Axillary lymph node ratio and total number of removed lymph nodes: predictors of survival in stage I and II breast cancer

Transcription

1 EJSO 2002; 28: 481±489 doi: /ejso , available online at on 1 Axillary lymph node ratio and total number of removed lymph nodes: predictors of survival in stage I and II breast cancer B. C. H. van der Wal, R. M. J. M. Butzelaar, S. van der Meij and M. A. Boermeester* Department of Surgery, St. Lucas-Andreas Hospital, Amsterdam, The Netherlands Aims: Presence of axillary lymph node metastases is considered the most important prognostic factor for breast cancer survival. In a period of increasing popularity for the sentinel node procedure, clarity about the possible relation between axillary dissection and survival is essential. This study investigated whether the total number of removed lymph nodes and the ratio of invaded/removed lymph nodes (lymph node ratio (LNR) would prove to be independent prognostic factors for survival. Methods: Data from 453 consecutive patients with stage I or II breast cancer were studied retrospectively. The total number of removed lymph nodes and the LNR were analysed for their prognostic value in comparison with known prognostic factors. Results: Node-negative patients with, 14 lymph nodes removed had a 10 year survival of 79% compared with 89% in patients with 14 lymph nodes removed (P ˆ 0.005). The 10 year survival for patients with an LNR 0.2 was 52%, compared with 73% for patients with an LNR, 0.2 (P ). A Cox proportional hazards model showed that, for node-negative patients, only age and total number of removed lymph nodes were significant prognostic factors. For node-positive patients, age, total number of removed lymph nodes and the LNR were significant risk factors for survival outcome. The LNR was also significantly associated with the presence of distant metastases during follow-up (hazard ratio 3.56, range 1.63±7.77). Conclusions: In stage I and II breast cancer, a favourable prognosis was found for node-negative patients with 14 removed lymph nodes. Before axillary lymph node dissection with its well-defined survival prognosis is replaced by less invasive staging methods, long-term survival using new staging techniques needs to be defined. For node-positive patients, the LNR proved to be an excellent predictor for survival outcome or development of metastatic disease. Selection of lymph node-positive patients based on the LNR may guide specific adjuvant treatment choices. # 2002 Elsevier Science Ltd. All rights reserved. Key words: breast carcinoma; metastases; axillary dissection; lymph node ratio; survival. INTRODUCTION Multimodality therapy is the standard treatment for the majority of primary breast carcinoms. Because of the established beneficial effects of adjuvant radiotherapy, chemotherapy or hormonal therapy and its increased use in women with breast cancer, one may argue that knowledge of the axillary lymph node status is no longer needed. 1 However, to date there is no better prognostic Correspondence to: MA. Boermeester MD, St Lucas Andreas Hospital, Department of Surgery (C4), Jan Tooropstraat 164, 1061 EA Amsterdam, The Netherlands. Tel.: ; Fax: ; m.a.boermeester@amc.uva.nl * Present Affiliation: Academic Medical Centre, Department of Surgery, Amsterdam, The Netherlands. marker than lymph node involvement to recognize metastatic potential of breast cancer. 2,3 The presence of lymph node metastases decreases 5 year survival by approximately 40%. Appropriate treatment may not be given if axillary staging is inadequate and tumour stage is underestimated. 4 The number of lymph nodes needed for accurate staging of the axilla is still controversial. 5,6 Clearly, the accuracy of staging increases with the number of lymph nodes resected. 7 Besides the importance of axillary nodal staging, axillary lymph node dissection (ALND) provides excellent local disease control. 8 However, it has been associated with considerable morbidity including, pain, infection, seroma formation, restricted shoulder mobility and sometimes lymphoedema. 9 Overall, ALND is only justified when it 0748±7983/02/$35.00 # 2002 Elsevier Science Ltd. All rights reserved.

2 482 B. C. H. VAN DER WAL ET AL. has therapeutic implications and/or improves local control. The potential beneficial effects of ALND on overall survival are controversial. Recently, it has been found for node-negative patients that removing 15 nodes or more is associated with an improved overall and disease-free survival. 10 This finding is supported by four large nodenegative trials showing decreased regional recurrence rates in ALND-treated patients. 11±14 Whether this translates into a survival benefit is uncertain. In contrast, Camp et al. have noted a disadvantage in survival for node-negative patients with more than 20 axillary nodes removed. 15 In a period of increasing popularity of the sentinel lymph node biopsy, clarity is needed about the therapeutic effects of axillary dissection. Also, immunocytochemical detection of micrometastases in bone marrow has been proposed as a possible alternative to nodal dissection. 16 The removal of very early nodal metastases (micrometastases) may have clinical significance for loco-regional disease control or mortality in these patients compared with true node-negative patients. 17 Breast cancer is known for its possible late metastatic manifestations even after 10 years or more. Therefore, long-term follow-up should be known from new staging and survival parameters before ALND is abandoned. Numerous studies have been undertaken to identify clinical or pathological prognostic indicators of breast cancer survival in node-positive patients. Patients with four or more positive axillary lymph nodes have a dismal prognosis compared with those with one to three positive lymph nodes. 2,3 This subdivision is hampered by a large variety in the number of harvested lymph nodes during axillary dissection. Partially, this may be attributed to the extent of the axillary dissection. This problem may be reduced by assessing the number of invaded lymph nodes in the context of the total number of removed lymph nodes, i.e. the lymph node ratio (LNR) Recently, this parameter has been proven a reliable predictor of survival outcome in patients with oesophageal cancer. 18 In node-positive patients, the discussion is ongoing whether the extent of the ALND (level I and II or level I±III) or the number of removed lymph nodes is related to breast cancer survival. In this retrospective study, we examined the relationship between survival and (1) the total number of removed lymph nodes and (2) the ratio of invaded/removed lymph nodes in patients with stage I or II breast cancer. METHODS Patients All 453 patients with stage I or II breast cancer and complete follow-up, treated at the St Lucas Andreas Hospital in Amsterdam between 1980 and 1997, were included in this series. Predominantly during the last 7 years, an increasing number of patients with small lesions, detected by the national breast cancer screening programme (mammography every 2 years for patients aged. 50 years), were referred to our hospital because of its special outpatient clinic with 1 day work-up of newly found breast lesions. Hospital databases and medical records were reviewed for all patients. Tumours were categorized using the tumour, node, metastasis (TNM) system of the American Joint Committee on Cancer. 19 Patients underwent breast-conserving surgery combined with radiotherapy (50 Gy in 25 fractions in 5 weeks) or a modified radical mastectomy (Madden procedure), depending on tumour size and patients' preferences. A complete ALND (level I±III) was performed in all patients. Pathological characteristics of the primary tumour (size, histology, radicality of the resection, vascular invasion, oestrogen receptors) and the axillary lymph nodes (lymphatic invasion, total number of lymph nodes, extracapsular invasion) were determined on routine haematoxylin and eosin slides. During the study period, microscopic examination of the axillary specimens was performed by only two dedicated, experienced pathologists. For each ALND, the LNR was calculated. The LNR is the ratio of the number of tumour-involved lymph nodes to the total number of removed lymph nodes. 18 The primary endpoints were overall survival, disease-free survival and metastasis-free survival. Overall survival was calculated as the length of time until death, irrespective of the cause. Disease-free survival was defined as the duration of survival without confirmed loco-regional recurrence or distant metastases. Statistical analysis The study was designated to determine the prognostic importance of the LNR (for node-positive patients) and the total number of removed lymph nodes (for nodenegative patients) in addition to known prognostic factors in relation to disease recurrence or survival. Univariate analyses of survival were carried out by the method of Kaplan and Meier 20 and compared with the use of the log-rank test. The Cox proportional hazards model was used to calculate hazard ratios and 95% confidence intervals. 21 A model was obtained with stepdown variable selection in which all prognostic factors with univariate P, 0.1 were initially entered into the model. All factors were treated as simple categorical variables. Survival curves were obtained using the Cox proportional hazards model, depicted as proportional survival. All reported P values are two sided, and a P value, 0.05 was considered statistically significant. All statistical analyses were performed using the commercial statistical package SPSS 9.0.

3 PROGNOSTIC VALUE OF AXILLARY LYMPH NODE RATIO 483 RESULTS Patient characteristics and pathology data All 453 evaluated patients with stage I and II breast cancer had a complete follow-up until January The patients were seen on a regular basis during follow-up with yearly routine mammography and at least once a year a physical examination. The follow-up was continued during lifetime and was mainly limited by death. The median follow-up time was 6.1 years (range 0±19.6 years). Mean age at surgery was 65.6 years (range 29±92 years). Of the study population, 301 patients (66.4%) underwent modified radical mastectomy and 152 patients (33.6%) underwent breast-conserving therapy (including lumpectomies with ALND and radiotherapy). In the period 1980±1989, 153 patients were operated of which 74.5% underwent a Madden procedure and 25.5% breast-conserving therapy. Of the 300 patients operated in the period 1990±1997, 62.3% underwent a Madden procedure and 37.7% breastconserving therapy. There was no overall significant difference in survival (10 year survival for Madden group 73.1% + 2.7% and for the breast-conserving group 78.4% + 5.2%; P ˆ 0.3). In addition, no differences were found when a comparison was made between the two procedures within each time period. Among the 453 patients, there were 296 nodenegative and 157 node-positive patients. The median total number of lymph nodes (LNN) dissected was 14 (range 1±26). In 24 patients (24/453 patients, 5.3%), the total number of removed lymph nodes was not specified. In the six node-positive patients with unspecified total number of lymph nodes, all lymph nodes were confluent and tumour involved. The lymph node status was retrieved in all 453 patients. The median LNR of nodepositive patients was 0.20 (range 0.04±1). In nodenegative patients, a comparison was made between two subgroups based on the median number of harvested lymph nodes:, 14 or 14 LNN. In node-positive patients, a similar analysis was made according to the median LNR, distinguishing patients with a LNR, 0.2 and those with a LNR 0.2. A ductal carcinoma was found in 363 patients (80.1%) and a lobular carcinoma was found in 52 patients (11.5%). The oestrogen receptor status was negative in 18.8% (85/453), intermediate in 5.3% (24/453) and positive in 47.9% (217/453) of patients (Table 1). Oetstrogen receptor status was unknown in 28% of all patients and in 26.7% of node-negative patients. Adjuvant therapy was recommended to patients according to the guidelines of the Comprehensive Cancer Centre, Amsterdam. Seventy-four patients underwent no subsequent adjuvant therapy, 140 patients underwent radiation therapy only, 107 patients underwent radiation therapy combined with tamoxifen, 18 patients underwent radiation therapy combined with chemotherapy, 87 patients received tamoxifen only and three patients underwent chemotherapy only. Four patients underwent radiation therapy combined with both chemotherapy and tamoxifen and one patient underwent radiation therapy combined with oophorectomy. Patients were given tamoxifen for at least 2 years and the chemotherapy regimen consisted of CMF. Only node-positive ( 4 LNN), pre-menopausal women were given chemotherapy during the study period. The majority of patients received radiotherapy in the context of breast-conserving therapy. Importantly, the two node-negative subgroups (, 14 or 14 LNN removed) were comparable with respect to adjuvant therapy received, age, tumour size, operation type and oestrogen receptor status. Also, similar distributions of these parameters were found for the node-positive patients with a LNR, 0.2 compared with those with a LNR 0.2. Survival rates and tumour- or axillary derived predictive factors The overall 5 and l0 year survival rates for all 453 patients were 85% + 2% and 75% + 3%, respectively. Disease-free survival rates were 87% + 2% and 72% + 3% after 5 and 10 years respectively. In overall univariate analysis, age (P ), vascular invasion (P ˆ 0.003), tumour size (P ˆ ), axillary lymph node metastases (P ), extracapsular invasion (P ) and LNR (P ) were associated significantly with survival rates (Table 1). For T1a tumours, a 5 year survival of 93% was found, which was slightly higher than a 5 year survival rate of 90% for T1b/c tumours. When all 453 breast cancer patients were considered, the survival rates were comparable in patients with, 14 LNN and those with 14 LNN removed (Table 1). In 296 node-negative patients, it was shown that age (P ), vascular invasion (P ˆ 0.04) and the total number of harvested lymph nodes (P ˆ 0.005) were significantly associated with survival (Table 2). For node-negative patients with 14 LNN resected, 5 and 10 year survival rates were significantly higher compared with those with, 14 LNN resected (96% and 88% vs 87% and 73% respectively; P ˆ 0.005; Fig. 1A). Comparable results were found for distant metastasis-free survival in these groups (Fig.1B). Also, the 5 and 10 year disease-free survival rates for node-negative patients with 14 LNN resected were significantly higher than in those with, 14 LNN resected (94% and 88% vs 88% and 75% respectively; P ˆ 0.01). In node-positive patients, a newly introduced axillary parameter, i.e. LNR, was examined in relation to breast cancer survival. Patients with an LNR 0.2 had a 5 year survival of 66%, which was significantly lower than that of patients with an LNR, 0.2 (81%). The 10 year survival with an LNR 0.2 was 52% compared with 73%

4 484 B. C. H. VAN DER WAL ET AL. Table 1 Univariate analysis of predictive factors for survival in 453 patients with stage I and II breast cancer Prognostic factor Number of patients 5 year survival (%) 10 year survival (%) P value Patient Age (at operation), 50 years ±70 years > 70 years Operation Type of operation 0.2 Lumpectomy ALND Madden Radical resection 0.3 Yes No Tumour features Tumour type 0.7 Ductal carcinoma Lobular carcinoma Mixed type Other 6 ND ND Vascular invasion No Yes Oestrogen receptor 0.8 Negative Intermediate Positive Tumour stage I IIa IIb Axillary lymph node features Lymph node metastases No Yes Extracapsular invasion NA (node-negative patients) No (node-positive patients) Yes LNR LNR ˆ LNR, LNR Number of lymph nodes removed (all patients) 0.9 Number, Number Unknown Plus±minus values are means standard error. ND denotes not done. NA denotes not applicable. The log-rank test was used to calculate P values. with an LNR, 0.2 (P0.0001; Fig. 2A). Comparable differences in survival curves between these groups were found with respect to distant metastasis-free survival (Fig. 2B). Also, 5 and 10 year disease-free survival rates were significantly different, between patients with LNR 0.2 and those with LNR, 0.2 (73% + 5% and 50% + 8% vs 86% + 4% and 61% + 7% respectively; P ˆ 0.03). It is of note that in both LNR subgroups a comparable percentage of patients had more than 14 lymph nodes removed (68.7% and 59.5%, respectively; P ˆ NS). This is important because, in node-positive patients, long-term survival decreased with the number of LNN resected. For node-positive patients with 14 LNN resected, 5 and 10 year survival rates were 69% + 5% and 57% + 6% respectively, compared with 83% + 5% and 76% + 7% respectively for those with

5 PROGNOSTIC VALUE OF AXILLARY LYMPH NODE RATIO 485 Table 2 Univariate analysis of predictive factors for survival in 296 node-negative patients with breast cancer Prognostic factor Number of patients 5 year survival (%) 10 year survival (%) P value Patient Age (at operation), 50 years ±70 years > 70 years Operation Type of operation 0.6 Lumpectomy ALND Madden Radical resection 0.2 Yes No Tumour features Tumour type 0.7 Ductal carcinoma Lobular carcinoma Mixed type 19 ND ND Other 4 ND ND Vascular invasion 0.04 No Yes Oestrogen receptor 0.1 Negative Intermediate 17 ND ND Positive Tumor stage 0.06 I II Axillary lymph node features Number of lymph nodes removed Number, Number Unknown , 14 LNN resected (P ˆ 0.04). Furthermore, the distribution of patients with 4 tumour-involved lymph nodes among the two groups with a total number of 14 or, 14 LNN removed was examined. Importantly, 38.5% of node-positive patients with 14 LNN removed had 14 tumour-involved LNN compared with 25.4% in those with, 14 LNN removed (P ˆ 0.001). Multivariate analysis of prognostic factors for survival A Cox proportional hazards model showed in the total study group that age (, 50 years vs > 70 years, hazard ratio 3.72), presence or absence of extracapsular invasion in lymph nodes (hazard ratio 2.74) and vascular invasion of the tumour (hazard ratio 1.69) were significant predictive factors of survival (Table 3). For node-negative patients, only age and the total number of resected lymph nodes were independent predictors of survival outcome (Table 4). Patients older than 70 years had an increased mortality risk of 4.07 times that of patients younger than 50 years. Importantly, node-negative patients with 14 resected axillary LNN had a decreased mortality risk (hazard ratio 0.40) compared with that of patients with, 14 resected LNN (Table 4, Fig. 1A). All other potential predictive factors were not significantly associated with an increased or decreased risk of death. For node-positive patients, age and in particular the total number of removed LNN and the LNR were independent prognostic factors for survival outcome. For the axillary parameters in these node-positive patients, it was found that 14 removed axillary LNN translated into an increased risk of death (hazard ratio 2.46; Table 5). Also, an LNR 0.2 was associated with a 2.1 times higher risk of death compared with an LNR, 0.2 (Table 5, Fig. 2A). Tumour size, type of operation, vascular invasion, radical resection and nodal extracapsular invasion did not significantly predict mortality risks in node-positive patients.

6 486 B. C. H. VAN DER WAL ET AL. Figure 1 (A) Percentage overall survival and (B) percentage distant metastases-free survival in node-negative breast cancer patients in relation to the number of removed lymph nodes, 14 or 14 (Cox proportional hazards curves). The numbers in the box refer to the numbers of patients at risk. LNN, lymph nodes. Figure 2 (A) Percentage overall survival and (B) percentage distant metastases-free survival in node-positive breast cancer patients in relation to the LNR, 0.2 or 0.2 (Cox proportional hazards curves). The numbers in the box refer to the numbers of patients at risk. Multivariate analysis of prognostic factors for distant organ metastases In the node-negative group, only the total number of resected lymph nodes was significantly predictive for the development of metastases to distant organs. When 14 lymph nodes had been removed the chance of developing distant metastases was significantly lower (hazard ratio 0.41) compared with, 14 removed lymph nodes (Table 6). Other parameters such as type of operation, vascular invasion, tumour size, age at operation and radicality of the resection were not significantly associated with development of distant metastases. Contrasting the results in node-negative patients, the chance of developing distant metastases increased when more than 14 LNN had been removed in node-positive patients (hazard ratio 6.75). Also, the lymph node ratio (LNR 0.2; hazard ratio 3.56) and vascular invasion of the tumour (hazard ratio 3.88) proved to be independent predictive factors for development of distant metastases during follow-up (Table 6). DISCUSSION In our series, patient age, extracapsular invasion, tumour size and vascular invasion were predictive for survival outcome. In proportional hazards analysis of the 296 node-negative patients, the total number of removed lymph nodes and age both independently predicted disease outcome. Removal of more than 14 lymph nodes was associated with a higher survival rate and a decreased development of distant metastases. The total number of lymph nodes being a predictor of

7 PROGNOSTIC VALUE OF AXILLARY LYMPH NODE RATIO 487 Table 3 Significant predictive factors for survival in a Cox proportional hazards Analysis of 453 patients with stage I and II breast cancer Table 5 Significant predictive factors for survival in a Cox proportional hazards analysis of 157 node-positive patients with breast cancer Prognostic factor Hazard ratio (95% CI) * P value y Prognostic factor Hazard ratio (95% CI) * P value y Age, 50 years ±70 years 1.61 (0.80±3.25) 70 years 3.72 (1.83±7.56) Extracapsular invasion NA (node negative) 1.00 No (node positive) 1.35 (0.76±2.41) Yes 2.74 (1.73±4.34) Vascular invasion, 0.05 Yes 1.69 (1.01±2.82) All other univariate significant predictive factors were rejected in this model obtained with stepwise variable selection. CI denotes confidence interval. * Hazard ratios less than 1.00 represent a decreased risk of death, whereas hazard ratios greater than 1.00 represent an increased risk of death { P values result from the hypothesis that the hazard ratio as determined by a Cox proportional hazard analysis equalled 1.0. Age 0.01, 50 years ±70 years 1.93 (0.79±4.73) > 70 years 3.49 (1.41±8.61) 14 lymph nodes removed, 0.01 Yes 2.46 (1.28±4.75) LNR, , (1.20±3.66) All other univariate significant predictive factors were rejected in this model obtained with stepwise variable selection. CI denotes confidence interval. * Hazard ratios less than 1.00 represent a decreased risk of death, whereas hazard ratios greater than 1.00 represent an increased risk of death. { P values result from the hypothesis that the hazard ratio as determined by a Cox proportional hazard analysis equalled 1.0. Table 4 Significant predictive factors for survival in a Cox proportional hazards analysis of 296 node-negative patients with breast cancer Prognostic factor Hazard ratio (95% CI) * P value y Age , 50 years ±70 years 1.05 (0.41±2.71) 70 years 4.07 (1.56±10.6) 14 lymph nodes removed 0.03 Yes 0.40 (0.20±0.79) All other univariate significant predictive factors were rejected in this model obtained with stepwise variable selection. CI denotes confidence interval. * Hazard ratio less than 1.00 represent a decreased risk of death, whereas hazard ratios greater than 1.00 represent an increased risk of death. { P values result from the hypothesis that the hazard ratio as determined by a Cox proportional hazard analysis equalled 1.0. survival may be explained by a more accurate staging when an extensive dissection is applied. Alternatively, removal of nodal micrometastases may improve survival. This is in agreement with recent findings that occult disease may be a predictor of recurrence for the 25% of lymph node-negative patients that develop recurrences. 22 In our series up to 1997, no additional Table 6 Significant predictive factors for occurrence of distant metastases in a Cox proportional hazards analysis of patients with stage I and II breast cancer Prognostic factor Hazard ratio (95% CI) * P value y Node-negative patients 14 lymph nodes removed 0.03 Yes 0.41 (0.41±0.82) Node-positive patients 14 lymph nodes removed Yes 6.75 (2.42±18.8) Vascular invasion Yes 3.88 (1.71±8.81) LNR, 0.005, (1.63±7.77) All other univariate significant predictive factors were rejected in this model obtained with stepwise variable selection. CI denotes confidence interval. * Hazard ratios less than 1.00 represent a decreased risk of death, whereas hazard ratios greater than 1.00 represent an increased risk of death. { P values from the hypothesis that the hazard ratio as determined by a Cox proportional hazard analysis equalled 1.0.

8 488 B. C. H. VAN DER WAL ET AL. cytokeratin staining was used and the lymph node status was assessed by conventional methods only. ALND is useful in patients with breast cancer when accurate staging is needed for adjuvant treatment planning. Also, patients may benefit from ALND by a reduced risk of regional recurrence or a reduced mortality risk. The NSABP B-04 trial has found no survival benefit from prophylactic ALND in nodenegative patients. However, a recent meta-analysis has shown that ALND resulted in a survival advantage of 5.6% in stage I disease. 23 In lymph node-negative patients, sentinel node biopsy procedures have shown promising results. The sentinel node can be easily located in the majority of patients. Nearly all studies reach detection rates of 90±95% or even higher. 24 On the contrary, Krag et al. showed that identifying the sentinel lymph node varies among experienced surgeons from 79% to 98%. 25 The pathological evaluation of the sentinel node is still controversial. It has been described that in 5±32% of cases the sentinel node were falsely negative during intraoperative staining. 23 Weaver et al. showed that on pathological review additional metastases were detected in 10% of sentinel lymph nodes after examining serial sections and cytokeratin immunohistochemical stains. 22 Before completely abandoning axillary dissection in lymph node-negative patients, clarity is needed about the therapeutic effects of axillary dissection. When in node-positive patients the number of tumour-invaded lymph nodes was considered in the context of the total number of harvested lymph nodes, i.e. the lymph node ratio, this ratio appeared to be significantly predictive for disease outcome. The mortality risk was 2.1 times higher for the lymph node ratio 0.2 compared with, 0.2. Contrasting the results of the node-negative patients, it was found for nodepositive patients that 14 removed LNN was associated with a 2.46 times higher mortality risk compared with those with, 14 removed LNN. However, this was associated with tumour burden of these lymph nodes. A significant relationship to higher LNRs was found in the node-positive group with 14 removed LNN, which probably explains their dismal prognosis. Recently, Camp et al. postulated that finding a higher number of axillary lymph nodes in node-negative patients suggests hyperplasia of lymph nodes and this is predictive for aggressive disease. 15 We could not confirm their findings in node-negative patients, although we did find such an inverse relation between high numbers of lymph nodes and survival in our node-positive cohort. Earlier studies have indicated that germinal centre hyperplasia in axillary lymph nodes is associated with local suppression of cytotoxic cell activity, thereby negatively affecting survival. 26,27 This may be an alternative explanation for the worse survival outcome found in patients with a large total number of removed lymph nodes. In both node-negative and node-positive groups, patients older than 70 years had a decreased survival compared with those from 51 to 70 years. The former group had a more favourable stage distribution. For example, 30.9% of patients older than 70 years had stage I disease compared with 48.4% of patients between 50 and 70 years. This difference may be explained by the participation of the 51±70 year age group in the national breast cancer screening programme. An additional, more obvious explanation is the higher age-related mortality of patients older than 70 years compared with the group 51± 70 years. Evaluation of prognostic factors for node-positive patients may be complicated by differences in adjuvant therapy. In our series, patients received adjuvant chemo-, radio- or hormonal therapy according to the guidelines of the Regional Comprehensive Cancer Centre. All cases were discussed in a multidisciplinary working party. No significant differences were found among patient groups subdivided according to prognostic factors. The use of systemic therapy was low by current standards and may have influenced overall and diseasefree survival. However, within this study group, indications for use of systemic therapy were consistent throughout the first and second parts of the study period. In addition, overall as well as within each time period, no significant differences were found among groups, making performance bias due to differences in treatment unlikely. Overall, survival rates according to tumour stage were comparable with data from the literature. 28 Within the node-negative group or the node-positive group, a differential probability of survival could be found based on axillary lymph node characteristics. Critics of ALND claim that the overall survival may not be influenced by axillary dissection but depends on the development of distant metastases. 29 However, we found that for nodenegative patients the number of removed lymph nodes was clearly related to a better survival outcome, an effect which could not be explained by differences in therapeutic regimens. Effects of a more accurate staging or an improved local control by removal of micrometastatic nodal involvement might explain better survival rates. Definite outcome in breast cancer is known only after at least 10 years. Therefore, before risk prediction derived from axillary dissection results is replaced by less invasive staging methods, caution must be taken to define long-term survival of new staging techniques such as sentinel lymph node biopsy in randomized trials. In node-positive patients with primary operable breast cancer, an unfavourable prognosis was found for patients with an LNR 0.2. With this knowledge one can argue in favour of individual risk calculation based on the axillary parameters described in this study. Even when all node-positive patients are recommended adjuvant therapy, selection of patients based on the LNR may guide choices in duration and/or dosage and/or type of adjuvant therapy.

9 PROGNOSTIC VALUE OF AXILLARY LYMPH NODE RATIO 489 REFERENCES 1. McMasters KM, Giuliano AE, Ross MI et al. Sentinel-lymph-node biopsy for breast cancer ± not yet the standard of care. N Engl J Med 1998; 339(14): 990±5. 2. Carter CL, Allen C, Henson DE. Relation of tumor size, lymph node status and survival in breast cancer cases. Cancer 1989; 63: 181±7. 3. Ragaz J, Jackson SM, Le N et al. Adjuvant radiotherapy and chemotherapy in node-positive premenopausal women with breast cancer. N Engl J Med 1997; 337: 956± Kutiyanawala MA, Sayed M, Stotter A, Windle R, Rew D. Staging the axilla in breast cancer: an audit of lymph-node retrieval in one UK regional centre. Eur J Surg Oncol 1998; 24(4): 280±5. 5. Graversen HP, Blichert-Toft M, Andersen JA, Zedler K, and the Danish Breast Cancer Cooperative Group. Breast cancer: risk of axillary recurrence in node-negative patients following partial dissection of the axilla. Eur J Surg Oncol 1988; 14: 407± Axelsson CK, Mouridsen HT, Zedeler K. Axillary dissection of level I and II lymph nodes is important in breast cancer classification. The Danish Breast Cancer Cooperative Group (DBCG). Eur J Cancer 1992; 28A(8±9): 1415± Moffat FL Jr, Senofsky GM, Davis K, Clark KC, Robinson DS, Ketcham AS. Axillary node dissection for early breast cancer: some is good, but all is better. J Surg Oncol 1992; 51(1): 8± O'Dwyer PJ. Axillary dissection in primary breast cancer. Br Med J 1991; 302: 360±1. 9. De Kanter AY, van Eijck CH, van Geel AN et al. Multicentre study of ultrasonographically guided axillary node biopsy in patients with breast cancer. Br J Surg 1999; 86(11): 1459± Sosa JA, Diener-West M, Gusev Y et al. Association between extent of axillary lymph node dissection and survival in patients with stage I breast cancer. Ann Surg Oncol 1998; 5(2): 140± Atkins H, Hayward JL, Klugman DJ, Wayte AB Treatment of early breast cancer: a report after ten years of a clinical trial. Br Med J 1972; 20;2(811): 423± Fisher B, Bauer M, Margolese R et al. Five-year results of a randomized clinical trial comparing total mastectomy and segmental mastectomy with or without radiation in the treatment of breast cancer. N Engl J Med 1985; 14;312(11): 665± Hayward J, Caleffi M. The significance of local control in the primary treatment of breast cancer. Arch Surg 1987; 122(11): 1244± Cabanes PA, Salmon RJ, Vilcoq JR et al. Value of axillary dissection in addition to lumpectomy and radiotherapy in early breast cancer. The Breast Carcinoma Collaborative Group of the Institut Curie. Lancet ;339(8804): 1245± Camp RL, Rimm EB, Rimm DL. A high number of tumor free axillary lymph nodes from patients with lymph node negative breast carcinoma is associated with poor outcome. Cancer 2000; 1;88(1): 108± Diel IJ, Kaufmann M, Costa SD et al. Micrometastatic breast cancer cells in bone marrow at primary surgery: prognostic value in comparison with nodal status. J Natl Cancer Inst 1996; 20;88(22): 1652± Clare SE, Sener SF, Wilkens W, Goldschmidt R, Merkel D, Winchester DJ. Prognostic significance of occult lymph node metastases in node-negative breast cancer. Ann Surg Oncol 1997; 4(6): 447± Van Sandick JW, van Lanschot JJB, ten Cate FJW, Tijssen JGP, Obertop H. Indicators of prognosis after transhiatal esophageal resection without thoracotomy for cancer. J Am Coll Surg in press. 19. Fleming ID, Cooper JS, Henson DE et al. (eds). American Joint Committee on Cancer Staging Manual (5th edn). Philadelphia, PA: JB Lippincott, Kaplan EL, Meier P. Nonparametric estimation from incomplete estimation. J Am Stat Assoc 1958; 53: 457± Cox DR. Regression models and life tables. J R Statist Soc Bull 1992; 34: 187± Weaver D, Krag DN, Ashikaga T, Harlow SP, O'Connell M. Pathologic analysis of sentinel and nonsentinel lymph nodes in breast carcinoma. Cancer 2000; 88: 1099± Orr RK. The impact of prophylactic axillary node dissection on breast cancer survival ± a Bayesian meta-analysis. Ann Surg Oncol 1999; 6(1): 109± Sandrucci S, Casaiegno PS, Percivale P, Mistrangelo Bombardieri E, Bertoglio S. Sentinel lymph node mapping and biopsy for breast cancer: a review of the literature relative to 4791 procedures. Tumori 1999; 85: 425± Krag D, Weaver D, Ashikaga T et al. The sentinel node in breast cancer: a multicenter validation study. New Engl J Med 1998; 339: 941± Hunter RL, Ferguson DJ, Coppleson LW. Survival with mammary cancer related to the interaction of germinal center hyperplasia and sinus histiocytosis in axillary and internal mammary lymph nodes. Cancer 1975; 36(2): 528± Check IJ, Cobb M, Hunter RL. The relationship between cytotoxicity and prognostically significant histologic changes in lymph nodes from patients with cancer of the breast. Am J Pathol 1980; 98(2): 325± Early Breast Cancer Trialists' Collaborative Group. Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy. 133 randomised trials involving recurrences and deaths among women. Lancet 1992; 11;339(8785): 71± Haffty BG, McKhann C, Beinfield M, Fischer D, Fischer JJ. Breast conservation therapy without axillary dissection. A rational treatment strategy in selected patients. Arch Surg 1993; 128(12): 1315±9. Accepted for publication 12 November 2001