Time trends in axilla management among early breast cancer patients: Persisting major variation in clinical practice across European centers

Abstract

Background We examined time trends in axilla management among patients with early breast cancer in European clinical settings.

Material and methods EUROCANPlatform partners, including population-based and cancer center-specific registries, provided routinely available clinical cancer registry data for a comparative study of axillary management trends among patients with first non-metastatic breast cancer who were not selected for neoadjuvant therapy during the last decade. We used an additional short questionnaire to compare clinical care patterns in 2014.

Results Patients treated in cancer centers were younger than population-based registry populations. Tumor size and lymph node status distributions varied little between settings or over time. In 2003, sentinel lymph node biopsy (SLNB) use varied between 26% and 81% for pT1 tumors, and between 2% and 68% for pT2 tumors. By 2010, SLNB use increased to 79–96% and 49–92% for pT1 and pT2 tumors, respectively. Axillary lymph node dissection (ALND) use for pT1 tumors decreased from between 75% and 27% in 2003 to 47% and 12% in 2010, and from between 90% and 55% to 79% and 19% for pT2 tumors, respectively. In 2014, important differences in axillary management existed for patients with micrometastases only, and for patients fulfilling the ACOSOG Z0011 criteria for omitting ALND.

Conclusion This study demonstrates persisting differences in important aspects of axillary management throughout the recent decade. The results highlight the need for international comparative patterns of care studies in oncology, which may help to identify areas where further studies and consensus building may be necessary.

The sentinel lymph node biopsy (SLNB) technique, introduced over 20 years ago [1], has had a major impact on breast cancer treatment [2] following the demonstration of its technical feasibility [3], efficacy in staging [4] and clinical safety [5,6]. Of major importance, this technique also improved patients’ quality of life, by reducing morbidity from axillary lymph node dissection (ALND) side effects [7–9] due to lower proportions undergoing the latter procedure. Monitoring the translation of innovations into clinical practice is a critical last step in translational research, which requires clinically detailed cancer registries data.

The EUROCANPlatform is a collaboration of European cancer institutions, including cancer research centers, comprehensive cancer centers and cancer organizations with the aim of advancing translational cancer research along the entire cancer research continuum by fostering research coordination, collaboration and the development of synergies between the participants [10–12]. For this study, population-based or cancer center-specific clinical cancer registries operating among EUROCANPlatform Work Package 11 – Clinical Epidemiology partners were invited to provide data for examining uptake and time trends in SLNB use and axilla management among patients with primary non-metastatic breast cancer.

Materials and methods

EUROCANPlatform Work Package 11 (WP11) is responsible for clinical epidemiology and outcome research [12]. Data were obtained from four population-based and four institute-specific clinical cancer registries. The former group includes the national database of the Danish Breast Cancer Cooperative Group (DBCG) from Denmark [13], the Norway Cancer Registry, and regional databases of the Stockholm Regional Cancer Registry in Sweden and the Eindhoven Cancer Registry (ECR) from the Netherlands, while the latter group includes the Netherlands Cancer Institute (NKI) in Amsterdam, the Institute Jules Bordet (IJB) in Brussels (Belgium), the Istituto Nationale dei Tumori (INT) in Milan (Italy), and the Womens’ Hospital at the University of Heidelberg (Germany). Data on basic patient and tumor characteristics, clinical and pathological TNM stage, type of axilla treatment and use of neoadjuvant therapy were requested from all partner registries for consecutive years with available data. In all analyses, only first invasive female breast cancer with no evidence of metastasis at diagnosis was included. Beyond the data provided, we also asked colleagues to describe clinical practice patterns in axillary management in 2014 using a short questionnaire, to which contribution was received from the National Institute of Oncology in Hungary as well.

For the comparison of patient profiles, we grouped patients into four age groups (15–49, 50–59, 60–69, 70+), four pathological tumor stage groups (pT1, pT2, pT3, pT4), and three pathological lymph node (pN) status groups (negative, positive, unknown). Age, pT stage and pN status of patients were examined in two-year periods between 2003–2004 and 2009–2010, while trends in SLNB and ALND use were examined by calendar year. For comparability, the comparisons of subgroups defined by pT and pN status and axilla management were done after excluding patients with neoadjuvant therapy, where neoadjuvant treatment status was assessed either by a corresponding variable or by the pT status prefix ‘y’.

Data from Stockholm region were available starting with 2007, while from the INT data were only available for the period between October 2011 and April 2013 and consequently could not be considered for trend analysis. Data provided from Denmark did not allow the classification of pT4 tumors, while for Sweden, pT4 tumors, which are frequently not operated, were a priori not included. Furthermore, patients with neoadjuvant treatment were not included from Denmark and Norway due to very few cases and no corresponding registry variable, respectively. For the description of lymph node status, patients with non-macro level lymph node metastasis [clearly classified as micrometastasis (Nmic) positive only, or isolated tumor cells (ITC) positive only] were considered as lymph node negative (LN−) in order to remove the effect of possible variation in the intensity of diagnostic work-up between registries and over time.

Results

The age distribution and the calendar period-specific numbers of patients with first invasive breast cancer after the exclusion of those with metastasis at diagnosis is shown for population-based and cancer center-based registries in Supplementary Figure 1a and b (available online at http://www.informahealthcare.com), respectively. Apart from Denmark, where the sequential introduction of a national level screening program from 2007 to 2010 increased the proportion of screening age patients (50–69 years) and decreased the proportion of younger and older patients, no major changes occurred in the age distribution of patients. All institute-specific populations were characterized by a lower proportion of older patients (70+ years) compared to younger age groups, and this was most striking for the NKI, where the patients aged 70 or older contributed just 7% of all patients in 2009–2010, compared to up to 30% seen in Stockholm.

The pT distribution of patients in two-year calendar periods between 2003–2004 and 2009–2010 is shown for population-based and cancer center-based registries in Supplementary Figure 2a and b (available online at http://www.informahealthcare.com), respectively (note that the data for the Stockholm region and the INT population is included for the years 2007–2010 and 2011–2013, respectively). In all populations, after taking missing values into account, 60% to slightly more than 70% of patients had pT1 tumors, while the vast majority of the remaining patients had pT2 tumors. In all populations, not more than 5% of patients had tumors with a diameter of at least 5 cm and an even smaller proportion was found to have pT4 tumors involving the chest wall and/or the skin. A moderate rise in the proportion of pT1 tumors was seen in Denmark and the NKI over time, with a corresponding decrease of pT2 tumors, while the other registry populations displayed rather stable tumor size distribution over time.

Figures 1 and 2 provide calendar period-specific distributions of pN status among patients with pT1 and pT2 tumors, respectively [patients with macrometastasis were considered as lymph node positive (LN+)]. Among patients with pT1 tumors, the large majority (∼70–80%) had macroscopically pN LN−, and this picture did not change in any major way over time, although a slightly higher proportion of LN − patients were observed in Denmark and Norway in later calendar periods. Among patients with pT2 tumors, those with LN + comprised generally 40–50% of all patients, with exceptions in some calendar periods and registries: consistently slightly more than 50% LN + were observed in Denmark, a clear downward trend in the proportion of patients with LN + tumors was seen in the NKI and only slightly more than 30% LN + were seen in the IJB in 2005–2006 and 2009–2010.

Figure 1. Lymph node status (negative “−”; positive “+”; unknown/missing “x”) among patients with first malignant breast cancer (pT1) without neoadjuvant treatment.

Figure 2. Lymph node status (negative “−”; positive “+”; unknown/missing “x”) among patients with first malignant breast cancer (pT2) without neoadjuvant treatment.

Figure 3 shows the proportion of patients with pT1 and pT2 tumors who received a SLNB and an ALND by calendar year between 2003 and 2011. In 2003, the proportion of patients undergoing SLNB with pT1 tumors ranged between 26% and 81%. Over time, SLNB utilization increased among all participants, and by 2010, between 79% and 96% of pT1 patients received a SLNB. The proportion of patients undergoing ALND varied between 75% and 27% in 2003. ALND use declined over time, but use remained heterogeneous with between 47% and 12% of patients undergoing ALND by 2010. In the INT, for the 2011–2013 period, SLNB and ALND use were 73% and 23%, respectively, for pT1 tumors.

Figure 3. Trends in SLNB and ALND use in patients with first non-metastatic breast cancer without neoadjuvant treatment between 2003 and 2011.

Among patients with pT2 tumors, SLNB use varied between 2% and 68% in 2003 and between 49% and 92% in 2010, while ALND use varied between 90% and 55% and 79% and 19%, respectively. Changes in ALND use were heterogenous, e.g. ALND use started to decline earlier in Denmark than Heidelberg, but the latter showed a very large change in clinical practice after 2007 with very rapid declines and lower levels by 2010. The highest recorded level of ALND use was found in the IJB in 2007, followed by a slow decline. The NKI showed the strongest decline, while ALND use generally decreased in a constant and moderate way elsewhere. In the INT, for the 2011–2013 period, SLNB and ALND use were at 51% and 54%, respectively, for pT2 tumors.

Supplementary Table 1 (available online at http://www.informahealthcare.com) summarizes the results of the questionnaire on current clinical axillary management. In 2014, the most important differences existed in the management of patients with micrometastasis as well as of patients with early breast cancer (pT1-pT2) with not more than two LN + who were receiving breast conserving surgery and are scheduled for post-operative radiation therapy (ACOSOG Z0011 criteria [14]). Among patients with micrometastatic disease only, ALND is carried out in IJB, while in KI a trial is underway. In Denmark, ALND is offered in this group when three or more lymph nodes are found to have micrometastatic lymph node infiltration. The ACOSOG Z0011 criteria have been incorporated into clinical practice in Heidelberg since 2010 September and since 2013 in the INT, although with slight differences and more exceptions in the former. A trial is underway in KI, while implementation was not yet considered in Denmark, the IJB or in the NIO in Hungary in 2014.

Discussion

In this translational study of axillary management among patients with first malignant breast cancer among EUROCANPlatform participants, we found considerable differences both in the timing of uptake of the SLNB technique as well as persisting variation in ALND use, despite overall small between-center differences in baseline tumor characteristics. Overall, during the study years, a high degree of alignment in clinical practice has taken place in the use of SLNB for pT1 tumors, while variation in SLNB use for pT2 tumors has remained rather large until 2010. With increases in SLNB use, proportions of patients undergoing ALND generally declined, yet with rather substantial variation, particularly for pT2 tumors, for which between-participant differences in use actually increased between 2003 and 2010, suggesting substantially lower degree of alignment in clinical practice during the study period. In 2014, based on the questionnaire responses, differences in clinical patterns of care existed regarding the treatment of patients with micrometastasis and particularly in variance of ALND use among patients satisfying the ACOSOG Z0011 trial criteria [14].

The actual magnitude of variation in clinical practice in ALND use is clearly demonstrated by data from the years 2009–2010, when around 20% of pT1 patients had macroscopically LN + in all centers, yet ALND use varied between 12% and 47%. Furthermore, for pT2 tumors, the highest ALND use was seen in the IJB, where proportions of macrometastatic LN + patients were lower than at most other sites. Finally, among patients in Denmark, ALND use was around 20% units higher than in Heidelberg, the ECR, Stockholm or Norway, while the proportion of patients with LN + disease was only around 10% units above the levels seen in the same group of other registries. These patterns clearly suggest substantial variation in the clinical practice of ALND use among in 2009–2010.

Previous studies of breast cancer clinical care as well as patient outcomes have demonstrated variation between countries [15] and for specific patient populations [16]. These variations might be caused by differences in national treatment guidelines or by the implementation of the guidelines. In 1993, SLNB was first described as an alternative method for axillary staging [17] and, in the following years, SLNB as an alternative for ALND in clinically node negative patients was discussed. In 2003, the St. Gallen expert consensus stated that ALND can be avoided in case of a negative SLNB [18]. According to national guidelines, use of SLNB in pT1 and pT2 clinical node negative breast cancers was recommended during the whole study period in the Netherlands, Sweden, Italy, and Hungary. The Norwegian Breast Cancer Group recommended SLNB as standard procedure in 2013. In Denmark, prior to 2006, SLNB was only recommended for unifocal tumors ≤40 mm and, prior to 2012, all patients with micrometastasis or with ITCs in the lymph node were offered ALND. In Germany, the guideline of the German Cancer Society from 2004 still recommended ALND as a standard procedure [19], which was changed in the revision in 2008 [20]. In the annually updated treatment recommendations from the German Gynecological Oncology Group [Arbeitsgemeinschaft Gynäkologische Onkologie (AGO)], SLNB was first recommended in 2006 [21]. In Belgium, the first national guideline was published in 2013, but local guidelines at the IJB considered SLNB before 2010 only if the tumor size was suspected to be in the range of 20–22 mm.

Thus, while the national guidelines are, to a large extent, comparable, some observed differences might at least be partly explained by differences in the guidelines. For example, the observed higher proportion of ALNDs in Denmark, the late increase of SLNB use in Heidelberg and the low proportion of SLNBs in the IJB might partly be based on differences in the guidelines. Furthermore, use of SLNB was first recommended to be performed only by experienced surgeons and, thus, implementation of SLNB in clinical practice depended on the training of the surgeons. To our knowledge, an organized nationwide implementation of SLNB was conducted in Denmark only [22]. Thus, the observed variations are likely to reflect mostly the variation in the interpretation of currently available evidence and the implementation of the guidelines in clinical practice. Notwithstanding these variations, the proportion of patients undergoing ALND might be expected to decrease in the future, which may further contribute to avoiding arm lymph edema [8] as well as contribute to lowering the cost of care [23].

Despite the overall similarity of the national guidelines, several breast cancer related issues are controversial among individual oncologists [24], including aspects of axillary management [25,26]. For example, the use of ALND in patients with sentinel node micro- or macrometastasis is still unclear. Results from our questionnaire on clinical practice in 2014 also show differences in treatment recommendations, especially with respect to treatment of patients with micrometastasis and of patients satisfying the ACOSOG Z0011 trial criteria. Thus, more studies are needed to clarify in detail when it is safe to avoid ALND in breast cancer patients. While two large Swedish studies on the use of ALND in patient with sentinel node micro- [27] and macrometastasis [28], respectively, are currently been conducted, first results from these studies cannot be expected before 2017 and 2020, respectively. Collaborations among European centers and population-based registries could help to provide evidence to open questions more rapidly, as joint prospective studies could achieve sufficient sample sizes faster. For example, the Swedish randomized trial investigating whether ALND is necessary in patients with macrometastatis (SenOMac, [28]) is currently recruiting patients and is open for inclusion from all countries in order to achieve the required sample size of 3700 patients faster. Furthermore, retrospective studies on already available treatment and outcome data from European cancer centers and population-based cancer registries could provide first evidence for the evaluation of axilla management for specific patient groups.

Although several previous studies have examined patterns of care using population-based samples extended with additionally collected clinically detailed data [15,29], this is the first European level study that used routinely available clinical cancer registry data. Strengths of the study include the presence of high quality population-based [30–32] as well as institutional clinical cancer registry data, generally high completeness of data on tumor size and lymph node status (as evidenced by low and mostly negligible proportions with missing data in Figures 1 and 2) as well as axillary treatment, which overall suggest that the results obtained here truly reflect the actual patterns of care in the registries and time periods examined in this study.

Furthermore, the examination of concurrent patterns of care increased the timeliness of the study beyond the years for which data were available. A notable limitation of the study is the fact that not all EUROCANPlatform WP11 participants could take part in this study due to the lack of available relevant data or other administrative barriers. Furthermore, several data items, including clinical TNM, an important determinant of initial axillary management, were not available in sufficient completeness across the registries to be considered in the analysis. We considered lymph node status without micrometastatic and ITC status, and found that populations were largely comparable. The assessment of micrometastatic and ITC status, in themselves possibly influenced by the intensity of pathologic work-up, could possibly contribute to and explain some of the variation in clinical practice identified in this study. Further research may be warranted to identify how pathological practice may impact patterns of care. Patients with neoadjuvant therapy were not included in the study data from Denmark due to the very limited use of neoadjuvant therapies in the study period, while Norway did not have available data on this in the registry. While both tumor size and lymph node status may be altered by neoadjuvant therapy, SLNB has been shown to be feasible and accurate among patients who have received neoadjuvant therapy [33], and it appears unlikely that the consistent and large differences in ALND use, which persisted between the two countries during the entire study and also after SLNB use has aligned for both pT1 and pT2 tumors would be explained by this factor.

In conclusion, this study found considerable differences in the time trends of axillary management both among leading European comprehensive cancer centers and populations covered by detailed high quality cancer registries. The results underline the need for international comparative studies in patterns of care in oncology, the prominent benefit of detailed clinical cancer registry data for this purpose, but also the need for efforts to improve general data availability and standardization in this area. Efforts to improve institutional, regional and national clinical cancer registration, conducting clinical epidemiological studies using routine registry data, and the coordination of these efforts on European and local levels remain a high priority.

Acknowledgments

We thank all the contributing registries for their efforts to collect and process detailed clinical and for providing data for this study.

Disclosure statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Funding information

This work was supported by the FP7 program of the European Commission as part of a EUROCANPlatform project (grant number 260791).