- Email: [email protected]
Selective analysis of the sentinel node in breast cancer
The American Journal of Surgery 182 (2001) 372–376
Selective analysis of the sentinel node in breast cancer Emmanuel E. Zervos, M.D.a,*, Brian D. Badgwell, M.D.b, Shahab F. Abdessalam, M.D.b, William B. Farrar, M.D.b, Michael J. Walker, M.D.b, Lisa D. Yee, M.D.b, William E. Burak, Jr., M.D.b b
a Digestive Disorders Center, Tampa General Hospital, University of South Florida, P.O. Box 1289, Rm. F145, Tampa, Florida 33601, USA Division of Surgical Oncology, Arthur G. James Cancer Hospital and Richard J. Solove Research Institute, Ohio State University, Columbus, OH, USA
Manuscript received June 29, 2001; revised manuscript July 16, 2001
Abstract Background: This study was designed to determine the minimum number of sentinel nodes necessary to accurately stage patients with breast cancer. Methods: Between August 1997 and February 2001, 509 consecutive patients were enrolled in a prospective sentinel node database. Nodes were characterized as either blue or hot (⬎2 times background), or both, and ranked based on the order harvested. Predictive value of the sentinel node based on these characteristics was evaluated to determine the minimum number necessary to stage the basin. Results: In all, 990 sentinel nodes were harvested from 465 basins. Pathologic stage in 126 of 128 positive basins was predicted by the first or second node harvested. The remaining 2 patients were positive by immunohistochemistry only. The hottest node predicted the status in 114 of 128 basins. Conclusions: Although all nodes should be examined, these data suggest that limiting frozen section analysis to the first two sentinel nodes identified will not compromise the accuracy of staging and may provide a vehicle for resource savings. © 2001 Excerpta Medica, Inc. All rights reserved. Keywords: Sentinel node; Breast cancer; Lymphatic mapping
Sentinel lymph node biopsy (SLNB) continues to gain momentum in becoming the standard of care for staging breast cancer. Several large prospective studies have now documented that a sentinel node can be localized in more than 90% of cases with false negative rates of 1% to 7% [1–7]. These studies have further documented that optimal execution of this technique involves the use of both vital blue dye and radio-labeled sulfur colloid to improve rates of localization and subsequent accuracy in staging. In the initial report of the combination technique, Albertini et al [8] reported an increase in the number of sentinel nodes identified. By definition, all “sentinel” nodes should be subjected to the same pathologic scrutiny, and while this has allowed the sentinel node concept to reach its nadir with regard to false negative rates, it has de facto increased the resources and overall costs associated with the technique. Presented at the Second Annual Meeting of the American Society of Breast Surgeons, La Jolla, California, May 5, 2001. * Corresponding author. Tel.: 1-813-251-7075; fax: 813-251-7396. E-mail address: [email protected]
At least two studies in melanoma have confirmed that not all sentinel nodes provide useful pathologic information. These studies go on to suggest that defining characteristics of sentinel nodes (ie, order harvested or level of radioactive uptake) may someday be employed to limit the extent of sentinel lymphadenectomy [9,10]. As with the sentinel node concept in general, many investigators are beginning to extrapolate these data to breast cancer patients. Two recent abstracts have identified the most radioactive or “hottest” sentinel node in breast cancer as comprising 70% to 80% of those nodes ultimately showing metastatic disease [11,12], but there exists a paucity of published data on mapping characteristics or order of harvest of predictive sentinel nodes in breast cancer. Since 1997, our institution has routinely applied SLNB and has enrolled all patients undergoing this procedure into a prospective database. Working closely with our colleagues in pathology, a low (5%) false negative rate on frozen section of the sentinel node by multilevel sectioning has been reported [13]. Such sectioning, however, is both labor intensive and costly. Because multiple sentinel nodes are often harvested from breast cancer patients, we hypothesized that a threshold
0002-9610/01/$ – see front matter © 2001 Excerpta Medica, Inc. All rights reserved. PII: S 0 0 0 2 - 9 6 1 0 ( 0 1 ) 0 0 7 4 0 - 1
E.E. Zervos et al. / The American Journal of Surgery 182 (2001) 372–376
may exist beyond which additional intense scrutiny of sentinel nodes would yield no additional useful pathologic information. We further hypothesized that identification of this threshold may allow us to minimize the number of nodes subjected to this scrutiny. In this review of our prospective database, we sought to determine exactly how many sentinel lymph nodes are necessary to accurately stage the axilla, and to propose a paradigm describing how these data might be employed to effect resource savings at our institution.
373
Table 1 Patient demographics (n ⫽ 509) Age (years) Sex Size of primary tumor Localization rate Total SLN harvested Number of SLN per patient (avg) Number of patients with positive SLN Axillary dissection False negative rate
57.9 ⫾ 12.53 505 female (99%) 1.75 cm ⫾ 1.26 cm 465 (91%) 990 1.94 ⫾ 1.6 128 (25%) 199 (39%) 4.8%
SLN ⫽ sentinel lymph node.
Patients and methods From August 1997 to January 2001, 509 consecutive patients underwent attempted SLNB for invasive breast cancer. Eight surgeons contributed to this database, each of whom completed the American Society of Breast Surgeons (ASBS) recommended guidelines for validation and proficiency of the sentinel node technique. All patients had cytologically or histologically proven breast cancer obtained from core needle biopsy, fine needle aspiration, or excisional biopsy prior to the sentinel node procedure. Technique of sentinel node biopsy Patients are injected at least 2 hours prior to their surgery with 400 Ci of filtered, 0.2 m, technetium sulfur colloid in equal aliquots either around the tumor or through localization needles placed for the purpose of injection. In the operating room, 5 cc of lymphazurin (vital blue) dye was injected around the tumor or through the localization needles. The hand-held gamma probe (Navigator; U.S. Surgical, Norwalk, Connecticut; or Neoprobe 2000, Dublin, Ohio) is used to localize the area in the axilla of increased radioactivity over which a small incision is made. Sentinel nodes are characterized as hot only or blue only, or both. Nodes that had ex-vivo radioactivity counts greater than 2 times background tissue were defined as hot, and ex-vivo counts were used to define the hottest node. All sentinel nodes were assigned a rank (1, 2, 3, 4, and so forth) based on the order harvested irrespective of the level of radioactivity or intensity of blue staining. In some cases, larger blocks of radioactive tissue were labeled as one node and attempts to subdivide such clusters of nodes were not made in the operating room. All sentinel nodes were sent for frozen section analysis and, when metastases were present, a level I/II axillary lymph node dissection was performed under the same anesthetic. Pathologic handling of the sentinel node In pathology, all sentinel nodes were bivalved, and step sections of each half were taken at 25%, 50%, and 75% of block depth, creating a total of six sections for each node processed. All nodes were then submitted for permanent sectioning, and in the event that both frozen and permanent sections were negative for metastatic disease, immunohis-
tochemistry using a cytokeratin antibody cocktail was performed on all nodes. If a sentinel node was negative on frozen section but positive on permanent section, or if the sentinel node was found to be reactive to cytokeratin antibody, then patients were brought back (or offered) completion axillary lymph node dissection. Statistics and data analysis All patients were prospectively entered into a computerbased registry, which is updated with each subsequent patient encounter. Pathology reports, operative notes, and data collection sheets filled out in the operating room were cross referenced to verify the rank, pathologic status, and mapping characteristics (radioactive uptake, blue staining) of each node. False negative rates were calculated by dividing the number of false negative lymph nodes by the number of positive axillary basins times 100. Predictive value for individual nodes or cohorts of nodes were determined by assessing how accurately that node predicted the true pathologic status of the basin. Potential for resource savings was determined by dividing the number of nodes necessary to achieve a given predictive value by the actual number of nodes analyzed. Longer term follow-up data were obtained from the tumor registry maintained by our National Cancer Institute-designated comprehensive cancer center and is current to within 3 months. The Student’s t test was used to compare means where appropriate, and significance was accepted with P ⬍0.05.
Results Demographics Between August 1997 and February 2001, 509 patients underwent attempted SLNB. Demographic summary of these patients is provided in Table 1. At least one sentinel lymph node could be localized in 91% of patients. Of these 465 patients, 199 went on to axillary lymph node dissection either because they had pathologically positive sentinel nodes or they were part of a surgeon’s validation series (n ⫽ 71; some surgeons achieved partial or full validation prior to enrolling patients in this database).
374
E.E. Zervos et al. / The American Journal of Surgery 182 (2001) 372–376
Table 2 Demographics of patients with positive sentinel node (n ⫽ 128)
Table 4 Radioactivity of non-hottest predictive sentinel node (n ⫽ 13)
Female SLN per patient, median (range) Positive SLNs per patient Immunohistochemistry positive only* Patients having ALND Axillary nodes per patient Total nodes per patient Patients with positive non-SLN
Patient number
Node rank
Blue
Percent hottest sentinel node
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 2 2 2 1 1 3 2 1 2 1 1 1 1
n y y y n y y y y n y y y y
30 29 11 18 87 23 No uptake 30 77 36 58 80 No uptake 99
100% 2 (0–16) 1.3 ⫾ 0.81 15 (12%) 116 21 ⫾ 9.0 24 ⫾ 9.0 56 (44%)
* Ninety-eight patients enrolled in ACS-OG Z-010; cytokeratin reactivity unknown. SLN ⫽ sentinel lymph node; ALND ⫽ axillary lymph node dissertion.
The demographics for the 128 patients with positive sentinel nodes are displayed in Table 2. Not all of these 128 patients went on to completion axillary dissection: 5 declined further surgery upon learning that their involved node was immunohistochemistry positive only, 4 refused altogether, and 3 were enrolled in the American College of Surgeons Oncology Group Z-011 trial, which prospectively randomizes patients with histologically positive sentinel nodes to axillary dissection or observation.
axillary dissection, only 1 of whom had positive nonsentinel nodes. Five patients declined further surgery to stage the axilla, with no reported axillary recurrences in these patients after 26 ⫾ 12.2 months. The “hottest” sentinel lymph node was positive in 114 of 128 (89%) patients. In patients whose hottest node was not the predictive node, radioactive counts ranged from 11% to 93% of the hottest node (Table 4). The first node identified was positive in 114 (89%) of 128 patients and the second node identified was positive in 12 (9.4%) patients. Pathologic stage was determined by one of the first two nodes identified in 126 (98%) of 128 patients with positive sentinel nodes. Of the 128 predictive sentinel nodes, 19 were hot only, 2 were blue only, and the remaining nodes were both hot and blue.
Distribution and pathologic status of all sentinel nodes The distribution of positive nodes and number of sentinel nodes harvested in all patients undergoing attempted sentinel node biopsy is delineated in Table 3. There were 255 nodes beyond the first two sentinel nodes harvested, and 177 positive sentinel nodes were found in 128 patients. Mapping characteristics of positive sentinel nodes Of the 128 patients with positive sentinel nodes, 15 were positive by cytokeratin antibody staining only. Size of the primary tumor in these patients (18 ⫾ 10.7 mm) did not differ from that of the general population of patients (17.5 ⫾ 12.6 mm). Of these patients, 10 went on to completion
Outliers Two patients were positive by immunohistochemistry only, and this occurred in the third of three and the fourth of
Table 3 Distribution of patients undergoing attempted sentinel lymph node biopsy Number of positive SLN
Number of sentinel lymph nodes 0* 1 Number of patients
2
3
4
0 1 2 3 4 5 ⱖ6
45
140 53
110 21 12
45 15 7 5
24 3 3 1 0
9 2 1 0 0 0
Total patients
45
193
143
72
31
12
* Could not localize. SLN ⫽ sentinel lymph node.
5
ⱖ6 8 2 0 2 0 0 1 Total 13
Positive/ total nodes
96/465 46/270 24/128 0/56 0/25 11/46 177/990 509
E.E. Zervos et al. / The American Journal of Surgery 182 (2001) 372–376
375
Table 5 Resource utilization Node(s) analyzed
First node
First 2 nodes
First 3 nodes
Hottest node
Actual
Predictive value Number of nodes Resource utilization
89% 465 46%
98% 735 74%
99% 863 87%
89% 465 46%
100% 990 100%
four nodes identified. The patient whose third of three sentinel nodes harvested was positive went on to axillary dissection, which showed histologically positive involvement in an additional 10 of 38 lymph nodes. This was 1 of 2 patients whose positive sentinel lymph node did not take up radiolabeled sulfur colloid and was blue only. The second patient’s fourth of four nodes was positive by immunohistochemistry only and was also the hottest node. That patient had no additional positive nodes. Potential resource savings Table 5 outlines a hypothetical paradigm examining potential resource savings when only those lymph nodes producing a given predictive value are analyzed. This paradigm examines the predictive value and potential resource savings when only the first, first two, first three, or hottest nodes are analyzed. In our institution, if one were willing to accept a 1.5% increase in the false negative rate, a 26% reduction in resource utilization could be realized. Axillary recurrences A total of 337 patients had no metastatic disease in their sentinel lymph nodes. Of these patients, 71 underwent completion axillary lymph node dissection as part of the surgeon’s validation, leaving 266 patients at risk for axillary recurrence. Of these 266 patients, with average follow-up of 19⫾9.2 months, there were no axillary recurrences. Two patients had local recurrences that resulted ultimately in axillary recurrences, but the local event preceded the regional event by months in both cases.
Comments Sentinel lymph node biopsy is an established, minimally invasive means of staging the axilla in patients with breast cancer. The morbidity associated with SLNB is decidedly less than that with traditional axillary staging, and several studies with longer term follow-up are now being published documenting minimal morbidity and low recurrence rates among patients with uninvolved sentinel lymph nodes [14 – 16]. If a single node most likely to harbor metastatic disease from a breast primary truly exists, through exhaustive clinical trials and comprehensive follow-up, surgeons have perfected the best means of identifying a group of nodes that
contain that special node. The natural evolution of any minimally invasive staging technique is to make it less invasive and more predictive. The trend has begun with melanoma and is now being extrapolated to breast cancer. This is the first study to evaluate the mapping characteristics, pathology, and order of harvest in those sentinel nodes that revealed metastatic spread in patients undergoing sentinel lymph node biopsy for breast cancer. The patients comprising this database are typical of most large series of patients undergoing sentinel node biopsy for breast cancer. They are nearly all women with clinical stage I and II breast cancers. Metastatic disease by histologic criteria always manifested in the first or second lymph nodes harvested. Two patients with only cytokeratin reactive sentinel nodes were the outliers. In 1 patient, the positive sentinel node did not take up radio-labeled sulfur colloid (as all but two of our predictive nodes did) and led to a completion axillary dissection that yielded 10 histologically involved nodes—all replaced by tumor. These findings suggest that the sentinel node in this patient may have been misidentified, possibly because of tumor obstruction of lymphatics. The other case, with a cytokeratin positive sentinel node only, is more in keeping with what we have come to expect in these patients: no additional pathologically involved nodes. One hundred and twenty-eight patients in our series had more than two sentinel nodes harvested, 25 had more than five nodes harvested. In total, 255 sentinel nodes beyond the first two nodes were harvested in these patients. Limiting sentinel node biopsy to the first two nodes in this series of patients would result in a more than 25% reduction in resource utilization across the board. We acknowledge that most surgeons would not sacrifice accurate staging of even 1 patient to realize such resource savings based on post hoc data. Frozen section analysis of the first two nodes, however, was 100% accurate. Therefore, resource saving may still be possible if one simply limits the amount of tissue submitted for frozen section. In our institution, at $193 per lymph node, this strategy alone would result in almost $50,000 in cost savings without compromising accuracy or exposing a single patient to an unnecessary anesthetic. In this retrospective review, we believe these data to demonstrate minimal bias, as none of the participating surgeons knew that this analysis would ever be carried out. These surgeons went about their operations labeling sentinel nodes with no preconceived predilection toward harvesting
376
E.E. Zervos et al. / The American Journal of Surgery 182 (2001) 372–376
the hottest or the biggest or the bluest node first; they merely identified those nodes that met the criteria for being a sentinel node, as they were identified. Having done so, the predictive node was almost always one of the first two identified. Any surgeon faced with the task of clearing the axilla of all radioactivity may appropriately question the utility of harvesting the sixth, or seventh— or even the16th—“sentinel” node. Based on these data, such frustration may be circumvented by continuing the dissection as the first two nodes are analyzed by frozen section. In our institution, our pathologists’ ability to expeditiously analyze these two nodes would truncate further axillary dissection 100% of the time while providing resource savings and not compromising the accuracy of staging.
References [1] Krag D, Weaver D, Ashikaga T, et al. The sentinel node in breast cancer—a multicenter validation study. N Eng; J Med 1998;339: 941– 6. [2] Cox CE, Bass SS, McCann CR, et al. Lymphatic mapping and sentinel lymph node biopsy in patients with breast cancer. Annu Rev Med 2000;51:525– 42. [3] Haigh PI, Hansen NM, Qi K, Giuliano AE. Biopsy method and excision volume do not affect success rate of subsequent sentinel lymph node dissection in breast cancer. Ann Surg Oncol. 2000;7: 21–7. [4] Veronesi U, Paganelli G, Viale G, et al. Sentinel lymph node biopsy and axillary dissection in breast cancer: results in a large series. J Natl Cancer Inst 199917;91:368 –73. [5] Giuliano AE, Jones RC, Brennan M, et al. Sentinel lymphadenectomy in breast cancer. J Clin Oncol 1997;15:2345–50.
[6] Veronesi U, Paganelli G, Galimberti V, et al. Sentinel node biopsy to avoid axillary dissection in breast cancer with clinically negative nodes. Lancet 1997;349:1864 –7. [7] Barnwell JM, Arredondo MA, Kollmorgen D, et al. Sentinel node biopsy in breast cancer. Ann Surg Oncol 1998;5:126 –30. [8] Albertini JJ, Lyman GH, Cox CE, et al. Lymphatic mapping and sentinel node biopsy in the patient with breast cancer. JAMA 1996; 276:1818 –22. [9] Porter GA, Ross MI, Berman RS, et al. How many lymph nodes are enough during sentinel lymphadenectomy for primary melanoma? Surgery 2000;128:306 –11. [10] McMasters KM, Reintgen DS, Ross MI, et al. Sentinel lymph node biopsy for melanoma: how many radioactive nodes should be removed? Ann Surg Oncol 2001;8:192–7. [11] Martin RCG, Fey J, Borgen PI, Cody HS. Highest isotope count does not predict sentinel node positivity in all breast cancer patients. SSO Cancer Symposium, 2001, abstract no. 70, Washington, D.C., March 15–18, 2001. [12] Quan ML, McCready D, McKinnon JG, Lehr E. Sentinel node biopsy in breast cancer: is the hottest node always the positive one? SSO Cancer Symposium, 2001, abstract no. P31, Washington, D.C., March 15–18, 2001. [13] Zervos EE, Burak WE. Lymphatic mapping for breast cancer: experience at the Ohio State University. Breast Cancer 2000;7:195–200. [14] Kuehn T, Klauss W, Darsow M, et al. Long-term morbidity following axillary dissection in breast cancer patients— clinical assessment significance for life quality and the impact of demographic, oncologic and therapeutic factors. Breast Cancer Res Treat 2000;64:275– 86. [15] Giuliano AE, Haigh PI, Brennan MB, et al. Prospective observational study of sentinel lymphadenectomy without further axillary dissection in patients with sentinel node-negative breast cancer. J Clin Oncol 2000;18:2553–9. [16] Dessureault S, Dupont E, Shons A, et al. Early results of breast cancer lymphatic mapping from H. Lee Moffitt Cancer Center: no axillary recurrences in breast cancer patients after a negative sentinel node biopsy. Breast Cancer Res Treat 2000;64:3.
Selective analysis of the sentinel node in breast cancer Emmanuel E. Zervos, M.D.a,*, Brian D. Badgwell, M.D.b, Shahab F. Abdessalam, M.D.b, William B. Farrar, M.D.b, Michael J. Walker, M.D.b, Lisa D. Yee, M.D.b, William E. Burak, Jr., M.D.b b
a Digestive Disorders Center, Tampa General Hospital, University of South Florida, P.O. Box 1289, Rm. F145, Tampa, Florida 33601, USA Division of Surgical Oncology, Arthur G. James Cancer Hospital and Richard J. Solove Research Institute, Ohio State University, Columbus, OH, USA
Manuscript received June 29, 2001; revised manuscript July 16, 2001
Abstract Background: This study was designed to determine the minimum number of sentinel nodes necessary to accurately stage patients with breast cancer. Methods: Between August 1997 and February 2001, 509 consecutive patients were enrolled in a prospective sentinel node database. Nodes were characterized as either blue or hot (⬎2 times background), or both, and ranked based on the order harvested. Predictive value of the sentinel node based on these characteristics was evaluated to determine the minimum number necessary to stage the basin. Results: In all, 990 sentinel nodes were harvested from 465 basins. Pathologic stage in 126 of 128 positive basins was predicted by the first or second node harvested. The remaining 2 patients were positive by immunohistochemistry only. The hottest node predicted the status in 114 of 128 basins. Conclusions: Although all nodes should be examined, these data suggest that limiting frozen section analysis to the first two sentinel nodes identified will not compromise the accuracy of staging and may provide a vehicle for resource savings. © 2001 Excerpta Medica, Inc. All rights reserved. Keywords: Sentinel node; Breast cancer; Lymphatic mapping
Sentinel lymph node biopsy (SLNB) continues to gain momentum in becoming the standard of care for staging breast cancer. Several large prospective studies have now documented that a sentinel node can be localized in more than 90% of cases with false negative rates of 1% to 7% [1–7]. These studies have further documented that optimal execution of this technique involves the use of both vital blue dye and radio-labeled sulfur colloid to improve rates of localization and subsequent accuracy in staging. In the initial report of the combination technique, Albertini et al [8] reported an increase in the number of sentinel nodes identified. By definition, all “sentinel” nodes should be subjected to the same pathologic scrutiny, and while this has allowed the sentinel node concept to reach its nadir with regard to false negative rates, it has de facto increased the resources and overall costs associated with the technique. Presented at the Second Annual Meeting of the American Society of Breast Surgeons, La Jolla, California, May 5, 2001. * Corresponding author. Tel.: 1-813-251-7075; fax: 813-251-7396. E-mail address: [email protected]
At least two studies in melanoma have confirmed that not all sentinel nodes provide useful pathologic information. These studies go on to suggest that defining characteristics of sentinel nodes (ie, order harvested or level of radioactive uptake) may someday be employed to limit the extent of sentinel lymphadenectomy [9,10]. As with the sentinel node concept in general, many investigators are beginning to extrapolate these data to breast cancer patients. Two recent abstracts have identified the most radioactive or “hottest” sentinel node in breast cancer as comprising 70% to 80% of those nodes ultimately showing metastatic disease [11,12], but there exists a paucity of published data on mapping characteristics or order of harvest of predictive sentinel nodes in breast cancer. Since 1997, our institution has routinely applied SLNB and has enrolled all patients undergoing this procedure into a prospective database. Working closely with our colleagues in pathology, a low (5%) false negative rate on frozen section of the sentinel node by multilevel sectioning has been reported [13]. Such sectioning, however, is both labor intensive and costly. Because multiple sentinel nodes are often harvested from breast cancer patients, we hypothesized that a threshold
0002-9610/01/$ – see front matter © 2001 Excerpta Medica, Inc. All rights reserved. PII: S 0 0 0 2 - 9 6 1 0 ( 0 1 ) 0 0 7 4 0 - 1
E.E. Zervos et al. / The American Journal of Surgery 182 (2001) 372–376
may exist beyond which additional intense scrutiny of sentinel nodes would yield no additional useful pathologic information. We further hypothesized that identification of this threshold may allow us to minimize the number of nodes subjected to this scrutiny. In this review of our prospective database, we sought to determine exactly how many sentinel lymph nodes are necessary to accurately stage the axilla, and to propose a paradigm describing how these data might be employed to effect resource savings at our institution.
373
Table 1 Patient demographics (n ⫽ 509) Age (years) Sex Size of primary tumor Localization rate Total SLN harvested Number of SLN per patient (avg) Number of patients with positive SLN Axillary dissection False negative rate
57.9 ⫾ 12.53 505 female (99%) 1.75 cm ⫾ 1.26 cm 465 (91%) 990 1.94 ⫾ 1.6 128 (25%) 199 (39%) 4.8%
SLN ⫽ sentinel lymph node.
Patients and methods From August 1997 to January 2001, 509 consecutive patients underwent attempted SLNB for invasive breast cancer. Eight surgeons contributed to this database, each of whom completed the American Society of Breast Surgeons (ASBS) recommended guidelines for validation and proficiency of the sentinel node technique. All patients had cytologically or histologically proven breast cancer obtained from core needle biopsy, fine needle aspiration, or excisional biopsy prior to the sentinel node procedure. Technique of sentinel node biopsy Patients are injected at least 2 hours prior to their surgery with 400 Ci of filtered, 0.2 m, technetium sulfur colloid in equal aliquots either around the tumor or through localization needles placed for the purpose of injection. In the operating room, 5 cc of lymphazurin (vital blue) dye was injected around the tumor or through the localization needles. The hand-held gamma probe (Navigator; U.S. Surgical, Norwalk, Connecticut; or Neoprobe 2000, Dublin, Ohio) is used to localize the area in the axilla of increased radioactivity over which a small incision is made. Sentinel nodes are characterized as hot only or blue only, or both. Nodes that had ex-vivo radioactivity counts greater than 2 times background tissue were defined as hot, and ex-vivo counts were used to define the hottest node. All sentinel nodes were assigned a rank (1, 2, 3, 4, and so forth) based on the order harvested irrespective of the level of radioactivity or intensity of blue staining. In some cases, larger blocks of radioactive tissue were labeled as one node and attempts to subdivide such clusters of nodes were not made in the operating room. All sentinel nodes were sent for frozen section analysis and, when metastases were present, a level I/II axillary lymph node dissection was performed under the same anesthetic. Pathologic handling of the sentinel node In pathology, all sentinel nodes were bivalved, and step sections of each half were taken at 25%, 50%, and 75% of block depth, creating a total of six sections for each node processed. All nodes were then submitted for permanent sectioning, and in the event that both frozen and permanent sections were negative for metastatic disease, immunohis-
tochemistry using a cytokeratin antibody cocktail was performed on all nodes. If a sentinel node was negative on frozen section but positive on permanent section, or if the sentinel node was found to be reactive to cytokeratin antibody, then patients were brought back (or offered) completion axillary lymph node dissection. Statistics and data analysis All patients were prospectively entered into a computerbased registry, which is updated with each subsequent patient encounter. Pathology reports, operative notes, and data collection sheets filled out in the operating room were cross referenced to verify the rank, pathologic status, and mapping characteristics (radioactive uptake, blue staining) of each node. False negative rates were calculated by dividing the number of false negative lymph nodes by the number of positive axillary basins times 100. Predictive value for individual nodes or cohorts of nodes were determined by assessing how accurately that node predicted the true pathologic status of the basin. Potential for resource savings was determined by dividing the number of nodes necessary to achieve a given predictive value by the actual number of nodes analyzed. Longer term follow-up data were obtained from the tumor registry maintained by our National Cancer Institute-designated comprehensive cancer center and is current to within 3 months. The Student’s t test was used to compare means where appropriate, and significance was accepted with P ⬍0.05.
Results Demographics Between August 1997 and February 2001, 509 patients underwent attempted SLNB. Demographic summary of these patients is provided in Table 1. At least one sentinel lymph node could be localized in 91% of patients. Of these 465 patients, 199 went on to axillary lymph node dissection either because they had pathologically positive sentinel nodes or they were part of a surgeon’s validation series (n ⫽ 71; some surgeons achieved partial or full validation prior to enrolling patients in this database).
374
E.E. Zervos et al. / The American Journal of Surgery 182 (2001) 372–376
Table 2 Demographics of patients with positive sentinel node (n ⫽ 128)
Table 4 Radioactivity of non-hottest predictive sentinel node (n ⫽ 13)
Female SLN per patient, median (range) Positive SLNs per patient Immunohistochemistry positive only* Patients having ALND Axillary nodes per patient Total nodes per patient Patients with positive non-SLN
Patient number
Node rank
Blue
Percent hottest sentinel node
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 2 2 2 1 1 3 2 1 2 1 1 1 1
n y y y n y y y y n y y y y
30 29 11 18 87 23 No uptake 30 77 36 58 80 No uptake 99
100% 2 (0–16) 1.3 ⫾ 0.81 15 (12%) 116 21 ⫾ 9.0 24 ⫾ 9.0 56 (44%)
* Ninety-eight patients enrolled in ACS-OG Z-010; cytokeratin reactivity unknown. SLN ⫽ sentinel lymph node; ALND ⫽ axillary lymph node dissertion.
The demographics for the 128 patients with positive sentinel nodes are displayed in Table 2. Not all of these 128 patients went on to completion axillary dissection: 5 declined further surgery upon learning that their involved node was immunohistochemistry positive only, 4 refused altogether, and 3 were enrolled in the American College of Surgeons Oncology Group Z-011 trial, which prospectively randomizes patients with histologically positive sentinel nodes to axillary dissection or observation.
axillary dissection, only 1 of whom had positive nonsentinel nodes. Five patients declined further surgery to stage the axilla, with no reported axillary recurrences in these patients after 26 ⫾ 12.2 months. The “hottest” sentinel lymph node was positive in 114 of 128 (89%) patients. In patients whose hottest node was not the predictive node, radioactive counts ranged from 11% to 93% of the hottest node (Table 4). The first node identified was positive in 114 (89%) of 128 patients and the second node identified was positive in 12 (9.4%) patients. Pathologic stage was determined by one of the first two nodes identified in 126 (98%) of 128 patients with positive sentinel nodes. Of the 128 predictive sentinel nodes, 19 were hot only, 2 were blue only, and the remaining nodes were both hot and blue.
Distribution and pathologic status of all sentinel nodes The distribution of positive nodes and number of sentinel nodes harvested in all patients undergoing attempted sentinel node biopsy is delineated in Table 3. There were 255 nodes beyond the first two sentinel nodes harvested, and 177 positive sentinel nodes were found in 128 patients. Mapping characteristics of positive sentinel nodes Of the 128 patients with positive sentinel nodes, 15 were positive by cytokeratin antibody staining only. Size of the primary tumor in these patients (18 ⫾ 10.7 mm) did not differ from that of the general population of patients (17.5 ⫾ 12.6 mm). Of these patients, 10 went on to completion
Outliers Two patients were positive by immunohistochemistry only, and this occurred in the third of three and the fourth of
Table 3 Distribution of patients undergoing attempted sentinel lymph node biopsy Number of positive SLN
Number of sentinel lymph nodes 0* 1 Number of patients
2
3
4
0 1 2 3 4 5 ⱖ6
45
140 53
110 21 12
45 15 7 5
24 3 3 1 0
9 2 1 0 0 0
Total patients
45
193
143
72
31
12
* Could not localize. SLN ⫽ sentinel lymph node.
5
ⱖ6 8 2 0 2 0 0 1 Total 13
Positive/ total nodes
96/465 46/270 24/128 0/56 0/25 11/46 177/990 509
E.E. Zervos et al. / The American Journal of Surgery 182 (2001) 372–376
375
Table 5 Resource utilization Node(s) analyzed
First node
First 2 nodes
First 3 nodes
Hottest node
Actual
Predictive value Number of nodes Resource utilization
89% 465 46%
98% 735 74%
99% 863 87%
89% 465 46%
100% 990 100%
four nodes identified. The patient whose third of three sentinel nodes harvested was positive went on to axillary dissection, which showed histologically positive involvement in an additional 10 of 38 lymph nodes. This was 1 of 2 patients whose positive sentinel lymph node did not take up radiolabeled sulfur colloid and was blue only. The second patient’s fourth of four nodes was positive by immunohistochemistry only and was also the hottest node. That patient had no additional positive nodes. Potential resource savings Table 5 outlines a hypothetical paradigm examining potential resource savings when only those lymph nodes producing a given predictive value are analyzed. This paradigm examines the predictive value and potential resource savings when only the first, first two, first three, or hottest nodes are analyzed. In our institution, if one were willing to accept a 1.5% increase in the false negative rate, a 26% reduction in resource utilization could be realized. Axillary recurrences A total of 337 patients had no metastatic disease in their sentinel lymph nodes. Of these patients, 71 underwent completion axillary lymph node dissection as part of the surgeon’s validation, leaving 266 patients at risk for axillary recurrence. Of these 266 patients, with average follow-up of 19⫾9.2 months, there were no axillary recurrences. Two patients had local recurrences that resulted ultimately in axillary recurrences, but the local event preceded the regional event by months in both cases.
Comments Sentinel lymph node biopsy is an established, minimally invasive means of staging the axilla in patients with breast cancer. The morbidity associated with SLNB is decidedly less than that with traditional axillary staging, and several studies with longer term follow-up are now being published documenting minimal morbidity and low recurrence rates among patients with uninvolved sentinel lymph nodes [14 – 16]. If a single node most likely to harbor metastatic disease from a breast primary truly exists, through exhaustive clinical trials and comprehensive follow-up, surgeons have perfected the best means of identifying a group of nodes that
contain that special node. The natural evolution of any minimally invasive staging technique is to make it less invasive and more predictive. The trend has begun with melanoma and is now being extrapolated to breast cancer. This is the first study to evaluate the mapping characteristics, pathology, and order of harvest in those sentinel nodes that revealed metastatic spread in patients undergoing sentinel lymph node biopsy for breast cancer. The patients comprising this database are typical of most large series of patients undergoing sentinel node biopsy for breast cancer. They are nearly all women with clinical stage I and II breast cancers. Metastatic disease by histologic criteria always manifested in the first or second lymph nodes harvested. Two patients with only cytokeratin reactive sentinel nodes were the outliers. In 1 patient, the positive sentinel node did not take up radio-labeled sulfur colloid (as all but two of our predictive nodes did) and led to a completion axillary dissection that yielded 10 histologically involved nodes—all replaced by tumor. These findings suggest that the sentinel node in this patient may have been misidentified, possibly because of tumor obstruction of lymphatics. The other case, with a cytokeratin positive sentinel node only, is more in keeping with what we have come to expect in these patients: no additional pathologically involved nodes. One hundred and twenty-eight patients in our series had more than two sentinel nodes harvested, 25 had more than five nodes harvested. In total, 255 sentinel nodes beyond the first two nodes were harvested in these patients. Limiting sentinel node biopsy to the first two nodes in this series of patients would result in a more than 25% reduction in resource utilization across the board. We acknowledge that most surgeons would not sacrifice accurate staging of even 1 patient to realize such resource savings based on post hoc data. Frozen section analysis of the first two nodes, however, was 100% accurate. Therefore, resource saving may still be possible if one simply limits the amount of tissue submitted for frozen section. In our institution, at $193 per lymph node, this strategy alone would result in almost $50,000 in cost savings without compromising accuracy or exposing a single patient to an unnecessary anesthetic. In this retrospective review, we believe these data to demonstrate minimal bias, as none of the participating surgeons knew that this analysis would ever be carried out. These surgeons went about their operations labeling sentinel nodes with no preconceived predilection toward harvesting
376
E.E. Zervos et al. / The American Journal of Surgery 182 (2001) 372–376
the hottest or the biggest or the bluest node first; they merely identified those nodes that met the criteria for being a sentinel node, as they were identified. Having done so, the predictive node was almost always one of the first two identified. Any surgeon faced with the task of clearing the axilla of all radioactivity may appropriately question the utility of harvesting the sixth, or seventh— or even the16th—“sentinel” node. Based on these data, such frustration may be circumvented by continuing the dissection as the first two nodes are analyzed by frozen section. In our institution, our pathologists’ ability to expeditiously analyze these two nodes would truncate further axillary dissection 100% of the time while providing resource savings and not compromising the accuracy of staging.
References [1] Krag D, Weaver D, Ashikaga T, et al. The sentinel node in breast cancer—a multicenter validation study. N Eng; J Med 1998;339: 941– 6. [2] Cox CE, Bass SS, McCann CR, et al. Lymphatic mapping and sentinel lymph node biopsy in patients with breast cancer. Annu Rev Med 2000;51:525– 42. [3] Haigh PI, Hansen NM, Qi K, Giuliano AE. Biopsy method and excision volume do not affect success rate of subsequent sentinel lymph node dissection in breast cancer. Ann Surg Oncol. 2000;7: 21–7. [4] Veronesi U, Paganelli G, Viale G, et al. Sentinel lymph node biopsy and axillary dissection in breast cancer: results in a large series. J Natl Cancer Inst 199917;91:368 –73. [5] Giuliano AE, Jones RC, Brennan M, et al. Sentinel lymphadenectomy in breast cancer. J Clin Oncol 1997;15:2345–50.
[6] Veronesi U, Paganelli G, Galimberti V, et al. Sentinel node biopsy to avoid axillary dissection in breast cancer with clinically negative nodes. Lancet 1997;349:1864 –7. [7] Barnwell JM, Arredondo MA, Kollmorgen D, et al. Sentinel node biopsy in breast cancer. Ann Surg Oncol 1998;5:126 –30. [8] Albertini JJ, Lyman GH, Cox CE, et al. Lymphatic mapping and sentinel node biopsy in the patient with breast cancer. JAMA 1996; 276:1818 –22. [9] Porter GA, Ross MI, Berman RS, et al. How many lymph nodes are enough during sentinel lymphadenectomy for primary melanoma? Surgery 2000;128:306 –11. [10] McMasters KM, Reintgen DS, Ross MI, et al. Sentinel lymph node biopsy for melanoma: how many radioactive nodes should be removed? Ann Surg Oncol 2001;8:192–7. [11] Martin RCG, Fey J, Borgen PI, Cody HS. Highest isotope count does not predict sentinel node positivity in all breast cancer patients. SSO Cancer Symposium, 2001, abstract no. 70, Washington, D.C., March 15–18, 2001. [12] Quan ML, McCready D, McKinnon JG, Lehr E. Sentinel node biopsy in breast cancer: is the hottest node always the positive one? SSO Cancer Symposium, 2001, abstract no. P31, Washington, D.C., March 15–18, 2001. [13] Zervos EE, Burak WE. Lymphatic mapping for breast cancer: experience at the Ohio State University. Breast Cancer 2000;7:195–200. [14] Kuehn T, Klauss W, Darsow M, et al. Long-term morbidity following axillary dissection in breast cancer patients— clinical assessment significance for life quality and the impact of demographic, oncologic and therapeutic factors. Breast Cancer Res Treat 2000;64:275– 86. [15] Giuliano AE, Haigh PI, Brennan MB, et al. Prospective observational study of sentinel lymphadenectomy without further axillary dissection in patients with sentinel node-negative breast cancer. J Clin Oncol 2000;18:2553–9. [16] Dessureault S, Dupont E, Shons A, et al. Early results of breast cancer lymphatic mapping from H. Lee Moffitt Cancer Center: no axillary recurrences in breast cancer patients after a negative sentinel node biopsy. Breast Cancer Res Treat 2000;64:3.