- The trial design is not adequate to answer the questions that arise about screening.
- The statistical analysis strategy is flawed.
- The information brochure for women who might participate is dishonest.

MyPeBS poses at least 3 problems :

There are three options for the future of screening: move toward individualized screening, maintain current screening, or discontinue screening.

By comparing an individualized screening group to a standard screening control group, we can only answer the question : does individualized screening perform
better or worse than current screening ?

The answer will only provide a choice between moving toward individualized screening and continuing with current screening. It will not provide any information
to allow you to choose between individualized or standard screening or no screening at all.

This is especially unfortunate given that the value of organized screening no longer seems as obvious as when it was first fully implemented.

- In terms of benefits, the expected 20% reduction in mortality is based on old studies and has not been found in recent studies.
- Overdiagnosis may have been underestimated in terms of risks, as recent studies place its frequency closer to 40% than the 10% initially reported. Furthermore, we must remember that overdiagnosis= unnecessary treatment= side effects, sometimes severe, with no benefit in counterpart.

MyPeBS thus represents **a missed opportunity** : the opportunity to answer, using current data, the question of whether organized screening
should be abandoned or maintained in its current form, or whether it should evolve toward individualized risk-based screening.

It would have sufficed to plan and compare three groups :

- an individualized screening group,

- a standard screening group,

- **a no screening group**.

Of course, this implies agreeing to call screening into question if the study does not demonstrate that it is superior to no screening.

And it seems that MyPeBS sponsors are not ready for this challenge.

Let us first recall the analysis strategy foreseen in the MyPeBS protocol :

- 1st step = primary objective = demonstrate "non-inferiority" of individualized screening compared to standard screening

- 2nd step = secondary objective = if "non-inferiority" is confirmed, demonstrate the superiority of individualized screening over standard screening.

Contrary to what one might think, for statisticians, A non-inferior to B does not mean A is at least as good as B.

For statisticians, A non-inferior to B means that A may be inferior to B but that this inferiority does not exceed a certain threshold. In the case of MyPeBS,
this non-inferiority threshold is set, arbitrarily, at -25%.

To fully grasp what this "non-inferiority" means, keep in mind that clinical trial results are inevitably subject to random variability. In other words, if the trial is repeated exactly the same way but with a different sample, the results will not be identical to the first time. And, a priori, neither result is more accurate than the other. The outcome of a clinical trial cannot be said to be an exact reflection of reality. Statisticians, on the other hand, can calculate a range on either side of the result that has a 95 percent chance of containing the truth (this range is called the 95% confidence interval).

For "non-inferiority" at the -25% threshold to be considered as demonstrated by MyPeBS, it is sufficient that the lower bound of the 95% confidence
interval is greater than -25%.

As illustrated in the diagrams below (see Eventuality 2), this way of proceeding can lead to the conclusion of "non-inferiority" when in fact one is sure
that there is inferiority.

In the diagrams below, the red line represents the 95% confidence interval of the difference in performance between individualized and standard screening. The 6 possibilities represented correspond to the 6 possible ways of positioning this confidence interval with respect to 0% (no difference in performance) and -25% (the performance of individualized screening is 25% lower than that of standard screening). | ||

Eventuality 1: The lower limit of the 95% confidence interval (red line) is less than -25%. There is inferiority and MyPeBS will correctly conclude on the inferiority. |
||

Eventuality 2: The lower limit of the 95% confidence interval (red line) is greater than -25%
and the upper limit is less than 0%.There is definite inferiority but MyPeBS will nevertheless conclude non-inferiority.. |
||

Eventuality 3: The lower limit of the 95% confidence interval (red line) falls below 0% and
the upper limit above but with a larger part of the confidence interval below 0%. There is a doubt with a tendency to inferiority but MyPeBS will conclude to non-inferiority. | ||

Eventuality 4: The lower limit of the 95% confidence interval (red line) falls below 0%
and the upper limit above and the interval is equally distributed on either side of 0%. There is complete doubt but MyPeBS will conclude non-inferiority. |
||

Eventuality 5: The lower limit of the 95% confidence interval (red line) falls below 0%
and the upper limit above but with a larger part of the confidence interval above 0%. There is a doubt with a tendency to superiority and MyPeBS will conclude to non-inferiority. |
||

Eventuality 6: The lower limit of the 95% confidence interval (red line) is above 0%. There is superiority and MyPeBS will correctly conclude to superiority. |

Thus,**MyPeBS may conclude "non-inferiority" when the situation is at best doubtful (eventualities 3 and 4) or, worse, when the inferiority is
certain (eventuality 2)**.

Furthermore, with a non-inferiority threshold as low as -25 percent, it is clear that scenario 1 is extremely unlikely to occur. In other words, **the die
has been cast**, and a favorable outcome for individualized screening can already be predicted before the trial even begins. The only question is whether
MyPeBS will conclude superiority (eventuality 6) or simply "non-inferiority" (eventuality 2-5).

This real intellectual swindle is all the more unacceptable because we know that in the expression of results, we will see a double semantic shift : **non-inferior
to -25% will become non-inferior, and non-inferior will become at least as effective as**. To verify this, simply read the
patient information brochure or the
press kit.
It is specifically stated :

- on page 2 of the informational brochure, « The MyPeBS study was designed to evaluate, in women aged 40 to70, whether personalized breast cancer screening based on
individual risk of developing breast cancer in the next 5 years is at least as effective as current standard screening. »

- and in the press kit (page 4), « MyPeBS: European mobilization to offer personalized, more effective and safer
screening » and (page 5) « demonstrating that stratified screening is at least as effective in terms of detecting
advanced cancers at the time of diagnosis (stage 2 and above) as standard screening. »

- One of two reasons :
- The MyPeBS sponsors are truly convinced of the value of organized screening in its current form.

In this case, it is understandable that they did not consider a group without first screening it. But why conduct a non-inferiority study ?

If the current screening is effective, we will not change it unless there is a compelling reason to do so. In other words, demonstrating non-inferiority is pointless because we will not abandon a screening that works well for another screening that has not been shown to be superior. - Or the MyPeBS sponsors aren't entirely convinced of the value of current screening.

The non-inferiority study makes sense in this case. But, why not include a third group of women who are not screened to see if the concerns about the effectiveness of current screening are justified ?

The brochure refers to « large-scale studies that have shown that screening reduces breast cancer deaths by about 20% » in the section « Advantages
and disadvantages of current breast cancer screening ». It fails to mention that these studies are old and likely out of date, and, more importantly, it fails to
mention recent studies that show no significant reduction in mortality from screening.

MyPeBS sponsors have the right to be convinced of the effectiveness of screening. However, the purpose of the information brochure is not to express opinions; rather,
it should provide all available information to allow patients to make an informed, non-pressured decision about whether to participate in the clinical trial.

The brochure is no more honest when it comes to overdiagnosis. It merely mentions 10% overdiagnosis, failing to mention that the frequency of overdiagnosis is in
fact poorly known and that some studies estimate this frequency at around 50%.

Furthermore, the brochure carefully avoids mentioning the overtreatment that results from overdiagnosis. Yet it is these unnecessary treatments, along with their
associated side effects, that pose the greatest risk of screening and give concrete meaning to the term "overdiagnosis".

The current scientific uncertainties make it difficult to present the benefits and risks of screening. However, **nothing should be hidden, and the wish for
clarity does not justify presenting only the most advantageous figures while leaving out the others**.

The patient information brochure can be downloaded by clicking here (pdf, 500Kb)

The promoters of MyPeBS appear to be motivated primarily by a desire to promote breast cancer screening. This is explicitly acknowledged by Dr. Balleyguer, the principal
investigator for France, who states in the MyPeBS press kit
(page 14): « MyPeBS will almost certainly help increase the proportion of women who participate in national screening. Today, barely more than one in every two women
participates. »

This is most likely what caused the serious flaws in this study's design and statistical analysis plan.

Because of these flaws, there is little to expect from MyPeBS.

MyPeBS will not provide an answer to the value of screening in the absence of a no-screening group.

Because the primary goal is non-inferiority at the -25% threshold, it is unclear whether MyPeBS will be able to distinguish between standard and individualized screening.
It will simply be known that, at worst, individualized screening is no more than a -25% inferior to standard screening.

So there is little to expect from MyPeBS and it is all the more regrettable that the study will cost European taxpayers a lot. The study is indeed funded by Europe in
the amount of **12 million euros** (yes, you read that correctly, 12 million euros) through the European program Horizon 2020 of funding for research and
innovation (source press kit, page 5).

Dernière mise à jour le 15/09/2021