It has traditionally been accepted in oncology that once patients have measurable stage IV disease, removal of the primary tumor is unlikely to improve survival. Indeed, this has been likened to closing the barn doors after the horses have left. In this setting, surgery has been considered to be palliative, or to prevent symptoms due to uncontrolled local tumor growth. It is with this in mind that patients with both a primary tumor in the breast and documented metastatic disease are generally not considered for lumpectomy or mastectomy. In her article, Dr. Khan argues that this widely held belief may be a misconception, and that removal of the primary tumor may indeed improve survival.
Dr. Khan presents compelling data that aggressive local therapy in such patients is associated with longer survival—but is this cause or effect? The evidence for this comes from studies that retrospectively reviewed patients with stage IV disease and compared those patients who underwent extirpation of the primary tumor with those who did not.[1-5] Dr. Khan argues that these data support the design and implementation of a prospective, randomized trial to definitively answer this question.
Tangibles and Intangibles
Retrospective, historically controlled clinical research contains inherent biases. Simply because similar, statistically significant results were obtained across several studies representing diverse geographic and cultural populations does not make them more credible. Repeating an inherently biased study 4 or 5 or even 100 times would almost by definition give you the same statistically significant results, especially when the bias is related to patient selection.
Patients deemed less likely to succumb quickly to their disease are offered surgery on the primary tumor more commonly than are less healthy appearing patients. Although most authors have stratified patients by known prognostic factors such as age, site of metastatic disease, type of systemic therapy, or comorbidities, this information is not always available. Even when it is, numerous intangibles may still have led to surgery in one subset of patients over another.
We believe it is also likely that publication bias affects this analysis. In general, investigations that results in negative findings are often less likely to be reported. For example, a presentation of these Surveillance, Epidemiology, and End Results (SEER) data at the 2007 Society of Surgical Oncology meeting prompted a response that in another study, when patients were stratified by the site of their metastases, the advantage of surgery largely disappears. However, this report has not been published.
The 'Gold Standard'
The hallmark of evidence-based medicine is the prospective, randomized clinical trial. Dr. Khan argues that such a trial is not only feasible but of potential benefit. While we are generally quite supportive of prospective, randomized trials, we are not sure that Dr. Khan's estimates are realistic.
First, we have grave concerns about the feasibility of such a trial. Dr. Khan estimates that approximately 7,000 patients are seen in the United States with simultaneous new primary tumors and distant metastases. If we estimate that approximately one-half of these women would not be eligible for this type of study due to confounding factors, then only 3,500 patients would be available for entry. Randomization to highly diverse treatment arms, such as surgery vs observation, has always presented accrual problems, since doctor and patient biases are so strong.