- Email: [email protected]
Meta-Analysis of Sentinel Lymph Node Biopsy in Breast Cancer
Journal of Surgical Research 84, 138 –142 (1999) Article ID jsre.1999.5629, available online at http://www.idealibrary.com on
Meta-Analysis of Sentinel Lymph Node Biopsy in Breast Cancer Darlene M. Miltenburg, M.D., 1 Charles Miller, Ph.D., Tara B. Karamlou, B.S., and F. Charles Brunicardi, M.D. Baylor College of Medicine, Houston, Texas 77030 Presented at the Annual Meeting of the Association for Academic Surgery, Seattle, Washington, November 18 –22, 1998
INTRODUCTION Background. Sentinel lymph node biopsy (SLNB) is a minimally invasive way to diagnose axillary lymph node (ALN) metastases in breast cancer. The most important features are ability to identify the SLN (I.D. rate), how often the SLN and ALN pathology match (concordance), and how often the SLN is negative for cancer when the ALNs are positive (false negative). Technique and patient criteria for SLNB vary among studies. This study performed meta-analysis of published studies to determine the I.D., concordance, and false negative rate (1) overall and for (2) both blue dye and radiocolloid, (3) the injection method, (3) palpable and nonpalpable ALNs, and (4) invasive and in situ disease. Methods. Inclusion criteria were patients with breast cancer who had SLNB followed by ALN dissection with H&E staining. Meta-analysis was performed using analysis of variance with each observation weighted inversely to its variance. P < 0.05 was considered significant. Results. Eleven studies (n 5 912) met the inclusion criteria. Overall, 762 (84%) SLNs were identified, concordance was 747/762 (98%), and 15/296 (5%) were falsely negative. Highest I.D. rates (P < 0.05) were reported with albumin radiocolloid or dye 1 radiocolloid (97 and 94%, respectively), with injection around an intact tumor (96%), with invasive cancer (95%), and in the clinically negative axilla (96%). Concordance and false negative rates did not vary. Conclusions. The SLN can be identified in over 97% of patients if certain techniques and inclusion criteria are used. SLNB reflects the status of the axilla in 97% of cases and has a 5% false negative rate. © 1999 Academic Press Key Words: breast cancer; sentinel lymph node; meta-analysis.
1 To whom correspondence should be addressed at Department of Surgery, Baylor College of Medicine, 6550 Fannin, Suite 1628, Houston, TX 77030.
0022-4804/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
Sentinel lymph node biopsy (SLNB) is a minimally invasive technique that holds promise to become an alternative to routine axillary lymphadenectomy in patients with breast cancer. It has the potential to improve the accuracy of axillary staging while reducing the number of unnecessary lymph node dissections and decreasing morbidity. When the efficacy of SLNB is evaluated as a diagnostic tool, the most important points are how often the sentinel lymph node (SLN) can be successfully identified (I.D. rate), how often the pathology of the SLN is the same as the axillary lymph node dissection (concordance rate or accuracy), and how often the SLN is negative for malignancy when cancer is present in the rest of the axilla (false negative rate). Since first reported in breast cancer in 1993, there have been many studies comparing SLNB to axillary lymph node dissection (ALND). However, each study uses a different technique and different criteria for patient selection, thus making it difficult for an individual practitioner to know how to apply the technique to clinical practice. The purpose of this paper was to perform a metaanalysis of all published studies to determine the identification rate, concordance rate, and false negative rate (1) overall; (2) for blue dye, sulphur colloid, albumin colloid, and both dye and radiocolloid; (3) when the material is injected around an intact tumor and around a biopsy cavity; (4) for invasive and in situ cancer; and (5) for clinically negative and positive lymph nodes. METHODS All published series of patients with breast cancer who underwent SLNB were reviewed. Data were obtained from published material and not raw data. Criteria for inclusion in the meta-analysis were those studies where SLNB was followed by standard ALND. In all cases pathological analysis was performed using standard H&E staining. In studies that described their technique, the sentinel and
138
139
MILTENBURG ET AL.: META-ANALYSIS IN BREAST CANCER
TABLE 1 Technique and Patient Features Author
n
Dye or radiocolloid
Inject around tumor or biopsy cavity
Invasive or in situ disease
Clinical status of axilla
Krag et al., 1993 [1] Giuliano et al., 1994 [2] Uren et al., 1995 [3] Albertini et al., 1996 [4] Pijpers et al., 1997 [5] Giuliano et al., 1997 [6] Veronesi et al., 1997 [7] Borgstein et al., 1997 [8] Guenther et al., 1997 [9] Rubio et al., 1998 [10] Nawariaku et al., 1998 [11]
22 174 3 62 37 107 163 25 145 55 119
Sulfur colloid Dye Both Both Albumin colloid Dye Albumin colloid Both Dye Sulfur colloid Sulfur colloid
Tumor or cavity Tumor or cavity Around tumor Around tumor Around tumor Tumor or cavity Around tumor Not stated Tumor or cavity Tumor or cavity Tumor or cavity
Not stated Both Not stated Invasive only Invasive only Invasive only Invasive only Invasive only Both Invasive only Both
Positive or negative Positive or negative Not stated Negative only Negative only Positive or negative Negative only Negative only Not stated Negative only Not stated
nonsentinel nodes were bivalved and then examined histologically. In some studies immunohistochemistry for cytokeratin was used as well. However, cytokeratin results were not used in this metaanalysis. Meta-analysis is a quantitative statistical technique for combining studies and summarizing their results. Once it is determined that studies can be combined and analyzed, different factors common to one or more studies can be pooled and conclusions can be drawn which are stronger than those of individual studies. Meta-analysis was performed using analysis of variance with each observation weighted inversely to its variance. A P value less than 0.05 was considered significant. Identification rate was the number of times the SLN was identified divided by the number of axilla operated on. Concordance rate was the number of times the pathology found in the SLN was the same as that found in the ALND. It was calculated by dividing the number of times the SLN and ALN pathologies were the same, by the number of times the SLN was identified. The false negative rate was the percentage of times the SLN was negative for malignancy divided by the percentage of times the ALND was positive for malignancy.
RESULTS
Thirteen studies met the inclusion criteria. The total number of patients was 912. Age was specified in 854 patients and ranged from 25 to 90 years, with a mean of 54.8 years. Tumor size was specified in 621 patients and ranged from 0.4 to 8.0 cm, with a mean of 3.0 cm. The number of SLNs retrieved was specified in 64 patients and averaged 1.9 (range 1.3–2.2). The number of lymph nodes removed during axillary lymph node dissection was specified in 538 patients and averaged 19.5 (range 9.4 –28.1). These results are shown in Table 1. There was no significant difference in mean age, tumor size, or number of sentinel or axillary lymph nodes between studies. The histologic type of breast tumor was only specific in three studies and the location of the breast tumor was only stated in two studies. Therefore, these variables were not analyzed. Overall, the SLN was identified in 762 (83.6%) patients. The histology of the SLN matched that of the ALN in 747 cases giving a concordance rate of 98%. The
ALN specimen was positive for malignancy in 296 cases and the SLN was negative in 15, for an overall false negative rate of 5.1%. That is, the percentage of cases in which there was a negative sentinel lymph node when the axillary lymph node dissection was in fact positive was approximately 5%. These results are shown in Table 2. With respect to the technique of SLNB, the highest I.D. rate was achieved with albumin radiocolloid (97% I.D. rate, P 5 0.001 compared to blue dye or sulfur colloid) or a combination of blue dye and radiocolloid (94.4% I.D. rate, P , 0.05 compared to blue dye or sulfur colloid alone). The concordance rate and negative rate were not statistically different among the four groups. The identification rate was highest when the material was injection around the intact tumor (95.8%, P 5 0.001 compared to injection either around the intact tumor or around the biopsy cavity); however, the concordance and false negative rates were not statistically different. Concerning patient features, identification rate was highest when only patients with invasive cancer were studied (95.3%, P , 0.05 compared to patients with invasive or in situ cancer). Similarly, the I.D. rate was highest when only patients with a clinically negative axilla were included (95.9%, P , 0.05 compared to patients with or without clinical lymphadenopathy). The concordance and false negative rates were not statistically different among any of these groups. These results are shown in Table 3. TABLE 2 Overall Identification, Concordance, and False Negative Rate n
Successful SLN identification (%)
Concordance of SLNB and ALND (%)
912
762 (83.6)
747 (98.0)
False negative (%) 15 of 296
(5.1)
140
JOURNAL OF SURGICAL RESEARCH: VOL. 84, NO. 2, JUNE 15, 1999
TABLE 3 Identification, Concordance, and False Negative Rate According to Technique and Patient Features Successful SLN identification (%)
n Technique—Injection material Blue dye Sulfur colloid Albumin colloid Blue dye 1 radiocolloid Technique—Injection method Inject around breast tumor Inject around breast tumor or tumor cavity Not stated Patient features—Pathology Only invasive cancer Invasive or insitu cancer Not stated Patient features—Axilla Only clinically negative axilla Clinically negative or positive axilla Not stated *P ** P *** P **** P
5 5 5 5
0.001 0.001 0.001 0.001
compared compared compared compared
to to to to
False negative (%)
426 196 200 90
317 (74.4) 166 (84.7) 194 (97.0)* 85 (94.4)*
309 (97.5) 163 (98.2) 190 (97.9) 85 (100)
8 3 4 0
265 622
254 (95.8)** 483 (77.6)
250 (98.4) 472 (97.7)
4 of 116 (3.4) 11 of 166 (6.6)
25
25 (100)
of of of of
115 (7.0) 51 (5.9) 96 (4.2) 34 (0.0)
25 (100)
0 of 14 (0.0)
449 438 25
428 (95.3)*** 313 (71.5) 21 (84.0)
422 (98.6) 304 (97.0) 21 (100)
6 of 187 (3.2) 9 of 100 (9.0) 0 of 9 (0.0)
342
328 (95.9)****
322 (98.2)
6 of 145 (4.1)
303
232 (76.6)
227 (97.8)
5 of 91 (5.5)
267
202 (75.6)
198 (98.0)
4 of 60 (6.7)
either blue dye or sulfur colloid. injection around tumor or tumor cavity. invasive or in situ cancer. clinically negative or positive axilla.
The SLN was the only positive node in 146 (19%) of 762 successful SLN biopsies. These studies do not state whether or not micrometastases were present. Three studies (n 5 274) reported the status of the axillary lymph node dissection in cases where the SLN was not identified. The SLN could not be identified in 53 (19.3%) patients and in 17 (32%) of these, subsequent ALND revealed metastatic malignancy. The feature common to 15 of these 17 cases was that blue dye was used. These results are shown in Table 4. One complication was reported in 346 successful SLNBs. This was a transient fall in detected oxygen saturation due to systemic absorption of blue dye. DISCUSSION
One out of every eight women in the United States will develop breast cancer by the age of 70 years. The TABLE 4 When the SLN Is Not Identified
Krag et al., 1993 [1] Giuliano et al., 1997 [6] Guenther et al., 1997 [9] Total
Concordance of SLNB and ALND (%)
n
SLN not identified (%)
ALN malignant (%)
22 107 145 274
4 (18.2) 7 (6.5) 42 (29.0) 53 (19.3)
2 (50.0) 1 (14.3) 14 (33.3) 17 (32.1)
most common sites of metastases are the ipsilateral axillary lymph nodes. The National Cancer Institute recommends level 1 and 2 axillary lymph node dissection in all patients with invasive breast cancer [12]. The rationale for knowing whether or not axillary nodes contain metastases is that regional lymph node status (a) is the single most important predictor of survival, (b) is necessary for accurate staging, and (c) forms the basis for adjuvant treatment. There does not appear to be a therapeutic benefit to treating the axilla. In the NSABPB-04 trial, where patients were randomized to total mastectomy versus total mastectomy with radiation versus radical mastectomy, patients who did not have axillary treatment did not have decreased survival despite delayed development of clinically positive axillary lymph nodes requiring a delayed axillary lymph node dissection. Physical examination does not reliably predict the presence or absence of lymph node metastases [13]. Ultrasound has been used to help determine whether a clinically positive axillary lymph node is a reactive node or a node replaced with metastatic cancer. However, only a few centers have experience with this technique. Limited dissection decreases the morbidity of ALND, but may result in understanding [14]. Recently, the clinical utility of axillary lymph node dissection has been questioned. Widespread screening mammography has resulted in invasive breast cancer
MILTENBURG ET AL.: META-ANALYSIS IN BREAST CANCER
being diagnosed at an earlier stage. It is estimated that for every 100 ALNDs performed in patients with T1 and T2 tumors, 70 will not contain cancer. Furthermore, most, if not all, women with invasive breast cancer receive some form of adjuvant therapy regardless of the status of the axilla [15]. Finally, 10 to 20% of patients who undergo ALND will develop a complication such as seroma, neuropathy, neuroma, or lymphedema [16]. It would be ideal if lymph node metastases could be diagnosed using a minimally invasive technique while maintaining or improving diagnostic accuracy. Intraoperative lymphatic mapping and lymphadenectomy were first described in 1950 with gastric cancer [17] and in 1951 with lung cancer [18]. The concept of the sentinel node was described 12 years later [19]. In 1977 Cabanas recommended SLNB in the treatment of penile cancer [20] and in 1992 Morton et al. proposed SLNB for patients with melanoma. SLNB has subsequently become the standard of care for patients with intermediate thickness melanoma [21]. The use of radioactive tracers for SLNB was reported in 1992 [22]. The sentinel lymph node is the first node in the regional lymphatic basin that drains the primary tumor. The technique of SLN biopsy is based on the premise that cancer cells that detach from the primary tumor are likely to arrive at, and be sequestered by, the first or sentinel node to receive lymph from the tumor area. If the SLN can be identified and removed, its histologic characteristics should reflect those of the remaining nodes in the basin. Staging and treatment of the regional lymph node basin can be based on the status of the SLN. The objective, therefore, in performing sentinel lymph node biopsy is to identify, excise, and pathologically examine the sentinel lymph node. If the SLN is negative for malignancy, regional lymph node dissection is not necessary. If the SLN is positive, regional lymph node dissection is therapeutic. In breast cancer, SLNB has been proposed as a minimally invasive alternative to routine axillary lymphadenectomy. The potential advantages of SLNB is that it may reduce the number of negative ALNDs, thereby reducing the complication rate and eliminating the need for general anesthesia, hospital admission, and axillary draining. Moreover, it may improve the accuracy staging. The pathologist could focus his or her examination on only one or two nodes. More detailed analyses could include multiple sectioning to identify micrometasases, immunohistochemical staining (antibodies to cytokeratin), and reverse transcriptasepolymerase chain reaction (RT-PCR) analysis. Theoretically, micrometastases could be found in patients who previously were deemed to have a negative axilla. Recent studies have shown that micrometastases are associated with poorer survival compared to patients with no micrometastases [23]. Detection of microme-
141
tastases may alter therapy, especially in patients with T1 tumors. This meta-analysis showed that SLNB is an effective technique for staging the axilla in patients with breast cancer. Overall, the SLN could be identified over 85% of the time and once it was identified, the SLN reflected the status of the rest of the axilla in over 95% of cases. The chance that the SLN was negative when in fact there was metastatic disease in the axilla was about 5%. This is no higher than the false negative rate in routine level 1 and 2 axillary lymph node dissection. In this meta-analysis, the most successful results were obtained using albumin colloid and a hand-held gamma probe or a combination of blue dye and a radiocolloid. This may be because albumin passes rapidly through the SLN whereas sulfur colloid does not become concentrated for 1 or 2 hours [24]. Furthermore, the two techniques may be complementary because uptake of radiocolloid in the axilla is difficult to detect if the primary tumor is close to the axilla. In these cases, blue dye may be helpful. Identification was highest when SLNB was performed before breast tumor is removed, presumably because excision of the primary tumor disrupts the lymphatic drainage. The SLN was identified only three-quarters of the time when the patients with palpable lymphadenopathy were included. It has been demonstrated that substantial axillary tumor load blocks the uptake of dye or radiocolloid [25]. This is probably a moot point anyway, because patients with palpable axillary lymphadenopathy should undergo therapeutic ALND and not SLNB. It is not surprising that the SLN was identified only 71.5% of the time when patients with carcinoma in situ were included. Occasionally, however, patients thought to have carcinoma in situ will have a focus of invasive cancer and in these cases SLNB may be ideal. Finally, when the SLN could not be identified, the axillary lymph nodes were positive for malignancy onethird of the time. This suggests that if the SLN cannot be identified, one should proceed with formal ALND. Regardless of the technique used or the criteria for patient selection, once the SLN is identified, the concordance rate is uniformly high and the false negative rate is acceptably low. Out of 912 axillary lymph node dissections, 296 (32.5%) were positive for malignancy. Of the 762 cases where the sentinel lymph node was identified, 281 (36.9%) were positive for malignancy, P , 0.05. What is interesting is that in 146 cases, the SLN was the only node positive for malignancy. Thus, approximately one-third of all patients who undergo ALN dissection will have regional metastasis and approximately onehalf will have only one positive node which is the sentinel lymph node. It therefore seems important to identify this node, because if it is missed by sampling error, the axilla will be understaged. Furthermore, in
142
JOURNAL OF SURGICAL RESEARCH: VOL. 84, NO. 2, JUNE 15, 1999
early stage breast cancer, the sentinel lymph node may be the only positive lymph node on histologic examination, thus making it particularly applicable in this group of patients. The studies used in this meta-analysis are largely from experienced surgeons who have been trained in sentinel lymph node biopsy for treatment of melanoma patients and then have extended their work to the treatment of breast cancer patients. As with all new procedures, there is a learning curve for SLNB. It is recommended that surgeons take a course in the technique and then be proctored by an experienced surgeon in the first 10 cases. In addition, surgeons should perform at least 40 SLNB with an axillary lymph node dissection and have a false negative rate of no more than 5% before they consider using SLNB instead of axillary lymph node dissection. This is one of the reasons why SLNB is not yet ready to be considered the standard of care in breast cancer. In conclusion, SLNB is an accurate way to diagnose axillary lymph node metastases with a false negative rate of only 5%. It appears to be associated with fewer complications than standard axillary lymph node dissection. The most successful technique for identifying the SLN is albumin colloid or a combination of blue dye and radiocolloid. The SLN is easiest to identify when the technique is performed before the primary tumor is excised, in patients with invasive breast cancer and in patients who do not have clinically positive lymph nodes. It takes a multidisciplinary team to develop a successful sentinel lymph node program. It is not just the surgical technique alone but the combination of surgeons, nuclear medicine doctors, breast imaging personnel, and pathologists. Furthermore, a good proctor and accurate record keeping are essential to ensure that the surgeon has good results. Surgeons who are interested in doing sentinel lymph node biopsy may use the results of this meta-analysis to tailor their technique in order to achieve the highest possible success rate. REFERENCES 1.
2.
3.
4.
5.
Krag, D. N., Weaver, D. L., Alex, J. C., and Fairbank, J. T. Surgical resection and radiolocalization of the sentinel lymph node in breast cancer using a gamma probe. Surg. Oncol. 2: 335, 1993. Giuliano, A. E., Kirgan, D. M., Guenther, J. M., and Morton, D. L. Lymphatic mapping and sentinel lymphadenectomy for breast cancer. Ann. Surg. 220(3): 391, 1994. Uren, R. F., Howman-Giles, R. B., Thompson, J. F., et al. Mammary lymphoscintigraphy in breast cancer. J. Nucl. Med. 36(10): 1775, 1995. Albertini, J. J., Jyman, G. H., Cox, C., Yeatman, T., et al. Lymphatic mapping and sentinel node biopsy in the patient with breast cancer. J. Am. Med. Assoc. 276(22): 1818, 1996. Pijpers, R., Meijer, S., Hoekstra, O. S., Collet, G. J., et al. Impact of lymphoscintigraphy on sentinel node identification with technetium-99m-colloidal albumin in breast cancer. J. Nucl. Med. 38(3): 366, 1997.
6.
Giuliano, A. E., Jones, R. C., Brennan, M., and Statman, R. Sentinel lymphadenectomy in breast cancer. J. Clin. Oncol. 15(6): 2345, 1997.
7.
Vernonesi, U., Paganelli, G., Galimberti, V., Viale, G., et al. Sentinel-node biopsy to avoid axillary dissection in breast cancer with clinically negative lymph-nodes. Lancet 349: 1864, 1997.
8.
Borgstein, P. J., Meijer, S., and Pijpers, R. Intradermal blue dye to identify sentinel lymph node in breast cancer. Lancet 349: 1668, 1997.
9.
Guenther, J. M., Krishnamoorthy, M., and Tan, L. R. Sentinel lymphadenectomy for breast cancer in a community managed care setting. Cancer J. Sci. Am. 3(6): 336, 1997.
10.
Rubio, I. T., Korourian, S., Cowan, C., Krag, D., et al. Sentinel lymph node biopsy for staging breast cancer. Am. J. Surg. 176(6): 532, 1998.
11.
Nwariaku, F. E., Euhus, D. M., Beitsch, P. D., Clifford, E., et al. Sentinel lymph node biopsy, and alternative to elective axillary dissection for breast cancer. (abstract), 1998.
12.
National Institute for Health Consensus Conference. Treatment of early stage breast cancer. J. Am. Med. Assoc. 265: 391, 1991.
13.
Fisher, B., Wolmark, N., Bauer, M., et al. The accuracy of clinical nodal staging and of limited axillary dissection as a determinant of histologic nodal status in carcinoma of the breast. Surg. Gynecol. Obstet. 152: 765, 1981.
14.
Moffat, F., Senofsky, G., Davis, K., et al. Axillary node dissection for early breast cancer: Some is good, but all is better. J. Surg. Oncol. 51: 8, 1992.
15.
Baxter, N., McCready, D., Chapman, J. A., et al. Clinical behavior of untreated axillary nodes after local treatment for primary cancer. Ann. Surg. Oncol. 3: 235, 1996.
16.
Shaw, J. H. F., and Rumball, E. M. Complications and local recurrence following lymphadenectomy. Br. J. Surg. 77: 760, 1990.
17.
Weinberg, J., and Greaney, E. M. Identification of regional lymph nodes by means of vital staining dye during surgery of gastric cancer. Surg. Gynecol. Obstet. 90: 561, 1950.
18.
Weinberg, J. A. Identification of regional lymph nodes in the treatment of bronchiogenic carcinoma. J. Thorac. Surg. 22: 517, 1951.
19.
Busch, R. M., and Sayegh, E. S. Roentgenographic visualization of human testicular lymphatics: A preliminary report. J. Urol. 89: 106, 1963.
20.
Cabanas, R. M. An approach for the treatment of penile carcinoma. Cancer 39: 456, 1977.
21.
Morton, D., Wen, D., and Cochran, A. Management of earlystage melanoma by intraoperative lymphatic mapping and selective lymphadenectomy: An alternative to routine elective lymphadenectomy or “watch and wait.” Surg. Oncol. Clin. North Am. 1: 247, 1992.
22.
Terui, S., Kato, H., Hirashima, T., Lizuka, T., et al. An evaluation of the mediastinal lymphoscintigram for carcinoma of the esophagus studied with 99mTc rhenium sulfur colloid. Eur. J. Nucl. Med. 7: 99, 1982.
23.
International (Ludwig) Breast Cancer Study Group. Prognostic importance of occult axillary lymph node micrometastases from breast cancers. Lancet 335: 1565, 1990.
24.
Reintgen, D., Haddad, F., Pendas, S., Ku, N. N., et al. Lymphatic mapping and sentinel lymph node biopsy. Sci. Am., 1–17, 1998.
25.
Peyton, Crosbie, J., Bell, T. K., et al. High colloidal uptake in axillary nodes with metastatic disease. Br. J. Surg. 68: 507, 1981.
Meta-Analysis of Sentinel Lymph Node Biopsy in Breast Cancer Darlene M. Miltenburg, M.D., 1 Charles Miller, Ph.D., Tara B. Karamlou, B.S., and F. Charles Brunicardi, M.D. Baylor College of Medicine, Houston, Texas 77030 Presented at the Annual Meeting of the Association for Academic Surgery, Seattle, Washington, November 18 –22, 1998
INTRODUCTION Background. Sentinel lymph node biopsy (SLNB) is a minimally invasive way to diagnose axillary lymph node (ALN) metastases in breast cancer. The most important features are ability to identify the SLN (I.D. rate), how often the SLN and ALN pathology match (concordance), and how often the SLN is negative for cancer when the ALNs are positive (false negative). Technique and patient criteria for SLNB vary among studies. This study performed meta-analysis of published studies to determine the I.D., concordance, and false negative rate (1) overall and for (2) both blue dye and radiocolloid, (3) the injection method, (3) palpable and nonpalpable ALNs, and (4) invasive and in situ disease. Methods. Inclusion criteria were patients with breast cancer who had SLNB followed by ALN dissection with H&E staining. Meta-analysis was performed using analysis of variance with each observation weighted inversely to its variance. P < 0.05 was considered significant. Results. Eleven studies (n 5 912) met the inclusion criteria. Overall, 762 (84%) SLNs were identified, concordance was 747/762 (98%), and 15/296 (5%) were falsely negative. Highest I.D. rates (P < 0.05) were reported with albumin radiocolloid or dye 1 radiocolloid (97 and 94%, respectively), with injection around an intact tumor (96%), with invasive cancer (95%), and in the clinically negative axilla (96%). Concordance and false negative rates did not vary. Conclusions. The SLN can be identified in over 97% of patients if certain techniques and inclusion criteria are used. SLNB reflects the status of the axilla in 97% of cases and has a 5% false negative rate. © 1999 Academic Press Key Words: breast cancer; sentinel lymph node; meta-analysis.
1 To whom correspondence should be addressed at Department of Surgery, Baylor College of Medicine, 6550 Fannin, Suite 1628, Houston, TX 77030.
0022-4804/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
Sentinel lymph node biopsy (SLNB) is a minimally invasive technique that holds promise to become an alternative to routine axillary lymphadenectomy in patients with breast cancer. It has the potential to improve the accuracy of axillary staging while reducing the number of unnecessary lymph node dissections and decreasing morbidity. When the efficacy of SLNB is evaluated as a diagnostic tool, the most important points are how often the sentinel lymph node (SLN) can be successfully identified (I.D. rate), how often the pathology of the SLN is the same as the axillary lymph node dissection (concordance rate or accuracy), and how often the SLN is negative for malignancy when cancer is present in the rest of the axilla (false negative rate). Since first reported in breast cancer in 1993, there have been many studies comparing SLNB to axillary lymph node dissection (ALND). However, each study uses a different technique and different criteria for patient selection, thus making it difficult for an individual practitioner to know how to apply the technique to clinical practice. The purpose of this paper was to perform a metaanalysis of all published studies to determine the identification rate, concordance rate, and false negative rate (1) overall; (2) for blue dye, sulphur colloid, albumin colloid, and both dye and radiocolloid; (3) when the material is injected around an intact tumor and around a biopsy cavity; (4) for invasive and in situ cancer; and (5) for clinically negative and positive lymph nodes. METHODS All published series of patients with breast cancer who underwent SLNB were reviewed. Data were obtained from published material and not raw data. Criteria for inclusion in the meta-analysis were those studies where SLNB was followed by standard ALND. In all cases pathological analysis was performed using standard H&E staining. In studies that described their technique, the sentinel and
138
139
MILTENBURG ET AL.: META-ANALYSIS IN BREAST CANCER
TABLE 1 Technique and Patient Features Author
n
Dye or radiocolloid
Inject around tumor or biopsy cavity
Invasive or in situ disease
Clinical status of axilla
Krag et al., 1993 [1] Giuliano et al., 1994 [2] Uren et al., 1995 [3] Albertini et al., 1996 [4] Pijpers et al., 1997 [5] Giuliano et al., 1997 [6] Veronesi et al., 1997 [7] Borgstein et al., 1997 [8] Guenther et al., 1997 [9] Rubio et al., 1998 [10] Nawariaku et al., 1998 [11]
22 174 3 62 37 107 163 25 145 55 119
Sulfur colloid Dye Both Both Albumin colloid Dye Albumin colloid Both Dye Sulfur colloid Sulfur colloid
Tumor or cavity Tumor or cavity Around tumor Around tumor Around tumor Tumor or cavity Around tumor Not stated Tumor or cavity Tumor or cavity Tumor or cavity
Not stated Both Not stated Invasive only Invasive only Invasive only Invasive only Invasive only Both Invasive only Both
Positive or negative Positive or negative Not stated Negative only Negative only Positive or negative Negative only Negative only Not stated Negative only Not stated
nonsentinel nodes were bivalved and then examined histologically. In some studies immunohistochemistry for cytokeratin was used as well. However, cytokeratin results were not used in this metaanalysis. Meta-analysis is a quantitative statistical technique for combining studies and summarizing their results. Once it is determined that studies can be combined and analyzed, different factors common to one or more studies can be pooled and conclusions can be drawn which are stronger than those of individual studies. Meta-analysis was performed using analysis of variance with each observation weighted inversely to its variance. A P value less than 0.05 was considered significant. Identification rate was the number of times the SLN was identified divided by the number of axilla operated on. Concordance rate was the number of times the pathology found in the SLN was the same as that found in the ALND. It was calculated by dividing the number of times the SLN and ALN pathologies were the same, by the number of times the SLN was identified. The false negative rate was the percentage of times the SLN was negative for malignancy divided by the percentage of times the ALND was positive for malignancy.
RESULTS
Thirteen studies met the inclusion criteria. The total number of patients was 912. Age was specified in 854 patients and ranged from 25 to 90 years, with a mean of 54.8 years. Tumor size was specified in 621 patients and ranged from 0.4 to 8.0 cm, with a mean of 3.0 cm. The number of SLNs retrieved was specified in 64 patients and averaged 1.9 (range 1.3–2.2). The number of lymph nodes removed during axillary lymph node dissection was specified in 538 patients and averaged 19.5 (range 9.4 –28.1). These results are shown in Table 1. There was no significant difference in mean age, tumor size, or number of sentinel or axillary lymph nodes between studies. The histologic type of breast tumor was only specific in three studies and the location of the breast tumor was only stated in two studies. Therefore, these variables were not analyzed. Overall, the SLN was identified in 762 (83.6%) patients. The histology of the SLN matched that of the ALN in 747 cases giving a concordance rate of 98%. The
ALN specimen was positive for malignancy in 296 cases and the SLN was negative in 15, for an overall false negative rate of 5.1%. That is, the percentage of cases in which there was a negative sentinel lymph node when the axillary lymph node dissection was in fact positive was approximately 5%. These results are shown in Table 2. With respect to the technique of SLNB, the highest I.D. rate was achieved with albumin radiocolloid (97% I.D. rate, P 5 0.001 compared to blue dye or sulfur colloid) or a combination of blue dye and radiocolloid (94.4% I.D. rate, P , 0.05 compared to blue dye or sulfur colloid alone). The concordance rate and negative rate were not statistically different among the four groups. The identification rate was highest when the material was injection around the intact tumor (95.8%, P 5 0.001 compared to injection either around the intact tumor or around the biopsy cavity); however, the concordance and false negative rates were not statistically different. Concerning patient features, identification rate was highest when only patients with invasive cancer were studied (95.3%, P , 0.05 compared to patients with invasive or in situ cancer). Similarly, the I.D. rate was highest when only patients with a clinically negative axilla were included (95.9%, P , 0.05 compared to patients with or without clinical lymphadenopathy). The concordance and false negative rates were not statistically different among any of these groups. These results are shown in Table 3. TABLE 2 Overall Identification, Concordance, and False Negative Rate n
Successful SLN identification (%)
Concordance of SLNB and ALND (%)
912
762 (83.6)
747 (98.0)
False negative (%) 15 of 296
(5.1)
140
JOURNAL OF SURGICAL RESEARCH: VOL. 84, NO. 2, JUNE 15, 1999
TABLE 3 Identification, Concordance, and False Negative Rate According to Technique and Patient Features Successful SLN identification (%)
n Technique—Injection material Blue dye Sulfur colloid Albumin colloid Blue dye 1 radiocolloid Technique—Injection method Inject around breast tumor Inject around breast tumor or tumor cavity Not stated Patient features—Pathology Only invasive cancer Invasive or insitu cancer Not stated Patient features—Axilla Only clinically negative axilla Clinically negative or positive axilla Not stated *P ** P *** P **** P
5 5 5 5
0.001 0.001 0.001 0.001
compared compared compared compared
to to to to
False negative (%)
426 196 200 90
317 (74.4) 166 (84.7) 194 (97.0)* 85 (94.4)*
309 (97.5) 163 (98.2) 190 (97.9) 85 (100)
8 3 4 0
265 622
254 (95.8)** 483 (77.6)
250 (98.4) 472 (97.7)
4 of 116 (3.4) 11 of 166 (6.6)
25
25 (100)
of of of of
115 (7.0) 51 (5.9) 96 (4.2) 34 (0.0)
25 (100)
0 of 14 (0.0)
449 438 25
428 (95.3)*** 313 (71.5) 21 (84.0)
422 (98.6) 304 (97.0) 21 (100)
6 of 187 (3.2) 9 of 100 (9.0) 0 of 9 (0.0)
342
328 (95.9)****
322 (98.2)
6 of 145 (4.1)
303
232 (76.6)
227 (97.8)
5 of 91 (5.5)
267
202 (75.6)
198 (98.0)
4 of 60 (6.7)
either blue dye or sulfur colloid. injection around tumor or tumor cavity. invasive or in situ cancer. clinically negative or positive axilla.
The SLN was the only positive node in 146 (19%) of 762 successful SLN biopsies. These studies do not state whether or not micrometastases were present. Three studies (n 5 274) reported the status of the axillary lymph node dissection in cases where the SLN was not identified. The SLN could not be identified in 53 (19.3%) patients and in 17 (32%) of these, subsequent ALND revealed metastatic malignancy. The feature common to 15 of these 17 cases was that blue dye was used. These results are shown in Table 4. One complication was reported in 346 successful SLNBs. This was a transient fall in detected oxygen saturation due to systemic absorption of blue dye. DISCUSSION
One out of every eight women in the United States will develop breast cancer by the age of 70 years. The TABLE 4 When the SLN Is Not Identified
Krag et al., 1993 [1] Giuliano et al., 1997 [6] Guenther et al., 1997 [9] Total
Concordance of SLNB and ALND (%)
n
SLN not identified (%)
ALN malignant (%)
22 107 145 274
4 (18.2) 7 (6.5) 42 (29.0) 53 (19.3)
2 (50.0) 1 (14.3) 14 (33.3) 17 (32.1)
most common sites of metastases are the ipsilateral axillary lymph nodes. The National Cancer Institute recommends level 1 and 2 axillary lymph node dissection in all patients with invasive breast cancer [12]. The rationale for knowing whether or not axillary nodes contain metastases is that regional lymph node status (a) is the single most important predictor of survival, (b) is necessary for accurate staging, and (c) forms the basis for adjuvant treatment. There does not appear to be a therapeutic benefit to treating the axilla. In the NSABPB-04 trial, where patients were randomized to total mastectomy versus total mastectomy with radiation versus radical mastectomy, patients who did not have axillary treatment did not have decreased survival despite delayed development of clinically positive axillary lymph nodes requiring a delayed axillary lymph node dissection. Physical examination does not reliably predict the presence or absence of lymph node metastases [13]. Ultrasound has been used to help determine whether a clinically positive axillary lymph node is a reactive node or a node replaced with metastatic cancer. However, only a few centers have experience with this technique. Limited dissection decreases the morbidity of ALND, but may result in understanding [14]. Recently, the clinical utility of axillary lymph node dissection has been questioned. Widespread screening mammography has resulted in invasive breast cancer
MILTENBURG ET AL.: META-ANALYSIS IN BREAST CANCER
being diagnosed at an earlier stage. It is estimated that for every 100 ALNDs performed in patients with T1 and T2 tumors, 70 will not contain cancer. Furthermore, most, if not all, women with invasive breast cancer receive some form of adjuvant therapy regardless of the status of the axilla [15]. Finally, 10 to 20% of patients who undergo ALND will develop a complication such as seroma, neuropathy, neuroma, or lymphedema [16]. It would be ideal if lymph node metastases could be diagnosed using a minimally invasive technique while maintaining or improving diagnostic accuracy. Intraoperative lymphatic mapping and lymphadenectomy were first described in 1950 with gastric cancer [17] and in 1951 with lung cancer [18]. The concept of the sentinel node was described 12 years later [19]. In 1977 Cabanas recommended SLNB in the treatment of penile cancer [20] and in 1992 Morton et al. proposed SLNB for patients with melanoma. SLNB has subsequently become the standard of care for patients with intermediate thickness melanoma [21]. The use of radioactive tracers for SLNB was reported in 1992 [22]. The sentinel lymph node is the first node in the regional lymphatic basin that drains the primary tumor. The technique of SLN biopsy is based on the premise that cancer cells that detach from the primary tumor are likely to arrive at, and be sequestered by, the first or sentinel node to receive lymph from the tumor area. If the SLN can be identified and removed, its histologic characteristics should reflect those of the remaining nodes in the basin. Staging and treatment of the regional lymph node basin can be based on the status of the SLN. The objective, therefore, in performing sentinel lymph node biopsy is to identify, excise, and pathologically examine the sentinel lymph node. If the SLN is negative for malignancy, regional lymph node dissection is not necessary. If the SLN is positive, regional lymph node dissection is therapeutic. In breast cancer, SLNB has been proposed as a minimally invasive alternative to routine axillary lymphadenectomy. The potential advantages of SLNB is that it may reduce the number of negative ALNDs, thereby reducing the complication rate and eliminating the need for general anesthesia, hospital admission, and axillary draining. Moreover, it may improve the accuracy staging. The pathologist could focus his or her examination on only one or two nodes. More detailed analyses could include multiple sectioning to identify micrometasases, immunohistochemical staining (antibodies to cytokeratin), and reverse transcriptasepolymerase chain reaction (RT-PCR) analysis. Theoretically, micrometastases could be found in patients who previously were deemed to have a negative axilla. Recent studies have shown that micrometastases are associated with poorer survival compared to patients with no micrometastases [23]. Detection of microme-
141
tastases may alter therapy, especially in patients with T1 tumors. This meta-analysis showed that SLNB is an effective technique for staging the axilla in patients with breast cancer. Overall, the SLN could be identified over 85% of the time and once it was identified, the SLN reflected the status of the rest of the axilla in over 95% of cases. The chance that the SLN was negative when in fact there was metastatic disease in the axilla was about 5%. This is no higher than the false negative rate in routine level 1 and 2 axillary lymph node dissection. In this meta-analysis, the most successful results were obtained using albumin colloid and a hand-held gamma probe or a combination of blue dye and a radiocolloid. This may be because albumin passes rapidly through the SLN whereas sulfur colloid does not become concentrated for 1 or 2 hours [24]. Furthermore, the two techniques may be complementary because uptake of radiocolloid in the axilla is difficult to detect if the primary tumor is close to the axilla. In these cases, blue dye may be helpful. Identification was highest when SLNB was performed before breast tumor is removed, presumably because excision of the primary tumor disrupts the lymphatic drainage. The SLN was identified only three-quarters of the time when the patients with palpable lymphadenopathy were included. It has been demonstrated that substantial axillary tumor load blocks the uptake of dye or radiocolloid [25]. This is probably a moot point anyway, because patients with palpable axillary lymphadenopathy should undergo therapeutic ALND and not SLNB. It is not surprising that the SLN was identified only 71.5% of the time when patients with carcinoma in situ were included. Occasionally, however, patients thought to have carcinoma in situ will have a focus of invasive cancer and in these cases SLNB may be ideal. Finally, when the SLN could not be identified, the axillary lymph nodes were positive for malignancy onethird of the time. This suggests that if the SLN cannot be identified, one should proceed with formal ALND. Regardless of the technique used or the criteria for patient selection, once the SLN is identified, the concordance rate is uniformly high and the false negative rate is acceptably low. Out of 912 axillary lymph node dissections, 296 (32.5%) were positive for malignancy. Of the 762 cases where the sentinel lymph node was identified, 281 (36.9%) were positive for malignancy, P , 0.05. What is interesting is that in 146 cases, the SLN was the only node positive for malignancy. Thus, approximately one-third of all patients who undergo ALN dissection will have regional metastasis and approximately onehalf will have only one positive node which is the sentinel lymph node. It therefore seems important to identify this node, because if it is missed by sampling error, the axilla will be understaged. Furthermore, in
142
JOURNAL OF SURGICAL RESEARCH: VOL. 84, NO. 2, JUNE 15, 1999
early stage breast cancer, the sentinel lymph node may be the only positive lymph node on histologic examination, thus making it particularly applicable in this group of patients. The studies used in this meta-analysis are largely from experienced surgeons who have been trained in sentinel lymph node biopsy for treatment of melanoma patients and then have extended their work to the treatment of breast cancer patients. As with all new procedures, there is a learning curve for SLNB. It is recommended that surgeons take a course in the technique and then be proctored by an experienced surgeon in the first 10 cases. In addition, surgeons should perform at least 40 SLNB with an axillary lymph node dissection and have a false negative rate of no more than 5% before they consider using SLNB instead of axillary lymph node dissection. This is one of the reasons why SLNB is not yet ready to be considered the standard of care in breast cancer. In conclusion, SLNB is an accurate way to diagnose axillary lymph node metastases with a false negative rate of only 5%. It appears to be associated with fewer complications than standard axillary lymph node dissection. The most successful technique for identifying the SLN is albumin colloid or a combination of blue dye and radiocolloid. The SLN is easiest to identify when the technique is performed before the primary tumor is excised, in patients with invasive breast cancer and in patients who do not have clinically positive lymph nodes. It takes a multidisciplinary team to develop a successful sentinel lymph node program. It is not just the surgical technique alone but the combination of surgeons, nuclear medicine doctors, breast imaging personnel, and pathologists. Furthermore, a good proctor and accurate record keeping are essential to ensure that the surgeon has good results. Surgeons who are interested in doing sentinel lymph node biopsy may use the results of this meta-analysis to tailor their technique in order to achieve the highest possible success rate. REFERENCES 1.
2.
3.
4.
5.
Krag, D. N., Weaver, D. L., Alex, J. C., and Fairbank, J. T. Surgical resection and radiolocalization of the sentinel lymph node in breast cancer using a gamma probe. Surg. Oncol. 2: 335, 1993. Giuliano, A. E., Kirgan, D. M., Guenther, J. M., and Morton, D. L. Lymphatic mapping and sentinel lymphadenectomy for breast cancer. Ann. Surg. 220(3): 391, 1994. Uren, R. F., Howman-Giles, R. B., Thompson, J. F., et al. Mammary lymphoscintigraphy in breast cancer. J. Nucl. Med. 36(10): 1775, 1995. Albertini, J. J., Jyman, G. H., Cox, C., Yeatman, T., et al. Lymphatic mapping and sentinel node biopsy in the patient with breast cancer. J. Am. Med. Assoc. 276(22): 1818, 1996. Pijpers, R., Meijer, S., Hoekstra, O. S., Collet, G. J., et al. Impact of lymphoscintigraphy on sentinel node identification with technetium-99m-colloidal albumin in breast cancer. J. Nucl. Med. 38(3): 366, 1997.
6.
Giuliano, A. E., Jones, R. C., Brennan, M., and Statman, R. Sentinel lymphadenectomy in breast cancer. J. Clin. Oncol. 15(6): 2345, 1997.
7.
Vernonesi, U., Paganelli, G., Galimberti, V., Viale, G., et al. Sentinel-node biopsy to avoid axillary dissection in breast cancer with clinically negative lymph-nodes. Lancet 349: 1864, 1997.
8.
Borgstein, P. J., Meijer, S., and Pijpers, R. Intradermal blue dye to identify sentinel lymph node in breast cancer. Lancet 349: 1668, 1997.
9.
Guenther, J. M., Krishnamoorthy, M., and Tan, L. R. Sentinel lymphadenectomy for breast cancer in a community managed care setting. Cancer J. Sci. Am. 3(6): 336, 1997.
10.
Rubio, I. T., Korourian, S., Cowan, C., Krag, D., et al. Sentinel lymph node biopsy for staging breast cancer. Am. J. Surg. 176(6): 532, 1998.
11.
Nwariaku, F. E., Euhus, D. M., Beitsch, P. D., Clifford, E., et al. Sentinel lymph node biopsy, and alternative to elective axillary dissection for breast cancer. (abstract), 1998.
12.
National Institute for Health Consensus Conference. Treatment of early stage breast cancer. J. Am. Med. Assoc. 265: 391, 1991.
13.
Fisher, B., Wolmark, N., Bauer, M., et al. The accuracy of clinical nodal staging and of limited axillary dissection as a determinant of histologic nodal status in carcinoma of the breast. Surg. Gynecol. Obstet. 152: 765, 1981.
14.
Moffat, F., Senofsky, G., Davis, K., et al. Axillary node dissection for early breast cancer: Some is good, but all is better. J. Surg. Oncol. 51: 8, 1992.
15.
Baxter, N., McCready, D., Chapman, J. A., et al. Clinical behavior of untreated axillary nodes after local treatment for primary cancer. Ann. Surg. Oncol. 3: 235, 1996.
16.
Shaw, J. H. F., and Rumball, E. M. Complications and local recurrence following lymphadenectomy. Br. J. Surg. 77: 760, 1990.
17.
Weinberg, J., and Greaney, E. M. Identification of regional lymph nodes by means of vital staining dye during surgery of gastric cancer. Surg. Gynecol. Obstet. 90: 561, 1950.
18.
Weinberg, J. A. Identification of regional lymph nodes in the treatment of bronchiogenic carcinoma. J. Thorac. Surg. 22: 517, 1951.
19.
Busch, R. M., and Sayegh, E. S. Roentgenographic visualization of human testicular lymphatics: A preliminary report. J. Urol. 89: 106, 1963.
20.
Cabanas, R. M. An approach for the treatment of penile carcinoma. Cancer 39: 456, 1977.
21.
Morton, D., Wen, D., and Cochran, A. Management of earlystage melanoma by intraoperative lymphatic mapping and selective lymphadenectomy: An alternative to routine elective lymphadenectomy or “watch and wait.” Surg. Oncol. Clin. North Am. 1: 247, 1992.
22.
Terui, S., Kato, H., Hirashima, T., Lizuka, T., et al. An evaluation of the mediastinal lymphoscintigram for carcinoma of the esophagus studied with 99mTc rhenium sulfur colloid. Eur. J. Nucl. Med. 7: 99, 1982.
23.
International (Ludwig) Breast Cancer Study Group. Prognostic importance of occult axillary lymph node micrometastases from breast cancers. Lancet 335: 1565, 1990.
24.
Reintgen, D., Haddad, F., Pendas, S., Ku, N. N., et al. Lymphatic mapping and sentinel lymph node biopsy. Sci. Am., 1–17, 1998.
25.
Peyton, Crosbie, J., Bell, T. K., et al. High colloidal uptake in axillary nodes with metastatic disease. Br. J. Surg. 68: 507, 1981.