by Constantine Kaniklidis *, Research Director, No Surrender Breast Cancer Foundation (NSBCF) **
In a previous editorial on the mammography debate, I wrote: “that it is complex; that it is naïvely implausible to expect any decisive final resolution to the residual issues that will be convincing to the principle contending parties; and that behind it all, the devil is in the methodology”1. Almost two years later, the words stand true, but we have advanced significantly in illuminating the many dark cobwebbed corners of the debate. Center in that web is overdiagnosis (overdetection) with sister ills of false-positives and overtreatments, the harms that trouble – not illegitimately – so many critics of screening mammography, with heated disagreement about both the degree of overdiagnosis, and the validity of different estimation methodologies.
The seemingly simple question “How many screen-detected tumors would never surface in the absence of screening?” is the very devil to answer, as overdiagnosed tumors are themselves not simple at all, showing three distinct classes: (1) progressive tumors that may have presented clinically but for limited life expectancy (from advanced age and/or comorbidity); (2) nonprogressive tumors that would have remained indolent, and (3) nonprogressive tumors that would have regressed spontaneously, the legitimacy of this last class being highly questionable, collective experience of radiologists and clinicians not attesting to even one case of spontaneous tumor regression in any untreated patient2, and with the assumption of such “self-evaporative” tumors effectively challenged3. With all these, the patient, were it not for screening, would have died of some other non-cancer cause without ever having been diagnosed with malignancy4.
One approach to overdiagnosis estimation is the excess (or cumulative) incidence approach (comparing disease incidence with and without screening), as in the 25-year follow-up of the Canadian Breast Cancer Screening Study (CNBSS-25)5. This however has some serious complications: in CNBSS-25, screening behavior was not monitored beyond the initial five-year screening period in both the intervention and control arms, leaving undetermined whether screening pattern differences across the trial arms persisted beyond the end of this period6. Such trials with long-term post-screening period follow-up are problematic since overdiagnosis estimates relative to all cases detected during follow-up will necessarily diverge significantly from estimates relative to cases detected strictly during the screening period. With any such trial design not tracking screening after the intervention period, then if either the screen arm continues screening post-intervention period, and/or the control arm starts screening post-intervention, the validity of any overdiagnosis estimation can be undermined7. And this excess/cumulative approach, as well as the lead time approach that estimates the frequency of overdiagnosis from the time to nonscreen diagnosis (the lead time) – which itself must be derived indirectly through estimation of what’s called sojourn time (the latent duration of a disease before clinical symptoms present but during which it is detectable by a screening test) – are both fraught with many complex limitations and challenges8-11,4, from which we can conclude that:
- All such overdiagnosis estimation studies are biased in different ways and to different degrees (thus, in one recent study12, it’s suggested mammographic screening can overdiagnose as much as 81.5% of all women screened, but this is based on the SEER database that is barren of any individual-level data on screening use or compliance).
- Most overdiagnosis estimates are overestimations because of inadequate adjustment for lead time13 and we know that the chosen pre-screening period has the largest influence, so that in one study13, median overdiagnosis is 17.1% using 1975-1986 as the pre-screening period, but 44.7% when using 1975-1988.
- When using an excess/cumulative incidence approach, optimal adjustment for lead time requires that overdiagnosis be estimated in a population in which screening has terminated, and is followed up until the maximum length of lead time after termination of screening13,14.
- Unbiased estimation can only be achieved by waiting until screening stabilizes plus the maximum preclinical period15, with cumulative excess incidence being persistently biased in trials with continued screen arm screening and in settings in which screening is adopted by the population, showing that no trial or population setting automatically permits unbiased empirical estimation of overdiagnosis.
Normalization of Overdiagnosis
To try to extricate ourselves from these pitfalls, overdiagnosis estimates can be “normalized”16,1,17, via studies restricted to screening-attendant (as opposed to only screening-invited) cohorts, and which control for both lead time and background breast cancer incidence, as included in the EUROSCREEN review18 of breast screening programs, the combined estimate of overdiagnosis being just 6.5%.
The Price of Underdiagnosis
And we must avoid neglecting to weigh also the real harms of underdiagnosis, especially the progressive decline in survival for each omitted annual mammography screening: women who had missed any of their previous 5 annual screenings incurred more than a doubling of risk for all-cause mortality compared with subjects having no missed screenings, with HR statistically significant at even just 2 annual missed exams, arguing against a biennial schedule19.
Measuring Overdiagnosis Directly
We can directly observe some of the natural history of screen-detected breast cancers from two special sources2:
(1) patients who refuse treatment, but still elect to return for follow-up;
(2) historic autopsy studies that reveal undiagnosed disease prevalence from the pre-screening era,
and although supporting a small degree of overdiagnosis in DCIS (8.9% median prevalence), these lines of observation fail to support widespread overdiagnosis for invasive cancer (1.3% in the autopsy studies20. As this 1.3% mean invasive cancer incidence tracks disease prevalence closely, this suggests that breast cancers do not remain in quiescence until death and that invasive breast cancers do not regress in untreated patients, strongly entailing that high estimates for overdiagnosis are far likelier to reflect length bias from long – often very long, in hormone-positive disease – natural histories of breast cancers (presenting as “excess cancers”), rather than any true overdiagnosis.
New Understanding of DCIS
Ductal carcinoma in situ (DCIS) is another center-of-the-mammography-debate hot point. Does the detection of DCIS represents overdiagnosis and consequent overtreatment, or is DCIS in fact the ideal stage of the disease for early detection21? To answer, we need to discern the true connection of DCIS to disease progression22,23, which is not trivial. In exploring the impact of DCIS on overdiagnosis in breast cancer screening21, one study found that 50.9 % of all mammographically detected DCIS are high grade, and so with an associated high risk of progression, entailing that aggressive cancer is being detected earlier.
These facts suggests that a benign view of DCIS – that breast screening primarily increases rates of nonadvanced tumors and DCIS, and that DCIS is not usually precancerous and so contributes substantially to overdiagnosis24 – is unwarranted: to support this view, extrapolations were made not from disease-specific mortality data but from incidence of late-stage tumors24, an imperfect and often treacherous surrogate. The authors, to their credit, acknowledge still other significant limitations including inadequate registration of DCIS cases pre-2008, indeterminate extra-program screening, and the significant changes in mammography technology after study’s initiation in 1980. Against this again, consider a just-published study where recurrence-free survival (RFS) = 35% for high-grade, and 13% even for low-grade, DCIS25. On this reckoning, failure to treat even low-grade DCIS would lead to more than half the patients experiencing a local recurrence in 10 years. Worse, invasive local recurrence rates for the surveillance alone group were unacceptably high: at 10-years 26% for low-grade, 31% for non-low-grade, DCIS. It appears that most screening-detected cases of DCIS are in fact of medium and high grade26, with substantial invasivity potential. Thus the fact of high detection of DCIS by screening mammography cannot continue to be construed as unambiguous real harm.
Further problems arise in reliably determining invasive disease overdiagnosis: optimally, the lead time estimates should not be from screen-detected cancers, which of course will include overdiagnosed tumors. Since lead time is a function of sojourn time, one study27 ingeniously derived estimated sojourn time and sensitivity using only interval (symptomatic) cancers, guaranteeing lead time estimates uncontaminated by overdiagnosed cases, generating an estimate for invasive-only overdiagnosed cancers at no more than 7%, lower still if we use individual level data, correction for screening attendant (not just invited) women, and longer follow-up, at least as long as the time until screening stabilizes plus the maximum lead time16.
Overdiagnosis and False Positives in the Real World: What Do Women Really Think?
Data supports that women overwhelmingly elect annual mammographic screening, and beginning at no later than 40, with additional data decisively confirming that even when informed of potential harms, women show regret-avoidance/minimization behavior28 in preferring the risk of overtreatment to the risk of undertreatment29. The screening-eligible perceive overdiagnosis and false positives30-32 – with core-biopsy false-positives hovering around just 1%33 – and any associated anxiety, as acceptable consequences of screening in order to obtain its greater benefits, with adverse psychosocial impact from false positives comparable between women managed invasively (biopsy) and women managed noninvasively34.
Ultimately, the goal is minimization of harm from overdiagnosis and overtreatment through superior differentiation between malignancy-progressive subtypes of DCIS (and atypias) and lesions lacking that potential invasivity. A biomarker that is reliably predictive of what I call tumor militancy, differentiating treatment-mandatory evolutive cancers (including DCIS) from nonevolutive ones, would be fundamentally reductive of overdiagnosis, positively rebalancing the benefit–harm ratio in favor of mammographic screening. Coupled with this, as I have previously argued17, advances in especially digital breast tomosynthesis (DBT) may obsolete much of what we now see as core issues of the mammography debate. All to the benefit of our screening-eligible stakeholders, something I know all partisans in the debate are dedicated to.
Click here to read the Table of Contents for the series: Breast Cancer Screening, Mammography and “Alternative Facts”
- Kaniklidis C, No Surrender Breast Cancer Foundation. Through a glass darkly: the mammography debate. Curr Oncol 2015; 22:171–3.
- Hollingsworth A. Overestimating Overdiagnosis in Breast Cancer Screening. Cureus 2017 Jan 09; 9(1):e966.
- Kopans DB. More misinformation on breast cancer screening. Gland Surg 2017; 6(1):125-129.
- Etzioni R, Gulati R. Recognizing the Limitations of Cancer Overdiagnosis Studies: A First Step Towards Overcoming Them. J Natl Cancer Inst 2016; 108(3).
- Miller AB, Wall C, Baines CJ, et al. Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial. BMJ 2014; 348:g366.
- Etzioni R, Gulati R. Oversimplifying Overdiagnosis. J Gen Intern Med 2014; 29(9):1218–1220.
- Gulati R, Feuer EJ, Etzioni R. Conditions for Valid Empirical Estimates of Cancer Overdiagnosis in Randomized Trials and Population Studies. Am J Epidemiol 2016 Jul 15; 184(2):140-7.
- Biesheuvel C, Barratt A, Howard K, Houssami N, Irwig L. Effects of study methods and biases on estimates of invasive breast cancer overdetection with mammography screening: a systematic review. Lancet Oncol 2007; 8:1129-38.
- Puliti D, Duffy SW, Miccinesi G, de Koning H, Lynge E, Zappa M, et al. Overdiagnosis in mammographic screening for breast cancer in Europe: a literature review. J Med Screen 2012; 19(suppl 1):42-56.
- Etzioni R, Gulati R, Mallinger L, Mandelblatt J. Influence of study features and methods on overdiagnosis estimates in breast and prostate cancer screening. Ann Intern Med 2013; 158:831-8.
- Carter JL, Coletti RJ, Harris RP. Quantifying and monitoring overdiagnosis in cancer screening: a systematic review of methods. BMJ 2015; 350:g7773.
- Welch HG, Prorok PC, O’Malley AJ, Kramer BS. Breast-Cancer Tumor Size, Overdiagnosis, and Mammography Screening Effectiveness. N Engl J Med 2016 Oct 13; 375(15):1438-1447.
- Ripping TM, Verbeek AL, Ten Haff K, van Ravesteyn NT, Broeders MJ. Extrapolation of pre-screening trends: Impact of assumptions on overdiagnosis estimates by mammographic screening. Cancer Epidemiol 2016; 42:147-53.
-  [Ripping 2013] Ripping TM, Verbeek AL, van der Waal D, et al. Immediate and delayed effects of mammographic screening on breast cancer mortality and incidence in birth cohorts. Br J Cancer 2013 Oct 29; 109(9):2467-71.
- Gulati R, Feuer E, Etzioni R. Empirical Estimation Of Overdiagnosis In Trials And Population Studies. 38th Annual North American Meeting of the Society for Medical Decision Making (SMDM). October 23 – 26, 2016. Poster PS 4-42.
- Duffy SW, Parmar D. Overdiagnosis in breast cancer screening: the importance of length of observation period and lead time. Breast Cancer Res 2013; 15(3):R41.
- Kaniklidis C, No Surrender Breast Cancer Foundation. Beyond the mammography debate: a moderate perspective. Curr Oncol 2015; 22(3):220-9.
- Paci E, on behalf of the EUROSCREEN Working Group Summary of the evidence of breast cancer service screening outcomes in Europe and first estimate of the benefit and harm balance sheet. J Med Screen 2012; 19(suppl 1):5–13.
- Engel JM, Stankowski-Drengler TJ, Stankowski RV, Liang H, Doi SA, Onitilo AA. All-cause mortality is decreased in women undergoing annual mammography before breast cancer diagnosis. AJR Am J Roentgenol 2015; 204:898–902.
- Welch HG, Black WC. Using autopsy series to estimate the disease “reservoir” for ductal carcinoma in situ of the breast: how much more breast cancer can we find? Ann Intern Med 1997 Dec 1; 127(11):1023-8.
- van Luijt PA, Heijnsdijk EA, Fracheboud J, et al. The distribution of ductal carcinoma in situ (DCIS) grade in 4232 women and its impact on overdiagnosis in breast cancer screening. Breast Cancer Res 2016 May 10; 18(1):47.
- Collins LC, Tamimi RM, Baer HJ, Connolly JL, Colditz GA, Schnitt SJ. Outcome of patients with ductal carcinoma in situ untreated after diagnostic biopsy: results from the Nurses’ Health Study. Cancer 2005; 103(9):1778–84.
- Sanders ME, Schuyler PA, Dupont WD, Page DL. The natural history of low-grade ductal carcinoma in situ of the breast in women treated by biopsy only revealed over 30 years of long-term follow-up. Cancer 2005;103(12):2481–4.
- Jørgensen KJ, Gøtzsche PC, Kalager M, Zahl PH. Breast Cancer Screening in Denmark: A Cohort Study of Tumor Size and Overdiagnosis. Ann Intern Med 2017 Jan 10.
- Khan S, Epstein M, Lagios MD, Silverstein MJ. Are We Overtreating Ductal Carcinoma in Situ (DCIS)? Ann Surg Oncol 2017; 24(1):59-63.
- Feig SA. Overdiagnosis of breast cancer at screening is clinically insignificant. Acad Radiol. 2015; 22:961–6.
- Michalopoulos D, Duffy SW. Estimation of overdiagnosis using short-term trends and lead time estimates uncontaminated by overdiagnosed cases: Results from the Norwegian Breast Screening Programme. J Med Screen 2016; 23(4):192-202.
- Kaniklidis C. Mammography, Martin Yaffe, and me: response and appreciation. Curr Oncol 2015 Oct; 22(5): e404–e408.
- Waller J, Douglas E, Whitaker KL, Wardle J. Women’s responses to information about overdiagnosis in the U.K. breast cancer screening programme: a qualitative study. BMJ Open 2013; 3:e002703.
- Schwartz LM, Woloshin S, Sox HC, Fischhoff B, Welch HG. US women’s attitudes to false positive mammography results and detection of ductal carcinoma in situ: cross sectional survey. BMJ 2000; 320:1635–40.
- Thomson MD, Siminoff LA. Perspectives on mammography after receipt of secondary screening owing to a false positive. Womens Health Issues 2015; 25:128–33.
- Vyas A, Madhavan S, Sambamoorthi U. Association between persistence with mammography screening and stage at diagnosis among elderly women diagnosed with breast cancer. Breast Cancer Res Treat 2014; 148:645–54.
- Segnan N, Minozzi S, Ponti A, et al. Estimate of false-positive breast cancer diagnoses from accuracy studies: a systematic review. J Clin Pathol 2017 Jan 10.
- Heleno B, Siersma VD, Brodersen J. Diagnostic invasiveness and psychosocial consequences of false-positive mammography. Ann Fam Med. 2015;13:242–9.
The opinions expressed in this article are the author’s own and do not reflect the view of Cancer Knowledge Network or Multimed Inc.
* Constantine Kaniklidis is currently Director of Research for the No Surrender Breast Cancer Foundation (NSBCF), a not-for-profit organization providing high-quality critically reviewed and appraised information and guidance to the breast cancer community. His focus is on the most challenging of advanced / metastatic disease, especially triple negative breast cancer (TNBC) and inflammatory breast cancer (IBC). He is also the Editor of Evidence-based Medicine for the Open Directory Project. Research interests include epigenetic reprogramming, optimal treatment of CNS metastases (brain and leptomeningeal), drug interactions and resistance in oncology, and evidence-based integrative oncology. He is also active at the intersection of politics and medicine, as in his widely accessed paper “Cancer, Culture and Cooperation in the Middle East”.
** The No Surrender Breast Cancer Foundation is a U.S.-based 501(c)3 not-for-profit organization.