The identification and resection of sentinel lymph nodes has been found to be an effective tool for the detection of regional lymph node metastases in patients with clinically node-negative cutaneous melanoma. Multiple studies have now been reported using vital blue dyes, radiolabelled colloids or a combination of the two for localizing sentinel nodes. Success rates for locating and resection of sentinel nodes reported in these studies have ranged from 81 to 100%. 1–4 The accuracy of a sentinel node biopsy in predicting the pathological status of a regional nodal basin has been found to be high. Initial studies have shown a correlation between the sentinel node status and that of a full lymphadenectomy in 98–99% of cases. 1,2 However, the sentinel node status may be falsely negative in up to 6% of node-positive patients when simultaneous lymph node dissections are performed. 1,2 The pathological status of the sentinel node has been found to be an important prognostic factor in this patient population and may well be one of the most important predictors of survival. 5 Sentinel lymph node biopsies have been found to be associated with minimal morbidity compared with elective lymph node dissections. This low morbidity, combined with the findings of randomized trials showing little if any survival benefit for elective node dissections, 6–9 have led many centres to adopt these techniques as their standard treatment for clinically node-negative patients. At the present there are relatively few reports documenting the incidence of recurrence and survival rates of patients treated with sentinel lymph node biopsy. This report represents an analysis of a large multicentre prospective consecutive patient study of clinically node-negative melanoma patients treated with wide excision and sentinel lymph node biopsy. It represents an expansion of a previous study by the authors 4 to include additional patients and longer post-treatment follow-up. The goals of this study were to determine the success rate of identifying and removing sentinel lymph nodes in melanoma patients and to determine the rate of disease recurrence, location of recurrence and overall survival rates for patients treated in this fashion.
Materials and methods
To be eligible patients had to have a biopsy-proven invasive cutaneous melanoma (Clark level II or higher), clinically negative regional lymph nodes and no evidence of distant metastases (clinical stage I or II). Thin melanomas were included in this study to better define the rate of sentinel node metastasis in this otherwise low risk group of patients. The sentinel lymph node biopsy was to be performed at the same time or within 45 days of the wide excision of the primary tumour site. Patients with prior wide excisions of the primary tumour site were eligible. All wide excision margins were 2 cm or less and there were no prior skin grafts. Patients who were pregnant were not eligible. All patients gave informed consent for participation in this study and the trial was conducted in accordance with the University of Vermont Human Subject Protection Committee and the Radiation Safety Committees of the participating institutions.
In this study the surgeon, using a hand-held gamma detector, was primarily responsible for sentinel node localization. For sentinel node identification in melanoma a number of different radio-colloid agents have been used successfully and were allowed in this study. The volume and dose of radioactive agent were left to the discretion of the surgeons, to allow for variability in the area that needed to be injected due to differences in tumour size, location and prior excision size. In all the patients sentinel node identification was dependent on the careful evaluation of all potential lymph node drainage sites by the surgeon using the hand-held gamma detector. Before an incision was made the detection of a radioactive ‘hot spot’ with the gamma detector was required to localize a sentinel node. The use of lymphoscintigraphy or blue dye was intended as an adjunct to the surgeon in performing these evaluations and was not required in the protocol. Because lymphoscintigraphy was used as an adjunct and not as the primary means of sentinel node identification, the timing and method of scintiscanning was left to the discretion of the individual institution.
All patients were injected with a radiolabelled colloid tracer for sentinel node identification. The agents used in this study were either unfiltered technetium-99m-sulphur colloid (CIS-US Inc., Bedford, Massachusetts, USA) or technetium-99m-human albumin (Solco Nanocoll, Sorin Biomedica, Vercelli, Italy, or Microlite, Dupont, Billerica, Massachusetts, USA). The dose of technetium injected varied from 5.92 MBq (0.16 mCi) to 51.8 MBq (1.4 mCi). Injections of 0.1–0.2 ml were made intradermally in a circumferential fashion around the lesion or excision site at distances no greater than 2.0 cm apart. The total volume injected varied depending on the number of injections required to obtain circumferential deposition and ranged from 0.5–2.5 ml. Injections were performed at least 30 min and not more than 28 h before the surgical procedure.
Surgeons had the option to inject vital blue dye in addition to the radiopharmaceutical. When used, an intradermal injection of 2.0–5.0 ml of isosulfan blue dye (Lymphazurin 1%, Hirsch Industries Inc., Richmond, Virginia, USA) or Patent Blue V 2.5% (Laboratorie Guerbert, Aulney-Sous-Bois, France) was made circumferentially around the lesion or excision site.
Gamma camera images were obtained according to the protocol of the nuclear medicine department of each participating institution. Images were not required for the protocol but were performed in most patients. The goal of these studies was to assist in identifying the primary (sentinel) lymph nodes receiving direct lymphatic drainage from the primary tumour site.
All participants used a hand-held gamma detector (C-Trak, Care Wise Medical, Palo Alto, California, USA) to identify the location of the radiolabelled sentinel nodes intra-operatively. Instruments were initially set at a threshold of 130 keV and a window of 40 keV. The threshold level could be adjusted to allow easier identification of node sites if needed.
Probe localization techniques
All sites identified by preoperative lymphoscintigraphy, all potential regional node drainage sites and all skin within 15 cm of the injection sites were evaluated intraoperatively with the hand-held gamma detector. Pre-incision hot spots, defined as discrete areas of increased radioactivity compared with background (with at least 15 counts per 10 s), were marked on the skin. Pre-incision radioactivity counts (over 10 s) were recorded at the site of the maximum radioactivity of each hot spot. Radioactive background counts were taken between the hot spot and the injection sites. The location of each hot spot identified was also recorded. The hot spot counts in most patients were well over the minimum 15 counts, although these counts could drop if prolonged waits occurred after the injection of agent. A minimum count of 15 was required to define a ‘hot spot’ as this was the threshold at which it was felt that the surgeon could distinguish a true ‘hot spot’ over background with the audio feedback of the gamma detector.
Blue dye localization
Blue dye, when used, was injected intradermally around the tumour or excision site in the operating room at least 5 min before an incision was made. A sentinel node identified by the blue dye method was defined as any blue-stained lymph node or any node with a blue-stained lymphatic duct leading directly into it.
Sentinel node excision (radiolabelled technique)
The placement of the incision for sentinel node removal was guided by the gamma detector and was centred over the pre-incision hot spot. Incisions were generally 2–4 cm in length. The gamma detector was placed in a sterile sheath and utilized during the procedure to guide the dissection of the radiolabelled sentinel nodes. Radioactive counts (over 10 s) were taken of each node after excision (ex vivo counts) and recorded. After sentinel node removal, the gamma detector was used to detect any additional radiolabelled nodes in that basin. If additional radioactive nodes were identified that had counts greater than or equal to 10% of the hottest node removed (ex vivo count), they were removed and classified as sentinel nodes as well. In removing the less radioactive nodes it is possible that some second tier nodes were removed. By using the 10% count threshold we have previously found that false-negative results are minimized by ensuring all possible sentinel nodes are removed, while not greatly increasing the total number of nodes removed.
Sentinel nodes were evaluated by routine haematoxylin and eosin staining. Because there was no uniformly accepted method of evaluating sentinel nodes at the inception of this study, the number of sections from each node evaluated in this way was left to the discretion of the pathology departments of the individual institutions. Routine immunohistochemical staining was not performed but was allowed to confirm an area of suspected metastasis seen using standard techniques.
Any patient with a metastasis identified in a sentinel lymph node underwent a comprehensive regional lymphadenectomy. Patients with negative sentinel lymph nodes had no further surgery to that regional basin and were observed.
Postsurgical adjuvant therapies were left to the discretion of the treating physicians. However, because of the lack of any documented effectiveness for adjuvant treatment when most of these patients underwent surgery, most patients received no adjuvant treatment.
Patients were scheduled for follow-up clinical examinations on a 4 month basis for the first 3 years following surgery and on a 6 month basis for years 3 to 5. Tissue or cytological confirmation was required for local or regional recurrences. Patients with local or regional recurrences who were candidates for surgical resection had appropriate resection of the recurrent disease. Patients with suspected distant metastases were to have confirmation of the metastases by pathological evaluation if possible. If this was deemed to be unnecessary or involved undue risk, then confirmatory radiographic studies would be accepted. In the event of patient death, the death report form was used to document the cause of death.
Descriptive statistics were obtained for patient and primary tumour characteristics and location on sentinel nodes. A 95% confidence interval (CI) was obtained for the technical success rate. Fisher's exact tests were performed on the prevalence of pathologically positive sentinel nodes by Breslow depth, Clark's level, tumour ulceration, tumour location, age and gender. Sites of first recurrence were examined using Fisher's exact tests. Kaplan-Meier plots of survival times and times to first recurrence were obtained, along with log-rank and Wilcoxon tests of subgroups with Bonferroni adjusted P values when more than two subgroups were compared. Cox proportional hazard models were used to examine survival times and times to first recurrence for individual covariates, both singly and jointly. Individual model covariates were characterized with 95% CIs on the hazard ratio (HR) scale, while interaction effects were characterized with 95% CIs for selected joint covariate patterns. A 5% significance level was utilized for hypothesis testing purposes. All data management and statistical processing was conducted using the SAS 10 and BMDP 11 software systems.
From 1993 to 1998, 336 patients met the eligibility criteria and were enrolled in this study. The demographics of the patients are listed in Table 1. Sentinel lymph node identification and excision was successful in 329 patients, giving an overall success rate of 97.9% (95% CI 95.8–99.2%). Sentinel nodes were identified in one, two or three regional nodal basins in 282 (85.7%,), 44 (13.4%) and three (0.9%) patients, respectively. The frequency of sentinel nodes identified in each regional basin is shown in Table 2. There was a mean of 1.96 and median of 2 sentinel lymph nodes per patient, with a range of 1–10. A total of 39 patients had one or more sentinel node(s) containing metastatic disease, giving a positive sentinel node rate of 11.9%. The distribution of primary tumour locations, depth of invasion and frequency of positive sentinel nodes is shown in Table 3. In patients with positive sentinel nodes, 28 (71.8%) patients had one positive sentinel node, nine (23.1%) patients had two positive sentinel nodes, and two (5.1%) patients had three positive sentinel nodes. Two patients had positive sentinel nodes in two different regional nodal basins (bilateral axillary). Pathological evaluation of the complete lymph node dissection in 38 of the 39 patients with positive sentinel nodes showed 28 (73.7%) patients had no additional positive lymph nodes, five (13.2%) patients had one additional positive lymph node and five (13.2%) patients had two or more additional positive nodes.
Overall, in the 329 patients in whom sentinel nodes were found, disease recurrences occurred in 45 (13.7%) patients, and 27 (8.2%) patients died of the disease after a mean follow-up time of 35.8 months. The locations of the first recurrence in patients with negative sentinel lymph node biopsies in relation to primary tumour depth are shown in Table 4. In the negative sentinel node group there were a total of 31 sites of first recurrence in 26 patients who recurred, with five patients having multiple recurrence sites at first presentation. Regional recurrences occurred as the most advanced site of first recurrence in 10 patients, nine of which were classified as lymph node recurrences and one as an in-transit recurrence. Three patients presented with simultaneous systemic metastases and lymph node recurrences.
In the group of patients who had wide excisions of the primary tumour site prior to the sentinel node biopsy (59 patients), sentinel nodes were successfully identified and removed in 58 (98%) patients. These patients had similar rates of positive sentinel nodes (eight patients, 13.6%). Patients with negative sentinel nodes in this group had similar recurrence rates (11.8% overall) and similar rates of regional recurrences (one isolated [1.96%], two simultaneous distant and regional [3.9%]).
Disease-free and overall survival rates were significantly related to sentinel node status and primary tumour thickness, as shown in Figures 1–6. In addition, patient sex (female doing better) and the presence of tumour ulceration were found to be predictors of disease-free but not overall survival using univariate analysis (Table 5). Multivariate analysis demonstrated that sentinel node status and primary tumour depth (Breslow or Clark) were the most important covariates predicting disease recurrence and patient survival. These covariates could not be separated from each other in this model; their interactions are described in Table 6. In the group of patients with sentinel node metastases it was found that patients whose disease was confined to only the sentinel nodes had significantly better survival than patients where disease was found beyond the sentinel nodes (Figure 7). In the group with sentinel node only disease, the survival times were not different if one or more sentinel nodes showed disease. Metastatic disease beyond the sentinel nodes was found in 10 of the 39 patients. All these cases had Clark level IV or V primary tumours with a Breslow depth of 1.70 mm or greater.
A comparison of the method of sentinel node identification was made in patients who were injected with both radio-colloid and blue dye tracers. A total of 108 patients could be evaluated in this way. In these patients, 162 sentinel node specimens were removed, 125 (77.2%) specimens were identified by both radioactivity and blue dye, 37 (22.8%) specimens were identified by radioactivity alone and no specimens were identified by blue dye alone. Metastatic disease was found in 23 (18.4%) of the 125 specimens labelled by both agents, while four (10.8%) of the 37 nodes identified by radioactivity alone had metastatic disease. These rates of metastatic disease were not significantly different statistically. Two of the four positive sentinel nodes identified by radioactivity alone were found in patients where no other positive sentinel nodes were identified by blue dye.
The results of this prospective multicentre study support previous reports showing a high success rate for identifying sentinel lymph nodes in patients with cutaneous melanoma. The success rates were similar in surgeons from a spectrum of different clinical practice situations, demonstrating the wide applicability of these techniques in the general surgical community. Metastatic disease was identified in 11.9% of the patients in whom sentinel lymph nodes were identified and removed. This rate was higher, as expected, in patients with thicker primary tumours. In this series there were no patients with positive sentinel nodes if the primary tumour was Clark level II; however, there was a single patient with a thin lesion (Breslow depth < 0.75 mm, Clark III) who had a positive sentinel lymph node. The absence of detectable metastases in the sentinel nodes of Clark II, thin melanomas suggests that these procedures may not be necessary in this subset of patients, although we did not have a sufficient number of patients with Clark II primary lesions to accurately determine the true rate of sentinel node metastases in this group. This low risk needs to be weighed with the finding that the HR for dying of disease from a positive sentinel node is much greater in patients with thin (< 0.75 mm) primary tumours (HR = 12.1) than in patients with thick primary tumours (HR = 2.7). Therefore, while sentinel node metastases are much less common in thin lesions, the impact of this information is potentially much more significant than in thicker lesions.
Sentinel lymph nodes may be found in multiple named lymph node basins within a single patient (14% of the patients in this study) and may also be present in less commonly involved nodal sites such as epitrochlear or popliteal nodes or at in-transit locations. Because of the risk of multiple nodal basins having metastatic disease, all identified sentinel node sites should be explored if feasible.
The best method for sentinel node identification has been debated in the past. In this study all the identified sentinel nodes were found to be radiolabelled. In the group of patients who were also injected with blue dye, 22.8% of the excised sentinel nodes were not identified by the blue dye and four of these radioactive-only nodes were found to harbour metastatic disease. These findings are not meant as a direct comparison of the two techniques, as the surgical dissection used to identify and remove the radiolabelled nodes may have interfered with the blue dye localization by interrupting lymphatic channels, but they do lend support to the continued use of radiolabelled guidance methods for sentinel node identification, either alone or in combination with vital blue dyes.
The pathological status of the sentinel lymph node was found to be an important predictor of disease-free and overall survival. In a multifactorial model, sentinel node status and primary tumour depth (Breslow or Clark) were found to be the strongest covariates predicting prognosis. The presence of metastatic disease beyond the sentinel lymph nodes also predicted a worse prognosis compared with patients where metastatic disease was confined only to the sentinel lymph nodes, even if disease was present in multiple sentinel nodes.
An important question for the surgeon is what is the risk of disease recurrence in the regional lymph node basin if a sentinel lymph node(s) is negative and no further surgery is performed. The results of this study indicate that this risk is low, occurring as an isolated event in only 3.4% (4.4% regional recurrence overall) of sentinel node-negative patients. A false-negative rate cannot be determined in this study, as complete lymph node dissections were not performed in sentinel node-negative patients. If it is assumed that all patients who had regional recurrences (10 as the first site, 13 total) had false-negative sentinel nodes, the false-negative rate would be 20.4 to 25%. However, the actual observed regional recurrence rate was quite similar to the rates seen in other studies of sentinel node-negative patients and is similar to the rate of recurrence in lymph node basins that were node negative but previously treated by therapeutic lymphadenectomy (rates ranging from 2.7 to 3.4%). 12–18 This finding of similar regional recurrence rates in patients treated by regional lymphadenectomy or sentinel node biopsy gives support to the sentinel node concept. It also suggests that a portion of these recurrences may be due to re-seeding of the regional node basin site by tumour cells in transit to that area from the primary site or from a systemic source and may not be a true failure of the sentinel node technique. It is also possible that some of the patients who had recurrent disease may have had micrometastatic disease that was missed by the standard histopathological evaluation performed in this study. It is possible that some of these patients may have been upstaged if a more intensive pathological evaluation of the nodes, using techniques such as serial sectioning or immunohistochemistry, had been performed.
In summary, this study confirms the accuracy and reliability of sentinel lymph node biopsy in staging cutaneous melanoma. It confirms the importance of the sentinel lymph node status in predicting patient survival and the low risk of regional recurrence if the sentinel nodes have no evidence of metastatic disease. These techniques provide this important clinical information in a minimally invasive fashion and should be considered for all patients with invasive melanoma, as even thin lesions may have associated positive sentinel lymph nodes. The current data suggest that the likelihood of finding metastatic disease in Clark II primary tumours is remote and this investigation may not be needed in the majority of such patients. However, given the relative low morbidity associated with these procedures and the impact of identifying regional nodal disease on patient treatment, the surgeon should determine the relative risk/benefit ratio on an individual basis. Larger numbers of Clark level II patients will need to be evaluated before the true risk of nodal involvement in this subset of patients is known. The information obtained from these procedures will allow for improved tumour staging and aid in the selection of patients who may benefit from more aggressive surgical interventions or systemic adjuvant therapies. The magnitude of this benefit, if any, will be answered by ongoing prospective randomized clinical trials.
This work was supported by the William Hauke Sr Fund, the SD Ireland Fund and a Vermont Cancer Center Core Grant (PHS P30CA22435).
We are indebted to Ms Mary Krupski and Ms Megan Hacking from the Vermont Cancer Center and Ms Maureen O'Connell from the Department of Biometry, University of Vermont.
1. Morton DL, Wen D-R, Wong JH, Economou JS, Cagle LA, Storm FK, Foshag LJ, Cochran AJ. Technical details of intraoperative lymphatic mapping for early stage melanoma
. Arch Surg 1992; 127: 392 –399.
2. Thompson JF, McCarthy WH, Bosch CM, O'Brien CJ, Quinn MJ, Paramaesvaran S, Crotty K, McCarthy SW, Uren RF, Howman-Giles R. Sentinel lymph node
status as an indicator of the presence of metastatic melanoma
in regional lymph nodes. Melanoma
Res 1995; 5: 255 –260.
3. Reintgen D, Cruse CW, Wells K, Berman C, Fenske N, Glass F, Schroer K, Heller R, Ross M, Lyman G. The orderly progression of melanoma
nodal metastases. Ann Surg 1994; 220: 759 –776.
4. Krag DK, Meijer SJ, Weaver DL, Loggie BW, Harlow SP, Tanabe KK, Laughlin EH, Alex JC. Minimal access surgery for staging of malignant melanoma
. Arch Surg 1995; 130: 654 –658.
5. Gershenwald JE, Thompson W, Mansfield PF, Lee JE, Colome MI, Tseng C-H, Lee JJ, Balch CM, Reintgen DS, Ross MI. Multi-institutional melanoma
lymphatic mapping experience: the prognostic value of sentinel lymph node
status in 612 stage I or II melanoma
patients. J Clin Oncol 1999; 17 (3): 976 –983.
6. Veronesi U, Adamus J, Bandiera DC, Brennhovd IO, Caceres E, Cascinelli N, Claudio F, Ikonopisov RL, Javorskj VV, Kirov S, Kulakowski A, Lacoub J, Lejeune F, Mechl Z, Morabito A, Rode I, Sergeev S, van Slooten E, Szcygiel K, Trapeznikov NN. Inefficacy of immediate node dissection in stage I melanoma
of the limbs. N Engl J Med 1977; 297: 627 –630.
7. Sim FH, Taylor WF, Pritchard DJ, Soule EH. Lymphadenectomy in the management of stage I malignant melanoma
: a prospective randomized study. Mayo Clin Proc 1986; 61: 697 –705.
8. Balch CM, Soong S-J, Bartolucci AA, Urist MM, Karakousis CP, Smith TJ, Temple WJ, Ross MI, Jewell WR, Mihm MC, Barnhill RL, Wanebo HJ. Efficacy of an elective regional lymph node dissection of 1 to 4 mm thick melanomas for patients 60 years of age and younger. Ann Surg 1996; 224: 255 –266.
9. Cascinelli N, Morabito A, Santinami M, MacKie RM, Belli F. Immediate or delayed dissection of regional nodes in patients with melanoma
of the trunk: a randomized trial WHO melanoma
programme. Lancet 1998; 14: 793 –796.
10. SAS software: Version 6
, 1st edn. Cary, North Carolina: SAS Institute Inc., 1991.
11. Dixon WJ (ed.). BMDP Statistical Software Manual
, vol 2. Berkley: University of California Press, 1992.
12. Leong SP, Steinnetz I, Habib FA, McMillan A, Gans JZ, Allen RE Jr, Morita ET, el-Kadi M, Epstein HD, Kashani-Sabet M, Sagebiel RW. Optimal selective sentinel lymph node
dissection in primary malignant melanoma
. Arch Surg 1997; 132 (6): 666 –672.
13. Bostick P, Essner R, Saranto T, Kelley M, Glass E, Foshag L, Stern S, Morton D. Intraoperative lymphatic mapping for early stage melanoma
of the head and neck. Am J Surg 1997; 174 (5): 536 –539.
14. Gershenwald JE, Colome MI, Lee JE, Mansfield PF, Tseng C, Lee JJ, Balch CM, Ross MI. Patterns of recurrence following a negative sentinel lymph node
biopsy in 243 patients with stage I or II melanoma
. J Clin Oncol 1998; 16: 2253 –2260.
15. Gadd MA, Cosimi AB, Yu J, Duncan LM, Yu L, Flotte TJ, Souba WW, Ott MJ, Wong LS, Sober AJ, Mihm MC, Haluska FG, Tanabe KK. Outcome of patients with melanoma
and histologically negative sentinel lymph nodes. Arch Surg 1999; 134 (4): 381 –387.
16. Lenisa L, Santinami M, Belli F, Clemente C, Mascheroni L, Patuzzo R, Gallino G, Bergonzi M, Rao S, Polverelli M, Morelli R, Landi G, Cascinelli N. Sentinel node biopsy and selective lymph node dissection in cutaneous melanoma
patients. J Exp Clin Cancer Res 1999; 18 (1): 69 –74.
17. Warso MA, Das Gupta TK. Melanoma
recurrence in a previously dissected lymph node basin. Arch Surg 1994; 129 (3): 252 –255.
18. Slingluff CL Jr, Stidham KR, Ricci WM, Stanley WE, Seigler HF. Surgical management of regional lymph nodes in patients with melanoma
. : Experience with 4682 patients. Ann Surg 1994; 219 (2): 120 –130.