Breast cancer screening is associated with increased appropriateness of surgical treatment

Poster Sessions control group was never invited to screening, while in the ten youngest (born 1923-32, 45-54 years at entry) the control group was invited after 1990. Incident breast cancer cases for the period 1987-2001 were identified through record linkage with the Swedish Cancer Register. Results: At the end of the period of randomised design there was an excess of 24% (150 cases). In women aged 55-69 years at randomisation the excess was 32% and in women aged 45-54 it was16%. An excess number of cases of 10% (115 cases) remained in the former invited group at the end of follow-up in 2001. This includes invasive as well as in situ carcinoma. If only invasive carcinoma is considered, the over diagnosis reached a level of 7%. Conclusion: Screening by mammography is associated with over diagnosis. The rate of over-diagnosis was greater in older than in younger women and it was mainly caused by invasive breast cancer. Women opting to continue screening after 70 years of age, should be informed of a substantial risk of over diagnosis. 35 Poster German pathologists in preparation for population based mammography screening: Results of a nation-wide training course series T. Decker, D. Hungermann, H. B/irger, C. Kersting, W. B6cker. University

of Muenster, G. Domagk Institute of Pathology, MOnster, Germany Introduction: In 2005, a population based mammography screening program for Germany based on the European guidelines has been started. Without improving the pathology service, the benefit of screening may be reduced or even lost. Especially specimens from minimal invasive diagnosis (MIB) of screened women provide pathologists with problems. To prepare pathologists who intend to participate in the German screening project the authors give courses on breast screening pathology as the first feature within an external quality assurance scheme. Aims: 1. to evaluate the effect of the training courses on diagnostic performance, 2. to assess the performance of German pathologists interested in breast cancer screening, and 3. to identify the diagnostic problems. Methods: From February to November 2005, 297 pathologists participated in 15 diagnostic courses, each for maximum 20 participants. Each course included 4 microscopic tutorials and 8 presentations. It started and ended with a diagnostic test using the same set of 11 slides of MIB specimens selected within broad diagnostic categories by two of us (T.D. and W.B.) with complete agreement. Participants reading the slides used a form planned as standard reporting form for the German Screening Project including the B category according to the EU Guidelines. Kappa statistics were calculated separately for the entry and the closing test. The advantage of these statistics is that kappa values are independent of the diagnosis; they simply reflect the consistency of ratings by the participants. The following limits were used for interpretation: <0.0: none, 0.0-0.2: poor, 0.21-0.4: slight, 0.41-0.6: fair, 0.61-0.8: good, 0.81-0.92: very good, and 0.93-1.0: excellent. Results: 1. After the course the rate of cases with "good" and better consistency increased from 7 to 9/11 cases. 2. The overall consistency of the participants' B-categories is 0.62.3. There are 4 features with poor to fair agreement: columnar cell change, columnar cell hyperplasia, DCIS low grade, and phyllodes tumor. Conclusions: 1. The training course improved the consistency of the test results in MIB. 2. The consistence of attending German pathologists in MIB diagnosis is in general at least substantial (kappa >0.6) 3. To improve the performance, further education should focus on differential diagnosis of columnar cell lesions and non-high nuclear grade DCIS/ductal hyperplasia, respectively. 36 Poster Mammographic density and the size of breast cancers detected by programmatic screening C. Nickson, A.M. Kavanagh, G.B. Byrnes. University of Melbourne,

School of Population Health, Me/bourne, Australia We assess the effect of mammographic density on tumour size at detection in the Australian mamrnographic screening program, in order to examine the relative mortality benefits of mammographic screening. Brief description: Firstly we examine descriptive statistics of tumou r size according to density (threshold percent density or dense area) and mode of detection (screening or interval). We then use regression analyses to adjust for potential confounders and assess whether there is effect modification by mode of detection. Summary: The median and mean size of cancers detected by screening generally increased with density at first and subsequent rounds. No such effect was apparent in interval cancers.

Wednesday, 22 March 2006

55

Histograms of tumour size quintile groups by density quintile groups revealed a shift in size distribution towards larger cancers with increasing density, at first (Figure 1) and subsequent rounds. Cancers detected in women with lower density tended to be smaller and screen-detected, whereas cancers detected in women with higher density tended to be larger and more often interval-detected. T h e s e differences appeared to be incremental across the five density groups. Intervalcancers

....... Evendistnbutionline

~30%

|,oq . . . . . . . . . . . . . . . . . . . . . . . . . . .

~'~~

/

sss~ Density s

ss8~ DensitYQ2

sss~ DensitYs

ss8~ DensitY Q4

s88~8 Density Q5

Figure 1. Distribution of tumour size by density at first round screening. Regression showed that cancers detected in women with very high mammographic density were significantly larger than those detected in women with very low mammographic density. This effect was graded across the range of densities and was apparent both before and after adjustment for age, hormone replacement therapy (HRT) use, family history, symptomatic status and detection, however the trend was less clear in the middle range of densities after adjustment for effect modification by mode of detection. Of the factors that are potential targets for program improvement (such as age and family history), only density and HRT use were significantly associated with tumcur size in the fully adjusted model. The effects at subsequent round were less clear. Figure 2 shows the expected mean tumour size at first round screening according to density, for a 50-year-old asymptomatic woman who was not using HRT at the time of screening and had no family history of breast cancer. 30"1~25 . .~ 20 15 -

10 -

Intervaldetected Screendetected

~, 0Q1

Q2

Q3 Q4 Q5 Q6 Q7 (~8 Q9 Q10 Percent d ~ s i ~ decile Figure 2. Expected mean tumour size at first round screening, by density.

Conclusion: Tumours detected in women with higher density are larger than those detected in women with lower density, and this effect is somewhat graded across the range of densities at first round screening. 37 Poster A multidisciplinary team approach to family history risk assessment reduced clinic attendance by half J. Saldanha 1 , R. Garret 1 , M. Longmuir3, C. Watt 3, D. George 2, C. Wilson 2, J. Doughty 2, D. Smith 1, S. Stallard 1, R. Davidson 3. 1Victoria

Infirmary, Breast Surgery, Glasgow, United Kingdom; 2 Western Infirmary, Breast Surgery, Glasgow, United Kingdom; 3 Yorkhill Hospital, Genetics, Glasgow, United Kingdom The aim of this study was to accurately assess the breast cancer risk of family history surveillance patients at 2 breast units in Glasgow utilising a multidisciplinary team approach and national guidelines. In our practise most women had been attending for family history surveillance before the introduction of Scottish guidelines that have increased awareness about the importance of accurate risk assessment in this group of women. Many had therefore never undergone a formal risk assessment or verification of their family histories and had only ever been seen by the surgical team. Women attending for surveillance between November 2003 and February 2005 were included in this study. Genetics staff attended family history clinics and used Scottish guidelines in a 15 minute consultation to classify women already attending for annual follow up as being at low, moderate, or high risk of breast cancer. Where possible family members had their diagnoses verified through a national cancer registry. A surgical team that included specialist nurses examined all women and post consultation a summary letter including confirmation of the risk level was sent to the individual.

56

Wednesday, 22 March 2006

Poster Sessions

1005 clinic attendees were assessed 310 women (31%) were low risk, 584 (58%) were moderate risk. and 111 (11%) were high risk We continue to follow up 511 women (51%) These are the 386 (38%) moderate risk women aged 35 50. the 111 (11%) women over 35 years classified as high risk. and 10 women (1%) who declined discharge After accurate risk assessment 503 (50%) women attending for screening were eligible for discharge to primary care and/or the national breast screening programme. Overall 995 women (99%) complied with their revised management plan. Current guidelines and a multi disciplinary team approach have helped to more clearly define women who should have surveillance due to their family history, allowing them to avoid potentially harmful investigations, and significantly reducing the surgical clinic workload. 38 Breast cancer screening is associated with increased appropriateness of surgical treatment

Poster

M. Zorzi 1, D. Pulifi 2, E. Paci 2. The Impact Working Group ~Venetian Tumour

Registry, Padua Hospital, Padua, italy, 2Centre 1or Study and Prevenbon of Cancer, Research Institute ol Tuscany Region, Florence, Italy Introduction: On the basis of a review of the trials on breast screening with mammography, it has been claimed that screening increases mastectomy rates However. the point is not mastectomy rates, but whether screening modifies breast surgery appropriateness We describe the situation of breast cancer surgery in the late 1990s in Italy and the impact of screening services on its appropriateness. Materials and Methods: We collected population data on areas from six Italian regions on all incident cases of in situ and invasive breast cancer in women aged 40 to 79 years from 1997 to 2001. We classified cases according to method of detection in screen detected (SD) at first test, SD at subsequent tests, not screen detected (NSD) with a previous screening test, NSD never complfars and NSD never invited; we also evaluated screening by intention to treat (I]q) according to invited and non invited categories Results: We enrolled 2162 in situ and 21.148 invasive breast cancer cases Overall. 61 1% of cases underwent BCS. with a constant increase dudng the period of study The proportion of BCS was around 80% for pTis and pT1 a and pTlb, but was less than 50% for cases pT2 ~< 30 mm 75% cases SD at first test and 83% cases SD at subsequent tests had BCS. compared to two thirds cases NSD with a previous screening test and about 50% in those never invited to screening. At multinomial analysis on cases pTis and pT ~< 30mm. the OR for mastectomy (adjusted by year of diagnosis, age, size of the tumour, method of detection and Centre) decreased by 9% per calendar year. Compared to cases NSD never invited, the odd fur mastectomy showed a statistically significant reduction by one third and by half in cases SD at first and subsequent test. respectively, and was reduced by 15% in cases NSD with a previous test In the analysis by ITT the odd for mastectomy was significantly lower for cases invited to screening (OR 0 74. at p < 0 001) Conclusion: In the late 1990s in Italy there has been an increase in appropriateness of breast surgery, with a reduction of mastectomies in small lesions The risk of inappropriate mastectomy was lowest in cases screen detected: screening had a positive effect also on cases NSD with a previous screening test, partly due to patients referring to screening' diagnostic and therapeutic services. 39 Poster Histological upgrade of atypical ductal hyperplasia diagnosed with percutaneous core biopsy. A review of 58 Chinese patients B.Y.K. Lam I . W.K. Hung I . K.L. Mak 2. C.Y. Lui a, M.C.M. Chan 1, A W C Yip 1 ~Breast Centre, K~rang Wah Hospital, Surgery, Hong Kong,

Chtha. 2Kwong Wah Hospital. f>~thology,Hong Kong. China, 3Kwong Wah Hospital, Radiology. Hong Kong, Chtha Introduction: Management of atypical ductal hyperplasia (ADH) diagnosed with core biopsy/vacuum assisted biopsy device was usually fallowed with surgical excision because of its assoclafion with malignancy. However, this potentially leads to over treatment for patients. The present study was to review the upgrade rate for ADH and to identify any predicting factor for upgrade over 5 years pedod in our centre Design of study: Those patients presented with mammographic microcalcifications only and with percutaneous core biopsy yielded ADH were recruited Wire-guided biopsy was advised to all patients Histological upgrade was defined as detection of ductal carcinoma in situ (DCIS) or invasive ductal carcinoma (IDC) in subsequent surgical biopsy Predicting factors fur histological upgrade were analyzed using univarlate analysis and statistical significance was tested using Chi Square test. Results: From July 2000 to June 2005, there were ,58,436 patients attending our mammogram screening clinic. Stereotactic biopsies were

performed for 625 patients presented with abnormal microcalcifications 58 Chinese patients had ADH on stereotactic biopsy 48 patients had subsequent surgical excision and were included in the present study An upgrade rate of 29% was found 13 patients had DCIS and 1 patient had IDC Non-cribriform pattern in histology was found as a statistical significant predicting factor of upgrade when compared with cdbdform pattern; 53% (8/15) compared with 18.2% (6/33) (p= 0.013). Complete removal of microcalcl f/cations during core biopsy was associated with 20% upgrade, whereas incomplete removal with 31.6%, however, the result was not statistical significant (p = 0.474). Conclusions: Surgical excision is recommended alter atypical ductal hyperplasia is diagnosed with core biopsy fur mammographic microcal cifications. Our series demonstrated an overall upgrade of 29%. Non cdbdform histological pattern was found to be a significant predicting factor for histological upgrade 40 Poster Results of the new French breast cancer screening programme for year 2003 R. Ancelle Park ~, B. Seradour2, A.C. Patty 1, M. Julien ~, J. Bloch ~. ~/NVS,

Saint Maurtce, France; 2Arcades, Marsetlle, France The French national breast cancer screening programme set up in 1994 alter a pilot phase, was implemented in 32 districts. Evaluation of this decentralised programme showed a good quality but a poor attendance, cleady indicating that this programme was not accepted. The French department of health therefore developed a new protocol which was implemented in 2002. National coverage of the programme with this new protocol, targeted to women aged ,50 to 74, was achieved in the first quarter of 2004 The results of the new programme are presented Methods: The new protocol comprised, every two years, a systematic clinical breast examination, two views, immediate assessment by the first reader (R1) in case of abnormal mammography, central/sod double reading of all normal mammographies by a restricted number of radiologists (R2) Data is collected on a yearly base by the National public health institute (InVS) Data for year 2003 59 districts were available for analysis Results: A total of 831,44,5 women attended the new programme in 2003. The attendance rates reached 32% varying from 20 to 60% one district to another. Abnormal mammography rate befure assessment reached 13.8% and 3.6% after immediate assessment. True recall Fate alter R2 was 2%. Abnormal clinical examination with normal mammography reached 0.3% and decreased to 0.03% alter assessment. Overall cancer detection rates were 6.39'| (prevalence 6.81'~, incidence 5.91'~). Cancer rates of the new programme for women aged 50 to 69 were higher (5 8~'~) than the rates observed in the former programme (4 9~'~) A total of 8 3% of the cancers were detected by the second readen DCIS rates were 12 8 and rates of cancers ~<10mm were high 35 3% (prevalenca 35 8% incidence 38 9%) Positive predictive value of surgical biopsy reached 80% Conclusion: The new breast cancer screening programme targets women aged 50 to 74 and offers screening procedures in agreement with the French screening practices in a decentralised health care system. The main aims of the programme are, by promoting quality, to increase attendance and progressively replace individual screening by the organised screening programme. 41 Poster Cost implications of the first year of the Hungarian nation wide breast cancer screening programme I. Boncz 1'2'a A. Sebestyen a'4. G. Holler I . J. Sandor a, A. Budai 5. M. Pal 8, I Ember ~ /National Health thsurance Fund Administration (OEP),

Department of Health Policy, Budapest, Hungary, 2 University of P~cs, Institute of Diagnosbcs and Management, P#cs, Hungary, 9University of P#cs, thstitute of Public Health and Preventive Medicine, P~cs, Hungary, 4National Health Insurance Fund Admtmstra#on (OEP), County Baranya Health Insurance Fund, PCcs, Hungary, 5Nabonal Pubhc Health and Medical Officers Service (,4NTSZ), Budapest, Hungary, ~Nabonal Health insurance Fund AdmtmstratJon (OEP), Department ol Financing, Budapest, Hungary Background and aim: A nation wide organized breast cancer programme was introduced in January 2002 in Hungary for women aged 45 65 and a 2 years screening interval is applied The aim of this study is to calculate the annual financial burden of the breast cancer screening programme and to calculate the cost of finding one cancer case Data and Methods: The data denve from the database of the National Health Insurance Fund Administration containing routinely collected f/nan cial data reported by the health care providers (induding mammography centres). We calculated the cost of mammography screening, further