Discussion
In this retrospective study, the MSS of patients having WLE followed by OBS was not significantly different from the MSS of those having WLE and SNB. However, SNB patients had improved DFS and improved regional control compared with patients not undergoing SNB.
Several previous retrospective studies have also compared the outcomes of SNB patients with those of OBS patients ( Table 4 ). To overcome the limitations of retrospective studies, prospective RCTs are necessary. A large RCT addressing the issue of SNB or nodal observation is MSLT-I, the primary aim of which is to report the outcome following SNB or OBS in patients with intermediate thickness melanomas (1.2–3.5 mm). In the third interim analysis of MSLT-I, there was no statistically significant difference in MSS for patients in the SNB and OBS groups. However, the subgroup of SN-positive patients had significantly prolonged MSS after undergoing early CLND compared to WE-only patients who had a delayed TLND for regional node recurrence.
The results of the present study also suggest that patients with intermediate thickness melanomas (in this case defined according to the American Joint Committee on Cancer staging system as >1.0–4.0 mm) who undergo SNB and early CLND if they are SN-positive may have an overall survival benefit. SNB patients with melanomas 1 mm or less in thickness did not have a survival benefit in our data set. Patients with melanomas more than 1.0 to 4.0 mm who underwent SNB had significantly better DMFS and MSS than OBS patients in univariate analysis. Patients with T1 melanomas (≤1.0 mm) might not demonstrate a benefit from SNB because they have a very low rate of DM ( Table 2 ). Conversely, patients with melanomas more than 4 mm have a high rate of DM regardless of management, suggesting that they might be less likely to obtain a survival benefit from SNB. Nevertheless, these patients are likely to benefit not only from the prognostic information SNB provides, but also from the improved node field control provided by early CLND.
Of the 2515 SN-negative patients in the study, 89 (3.5%) had a first recurrence in the regional lymph node field. The clinical false-negative rate is conventionally reported as the rate of patients with a negative SN procedure who have a first recurrence in the same regional lymph node field as the SN procedure [false negative/(true positive + false negative)]. Calculated in this way, this study has an SN false-negative rate of 18.4% (89/483). This is in line with rates reported in the literature, which range from 7% to 24.8%. MSLT-I had an SN false-negative rate of 17.6%. The clinical false-negative rate of SNB may result from deficiencies in nuclear medicine, surgery, or pathology or may be a result of biologic events such as the presence of microscopic in-transit disease at the time of SNB, or subsequent metastasis from clinically apparent or occult locoregional recurrences. Our results indicate that the more experience a surgeon has with SNB, the lower the number of false-negative SNs will be. Between 1992 and 2000, the false-negative rate was 23.6% [35/(35+113)], but between 2001 and 2008 it was 16.1% [54/(54+281)]. Considering that SNB was first introduced in 1992, the reduction in the false-negative rate over time most likely reflects the learning curve of surgeons conducting a new, technically demanding procedure. Approximately one third (909/2909) of patients in the SNB group were treated during the earlier timeframe (1992–2000). A stratified analysis was conducted to assess whether changes in experience with SNB or management over time affected outcome. In the patient cohorts treated in the earlier (1992–2000) and the later (2001–2008) time frames, MSS was statistically similar, and DFS continued to be statistically different when comparing SNB and OBS groups (data not shown).
The rate of SN-positivity in this study was 13.5%, which is fairly low compared with rates reported in literature, which range from 14% to 29%. However, the median tumor thickness for the patients in the study was 1.70 mm, which is lower than in most other reported studies. Additional non-SN positivity was 19.5% (77/394). In the OBS group, 417 (14.2%) patients recurred in the regional lymph node field ( Table 1 ). The hypothesis being tested in MSLT-I is that SNB accurately identifies occult nodal metastases that will grow to palpable size if a "watch and wait" policy is adopted. In our study, SNB identified metastases in 13.5% of patients, with a false-negative result in 3.5% of patients (making a total of 17.0%). This is a higher percentage than the proportion of patients who subsequently had regional node metastases diagnosed clinically in the OBS group (14.2%). This difference of 2.8% requires explanation. The suggestion has been made that some patients with very low volume micrometastatic disease in a SN might never progress to clinically detectable metastatic disease in the node field However, the MSLT-I data indicating that all SN-positive patients will eventually develop clinically detectable nodal metastases do not support this concept. The lower rate of detection of metastases in the OBS group in this study is most likely due to the median follow-up time of only 42 months. Longer follow-up of the patients in this study will undoubtedly identify more patients with nodal metastases in the OBS group if there is the same pattern of time to nodal recurrence in the OBS group as occurred in MSLT-I, where it was not until 10 years of follow-up had elapsed that nodal recurrence in the OBS group reached a plateau. This was at a level that was virtually identical to the SN-positive plus false-negative value (20.5% vs 20.8%).
In addition to the inherent biases of any retrospective study, selection bias is an inevitable consequence of the design of the current study due to the exclusion of patients receiving either ELND or no immediate CLND after a positive SNB. The proportion of patients receiving ELND at MIA varied significantly over time, from 15% of all regional node operations during 1992–2000 to 2% during 2001–2008. On the basis of the results of the aforementioned analysis stratified by time frame, this bias likely did not significantly influence our results. In the study, 144 patients were excluded on the basis of no immediate CLND after a positive SNB, of which 113 (78%) were N1a and the remainder (31, 22%) were N2a. As a result, there are 82 more N1a (than N2a) patients removed from the data set who had a lower probability of benefit from SNB compared with N2a patients; however, this small number of patients is unlikely to have significantly influenced the results of the study. Nevertheless, selection bias should be considered when interpreting its results.
In this large, nonrandomized study, the overall outcome of patients having WLE alone was not significantly different from that of patients having additional SNB. However, the results indicate that the outcome for patients with T2 and T3 melanomas (>1.0–4.0 mm) may be improved if they undergo SNB, as did the interim results of MSLT-I, a large prospective trial. As well, SNB provided significantly improved regional disease control and overall DFS. SN status was the most important prognostic factor for survival, with disease recurrence and death approximately 3 times greater for SN-positive patients.
Pending the final results of MSLT-I, it is likely that most clinicians caring for patients with melanomas 1.0 mm or more in thickness will continue to recommend SNB as a staging procedure. SNB not only enables patients to be given more accurate prognostic information, but it also improves regional control and DFS. In addition, there may be an improvement of MSS in node-positive patients with intermediate thickness melanomas following SNB and early CLND, but long-term follow-up of patients in randomized trials (such as MSLT-I) will be necessary to confirm this. With trials of potentially more effective systemic adjuvant therapies likely to commence soon, SNB will also be necessary for selection and stratification of high-risk patients.