Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Overdiagnosis and overtreatment of breast cancer: Estimates of overdiagnosis from two trials of mammographic screening for breast cancer

Overdiagnosis and overtreatment of breast cancer: Estimates of overdiagnosis from two trials of... Breast Cancer Research December 2005 Vol 7 No 6 Duffy et al. Review Overdiagnosis and overtreatment of breast cancer Estimates of overdiagnosis from two trials of mammographic screening for breast cancer 1 1 2 3 4 Stephen W Duffy , Olorunsola Agbaje , Laszlo Tabar , Bedrich Vitak , Nils Bjurstam , 5 1 1 Lena Björneld , Jonathan P Myles and Jane Warwick Cancer Research UK Department of Epidemiology, Mathematics and Statistics, Wolfson Institute of Preventive Medicine, Queen Mary College, University of London, London, UK Department of Mammography, Falun Central Hospital, Sweden Department of Radiology, University Hospital, Linköping, Sweden Department of Radiology, Center for Breast Imaging, University Hospital of North Norway, Tromsø, Norway Department of Radiology, Sahlgrenska University Hospital, Göteborg, Sweden Corresponding author: Stephen W Duffy, stephen.duffy@cancer.org.uk Published: 10 November 2005 Breast Cancer Research 2005, 7:258-265 (DOI 10.1186/bcr1354) This article is online at http://breast-cancer-research.com/content/7/6/258 © 2005 BioMed Central Ltd Abstract continuing interest in the human costs associated with the mortality benefit, in particular, whether overdiagnosis occurs Randomised controlled trials have shown that the policy of in breast cancer screening and, if so, its magnitude [4,5]. In mammographic screening confers a substantial and significant reduction in breast cancer mortality. This has often been this context, overdiagnosis means the diagnosis of cancer as accompanied, however, by an increase in breast cancer incidence, a result of screening, usually histologically confirmed, that particularly during the early years of a screening programme, which would not have arisen clinically during the lifetime of the host has led to concerns about overdiagnosis, that is to say, the had screening not taken place. diagnosis of disease that, if left undetected and therefore untreated, would not become symptomatic. We used incidence When a mammographic screening programme is initiated, data from two randomised controlled trials of mammographic screening, the Swedish Two-county Trial and the Gothenburg Trial, usually a large increase in breast cancer incidence is to establish the timing and magnitude of any excess incidence of observed in the early years of the programme, and a relatively invasive disease and ductal carcinoma in situ (DCIS) in the study small increase later [4,6]. This in itself is not sufficient to imply groups, to ascertain whether the excess incidence of DCIS overdiagnosis, for the following reasons: reported early in a screening trial is balanced by a later deficit in 1. In most parts of the world, breast cancer incidence was invasive disease and provide explicit estimates of the rate of ‘real’ increasing prior to the epoch of mammography. Thus at and non-progressive ‘overdiagnosed’ tumours from the study groups of the trials. We used a multistate model for overdiagnosis least part of any excess incidence observed in the and used Markov Chain Monte Carlo methods to estimate the screening epoch is probably due to an existing increasing parameters. After taking into account the effect of lead time, we trend in incidence. estimated that less than 5% of cases diagnosed at prevalence 2. In addition, the early diagnosis of cancers due to lead screen and less than 1% of cases diagnosed at incidence screens time may exacerbate the underlying temporal increase by are being overdiagnosed. Overall, we estimate overdiagnosis to be bringing forward in time future higher rates of disease. around 1% of all cases diagnosed in screened populations. These estimates are, however, subject to considerable uncertainty. Our 3. In relation to this, screening also causes an artificial results suggest that overdiagnosis in mammography screening is a increase in age-specific incidence. With two years lead minor phenomenon, but further studies with very large numbers are time on average, we would observe age 52 incidence at required for more precise estimation. age 50, and so on. 4. There will be a substantial excess in incidence in the first Introduction few years of the programme due to the prevalence Randomised controlled trials have shown that the policy of screen: large numbers of asymptomatic tumours in the mammographic screening confers a substantial and prevalence pool will have their diagnosis date brought significant reduction in breast cancer mortality [1-3]. There is forward to the time of the prevalence screen. DCIS = ductal carcinoma in situ; MCMC = Markov Chain Monte Carlo. 258 Available online http://breast-cancer-research.com/content/7/6/258 5. There will be a continuing excess thereafter at the lower of DCIS is balanced by a later deficit in invasive disease; and end of the age range for screening, due to prevalence explicit estimation of rates of ‘real’ tumours and non- screens of subjects reaching the age for screening progressive ‘overdiagnosed’ tumours from the study groups eligibility. of the trials. That said, the increase could still be partly due to Methods overdiagnosis. The design features of the two trials have been described in detail elsewhere [1,8]. Briefly, in the Swedish Two-county One would expect the excess incidence due to lead time to Trial, 77,080 women aged 40 to 74 years were randomised be followed by a deficit in incidence in screened cohorts at to regular invitation to screening, and 55,985 to no invitation. ages higher than the upper age limit for screening, as was Screening was by single-view mammography, with an observed in the UK [6]. Estimation from the deficit, however, interscreening interval of 2 years in women aged 40 to is not straightforward, because usually one can identify 49 years and 33 months in women aged 50 to 74 years at screened cohorts only at aggregate rather than individual randomisation. The trial began in late 1977. Around 7 years level, and it takes some years after screening before the later, after approximately 3 rounds of screening in the older subsequent deficit becomes observable. group and 4 rounds of screening in the younger, a mortality reduction of 30% was observed and published [9], the An issue of particular interest is overdiagnosis of ductal control group invited to screening and the screening phase of carcinoma in situ (DCIS) [7]. Here, the question of most the trial closed. Follow-up was continued for mortality from interest is: how much of the DCIS diagnosed at screening the tumours diagnosed during the screening phase [1]. would be expected to progress to invasive cancer if left untreated? The DCIS that would have progressed represents In the Gothenburg Trial, 21,650 women aged 39 to 59 years invasive cancers prevented, a major benefit of screening. were randomised to invitation to screening and 29,961 to no Those that would not have progressed represent invitation [8]. The screening was by two-view mammography overdiagnosis and unnecessary treatment. at first screen, with number of views thereafter dependent on breast density. Screening took place at 18 month intervals. Essential to the concept and existence of overdiagnosis is the The trial began in 1982. After five rounds of screening in the duration of the preclinical screen-detectable period, the 1933 to 1944 birth cohorts (approximately the 39 to 49 year sojourn time. Overdiagnosis can be thought of as a age group at randomisation), the corresponding control combination of two disease entities. The first is the diagnosis group members were offered screening and the screening of a potentially progressive cancer in a subject who is going phase of the trial closed. In the 1923 to 1932 birth cohorts to die of other causes in the near future in any case, possibly (the 50 to 59 year age group), the control group was invited from an accident, another occult disease or an unexpected to screening after four rounds. As in the Swedish Two-county cerebrovascular or cardiovascular event, before the tumour Trial, follow-up has continued for mortality from the tumours would have given rise to clinical symptoms. The second is an diagnosed during the screening phase of the trial. extreme form of length bias whereby there are, in theory, subclinical tumours with little or no potential to progress to In both trials, the control group was offered screening at the symptomatic disease, that is, whose sojourn time has a close of the screening phase, so we cannot estimate radically different distribution from that of the general tumour overdiagnosis by a simple comparison of long term incidence population. rates in the study and control groups. We can, however, study the size and timing of excess incidence during the The first of these must undoubtedly happen, but given the low screening phase to obtain clues to when overdiagnosis may all-cause mortality rates of women in the age groups invited occur. Accordingly, our first analysis was to estimate for screening, and the likely mean and distribution of sojourn cumulative incidence rates of invasive, in situ and total time, this type of overdiagnosis is liable to be very rare [4]. It cancers in the study and control groups of each trial. It has would, therefore, seem more potentially productive in terms of already been noted that in both trials incidence equalised estimation to focus on the latter form of overdiagnosis, a between study and control groups with the first screen of the subpopulation of non-progressive or low-progression control group, suggesting that if there is overdiagnosis, it tumours. occurs mainly at the first screen [2,8]. In this paper, we use two randomised controlled trials of In the Gothenburg Trial, each individual year of birth cohort mammographic screening, the Swedish Two-county Trial and (from 1923 to 1944) was randomised in succession, with a the Gothenburg Trial, to address the following issues: the study to control ratio chosen on the basis of the capacity of timing and magnitude of excess incidence of invasive disease the mammography facilities to screen the study group [8]. and DCIS in the study groups compared to the control The variation of the randomisation ratio by year of birth groups; whether there is evidence that the excess incidence induced an age imbalance (albeit a very small imbalance) 259 Breast Cancer Research December 2005 Vol 7 No 6 Duffy et al. between study and control groups. To take account of this, The second component in the expected rate represents the the Gothenburg study group incidence is compared not with overdiagnosed cancers. the raw control group incidence but with the standardised incidence that would have been observed in the control Between second and third screen: group if it had had exactly the same year of birth distribution –λt –λt –λt as the study group [8]. {(1 – S) (1 – Se ) (1 – e ) + λt – (1 – e )} Our second analysis involved explicit estimation of the As these are symptomatic tumours there is no term for incidence of ‘real’ and ‘overdiagnosed’ cases from the overdiagnosis. numbers of cases detected at screening and between screens in the two trials. We assumed a uniform annual Third screen: incidence I of preclinical but screen detectable, truly SI –λt –λt –λt –µt progressive cancers, an exponential distribution of time from {(1 – S) (1 – Se ) e + (1 – e )} + (1 – e ) inception of these to clinical symptoms with rate λ, and a screening test sensitivity S. In addition, we assume The second component in the expected rate represents the exponential incidence of overdiagnosed (non-progressive) overdiagnosed cancers. preclinical screen-detectable cancers, with rate µ. Because a tumour is only overdiagnosed if it is actually detected at Interval after third screen: screening, we define the screening test sensitivity to be –λt (1 – S)II(1 – e ) –λt –λt –λt 100% for overdiagnosed cancers. In this model, there are {(1 – S) (1 – Se ) e + (1 – e )} + It – λλ four states: no detectable disease, non-progressive (overdiagnosed) preclinical disease, progressive preclinical Since these are symptomatic tumours there is no term for disease, and clinical symptomatic disease. The expected overdiagnosis. rates of cancers diagnosed at first, second and third screens, and in the intervals following those screens with an average From the data on screen-detected and interval cancers, we interval time of t are as follows. estimated I, λ, S and µ by fitting Poisson distributions to the numbers of cases at the three screens and in the three First screen: intervals with expectations as above. For the Swedish Two- county Trial, t = 2.56 years (the average interval for the SI –µa + (1 – e ) 19,844 women younger than 50 years and the 57,236 women aged 50 to 74 years). For the Gothenburg Trial, where a is average age (50 years in the Gothenburg Trial and t = 1.5 years. The estimation algorithm used was Markov 58 years in the Swedish Two-county Trial). The second Chain Monte Carlo (MCMC), implemented in the computer component in the expected rate represents the over- programme WinBUGS [10]. The diagnostic criteria of diagnosed cancers. Geweke, Raftry and Lewis, and Heldelberger and Elch in Convergence Diagnostics and Output Analysis Software This allows a constant incidence rate of non-progressive (CODA) were used to assess convergence of the MCMC disease from birth to age at first screen. This is arbitrary, parameters [11]. The results for the chain provided no biologically unverifiable and it may be wrong. However, the evidence against convergence for all the parameters. We expected rates predicted for any multiplier of µ from 15 or 20 intentionally chose uninformative prior distributions to years upwards are very similar, and it seemed to us less approximate a maximum likelihood solution. Results are arbitrary to allow the subjects’ age to dictate our time limit presented as mean posterior distribution values and 95% than to choose one ourselves, given the current low level of credible intervals. The WinBUGS program updated a single knowledge of non-progressive disease. chain with 15,000 samples (with thinning of 1), from which the first 5,000 samples were discarded (burn-ins) and the Between first and second screen: remaining 10,000 samples were used in estimation. Prior distributions used for the parameters I, λ, S and µ were as –λt {– S(1 – e ) + λt} follows: I, lognormal(0.0, 0.0001); λ, gamma(0.01, 0.01); S, logit(S) = α, α ~ normal(0.0, 0.0001); µ, lognormal(0.0, 0.01). As these are symptomatic tumours there is no term for Note that the second parameter in the normal and lognormal overdiagnosis. distributions is the precision, and not the variance or the standard deviation [10]. Second screen: Results SI –λt –µt (1 – Se ) + (1 – e ) Figure 1a-c shows the cumulative incidence of invasive 260 breast cancer, DCIS, and all breast cancers in the study and Available online http://breast-cancer-research.com/content/7/6/258 Figure 1 Figure 2 Cumulative incidence of breast cancers in study and control groups of the Swedish Two-county Trial. (a) Invasive cancers. (b) In situ cancers. (c) All cancers. control groups of the Swedish Two-county Trial. Figure 2a-c shows the corresponding absolute excesses/deficits in the study group over time, per thousand women randomised. As noted above, the overall rates equalised at years 8 to 9, once Cumulative excess incidence (study versus control) of breast cancers the first screen of the control group was complete. The study in the Swedish Two-county Trial. (a) Invasive cancers. (b) In situ cancers. (c) All cancers. group excess in DCIS rates peaked at 6 to 7 years and was balanced by a deficit in invasive tumours at 8 to 9 years, with the screening of the control group. The absolute excess of DCIS cases in the study group was 60 tumours, and the of all DCIS cases and 1% of all tumours. This can be deficit of invasive tumours was 68, suggesting no over- regarded as an upper limit on the amount of overdiagnosis of diagnosis at all. If, conservatively, we exclude DCIS cases DCIS in the trial. diagnosed at the first screen of the control group, there was an excess of 86 DCIS cases in the study group, suggesting a Figure 3a-c shows the corresponding cumulative incidences total overdiagnosis of 18 DCIS cases. This amounts to 15% in the Gothenburg Trial. 261 Breast Cancer Research December 2005 Vol 7 No 6 Duffy et al. Figure 3 Figure 4 Cumulative incidence of breast cancers in study and control groups of the Gothenburg Trial. (a) Invasive cancers. (b) In situ cancers. (c) All cancers. There was a substantial proportional excess, but very small absolute excess of in situ cancers, which was again balanced by a deficit in invasive cancers (Fig. 4). The excess of in situ cancers peaked at 4 to 5 years. Overall rates equalised at 6 Cumulative excess incidence (study versus control) of breast cancers in the Gothenburg Trial. (a) Invasive cancers. (b) In situ cancers. (c) to 7 years, around the time of screening the control group. All cancers. The absolute excess of DCIS cases was 10, and the deficit of invasive cases was 28, again suggesting no overdiagnosis of DCIS. After exclusion of DCIS cases diagnosed at the first screen of the control group, the excess in the study group Table 1 shows the numbers screened and cancers detected was 35, and the overall balance of all tumour types therefore at the first three screens and in the interval after each of the suggested 7 overdiagnosed cases, 18% of DCIS and 2% of first three screens in the study group of the Swedish Two- all study group cancers, a likely upper limit on overdiagnosis county Trial. Applying the overdiagnosis model to these data 262 of DCIS in this study. gives the results in Table 2. These results pertain to all Available online http://breast-cancer-research.com/content/7/6/258 Table 1 Table 3 Cancers diagnosed at and after the first three screens, Cancers diagnosed at and after the first three screens, Swedish Two-County Trial Gothenburg Trial Number Invasive Number Invasive Detection occasion screened cancers only All cancers Detection occasion screened cancers only All cancers First screen 68,770 384 426 First screen 18,197 55 70 First interval 68,770 123 134 First interval 18,197 7 9 Second screen 58,601 214 244 Second screen 17,005 27 28 Second interval 58,601 78 82 Second interval 17,005 14 14 Third screen 43,320 173 193 Third screen 17,093 36 41 Third interval 43,320 89 91 Third interval 17,093 21 23 Table 2 Table 4 Estimates from formal overdiagnosis modelling, Swedish Two- Estimates from formal overdiagnosis modelling, Gothenburg county Trial data, ages 40 to 74 years Trial data, ages 39 to 59 years Quantity Estimate 95% CI Quantity Estimate 95% CI Incidence (true cases)/1000 (I) 2.2 (2.0-2.4) Incidence (true cases)/1000 (I) 1.9 (1.6-2.3) Screening test sensitivity (S) as % 99.9 (99.6-100) Screening test sensitivity (S) 99.9 (99.6-100) Progression to clinical cancer (λ) 0.34 (0.31-0.38) Progression to clinical cancer (λ) 0.52 (0.41-0.67) Incidence of overdiagnosed cases/1000 (µ) 0.0038 (0.0001-0.0011) Incidence of overdiagnosed cases/1000 (µ) 0.0034 (0.0000-0.0238) Percent overdiagnosis (first screen) 3.1 (0.1-10.9) Percent overdiagnosis (first screen) 4.2 (0.0-28.8) Percent overdiagnosis (second screen) 0.3 (0.1-1.0) Percent overdiagnosis (second screen) 0.3 (0.0-2.0) Percent overdiagnosis (third screen) 0.3 (0.1-1.0) Percent overdiagnosis (third screen) 0.3 (0.0-2.0) CI, confidence interval. CI, confidence interval. cancers, invasive and in situ, but it should be noted that very diagnosed in the first three screening rounds. Restriction of similar results were obtained using invasive cancers only. the analysis to invasive tumours only reduces the Results indicate percentages of tumours overdiagnosed of overdiagnosis estimates by around one-third. 3.1%, 0.3% and 0.3% at the first, second and third screens, respectively. This implies a total of 14 tumours over- Discussion diagnosed, 1% of all tumours, screen-detected and clinical, We have derived formal estimates of overdiagnosis from arising during the period of observation. We also re- empirical breast screening data. The estimates take into estimated the parameters restricting the data to the 40 to account the effect of lead time and use direct estimation of 69 year age group, as the 70 to 74 year age group was only the underlying incidence of both ‘true’ and ‘overdiagnosed’ invited to the first two screens. Results were very similar, cases from the screened populations. We found over- giving overdiagnosis rates of 3%, 0.2% and 0.2% at the first diagnosis to be a minor phenomenon, with less than 5% of three screens, and an overall percentage overdiagnosed of cases diagnosed at prevalence screen and less than 1% of 1% of all tumours diagnosed in the programme. cases at incidence screens being overdiagnosed. Over- diagnosis was estimated at around 1% of all cases Table 3 shows the corresponding data for the Gothenburg diagnosed in the screened populations. Trial, and Table 4 the results of overdiagnosis modelling from the Gothenburg data. Results show 4.2% overdiagnosis at Examination of absolute incidence rates of DCIS and invasive first screen and 0.3% at subsequent screens. This corres- disease suggest further that overdiagnosis of DCIS is not the ponds to three cancers diagnosed, two percent of all tumours major problem it is claimed to be [12]. While large relative 263 Breast Cancer Research December 2005 Vol 7 No 6 Duffy et al. increases in DCIS rates have been cited as evidence for such This article is part of a review series on overdiagnosis [12], absolute rates of detection of DCIS Overdiagnosis and overtreatment of breast cancer, remain low, at around one per thousand screened [13]. edited by Nick E Day, Stephen Duffy and Eugenio Paci. Previous detailed estimation of DCIS progression is in agree- ment with our results [14]. Other articles in the series can be found online at http://breast-cancer-research.com/articles/ Other estimates of overdiagnosis in the literature range from review-series.asp?series=BCR_Overdiagnosis 5% or less [4] up to 30% [15]. The latter, however, does not formally take into account the lead time effect, and does not fully identify screened and unscreened cohorts. We would It would be of some interest to see estimates from formal suggest that simple estimation of rates at an aggregate level, models from other screening trials and service screening while useful, is not sufficient in itself to derive conclusive programmes. In the meantime, the results here suggest that estimates of overdiagnosis rates. overdiagnosis in mammography screening is a minor phenomenon. We need more data to reduce the uncertainty Our estimates of incidence of preclinical disease in the two around these estimates. trials are similar to the clinical incidence rates in the respective control groups before their exit screen (2.1 per Competing interests 1,000 and 1.8 per 1,000 for the Swedish Two-county and The author(s) declare that they have no competing interests. Gothenburg Trials, respectively). It should be noted that we have wide confidence intervals on our overdiagnosis Authors’ contributions estimates, and the estimate of screening test sensitivity tends SWD conceived and designed the study, supervised the to drift to its boundary at 100%. Also, there is some statistical analyses and drafted the manuscript. OA carried out sensitivity to the prior distribution for µ, the incidence rate of statistical analyses. LT led the Swedish Two-county Trial and overdiagnosed cancers, uniform priors tending to give higher contributed to the manuscript. BV conducted the Ostergotland estimates of µ. For more stable estimation, perhaps overview part of the Swedish Two-county trial and contributed to the estimates from several screening programmes, as in Yen et manuscript. NB led the Gothenburg Trial and contributed to al. [14], are indicated. the manuscript. LB managed the Gothenburg Trial administration and contributed to the manuscript. JPM In both trials, our estimate of sensitivity drifted towards its contributed to the statistical modelling. JW carried out upper bound of 100%. Two points should be noted here. statistical analyses and contributed to the manuscript. Firstly, the part of the likelihood related to the prevalence screen is monotonic increasing in S, as are the parts related Acknowledgements We thank the women who participated in the trials and the clinical and to incidence screens under most circumstances. The support staff who contributed time and expertise to them. We also likelihood component related to the interval cancers is not, thank Jayne Mead for secretarial support. but if there are very few interval cancers, this can be outweighed by the likelihood pertaining to screen-detected References cancers. This reflects the fact that a very high sensitivity is 1. Tabar L, Vitak B, Chen HH, Duffy SW, Smith RA: The Swedish Two-County Trial twenty years later: updated mortality results implied if there are very low interval cancer rates. Secondly, and new insights from long term follow-up. Radiol Clin Nth our sensitivity estimate is of test sensitivity, not program Am 2000, 38:625-651. sensitivity, which includes all interval cancers as false 2. Smith RA, Duffy SW, Gabe R, Tabar L, Yen AMF, Chen HHT: The randomized trials of breast cancer screening: what have we negatives. Our estimate differs from that of others [16], learned? Radiol Clin Nth Am 2004, 42:793-806. largely because it takes account of the sojourn time in 3. Nyström L, Andersson I, Bjurstam N, Frisell J, Nordenskjold B, estimation of the proportion of interval cancers that are really Rutqvist LE: Long-term effects of mammography screening: updated overview of the Swedish randomised trials. Lancet newly arising since the screen, as opposed to those missed 2002, 360:724. at the screen. As noted above, if the observed number of 4. Paci E, Warwick J, Falini P, Duffy SW: Overdiagnosis in service screening: should the increase in breast cancer incidence interval cancers is small, the estimate of S must be close to rates necessarily be a cause for concern? J Med Screen 2004, 100%. It should be noted that the maximum likelihood 11:23-27. estimate of S would also be 100%. 5. Anttila A, Koskela J, Hakama M: Programme sensitivity and effectiveness of mammography service screening in Helsinki, Finland. J Med Screen 2002, 9:153-158. The models we have fitted here are rather simple. Only a single 6. McCann J, Treasure P, Duffy SW: Modelling the impact of detecting and treating carcinoma in situ in a breast screening overdiagnosis parameter is estimated. There is room for programme. J Med Screen 2004, 11:117-125. improvement, in terms of estimation of age-specific over- 7. Evans AJ, Pinder SE, Ellis IO, Wilson AR: Screen detected diagnosis rates, for example. Multiple overdiagnosis parameters, ductal carcinoma in situ (DCIS): overdiagnosis or an obligate precursor of invasive disease? J Med Screen 2001, 8:149- and the small numbers resulting when analysis is restricted to age subgroups, both give rise to instability of estimation. Solving 8. Bjurstam N, Björneld L, Warwick J: The Gothenburg Breast 264 this problem is a target of ongoing research. Screening Trial. Cancer 2003, 97:2387-2396. Available online http://breast-cancer-research.com/content/7/6/258 9. Tabar L, Fagerberg CJ, Gad A, Baldetorp L, Holmberg LH, Grontoft O, Ljungquist U, Lundstrom B, Manson JC, Eklund G, et al.: Reduction in mortality from breast cancer after mass screening with mammography. Randomised trial from the Breast Cancer Screening Working Group of the Swedish National Board of Health and Welfare. Lancet 1985, 1:829- 10. WinBUGS 1.4.1 [http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/ contents.shtml] 11. CODA: Convergence Diagnostics and Output Analysis Soft- ware 0.4 [http://www.mrc-bsu.cam.ac.uk/bugs/classic/coda04/ readme.shtml] 12. Ernster VL, Barclay J: Increases in ductal carcinoma in situ (DCIS) of the breast in relation to mammography: a dilemma. J Natl Cancer Inst Monogr 1997, 22:151-156. 13. Ernster VL, Ballard-Barbash R, Barlow WE, Zheng Y, Weaver DL, Cutter G, Yankaskas BC, Rosenberg R, Carney PA, Kerlikowske K, et al.: Detection of ductal carcinoma in situ in women undergoing screening mammography. J Natl Cancer Inst 2002, 94:1546-1554. 14. Yen MF, Tabar L, Vitak B, Smith RA, Chen HH, Duffy SW: Quan- tifying the potential problem of overdiagnosis of ductal carci- noma in situ in breast cancer screening. Eur J Cancer 2003, 39:1746-1754. 15. Zahl PH, Strand BH, Maehlen J: Incidence of breast cancer in Norway and Sweden during introduction of nationwide screening: prospective cohort study. Br Med J 2004, 328:921- 16. Vainio H, Bianchim F: Breast Cancer Screening. Lyon: IARC Press; 2002. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Breast Cancer Research Springer Journals

Overdiagnosis and overtreatment of breast cancer: Estimates of overdiagnosis from two trials of mammographic screening for breast cancer

Loading next page...
 
/lp/springer-journals/overdiagnosis-and-overtreatment-of-breast-cancer-estimates-of-TXCd7pL0hv

References (23)

Publisher
Springer Journals
Copyright
2005 BioMed Central Ltd
eISSN
1465-542X
DOI
10.1186/bcr1354
Publisher site
See Article on Publisher Site

Abstract

Breast Cancer Research December 2005 Vol 7 No 6 Duffy et al. Review Overdiagnosis and overtreatment of breast cancer Estimates of overdiagnosis from two trials of mammographic screening for breast cancer 1 1 2 3 4 Stephen W Duffy , Olorunsola Agbaje , Laszlo Tabar , Bedrich Vitak , Nils Bjurstam , 5 1 1 Lena Björneld , Jonathan P Myles and Jane Warwick Cancer Research UK Department of Epidemiology, Mathematics and Statistics, Wolfson Institute of Preventive Medicine, Queen Mary College, University of London, London, UK Department of Mammography, Falun Central Hospital, Sweden Department of Radiology, University Hospital, Linköping, Sweden Department of Radiology, Center for Breast Imaging, University Hospital of North Norway, Tromsø, Norway Department of Radiology, Sahlgrenska University Hospital, Göteborg, Sweden Corresponding author: Stephen W Duffy, stephen.duffy@cancer.org.uk Published: 10 November 2005 Breast Cancer Research 2005, 7:258-265 (DOI 10.1186/bcr1354) This article is online at http://breast-cancer-research.com/content/7/6/258 © 2005 BioMed Central Ltd Abstract continuing interest in the human costs associated with the mortality benefit, in particular, whether overdiagnosis occurs Randomised controlled trials have shown that the policy of in breast cancer screening and, if so, its magnitude [4,5]. In mammographic screening confers a substantial and significant reduction in breast cancer mortality. This has often been this context, overdiagnosis means the diagnosis of cancer as accompanied, however, by an increase in breast cancer incidence, a result of screening, usually histologically confirmed, that particularly during the early years of a screening programme, which would not have arisen clinically during the lifetime of the host has led to concerns about overdiagnosis, that is to say, the had screening not taken place. diagnosis of disease that, if left undetected and therefore untreated, would not become symptomatic. We used incidence When a mammographic screening programme is initiated, data from two randomised controlled trials of mammographic screening, the Swedish Two-county Trial and the Gothenburg Trial, usually a large increase in breast cancer incidence is to establish the timing and magnitude of any excess incidence of observed in the early years of the programme, and a relatively invasive disease and ductal carcinoma in situ (DCIS) in the study small increase later [4,6]. This in itself is not sufficient to imply groups, to ascertain whether the excess incidence of DCIS overdiagnosis, for the following reasons: reported early in a screening trial is balanced by a later deficit in 1. In most parts of the world, breast cancer incidence was invasive disease and provide explicit estimates of the rate of ‘real’ increasing prior to the epoch of mammography. Thus at and non-progressive ‘overdiagnosed’ tumours from the study groups of the trials. We used a multistate model for overdiagnosis least part of any excess incidence observed in the and used Markov Chain Monte Carlo methods to estimate the screening epoch is probably due to an existing increasing parameters. After taking into account the effect of lead time, we trend in incidence. estimated that less than 5% of cases diagnosed at prevalence 2. In addition, the early diagnosis of cancers due to lead screen and less than 1% of cases diagnosed at incidence screens time may exacerbate the underlying temporal increase by are being overdiagnosed. Overall, we estimate overdiagnosis to be bringing forward in time future higher rates of disease. around 1% of all cases diagnosed in screened populations. These estimates are, however, subject to considerable uncertainty. Our 3. In relation to this, screening also causes an artificial results suggest that overdiagnosis in mammography screening is a increase in age-specific incidence. With two years lead minor phenomenon, but further studies with very large numbers are time on average, we would observe age 52 incidence at required for more precise estimation. age 50, and so on. 4. There will be a substantial excess in incidence in the first Introduction few years of the programme due to the prevalence Randomised controlled trials have shown that the policy of screen: large numbers of asymptomatic tumours in the mammographic screening confers a substantial and prevalence pool will have their diagnosis date brought significant reduction in breast cancer mortality [1-3]. There is forward to the time of the prevalence screen. DCIS = ductal carcinoma in situ; MCMC = Markov Chain Monte Carlo. 258 Available online http://breast-cancer-research.com/content/7/6/258 5. There will be a continuing excess thereafter at the lower of DCIS is balanced by a later deficit in invasive disease; and end of the age range for screening, due to prevalence explicit estimation of rates of ‘real’ tumours and non- screens of subjects reaching the age for screening progressive ‘overdiagnosed’ tumours from the study groups eligibility. of the trials. That said, the increase could still be partly due to Methods overdiagnosis. The design features of the two trials have been described in detail elsewhere [1,8]. Briefly, in the Swedish Two-county One would expect the excess incidence due to lead time to Trial, 77,080 women aged 40 to 74 years were randomised be followed by a deficit in incidence in screened cohorts at to regular invitation to screening, and 55,985 to no invitation. ages higher than the upper age limit for screening, as was Screening was by single-view mammography, with an observed in the UK [6]. Estimation from the deficit, however, interscreening interval of 2 years in women aged 40 to is not straightforward, because usually one can identify 49 years and 33 months in women aged 50 to 74 years at screened cohorts only at aggregate rather than individual randomisation. The trial began in late 1977. Around 7 years level, and it takes some years after screening before the later, after approximately 3 rounds of screening in the older subsequent deficit becomes observable. group and 4 rounds of screening in the younger, a mortality reduction of 30% was observed and published [9], the An issue of particular interest is overdiagnosis of ductal control group invited to screening and the screening phase of carcinoma in situ (DCIS) [7]. Here, the question of most the trial closed. Follow-up was continued for mortality from interest is: how much of the DCIS diagnosed at screening the tumours diagnosed during the screening phase [1]. would be expected to progress to invasive cancer if left untreated? The DCIS that would have progressed represents In the Gothenburg Trial, 21,650 women aged 39 to 59 years invasive cancers prevented, a major benefit of screening. were randomised to invitation to screening and 29,961 to no Those that would not have progressed represent invitation [8]. The screening was by two-view mammography overdiagnosis and unnecessary treatment. at first screen, with number of views thereafter dependent on breast density. Screening took place at 18 month intervals. Essential to the concept and existence of overdiagnosis is the The trial began in 1982. After five rounds of screening in the duration of the preclinical screen-detectable period, the 1933 to 1944 birth cohorts (approximately the 39 to 49 year sojourn time. Overdiagnosis can be thought of as a age group at randomisation), the corresponding control combination of two disease entities. The first is the diagnosis group members were offered screening and the screening of a potentially progressive cancer in a subject who is going phase of the trial closed. In the 1923 to 1932 birth cohorts to die of other causes in the near future in any case, possibly (the 50 to 59 year age group), the control group was invited from an accident, another occult disease or an unexpected to screening after four rounds. As in the Swedish Two-county cerebrovascular or cardiovascular event, before the tumour Trial, follow-up has continued for mortality from the tumours would have given rise to clinical symptoms. The second is an diagnosed during the screening phase of the trial. extreme form of length bias whereby there are, in theory, subclinical tumours with little or no potential to progress to In both trials, the control group was offered screening at the symptomatic disease, that is, whose sojourn time has a close of the screening phase, so we cannot estimate radically different distribution from that of the general tumour overdiagnosis by a simple comparison of long term incidence population. rates in the study and control groups. We can, however, study the size and timing of excess incidence during the The first of these must undoubtedly happen, but given the low screening phase to obtain clues to when overdiagnosis may all-cause mortality rates of women in the age groups invited occur. Accordingly, our first analysis was to estimate for screening, and the likely mean and distribution of sojourn cumulative incidence rates of invasive, in situ and total time, this type of overdiagnosis is liable to be very rare [4]. It cancers in the study and control groups of each trial. It has would, therefore, seem more potentially productive in terms of already been noted that in both trials incidence equalised estimation to focus on the latter form of overdiagnosis, a between study and control groups with the first screen of the subpopulation of non-progressive or low-progression control group, suggesting that if there is overdiagnosis, it tumours. occurs mainly at the first screen [2,8]. In this paper, we use two randomised controlled trials of In the Gothenburg Trial, each individual year of birth cohort mammographic screening, the Swedish Two-county Trial and (from 1923 to 1944) was randomised in succession, with a the Gothenburg Trial, to address the following issues: the study to control ratio chosen on the basis of the capacity of timing and magnitude of excess incidence of invasive disease the mammography facilities to screen the study group [8]. and DCIS in the study groups compared to the control The variation of the randomisation ratio by year of birth groups; whether there is evidence that the excess incidence induced an age imbalance (albeit a very small imbalance) 259 Breast Cancer Research December 2005 Vol 7 No 6 Duffy et al. between study and control groups. To take account of this, The second component in the expected rate represents the the Gothenburg study group incidence is compared not with overdiagnosed cancers. the raw control group incidence but with the standardised incidence that would have been observed in the control Between second and third screen: group if it had had exactly the same year of birth distribution –λt –λt –λt as the study group [8]. {(1 – S) (1 – Se ) (1 – e ) + λt – (1 – e )} Our second analysis involved explicit estimation of the As these are symptomatic tumours there is no term for incidence of ‘real’ and ‘overdiagnosed’ cases from the overdiagnosis. numbers of cases detected at screening and between screens in the two trials. We assumed a uniform annual Third screen: incidence I of preclinical but screen detectable, truly SI –λt –λt –λt –µt progressive cancers, an exponential distribution of time from {(1 – S) (1 – Se ) e + (1 – e )} + (1 – e ) inception of these to clinical symptoms with rate λ, and a screening test sensitivity S. In addition, we assume The second component in the expected rate represents the exponential incidence of overdiagnosed (non-progressive) overdiagnosed cancers. preclinical screen-detectable cancers, with rate µ. Because a tumour is only overdiagnosed if it is actually detected at Interval after third screen: screening, we define the screening test sensitivity to be –λt (1 – S)II(1 – e ) –λt –λt –λt 100% for overdiagnosed cancers. In this model, there are {(1 – S) (1 – Se ) e + (1 – e )} + It – λλ four states: no detectable disease, non-progressive (overdiagnosed) preclinical disease, progressive preclinical Since these are symptomatic tumours there is no term for disease, and clinical symptomatic disease. The expected overdiagnosis. rates of cancers diagnosed at first, second and third screens, and in the intervals following those screens with an average From the data on screen-detected and interval cancers, we interval time of t are as follows. estimated I, λ, S and µ by fitting Poisson distributions to the numbers of cases at the three screens and in the three First screen: intervals with expectations as above. For the Swedish Two- county Trial, t = 2.56 years (the average interval for the SI –µa + (1 – e ) 19,844 women younger than 50 years and the 57,236 women aged 50 to 74 years). For the Gothenburg Trial, where a is average age (50 years in the Gothenburg Trial and t = 1.5 years. The estimation algorithm used was Markov 58 years in the Swedish Two-county Trial). The second Chain Monte Carlo (MCMC), implemented in the computer component in the expected rate represents the over- programme WinBUGS [10]. The diagnostic criteria of diagnosed cancers. Geweke, Raftry and Lewis, and Heldelberger and Elch in Convergence Diagnostics and Output Analysis Software This allows a constant incidence rate of non-progressive (CODA) were used to assess convergence of the MCMC disease from birth to age at first screen. This is arbitrary, parameters [11]. The results for the chain provided no biologically unverifiable and it may be wrong. However, the evidence against convergence for all the parameters. We expected rates predicted for any multiplier of µ from 15 or 20 intentionally chose uninformative prior distributions to years upwards are very similar, and it seemed to us less approximate a maximum likelihood solution. Results are arbitrary to allow the subjects’ age to dictate our time limit presented as mean posterior distribution values and 95% than to choose one ourselves, given the current low level of credible intervals. The WinBUGS program updated a single knowledge of non-progressive disease. chain with 15,000 samples (with thinning of 1), from which the first 5,000 samples were discarded (burn-ins) and the Between first and second screen: remaining 10,000 samples were used in estimation. Prior distributions used for the parameters I, λ, S and µ were as –λt {– S(1 – e ) + λt} follows: I, lognormal(0.0, 0.0001); λ, gamma(0.01, 0.01); S, logit(S) = α, α ~ normal(0.0, 0.0001); µ, lognormal(0.0, 0.01). As these are symptomatic tumours there is no term for Note that the second parameter in the normal and lognormal overdiagnosis. distributions is the precision, and not the variance or the standard deviation [10]. Second screen: Results SI –λt –µt (1 – Se ) + (1 – e ) Figure 1a-c shows the cumulative incidence of invasive 260 breast cancer, DCIS, and all breast cancers in the study and Available online http://breast-cancer-research.com/content/7/6/258 Figure 1 Figure 2 Cumulative incidence of breast cancers in study and control groups of the Swedish Two-county Trial. (a) Invasive cancers. (b) In situ cancers. (c) All cancers. control groups of the Swedish Two-county Trial. Figure 2a-c shows the corresponding absolute excesses/deficits in the study group over time, per thousand women randomised. As noted above, the overall rates equalised at years 8 to 9, once Cumulative excess incidence (study versus control) of breast cancers the first screen of the control group was complete. The study in the Swedish Two-county Trial. (a) Invasive cancers. (b) In situ cancers. (c) All cancers. group excess in DCIS rates peaked at 6 to 7 years and was balanced by a deficit in invasive tumours at 8 to 9 years, with the screening of the control group. The absolute excess of DCIS cases in the study group was 60 tumours, and the of all DCIS cases and 1% of all tumours. This can be deficit of invasive tumours was 68, suggesting no over- regarded as an upper limit on the amount of overdiagnosis of diagnosis at all. If, conservatively, we exclude DCIS cases DCIS in the trial. diagnosed at the first screen of the control group, there was an excess of 86 DCIS cases in the study group, suggesting a Figure 3a-c shows the corresponding cumulative incidences total overdiagnosis of 18 DCIS cases. This amounts to 15% in the Gothenburg Trial. 261 Breast Cancer Research December 2005 Vol 7 No 6 Duffy et al. Figure 3 Figure 4 Cumulative incidence of breast cancers in study and control groups of the Gothenburg Trial. (a) Invasive cancers. (b) In situ cancers. (c) All cancers. There was a substantial proportional excess, but very small absolute excess of in situ cancers, which was again balanced by a deficit in invasive cancers (Fig. 4). The excess of in situ cancers peaked at 4 to 5 years. Overall rates equalised at 6 Cumulative excess incidence (study versus control) of breast cancers in the Gothenburg Trial. (a) Invasive cancers. (b) In situ cancers. (c) to 7 years, around the time of screening the control group. All cancers. The absolute excess of DCIS cases was 10, and the deficit of invasive cases was 28, again suggesting no overdiagnosis of DCIS. After exclusion of DCIS cases diagnosed at the first screen of the control group, the excess in the study group Table 1 shows the numbers screened and cancers detected was 35, and the overall balance of all tumour types therefore at the first three screens and in the interval after each of the suggested 7 overdiagnosed cases, 18% of DCIS and 2% of first three screens in the study group of the Swedish Two- all study group cancers, a likely upper limit on overdiagnosis county Trial. Applying the overdiagnosis model to these data 262 of DCIS in this study. gives the results in Table 2. These results pertain to all Available online http://breast-cancer-research.com/content/7/6/258 Table 1 Table 3 Cancers diagnosed at and after the first three screens, Cancers diagnosed at and after the first three screens, Swedish Two-County Trial Gothenburg Trial Number Invasive Number Invasive Detection occasion screened cancers only All cancers Detection occasion screened cancers only All cancers First screen 68,770 384 426 First screen 18,197 55 70 First interval 68,770 123 134 First interval 18,197 7 9 Second screen 58,601 214 244 Second screen 17,005 27 28 Second interval 58,601 78 82 Second interval 17,005 14 14 Third screen 43,320 173 193 Third screen 17,093 36 41 Third interval 43,320 89 91 Third interval 17,093 21 23 Table 2 Table 4 Estimates from formal overdiagnosis modelling, Swedish Two- Estimates from formal overdiagnosis modelling, Gothenburg county Trial data, ages 40 to 74 years Trial data, ages 39 to 59 years Quantity Estimate 95% CI Quantity Estimate 95% CI Incidence (true cases)/1000 (I) 2.2 (2.0-2.4) Incidence (true cases)/1000 (I) 1.9 (1.6-2.3) Screening test sensitivity (S) as % 99.9 (99.6-100) Screening test sensitivity (S) 99.9 (99.6-100) Progression to clinical cancer (λ) 0.34 (0.31-0.38) Progression to clinical cancer (λ) 0.52 (0.41-0.67) Incidence of overdiagnosed cases/1000 (µ) 0.0038 (0.0001-0.0011) Incidence of overdiagnosed cases/1000 (µ) 0.0034 (0.0000-0.0238) Percent overdiagnosis (first screen) 3.1 (0.1-10.9) Percent overdiagnosis (first screen) 4.2 (0.0-28.8) Percent overdiagnosis (second screen) 0.3 (0.1-1.0) Percent overdiagnosis (second screen) 0.3 (0.0-2.0) Percent overdiagnosis (third screen) 0.3 (0.1-1.0) Percent overdiagnosis (third screen) 0.3 (0.0-2.0) CI, confidence interval. CI, confidence interval. cancers, invasive and in situ, but it should be noted that very diagnosed in the first three screening rounds. Restriction of similar results were obtained using invasive cancers only. the analysis to invasive tumours only reduces the Results indicate percentages of tumours overdiagnosed of overdiagnosis estimates by around one-third. 3.1%, 0.3% and 0.3% at the first, second and third screens, respectively. This implies a total of 14 tumours over- Discussion diagnosed, 1% of all tumours, screen-detected and clinical, We have derived formal estimates of overdiagnosis from arising during the period of observation. We also re- empirical breast screening data. The estimates take into estimated the parameters restricting the data to the 40 to account the effect of lead time and use direct estimation of 69 year age group, as the 70 to 74 year age group was only the underlying incidence of both ‘true’ and ‘overdiagnosed’ invited to the first two screens. Results were very similar, cases from the screened populations. We found over- giving overdiagnosis rates of 3%, 0.2% and 0.2% at the first diagnosis to be a minor phenomenon, with less than 5% of three screens, and an overall percentage overdiagnosed of cases diagnosed at prevalence screen and less than 1% of 1% of all tumours diagnosed in the programme. cases at incidence screens being overdiagnosed. Over- diagnosis was estimated at around 1% of all cases Table 3 shows the corresponding data for the Gothenburg diagnosed in the screened populations. Trial, and Table 4 the results of overdiagnosis modelling from the Gothenburg data. Results show 4.2% overdiagnosis at Examination of absolute incidence rates of DCIS and invasive first screen and 0.3% at subsequent screens. This corres- disease suggest further that overdiagnosis of DCIS is not the ponds to three cancers diagnosed, two percent of all tumours major problem it is claimed to be [12]. While large relative 263 Breast Cancer Research December 2005 Vol 7 No 6 Duffy et al. increases in DCIS rates have been cited as evidence for such This article is part of a review series on overdiagnosis [12], absolute rates of detection of DCIS Overdiagnosis and overtreatment of breast cancer, remain low, at around one per thousand screened [13]. edited by Nick E Day, Stephen Duffy and Eugenio Paci. Previous detailed estimation of DCIS progression is in agree- ment with our results [14]. Other articles in the series can be found online at http://breast-cancer-research.com/articles/ Other estimates of overdiagnosis in the literature range from review-series.asp?series=BCR_Overdiagnosis 5% or less [4] up to 30% [15]. The latter, however, does not formally take into account the lead time effect, and does not fully identify screened and unscreened cohorts. We would It would be of some interest to see estimates from formal suggest that simple estimation of rates at an aggregate level, models from other screening trials and service screening while useful, is not sufficient in itself to derive conclusive programmes. In the meantime, the results here suggest that estimates of overdiagnosis rates. overdiagnosis in mammography screening is a minor phenomenon. We need more data to reduce the uncertainty Our estimates of incidence of preclinical disease in the two around these estimates. trials are similar to the clinical incidence rates in the respective control groups before their exit screen (2.1 per Competing interests 1,000 and 1.8 per 1,000 for the Swedish Two-county and The author(s) declare that they have no competing interests. Gothenburg Trials, respectively). It should be noted that we have wide confidence intervals on our overdiagnosis Authors’ contributions estimates, and the estimate of screening test sensitivity tends SWD conceived and designed the study, supervised the to drift to its boundary at 100%. Also, there is some statistical analyses and drafted the manuscript. OA carried out sensitivity to the prior distribution for µ, the incidence rate of statistical analyses. LT led the Swedish Two-county Trial and overdiagnosed cancers, uniform priors tending to give higher contributed to the manuscript. BV conducted the Ostergotland estimates of µ. For more stable estimation, perhaps overview part of the Swedish Two-county trial and contributed to the estimates from several screening programmes, as in Yen et manuscript. NB led the Gothenburg Trial and contributed to al. [14], are indicated. the manuscript. LB managed the Gothenburg Trial administration and contributed to the manuscript. JPM In both trials, our estimate of sensitivity drifted towards its contributed to the statistical modelling. JW carried out upper bound of 100%. Two points should be noted here. statistical analyses and contributed to the manuscript. Firstly, the part of the likelihood related to the prevalence screen is monotonic increasing in S, as are the parts related Acknowledgements We thank the women who participated in the trials and the clinical and to incidence screens under most circumstances. The support staff who contributed time and expertise to them. We also likelihood component related to the interval cancers is not, thank Jayne Mead for secretarial support. but if there are very few interval cancers, this can be outweighed by the likelihood pertaining to screen-detected References cancers. This reflects the fact that a very high sensitivity is 1. Tabar L, Vitak B, Chen HH, Duffy SW, Smith RA: The Swedish Two-County Trial twenty years later: updated mortality results implied if there are very low interval cancer rates. Secondly, and new insights from long term follow-up. Radiol Clin Nth our sensitivity estimate is of test sensitivity, not program Am 2000, 38:625-651. sensitivity, which includes all interval cancers as false 2. Smith RA, Duffy SW, Gabe R, Tabar L, Yen AMF, Chen HHT: The randomized trials of breast cancer screening: what have we negatives. Our estimate differs from that of others [16], learned? Radiol Clin Nth Am 2004, 42:793-806. largely because it takes account of the sojourn time in 3. Nyström L, Andersson I, Bjurstam N, Frisell J, Nordenskjold B, estimation of the proportion of interval cancers that are really Rutqvist LE: Long-term effects of mammography screening: updated overview of the Swedish randomised trials. Lancet newly arising since the screen, as opposed to those missed 2002, 360:724. at the screen. As noted above, if the observed number of 4. Paci E, Warwick J, Falini P, Duffy SW: Overdiagnosis in service screening: should the increase in breast cancer incidence interval cancers is small, the estimate of S must be close to rates necessarily be a cause for concern? J Med Screen 2004, 100%. It should be noted that the maximum likelihood 11:23-27. estimate of S would also be 100%. 5. Anttila A, Koskela J, Hakama M: Programme sensitivity and effectiveness of mammography service screening in Helsinki, Finland. J Med Screen 2002, 9:153-158. The models we have fitted here are rather simple. Only a single 6. McCann J, Treasure P, Duffy SW: Modelling the impact of detecting and treating carcinoma in situ in a breast screening overdiagnosis parameter is estimated. There is room for programme. J Med Screen 2004, 11:117-125. improvement, in terms of estimation of age-specific over- 7. Evans AJ, Pinder SE, Ellis IO, Wilson AR: Screen detected diagnosis rates, for example. Multiple overdiagnosis parameters, ductal carcinoma in situ (DCIS): overdiagnosis or an obligate precursor of invasive disease? J Med Screen 2001, 8:149- and the small numbers resulting when analysis is restricted to age subgroups, both give rise to instability of estimation. Solving 8. Bjurstam N, Björneld L, Warwick J: The Gothenburg Breast 264 this problem is a target of ongoing research. Screening Trial. Cancer 2003, 97:2387-2396. Available online http://breast-cancer-research.com/content/7/6/258 9. Tabar L, Fagerberg CJ, Gad A, Baldetorp L, Holmberg LH, Grontoft O, Ljungquist U, Lundstrom B, Manson JC, Eklund G, et al.: Reduction in mortality from breast cancer after mass screening with mammography. Randomised trial from the Breast Cancer Screening Working Group of the Swedish National Board of Health and Welfare. Lancet 1985, 1:829- 10. WinBUGS 1.4.1 [http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/ contents.shtml] 11. CODA: Convergence Diagnostics and Output Analysis Soft- ware 0.4 [http://www.mrc-bsu.cam.ac.uk/bugs/classic/coda04/ readme.shtml] 12. Ernster VL, Barclay J: Increases in ductal carcinoma in situ (DCIS) of the breast in relation to mammography: a dilemma. J Natl Cancer Inst Monogr 1997, 22:151-156. 13. Ernster VL, Ballard-Barbash R, Barlow WE, Zheng Y, Weaver DL, Cutter G, Yankaskas BC, Rosenberg R, Carney PA, Kerlikowske K, et al.: Detection of ductal carcinoma in situ in women undergoing screening mammography. J Natl Cancer Inst 2002, 94:1546-1554. 14. Yen MF, Tabar L, Vitak B, Smith RA, Chen HH, Duffy SW: Quan- tifying the potential problem of overdiagnosis of ductal carci- noma in situ in breast cancer screening. Eur J Cancer 2003, 39:1746-1754. 15. Zahl PH, Strand BH, Maehlen J: Incidence of breast cancer in Norway and Sweden during introduction of nationwide screening: prospective cohort study. Br Med J 2004, 328:921- 16. Vainio H, Bianchim F: Breast Cancer Screening. Lyon: IARC Press; 2002.

Journal

Breast Cancer ResearchSpringer Journals

Published: Dec 1, 2005

Keywords: Cancer Research; Oncology; Surgical Oncology

There are no references for this article.