Huge decreases in the risk of breast cancer relapse over the last three decades

Introduction The aim of this study was to evaluate local and systemic breast cancer control by comparing the risk of relapse in breast cancer patients in 2003– 2004 with that in 1972–1979 and in 1980–1986. Materials and methods About 8,570 women diagnosed with invasive breast cancer in 2003–2004 were selected from the populationbased Netherlands Cancer Registry and compared with 133 patients treated in 1972–1979 and 174 in 1980–1986. Five-year risk of relapse was calculated by the Kaplan–Meier method. Cox-proportional hazard models were applied to adjust for tumour size, nodal status and age at diagnosis. Results Patients diagnosed in 2003–2004 had smaller tumours and a less advanced nodal stage than patients diagnosed in 1972–1986. In 1972– 1979, 1980–1986 and 2003–2004, treatment included mastectomy in 94%, 72% and 47%; postmastectomy radiotherapy in 75%, 70% and 30%; chemotherapy in 9%, 14% and 37% and hormonal therapy in 3%, 3% and 42% of patients, respectively. Fiveyear risk of locoregional and distant recurrence decreased from 37% and 34% to 15%, respectively. The 5-year risk of second primary breast cancer did not differ and was 1%, 4% and 2%, respectively. The improved relapse-free survival in patients diagnosed in 2003–2004 as compared with 1972–1979 hardly changed after adjustment (HR = 0.38, 95% CI = 0.28–0.52). Conclusion Over the last decades, local breast cancer therapies have become less rigorous, whereas systemic therapy use has increased. Simultaneously, the risk of breast cancer relapse has tremendously decreased. Future novel therapies may lead to such small additional decreases in relapse rates, while the long-term side effects in breast cancer survivors will increase. Introduction The breast cancer relapse rate is the key outcome when evaluating local and systemic disease control achieved by primary breast cancer management. Changing relapse rates may have implications for patient management and post-treatment follow-up. In contrast to breast cancer mortality, population-based studies on the time trend in relapse-free survival are rare and generally focus on locoregional recurrence1,2, second primary breast cancer3,4 or distant metastasis5,6 separately, instead of studying overall time trends on relapse-free survival. Relapse rate estimates derived from randomized controlled trials (RCTs) may differ from daily practice, as trials are usually carried out in specialized clinics. Patients and physicians participating in a trial are likely to be more motivated, and patient care is expected to be of higher quality7,8. For that reason, population-based outcome research is of pivotal importance to assess the real impact of innovations after translation in daily practice. The purpose of this populationbased study was to determine the risk of relapse in breast cancer patients treated with curative intent diagnosed in 2003–2004 and in 1972–1986. Materials and methods This work conforms to the values laid down in the Declaration of Helsinki (1964). The protocol of this study has been approved by the relevant ethical committee of the Netherlands Cancer Registry. Two cohorts of female breast cancer patients operated for invasive nonmetastatic breast cancer were available. The oldest cohort comprised 307 patients consecutively diagnosed from 1972 to 1986 in the St Anna hospital in Oss, the Netherlands. Occurrences of relapse and death were registered in 1993. A recent population-based cohort was selected from the Netherlands Cancer Registry and comprised 8,570 patients diagnosed in 2003 or 2004 in 53 hospitals in the Netherlands. The Netherlands Cancer Registry has registered all primary tumours since 1984, but does not routinely register cancer recurrences9. Trained registration clerks gathered data from the patients’ files based on * Corresponding author Email: S.Geurts@ebh.umcn.nl 1 Department for Health Evidence, Radboud University Medical Centre, Nijmegen, The Netherlands 2 Department of Surgery, Hospital Bernhoven, Oss, The Netherlands 3 Comprehensive Cancer Centre the Netherlands, Utrecht, The Netherlands 4 Health Technology and Services Research, Twente University, Enschede, The Netherlands 5 Department of Internal Medicine, GROW – School for Oncology and Developmental Biology, Division of Medical Oncology, Maastricht University Medical Centre, Maastricht, The Netherlands


Introduction
The breast cancer relapse rate is the key outcome when evaluating local and systemic disease control achieved by primary breast cancer management.Changing relapse rates may have implications for patient management and post-treatment follow-up.In contrast to breast cancer mortality, population-based studies on the time trend in relapse-free survival are rare and generally focus on locoregional recurrence 1,2 , second primary breast cancer 3,4 or distant metastasis 5,6 separately, instead of studying overall time trends on relapse-free survival.Relapse rate estimates derived from randomized controlled trials (RCTs) may differ from daily practice, as trials are usually carried out in specialized clinics.Patients and physicians participating in a trial are likely to be more motivated, and patient care is expected to be of higher quality 7,8 .For that reason, population-based outcome research is of pivotal importance to assess the real impact of innovations after translation in daily practice.
The purpose of this populationbased study was to determine the risk of relapse in breast cancer patients treated with curative intent diagnosed in 2003-2004 and in  1972-1986.

Materials and methods
This work conforms to the values laid down in the Declaration of Helsinki (1964).The protocol of this study has been approved by the relevant ethical committee of the Netherlands Cancer Registry.
Two cohorts of female breast cancer patients operated for invasive nonmetastatic breast cancer were available.The oldest cohort comprised 307 patients consecutively diagnosed from 1972 to 1986 in the St Anna hospital in Oss, the Netherlands.Occurrences of relapse and death were registered in 1993.A recent population-based cohort was selected from the Netherlands Cancer Registry and comprised 8,570 patients diagnosed in 2003 or 2004 in 53 hospitals in the Netherlands.The Netherlands Cancer Registry has registered all primary tumours since 1984, but does not routinely register cancer recurrences 9  Competing interests: none declared.Conflict of interests: none declared.
All authors contributed to conception and design, manuscript preparation, read and approved the final manuscript.
All authors abide by the Association for Medical Ethics (AME) ethical rules of disclosure.
using the method described by Dickman 14 .Five-year risk of relapse was calculated using the Kaplan-Meier method.Relapse was defined as a second primary breast cancer, locoregional recurrence or distant metastasis.Date of surgery was defined as the starting point of follow-up.When locoregional recurrence was considered as event of interest, patients' follow-up was censored at the date of last follow-up, distant metastasis or death.For risk of distant metastasis, patients were censored at the last follow-up date, occurrence of a local or regional recurrence or death.We therefore calculated the risk of locoregional recurrence or distant metastasis.Patients with second primary breast cancer were censored at the time of last follow-up or death.
Patients with synchronous bilateral breast cancer at primary diagnosis were excluded from the risk of second primary breast cancer analyses.
Of patients who had a second primary breast cancer diagnosed within 3 months of the first breast cancer, the tumour with the highest Nottingham prognostic index 12 was included for further analyses.
Breast cancer relapse was defined as a local or regional recurrence, distant metastasis or second primary breast cancer 13 .Relapses diagnosed within 3 months were considered to be diagnosed simultaneously.

Data analyses
To examine the time trend in patient, tumour and treatment characteristics and breast cancer relapse rates, the historical cohort was divided into two groups.Analyses were consequently reported for three periods: 1972-1979, 1980-1986 and 2003-2004.The utilization of radiotherapy was stratified by the type of surgery.
Five-year relative survival by age, sex and calendar year was calculated pathological notification through an automated pathology archive.Information about relapse and death was retrieved until December 31, 2009, corresponding with at least 5 years of follow-up since diagnosis.Only first occurrence of either locoregional recurrence or distant metastasis was registered.
Patients diagnosed in 1972-1986 were staged according to the TNM classification of 1977 10 .In 2003-2004, the TNM classification of 2002 was used 11 .Staging of tumour size (T) was similar for both periods, but categorization of patients by lymph node status (N) differed between 1977 and 2002 (Table 1).Nodal status was, therefore, categorized as N0, that is no lymph node metastases, versus N1-3, that is at least one lymph node metastasis.Patients with unknown pathological TNM stage were classified according to their clinical TNM stage.The hazard ratio of 5-year risk of relapse in 1980-1986 and 2003-2004  versus 1972-1979 was adjusted for age at diagnosis, tumour size, nodal status and tumour histology using Cox-proportional hazard modelling.
The SAS system was used to perform the analyses (release 9.2; SAS Institute Inc., Cary, NC, USA).

Discussion
In among 8,877 Dutch breast cancer patients treated with curative intent, this study showed a decrease in 5-year risk of relapse from 38% in 1972-1979 and 35% in  18 .We decided not to adjust for treatment for several reasons.First, adjustment for stage includes part of the treatment effect.Furthermore, not only frequency of use changed, but the treatments themselves have changed over the years as well.In addition, we expect an interaction between therapy and tumour stage.Unfortunately, we were not able to consider other important prognostic factors, like the mitotic activity index 19,20 , tumour grade, hormonal receptor status, HER2 and resection margin.
The differences in histology between the periods may reflect changes in pathological assessment and classification rather than true changes.The period effect (HR=0.38) after adjustment for tumour characteristics and patient's age therefore includes the issues described above.
The 5-year risk of locoregional recurrence as first event decreased from 24% in the 1970s to 14% in the 1980s and was only 4% in 2003-2004.This trend was also observed in previous Dutch studies 1,2 .We can therefore conclude that current treatment, despite the increased use of breast conserving therapy, has resulted in a good local control.In the late 1980s, a positive resection margin was identified as an important risk factor for local recurrence.This had led to more careful breast imaging before surgery, routine pathological assessment of the resection margins and re-excision in case of a positive margin 21 .In addition, adjuvant treatment 13 and a lower tumour stage 16 were associated with a decreased risk of locoregional recurrence.Breast cancer treatment is currently changing from maximal tolerable to optimally needed 22 .We recognize the importance of preventing overtreatment, yet, it is important to monitor relapse rates closely to prevent underuse of (axillary) surgery or radiotherapy 23 .This impressive improvement equals a risk reduction of 1% per year.The increasing use of systemic therapy in clinical practice has, to a large extent, contributed to the improved breast cancer outcome at a population level.Especially, since this improvement was largely independent of tumour size, nodal status and age.In contrast to systemic therapy, the strategy with respect to local treatment became less aggressive.Apparently, the less aggressive local approach could safely be implemented and possibly may have been facilitated by the availability and increased use of systemic therapy.
In the Netherlands, breast cancer screening was introduced in 1989.Previous studies showed that both a more early diagnosis and the use of systemic treatment are effective in preventing breast cancer relapse 13,15,16 .Discerning the impact of changed breast cancer therapy from that of a more early diagnosis goes beyond the scope of this study.Nevertheless, we attempted to evaluate the effect of a more early diagnosis by adjusting period-specific breast cancer relapse for age at diagnosis, tumour size and nodal status.Increasing tumour size and a positive lymph node status were associated with an increased risk of recurrence.Adjustment for these factors changed the period effect only slightly.Lymph node staging has, however, changed with the introduction of the sentinel node biopsy procedure in 1993 17 .As accurate staging became increasingly important for treatment planning, pathological lymph node examination was intensified.This may explain the incomplete correction for lymph node status in the present study.Competing interests: none declared.Conflict of interests: none declared.
All authors contributed to conception and design, manuscript preparation, read and approved the final manuscript.
All authors abide by the Association for Medical Ethics (AME) ethical rules of disclosure.
further counterbalance the increasing population risk of breast cancer 27 .
The trends in breast cancer treatment observed in our study was also observed in previous Dutch studies [28][29][30][31] and in other countries 5,[32][33][34][35][36][37] .Nowadays, treatment strategies are quite similar for different countries, although the use of BCS in China was only 30% in 1994-2007 38 .Notably, radiotherapy following BCS is standard practice in the Netherlands [28][29][30] , but underused in the United States 39 and in Korea 40 , even though numerous studies have shown a survival benefit from adjuvant radiotherapy after BCS 41 .Radiotherapy after mastectomy has decreased in the Netherlands, which was in line with guideline recommendations 42 .In contrast, the use of postmastectomy radiotherapy was lower and increased over time in the United States 43 and Korea 40 .Of patients diagnosed in 2003-2004 in the Netherlands, 37% received chemotherapy and 42% hormonal therapy.Similar rates were observed in Australia 32 and Sweden 36 .
This population-and hospitalbased study reflects changes in breast cancer management in practice.We studied risk of relapse in total, and for the different sites separately, providing a complete overview of trends in relapse.A limitation of this study is that in patients diagnosed in 2003-2004, only the first occurrence of locoregional recurrence or distant metastasis was registered.Site-specific risk estimates for locoregional recurrence and distant metastasis were, therefore, underestimated in this study.Another limitation of this study is the use of only one hospital as a historical control.Findings were, however, in line with previous Dutch studies 1,2,28-30 , suggesting good representativeness of our data.

Conclusion
This hospital-and population-based study demonstrates a huge improvement in breast cancer outcome when The occurrence of distant metastases has markedly decreased between the 1980s and 2003-2004, which was also observed in Canada 5 and Sweden 6 .This decrease may be attributable to a more early detection at initial diagnosis and the increased use of systemic therapy in primary breast cancer treatment.In addition, more and more effective chemotherapeutic options have become available 13 .The risk of distant metastasis is expected to decrease further because other systemic treatment options become available, for example trastuzumab 24 , and more patients are advised to be treated with systemic therapy.As prognosis improves, the impact of more and more aggressive systemic therapy on survival time will become smaller, whereas the occurrence of long-term toxic side effects will increase.Implementation of new systemic treatment options should therefore be preceded by a careful consideration of these pros and cons.
We observed no large difference in risk of second primary breast cancer between the 1970s, 1980s and 2000s, despite the increase in breast cancer incidence in the general population.In the United States, second primary breast cancer rate decreased since 1985 3,4 .The increased risk may have been counterbalanced by the use of systemic therapy for the primary tumour 13,25 .In addition, contralateral prophylactic mastectomy (CPM) in patients at high risk of developing a second breast cancer is increasingly performed.In the United States, the CPM rate increased from 0.4% in 1998 to 4.7% in 2007 26 .In the future, increased use of systemic therapy and CPM might Cancer Registry and Dr P. Luning for the patient registration.This study was supported by a grant from the Dutch Cancer Society, The Netherlands (Grant No. KUN 2008-4086), no involvement.comparing the patients diagnosed in the 1970s with those diagnosed in the early 2000s.Moreover, since 2004 as the number of patients selected for systemic therapy has further increased, it is likely that outcome for current patients is even better.On the one hand, this implies that there may still be room to reduce the extent of locoregional breast treatment.On the other hand, however, it may be time to reconsider the indications for systemic therapy.When the benefit of systemic therapy is only 1-2% for subgroups of patients, preventing long-term toxicities in cancer survivors may become equally or more important than preventing locoregional recurrences.This is, in our opinion, an important challenge for clinical practice and future clinical trials.
. Trained registration clerks gathered data from the patients' files based on Licensee OA Publishing London 2013.Creative Commons Attribution License (CC-BY) For citation purposes: Geurts SME, van Dijck JAAM, de Vegt F, Paquay Y, Siesling S, Verbeek ALM, et al.Huge decreases in the risk of breast cancer relapse over the last three decades.OA Epidemiology 2013 Apr 19;1(1):1.

Table 1 TNM in 1977 and in 2002: pathological classification of breast cancer 1977 2002 Tumour size
11tastasis in 10 or more ipsilateral axillary lymph nodes; or in clinically apparent ipsilateral internal mammary lymph nodes in the presence of one or more positive axillary lymph nodes; or in more than three axillary lymph nodes with clinically negative, microscopic metastasis in internal mammary lymph nodes; or in ipsilateral supraclavicular lymph nodes Sources: American Joint Committee for Cancer Staging and End-Results Reporting 197710and American Joint Committee on Cancer 200211.LicenseeOA Publishing London 2013.Creative Commons Attribution License (CC-BY) For citation purposes: Geurts SME, van Dijck JAAM, de Vegt F, Paquay Y, Siesling S, Verbeek ALM, et al.Huge decreases in the risk of breast cancer relapse over the last three decades.OA Epidemiology 2013 Apr 19;1(1):1.Competing interests: none declared.Conflict of interests: none declared.All authors contributed to conception and design, manuscript preparation, read and approved the final manuscript.All authors abide by the Association for Medical Ethics (AME) ethical rules of disclosure.

Table 2 Patient, tumour and treatment characteristics of nonmetastatic breast cancer patients by period of diagnosis, n (%) Characteristics Period of Diagnosis 1972-1979 n = 133 1980-1986 n = 174 2003-2004 n = 8570
Licensee OA Publishing London 2013.Creative Commons Attribution License (CC-BY) For citation purposes: Geurts SME, van Dijck JAAM, de Vegt F, Paquay Y, Siesling S, Verbeek ALM, et al.Huge decreases in the risk of breast cancer relapse over the last three decades.OA Epidemiology 2013 Apr 19;1(1):1.Competing interests: none declared.Conflict of interests: none declared.All authors contributed to conception and design, manuscript preparation, read and approved the final manuscript.All authors abide by the Association for Medical Ethics (AME) ethical rules of disclosure.

Table 2 (Continued)
*In case the pathological status was unknown, the clinical status was taken.