Breast Cancer Patients Treated Without Axillary Surgery

Abstract

Objective

To evaluate the impact of breast carcinoma (T1–2N0) surgery without axillary dissection on axillary and distant relapses, and to evaluate the usefulness of a panel of pathobiologic parameters determined from the primary tumor, independent of axillary nodal status, in planning adjuvant treatment.

Methods

In a prospective nonrandomized pilot study, 401 breast cancer patients who underwent breast surgery without axillary dissection were accrued from January 1986 to June 1994. At surgery, all patients were clinically node-negative and lacked evidence of distant metastases after clinical or radiologic examination. A precise 4-month clinical and radiologic follow-up was performed to detect axillary or distant metastases. Patients with clinical evidence of axillary nodal relapse were considered for surgery as salvage treatment. Biologic characteristics of primary carcinomas were investigated by immunohistochemistry, and four pathologic and biologic parameters (size, grading, laminin receptor, and c-erbB-2 receptor) were analyzed to determine a prognostic score.

Results

The 5-year follow-up of these patients revealed a low rate of nodal relapses (6.7%), particularly for T1a and T1b patients (2% and 1.7%, respectively), whereas T1c and T2 patients showed a 10% and 18% relapse rate, respectively. Surgery was a safe and feasible salvage treatment without technical problems in all 19 cases of progressive disease at the axillary level. The low rate of distant metastases in T1a and T1b groups (<6%) increased to 15% in T1c and 34% in T2 patients. Analyzing the primary tumor with respect to the panel of pathologic and biologic parameters was predictive of metastatic spread and therefore can replace nodal status information for planning adjuvant treatment.

Conclusions

Middle-term follow-up shows that the rate of axillary relapse in this patient population is lower than expected, suggesting that only a minimal number of microembolic nodal metastases become clinically evident. Avoidance of axillary dissection has a negligible effect on the outcome of T1 patients, particularly in T1a and T1b tumors with no palpable nodes, because the rate of axillary node relapse is very low for both. In T1 breast carcinoma, postsurgical therapy should be considered on the basis of biologic characteristics rather than nodal involvement. The authors’ prognostic score based on the primary tumor identified patients who required postsurgical treatment, providing a practical alternative to axillary status for deciding on adjuvant treatment. Conversely, in the T2 group, the high rate of salvage surgery for axillary relapses, which is expected in tumors larger than 2.5 cm or 3.0 cm, represents a limit for avoiding axillary dissection. Preoperative evaluation of axillary nodes for modification of surgical dissection in this subgroup would be more useful more than in T1 breast cancer because of the high risk. Complete dissection is feasible without technical problems if precise follow-up detects progressive axillary disease.

In recent years, axillary dissection in patients with T1 breast carcinoma has been intensively discussed in terms of usefulness and type of surgery (sampling, low or total axillary dissection). Indeed, it is well known in clinical practice that axillary dissection represents an overtreatment in the overwhelming majority of T1 breast cancer patients, because more than 75% of axillary nodes in this group of patients are histologically negative. The decision to excise clinically negative nodes is based mainly on the consideration that axillary node status is the most important predictive factor, and thus is essential for planning postsurgical treatment. This information is considered of such relevance that clinically uninvolved axillary nodes are routinely dissected or at least extensively sampled, despite the demonstration in several studies that axillary node dissection has a negligible effect as a prophylactic procedure. Because axillary surgery is responsible for most of the postsurgical sequelae in breast cancer treatment, there is a need to determine whether a subset of patients can be identified in whom axillary node dissection is not required. Further, it must be determined whether adequate prognostic information can be obtained by alternative and less invasive procedures that would be useful both in identifying patients likely to develop distant metastasis and in planning adjuvant therapy.

In recent efforts to modify the requirement for complete axillary surgery, lymphatic mapping and sentinel node biopsy have been evaluated in terms of detection rate and prediction of axillary status. However, this is merely a different surgical approach for obtaining information on axillary status; thus, the question remains of a viable alternative to axillary surgery and whether it is possible to disregard axillary status in breast cancer management. Several studies support a “wait and see” policy, although salvage treatment in the event of progressive disease remains an issue as well as the reliability of criteria other than axillary node involvement in planning adjuvant treatment.

In the present study, we have evaluated the effect of avoiding axillary surgery on local control and on the risk of developing distant metastases in selected breast cancer patients. We also describe a panel of pathologic and biologic parameters evaluated on the primary tumor that are predictive of metastatic spread, are useful in planning adjuvant treatment, and promise to obviate the need for axillary nodal status information.

PATIENTS AND METHODS

Patient Population

From 1986 to 1994, 401 breast carcinoma patients did not participate in trials and underwent breast surgery without axillary dissection at the National Cancer Institute of Milan. As previously described, ¹ criteria for enrollment of patients included breast tumor size of 3 cm or less at its largest diameter at clinical examination, with no evidence of skin involvement or fixation to the underlying muscle; axilla free of palpable nodes likely to be involved by metastatic deposit; no inflammatory breast cancers; and no evidence of distant metastases at clinical or laboratory examination at the time of diagnosis. Axillary dissection was avoided based on informed consent from the patient. The presence of other unrelated diseases (i.e., heart disease or neurologic disease) contraindicating general anesthesia or axillary dissection and its sequelae, previous contralateral breast tumors, and the patient’s refusal of axillary dissection eliminated these patients with respect to other clinical trials, but they were still considered for this pilot study.

Breast surgery and radiation therapy were carried out as described in detail. ² Total mastectomy was performed when an extensive intraductal component was present.

Radiation therapy on the operated breast was given in patients younger than 70 years of age because it has been shown to have lower efficacy in preventing local relapse in elderly patients. ^3,4 Technically, radiation therapy was performed using supervoltage equipment (6 MV) or cobalt 60. Treatment planning included the use of wedges to minimize dose inhomogeneity with isodose calculations for two tangential fields that were treated daily. The dose to the entire breast was 5,000 cGy given during a period of approximately 5 weeks; the daily fraction size was 200 cGy. A booster dose of radiation of 1,000 cGy was delivered to the tumor bed, thus making the total dose 6,000 cGy. The axillary region was excluded from the irradiated area. Supraclavicular and internal mammary regions were never considered as the target of an irradiation plan.

Adjuvant hormone therapy (20 mg/day tamoxifen) was given in postmenopausal patients, and generally in patients older than 50 years of age. Chemotherapy was not administered to any patients in this series. Follow-up included routine physical examination every 4 months, with particular attention to the ipsilateral axilla and in general to the regional lymph nodal fields. Periodic mammography, chest radiography, bone scan, and liver ultrasonography were also performed.

Pathologic Assessment

Primary tumor size was determined by gross or microscopic examination. All histologic specimens (stained with hematoxylin and eosin) included in this study were reviewed with respect to previous reports for diagnostic reassessment of histotype, grading, necrosis, lymphocyte infiltration, tumor margins, and multifocality, and to add any missing data. Tumor margins were classified as round or stellate. Lymphoid cell infiltration was evaluated morphologically and scored as “strong” or “moderate” when a cellular population was present and “negative” when only the occasional inflammatory cell was found. Histologic grading was performed according to the method of Elston and Ellis, ⁵ considering three morphologic features: percentage of tubule formation, degree of nuclear pleomorphism, and mitotic count. The latter parameter was scored as “high” for more than 20 mitoses per high-power field and as “low” for less than 20 mitoses per high-power field after analysis of at least 10 microscopic fields at ×400 magnification. Presence or absence of noncomedo necrosis was also recorded.

Immunohistochemistry

The following monoclonal antibodies (mAbs) were used: anti-c-erbB-2 mAb cB11 (1:60 diluted; Ylem, Avezzano, Italy), anti-p53 mAb DO7 (1:6 diluted; Novocastra, Newcastle-upon-Tyne, UK), anti-bcl2 mAb 100 (1:20 diluted; kindly provided by Dr. David Mason), antiprogesterone receptor mAb 1A6 (1:100 diluted; DBA, Segrate, Italy), antiestrogen receptor mAb 1D5 (1:200 diluted; DBA, and antilaminin receptor (LR) MLuC5 (1:100 diluted; produced in our laboratory).

Immunostaining was performed using the peroxidase–streptavidin method ⁶ on representative formalin- or Bouin-fixed, paraffin-embedded samples of the tumor. Briefly, consecutive sections were cut, deparaffinized, rehydrated, and pretreated using the heat-induced epitope retrieval method. After washes in phosphate-buffered saline, immunostaining was carried out using an automated immunostainer (Dako TechMate 1000, Copenhagen, Denmark). Adequate positive and negative controls were performed. Sections were considered positive when more than 5% to 10% of the tumor cells were labeled.

Statistical Methods

All variables used in the analysis were categorical. Relationships between variables were studied in a two-by-two setting by means of contingency tables and evaluated with the chi-square test.

RESULTS

In this study, 401 breast carcinoma patients treated surgically for the primary tumor without axillary dissection were considered (Table 1). Median follow-up time was 62.6 ± 23.4 months (range 36–132); 58 patients died (14.5%), 32 of unrelated or unknown causes.

Table 1. PATIENT PROFILE

DCIS, ductal carcinoma in situ; ER, estrogen receptors; IDC, infiltrating ductal carcinoma; ILC, infiltrating lobular carcinoma; PgR, progesterone receptors.

Clinical examination during follow-up revealed palpable axillary nodes in 35 patients. Among these 35 patients, 8 with cytologically positive axillary nodes did not undergo surgery because of synchronous visceral metastases (5 patients) or cardiovascular disease contraindicating general anesthesia (3 patients). The remaining 27 patients underwent complete axillary dissection with en bloc removal of the three axillary levels. Technical problems stemming from the delay in the surgical procedure were negligible. In all operations performed, we never detected infiltration of the axillary vein or thorax wall, nor was there any other need to resect nervous, vascular, or muscular structures generally spared in elective axillary surgery, even with massive nodal involvement. No further axillary relapses were seen in these operated patients. Of these 27 patients, 8 (30%) underwent axillary dissection for clinical suspicion of involved nodes, but such lymph nodes were histologically negative for metastatic disease. In contrast, 19 (70%) were positive for axillary metastases: the mean number of metastatic nodes was 6 (range 1–32) and the mean number of nodes removed was 21 (range 15–35). The median axillary-free elapsed time was 30.6 months (range 6–72 months). Eleven of the 19 operated patients with metastatic axillary nodes are currently alive with no evidence of disease, whereas 8 patients showed distant metastases on further clinical and laboratory follow-up. The group without further distant relapses had a median axillary-free elapsed time of 36.1 months, compared with 22.8 months in the patients with distant relapses, which became clinically evident in a median time of 16.2 months (range 2–44 months). The median follow-up time after axillary surgery was similar in the two groups (47.6 vs. 42.6 months), and the mean number of metastatic nodes was more than four nodes in both groups (five vs. eight, respectively). In general, in patients with axillary recurrence who had not received tamoxifen before, the decision was made to give tamoxifen if axillary lymph nodes were histologically involved. However, tamoxifen therapy had no effect on the frequency of axillary recurrence or on the ultimate outcome.

No significant correlation was seen between axillary relapses and breast side, site, or histotype of the primary tumor. Moreover, there was no difference in the axillary recurrence rate between patients who received radiation therapy and those who did not. Also, there was no difference in the axillary recurrence rate between patients who underwent quadrantectomy rather than mastectomy. Overall, axillary progressive disease occurred in 6.7% of the patient population.

Distant metastases were found in 44 patients (11%), with the most frequent sites in bone (37%) and lung (32%), followed by liver (14%), brain (11%), and supraclavicular nodes (7%). The number of axillary and distant events was similar in all the groups receiving different postsurgical therapies.

The rate of axillary metastases increased progressively with size of the primary tumor, ranging from approximately 2% in T1a patients to 18% in T2 patients. The rate of distant metastases was also related to primary tumor size: in T1a-b patients, the mean rate was 4%, whereas in T1c and T2 patients, the metastatic rate was respectively 15% and 34% (Table 2). Only 13 of 44 patients with distant metastases clinically showed synchronous or previous axillary relapses.

Table 2. AXILLARY NODAL RELAPSES AND DISTANT METASTASIS IN PATIENTS GROUPED ACCORDING TO TUMOR SIZE

The frequency of large tumors (>2 cm) was significantly higher in patients who had axillary metastasis compared with patients who remained free of relapses (Table 3). Patients with grade III tumors or tumors with either necrosis or high proliferation rates also had more frequent relapses at the axillary node level. The other clinical and pathologic parameters examined were not significantly associated with nodal relapses. Tumor grade, number of mitoses, necrosis, tumor size, and proliferation rate (labeling index) were significantly more frequently associated with development of distant metastases compared with the patients free of metastases.

Table 3. FREQUENCY OF CLINICAL AND PATHOLOGIC PARAMETERS IN PATIENT SUBGROUPS

LPI, lymphoplasmacytic infiltration; NS, not significant.

To compare the prognostic information given by nodal status with that obtained from the biologic characterization of the primary tumor, six parameters were evaluated retrospectively by immunohistochemistry (Table 4). Expression of the laminin receptor and negativity for Bcl2 were significantly associated with axillary node relapses, whereas only progesterone receptor expression gave significant information concerning distant relapses. A logistic regression analysis of the standard pathologic factors and the molecular markers listed in Table 4 indicated that joint significance of the explanatory variables is not predictive for distant metastasis occurrence. These data indicate that no single biologic parameter can replace nodal status prognostic information.

Table 4. FREQUENCY OF BIOLOGIC AND PATHOLOGIC PARAMETERS IN PATIENT SUBGROUPS

NS, not significant.

By contrast, analysis of a score based on four pathobiologic parameters (tumor size, grading, laminin receptor, and HER2 overexpression), previously shown to have the same value as nodal status in breast cancer prognosis, ⁷ indicated that this score was predictive of axillary node relapses as well as distant relapses in the present patient series (Fig. 1). Indeed, the frequency of nodal relapses and of metastasis increased significantly (P < .01) from 2% to 18% and from 5% to 28%, respectively, with score increases from 1 to score 3–4. In the T1c subgroup (T1 patients with a higher risk of nodal relapses and distant metastasis), scores were strongly associated with both axillary node relapse and distant relapse (P < .01) (see Fig. 1).

Figure 1. Frequency of axillary node relapses (A1 and B1) and distant metastasis (A2 and B2) according to the prognostic score in all cases (A) and in T1c cases (B).

DISCUSSION

Axillary dissection has remained a permanent part of breast cancer care, because it is still considered mandatory for staging breast cancer and planning adjuvant treatment. ^8,9 Some efforts have been made to identify patients who might avoid axillary surgery—that is, patients at low risk of nodal metastases and patients whose therapeutic planning would not be changed by performing axillary dissection. ¹⁰ The low-risk group includes T1a tumors, ^11,12 and possibly T1b tumors if mammographically detected, ¹³ although this remains controversial ¹⁴; tumor size evaluated even in association with other pathologic or biologic parameters ^15,16 has failed to identify a subgroup of breast cancer patients with less than a 5% to 10% risk of axillary metastases. The second group includes patients who would receive chemotherapy or hormone therapy independent of nodal status; instead, the decision is based on primary tumor size, estrogen receptor status, ploidy, and proliferation index. ¹⁷ One study of 240 patients showed that axillary nodal status was a factor in the decision to provide adjuvant therapy for only 5% of the clinically node-negative patients. ¹⁸

More recently, new diagnostic approaches have been used to achieve the staging and therapeutic planning goals of axillary surgery. One of these is lymphoscintigraphy with sentinel node biopsy, which allows modification of further axillary surgery consistent with the pathology results of the biopsied node. However, in several studies, detection rates ranged from 66% to 98%, whereas false-negative rates ranged from 1% to 15%, with a prediction of axillary status ranging from 95% to 100%. ^19–23 In sentinel node biopsy, the goal is to provide sufficient information in the absence of axillary dissection to treat patients reliably as having node-negative breast cancer.

However, even these approaches consider the information derived from axillary node status necessary in planning further therapy. In the absence of this information, the question arises as to whether any reproducible parameters of the primary tumor exist that can predict distant metastatic spread with the same prognostic power as axillary nodal status. A recent study reported that the information derived from sentinel node biopsy increased the percentage of T1 breast cancer patients recommended for treatment based only on primary tumor characteristics to 37%. ²⁴ This may be a function of the parameters considered, which may have low predictive power and are insufficient alternatives to axillary status. ²⁴

Fisher et al ^25,26 and Baum and Coyle ²⁷ were the first to show that a “wait and see” policy regarding axillary nodes is possible in breast cancer without loss of local control of disease, and that it is sufficiently safe in terms of overall survival. Other studies in the past decade, in which different therapeutic strategies on the axilla were tested, ranging from no intervention ^28,29 to radiation therapy of the lower ³⁰ or entire axilla, ³¹ have generally confirmed a good confinement of disease and overall survival. The good regional control of disease was not only in terms of axillary relapses, ^29–31 but also with respect to a low complication rate after salvage surgery. ²⁸ Our present data on 401 patients show that axillary progressive disease occurred in only 6.7% of the patients after a mean follow-up of more than 5 years, suggesting that only a limited number of microembolic axillary metastases become clinically evident at that time. The role of such microembolic metastases in clinical practice and in the natural history of the disease should be considered as a result of this observation. Moreover, the risk of relapse at the axilla is mainly related to the size of the primary tumor (2% for T1a cancer to 18% for T2). However, it has been well established that lymph node metastases are “indicators but not governors” of survival and may represent organ-specific metastatic cells, unable to lodge or grow in other sites. ³² The different clinical behaviors in operated patients in our series with axillary metastases support the hypothesis that there were two populations with different axillary disease-free elapsed time and different further follow-up times. This further suggests that the appearance of axillary metastases does not necessarily indicate a poor prognosis or a potential source of tertiary spread. Clinical axillary nodal involvement was found in only 13 of the 44 patients who had distant metastases, arguing against a sequential model of metastatic spread in breast cancer, and arguing for the validity of planning cancer management without treating the axilla.

Our data indicate that elective axillary dissection can be considered an overtreatment in T1 breast cancer and particularly in T1a and T1b tumors with no palpable nodes, because the rate of axillary node relapses is actually very low in both. T1c patients who did not undergo axillary dissection actually had, as expected, a higher rate than T1a and T1b patients for both nodal relapses (10.5%) and distant metastases (15.4%). In T1 breast cancer, postsurgical therapy should be considered on the basis of biologic characteristics rather than nodal involvement: our prognostic score of the primary tumor could be used to identify patients who required postsurgical treatment, providing a practical alternative to axillary status for making decisions on adjuvant treatment. Conversely, in the T2 group, the high rate of salvage surgery for axillary relapses, expected with tumor sizes of more than 2.5 to 3.0 cm, represents a limit for avoiding axillary dissection. In this subgroup especially, a preoperative evaluation of axillary nodes for a modification of surgical dissection would be useful. Nevertheless, even delayed axillary dissection proved to be an effective salvage treatment if a regular clinical follow-up was performed every 4 months, and no technical problems occurred in performing a complete clearance.

Setting criteria by which to plan adjuvant treatments without information on nodal involvement is crucial, particularly for T1c patients, for whom the risk of undertreatment is real. Nevertheless, in our series, patients did not receive adjuvant chemotherapy, and tamoxifen was given to only a few elderly patients.

The need for adjuvant treatments is related to the biologic aggressiveness of the disease, and subsets of patients at high risk of local or distant relapse should be identified on the basis of the characteristics of the primary tumor, when information on nodal involvement is not available. Thus, we retrospectively investigated several biologic and pathologic parameters of the primary tumors in an effort to identify subsets of patients with different outcomes and to establish criteria, other than nodal involvement, for a decision on adjuvant treatment. Because of the relatively few patients in our series, analysis of the data revealed biologic trends for single parameters more than statistically significant values. Nevertheless, such trends were observed for nearly all the parameters that were tested, whereas clinical features (age and menopausal status) showed neither significant values nor any trend. Thus, the use of traditional parameters in association with nodal involvement does not appear to be a viable alternative to axillary dissection for planning adjuvant treatment.

Moreover, previous data from our institute showed that a panel of pathobiologic parameters had the same predictive value on metastatic risk as nodal involvement. ⁷ Those results identified a biologic subpopulation of tumors with a poor prognosis, particularly when biologic (c-erbB-2 and 67LR overexpression) and pathologic (grading and tumor size) parameters were evaluated together and a new distribution of the patient population was observed. Our present evaluation of the same parameters revealed a significant trend for metastatic risk in patients who showed the progressive presence of one to four parameters, in the sense of a worse prognosis. This finding represents the first demonstration that pathobiologic parameters of primary tumors from a consecutive series of patients treated without axillary dissection are statistically and clinically significant and predictive of metastatic risk.

Based on these promising results, our institute has initiated a prospective randomized clinical trial of T1 breast cancer patients, with or without axillary dissection, in which adjuvant treatments in the no-dissection group are based on pathobiologic parameters. Our goal remains the establishment of new guidelines in breast cancer therapeutic strategies that will make clinicians less dependent on pathologic axillary nodal status.