AgeX: a clinical trial of breast cancer screening

Seven years into the Age Extension trial of mammography screening, in September 2016, the target number of trial participants was quietly doubled, to "at least six million". The upper limit to the age range was extended from 73 to 79, inflicting extra mammograms on women already at increased risk from overdiagnosis and overtreatment. It has been re-named and re-branded; there have been changes to the protocol, the research hypothesis, and to the primary and secondary outcomes.

HealthWatch has been highlighting concerns over this mammoth trial almost since its inception, and our recent BMJ Analysis* describes the ethical flaws. In this HealthWatch Update we describe these and the procedural issues around the trial in more detail. We ask, will this research, with its constantly-shifting goalposts, ever be fit for publication in a serious peer-reviewed journal?

 So many protocols

The trial just keeps on changing. This is a problem because changing aspects of a trial after it has started risks making the final results biased. In a trial like this, where you need huge number of participants to achieve a tiny difference in outcome, small amounts of bias or contamination could have a serious effect on the results. It makes for bad science. And bad science is a bad foundation on which to base public health decisions.

The changes over time are listed on public registries. Here are the trial's pages on ISRCTN, and ClinicalTrials.gov. On clinicaltrials.gov there's also an archive page for the trial where you can compare the different versions side-by-side.

So far as we know there have been 4 major versions of the protocol to date. We obtained v1 and v3 only after Freedom of Information requests (we have never seen v2), but v4 is now freely available online. When it appeared in September 2016 it was not numbered or dated, as per Good Clinical Practice (GCP), so it was going to be difficult to spot future version changes. Hence we are archiving the original iteration of v4 on our website below, along with the previous versions that we obtained.

- Version 1: Evaluating the net effects of extending the age range for breast screening in the NHS breast screening programme in England from 50-70 years to 47-73 years

- Version 3: Nationwide cluster-randomised trial of extending the NHS breast screening age range in England: protocol (redacted version released following Freedom of Information request)

- Version 4: Nationwide cluster-randomised trial of extending the NHS breast screening age range in England: AgeX trial protocol of 28 September 2016

Note: on the subject of GCP, because the trial is not of a treatment or device, GCP does not strictly apply. However the Department of Health's Research Governance Framework (2005) says "The principles of Good Clinical Practice apply to all research involving patients, not just clinical trials".

The target number of trial participants has quietly been doubled

We say "quietly", because this vast increase in participation does not seem to be mentioned on the July 2016 application to the ethics committee that shortly preceded the increase, nor in the 5 September 2016 letter that communicated the ethics committee's favourable opinion on that application. Yet, on ISRCTN it states that on 28th September 2016 the number of trial participants was increased from 3 million to 6 million. That means now around two thirds of the female population of England who are in the target age groups during the trial’s duration will be part of the study.

But wait, to be fair, the current version of the protocol says: "The total number of women entering the trial (half invited for additional screening and half not) is not a fixed sample size. If substantial uncertainty still persists, randomisation may well continue." Is it OK then, to keep enlarging the sample size as you go along? HealthWatch is short on statistical expertise in-house, so we asked an expert. Adam Jacobs, who blogs as StatsGuy, commented, "speaking as both a biostatistician and a former ethics committee member, on the subject of doubling the study population from 3 million to 6 million without mentioning it in the ethics application: 'WTF?'" We thought so too.

Remember, the original trial protocol stated: "The age extension will proceed regardless of whether this study goes ahead or not, and therefore regardless of whether the phasing-in is randomised or not."

Plans for extra screens putting older women at even more risk

As well as inviting more younger women for screening, women aged over 70 are now being targeted for extra screens. As it says in the Sept 2016 protocol: "Although AgeX began as a trial of additional screening at ages 47-49 and at ages 71-73, it has therefore become a trial in which the older women allocated additional screening can, where resources are available, continue be invited triennially at ages 71-76 or at ages 71-79, thereby assessing the effects of continuing triennial screening for several years after age 70." Note: the new move to include women aged 74-79 will not in itself add to the number of trial participants as it will be repeat screens on the group who were enrolled at 71-73.

The rationale for this change in a research trial is the recommendations of the All-Party Parliamentary Group on Breast Cancer's (APPGBC) enquiry into older age and breast cancer. First in 2013 and again in 2015, the APPGBC called for the Age Extension trial to include additional rounds of screening to women aged 74-76 and, if uptake is sufficient, 77-79. In the 2015 report, disappointment was expressed that there was little possibility of securing the funding necessary to extend the trial to older age groups (two-thirds of the trial funding was committed to the 47-49 age group due to the greater numbers). It seems the money has been found.

Why push to include older women in AgeX? The APPGBC justifies the call by the need for evidence on which to base screening advice for older women. Yet little account seems to have been taken of evidence that already exists of increased overdiagnosis and increased dangers from overtreatment in this age group. For example, this Dutch population-based study of breast screening in women aged 69-75, found only very limited benefits of screening and high levels of overdiagnosis, and concluded with a recommendation that screening of women over 70 should not be conducted on a large scale. Along with the distress and suffering caused by treatment of overdiagnosed cancers, older women are less able to tolerate cancer treatments and more likely to suffer from side effects and post-operative complications.

If that should not persuade researchers to think hard about subjecting elderly women to unnecessary harms, UK NIHR research published a modelling study that went online in September 2015. Although it said that extending screening to age 78 would be a cost effective use of NHS resources, how it reached that conclusion is open to question, given the lack of evidence of benefit in survival, and the fact that its model predicted that rates of overdiagnosis at age 72 would be greater than 6%, rising to 15% at age 78.

New information leaflets still underplay harms

Since our last update in June 2015 we can claim some more but modest 'success' in forcing the issue of women being given transparent information about the Age Extension trial. There’s more information online at a new website www.agex.uk, and an information leaflet for trial participants has grown from 1 to 4 pages long.

Unfortunately, the information given in the leaflet is vague about the risks to which participants may be subjected. There is the chance "of being diagnosed and treated for a non-life-threatening cancer" without description of the nature of the treatments or their potentially deadly or disfiguring side effects; "being asked to return for more tests" again, these tests and any associated risks are not described; and a "slightly lower or higher" chance of having your death from breast cancer prevented. Participants are then cross-referred to the pink "Helping you decide" breast screening for 50-70 leaflet for more information, whose risk data is not relevant to women outside that age range.

The pink 50-70 leaflet and the trial participant leaflet state that 1 breast cancer death is prevented for every 200 women screened. They do not give details of harms resulting from overdiagnosis or, crucially, for deaths caused by treatment, which may result from radiotherapy-related cardiovascular damage or generation of other cancers. We will continue to push for the use of better decision aids, such as the icon array diagrams produced by the Harding Center for Risk Literacy. Their simple and truthful guide: Early detection of breast cancer by mammography screening, is based on the German health system, in which women aged 50-69 are screened every two years, and presents the key facts based on recent evidence that is assessed as moderate to high quality. It concluded that mammography screening had no effect on the total number of deaths when all deaths from all types of cancer are considered.

In 2015 we joined a group letter calling for the Gov.UK website to include up to date information about overdiagnosis and overtreatment and a more balanced view better information about breast cancer screening, by including links to the Cochrane leaflet "Screening for breast cancer with mammography". NHS Screening Programmes responded that there were no plans to update the Gov.UK screening website content.

Latest screening rules demand evidence of lives saved or less disease

Following the publication of the House of Commons Science and Technology Select Committee report in June 2015 ("Review of the UK National Screening Committee 2015"), cancer and non-cancer screening have been combined under Public Health England. Breast screening is now in the hands of the National Screening Committee. We are pleased about this change, which we had called for some years ago.

The UK National Screening Committee (UK NSC) criteria for appraising the viability, effectiveness and appropriateness of a screening programme were updated in October 2015. In line with the UK NSC review recommendations above, they say (point 4.11):

"There should be evidence from high quality randomised controlled trials that the screening programme is effective in reducing mortality or morbidity".

Mortality means lives saved overall, not lives saved from that particular disease. It should take into account deaths from conditions that result from treatment.

But evidence from high quality trials says that breast screening does not reduce overall mortality or morbidity. So it doesn't meet UK NSC criteria to be a screening programme, right?

But the screening programme does not aim to reduce deaths

Julietta Patnick is no longer director of the National Health Service Breast Screening Programme (NHS BSP) in Cancer directorate. In 2015 the Public Health England blog announced her retirement as head of NHS BSP and she moved to Oxford to work with the trial research team. The new director of the National Screening Committee is Anne Mackie.

So it was Anne Mackie to whom we sent our concerns at the end of 2015. You can read our letter to Anne Mackie of 7th November 2015, along with her response of 18th November 2015 in full. She told us that the requirement of the screening programme is not to reduce all cause mortality but rather that "The purpose of the trial is to establish whether there is benefit or not to those offered or taking up screening." We disagree that benefit only should be looked at without countervailing harm. Deaths can be caused as well as prevented. It is wrong to only look at one side of the equation.

Who 'owns' the Age Extension trial?

In recent years we have written to every authority associated with the Age Extension trial. None would claim responsibility for the quality of the science. None share our concern that there has been no transparent peer review at any stage, nor no systematic review of the existing evidence before framing the research question and constructing a protocol.

There has even been confusion about where the money is coming from.

The ISRCTN currently lists the trial funders as the University of Oxford and Public Health England, but in 2015 the list of funders also included Cancer Research UK and the Medical Research Council.

We contacted both at the time to ask about the extent of this funding. It seems clear that neither played a part in the genesis of the trial and would rather keep their distance from it. In their replies both said that they had no direct financial input into the trial but instead made available a single member of their research staff to work on the trial as a shared resource with their other projects.

We asked the MRC about the trial’s governance processes and the extent to which they were funding the trial. The response from the MRC on 21st December 2015 explained that their support is limited to allowing one of their full-time researchers, Professor Sir Richard Peto, to devote some of his time to the trial. They said they do not provide funding. Neither do they have any involvement with trial governance. They did not have any involvement with peer review of the trial.

The CRUK reply of 9th December 2015 was more emphatic, complete with underline: “It is important to clarify that CRUK is not a funder of the trial itself and thus the questions you raise in your email below do not apply.” The limit of CRUK’s involvement is providing programme grant support to Professor Dame Valerie Beral at the University of Oxford.

We complained to the ISRCTN registry about the reluctance of these funding bodies to be considered as such, and the registry entry was subsequently changed. The funders are currently shown as “Department of Health funds allocated to Public Health England” and “Study co-ordination and data analysis is funded from the quinquennial core support for the Cancer Epidemiology Unit and for the Clinical Trial Service Unit (both in the University of Oxford’s Nuffield Department of Population Health) from Cancer Research UK and the Medical Research Council”. The new AgeX website (of 2016) repeats the same text.

But the University of Oxford remains as the trial sponsor. Should they not have a role in overseeing the research they sponsor?

Who will take responsibility for the scientific content of the Age Extension trial?

In documents we obtained via Freedom of Information requests, the investigators say, referring to the 2012 Marmot report by the Independent Breast Screening Review Panel, "The Panel supports the principle of the ongoing trial in the UK for randomising women under age 50 and above age 70 to be invited for breast screening". The office of Sir Michael Marmot wrote to us on 7 October 2014, making it clear that any support for the Age Extension trial had been given only in principle, not having seen the details of the trial that was already underway. As the panel is now disbanded there can be no further comment, the letter says.

There can be no dispute now that the trial sponsor is the University of Oxford, and the funder is Public Health England. But which is responsible for the scientific content? The University of Oxford claim that responsibility for formal review of the trial is the remit of the Ethics Committee, and not the University’s (see 20 August 2015 letter from Prof Ewan McKendrick).

On pressing, he sent us a letter from Professor Michael Parker, professor of bioethics at the University of Oxford, who had been tasked with reviewing our complaint. In Professor Parker’s response, it is clear that he was keen to align with the NHS Health Research Authority (HRA), the body that oversees UK government-funded health research, and under whom the relevant Ethics Committee operates. He conferred with them before responding, and said that the responsibility for scientific oversight lay with the Research Ethics Committee.

You will remember, the HRA previously addressed our complaint in detail (see our earlier update 4 from June 2015). They rejected our contention that the protocol was flawed with the statement "appropriate peer review is the responsibility of the Sponsor". So, we come full circle.

But is Professor Parker a completely independent investigator? Up to April 2016 he was working in the same department as other trial protagonists Valerie Beral, Richard Peto and Julietta Patnick: all were senior staff at the Nuffield Department of Population Health Medical Sciences Division (see screen shots).

Public Health England seems to reject any independent scrutiny of the trial

Independence is clearly hard to come by. Our complaints to Public Health England, including our concern that the principal investigator is not a doctor or even scientifically trained, were answered by its chief executive Duncan Selbie (17 December 2014 and 19 February 2015). He'd initially said her boss, Kevin Fenton (then director of Wellbeing) would be investigating. But in the end it was dealt with by way of inviting a briefing on the history of the trial to his Board from none other than Julietta Patnick, the trial's principal investigator.

We do not believe this is proper process.

Our next complaint, made direct to the then health secretary Jeremy Hunt, did not even merit a response. Jeremy Hunt’s staff simply referred us to Duncan Selbie’s letter above.

It may be coincidence, but early in 2017 we noticed a link to a new document titled "PHE External Review of AgeX" had appeared, without fanfare, on the public AgeX website. The document, in the form of a letter from NSC director Anne Mackie to AgeX investigator Valerie Beral, begins: "PHE consulted three external academics to review the revised AgeX protocol with regards to its scientific merit". It goes on to report their very limited assessments of the scientific methods, research methods including recruitment and the research team, resources and management.

Under a heading "Conditions" it says: "All reviewers were informed that their identities and affiliations would remain anonymous." The letter gives no information about the level or area of expertise of the academics, or on what basis they were selected. Although the reviewers are said to be "external", nowhere does the letter claim they are independent. Their verdict? “The three external reviewers all support the trial methodology and consider it has much scientific merit”. Anonymously.

We maintain that this post-hoc fig-leaf is no substitute for proper, independent, scientific peer review.

Ethical oversight of the trial is inadequate

Our previous updates have detailed how the Age Extension trial has struggled to achieve ethics committee approval for its various amendments, with several rejected before being cleared. We don’t know whether and in what detail the original REC examined the science (beyond the fact that they didn’t appear to have a statistician on the REC, which itself raises questions). We struggle to obtain minutes of individual meetings but we have so far received those from the Data Monitoring and Ethics Committee (DMEC) meetings of 8th November 2011, 9th January 2013, and 28th January 2014, and the REC meeting of 14/10/2014. While HRA assure us (see above) that our complaints have been considered by committee, we cannot find mention of them in the minutes.

In 2016 we wrote to the British Medical Association’s Ethics Committee who said they couldn’t deal with individual ethical problems, but referred us to the BMA Board of Science. She gave us a number of names, which took us to the National Cancer Research Institute, who then referred us back to the NHS HRA.

It seems that once a trial has received the go-ahead from the Research Ethics Committee, there is no process by which it can be objected to, or halted. Although scientists may appeal a rejection from an Ethics Committee, an approval is rarely, if ever, objected to. In fact, NHS HRA had to come up with a special appeal process for us as an interested third party, although we were not invited in person to explain our concerns.

The pilot study into "feasibility" and "acceptability"

The main trial’s original registration stated: “The study builds on the pilot study (Pilot study of the feasibility and acceptability of randomising the phasing-in of the age extension of the NHS Breast Screening Programme in England: ISRCTN50037017) which investigated in several pilot sites the feasibility and acceptability of randomising the phasing-in of the age extension.” 

The results of the pilot study have been called into question. Biostatistician Adam Jacobs posted on PubMed Commons on 17 July 2014 that the pilot, which was published in the Journal of Medical Screening in 2011, had concluded "No major problems of feasibility or acceptability of randomization were found. This pilot study has informed the randomized phasing-in of the age extension across the whole of England." Were the conclusions justified? Jacobs writes: "I fail to see how the authors' conclusions of 'no major problems of feasibility or acceptability of randomisation were found' ... With no attempt made by the authors to quantify the increased workload, it is hard to judge how it might impact on feasibility", and as for acceptability, "it is not surprising that the authors found no problems, as they did not look for them. I could not find any mention of the study participants having been asked questions about acceptability." Indeed, as women were at that stage not explicitly told they were in a trial, it is unlikely they would object. It is also not correct that there were no problems of acceptability, because the medical author Mitzi Blennerhassett realised that her mammogram invitation was part of a research trial and objected vigorously. We are grateful to her for being the first to draw the Age Extension trial to our attention.

Clearly no such problems were anticipated, as the triallists did not wait for results before the main trial was begun.

Research into excess surgery resulting from the trial

This paper published in the International Journal of Surgery in 2013 estimated the impact of the Age Extension trial’s screening of 47-49-year-olds on surgical departments. It’s open access, so you can download it free to read in full. But here’s the gist. Of 4250 women screened, almost one in 10 were recalled, and a tenth of those recalled underwent surgery. Seven had mastectomies.

The study was measuring excess surgical burden arising from the screening of this group who would not otherwise have been screened. Most lesions detected, it says, were impalpable (could not be felt). It is impossible to know for certain whether the screening and subsequent treatment affected the outcomes for the women concerned. The paper speculates that early detection will have made the lesions more amenable to breast conserving surgery. But this runs counter to more recent evidence that mammography screening does not seem to reduce the number of tumours reaching late stage, or lead to less invasive treatment - screened groups have more, not fewer, radiotherapy, lumpectomies and mastectomies.

A point to note is that the data sample for this paper was taken from The North & East Devon Breast Screening Unit (BSU), which was the first unit to invite women in the age group of 47–49, starting January 2011. But although the Age Extension trial protocol describes a randomised controlled trial, women here are NOT randomised – InHealth Group Ltd, which operates the stand alone North & East Devon Breast Screening Unit, does not accept randomisation, hence every women aged 47-49 will be invited. We could find no mention in the trial protocols of how the data would be adapted in the event of randomization not being possible in any region.

Radiologists’ workload

The Royal College of Radiologists (RCR) continues to experience challenges in staffing and workload. BBC News recently reported that workload had increased by 30% between 2012 and 2017, while the number of consultant radiologists had increased by only 15% and the mismatch could be a threat to cancer care.

In 2016 they were already reporting a "workforce crisis in UK breast cancer screening and diagnostic services". Their survey of UK diagnostic breast screening services revealed "major challenges facing services as they cope with an ever-increasing demand for tests". In view of this, we requested a meeting with Giles Maskell, then president of the RCR. It was disappointing to receive his reply, which mentioned the trial’s ethical approval and said: "The pressures on the symptomatic service will not be resolved by cessation of the trial early before evidence for or against roll out of age extension is accrued." It is baffling to be told that the removal of millions of breast screens would not alleviate radiologists' workload. Even more so to see, months later, a BMJ article entitled "Radiology services are 'on their knees,' royal college warns". Growth in the workforce has not kept pace with demand, says the report, quoting Dr Maskell: "there is currently no light at the end of the tunnel."

Following a change in RCR President, we approached them again and were pleased that a member, Professor Fiona Gilbert, took matters seriously. A face-to-face discussion with new president Nicola Strickland took place in September 2017. However, the RCR could not broker a meeting with the trialists with themselves as neutral third parties, due to their refusal to meet HealthWatch.

The tide is turning, or is it?

The tide of expert opinion seems to be slowly turning. The major medical journals are increasingly questioning mammography screening. And here are a couple of great simply explained blogs, so it can be explained to the general public: in Scientific American, and HealthNewsReview.org.

But the pushback will be substantial. In the US where private healthcare systems can profit from overtreatment, US health insurance companies are offering women cash incentives to take unnecessary mammograms, according to a New York Times article from 2016.

But, in our publicly funded NHS, things are different, or are they? The public tends to overestimate the benefits of screening and underestimate its risks, a misconception that has been broadly encouraged by cancer charities and government bodies. It's free, girls, it shows we care about you. Is there any political party who holds the public trust with such confidence they would dare to snatch those mammograms away?

In a nutshell

The UK government (NHS breast screening programme) and Oxford University are performing the largest randomised controlled trial in history – a human experiment on 6 million women without their fully informed consent – that appears to have been 'cooked up' by a small group of people, possibly as a political PR exercise, and which has an inadequate and changing rationale. 

The randomised controlled trial (RCT) arose from not having enough money to 'roll out' the extended age range of women proposed by the Labour Government’s 2007 Cancer Strategy which, recognising that late diagnosis is a factor contributing to poor cancer survival rates in the UK, promised to extend breast screening to 47-73 year olds by 2012. The strategy was based on the assumption that mammography screening was effective at preventing deaths from breast cancer, which was the UK establishment’s view at that time.

This cluster RCT did not go through any recognisable scientific peer review process. The principal investigator is neither a scientist nor a clinician. The trial's research governance committees were only bolted on at a late stage, following our protests and after the 'belief of benefit' came into question following the publication of the Marmot review in 2012, which highlighted and quantified the problem of overdiagnosis. We believe this evidence should have been sufficient to halt the trial or at least pause for a moratorium whilst independent scientific review was undertaken.

The trial is being funded by the taxpayer and is a burden on hard pressed screening services.

And worst, worst of all – more women than ever are being harmed unnecessarily, and unknowingly, and in the belief that breast screening for cancer is in their own best interests. Thousands will have needless disfiguring surgery, accepting it gratefully, believing their lives have been saved.

Appeals we have made about the science and ethics have been passed between Oxford University, Public Health England, and the NHS Health Research Authority, who seem to rely on one another’s approvals to pass around major chunks of the responsibility for the poor science and absent peer review. Organisations who have been publicly 'badged' as having funded the trial wash their hands of it. Reputable scientists who’ve looked at the protocol on our request expressed serious concerns.

We've exposed a problem that just won't go away – is this clinical trial simply "too big to fail"?

Mandy Payne

*A full text of the BMJ Analysis can be accessed here