Gene Expression Profiling in Malignant Melanoma
David M. Hyams, M.D., F.A.C.S.
Malignant melanoma is a high-risk cancer, derived from the pigment cells of the skin. There are 100,000 new cases of melanoma in the US every year resulting in nearly 7,000 deaths. Understanding the individual risk of any tumor is an important part of designing treatment and followup plans. Although pathology features have long been the primary determinant of risk, techniques that incorporate a better understanding of tumor biology are now becoming more widely used.
There are practical limits to risk assessment by pathology examination alone. For example, lymph node involvement has been a significant risk factor for melanoma recurrence. Patients with lymph node-positive disease (Stage III) have greater overall risk. However the largest proportion of melanoma deaths in the US, occur in patients who originally presented without lymph node metastases (Stage I and Stage II). This shows the importance of identifying the small percentage of high-risk patients that exist among the very large population of Stage I and Stage II lymph node-negative patients.
The “staging” of localized melanoma is based on an assessment of just three pathologic features. These include the measured thickness of the tumor, the presence of tumor ulceration, and the status of regional lymph node involvement.
Patients having thin tumors, with even a small deposit of cancer in a single lymph node, are currently classified as Stage III. Yet, their clinical course is much more favorable than patients with only slightly thicker tumors and negative lymph nodes (Stage II).
Among Stage I and Stage II patients having identical external pathologic features, different biologic features give some a 5% risk of recurrence at 5 years, while others may have as much as a 50% chance of recurrence over the same time period.
To make physical assessment more complicated, dermatologic biopsies often cut across the deepest portion of the melanoma, limiting the accurate assessment of tumor thickness. In some patients, ulceration may be confused with trauma caused by the scratching an itching tumor.
Because of the importance of lymph node involvement in staging and prognosis, a large number of sentinel lymph node biopsies are performed to find a small number of patients with any lymph node tumor. Using contemporary biopsy recommendations, 88% of lymph node biopsies will be negative. As a result many patients undergo surgery that does not contribute to their outcome.
Gene expression is how cells relay information in the DNA “master blueprint” to the protein machinery of the cell. Proteins produced provide the basis of cell function, including growth, metabolism, etc.
Gene expression profiling is a technique that evaluates the functional activity of cancer cells. In this case a limited group of genes relating to melanoma growth and spread are evaluated. These genes are used to create a biological fingerprint of the tumor.
When we order gene expression profiling of a patient’s melanoma, a sample of the original tumor specimen is sent to an independent laboratory where the assay is performed. Based on the pattern of genes that are turned “on” or turned “off”, a validated calculation creates a probability score for that tumor. The score suggests that the tumor will behave more like a group of patients with poor outcome who recurred, or more like a group of patients who were recurrence-free with a favorable outcome.
Each patient is assigned to a low-risk group (Class 1) or high-risk group (Class 2). As with any probability assessment, there is a an inflection point were the risk assignment is a statistical “toss-up” . Fortunately, that result almost never happens.
Scores within a single standard deviation of the mid-point produce a class-assignment of somewhat diminished confidence. These are termed Class 1B, or 2A depending on which side of the 50% cut-point their fall on. These groups appear to have a more intermediate risk, than the more confidently assigned Class 1A and Class 2B patients.
Interestingly, this information isn’t only relevant for prognosis. It appears to be useful for predicting the likelihood of lymph node positivity at the time of sentinel lymph node biopsy. When a patient is over 55 and the risk class assignment is low-risk Class 1A, the chance of a positive sentinel node drops to 5% or less. With sentinel lymph node biopsy having a false-negative rate of 4 – 5%, it makes little sense to perform sentinel lymph node biopsy on these low-likelihood patients.
Gene expression profiling is a tool that helps us better help our melanoma patients. When combined with other relevant clinical features and demographic information, we avoid unnecessary operations. This can reduce cost, inconvenience, and side effects. In other patients, gene expression profiling can help us plan better follow-up programs, limiting expensive and intrusive testing to where it is needed. In still other patients, gene expression profiling can contribute to decision-making regarding the use of expensive and lengthy drug treatments.
Understanding tumor biology is key to providing good state-of-the-art surgical care. It is one of the ways to bring patients back to better.
