Are small breast cancers good because they are small or small because they are good?

The New England Journal of Medicine

Donald R. Lannin, M.D., and Shiyi Wang, M.D., Ph.D.

http://www.nejm.org/doi/full/10.1056/NEJMsr1613680

Summary by Cécile Bour, MD

Presentation, objectives and conclusions of the study

Under this rather provocative title, the authors of this study, Donald Lannin and Shiyi Wang of the Yale Cancer Center (New Haven, Connecticut) use a database from the U.S. Surveillance, Epidemiology, and End Results (SEER) to evaluate the screening.

The study is original in that for the first time, biological data are being used for an epidemiological evaluation, particularly for over-diagnosis.

The study links the biological factors of tumors, their size, and both life expectancy and latency of cancers in order to evaluate overdiagnosis, and to support Welch's thesis that the smallest breast cancers, as often detected by screening, have disproportionately favorable biological characteristics, a very long lead time, and would not compromised women's health or lives if undetected.

Welch quantified over-diagnosis at around 22%.

The authors reached this conclusion:

Many of the small tumors that are excessively detected by screening have a very good prognosis due to their intrinsically slow growth, which means that they are unlikely to become large tumors and are inherently favorable. They are the ones that constitute the over-diagnosis as a direct result of the screening activity. They will not grow enough to be dangerous.

Conversely, large tumors, which are responsible for death and most often have an unfavorable prognosis, are dangerous from the outset. Unfortunately, they escape mammographic detection because their growth kinetics are too rapid.

This theory, that originated from epidemiology a long time ago and which explains why screening does not improve the prognosis of women with breast cancer, is substantiated here.

Procedures, methods, data to be understood

The study concerns only invasive cancers. Lesions are divided into three prognosis groups based on the following biological factors: their grade, the presence of estrogen receptors and progesterone receptors (knowing that tumors with these hormone receptors have a better prognosis).

Twelve combinations are possible based on these three variables, each of these 12 groups having a distinct prognosis.

Four groups of cancers with poor prognosis :

- grade 2, negative receptors

- grade 3, negative receptors

- grade 3, estrogen receptor positive and progesterone receptor negative

- grade 3, estrogen receptor negative and progesterone receptor positive

Three groups of cancers with a good prognosis:

- grade 1, positive receptors

- grade 1, estrogen receptors positive and progesterone receptors negative

- grade 1, estrogen receptors negative and progesterone receptors positive

The other five groups are intermediate prognosis groups.

 The authors investigate the correlation of tumor size and biological features of tumors in relation to cancer-specific survival rates.

A "favorable" tumor is one where the biological features presume a good prognosis and an "unfavorable" tumor if otherwise.

The "mean lead time " that enables the quantification of over-diagnosis is the length of time between when a cancer can be detected by mammography and when it would have become clinically apparent without early screening by mammography (adjustment to the age being made).

The fraction of women with a life expectancy less than the lead time represented the percent of overdiagnosis.

(This means that these women will die from causes other than their breast cancer, which would have been unnecessarily detected because it did not endanger them).

Study results 

I-Biological characteristics of tumors as a function of their size:

The percentages of favorable, intermediate, or unfavorable tumors for each tumor size are examined for women over 40 years of age and for women under 40 years of age.

For women aged 40 and over :

For tumors 1cm and less: 38.2% are "favorable" tumors.

Among the lesions of 1cm and less, only 14.1% are "unfavorable" tumors. On the other hand, among tumors larger than 5 cm, 35.8% are "unfavorable" tumors.

Results for women under 40 years of age:

The  favorable tumors  were only about half as common and the unfavorable tumors were much more common, for each tumor size examined.

II-Study of specific survival as a function of both biological characteristics and tumor size.

Classification into small tumors for those between 0.1 and 2 cm = T1; and into large tumors for those between 2.1 and 5 cm = T2.

The diagram below shows that both tumor size and biological factors influence the prognosis, but that large tumors with a favorable biological features had a better prognosis than small tumors with an unfavorable biological features.

This means that the difference in survival is less dependent on size than on biological factors, while larger size will be more critical when the biological factors are already unfavorable.

III-Evaluation of over-diagnosis according to the lead time

The approach finds a close link between the over-diagnosis rate and the lead time, by identifying the average lead time most in line with a given frequency of over-diagnosis.

It is important to remember that the lead time is the period of time between the time when cancer could be detected if a mammogram was performed and the time when clinical signs appear if a mammogram is not performed.

When life expectancy is known (estimated according to several factors including age), and when one of the two data, overdiagnosis rate or lead time, is known, it is possible to estimate the one of these two data that is unknown, and this for each of the 12 biological groups listed above.

The lead time, according to all models, is longer for tumors with favorable than unfavorable factors.

Thus, the percentage of overdiagnosis could be estimated at 53% for favorable tumors, 44% for intermediate tumors, 3% for unfavorable ones; in fact there is no evidence that unfavorable tumors do not progress, the small overdiagnosis observed in the unfavorable group is due to deaths occurring in these patients from unrelated causes before the lead time, which is short for these forms.

We find the figure proposed by Welch of 22% over-diagnosis overall, again concerning only invasive forms (not the cancers in situ).

IV-Tumor size may be an indirect indicator of good or poor biological features

If all tumors were progressing we would expect a steady state to be reached in which there would be a similar distribution of tumor biological features across size categories.

There would be favorable and unfavorable forms in similar rates in each size category.

Instead, these data provide fairly direct evidence that many small tumors with favorable biological features do not progress to large tumors over the lifetime of the patient.

Furthermore, the data imply that large tumors do not arise equally from all small tumors but preferentially develop from a distinct subpopulation of small tumors with unfavorable biological features.

The figure below shows the distribution of tumors of the three different prognostic values according to tumor size; on the x-axis the number of patients, on the y-axis the tumor size. The colors correspond to a prognostic value.

The upper part corresponds to women under 40 years of age and the lower part for women over 40 years of age.

There are generally more tumors with good prognosis in the small tumor size category and vice-versa.

Click to enlarge

Reminder of the conclusions, in detail

- Both data: tumor size and biological features impact the prognosis, but a large tumor with favorable biological features is likely to have a better prognosis than a small lesion with unfavorable features. Furthermore, tumor size is more decisive for the prognosis in the category of tumors with unfavorable biological features.

- Many small tumors with favorable biological factors do not progress to a large size during the patient's lifetime. Large tumors do not arise from the small ones, but from a subpopulation of small lesions with pejorative biological factors from the outset. The higher incidence of favorable tumors in women aged 40 and over suggests that these tumors are precisely and preferentially detected by systematic mammography, and that this explains the obviously very favorable survival rates for this detected age group.

- The authors' estimate of over-diagnosis by all biological groups combined is similar to Welch's estimate of 22%, and indicates that the lead time for the most favorable cancers is 19 years (between 10 and 20 years), compared to 2 years or less for unfavorable cancers.

- Lead time: due to a very long lead time in the favorable tumor group, mammography is necessarily very effective for these tumors with a very good prognosis, long time latent, which find themselves over-represented among small tumors. A considerable number of these tumors would never manifest themselves during a woman's lifetime. And those of these less aggressive tumors that are likely to grow, keep this excellent prognosis. Thus their detection by mammography is of little interest but leads to a perception of efficiency and exaggerated survival.

In the most favorable group, which certainly contains the largest proportion of over-diagnosed cancers (grade 1, positive receptors and size < 2cm), the 10-year survival rate is 97%. The proportion of over-diagnosis can be estimated to be at least 50%. (In fact, the smaller the tumor in this group, the greater the likelihood of over-diagnosis. It is therefore not surprising to see these very good survival figures highlighted in  communication on screening, Editor’s note).

- Still in this very favorable group, over-diagnosis predominates in older women compared to younger women. Indeed, because of this very long lead time of 15 to 20 years, many of these cancers could have been diagnosed only around the age of 70, but are now detected around 50 to 60 years because of screening. And a large proportion of these non-aggressive and indolent lesions detected by screening in these 70-year-olds may never have been detected during a person's lifetime in the absence of mammography.

- For tumors with unfavorable biological features, the prognosis is considerably better if the detection is early (less than 2cm), unfortunately this is very rarely the case due to their short lead time, so they are often diagnosed late and few of these unfavorable tumors are found in the small tumor group because of their rapid growth.

 Early detection is not a universal benefit

In sum

====>>>> Low-grade  and high-grade malignancies tumors result from different molecular mechanisms; a low-grade tumor almost never dédifferentiates into a high-grade tumor. It seems to be arguable that the biological features of a tumor represent a constant natural factor.

=====>>>> The biggest problem resulting from the over-diagnosis of small, favorable lesions, over-represented in tumors detected by screening, is the over-treatment and anxiety that these unnecessary diagnoses cause.

There is a need to educate physicians, patients, and the general public that some cancers are indolent.

Editor's note: The partial indicators that are the five-year survival and the increase in the proportion of cancers detected at an "early" stage are not relevant criteria for judging the effectiveness of existing diagnostic and therapeutic practices. Once again, only mortality is a relevant indicator.

N.B. :

Sparano, J.A, Gray R et al. Prospective Validation of a 21-Gene Expression Assay in Breast Cancer. The New England Journal of Medicine. 2015;373(21):2005-14. https://www.nejm.org/doi/full/10.1056/NEJMoa1510764

The majority of cancers are diagnosed at an early stage.

However, diagnosing increasingly smaller cancers was not accompanied by any decrease in breast cancer mortality until the 1990s. The prognosis of a cancer varies more according to its molecular characteristics than to its size, whether it is larger or smaller than 2 cm.

Extract from the study :

"In multivariate analysis including age ...., tumor size (2.1 to 5.0 cm vs. ≤ 2 cm in the largest dimension), histological grade (high vs. intermediate vs. low) and type of operation ... only histological grade showed a significant correlation with rate of recurrence."

Cancer Rose est un collectif de professionnels de la santé, rassemblés en association. Cancer Rose fonctionne sans publicité, sans conflit d’intérêt, sans subvention. Merci de soutenir notre action sur HelloAsso.


Cancer Rose is a French non-profit organization of health care professionals. Cancer Rose performs its activity without advertising, conflict of interest, subsidies. Thank you to support our activity on HelloAsso.

30,000 cancers overdiagnosed in an Australian study

30,000 cancers overdiagnosed each year in an Australian study : a public health issue

Summary , Cécile Bour, MD
28 January 2020
https://www.mja.com.au/journal/2020/212/4/estimating-magnitude-cancer-overdiagnosis-australia
Authors of australian study published in Medical Journal of Australia
Paul P Glasziou, Mark A Jones, Thanya Pathirana, Alexandra L Barratt and Katy JL Bell
Med J Aust || doi: 10.5694/mja2.50455

The obsession of our modern societies to track diseases through ever earlier detection is leading to a public health crisis, according to Australian authors who published in January 2020 their evaluation of over-diagnoses for five different cancers.


Method

National data routinely collected by the Australian Institute of Health and Welfare (Australia's national agency for health and welfare information and statistics) were analyzed to estimate risks in population to have a cancer during lifetime by comparing the current period (2012 data) with the past (1982 data).
This allowed to measure changes in these risks over the last 30 years. An adjustment has been made to take into account the risk of death and changes in risk factors over time.
The five different studied cancers are: prostate, breast, renal, thyroid cancers, and melanoma.

The authors propose an estimate of the proportion of cancer diagnoses in Australia which are reasonably due to over-diagnosis.

Results


For women, an estimated 22% of breast cancers (invasive cancers, 13%), 58% of renal cancers, 73% of thyroid cancers, and 54% of melanomas (invasive melanoma, 15%) were overdiagnosed.

For men an estimated 42% of prostate cancers, 42% of renal cancers, 73% of thyroid cancers, and 58% of melanomas (invasive melanomas, 22%) were overdiagnosed.
Despite the relative uncertainty conceded by the authors themselves about these estimates, this result would be equivalent to an overdiagnosis representing about 18% of women's cancer diagnoses and about 24% of men's cancer diagnoses in Australia in 2012, i.e. about 11,000 women's cancers and 18,000 men's cancers overdiagnosed in Australia each year.

Absolute rates of over-diagnosis were highest for breast cancer and prostate cancer due to their higher baseline prevalence (rate of new cases + rate of cases already present).

Causes

The population is certainly aging, but screening programs are the big providers of overdiagnoses.
Cancer can also be overdiagnosed outside screening programs; for example overdiagnosis of thyroid cancer is attributable to incidental detection during imaging investigations of unrelated problems or the overdiagnosis of abdominal scans carried out for other causes.

Problems

Overdiagnosis is important to know and control because of the adverse effects on iatrogeny (pathology induced by treatments) and the associated additional costs.

Harms include the psychosocial impact of unnecessary cancer diagnoses, such as the increased suicide risk for men after being diagnosed with prostate cancer.
Cancer treatments such as radiotherapy, endocrine therapy and chemotherapy can cause physical harm, but the risks are considered acceptable if diagnosis is appropriate. Contrary, when someone is unnecessarily diagnosed with a cancer which would not be harmful for its health or life (overdiagnosis definition), this person will suffer harm as a result of the treatment, instead of enjoying the benefit of having been detected.

In other countries

This study is the first study to estimate overall cancer overdiagnosis on a national level.
A recent British analysis found that the “incidence of 10 of the 20 most common cancers in the UK has increased by more than 50% in both sexes since the 1980s.” These cancers included breast, kidney, prostate, thyroid cancers and melanoma, but also non‐Hodgkin lymphoma, oral, cervical, liver, and uterine cancers.
The authors therefore estimated overdiagnosis only for cancers with the typical epidemiologic signature of overdiagnosis: breast, prostate, kidney, thyroid cancers and melanoma (for which lifetime mortality has changed little in absolute terms).
UK cancer statistics released in January 2019 show very high survival rates for people with early stage cancers of these types, providing further evidence of probable overdiagnosis: 5‐year survival of 99% for stage 1 breast cancer, 100% for stage 1 prostate cancer, 100% for stage 1 melanoma, 89% for stage 1 kidney cancer, and 88% for thyroid cancer of any stage.

Editor’s note: Survival, often emphasized by INCa to justify the screening for breast cancer, is not an indicator of screening effectiveness, but a good over-diagnosis marker.
Survival measures the length of life with a cancer, If the cancer is not meant to kill its carrier, as is the case for cancers detected predominantly by screening, which are at low stage, survival can indeed be important since these cancers detected would never have led to the death.

The more over-diagnosis emerges, the more cancers that could have been ignored, and the better the survival rates, automatically.
The only appropriate indicator of the effectiveness of screening is mortality, or more precisely total mortality.

Author’s conclusion

Cancer overdiagnosis has important implications for health policy.
First, rates of avoidable overdiagnosis need to be reduced to the lowest level compatible with targeted screening and appropriate investigation, instead of mass screening.

Strategies to reduce overtreatment of low-stage, therefore low-risk prostate, breast and thyroid cancers should be addressed.

A second, and perhaps more important implication is that health services need to be alert to new areas of overdiagnosis and to detect them early.

Editor’s note: Australia has already initiated an over-diagnosis action plan : https:/cancer-rose.fr/2018/10/15/un-plan-daction-national-contre-le-surdiagnostic-en-australie/

Editor’s note: The Australian example should prompt us to look downwards, in particular to the pink October commercial campaigns that culpably promote and train crowds to practice mass screening, encouraging women to be screened even outside the screening age range.

Cancer Rose est un collectif de professionnels de la santé, rassemblés en association. Cancer Rose fonctionne sans publicité, sans conflit d’intérêt, sans subvention. Merci de soutenir notre action sur HelloAsso.


Cancer Rose is a French non-profit organization of health care professionals. Cancer Rose performs its activity without advertising, conflict of interest, subsidies. Thank you to support our activity on HelloAsso.

Absence of benefit from mammograms in women aged 40-50 years confirmed by final results of UK Age Trial

Cancer Rose, 16 August 2020


In August 2020, the final results of the UK Age Trial [I] were published.

For women aged 40-49 years


Between 1990 and 1997, this British trial included approximately 161,000 women aged 39 to 41.Women were randomly selected and an annual mammography up to the age of 48 was proposed for about one in three (53,883 women), while the remainder had no screening. All the women then joined the standard British screening program, which includes a mammogram between the ages of 50 and 69 every three years. The main aim of the trial was to determine whether screening could reduce mortality from breast cancer before the first mammogram of standard screening program, which starts after the age of 50.

Prior to these results, no evidence of a benefit


After 10 years, the UK Age Trial results showed a statistically significant decrease in the number of breast cancer deaths before routine screening program. The authors announced a 25% decrease in relative value, but this actually corresponds to a gain of only 4 deaths from breast cancer per 10,000 women screened and followed for 10 years. Furthermore, the results did not demonstrate a decrease in total mortality (or all-cause mortality).*
There was no statistically significant decrease in deaths from cancer when the results of all nine studies that included women aged 40-49 (not just the UK Age Trial) were considered.

*Only the total mortality includes all elements of patient management, hence also the effects of treatment, overdiagnosis and overtreatment.
This makes more sense because any cancer detected will be treated, the treatments themselves sometimes causing deaths, which will be counted in the «all cause mortality» , thereby better reflecting the screening reality.

After these results, even less evidence of a benefit

After 23 years, the UK Age Trial results no longer indicate a significant decrease in the number of breast cancer deaths in women screened between the ages of 40 and 49.
The authors of the trial write: «Overall, there was no significant reduction in breast cancer mortality in the intervention group compared with the control group» (p4 penultimate paragraph (ref1 PDF)).

Prior to these final results, analyses that considered the results of all trials already concluded that there was no measurable benefit. Or, at the time, the UK Age was one of those trials that supported screening program.
With these results, we can be even more affirmative in saying that attempting to screen for breast cancer before the age of 50 does not have any tangible benefit.

When should we expect a re-assessment of these results?

Another trial is underway to evaluate the possible benefit of expanding screening to women before age 50 and after age 69: the Age X Trial. Its results are not expected before 2026.

Controversy in Great Britain

The publication of these results caused a great deal of controversy in Great Britain. Not because they were challenged or questioned, because the trial is of a good methodological level.
But because the authors [iii], no doubt disappointed with the results, tried to hide their negative character by insisting on the results obtained at the end of a 10-year follow-up and not at the end of the 23-year follow-up.
Some popular media journalists were thus prompted to write that the results were in favor of the effectiveness of the screening, hence the controversy [iv].

Read more about:

The international reactions to attempts to cover up screening failure in a publication

REFERENCES

[i] Duffy SW et coll. "Effects of mammogrpahic screening from age 40 years on breast cancer mortality (UK Age trial) : final results of a randomised, controlled trial" The Lancet Oncology online. 12 août 2020. Website www.thelancet.com/oncology. Doi : 10.1016/S1470-2045(20)30398-3
Document PDF of the study


[ii] Nelson HD et coll. "Screening for Breast Cancer: A Systematic Review to Update the 2009 U.S. Preventive Services Task Force" Recommendation. Evidence Synthesis No. 124. AHRQ Publication No. 14-05201-EF-1. Rockville, MD: Agency for Healthcare Research and Quality; 2016.
See particularly table 28 page 128.


[iii] Who are the authors of this study?
They are Duffy’s team and his staff from Queen Mary University in London.
Dr.Duffy is already well known in the world of screening, as he is one of the oldest pioneers in the promotion of screening and has published several studies seeking to quantify overdiagnosis, most often at its lowest range, according to him from 1 to 10%. (Overdiagnosis in mammographic screening for breast cancer in Europe: a literature review. Puliti D, Duffy SW, Miccinesi G, by Koning H, Lynge E, Zappa M and the EUROSCREEN Working Group. J Med Screen 2012;19 Suppl1:42-56.)

This was a review of studies, and this work had been highly controversial as a source of multiple rather crude biases. In their analysis the authors, Duffy and Puliti, had deliberately excluded many reference studies, Zahl’s in 2008 and Junod’s in 2011( Junod B, Zahl P-H, Kaplan Rm, Olsen J, Greenland S. An investigation of the apparent breast cancer epidemic in France: screening and incidence trends in birth cohorts. BmC Cancer. 2011 Sep 21;11(1):401.)

The Prescrire Review in 2006, after a careful analysis, as well as the exhaustive analysis made by Professor Autier, and many others even more recent, currently conclude that the rate of overdiagnosis may be between 30 and 50%.
Revue Prescrire :
*Dépistage des cancers du sein par mammographie Deuxième partie Comparaisons non randomisées : résultats voisins de ceux des essais randomisés. Rev Prescrire. 2014 Nov;34(373):842–6.
*Dépistage des cancers du sein par mammographie Première partie Essais randomisés : diminution de la mortalité par cancer du sein d’ampleur incertaine, au mieux modeste. Rev Prescrire. 2014 Nov;34(373):837–41.
*Dépistage des cancers du sein par mammographies Troisième partie Diagnostics par excès : e et indésirable insidieux du dépistage. Rev Prescrire. 35(376):111–8.


[iv]
How screening promoters try to justify the alleged success:
In the screening group, after 10 years, for every 10,000 women in the screening group, there were 14 deaths from breast cancer, while for every 10,000 women in the control group, there were 20 deaths.This relatively small benefit (including adverse effects of screening) was statistically significant. It is this difference that the authors rely on to argue that there could be a benefit, which they proclaim as “25%” (20-14 / 20 = 25%).

However, at the end of the follow-up, after 23 years, for 10,000 women in the screening group, there were 39 deaths from breast cancer, while for 10,000 women in the control group, there were 44. This time the difference is not statistically significant. And in relative terms it is 44 – 39 / 44 = 11%.
Above all, the important thing to remember: no significant difference in all-cause mortality was found between the two groups, neither after 10 years, nor at the end of the follow-up, after 23 years
At the end of the trial, there were 650 deaths per 10,000 women in the screening group and 648 deaths per 10,000 women in the control group.

Cancer Rose est un collectif de professionnels de la santé, rassemblés en association. Cancer Rose fonctionne sans publicité, sans conflit d’intérêt, sans subvention. Merci de soutenir notre action sur HelloAsso.


Cancer Rose is a French non-profit organization of health care professionals. Cancer Rose performs its activity without advertising, conflict of interest, subsidies. Thank you to support our activity on HelloAsso.