Sunday, April 14, 2013

Three Decades of SEER Data Confirms That Mammogram Screening Does More Harm Than Good

Nearly one-third or 1.3 million women over the past 30 years were overdiagnosed with breast cancer (ie, their tumors would have never led to clinical symptoms in their lifetimes) in the United States,  according to the research published in the November 22, 2012 issue of the New England Journal of Medicine. 

Two highly respected oncologists/epidemiologists Archie Bleyer of St. Charles Health System, Bend, Ore. and H. Gilbert Welch of the Dartmouth Institute of Health Policy and Clinical Practice, Hanover, New Hampshire analyzed three decades of data collected by Surveillance, Epidemiology, and End Results (also known as SEER)  registry of the National Cancer Institute (NCI) to come to this damning conclusion about mammogram screening. SEER registry is the premier source of cancer incidence, prevalence and survival data and covers 28% of the US population, and the earliest recorded data in the registry dates back to 1973.

Comparing incidence of breast cancer from 1976-1978 (years before mammography screening) to 2006-2008 in women 40 years and older, Bleyer and Welch, concluded that the detection of early-stage breast cancer increased from 112 to 234 cases per 100,000 women or 122 additional cases per 100,000 women. 

However, mammogram screening fails the litmus test: how many late-stage cancers or deaths were avoided because of early screening? Apparently not much -- the decrease in late-stage cancer was negligible (from 102 to 94 cases per 100,000 women or 8 fewer cases!)

The experimental method and the evidence

To stratify SEER data, the authors divided the 4 stages of breast cancer into early-cancer and late-cancer as follows: 

  • Early cancer included the in situ disease (analysis was restricted to ductal carcinoma in situ [DCIS]) and localized disease (confined to breast) -- they excluded lobular in situ disease
  • Late cancer included regional disease (ie, cancer spread to draining lymph nodes) and distant (or metastatic) disease

The authors then made the following adjustments to how the data was analyzed:

  • While the earliest data in SEER registry dates back to 1973, the first year data should be considered spurious (there is a learning curve)
  • The years 1974-1975 overestimates the number of diagnoses due to the awareness as a result of the First Lady Betty Ford's breast cancer diagnosis at that time. Therefore, Bleyer and Welch chose 1976-1978 years for baseline comparison
  • Then there was an uptick in beast cancer diagnosis due to the widespread use of hormone replacement therapy during 1990-2005. . . the authors discounted those increases in their calculations
  • Taking all these adjustments into account, the comparison was made between 1976-1978 (before mammography availability) and 2006-2008 (after mammogram screening)

With the introduction of mammography screening in the 1980s, the incidence (=diagnosis) of early-stage cancer in women 40 and older picked up as shown above inn Figure A. This is the expected outcome of screening. This increase was mostly attributed to DCIS which would not have been detected without mammography. (A peak in incidence ~1990 in the Figure is due to HRT use.) 

The other expected outcome of effective screening should have been a decrease in the late-stage cancer incidence -- and, yes, there was a small decrease, though not dramatic (look at the red curve in Figure A). This decrease was just 8 cases per 100,000 women (a decrease from 102 to 94 cases per 100,000 women. This is a key evidence that mammogram screening had not been useful when it comes to decreasing death.

A great number of women (about 1.3 million over the last 30 years) who were given the diagnosis of early-stage cancer probably went through needless surgeries, chemo, radiation, psychological and family stress. This is where the "harm" of overdiagnosis occurred.

The bottom figure (B) is early and late-stage cancer incidence in women younger than 40 years. These women are generally not advised to undergo routine mammogram screening. Here the incidence just follows an average of 0.25% increase per year. . . nowhere near the overdiagnosis seen in the 40 and older women group.

Widespread adoption of mammogram screening has not decreased the incidence of death from metastatic disease

The late-stage disease has registered a small decrease, but all of this is attributed to better drugs and treatment protocols. In fact, today only 1 in 4 women with metastatic disease in the US survive beyond 5 years [SEER website quoted as Ref 15 in Bleyer, Welch NEJM report]. And, this statistics has not budged in spite of the introduction of mammogram screening.

Second, women younger than 40 (who rarely participate inn mammogram screening) have seen overall decrease in death from breast cancer by 42% compared to just 28% percent in 40 and older. Thus, mammogram screening can't be responsible in any major way in preventing deaths -- the credit goes to better drugs and surgical methods.

Other countries with mammogram screening programs have also seen the balance shift to "harm" of overdiagnosis

In the letters to editors, several investigators concurred with the NEJM conclusions and added the following countries with similar experiences: the United Kingdom, the Netherlands, Italy, Switzerland, Norway, Australia and Denmark.

But, the resistance continues. . . 

One letter to this NEJM report was from Debra Monticciolo, MD, and Barbara Monsees, MD, both representing the mammogram imaging interests (Society of Breast Imaging, Reston, VA, and American College of Radiology Commission on Breast Imaging, Reston, VA).

Drs Monticciolo and Monsees quoting 8 older clinical trials called the Bleyer and Welch's analysis flawed and misleading. Bleyer and Welch in response, were disappointed that the experts/spokespersons from the imaging community relied on older trials' data, by now likely irrelevant,  and called their analysis based on 30 years of cancer surveillance registry (the best in the world) as misleading.  Bleyer and Welch said, "We would argue that it is hard to get more 'real' than three decades of data from the world's preeminent cancer surveillance program. . . And yet the authors of this letter [Drs Monticciolo and Monsees] characterize our research as 'dangerous' ."

So, unfortunately, the debate continues. . . For example, read here and here. Meanwhile, the advice that Bleyer and Welch gave in the closing lines holds true for all women: "Women should recognize that our study does not answer the question 'Should I be screened for breast cancer?' However, they can rest assured that the question has more than one right answer."

  • Bleyer A, & Welch HG (2012). Effect of three decades of screening mammography on breast-cancer incidence. The New England journal of medicine, 367 (21), 1998-2005 PMID: 23171096 | Find Free PDF via Google Scholar |
  • Simoncini, Tommaso (2012). Faculty of 1000 evaluation for Effect of three decades of screening mammography on breast-cancer incidence (04 Dec 2012) F1000 Women's Health DOI: 10.3410/f.717964951.793466046

No comments:

Post a Comment