Annals of Surgical Oncology

, Volume 14, Issue 10, pp 2861–2866

Isolated Tumor Cells in the Sentinel Node Affect Long-Term Prognosis of Patients with Melanoma

Authors

  • Randall P. Scheri
    • Department of Surgical OncologyJohn Wayne Cancer Institute at Saint John’s Health Center
    • Department of Surgical OncologyJohn Wayne Cancer Institute at Saint John’s Health Center
  • Roderick R. Turner
    • Department of PathologySaint John’s Health Center
  • Xing Ye
    • Biostatistical UnitJohn Wayne Cancer Institute at Saint John’s Health Center
  • Donald L. Morton
    • Department of Surgical OncologyJohn Wayne Cancer Institute at Saint John’s Health Center
Melanomas

DOI: 10.1245/s10434-007-9472-y

Cite this article as:
Scheri, R.P., Essner, R., Turner, R.R. et al. Ann Surg Oncol (2007) 14: 2861. doi:10.1245/s10434-007-9472-y

Abstract

Background

The clinical significance of isolated tumor cells (ITCs) in the melanoma-draining sentinel nodes (SNs) is unclear.

Methods

Records of patients who underwent SN biopsy (SNB) for stage I/II melanoma at our institute between 1991 and 2003 were reviewed to identify patients whose SNs were tumor-free or contained only ITC (≤0.2 mm). Tumor-positive SNs were reevaluated by the study pathologist to confirm the diagnosis and microstage the SN. Characteristics of the primary melanoma, tumor status of regional lymph nodes, and other prognostic variables were recorded. Melanoma-specific survival (MSS) rates were compared by the log-rank test.

Results

Of 1382 patients who underwent SNB, 1168 (85%) had tumor-free SNs; among the 214 remaining patients with tumor-positive SNs, 57 had metastases limited to ITC. Completion lymphadenectomy (CLND) was performed in 52 of 57 patients: six (12%) had metastases in nonsentinel nodes (NSNs). At a median follow-up of 57 months, 5-year and 10-year MSS was significantly higher (P = .02) for the 1168 patients with tumor-negative SNs (94 ± 1% and 87 ± 2%, respectively) than the 57 patients with ITC-positive SNs (89 ± 4% and 80 ± 7%, respectively). Multivariate analysis identified ITC (P = .002), Breslow’s thickness (P < .0001), ulceration (P < .0001), and primary site (P = .04) as significant for MSS.

Conclusion

Patients with ITC in SNs have a significantly higher risk of melanoma-specific death than those with tumor-negative SNs. The 12% incidence of nonsentinel node metastasis is similar to rates reported for patients with more extensive SN involvement. Patients with ITC should be considered for CLND.

Keywords

MelanomaSentinel lymph nodeMetastasisIsolated tumor cells

Since the initial report by Morton et al. more than 15 years ago,1 lymphatic mapping and sentinel lymph node biopsy (SNB) has drastically altered the management of early-stage melanoma as well as a variety of other solid tumors.2,3 SNB was initially developed as a minimally invasive procedure to identify those patients with lymph node metastasis who could potentially gain benefit from completion lymph node dissection (CLND), while allowing those patients without lymph node metastasis to avoid the costs and morbidity of CLND. The Multicenter Selective Lymphadenectomy Trial (MSLT-1)4,5 initiated by our group in 1994 validated the accuracy and reliability of SNB in melanoma on an international scale. Results from this and other studies6,7 have revealed that the majority of patients (80–90%) with sentinel node (SN) metastases will have only a single tumor-positive SN and no other lymph node metastases. As a result, surgeons812 have attempted to determine if particular patients with a tumor-positive SN could safely avoid CLND. If any patients could avoid CLND, it would be expected that those with isolated tumor cells in the SN (ITC) would be candidates. As yet, no method can accurately identify those patients whose nodal disease is limited to the SN.

For breast cancer, the American Joint Committee on Cancer (AJCC) has recently altered the staging system, categorizing SN metastases less than or equal to 0.2 mm in size as isolated tumor cells (ITC), designated as N0(i+) due to their excellent prognosis and outcome similar to that associated with tumor-negative SNs.13 These results have prompted studies in melanoma, questioning the significance of SN metastases ≤0.2 mm in size. Recent reports14,15 have suggested that, as do patients with breast cancer, patients with ITC in melanoma-draining SNs may have a prognosis similar to SN-negative patients and seldom have nonsentinel node (NSN) metastases. We found this surprising because breast cancer and melanoma are biologically quite different; excellent adjuvant therapy is available for breast cancer but none exists for melanoma. Furthermore, these studies suffer from short follow-up and small size, defects compounded by late and infrequent onset of recurrence in patients with low-volume disease.

We reviewed our prospectively collected melanoma database of nearly 1400 patients who underwent SNB at our institution from 1991–2003 to identify the long-term prognostic significance of ITC-positive SNs in melanoma and the rate of NSN metastasis.

METHODS

Approval of the study was granted by the John Wayne Cancer Institute (JWCI)/Saint John’s Health Center joint institutional review board. Our single institutional database was queried for all patients who underwent SNB for American Joint Committee on Cancer (AJCC) clinical stage I and II melanoma between July 1991 and December 2003. Patients with mucosal or ocular melanomas were excluded. Clinicopathological data included age, gender, primary site, Breslow’s thickness, Clark’s level, ulceration status, and nodal status.

Preoperatively, all patients underwent thorough history and physical, chest radiography, and blood profiles. In general, patients with primaries ≥1.0 mm or <1.0 mm with adverse risk factors including deep positive margin, regression, ulceration, and young age were offered SNB. SNB was performed by our group as previously described.1 Briefly, patients underwent cutaneous lymphoscintigraphy on the day of surgery, at which time the SN site was marked by the nuclear medicine physician. After general anesthesia was induced, isosulfan blue dye (Lymphazurin 1%, Tyco International, Norwalk, CT) was injected intradermally around the primary site, and the area was gently massaged to enhance lymphatic flow. After 10 minutes, an incision was made over the skin site identified by the nuclear medicine physician. The SN was identified by a handheld gamma probe that measured radioactivity and by visualization of blue dye. All radioactive and blue-stained nodes were excised and examined by an experienced pathologist.

SNs were cut at two histological levels separated by 200 μm and stained with hematoxylin and eosin (H&E); if results were negative for tumor, unstained sections from the two levels were stained with antibodies to S-100, HMB-45, and Melan-A.16 Slides assessed as positive for tumor were reevaluated by our study’s pathologist, who was blinded to the patient’s clinical outcome. Melanoma metastases were classified by the greatest dimension of the largest tumor cluster, as measured with an ocular micrometer. Clusters were measured separately; clusters greater than 0.2 mm did not qualify as ITC.

If the SN contained metastases, CLND was recommended and performed in the majority of cases. Lymph nodes removed at the time of CLND were sectioned and stained with H&E only, a potential flaw in our study since some metastases are found only by immunohistochemistry.17

Our study was limited to patients with tumor-negative SNs or ITC-positive SNs. Patients were followed up postoperatively by routine clinical examination, blood work, and chest radiography. Follow-up time was calculated from the time of initial diagnosis until last follow-up or death. All of the SNB and CLND procedures were performed within 3 months of diagnosis of the primary melanoma. Recurrences were grouped into local/in-transit, lymph node basin, or distant. The local and in-transit groups were combined into a single group due to the frequent inability to consistently separate the two groups. Estimated survival curves were constructed by the Kaplan-Meier method. Differences in survival distributions were tested by the log-rank method. Differences in intergroup frequency distributions and proportions were analyzed by Fischer’s exact test or Cochran-Armitage trend test. The prognostic significance of individual clinicopathological features was determined by Cox-regression analysis. Statistical significance was determined by P < .05.

Results

Complete data were available for 1382 patients with AJCC stage I or II melanoma who underwent SNB during the 13-year study period (Fig. 1). Of these, 1168 (85%) had negative SNs and 214 (15%) had tumor-positive SNs. Of the 57 (4%) patients whose SN metastases were limited to ITC, five had two ITC-positive SNs. Follow-up ranged from 1–182 months (median, 57 months). Five of the patients with ITC-positive SNs did not undergo CLND due to either patient or physician choice. The average number of lymph nodes removed during CLND was 17 for neck dissections, 23 for axillary dissections, and 15 for groin dissections.
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-007-9472-y/MediaObjects/10434_2007_9472_Fig1_HTML.gif
FIG. 1.

Flow chart summarizing study patients. SNB, sentinel node biopsy; SN, sentinel node; ITC, isolated tumor cells; CLND, completion lymph node dissection.

The majority (60%) of the 1225 patients with ITC-positive or tumor-negative SNs were 60 years or younger (range 15–86, median 53.2) (Table 1). The majority of the patients were men (60%). Most primary tumors were located on the trunk or extremities (80%), and more than 80% were Clark’s level III or IV. Complete data on Breslow’s thickness was available for 1205 (98%) of 1225 patients. The primary tumors were significantly thicker in the ITC-positive group (median 1.7 mm) than the tumor-negative group (mean 1.2 mm) (P = .0013). The rate of primary site ulceration was similar for the ITC-positive (14%) and tumor-negative (15.5%) groups.
TABLE 1.

Distribution of prognostic factors in the two study groups

Prognostic factor

Tumor-negative SNs (N = 1168)

ITC-positive SNs (N = 57)

P value

Age, years

  

.48

  ≤60

742 (64%)

39 (68%)

 

  >60

426 (36%)

18 (32%)

 

Gender

  

.097

  Female

461 (40%)

29 (51%)

 

  Male

707 (60%)

28 (49%)

 

Primary site

  

.09

  Head and neck

235 (20%)

6 (11%)

 

  Trunk

466 (40%)

21 (37%)

 

  Extremity

467 (40%)

30 (53%)

 

Clark’s level

  

.057 a

  I

8 (1%)

0 (0%)

 

  II

162 (14%)

4 (7%)

 

  III

414 (35%)

17 (30%)

 

  IV

495 (42%)

32 (56%)

 

  V

46 (4%)

2 (4%)

 

  Unknown

43 (4%)

2 (4%)

 

Breslow’s thickness

  

.0013 a

  Median (range)

1.20 (0.15, 18.00)

1.70 (0.05, 5.25)

 

  ≤1 mm

487 (42%)

12 (21%)

 

  1–2 mm

401 (34%)

24 (42%)

 

  2–3 mm

125 (11%)

8 (14%)

 

  3–4 mm

58 (5%)

8 (14%)

 

  >4 mm

78 (7%)

4 (7%)

 

Ulceration

  

.72

  Yes

181 (16%)

8 (14%)

 

  No

856 (73%)

46 (81%)

 

  Unknown

128 (11%)

3 (5%)

 

SN, sentinel node; ITC, isolated tumor cell.

a Cochran-Armitage Trend Test. All others are Fisher’s exact test.

Six (12%) of the 52 patients who underwent CLND had at least one tumor-positive NSN; four patients had only one positive NSN, and two patients had three positive NSNs (Table 2). All NSN metastases were ≤1.0 mm. Of the five patients with two ITC-positive SNs, only one had a tumor-positive NSN. Fifteen (26%) of the 57 patients with ITC-positive SNs and 143 (12%) of 1168 patients with tumor-negative SNs developed recurrence (Table 3). In the ITC group, the first site of recurrence was more often at local/in-transit or distant sites than in the regional lymph nodes. In the tumor-negative group, first recurrence was more common at distant sites than local/in-transit or regional sites, but the frequency of first recurrence at a distant site was lower in this group than in the ITC group. Moreover, the median time to distant recurrence was 29 months (range 4–144 months) in the tumor-negative group as compared with 20 months (range, 7–79 months) in the ITC group. Nine (16%) patients in the ITC group and 90 (8%) patients in the tumor-negative group have died from melanoma.
TABLE 2.

Characteristics of patients with ITC-positive sentinel nodes and tumor-positive nonsentinel nodes

Patient

Primary melanoma

Size of nodal metastases, mm

Anatomic site

Breslow’s thickness, mm

Ulceration

Sentinel

Nonsentinel a

1

Extremity

1.1

No

0.15

<0.2

2

Head and neck

1.3

No

0.15

<0.2

3

Extremity

1.9

No

0.04

1.0 b

4

Extremity

3.0

Yes

0.10

1.0

5

Trunk

3.9

Unknown

0.03

<0.2

6

Extremity

4.0

Yes

0.10

1.0 b

ITC, isolated tumor cell.

a Evaluated by hematoxylin and eosin alone.

b Three tumor-positive NSNs.

TABLE 3.

First site of recurrence for patients with tumor-negative SNs and patients with ITC-positive SNs

SN status

Number of recurrences

Local and in-transit

Lymph node

Distant

Total

Tumor-negative

46 (4%)

26 (2%)

70 (6%)

142 (12%)

Onset: 5–142 months

Onset: 4–69 months

Onset: 4–144 months

Onset: 4–144 months

ITC only

7 (12%)

2 (4%)

6 (11%)

15 (26%)

Onset: 5–35 months

Onset: 7–114 months

Onset: 7–79 months

Onset: 5–114 months

SN, sentinel node; ITC, isolated tumor cell.

The mean 5-year and 10-year disease-free survival (DFS) was significantly worse for patients with ITC in the SN (74 ± 6% and 61 ± 11%, respectively) than for those with a tumor-negative SN (89 ± 1% and 83 ± 2%, respectively) (P = .0008) (Fig. 2). The difference in DFS translated into a similar trend at 5 and 10 years for melanoma-specific survival (MSS): 89 ± 4% and 80 ± 7%, respectively, for the ITC-positive group compared with 94 ± 1% and 87 ± 2%, respectively, for the tumor-negative group (P = .02) (Fig. 3).
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-007-9472-y/MediaObjects/10434_2007_9472_Fig2_HTML.gif
FIG. 2.

Disease-free survival estimates at 5 and 10 years for patients with tumor-negative SNs (89 ± 1.% and 83 ± 2%, respectively) and ITC-positive SNs (74 ± 6.% and 61 ± 11%, respectively) (P = .0008).

https://static-content.springer.com/image/art%3A10.1245%2Fs10434-007-9472-y/MediaObjects/10434_2007_9472_Fig3_HTML.gif
FIG. 3.

Melanoma-specific survival estimates at 5 and 10 years for patients with tumor-negative SNs (94 ± 1% and 87 ± 2%, respectively) and ITC-positive SNs (89 ± 4% and 80 ± 7%, respectively) (P = .02).

Of the standard clinipathological features evaluated for their role in predicting melanoma-specific survival, age, primary site, Breslow’s thickness, ulceration status, and ITC were significant (P < .05) (Table 4).
TABLE 4.

Univariate and multivariate analysis of factors predicting melanoma-specific survival of 1168 patients with tumor-negative sentinel nodes (SNs) and 57 patients with ITC-positive SNs

Factor

Univariate analysis

Multivariate analysis

Gender (male vs female)

.03

NS

Age (>60 vs ≤60 years)

.006

NS

Breslow’s thickness

<.0001

<.0001

Ulceration (yes vs no)

<.0001

<.0001

Primary site (head/neck vs other)

<.0001

.04

ITC in SN

.03

.002

ITC, isolated tumor cell.

Discussion

SNB has become standard of care in early-stage melanoma. The SN provides accurate staging, and its tumor status has been well established as the most important prognostic factor in early-stage melanoma.5,6,18 SNB facilitates early CLND for patients with a positive SN; however only 10–20% of these patients will have positive NSNs.6,7 Several studies have attempted to identify subgroups of patients whose disease is limited to the SN and who therefore can avoid CLND.812 Results have been inconclusive; none of the predictors of NSN metastases have been validated.

Recent reports have questioned whether all tumor-positive SNs are true positives and have suggested that SNs with minimal evidence of disease, categorized as ITC (≤0.2 mm), should be treated as tumor-negative. Specifically, Van Akkooi et al.14 reported no NSN metastases in patients with SN metastases <0.1 mm, and Govindarajan et al.15 reported no NSN metastases in patients with SN metastases <0.2 mm; projected 5-year survival was 100%. However, both studies had fewer than 20 patients and a median follow-up less than 3 years. Both studies lacked sufficient statistical power due to their size and had wide confidence intervals indicating that the absence of NSN metastases could occur by chance alone. The short follow-up is a particular concern because the onset of recurrence is often late for low-volume (early-stage) disease.

To evaluate the significance of SN ITC in terms of recurrence and likelihood of positive NSN found on CLND, we evaluated our experience for which we had lengthy follow-up (median 57 months). The frequency of ITC in SN was low: only 4% (57 patients) of all patients who underwent SNB. This illustrates the infrequency of this pattern of metastases and the likely difficulty of identifying small-volume disease. The incidence of NSN metastases in the 52 patients who underwent CLND was 12% (6 of 52), only slightly less than the incidence of NSN metastases reported for all patients with tumor-positive SNs. Four of these six patients had SN metastases of ≤0.1 mm, the criterion used by Van Akkooi et al.,14 which suggests that diminishing size is not related to NSN involvement. Furthermore, the studies reported by Van Akkooi et al.14 and Govindarajan et al.15 were so small that only one or two patients would be expected to have a tumor-positive NSN, and it is not surprising that all NSNs were tumor-negative. The absence of standard criteria for ITC likely contributed to the differences between their findings and our report. In addition, in all three studies the NSNs were simply bisected and stained by H&E alone, which underestimates the frequency of metastases seen by enhanced pathological evaluation including immunohistochemical stains.17 Nodal understaging in this group may be particularly risky because it is these patients with minimal lymph node disease who may receive the most benefit from CLND, since metastatic disease may be confined to the lymph node basin.

The recurrence rate for patients with ITC in the SN was 26%. The median time to distant recurrence was 20 months (range, 7–79 months), illustrating the importance of long follow-up for this patient population.

Our 5-year MSS and DFS for the tumor-negative SN group were slightly higher than survival rates observed in MSLT-I.5 The improved survival can easily be explained by the thinner primaries in our study (median thickness 1.2 vs 1.8 mm for MSLT-I). Similarly, the 5-year MSS and DFS for the ITC-positive SN group were higher than those for the tumor-positive SN group in MSLT-I and can be explained by the former group’s lower volume of nodal disease and thinner primary lesions (median thickness 1.7 vs 2.1 mm for MSLT-I). The multivariate analysis identified SN status, ulceration status, Breslow’s thickness, and head and neck primary site as independently predictive of melanoma-specific survival, consistent with prior studies. These findings support the prognostic importance of ITC in the SN. Furthermore, studies at our institution have shown that patients with tumor-positive polymerase chain reaction (PCR) but no other evidence of SN involvement have a significantly higher rate of recurrence and poorer survival than patients who are PCR negative, suggesting that even submicroscopic disease in the SN is significant.19 Other studies, however, have not reached the same conclusions, and the significance of positive PCR findings remains controversial.20,21

Little was known about the natural history, recurrence pattern, and survival for patients who did not undergo CLND after a positive SNB until a recent report by Wong et al.22 examined a multi-institutional experience. Although these patients had a variable extent of SN disease, the rate of nodal recurrence without CLND was 15%, similar to a contemporary cohort who underwent CLND. This similarity is evidence to suggest the biological significance of most NSN metastases. Disease-specific survival was similar with and without CLND, suggesting clinical equipoise of CLND and observation and the importance of a randomized trial to evaluate the role of CLND for patients with a tumor-positive SN.

In conclusion, review of 1382 patients during a 13-year period has shown that melanoma-draining SNs that contain ITC as the only evidence of tumor are infrequent but clinically significant and should not be considered false-positive SN biopsies. These patients have a 12% incidence of NSN metastasis and higher rates of recurrence and melanoma-specific death than do patients with tumor-negative SNs. The second MSLT (MSLT-II) in which patients with a tumor-positive SN are randomized to CLND or nodal observation with ultrasound monitoring will determine which patients have biologically insignificant NSN involvement and can safely avoid CLND. Until the results of this trial are available, all patients with ITC in the SN should be considered for CLND.

Acknowledgments

Supported by grant CA29605 from the National Cancer Institute and by funding from the Wayne and Gladys Valley Foundation (Oakland, CA), the Harold J. McAlister Charitable Foundation (Los Angeles, CA), the Wrather Family Foundation (Los Alamos, CA), the Family of Robert Novick (Los Angeles, CA), the Samueli Foundation (Corona del Mar, CA) and the Saban Family Foundation (Los Angeles, CA).

Copyright information

© Society of Surgical Oncology 2007