should be noted that the length of based trial management system with automation of key
follow-up and, therefore, the dilution effect in the screen processes, remote data entry, and concurrent central
groups of inclusion of cancers diagnosed clinically after monitoring; completeness of follow-up through linkage
end of screening varied in these reports. An analysis of to national registries and administrative databases; and
breast cancer screening trials found no reduction in independent assignment of site and cause of death. The
breast cancer mortality in trials that achieved less than longitudinal algorithm we used since 2001 to interpret
10% reduction in stage III or IV disease and a mean CA125 concentrations was innovative and forward
reduction of 28% in trials that saw a 20% or greater thinking. The UK Government’s Accelerating Detection
reduction.25 In colorectal and lung cancer screening, of Disease Programmme includes collection of repeat
much of the mortality reduction is related to reductions biological samples that will enable building of such
in stage IV disease, which has much higher mortality algorithms. We re-staged all cases using the latest
than earlier stages.26,27 For ovarian and tubal cancer, the FIGO 2014 criteria and revised our ovarian and tubal
high mortality associated with stage III and IV disease, cancer site assignment using WHO 2014 classification to
combined with most women clinically detected with reflect current understanding of disease biology. We also
stage III disease, requires a substantial reduction in the changed our primary analysis approach from a constant-
incidence of both stages before a mortality reduction is effect approach (proportional hazards Cox-model) to
possible. one that allows for a delayed effect (the versatile test) to
There are previous instances in which increased reflect growing evidence that the mortality reduction
incidence of stage I or II cancers in the screen group in in cancer-screening trials, if present, is delayed. We did
screening trials did not translate to a mortality reduction. this through a transparent process with publication of
In the four early randomised controlled trials of lung our methods and the expert opinions that underpinned
cancer,28 compared with the control group, the screened our decision.18
groups showed significant improvements in stage Much has been learnt from the design, conduct, and
distribution. However, much of the screen-detected early analysis of UKCTOCS, which is relevant to future
stage disease cases in these trials were indolent. This large-scale trials. In addition, a large bioresource of For details on how to access the
finding in the lung cancer trials was accompanied by a serum samples (>550 000) and linked data (UKCTOCS resources see http://uklwc.
mrcctu.ucl.ac.uk/
significant increase in cancer incidence in the screen Longitudinal Women’s Cohort) has been built through
groups, but no reduction in late-stage incidence. Both the generosity of the participants. The resource includes
together suggest that overdiagnosis was the main con­ a rare sample set of up to 11 annual blood samples from
tributor to the absence of reduction in mortality.28 This women in the MMS group. In ovarian cancer, it has
result differs from UKCTOCS, in which no significant allowed us to collaborate to explore new biomarkers and
increase in cancer incidence was observed in either of develop longitudinal algorithms.24 The data provides a
the screen groups. In the MMS group of UKCTOCS, unique opportunity to study the natural history of ovarian
it seems probable that the cancers shifted to an earlier and tubal cancer. Important research is also underway on
stage had an intrinsic poor prognosis, which was not early detection of other cancers and risk stratification in
altered by earlier detection and the available treatments cardiovascular disease using this bioresource.
for early stage disease. Further histo­ pathological and A key limitation of the trial is the interval from end of
genetic analysis of these early stage screen-detected screening in 2011 to censorship in 2020, which raises the
cancers could yield important infor­ mation about the possibility of dilution of the screening effect. However,
biology of ovarian cancer. extended follow-up after screening is the norm in
The 47·2% increase in incidence of stage I and screening trials and did not affect mortality reduction
24·5% decrease in incidence in stage IV disease was seen in the European Randomized Study of Screening
accompanied by a higher stage I and lower stage IV case for Prostate Cancer.29 Also, there was no variation in the
fatality rate in the MMS group. This finding persisted in ovarian cancer mortality HR at 15 years and 18 years
subgroup analysis of invasive epithelial cancers. The from randomisation in our Royston-Parmar model
findings are unlike that described previously in cancer (appendix p 10).
screening trials.26 The result suggests that in the MMS In addition, most screen-detected women were
group, although earlier detection of stage IV invasive diagnosed and treated more than a decade ago, before
epithelial ovarian and tubal cancers improved outcomes, many of the advances in clinical management (eg,
earlier detection in stage I cancers that might have widespread use of ultraradical surgery, earlier treatment
presented in later stages in the absence of screening, did modulation based on better prognostic indicators, and

www.thelancet.com Vol 397 June 5, 2021 2191

 Articles

targeted therapies), which could have improved out­ Freenome, outside the submitted work. IJJ reports grants from Eve
comes. In retrospect, second-line tests could have been Appeal Charity, Medical Research Council, Cancer Research UK, and
NIHR during the conduct of the study. He co-invented the ROCA
further optimised so that time to diagnosis after an in 1995, it was patented by Massachusetts General Hospital and
abnormal screen was reduced.23 Finally, clinicians could Queen Mary University of London and is owned by these universities.
have been encouraged to intervene earlier when faced Massachusetts General Hospital and Queen Mary University of London
with rising biomarker concentrations and normal granted a licence to ROCA to Abcodia in 2014. IJJ is non-executive
director, shareholder, and consultant to Abcodia and has rights to
imaging.30 This theory is supported by modelling, which royalties from sales of the ROCA. He founded (1985), was a trustee of
suggests that the majority of stage I, type II epithelial (2012–14), and is now an Emeritus trustee (2015–present) of The Eve
cancers are 0·4–1·3 cm in diameter and, therefore, Appeal, one of the funding agencies for UKCTOCS. All other authors
difficult to reliably image.31 declare no competing interests.
We began our mission to reduce deaths from ovarian Data sharing
For the protocol see http:// cancer in the 1980s,5,6 started our pilot randomised The protocol is available on the study website. The individual participant
ukctocs.mrcctu.ucl.ac.uk/ data that underlie the results reported in this Article, after
media/1066/ukctocs-protocol_
controlled trial in the mid-1990s,32 and then undertook de-identification, will be available beginning 12 months after publication.
v90_19feb2020.pdf UKCTOCS over the next 20 years. The journey has Researchers will need to state the aims of any analyses and provide a
involved more than 200 000 women who have trusted us methodologically sound proposal. Proposals should be directed to
and generously given their time, multiple UK funding [email protected]. Data requestors will need to sign a data access
agreement and in keeping with patient consent for secondary use, obtain
agencies, numerous NHS staff both at the trial centres ethical approval for any new analyses.
and more widely, and the support of many UK and
Acknowledgments
international charities, and expert groups. To them, we We thank the volunteers without whom the trial would not have been
are hugely grateful. While disappointing, the trial has possible and everyone involved in conduct and oversight of UKCTOCS.
provided a clear answer that our screening strategies We are very grateful to the current members of the UKCTOCS Trial
Steering Committee: Henry Kitchener (Chair), Julietta Patnick,
coupled with treatment protocols available in 2001–11
Jack Cuzick, and Annwen Jones. We are indebted to the administrative
(the active screening phase) did not save lives. Currently, support provided by Anna Widdup, Roxanne Payne, and Jasvinder
general population screening for ovarian and tubal Dinza. We are indebted to the funding agencies for working together to
cancer cannot be recommended. We remain optimistic support this long and challenging trial, and for their sustained support
for UKCTOCS over many years. The long-term follow-up UKCTOCS is
that further research will develop more effective ways to
supported by National Institute for Health Research (NIHR HTA grant
detect and treat this lethal disease. Meanwhile, the 16/46/01), Cancer Research UK, and The Eve Appeal. UKCTOCS was
UKCTOCS biorepository with longitudinal samples funded by Medical Research Council (G9901012 and G0801228), Cancer
provides a unique opportunity to advance early detection Research UK (C1479/A2884), and the UK Department of Health, with
additional support from The Eve Appeal. Researchers at UCL are
biomarker research.
supported by the NIHR UCL Hospitals Biomedical Research Centre and
Contributors by Medical Research Council Clinical Trials Unit at UCL core funding
UM was the chief investigator from 2015 to present and was co-chief (MR_UU_12023). The views expressed are those of the authors and not
investigator from 2001–14. MP was the trial statistician. IJJ was chief necessarily those of the NHS, the NIHR, or the UK Department of
investigator from 2000–14 and was a co-investigator from 2015. MP, SJS, Health and Social Care.
SC, AJM, and LF were co-investigators for whole study, 2001 to present.
References
UM, MP, IJJ, SJS, SC, AJM, LF, AG-M, MB, AR, and JKK contributed to
1 Cancer Research UK. Ovarian cancer survival statistics. 2018.
current study concept and design. AG-M, MB, and UM did the literature https://www.cancerresearchuk.org/health-professional/cancer-
search. AR, AG-M, CK, GC, JT, and SKM collected the data. Outcomes statistics/statistics-by-cancer-type/ovarian-cancer/survival#heading-
review was done by NS (chair), RM, RW, RA, AS, KW, and LC. AR and Three (accessed Jan 21, 2021).
UM prepared the dataset used for the analysis. UM and AR verified the 2 Jacobs IJ, Skates SJ, MacDonald N, et al. Screening for ovarian
data. All statistical analysis was done by MB under the supervision of cancer: a pilot randomised controlled trial. Lancet 1999;
MP and independently verified by YL under the supervision of SJS. 353: 1207–10.
The primary analysis was also independently repeated by MR under the 3 van Nagell JR Jr, Burgess BT, Miller RW, et al. Survival of women
supervision of AJM. AR, MB, AG-M, and UM prepared the figures and with type I and II epithelial ovarian cancer detected by ultrasound
tables and drafted the manuscript. UM and AR extracted the dataset. screening. Obstet Gynecol 2018; 132: 1091–100.
MB, SJS, YL, AJM, MR, UM, and AR had full access to the dataset. 4 Kobayashi H, Yamada Y, Sado T, et al. A randomized study of
UM had final responsibility for the decision to submit for publication. screening for ovarian cancer: a multicenter study in Japan.
All authors contributed to the interpretation of the data and revision of Int J Gynecol Cancer 2008; 18: 414–20.
the manuscript. All authors approved the report before submission. 5 Campbell S, Bhan V, Royston P, Whitehead MI, Collins WP.
Transabdominal ultrasound screening for early ovarian cancer. BMJ
Declaration of interests 1989; 299: 1363–67.
UM has stock ownership awarded by University College London (UCL) 6 Jacobs I, Stabile I, Bridges J, et al. Multimodal approach to
in Abcodia, which holds the licence for ROCA. She has received grants screening for ovarian cancer. Lancet 1988; 1: 268–71.
from the Medical Research Council (MRC), Cancer Research UK, 7 Buys SS, Partridge E, Black A, et al. Effect of screening on ovarian
the National Institute for Health Research (NIHR), and The Eve Appeal. cancer mortality: the Prostate, Lung, Colorectal and Ovarian (PLCO)
She holds patent number EP10178345.4 for Breast Cancer Diagnostics. cancer screening randomized controlled trial. JAMA 2011;
MP has received grants and AG-M, MB, AR, CK, GC, and SKM have 305: 2295–303.
been funded by grants from MRC, Cancer Research UK, NIHR, and 8 Pinsky PF, Yu K, Kramer BS, et al. Extended mortality results for
The Eve Appeal. RM has received grants from The Eve Appeal, Rosetrees ovarian cancer screening in the PLCO trial with median 15 years
Charity, and Barts Charity, and personal fees from AstraZeneca. follow-up. Gynecol Oncol 2016; 143: 270–75.
SJS holds the (expired) patent for ROCA, patented and owned by 9 Jacobs IJ, Menon U, Ryan A, et al. Ovarian cancer screening and
Massachusetts General Hospital and Queen Mary University of London, mortality in the UK Collaborative Trial of Ovarian Cancer Screening
licenced to Abcodia. He reports stock options from SISCAPA Assay (UKCTOCS): a randomised controlled trial. Lancet 2016; 387: 945–56.
Technologies, and personal fees from Abcodia, Guardant Health, and

2192 www.thelancet.com Vol 397 June 5, 2021

 Articles

10 Menon U, Gentry-Maharaj A, Ryan A, et al. Recruitment to 22 Dilley J, Allen P, Gentry-Maharaj A, et al. Symptoms reported by
multicentre trials—lessons from UKCTOCS: descriptive study. BMJ women with screen detected invasive epithelial ovarian cancer in
2008; 337: a2079. the United Kingdom Collaborative Trial of Ovarian Cancer
11 Menon U, Gentry-Maharaj A, Hallett R, et al. Sensitivity and Screening (UKCTOCS). Int J Gynecol Cancer 2015;
specificity of multimodal and ultrasound screening for ovarian 25 (suppl 2): 423–24.
cancer, and stage distribution of detected cancers: results of the 23 Menon U, Ryan A, Kalsi J, et al. Risk algorithm using serial
prevalence screen of the UK Collaborative Trial of Ovarian Cancer biomarker measurements doubles the number of screen-detected
Screening (UKCTOCS). Lancet Oncol 2009; 10: 327–40. cancers compared with a single-threshold rule in the United
12 UKCTOCS. Protocol for the United Kingdom Collaborative Trial of Kingdom Collaborative Trial of Ovarian Cancer Screening.
Ovarian Cancer Screening (UKCTOCS) and the Long Term Impact J Clin Oncol 2015; 33: 2062–71.
of Screening on Ovarian Cancer Mortality (LTFU UKCTOCS). 2020. 24 Blyuss O, Burnell M, Ryan A, et al. Comparison of longitudinal
http://ukctocs.mrcctu.ucl.ac.uk/media/1066/ukctocs-protocol_ CA125 algorithms as a first-line screen for ovarian cancer in the
v90_19feb2020.pdf (accessed Jan 1, 2021). general population. Clin Cancer Res 2018; 24: 4726–33.
13 Skates SJ, Pauler DK, Jacobs IJ. Screening based on the risk of 25 Tabár L, Yen AM, Wu WY, et al. Insights from the breast cancer
cancer calculation from bayesian hierarchical changepoint and screening trials: how screening affects the natural history of breast
mixture models of longitudinal markers. J Am Stat Assoc 2001; cancer and implications for evaluating service screening programs.
96: 429–39. Breast J 2015; 21: 13–20.
14 Kalsi JK, Ryan A, Gentry-Maharaj A, et al. Completeness and accuracy 26 Schoen RE, Pinsky PF, Weissfeld JL, et al. Colorectal-cancer
of national cancer and death registration for outcome ascertainment incidence and mortality with screening flexible sigmoidoscopy.
in trials-an ovarian cancer exemplar. Trials 2021; 22: 88. N Engl J Med 2012; 366: 2345–57.
15 Meinhold-Heerlein I, Fotopoulou C, Harter P, et al. The new WHO 27 de Koning HJ, van der Aalst CM, de Jong PA, et al. Reduced lung-
classification of ovarian, fallopian tube, and primary peritoneal cancer mortality with volume CT screening in a randomized trial.
cancer and its clinical implications. Arch Gynecol Obstet 2016; N Engl J Med 2020; 382: 503–13.
293: 695–700. 28 Strauss GM, Gleason RE, Sugarbaker DJ. Chest X-ray screening
16 Daya D, Cheung AN, Khunamornpong S. Tumors of the improves outcome in lung cancer. A reappraisal of randomized
peritoneum: epithelial tumors of Müllerian type. In: Kurman RJ CM, trials on lung cancer screening. Chest 1995; 107 (suppl): 270–79S.
Herrington CS, Young RH, eds, ed. WHO classification of tumors of 29 Schröder FH, Hugosson J, Roobol MJ, et al. Prostate-cancer
female reproductive organs. 4th edn. Lyon: International Agency for mortality at 11 years of follow-up. N Engl J Med 2012; 366: 981–90.
Research on Cancer, 2014: 92–93. 30 Menon U, Gentry-Maharaj A, Ryan A, et al. Serial CA125
17 Tavassoli FA, Devilee P. Tumors of the breast and female genital interpreted using the risk of ovarian cancer algorithm can detect
organs. World Health Organization Classification of Tumours: ovarian cancer in absence of ultrasound abnormalities Innovations
Pathology and Genetics. Lyon: World Health Organization, 2003. & Progress in Healthcare for Women, 2nd International Meeting:
18 Burnell M, Gentry-Maharaj A, Skates SJ, et al. UKCTOCS update: Prevention, Screening and Risk Prediction in Women’s Health;
applying insights of delayed effects in cancer screening trials to the London; Oct 16, 2009. AB.021 16 (abstr).
long-term follow-up mortality analysis. Trials 2021; 22: 173. 31 Brown PO, Palmer C. The preclinical natural history of serous
19 Karrison TG. Versatile tests for comparing survival curves based on ovarian cancer: defining the target for early detection. PLoS Med
weighted log-rank statistics. Stata J 2016; 16: 678–90. 2009; 6: e1000114.
20 Rosenthal AN, Fraser LSM, Philpott S, et al. Evidence of stage shift 32 Menon U, Skates SJ, Lewis S, et al. Prospective study using the risk
in women diagnosed with ovarian cancer during phase II of the of ovarian cancer algorithm to screen for ovarian cancer.
United Kingdom familial ovarian cancer screening study. J Clin Oncol 2005; 23: 7919–26.
J Clin Oncol 2017; 35: 1411–20.
21 Skates SJ, Greene MH, Buys SS, et al. Early detection of ovarian
cancer using the risk of ovarian cancer algorithm with frequent
CA125 testing in women at increased familial risk—combined
results from two screening trials. Clin Cancer Res 2017; 23: 3628–37.

www.thelancet.com Vol 397 June 5, 2021 2193