By Abby Van Voorhees, MD, August 01, 2011
In this month’s Acta Eruditorum column, Physician Editor Abby S. Van Voorhees, MD, talks with Arthur Rhodes, MD, MPH, about his recent Archives in Dermatology article, “Diagnostic Usefulness of Sentinel Node Biopsy for Patients Who Have Clinically Node Negative, Localized, Primary Invasive Cutaneous Melanoma.”
Dr. Van Voorhees: Tell us about the goal of your study.
Dr. Rhodes: The sentinel lymph node biopsy (SLNB) is being recommended as a standard of care for asymptomatic patients with primary localized invasive cutaneous melanoma (CM), highlighting its prognostic usefulness and perhaps providing a survival benefit and preventing the dire complications of nodal recurrence. Standard of care implies that physicians are obligated to offer a SLNB on such CM patients for prognostic reasons, and to recommend therapeutic lymph node dissection (TLND) for positive cases. This standard of care “dictum” is reminiscent of routine hemi-pelvectomies, amputations, and excessive excision margins during the past 30 to 75 years, because those therapies were also unsubstantiated and offered no survival benefit. The SLNB issues, in turn, have created medical-legal concerns. My study, which was a Bayesian analysis using raw numbers available in informative published reports, was designed to determine the prognostic usefulness of SLNB — i.e. to determine if SLNB status is indeed a useful predictor of CM-related death.
Dr. Van Voorhees: What did the Multicenter Selective Lymphadenectomy Trial tell us?
Dr. Rhodes: This international randomized cohort study of patients with intermediate thickness primary CM, the Multicenter Selective Lymphadenectomy Trial (MSLT), was designed to determine the survival impact of wide local excision followed by immediate lymphatic mapping and SLNB (and TLND for positive cases), compared to nodal observation alone after wide local excision of the primary tumor. According to this study, there was no significant difference in disease-specific survival (DSS) when the two groups were compared. Retrospective studies by other groups have shown similar results. However, the authors of the MSLT emphasized secondary outcomes of their study, namely, that the disease-free survival (DFS) was better in patients who had a TLND after a positive SLNB compared with the DFS among patients who developed macroscopic disease during nodal observation. This type of comparison is spurious. It is not appropriate to compare the prognosis of a group of patients selected for microscopic disease based on SLNB, with the prognosis of patients who have developed macroscopic disease — these two patient groups have very different prognoses to begin with that precludes a legitimate comparison.[pagebreak]
Dr. Van Voorhees: You’re suggesting that death is the most important endpoint.
Dr. Rhodes: Disease-specific survival (DSS) is defined as the proportion of patients not dying from CM, usually within a given time-frame, e. g. three years, five years, or 10 years (or longer). Disease-free survival (DFS) is defined as the proportion of patients surviving without evidence of disease, also within a given time-frame. Using DFS as an endpoint when lymph nodes are removed in one subgroup of patients but not the other subgroup is not a valid comparison. In the MSLT, regional lymph node recurrence is included in DFS. If regional node recurrence is included in DFS, then surgical removal of regional nodes will strongly influence the rate of DFS. DFS will be worse in patients whose tumor recurs in regional nodes compared to patients who have had TLND for microscopic disease (and thus have no nodes or very few nodes susceptible for recurrence). To recall, among patients who have intact regional lymph nodes, the first site of recurrence after wide local resection of the primary tumor is regional lymph nodes in 50 to 60 percent of cases, distant metastases in 18 percent, and simultaneous regional lymph nodes and distant sites in 31 percent. For patients who have had regional lymph nodes removed, the first sites of recurrence are regional lymph nodes in only 5 percent of cases, and soft tissue or visceral sites in 50 to 72 percent of cases. Other information demonstrates that the time from initial diagnosis of primary localized CM to distant recurrent disease or death is the same regardless of regional lymph node involvement. Reliance on DSS, or CM death rate according to SLNB status instead of DFS, eliminates this potential bias and will more likely permit replication in validation studies. Distant disease-free survival would have been a fairer comparison than DFS in the MSLT.
Dr. Van Voorhees: Tell us about the data sources you used and what other criteria you used for inclusion.
Dr. Rhodes: The English language Medline database was searched electronically using PubMed for original reports containing the terms CM and SLNB, published between Jan. 1, 1993 and June 3, 2010. Of the 480 publications found, 25 provided the raw numbers required for calculations of sensitivity and specificity of SLNB status for CM-related death. The study required reports having at least 50 asymptomatic patients with localized primary CM undergoing SLNB, stated raw numbers of patients with CM-related death according to SLNB status, and a clearly stated follow-up interval.[pagebreak]
Dr. Van Voorhees: What were the findings for the various thicknesses of tumors — thin, intermediate, and thick?
Dr. Rhodes: For patients with tumors 1 mm thick or less, the chance of CM-related death was less than 1 percent whether the SLNB was positive or negative, based on 4 informative studies. Thus, SLNB is not a useful nor appropriate prognostic test for CM patients with thin tumors. For patients with thick tumors (at least 4 mm), risk of death was one in three regardless of SLNB status. SLNB does not appear to be a useful prognostic test for patients who have thick tumors. There was only one informative study of patients with thick tumors.
For patients with intermediate thickness tumors (1-4 mm) based on two informative reports, risk of CM-related death ranged from 26.2 percent to 31.6 percent for SLNB positive cases, and 9.7 percent to 15.6 percent for SLNB negative cases. Thus, for intermediate thickness tumor patients, risk of death is about one in four to one in three for SLNB positive cases, and between one in 10 and one in six for SLNB negative cases. This variably useful prognostic information must be balanced against the significant morbidity and economic costs associated with the double jeopardy of TLND and adjuvant interferon therapy which are usually offered to SLNB positive patients.
Dr. Van Voorhees: Does SLNB status add significantly to prognosis when compared to a detailed histopathological investigation of the primary tumor, assessing all available features currently at our disposal?
Dr. Rhodes: I do not know, and this question is worth further detailed investigation. To recall, SLNB status is highly dependent on histopathologic features of the primary tumor. It must also be emphasized that SLNB itself may be curative when there are no other nodes involved following TLND. When the SLNB is positive, other regional nodes are involved with tumor (following TLND) only about 20 percent of the time. Thus, in 80 percent of patients who have a positive SLNB, the SLNB itself may be curative when there are no other involved nodes. Thus, the vast majority of SLNB patients are being unnecessarily subjected to the additional procedure of TLND without a documented survival benefit. A trial is currently underway investigating this issue because we need to know if a TLND has a survival benefit for patients who have had a positive SLNB. However, it is important to recall that CM patients can have a negative SLNB and still die of metastatic disease. We will need to wait for the results of ongoing studies to determine whether managing SLNB-positive patients without a TLND has a role in melanoma treatment.[pagebreak]
Dr. Van Voorhees: So, does tumor thickness solely drive the prognosis? Are there other factors that should be considered, such as location of tumor or atypia of the cells?
Dr. Rhodes: Breslow tumor thickness is the single best determinant of CM-related survival. Other features that have been significantly associated with disease-specific survival (DSS) based on histopathologic features of the primary tumor include mitotic rate (number of mitotic figures per square millimeter), presence of ulceration, vertical growth phase tumor, microscopic satellitosis, angiotropism, vascular invasion, and fibrotic regression. Other significant prognostic features include tumor volume, quality of tumor-associated lymphohistiocytic infiltrate, anatomic site of the primary tumor, predominant cell type of the primary tumor (spindle cell vs. epithelioid cell), patient sex and age, Clark’s level of invasion, tumor vascularity, and extent of lymphangiogenesis, among others. Prognostic models are available that have incorporated some or many of these features to provide an assessment of DSS. It is anticipated that molecular markers on the primary tumor may eventually be useful in the assessment of DSS.
Dr. Van Voorhees: What are the risks of doing SLNB? What are the economic considerations? What is the false negative rate?
Dr. Rhodes: The false negative rate can vary from 2.5 percent to 20 percent, dependent on available studies and various definitions. The cost of doing a SLNB ranges from $15,000-19,000, with an average 13 percent risk of surgical complications according to various studies. Moreover, the procedure usually requires a general anesthetic, patient downtime, and significant wound care. SLNB-positive patients will also be offered the additional procedures of TLND and adjuvant interferon therapy, with their associated morbidity and costs but without a clinically significant survival benefit.
Dr. Van Voorhees: How does your study fit with current guidelines of care for melanoma and recommended quality measures related to how much interventional care is given for patients who are asymptomatic?
Dr. Rhodes: My study deals only with asymptomatic patients. It’s safe to say that for patients with no clinical evidence of disease, doing x-rays, blood tests, or any imaging studies [which Medicare’s quality measures specifically address] is not only useless in terms of enhancing survival, but the false positive rates for these tests will be responsible for additional expense and needless patient worry.
Dr. Van Voorhees: A “Practice Gaps” piece accompanied your article; one of the limitations they noted was the small number of studies on which your analysis is based.
Dr. Rhodes: I agree with the criticism, which is highlighted in the discussion of my manuscript. Of the large number of SLNB studies I was able to find in the electronic search, there were only 25 informative papers which provided detailed raw numbers required for the Bayesian analysis. Most of the discarded articles provided univariate and/or multivariate analyses. All of the 25 informative published studies had relatively short patient follow-up intervals, and most of the studies included all tumor thicknesses. Relatively few studies I found segregated patients according to tumor thickness, i.e. thin, intermediate, and thick. None of the studies reported the proportion of patients followed for at least five years (or intervening death). Rather than to continue recommending routine SLNB for our patients, I believe that we need a multicenter cooperative study with collation of raw numbers and segregation of data according to tumor thickness, patient age, and the requirement of at least seven-10 years of patient follow-up (or intervening CM-related death). This type of a multicenter study is needed to substantiate or refute the prognostic usefulness of SLNB. For the issue of a minimum required time interval of follow-up in such a study, it is useful to recall the Duke University melanoma follow-up study by Gamel et al (Cancer 2002;95(6):1286-1293), in which it was shown that for patients who die from metastatic melanoma by 23 years, only 66 percent of these patients died by five years, while 92 percent died by 10 years. I would like to see all future publications about SLNB for CM include raw numbers, including details about follow-up (preferably at least five years or intervening death), that would be amenable to additional studies like the Bayesian analysis provided in my paper.
Dr. Rhodes is a professor and director of the melanoma intervention clinic in the department of dermatology at Rush University Medical Center. His article was published in Archives of Dermatology, 147(4), 408415 (April 2011) after being published online Dec. 20, 2010. doi:10.1001/archdermatol.2010.371.