January 29, 2021 — The COVID-19 pandemic has demonstrated the tragic consequences that result from ignoring science, evidence, and the analysis and advice of experts while being guided by inexpert advice.
The history of the U.S. Preventive Services Task Force (USPSTF) on breast screening is another case in point, in which decisions on mammography guidelines have been based on flawed analyses of scientific data.
When the USPSTF last week opened a new review of breast screening guidelines, it created the opportunity to revisit the group’s flawed decision in 2009 to not recommend breast screening for women in their 40s. However, a number of factors need to be addressed for the 2021 review to present a fair and impartial review of the science and evidence in favor of breast screening.
A brief history of USPSTF and breast screening
In 2005, the USPSTF supported annual screening for all women ages 40 and over.1 Then in 2009, the USPSTF dropped support for screening women ages 40-49, instead advising them to consult with their physician whether to be screened. It urged women ages 50-74 to be screened every two years.2
In 2016, the USPSTF advised that women ages 50-74 be screened every two years,3 despite the fact that the task force admitted that the most lives are saved by screening beginning at the age of 40 — “the USPSTF found adequate evidence that mammography screening reduces breast cancer mortality in women aged 40 to 74 years.”4
These decisions came in spite of facts demonstrating the proven effectiveness of breast screening. Randomized controlled trials (RCTs) of breast cancer screening proved that screening and early detection of breast cancer reduces deaths for women ages 40-74 (the ages of the women who participated in the trials).5 Confusion had been created in 1993 by the inappropriate use of subgroup analysis6 to claim no benefit for women ages 40-49, but this was later refuted with longer follow-up7 (the 1993 claims had ignored the fact that an immediate benefit is not expected from a screening program).
There is clear benefit for breast screening in women ages 40-74. Although it has been suggested that these randomized controlled trials are old, they provide the fundamental proof that early detection reduces deaths.
Because of “noncompliance” and “contamination,” the RCTs have underestimated the benefit of early detection of breast cancer. Therefore, the results of these trials should be viewed as the lower level of the likely benefit of mammography.
The USPSTF should be aware that the Edinburgh Trial is no longer cited with the RCTs because of an apparent imbalance in the socioeconomic factors of participants.
Another RCT, the Canadian National Breast Screening Studies (CNBSS), should also have been dropped from guidelines analyses years ago. Not only are their results major outliers among the RCTs, but numerous critical analyses over the years have challenged their validity.
The CNBSS trials were compromised by poor-quality mammography,8,9 and their data compromised by the fact that they violated the fundamental requirements of RCTs by having a nonblinded allocation process.10,11 This resulted in a statistically significant excess of women with advanced cancers being assigned to the screening arm of CNBSS1.12,13
It has been claimed that the CNBSS trials showed a major, 22% rate of “overdiagnosis,” when, in fact, data from the trial show that there was only a 4% difference in cancers diagnosed between the two study arms.14 The CNBSS results are compromised and unreliable and should not factor into the USPSTF review.15
Numerous observational studies have validated the benefit of screening women starting at the age of 40 in the general population with reductions in deaths of 40% or more.16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33
In a review of the incidence of death among more than 500,000 women in Sweden, the researchers found some benefit from improvements in therapy, but those who participated in mammography screening had a 41% reduction in their risk of dying of breast cancer within 10 years compared with those who had not participated in screening.34
In addition, there are no data that show that any of the parameters of screening change abruptly at age 50 or any other age.35 The RCTs proved mortality reduction for women ages 40-74. The threshold for initiating screening at the age of 50 is completely arbitrary, with NO scientific support. Grouping of data and averaging has falsely suggested a legitimate threshold when the data show that none exists. The only reason to use the age of 50 as a threshold is based on individual and scientifically unsupportable biases by analysts.
In fact, all major groups — including the USPSTF — agree that the most lives are saved by annual screening starting at the age of 40.
What’s more, studies have shown that there are more years of life lost to breast cancer among women ages 40-49 than for women ages 50-59.36
The risk of developing cancer due to radiation delivered to the breast from mammography (there is little exposure to any other susceptible organs) drops rapidly with increasing age, such that by the age of 40 the risk is unmeasurable and may be nonexistent. Even the extrapolated risk is below the smallest amount of benefit from screening.37,38
Statistical models from the U.S. National Cancer Institute and Cancer Intervention and Surveillance Modeling Network all predict that the most lives are saved by annual screening starting at the age of 40.39
Despite specious arguments to the contrary, breast screening has been shown to reduce the rate of advanced cancers (see references)40,41,42,43,44,45,46,47,48,49,50,51 that has been used as a surrogate for death since these are incurable cancers.
In the Harvard Hospitals, 71% of deaths from breast cancer were among the 20% of women who were not participating in screening despite having access to modern therapies.51 Spencer et al had similar results.52
Mammography and all-cause mortality
Critics of mammography claim that the RCTs of breast screening did not show a reduction in “all-cause mortality” — but this is a specious argument.53
All-cause mortality is an appropriate metric in treatment trials where everyone has breast cancer and most of the deaths will be due to breast cancer. You want to be certain that the treatment is not causing an unforeseen risk. In radiation therapy trials, for example, analysis of all-cause mortality revealed the unexpected risk that radiation therapy damaged the coronary arteries.
In screening trials, however, most deaths will be due to other causes, since breast cancer only accounts for 3% of deaths each year from “all causes.” If you reduce breast cancer deaths by 30%, then this will reduce “all-cause mortality” by 1%. It would take a 2.5 million-woman trial to prove that a major decrease in breast cancer deaths significantly reduces all-cause mortality.
It would be more appropriate to look at all-cause mortality among women with breast cancer in the RCTs, and this does show that screening reduces the rate of all-cause mortality.54
The CISNET models show that as many as 100,000 women now in their 30s will die by waiting until the age of 50 and being screened every two years whose lives could be saved by annual screening starting at 40.55 Among just the women who are 40 today, if they wait until 50 to be screened every two years as many as 13,770 will die whose lives could be saved by annual screening beginning at the age of 40.56
Overdiagnosis
What about overdiagnosis — the claim that breast screening discovers cancers that would never present a health risk to a woman during her lifetime? This is another criticism that is frequently levied against breast screening.
The claim of massive overdiagnosis has been manufactured by “guessing” that the incidence of breast cancer was not steadily increasing as screening was being introduced. Since no one has ever seen a mammographically detected invasive breast cancer disappear on its own (the few “miracles” have all been clinically evident), and Arleo et al showed that none of almost 250 invasive cancers that were untreated did not regress or disappear,57 then waiting until age 50 and screening every two years will not reduce “overdiagnosis,” if it even exists, because the cancers will still be there.
Delaying screening will reduce recalls from screening (inappropriately called “false positives”) for a few extra pictures or an ultrasound. The recall rate is approximately 10% (which is, approximately, the same recall rate as cervical cancer screening using Pap tests) and a very small chance of having an imaging-guided needle biopsy using local anesthesia with a fairly high yield of cancer. Approximately 2%-4% of women screened will be advised to have an imaging-guided needle biopsy, and 20%-40% of these lesions will prove to be malignant.
There is no question that recalls make all of us anxious, and recalls from screening are no exception, but, for most, the anxiety is short-lived.58 Given that the major “harm” (“harm” is pejorative, it should be called “risk”) from screening is the anxiety of being recalled, it is beyond paternal/maternalistic to advise women that it is preferable to let them die an avoidable death than to be made anxious by a recall.
Conflict of interest
The makeup of the USPSTF panel almost ensures that breast screening will not receive a fair review, and changes are needed.
In a supposed effort to avoid biases from panel members who have a conflict of interest (COI), the USPSTF has prevented anyone with actual expertise in breast cancer screening from serving on its panel. Consequently, panel members have been unable to critically sort through the available data and understand the validity or lack of validity of the material they have been asked to review.
This has also given “advisers” to the panel extraordinary influence to “guide” an experientially and factually naive panel. In the past, the vast majority of advisers to the USPSTF panel reviewing breast cancer screening have been individuals who have expressed their opposition to screening and have clearly had great influence on the panel. Many of the advisers have been viewed as having no conflicts of interest when, in fact, this is not true.
Without an obvious COI it is impossible to gain expertise in a field. Experts, such as I, who earn a living related to breast cancer screening, have an obvious and open conflict of interest. However, advisers who have received and continue to receive grant support for their research efforts are viewed as free of conflict, when in fact, theirs is a far less obvious conflict of interest.
Granting agencies, including the U.S. National Cancer Institute (NCI) and foundations have undeclared biases. When the work of a grantee supports the biases of the grantor, grants are likely to be renewed. Grant or foundation support is a far more insidious conflict of interest than those that are out in the open.
The practice of excluding experts from the USPSTF panel should stop. Conflicts of interest should be detailed, but rather than being excluded, experts are critical for an accurate analysis of the data to provide the most factual and evidence-based advice. Guidelines panels, including the USPSTF, should have leading experts involved in their decisions and the public should be provided with “minority” reports should there be unresolvable disagreements.
Problems with the USPSTF 2021 review
There are numerous potential problems with the USPSTF’s current review of breast screening. The task force is not going to include the NCI/CISNET computer models to project the potential outcomes of various screening protocols. Without these, the task force will be guessing in their predictions. It would appear that CISNET modeling has been dropped because the models all showed that the most lives are saved by annual screening beginning at the age of 40.
Although a reduction in advanced stage disease is a potentially useful “surrogate endpoint,” it is critical to remember that lives are lost among women diagnosed at all stages of breast cancer. It has been shown that reducing the size of cancers within stages is also a major benefit from screening that reduces deaths.
A critical fact that has been, repeatedly, overlooked by analyses that denigrate the value of screening and (falsely) suggest massive overdiagnosis, is the false claim that the background incidence of breast cancer has not increased over time. This has been the fundamental piece of misinformation that has been used to promote the false concepts of massive overdiagnosis and the false claim that there has not been a reduction in advanced cancers.
The data clearly show that the baseline incidence of breast cancer has increased steadily by 1%-1.3% per year dating back to at least 1940,59 long before there was any screening, which did not start until the mid-1980s. If the correct increasing baseline is used, not only is there no apparent “overdiagnosis” of invasive cancers, but it appears that there has been a major reduction in the incidence of invasive cancers.60 Although unproven, this is likely due to the removal of ductal carcinoma in situ (lesions almost exclusively detected by mammography) precluding the future development of invasive cancers.
By using the correct baseline incidence and extrapolation it is also clear that there has been a major reduction in the rate of advanced cancers.