Diana B. Petitti

Comment on Misguided Research Misinforming Public Policies

De Gruyter | Published online: November 5, 2015

Abstract

I’d like to highlight some of the points about research on mammography screening that Dr. Miettinen and I agree on. But I disagree with the main point he tries to make.

1 Introduction

I’d like to highlight some of the points about research on mammography screening that Dr. Miettinen and I agree on. But I disagree with the main point he tries to make.

2 Agree: tests, absent treatment, do not prevent mortality

Dr. Miettinen reminds us that tests do not prevent mortality due to cancer. It is the appropriate management of abnormalities found by tests that has the potential to prevent mortality due to cancer. Management includes follow-up to evaluate whether the abnormalities detected by the test are cancer (or pre-cancer) and the appropriate treatment of proven cancers and pre-cancers.

Mammography is a test – called a screening test when it is done in women without symptoms or signs of breast cancer – that detects lesions some of which are cancer and some of which are called early cancer (e.g., ductal carcinoma in situ – DCIS) but may not be cancer. Mammography also detects lesions that look like they could be cancer or DCIS but turn out not to be either. After a lesion that could be cancer or DCIS is found by a mammography test, management includes other tests done to try to prove that the lesion either is not a cancer or is a cancer (or DCIS). Treatment of the screen-detected lesions first discovered by mammography that turn out to be cancer or DCIS includes surgical excision of the cancer or the DCIS (or the breast containing DCIS) and/or chemotherapy and/or radiation.

Absent appropriate management of lesions first detected by mammography, there will be no reduction in mortality due to breast cancer in a woman who had a lesion detected by the mammogram and breast cancer mortality will not be reduced in the population of women who were the target of a screening program. Dr. Miettinen is correct that some programs aimed at reducing mortality due to breast cancer have focused on the screening component and ignored downstream management actions – follow-up and treatment – that are required to prevent mortality due to breast cancer.

3 Agree: there are problems of randomization in mammography clinical trials

Discussing randomized trials of mammography, Dr. Miettinen states that “the randomizations have repeatedly been disastrously flawed.” He cites statements published in a book by Siddartha Mukherjee (Mukherjee, 2012) in support of this assertion. I wish that Dr. Miettinen had elaborated on the evidence for his statement about randomization in trials of mammography. It is my belief that many systematic reviewers, meta-analysts, clinical epidemiologists, and policy-makers do not understand fully the types of randomization in mammography trials and the problems that some of these randomizations pose for assessment of the benefits of screening. All trials are considered as if women were individually randomized and assumed to have used the tamper-proof methods we have come to expect in randomized trials. Table 1 of Humphrey et al.’s 2002 systematic evidence review done for the United States Preventive Services Task Force (Humphrey et al., 2002) discusses the problem of randomization for the early trials comprehensively.

Dr. Miettinen correctly, in my opinion, singles out the Canadian National Breast Cancer Screening Trial as an example of flawed randomization but provides no detail. As has been previously described (Freedman et al., 2004; Kopans and Feig, 1993; Tarone, 1995), prior to randomization to screening mammography, women recruited to this trial first had a clinical breast examination done by a nurse who had been trained in the detection of breast abnormalities. Women were assigned to mammography or usual care after they had been examined by study nurses. Study coordinators, who were not blinded to the results of the physical examination, used paper lists with preprinted group designations (mammography or usual care) to make assignments. Of the 21 women who had breast cancer with four or more positive lymph nodes detected by the physical examination prior to randomization, women who had advanced cancer at entry to the trial, 17 were assigned to mammography and 4 to usual care. This difference was statistically significant (Freedman et al., 2004; Tarone, 1995).

That this trial is generally considered high “quality” bewilders me.

4 Agree: time affects the magnitude of any benefit of mammography screening and treatment

Dr. Miettinen points out that many meta-analyses of randomized trials of mammography describe the effect of mammography screening as the proportional reduction in breast cancer mortality derived from the ratio of breast cancer death rates measured with a person-time denominator (ie., a 20% reduction in the relative risk of breast cancer mortality in women assigned to mammography) and treat the ratio as though it is constant over the duration of follow-up. It is obvious that a mammogram and treatment of cancer detected by the mammogram does not prevent a woman from dying from breast cancer the day after her mammogram, or even the week, month, or year after. Dr. Miettenin’s criticism of descriptions of meta-analyses reporting on the effect of mammography screening using the percentage reduction in mortality without reference to time is an apt criticism. It applies to meta-analyses done by the Cochrane Collaboration (Gøtzsche and Jørgensen, 2013) and the United States Preventive Services Task Force (Nelson et al., 2009).

A more recent (2014) meta-analysis discusses the reduction in breast cancer mortality in the context of time (Lee et al., 2013) with important implications for advising women about whether to undergo screening. I hope that the practice of considering time when discussing screening benefit will become universal.

5 Disagree: the policy perspective requires use of the population effect of programs of screening and treatment

Dr. Miettinen criticizes policy-makers for judging the benefit of breast cancer screening programs based on analyses that assess the effect of the program on breast cancer mortality in the population of women who have been the target of the program. He champions adoption of a “clinical” perspective on screening and use of a “clinical” metric to assess the benefit of screening. He suggests that screening – not screening programs but screening – should be judged based on a comparison of the effect of screening (and treatment) on the case-fatality rate of the cancers that were detected early and treated with the case fatality rate of these cancers had they not been detected early (and thus could not be treated early). Of course, as Dr. Miettinen notes, the case-fatality rate in cancer not detected early is unobservable or is estimable only using a research study design that would be rejected as unethical.

The effect on breast cancer mortality for a program that involves mammography as a first step will depend on the success of the program in getting women to undergo mammography, the success of the program in follow-up of positive screening tests, and the success of the program in assuring of appropriate treatment of cancer and pre-cancerous lesions initially detected by mammography. If the clinical enterprise fails downstream in its delivery of follow-up and treatment, breast cancer mortality may not decrease, even if every woman in a population has a mammogram and all of the early cancers and pre-cancers are found. In my opinion, policy-makers, who are stewards of the use of societal resources, are correct when they question the value of a program aimed at reducing mortality due to breast cancer in a population if the program does not reduce mortality due to breast cancer in that population. A program that does not deliver on its stated aim is, whatever the reason, a program that has failed.

It is not that policy-makers are using an incorrect metric. They are using a metric that permits them to answer a question that is not Dr. Miettinen’s question.

References

Freedman, D. A., Petitti, D. B. and Robins, J. M. (2004). On the efficacy of screening for breast cancer. International Journal of Epidemiology, 33:43–55. Search in Google Scholar

Gøtzsche, P. C. and Jørgensen, K. J. (2013). Screening for breast cancer with mammography.. Cochrane Database of Systematic Reviews, 6:CD001877. doi:10.1002/14651858.CD001877.pub5. Search in Google Scholar

Humphrey, L., Chan, B. K. S., Detlefsen, S. and Helfand, M. 2002. “Screening for Breast Cancer. Systematic Evidence Reviews, No. 15.” Rockville (MD): Agency for Healthcare Research and Quality (US); 2002 August. Search in Google Scholar

Kopans, D. B. and Feig, S. A. (1993). The Canadian national breast screening study: A critical review. American Journal of Roentgenology, 161:755–760. Search in Google Scholar

Lee, S. J., John Boscardin, W., Stijacic-Cenzer, I., Conell-Price, J., O‘Brien, S. and Walter, L. C. (2013). Time lag To Benefit after screening for breast and colorectal cancer: Meta-analysis of survival data from the United States, Sweden, United Kingdom, and Denmark. British Medical Journal, 346:e8441. doi: 10.1136/bmj.e8441. Search in Google Scholar

Mukherjee, S. (2012). The Emperor of All Maladies. A Biography of Cancer. New York: Scribner. Search in Google Scholar

Nelson, H. D., Tyne, K., Naik, A., Bougatsos, C., Chan, B. K. and Humphrey, L. L. (2009). U.S. preventive services task force. Screening for breast cancer: An update for the U.S. preventive services task force. Annals of internal medicine, 151:727-37, W237-42. doi: 10.7326/0003-4819-151-10-200911170-00009. Search in Google Scholar

Tarone, R. E. (1995). The excess of patients with advanced breast cancer in young women screened with mammography in the Canadian national breast screening study. Cancer, 75:997–1003. Search in Google Scholar

Published Online: 2015-11-5
Published in Print: 2015-12-1

©2015 by De Gruyter