BREAST CANCER SCREENING: controversies D David Dershaw Memorial Sloan Kettering Cancer Center New York, NY Areas of general agreement about mammographic screening Screening mammography has been demonstrated to save lives of women 39 65 in meta analysis of RCTs Screening mammography is recommended for normal risk women ages 50 74 by all major medical groups The test is imperfect with harms Screening Concepts Breast cancer screening is like using seatbelts: most users will never need it must be used chronically to have an impact it does not prevent even thought you use it, you may die it can harm you 1
US BREAST CANCER MORTALITY 1975 2007 Mammo screening starts in USA ~30% reduction Breast cancer 5 year survival by stage at diagnosis 99% 84% 23% US Breast Cancer Incidence Rates By Stage of Disease 60 50 per 100, 000 Cases 40 30 20 10 0 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 Year In Situ Incidence Rates Stage IV Incidence Rates Stage I Incidence Rates Source: SEER 2
Swedish 2 county trial: Screening Impact on Breast Cancer Mortality with 29 Year Followup Tabar. Radiology 2011; 260:658 663 Relative risk of breast cancer death for those invited to screen vs not invited = 0.69 0.71 Screening 300 women for 10 years prevents one breast cancer death. Longer followup (at least 20 years) shows an increasing advantage to screening. 2011 by Radiological Society of North America Norwegian Breast Cancer Screening Program: Breast cancer mortality screened vs nonscreened attending 1996 2010 Non attenders Ever attended Norwegian Breast Cancer Screening Program: Breast cancer mortality screened vs nonscreened attending 1996 2010 Non attenders 62% reduction at 13 years Ever attended 3
Breast Cancer Death Rate: vs THE SAME THERAPIES ARE AVAILBLE TO MEN AND WOMEN WITH BREAST CANCER Since the mid 1990 s death rate from breast cancer for women has fallen 34% (34 22/100,000) Since mid 1990 s death rate from breast cancer for men has not changed (0.275 0.300/100,000) CONCEPT: M REBNER US DEATHS FROM BREAST CANCER/100,000 WOMEN: BY AGE 250 35-39 40-44 200 45-49 50-54 150 55-5959 100 60-64 65-69 50 70-74 0 75-79 80-84 85+ 1930 1940 1950 1960 1970 1980 1990 1998 source: ACS, 2003 Mortality Reduction in Prospective Randomized Trials of Mammography 35 HIP 40-64 24 (7-38) 30 Malmo 45-69 19 (-8-39) 25 Two county 40-74 32 (20-41) 20 Edinburgh 45-64 21 (-2-40) 15 Gothenburg 39-59 16 (-39-49) 10 NBSS-1 40-49 -3 (-26-27) 5 NBSS-2 50-59 -2 (-33-22) 0 Stockholm 40-64 26 (-10-50) -5 All 39-74 24 (18-30) 4
Limitations in Prospective Randomized Trials for Mammographic Screening Comparison of invited for mammography vs not invited for mammography not comparison of got screened vs didn t get screened Only 1/3 of invited underwent all screening About 10% of uninvited underwent screening outside the trial Interval between mammograms was variable, ranging from 12 months to three years Quality of mammography in some trials, particularly NBSS, was judged to be poor by outside experts Therefore, these trials underestimate the value of mammographic screening Problems in Demonstrating Mortality Reduction by Mammographic Screening for Women 40 49 Lower breast cancer incidence Only 1/3 of women in studies are in this decade of age Higher rates of DCIS with longer time to impact on mortality requires longer followup to demonstrate advantage Shorter lead time for invasive cancers requires shorter interval mammography BREAST CANCER INCIDENCE VS AGE SEER data 1993 7 5
If current age is... Probability of developing breast cancer in next 10 years is Or 1 in: 20 0.05% 05% 2,152 30 0.40% 251 40 1.45% 69 Biennial (USPSTF) vs Annual Lives Saved per 1000 Women Screened Hendrick, Helvie. AJR 2012 Age Biennial Annual % Improvement 40 s 0.97 1.34 38% 50 s 2.20 2.85 30% 60 s 3.30 4.30 30% 70 s 2.35 2.65 13% Biennial 50 74 (USPSTF) vs Annual 40 80 (ACS, NCCN)Life Years Gained (LYG) Henrick, Helvie AJR 2011; Mandelblatt Ann Int Med 2009 LYG/1000 % MORTALITY REDUCTION B 50 74 110 23% A 40 84 189 40% Difference 79 (72%) 16.3 (71%) 6
Harms For women ages 40 79 undergoing annual screening incidence of these harms is: False positive biopsy 6.7 4.3/1000 screened 149 233 years Additional imaging 84 64/1000 screened 12 16 years Annual vs Biennial Harms Assuming annual = q 9 18 months and biennial = q 19 30 months: At 10 years: annual biennial RECALL 61% 42% 47% BIOPSY 7% 5% 42% Stage III,IV(40 49) 14.9 10.1 32%* Hubbard. Ann Intern Med 2011 *statistically significant value Natural History of DCIS Diagnosis of DCIS has increased from 2.4 to 27.7/100,000 women from 1981 to 2001 due to screening Discovery and treatment of DCIS in premenopausal women results in decreased breast cancer mortality at > 7 years Incidence of DCIS in screen detected cancers is: Up to 50% premenopausal 25% postmenopausal 7
Evidence for the progression of DCIS to invasive cancer 50% of all recurrences after DCIS conservation treatment are invasive Longterm followup of low grade DCIS treated only by biopsy without definitive excision or RT have at 30 years 30 60% incidence of invasive carcinoma, usually at or near the DCIS site The grade and histology of DCIS and related IFDC is usually similar DCIS decreases with advancing age as IFDC increases Cady. J Surg Oncol 1998; 69:60 Evidence for the progression of DCIS to invasive cancer Risk factors for DCIS and invasive cancer are similar Thereis a progression of genetic changes from atypia to DCIS to invasive cancer suggesting progression of atypia to DCIS finally resulting in invasive cancer Cady. J Surg Oncol 1998; 69:60 Percentage of Cancers (A) and Absolute # (B) of ductal carcinoma in situ (DCIS) per 1000 mammograms. Percentage of all cancers found at screening Number of cases of DCIS found at screening Ernster V L et al. JNCI J Natl Cancer Inst 2002;94:1546-1554 Oxford University Press 8
Recurrence rates after treatment for DCIS varying by grade as indicated by tumor necrosis With necrosis Without t necrosis Fisher ER. Cancer 1999; 86:429 BMJ: NBSS 25 YEAR F/U Twenty five year follow up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial BMJ 2014; 348 doi: http://dx.doi.org/10.1136/bmj.g366 (Published 11 February 2014) All cause mortality, by assignment to mammography or control arms (all participants) Miller A B et al. BMJ 2014;348:bmj.g366 2014 by British Medical Journal Publishing Group 9
OVER EXACTLY THE SAME TIME PERIOD THESE TWO OUTCOMES REPORTED SURVIVAL IN CANADA MORTALITY IN NORWAY NONATTENDERS ATTENDERS OVER EXACTLY THE SAME TIME PERIOD THESE TWO OUTCOMES REPORTED SURVIVAL IN CANADA MORTALITY IN NORWAY NO DIFFERENCE 62% MORTALITY REDUCTION NONATTENDERS ATTENDERS HOW IS THIS POSSIBLE? Mortality Reduction in Prospective Randomized Trials of Mammography 35 HIP 30 Malmo 25 Two county 20 Edinburgh 15 Gothenburg 10 NBSS-1 5 NBSS-2 Stockholm 0 All -5 10
Mortality Reduction in Prospective Randomized Trials of Mammography 35 HIP 30 Malmo 25 Two county 20 Edinburgh 15 Gothenburg 10 NBSS-1 5 NBSS-2 Stockholm 0 All -5 CANADIAN STUDIES Mortality Reduction in Prospective Randomized Trials of Mammography 35 HIP 30 Malmo 25 Two county 20 15 10 5 0-5 Edinburgh Gothenburg NBSS-1 NBSS-2 Stockholm All EVERY OTHER PROSPECTIVE RANDOMIZED TRIAL CANADIAN STUDIES NBSS MAMMOGRAPHY QUALITY Documented poor quality mammography. 1.Old devices (10 years old in Vancouver) at least 1 second hand. 2Nogrids 2.No 3.No training for techs used straight lateral not MLO 4.No training for the radiologists 5.Mammo size cancers = 1.9 cm Controls = 2.1 cm 11
"..in my work as reference physicist to the NBSS, [I] identified many concerns regarding the quality of mammography carried out in some of the NBSS screening centers. That quality [in the NBSS] was far below state of the art, even for that time (early 1980's). Yaffe MJ. Correction: Canada Study. Letter to the Editor JNCI 1993;85:94). NBSS Study Design A positive mammogram did not lead to a biopsy but to a surgical consultation to determine the need for biopsy. About 25% of recommended biopsies based on an abnormal mammogram were never performed. NBSS: FAULTY RANDOMIZATION For unexplained reasons, more women with advanced breast cancers were randomized to the screening arm vs control arm: While Canada in general had a 75% 5 year breast cancer survival rate, the control arm had 90% 5 year survival Women diagnosed with >4 positive nodes in the first year were 19 in the study arm and 11 in the control At seven years there were 29 breast cancer deaths in the study arm vs 18 in the control group 12
Baseline Incidence The incidence data from the first year in which breast cancer incidence was recorded (1973) were almost certainly spuriously low (which would bias our estimates of excess detection upward). The data from the subsequent 2 years (1974 and 1975) were above average for the decade Since these years show low and high incidences of breast cancer, using them would estimate a higher incidence of expected breast cancer than the authors published. Therefore, they chose different years to calculate baseline. Consequently, we chose the 3 year period 1976 through 1978 to obtain our estimate of the baseline incidence of breast cancer In 1973 Happy Rockefeller and in 1974 Betty Ford announced they had breast cancer. The increased use of mammography at that time resulted in the early diagnosis of breast cancers that would have presented clinically in the following few years and resulted in a decrease in breast cancer incidence during the 1976 1978 period. N Engl J Med 2012; 367:1999 2005 Due to screening alone, in 2008 breast cancer was overdiagnosed in more than 70,000 women; this accounted for 31% of all breast cancers diagnosed Bleyer and Welch claim that t since there were more cancers diagnosed in 2008 than they ESTIMATED should have occurred, that excess must be fake cancers and due to overdiagnosis. ov er di ag no sis They estimated a 0.25% increase per year in breast cancer incidence 13
Breast cancer incidence rates: SEER data 1973 1989 Garfinkel el at. Cancer 1994; 74:222 7 Breast cancer incidence rates: SEER data 1973 1989 Garfinkel el at. Cancer 1994; 74:222 7 BREAST CANCER IS INCREASING 1% PER YEAR US BREAST CANCER INCIDENCE 1940: 60 INVASIVE CANCERS/100,000 WOMEN 1980: 100 INVASIVE CANCERS/100,000 WOMEN Over 40 years there was an increase of 40 invasive cancers/100,000 women which is a 1% increase annually 14
Actually fewer invasive breast cancers than expected have occurred US BREAST CANCER INCIDENCE Since the start of mammographic screening, the increase in invasive cancers has slowed, although the diagnosis of DCIS has increased. This suggests that DCIS treatment interrupts the development of invasive cancers. And that the diagnosis of invasive cancer is not overaggressive, but that the treatment of in situ cancers is preventative. Mammographic screening: Reduces breast cancer deaths in women 40 and older Is best done as an annual examination Results in a small number of women having additional imaging and, rarely, biopsies for benign conditions Earlier diagnosis improves treatment options Earlier diagnosis decreases medical costs 15
Bad science leads to bad medicine 16