Loco-regional recurrences after mastectomy in breast cancer: prognostic factors and implications for postoperative irradiation

Loco-regional recurrences after mastectomy in breast cancer: prognostic factors and implications for postoperative irradiation

Radiotherapy and Oncology 50 (1999) 267±275 Loco-regional recurrences after mastectomy in breast cancer: prognostic factors and implications for post...

154KB Sizes 0 Downloads 2 Views

Radiotherapy and Oncology 50 (1999) 267±275

Loco-regional recurrences after mastectomy in breast cancer: prognostic factors and implications for postoperative irradiation Jos J. Jager a,*, Lex Volovics b, Leo J. Schouten c, Jos M.A. de Jong a, Pierre S.G.J. Hupperets d, Maarten F. von Meyenfeldt e, Bert Schutte f, Geert H. Blijham g a Institute for Radiation Oncology Limburg, Heerlen, The Netherlands Department of Methodology and Statistics, University of Maastricht, Maastricht, The Netherlands c Department of Registration and Epidemiology, Comprehensive Cancer Centre Limburg, Maastricht, The Netherlands d Department of Internal Medicine, Section of Hematology±Oncology, University Hospital Maastricht, Maastricht, The Netherlands e Department of Surgery, Section of Surgical Oncology, University Hospital Maastricht, Maastricht, The Netherlands f Department of Molecular Cell Biology & Genetics, University of Maastricht, Maastricht, The Netherlands g Department of Internal Medicine, University Hospital Utrecht, Utrecht, The Netherlands b

Received 31 March 1998; received in revised form 24 July 1998; accepted 10 September 1998

Abstract Purpose: Potential risk factors including DNA ¯ow cytometric-derived parameters predicting loco-regional recurrence (LRR) in early breast cancer were investigated. Materials and methods: This study included 608 patients treated by modi®ed radical mastectomy between 1982 and 1987. Recommendations regarding local treatment as well as adjuvant systemic therapy did not change during this period. Patients treated by adjuvant chemotherapy were randomized to receive additional medroxyprogesterone acetate (MPA) treatment. Only 59 (10%) patients received postoperative irradiation (XRT) to the chest wall and/or axillary lymph nodes; another 121 (20%) patients received XRT to the internal mammary nodes because of centromedially located tumours. Results: Patients were followed for a median period of 7.5 years. The event-free survival at 10 years was 50%. The cumulative incidence rate of LRR at 10 years was 18% (n ˆ 93), either with (n ˆ 30) or without (n ˆ 63) concurrent distant metastases. The chest wall, regional lymph nodes or both were involved in 41 (44%), 38 (41%) and 12 (13%) patients, respectively. Multivariate analysis according to the Cox model revealed two factors associated with LRR, i.e. pT (P , 0:05) and nodal status (P , 0:05). In node-positive patients extracapsular tumour extension (ECE) and pT were independent risk factors. DNA ploidy and S-phase fraction did not yield additional information. Based on pT, nodal status and extracapsular extension of tumour growth a high risk ( . 10%) and low risk ( , 10%) group for LRR could be identi®ed. Conclusions: Results indicate that T-stage and nodal status, combined with ECE, may help to identify patients at risk for loco-regional recurrence, whereas DNA ¯ow cytometry does not. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Breast cancer; Loco-regional recurrence; Flow cytometry; Extracapsular tumour extension

1. Introduction The rate of loco-regional recurrence is approximately 20% after modi®ed radical mastectomy (MRM). Two-thirds of the loco-regional recurrence is isolated [33]. By adding postoperative radiotherapy a two-thirds reduction can be achieved [6,9,12,19,21,33,40,45]. In spite of an improved local control and disease-free survival, a clear impact on overall survival was absent until now [9,33,42]. The debate regarding potential survival bene®t, however, is still * Corresponding author. Department of Radiation Oncology, R.T.I.L., Henri Dunantstr. 5, 6419 PC Heerlen, The Netherlands.

ongoing, supported by some evidence of reduction of secondary dissemination [3,38] and more recently after publication of two randomized studies demonstrating a clear survival bene®t of postoperative irradiation in a selected group of high risk patients [34,36]. Although postoperative irradiation is widely used, few good guidelines for its application exist. Currently, selection for postoperative irradiation is largely based on nodal status, T-stage and completeness of resection [4,5,8,13,16,17,19, 24,25,27,28,30,32,39,43,44,46±48] but in actual practice a wide variety of treatment recommendations can be found [19,20,24,25,47,48]. The present study was undertaken to investigate patterns

0167-8140/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0167-814 0(98)00118-2

268

J.J. Jager et al. / Radiotherapy and Oncology 50 (1999) 267±275

of and prognostic factors for loco-regional recurrence and to de®ne the appropriate treatment volume. The patient population had been entered in a prospective study, in which indications for postoperative irradiation were very much restricted as compared to most currently used recommendations. This allowed the assessment of risk factors for locoregional recurrences in a virtually unselected (apart from age) population of stage III breast cancer patients. Apart from irradiation of the internal mammary chain in case of centromedial tumours, only 10% of patients had received radiotherapy; moreover the analysis was done against the background of the use of anthracyclin-containing adjuvant chemotherapy for patients with positive axillary nodes. Our aim was to de®ne from these data a simple clinical decision rule regarding the selection of patients for routine postoperative radiotherapy. 2. Materials and methods 2.1. Patient population Between 1982 and 1987, patients less than 70 years of age with operable T1±3 breast cancer were entered in a prospective study on prognostic factors and treatment results (IKL8201). Within this study the node-positive patients were randomized to receive, in addition to adjuvant chemotherapy, hormonal treatment. The trial design and treatment results have been reported in a previous article [22]. The present analysis is based on the subgroup of 608 patients with operable invasive breast cancer who were treated by modi®ed radical mastectomy (MRM). Patients with clinical T1 tumours treated by breast conserving treatment (n ˆ 176) have been excluded. The study was conducted in the region of the Comprehensive Cancer Centre Limburg (IKL), The Netherlands, in which seven hospitals (one university and six community) are collaborating. All patients received irradiation at the Institute for Radiation Oncology Limburg (RTIL), the only radiotherapy facility present in the region. 2.2. Treatment Patients underwent a modi®ed radical mastectomy (MRM) with complete axillary dissection according to Patey/Madden. The surgeons were asked to do a complete axillary dissection rather than to con®ne themselves to levels I and II. In case of pathological positive axillary nodes they received adjuvant chemotherapy (CAF) irrespective of their age and were randomly assigned to concurrent high dose MPA during 6 months postoperatively [22]. Chemotherapy was started usually within 2 weeks after surgery. Except for irradiation of the internal mammary chain (IMC) in case of central or medial located tumours, the draining lymphatics were not routinely irradiated. It was recommended to irradiate the chest wall in case of nearpositive or positive resection margin or spill of tumour

cells during surgery. The chest wall was either irradiated with orthovoltage (250 kV) (n ˆ 38) to a total dose of 30 Gy with four daily fractions of 3 Gy or 5 MV photon beams (n ˆ 19) to 50 Gy with four daily fractions of 2.5 Gy a week, in both situations using tangential ®elds. Treatment was delivered and prescribed according to the ICRU29 recommendations. All patients were evaluated for disease outcome by the treating physicians at 3±6 month intervals by means of physical examination and annual mammography. Data were abstracted from the data base of the cancer registry of the IKL and if necessary, additional information was retrospectively abstracted from the medical records. LRR was de®ned as a recurrence as the ®rst event within the chest wall or the axillary, supraclavicular or the internal mammary nodes, whether or not simultaneously within a period of 2 months with distant metastases. The endpoints of the study were the occurrence of loco-regional recurrence in terms of cumulative incidence rates and event-free survival (EFS). LRR (n ˆ 63), distant metastases (n ˆ 189), both (n ˆ 30) or death due to breast cancer but not speci®ed (n ˆ 5) were determining events of EFS. Survival was calculated from the date of diagnosis. 2.3. Flow cytometry DNA histograms were obtained by ¯ow cytometry of paraf®n-embedded specimens as has been described previously [22,23,43]. Isolated nuclei from 50 mm sections were obtained from the primary tumours and adjacent 5 mm H&E stained sections were used for histological control. All specimens were analyzed on a FACS IV Cell Sorter (Becton and Dickinson, Sunnyvale, CA). The resulting DNA histograms were semi-automated analyzed. Ploidy status was expressed by the DNA index and calculated as the ratio of aneuploid and diploid G1/0 peak channel. Tumours with a single G1 peak were considered to be diploid, whereas evidence of an additional peak indicated aneuploidy. Histograms with coef®cients of variation of less than 8% were considered of good quality. The proliferative activity (SPF) was calculated by counting the number of cells between the inclination points of the descending G1 peak and the ascending G2/M peak [22]. In cases of less than 25% admixture of diploid cells, the percentage of aneuploid S-phase cells was calculated without corrections for the presence of diploid S- and G2/M-phase cells. In cases of more than 25% admixture of diploid cells overlapping in diploid and hyperdiploid histograms the percentage of S-phase cells was not calculated. 2.4. Patient characteristics All patients have been followed for a median period of 7.5 years (range 0±11 years). The mean age at presentation was 53 years (range 26±71 years). The main patient characteristics are summarized in Table 1. ER and PgR were measured by cytosol assay. The tumour was located centromedially in 191 (32%) patients and in the outer part in 304

J.J. Jager et al. / Radiotherapy and Oncology 50 (1999) 267±275

in 205 (59%) patients and not explicitly mentioned in 59 (17%) patients. DNA ploidy could be determined in 482 tumours and SPF of 306 specimens. The tumour was diploid in 169 (28%) patients and aneuploid in 313 (52%) patients. SPF ranged between 0.7% and 43.5%, with mean and median values of 10% and 8%, respectively. Adjuvant chemotherapy was actually given in 351 of the 353 nodepositive patients; within this group 169 patients were randomized for additional MPA. Postoperative radiotherapy was administered in 180 (30%) patients. The target volumes of radiotherapy are given in Table 1. The internal mammary chain (IMC) only was irradiated in 121 (20%) patients, the chest wall whether or not combined with irradiation of the IMC was irradiated in 47 (8%) patients, the chest wall with axillary and supraclavicular irradiation was irradiated in 10 (2%) patients and the regional lymph nodes were irradiated only in two (0.3%) patients. If irradiation of the IMC only is not taken into account, the rate of postoperative radiotherapy following MRM was only 10%.

Table 1 Patient characteristics (n ˆ 608) Characteristic Age (years) Mean (range) # 50 . 50 Histologic type Ductal Non-duct pT T1 T2 T3 pN N0 N1±3 N4 1 ER Unknown 0±10 . 10 PgR Unknown 0±10 . 10 Radiotherapy Intermammary Chest wall CW and IMC Loco-regional Regional lymph nodes

n (%) 53.5 (26±71) 226 (37) 382 (63) 498 (82) 110 (18) 187 (31) 381 (63) 40 (6) 255 (42) 201 (33) 152 (25) 40 (7) 160 (26) 408 (67)

2.5. Statistics

147 (24) 182 (30) 279 (46) 180 (30) 121 (20) 24 (4) 23 (4) 10 (2) 2 (0.3)

(50%) patients; the site of origin overlapped in 100 (17%) patients. The resection margin was positive in six (1%) patients, near-positive ( , 5 mm) in 51 (8%) patients and free in 551 (91%) patients. The median number of axillary nodes examined was 10 (range 0±35). Of the 353 nodepositive patients extracapsular tumour extension (ECE) was present in 95 (27%), absent in 102 (29%) and not explicitly mentioned in 149 (42%). A positive axillary node at the highest level (HPN) was found in 89 (25%) patients, absent Table 2 Loco-regional recurrence (LRR) following modi®ed radical mastectomy a Characteristic

Isolated With distant disease Chest wall Chest wall 1 lymph nodes Lymph nodes Axillary Infraclavicular Supraclavicular a

LRR (n)

63 30 41 12 38 16 12 35

269

Relative occurrence (%) In all patients

In patients with LRR

10 5 7 2 7 3 2 6

68 32 44 13 41 17 13 38

In two patients the site of recurrence was unknown.

As loco-regional recurrences, distant metastases and malignancy-related death can be considered as competing risks, survival curves were not estimated by the Kaplan± Meier method [26]. To describe the differences between different subgroups of patients, use was made of cumulative incidence curves, which have a direct probability interpretation [2]. The following variables were investigated in the univariate and multivariate analyses: age, menopausal status, lateralization, subsite, clinical stage, clinical Tstage, clinical N-stage, histology, resection margin, number of axillary nodes removed, number of positive axillary nodes, extranodal tumour extension of nodal metastases, positive node at the highest level, ER, PgR, DNA ploidy, S-phase fraction and treatment. The prognostic signi®cance of clinical parameters was evaluated using proportional hazards regression, testing the proportional hazards assumption in diverse ways and testing for undue in¯uence of extreme patients, linearity of effects and time-dependent effects [2]. The statistical package Stata was used. 3. Results 3.1. The occurrence of loco-regional recurrences At last follow-up 362 patients were alive; 301 patients were without evidence of disease, 22 patients had a locoregional recurrence, 31 patients had distant disease and eight patients had both. Two hundred forty-six patients died, 217 patients due to breast cancer and 22 patients without evidence of disease. In eight patients the cause of death was unknown. Forty-one patients died after a loco±regional recurrence, 158 patients died with distant disease only and 22 patients died with both. With a median follow-up time of 7.5 years, of the 608 patients 93 experienced a loco-regional

270

J.J. Jager et al. / Radiotherapy and Oncology 50 (1999) 267±275

Fig. 1. Event free survival (efs) and cumulative incidence of loco-regional recurrences of all (ci total Irr) and isolated loco-regional recurrences (ci isolated Irr).

recurrence with or without distant metastases as the ®rst site of failure (Table 2). Event-free survival (EFS) and the cumulative incidence rate of loco-regional recurrence are shown in Fig. 1. At 10 years the EFS was 50%. Sixty percent of the loco-regional recurrences occurred within the ®rst 3 years, with an 18% cumulative incidence rate at 10 years. A loco-regional recurrence was attributed to approximately 36% of the total number of events. Loco-regional recurrences were isolated in 63 (68%) patients and with concurrent distant metastases in 30 (32%) patients; the cumulative incidence of isolated recurrences after 10 years is 12% and is given in Fig. 1. An in-®eld recurrence was observed in four patients, all located in the chest wall. The anatomic distribution of the loco-regional recurrences is given in Table 2. The chest wall and lymph node regions were approximately equally involved. Of the regional recurrences the supraclavicular fossa was the most common site involved. Axillary recurrences were seen in only 16 patients, which is 3% of all patients and 17% of those with loco-regional recurrences. 3.2. Prediction of loco-regional recurrences in the total group of patients (n ˆ 608) A multivariate analysis according to the Cox model was performed that included age, pathological T- and N-stage,

receptor status, ploidy and SPF and treatment with locoregional recurrence as endpoints. Pathological T- and Nstage were the only factors signi®cantly associated with loco-regional recurrence. Similar results were observed when the 180 patients treated by radiotherapy (mainly parasternally) were withdrawn from the model, so all patients were pooled for the remaining analysis (Table 3). No difference was found whether or not MPA was given. An age over 50 years (HR 0.72, 95% CI 0.47±1.1, P ˆ 0:13), ER more than 10% (HR 1.1, 95% CI 0.64±1.88, P ˆ 0:74) and PgR more than 10% (HR 0.8, 95% CI 0.49±1.33, P ˆ 0:39) did not contribute to the risk signi®cantly. No difference was found between the specimens for which the resection margin was classi®ed as near-positive or positive compared to tumour-free (HR 0.99, 95% CI 0.5±1.95, P ˆ 0:98). Similar results were obtained regarding ploidy (HR 1.59, 95% CI 0.83±3.03, P ˆ 0:16) and SPF of more than 8% (HR 0.79, 95% CI 0.4±1.54, P ˆ 0:49). Interaction between relevant variables was investigated; none of them were signi®cant (for example, age versus radiotherapy P ˆ 0:44, a positive resection margin versus radiotherapy P ˆ 0:55, chemotherapy versus age P ˆ 0:25, MPA versus ER P ˆ 0:74 and MPA versus PgR P ˆ 0:84). Distribution of ploidy and SPF with respect to loco-regional recurrence is summarized in Table 4. Ploidy and SPF did not yield additional information regarding the risk of developing a loco-regional recurrence, even if analyzed in the aneuploid group only (data not shown). Cumulative incidence rates of loco-regional recurrence according to pathological T- and N-stage are shown in Figs. 2 and 3. Cumulative incidence rates at 10 years were 9%, 18% and 28% for T1, T2 and T3, respectively. Cumulative incidence rates for N0, N1±3 and N4 1 were 7%, 20% and 25%, respectively. 3.3. Prediction of loco-regional recurrences in nodepositive patients (n ˆ 353) Of the 93 loco-regional recurrences 76 occurred in the node-positive group. A Cox analysis of the node-positive patients showed similar results, pathological T-stage and the number of positive axillary lymph nodes being the only signi®cant risk factors associated with loco-regional recurrence. Again no signi®cant difference was observed when the analysis was restricted to the patients who did not

Table 3 Results of multivariate analysis of all patients Covariate pT T2 versus T1 T3 versus T1 pN N1±3 versus N0 N4 1 versus N0 a

HR a (all patients) (n ˆ 608)

95% CI

1.71 2.58

0.98±2.98 1.13±5.89

2.56 5.08

1.40±4.68 2.79±9.24

HR, hazard ratio; CI, con®dence interval; RT, radiotherapy.

P-value

HR (no RT) (n ˆ 428)

95% CI

P-value

0.06 0.03

1.69 2.82

0.92±3.10 1.09±7.28

0.09 , 0.05

, 0.05 , 0.05

1.83 3.53

0.92±3.64 1.78±6.99

0.08 , 0.05

J.J. Jager et al. / Radiotherapy and Oncology 50 (1999) 267±275 Table 4 Loco-regional recurrence and DNA ¯ow cytometry after modi®ed radical mastectomy Variable SPF (%) #8 .8 Not assessable Ploidy Diploid Aneuploid/multiploid Not assessable

n

LRR (%)

146 160 302

22 (15) 22 (14) 49 (16)

169 313 126

21 (13) 48 (15) 24 (19)

receive radiotherapy (data not shown). When extracapsular tumour extension of axillary lymph node metastases and highest positive axillary node (HPN) were introduced in the model, pathological T-stage and ECE appeared signi®cant (Table 5). The in¯uence of ECE in the node-positive patients is shown in Fig. 4. The cumulative incidence rate at 10 years was 20% in the absence of ECE and 38% in the presence of ECE. An intermediate incidence was found for the subgroup in which ECE was not assessed (data not shown). 3.4. Identi®cation of patients at high risk of loco-regional recurrence Based on the information of the Cox analysis a subgroup of patients at high risk of loco-regional recurrence could be identi®ed, i.e. all patients with four or more positive nodes or T2±3 with one to three positive axillary nodes in contrast to the remaining patients who had a relatively good prognosis with respect to the risk of loco-regional recurrence. The difference in the cumulative incidence for the high risk group at 10 years was 34% compared to 8% for the good prognostic group (Fig. 5). The in¯uence of ECE was investigated in all subgroups. Results suggest an additional value of ECE. The number of events was too small to draw ®rm

Fig. 2. Cumulative incidences of loco-regional recurrences according to Tstage. Numbers of patients at risk are under the curve.

271

conclusions as to whether or not ECE is of additional value in selecting patients at high risk in the group of patients with T1 N1±3 tumours. The crude values for the LRR rates in relation to T- and N-status as well as ECE are given in Table 6. 3.5. Difference in the risk pattern for chest wall and lymph node recurrence A separate Cox regression analysis was performed with chest wall recurrence as the endpoint. This procedure could not be performed for lymph node recurrence only due to the low number of events. Regarding chest wall recurrence a similar risk pattern was found compared to overall recurrence. Again pathological T-stage and particular pN appeared to be independent prognostic factors (Table 7). 4. Discussion The 18% 10-year rate of loco-regional recurrence in our 608 patients with T1±3, N-negative or N-positive breast cancer is in agreement with results from older studies of mastectomy patients who were not irradiated [4,9,10,19, 21,40,44,45]. The majority of these loco-regional recurrences were isolated and occurred within the ®rst 3 years following surgery. Data from radiotherapy studies indicate that this 18% can be brought down to between 5% and 10% with loco-regional radiotherapy [3,6,9,19,21,33,40,45]. This study was undertaken to identify factors that could help to select the patients who were most likely to bene®t from this added radiotherapy. In contrast to other studies this was done against the background of added adjuvant chemotherapy in node-positive patients and of radiotherapy to the non-surgically treated intermammary lymph nodes. Moreover, the population was virtually unselected except for an age of less than 70 years. The most important factors to predict loco-regional recurrence were nodal status and pathological T-status. After 10 years almost 30% of the patients with four or more positive nodes or patients with pT3 tumours will have experienced a loco-regional recurrence. In contrast, patients with negative nodes have an approximately 7% incidence of loco-regional recurrence. It appears reasonable to give routine postoperative radiotherapy to the two former groups and not to the latter group. With that, a clinical decision rule has been reached for about 70% of the patients. The data reported here were obtained against the background of all node-positive patients, whether pre- or postmenopausal, receiving anthracyclin containing adjuvant chemotherapy. Although this is common practice in some other parts of the world, in particular in the USA, in other countries the use of CMF-adjuvant therapy limited to premenopausal patients is still prominent. Available data suggest that adjuvant chemotherapy is not associated with a signi®cant decrease in the local recurrence rate after MRM [7,19,25,48]. Furthermore, in trials comparing CMF-type

272

J.J. Jager et al. / Radiotherapy and Oncology 50 (1999) 267±275

Fig. 3. Cumulative incidences of loco-regional recurrences according to nodal stage. Numbers of patients at risk are under the curve.

adjuvant chemotherapy with CAF-type adjuvant chemotherapy no differences in the local recurrence rate were observed [13]. Our results show that ECE may help to identify those node-positive patients who are at high risk for loco-regional recurrence. The UICC TNM system provides a distinction between extracapsular and intracapsular tumour extension. At least in The Netherlands a highest positive axillary node and/or ECE is often used to select patients at high risk for loco-regional recurrence for postoperative irradiation. However, results with respect to loco-regional control as well as survival are scarce and even con¯icting [8,15,30,35]. In the present series, 49% of patients with positive nodes had extracapsular extension; similar percentages of between 40% and 60% have been described by others [8,30]. The 10-year cumulative incidence of loco-regional recurrence in the node-positive group of patients and ECE-positive was 38% compared to 20% when ECE was absent; in the multivariate analysis of the node-positive patients it remained associated with prognosis, whereas the effect of the number or site of positive nodes disappeared. A simple decision rule can be derived from the identi®cation of a high and low risk group based on tumour and nodal information, the factors that remained independently associated with the 10-year risk. The group of patients with four or more positive nodes or T2±3 tumours with one to three

Fig. 4. Cumulative incidences of loco-regional recurrences in node-positive patients according to the presence of extracapsular tumour extension (ECE) of axillary node metastases. Numbers of patients at risk are under the curve.

positive nodes had a cumulative risk of 34% at 10 years compared to an 8% risk for the remaining patients. The introduction of ECE in the model suggested additional value of ECE, identifying different subsets in the patients already at high risk (.10%) for loco-regional recurrence. Patients with T1 tumours and one to three positive nodes were a subgroup in which ECE may be of clinical relevance. From these data one can make his own decision rule by ®rst deciding what loco-regional recurrence rate can be accepted without radiotherapy. This obviously depends on the expected bene®t, side-effects, possibilities for salvage therapy and costs. If we assume that the risk of loco-regional recurrence at 10 years should not exceed 10%, as is found in the treatment arm of several randomized studies, the Danish study inclusive, we have to irradiate all node-positive tumours except patients with T1 N1±3 node-positive and ECE-negative tumours (see Table 6). This would mean irradiation for about 84% of the node-positive patients.

Table 5 Results of multivariate analysis of N 1 patients (n ˆ 353) Covariate

HR a

95% CI

P-value

pT T2 versus T1 T3 versus T1 ECE (present versus absent) HPN (present versus absent)

2.41 2.77 1.93 1.69

1.12±5.15 1.01±7.58 0.99±3.75 0.93±3.06

, 0.05 0.05 0.05 0.93

a

HR, hazard ratio; CI, con®dence interval.

Fig. 5. Cumulative incidences of loco-regional recurrences according to high and low risk groups. Numbers of patients at risk are under the curve.

J.J. Jager et al. / Radiotherapy and Oncology 50 (1999) 267±275

273

Table 6 LRR rate in relation to T- and N-status and ECE N0

N1±3

Crude rate (%)

ECE 2 (%)

N4 1 ECE 1 (%)

T1 8/110 (7) 3/30 (10) 2/7 (29) T2±3 9/145 (6) 8/50 (16) 8/31 (26) LRR rate according to T- and N-status and ECE for isolated recurrences T1 7/110 (6) 2/30 (7) 1/7 (14) T2±3 6/145 (4) 7/50 (14) 7/31 (23) LRR rate according to T- and N-status and ECE for patients without radiotherapy T1 6/73 (8) 2/22 (9) 1/6 (17) T2±3 3/74 (4) 5/40 (13) 5/21 (24)

A number of other variables that could potentially in¯uence this decision rule were not found to predict for locoregional recurrence. This is true for the presence of a positive or near-positive resection margin. A potential association could, however, have been obscured by the fact that irradiation was recommended in these situations. Previous investigations have also failed to demonstrate an effect of the resection margin status [1,31], but it should be noted that these results are based on a relatively small number of events. We analyzed the pattern of relapse and found a virtually equal recurrence rate in the chest wall and in the regional nodes. The incidence of an axillary recurrence was extremely low (3% of all patients), although only 2% of all patients had received radiotherapy to this area. This low axillary recurrence rate is related to the application of a complete axillary dissection; without axillary dissection and radiotherapy the rate may be as high as 18% [12], whereas with incomplete dissections and no radiotherapy an intermediate axillary recurrence rate may be expected [17,33,34,40]. The intention was to perform a separate analysis for chest wall and nodal recurrences to identify new factors that could possibly predict exclusively for one of these events. Due to the low numbers of events we were not able to perform this for lymph node metastasis as the endpoint. The risk pattern for chest wall recurrence was similar compared to overall loco-regional recurrence with pT and particularly pN the most important predictive factors. Flow cytometric-derived factors did not in¯uence the results for chest wall recurrence. Table 7 Multivariate analysis for chest wall recurrence as the endpoint Covariate pT T2 versus T1 T3 versus T1 pN N1±3 versus N0 N . 3 versus N0 a

HR a

95% CI

2.14 1.45

0.93±4.93 0.28±7.45

0.07 0.60

1.40 2.49

0.62±3.16 1.11±5.62

0.41 , 0.05

HR, hazard ratio; CI, con®dence interval.

P-value

ECE 2 (%)

ECE 1 (%)

1/4 (25) 4/18 (22)

1/9 (11) 15/48 (31)

1/4 (25) 4/18 (22)

0/9 (0) 12/48 (25)

1/4 (25) 3/17 (18)

0/7 (0) 11/43 (26)

Finally, the issue of the treatment volume should be addressed. In the literature a wide variety of recommended treatment volumes can be found [8,16,19,20,24,25,30, 37,42,44,46,48]. Irradiation of the internal mammary chain is controversial [24,29,42] and the subject of an ongoing large European study. Treatment of the supraclavicular fossa, advocated by the majority of authors, is questioned by one institute [11] and irradiation of the chest wall is omitted by another [24]. We found a very low incidence of axillary recurrences after complete axillary dissection, as has also been described by others [7]. In contrast, in 4-year results of the Danish studies published in 1988, an axillary recurrence was observed in 44% of the patients who developed a loco-regional recurrence after dissection of levels I and II only [33]. Therefore, it seems to be reasonable to limit treatment volumes to the chest wall and the supraclavicular fossa, perhaps with the apical axilla, as is current practice in some institutes mainly in the USA [14,24]. This would result in a rather simple technique without problems of matching ®eld arrangements and with a far more homogeneous dose distribution. It may also result in a decrease of associated morbidity or complications compared to complete regional irradiation [29,41]. Our conclusions regarding the need to irradiate the axilla may change if the sentinel node biopsy approach is implemented in daily practice [18]. In summary, our results con®rm the importance of the pTstage and number of positive axillary nodes in predicting loco-regional recurrence. Extracapsular tumour extension may be helpful in identifying subgroups of node-positive patients at high risk. A highest positive node did not affect the risk of loco-regional recurrence. DNA-ploidy and Sphase fraction did not contribute additional information. A simple clinical decision rule has been developed recommending postoperative radiotherapy to all node-positive patients except patients with pT1 N1±3 and ECE-negative tumours. With such a policy, local recurrence rates can be achieved that are similar to those seen in the Danish study, in which radiotherapy was given to all node-positive patients and a survival bene®t was obtained [34]. In view of the low incidence of axillary recurrences following axillary clearance, we also suggest that the irradiation volume to

274

J.J. Jager et al. / Radiotherapy and Oncology 50 (1999) 267±275

the chest wall and the supraclavicular fossa is limited, at least in patients operated on with the intent to obtain a complete axillary clearance. References [1] Ahlborn TN, Gump FE, Bodian C, Habif DV, Kister S. Tumor to fascia margin as a factor in local recurrence after modi®ed mastectomy. Surg. Gynecol. Obstet. 1988;166:523±526. [2] Andersen PK, Borgan é, Gill RD, Keiding N. Statistical models based on counting processes. New York: Springer-Verlag, 1993. [3] Arriagada R, Rutqvist LE, Mattson A, Kramar A, Rotstein S. Adequate locoregional treatment for early breast cancer may prevent secondary dissemination. J. Clin. Oncol. 1995;13:2869±2878. [4] Axelsson C, Blichert-Toft M. Low-risk breast cancer patients treated by mastectomy and lower axillary dissection. Acta Oncol. 1988;27:605±609. [5] Borger J, Kemperman H, Hart A, Peterse H, van Dongen J, Bartelink H. Risk factors in breast-conservation therapy. J. Clin. Oncol. 1994;12:653±660. [6] Cancer Research Campaign Working Party. Cancer research campaign (King's/Cambridge) trial for early breast cancer; a detailed update at the tenth year. Lancet 1980;12:55±60. [7] Dahlstrùm KK, Andersson AP, Krag C. Wide local excision of recurrent breast cancer in the thoracic wall. Cancer 1993;72:774±777. [8] Donegan WL, Stine SB, Samter TG. Implications of extracapsular nodal metastases for treatment and prognosis of breast cancer. Cancer 1993;72:778±782. [9] Early Breast Cancer Trialists' Collaborative Group. Effects of radiotherapy and surgery in early breast cancer. An overview of the randomized trials. N. Engl. J. Med. 1995;333:1444±1455. [10] Ewers S-B, Attewell R, Baldetorp B, et al. Flow cytometry DNA analysis and prediction of loco-regional recurrences after mastectomy in breast cancer. Acta Oncol. 1992;31:733±740. [11] Fairlamb DJ, Machin D. Prophylactic supraclavicular fossa radiotherapy in early breast cancer: is it worthwhile?. Clin. Oncol. 1994;6:227± 231. [12] Fisher B, Redmond C, Fisher ER, et al. Ten-year results of a randomized clinical trial comparing radical mastectomy and total mastectomy with or without radiation. N. Engl. J. Med. 1985;312:674±681. [13] Fisher B, Brown AM, Dimitrov N, et al. Two months of doxorubicincyclophosphamide with and without interval reinduction therapy compared with 6 months of cyclophosphamide, methotrexate and ¯uorouracil in positive-node breast cancer patients with tamoxifennonresponsive tumors: results from the national surgical adjuvant breast and bowel project B-15. J. Clin. Oncol. 1990;8:1483±1496. [14] Fisher BJ, Perera FE, Cooke AL, et al. Long-term follow-up of axillary node-positive breast cancer patients receiving adjuvant systemic therapy alone: patterns of recurrence. Int. J. Radiat. Oncol. Biol. Phys. 1997;38:541±550. [15] Fisher BJ, Perera FE, Cooke AL, et al. Extracapsular axillary node extension in patients receiving adjuvant systemic therapy: an indication for radiotherapy?. Int. J. Radiat. Oncol. Biol. Phys. 1997;18:551± 559. [16] Fowble B, Glick J, Goodman R. Radiotherapy for the prevention of loco-regional recurrence in high risk patients post mastectomy receiving adjuvant chemotherapy. Int. J. Radiat. Oncol. Biol. Phys. 1988;15:627±631. [17] Fowble B, Gray R, Gilchrist K, Goodman RL, Taylor S, Tormey DC. Identi®cation of a subgroup of patients with breast cancer and histologically positive axillary nodes receiving adjuvant chemotherapy who may bene®t from postoperative radiotherapy. J. Clin. Oncol. 1988;6:1107±1117. [18] Giuliano AE, Jones RC, Brennan M, Statman R. Sentinel lymphadenectomy in breast cancer. J. Clin. Oncol. 1997;15:2345±2350.

[19] Griem KL, Henderson IC, Gelman R, et al. The 5-year results of a randomized trial of adjuvant radiation therapy after chemotherapy in breast cancer patients treated with mastectomy. J. Clin. Oncol. 1987;5:1546±1555. [20] Halverson KJ, Taylor ME, Perez CA, et al. Regional nodal management and patterns of failure following conservative surgery and radiation therapy for stage I and II breast cancer. Int. J. Radiat. Oncol. Biol. Phys. 1993;26:593±599. [21] Hùst H, Brennhovd IO, Loeb M. Postoperative radiotherapy in breast cancer ± longterm results from the Oslo study. Int. J. Radiat. Oncol. Biol. Phys. 1986;12:727±732. [22] Hupperets PSGJ, Wils JAJM, Volovics L, et al. Adjuvant chemohormonal therapy with cyclofosphamide, doxorubicin and 5-¯uorouracil (CAF) with or without medroxyprogesterone acetate for nodepositive breast cancer patients. Ann. Oncol. 1993;4:295±301. [23] Hupperets PS, Schutte B, van Assche C, et al. Ploidy and S-phase fractions (SPF) of primary breast cancers and their nodal metastases. Cancer Res. Treat. 1994;32:197±202. [24] Jacobson JA, Danforth DN, Cowan KH, et al. Ten-year results of a comparison of conservation with mastectomy in the treatment of stage I and II breast cancer. N. Engl. J. Med. 1995;332:907±911. [25] Janjan NA, McNeese MD, Buzdar AU, Montague ED, Oswald MJ. Management of locoregional recurrent breast cancer. Cancer 1986;58:1552±1556. [26] Kalb¯eisch JD, Prentice RL. Competing risk in multivariate analysis. In: Kalb¯eisch JD, Prentice RL, editors. The statistical analysis of failure time data, New York: Wiley, 1980. pp. 163. [27] Kennedy MJ, Abeloff MD. Management of locally recurrent breast cancer. Cancer 1993;71:2395±2409. [28] Korzeniowski S, Dyba T, Skolyszewski J. Classical prognostic factors for survival and loco-regional control in breast cancer patients treated with radical mastectomy alone. Acta Oncol. 1994;33:759±765. [29] Larson D, Weinstein M, Goldberg I, et al. Edema of the arm as a function of the extent of axillary surgery in patients with stage I±II carcinoma of the breast treated with primary radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 1986;12:1575±1582. [30] Leonard C, Corkill M, Tompkin J, et al. Are axillary recurrence and overall survival affected by axillary extranodal tumor extension in breast cancer? Implications for radiation therapy. J. Clin. Oncol. 1995;13:47±53. [31] Mentzer SJ, Osteen RT, Wilson RE. Local recurrence and the deep resection margin in carcinoma of the breast. Surg. Gynecol. Obstet. 1986;163:513±517. [32] Ottesen G, Andersen JA, Blichert-Toft M, Axelsson C. Frequency and types of chest wall recurrences among node negative breast cancer patients. Acta Oncol. 1988;27:601±609. [33] Overgaard M, Christensen JJ, Johansen H, et al. Postmastectomy irradiation in high-risk breast cancer patients. Present status of the Danish Breast Cancer Cooperative Group trials. Acta Oncol. 1988;27:707±714. [34] Overgaard M, Hansen PS, Overgaard J, et al. Postoperative radiotherapy in high-risk premenopausal women with breast cancer who receive adjuvant chemotherapy. N. Engl. J. Med. 1997;337:949±955. [35] Perera F, Fisher BJ, Cooke A, Venkatesan VM, Dar AR, Stitt L. Locoregional recurrence and extranodal extension in patients receiving systemic therapy for axillary node positive breast cancer (abstract). Int. J. Radiat. Oncol. Biol. Phys. 1993;27:265. [36] 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±962. [37] Recht A. Nodal treatment for patients with early stage breast cancer: guilty or innocent? 1992;Radiother. Oncol. 25:79±82. [38] Recht A. The return (?) of postmastectomy radiotherapy. J. Clin. Oncol. 1995;13:2861±2864. [39] Rosenman J, Bernard S, Kober C, Ieland W, Varia M, Newsome J. Local recurrences in patients with breast cancer at the North Carolina Memorial Hospital (1970±1980). Cancer 1986;57:1421±1425.

J.J. Jager et al. / Radiotherapy and Oncology 50 (1999) 267±275 [40] Rutqvist LE, Pettersson D, Johansson H. Adjuvant radiation therapy versus surgery alone in operable breast cancer: long-term follow-up of a randomized clinical trial. Radiother. Oncol. 1993;26:104±110. [41] Ryttov N, Holm NV, Qvist N, Blichert-Toft M. In¯uence of adjuvant irradiation on the development of late arm lymphedema and impaired shoulder mobility after mastectomy for carcinoma of the breast. Acta Oncol. 1988;27:667±670. [42] Sacks NPM, Baum M. Primary management of carcinoma of the breast. Lancet 1993;342:1402±1408. [43] Schutte B, Reynders MMJ, Bosman FT, Blijham GH. Flow cytometry determination of DNA ploidy level in nuclei isolated from paraf®nembedded tissue. Cytometry 1985;6:26±30. [44] Stefanik D, Goldberg R, Byrne P, et al. Local-regional failure in patients treated with adjuvant chemotherapy for breast cancer. J. Clin. Oncol. 1985;3:660±665.

275

[45] Tennvall-Nittby L, Tengrup I, Landberg T. The total incidence of loco-regional recurrence in a randomized trial of breast cancer TNM stage II. The South Sweden Breast Cancer Trial. Acta Oncol. 1993;32:641±646. [46] Toonkel LM, Fix I, Jacobson LH, Schneider JJ, Wallach CB. Postoperative radiation therapy for carcinoma of the breast: improved results with elective irradiation of the chest wall. Int. J. Radiat. Oncol. Biol. Phys. 1982;8:977±982. [47] Toonkel LM, Fix I, Jacobson LH, Wallach CB. The signi®cance of local recurrence of carcinoma of the breast. Int. J. Radiat. Oncol. Biol. Phys. 1983;9:33±39. [48] Uematsu M, Bornstein BA, Recht A, et al. Long-term results of postoperative radiation therapy following mastectomy with or without chemotherapy in stage I±III breast cancer. Int. J. Radiat. Oncol. Biol. Phys. 1993;25:765±770.