How can we Guide Patient Choice between “Minimally Invasive” Radiosurgery versus Resective Epilepsy Surgery?

Commentary

Informing patient choice is the ultimate goal of clinical research. In comparative therapeutic studies, the objective is to directly evaluate how two or more treatment options “stack up” against each other. In epidemiological research or outcomes analyses, the objective is to better understand drivers of success to optimize interventions and eventually tailor them toward patient needs and preferences. Either way, knowing the short and long-term implications of a therapeutic approach are necessary to guide decision-making. This knowledge of effectiveness and safety should be “hard-knowledge” rather than sophisticated speculation or selective memory: we should know what actually happens to patients when a group gets treatment A versus treatment B, rather than what we think should happen with treatment A versus treatment B, or what we recall happened based on our clinical practice. The need for such rigorous analyses is what puts randomized clinical trials on the top echelon of evidence-seeking clinical research study designs. However, the randomized, controlled trial for radio-surgery versus open surgery for mesial temporal lobe (ROSE) illustrates the challenges of advancing the practice of epilepsy surgery through randomized clinical trials.

First, let's get through the facts. The stated rationale for ROSE—and usually for any “innovative” surgical method or device—is that epilepsy surgery is underutilized, largely because of true or perceived suboptimal outcomes of the traditional approach (anterior temporal lobectomy or ATL) where postoperative seizure-freedom rates need to be improved and complications lowered. The hope is that the innovation (in this case stereotactic radiosurgery or SRS) will be safer and better tolerated than is resective surgery without compromising the odds of seizure freedom. The study approach is the gold standard randomized clinical trial design. The expectation is that we will get the definitive answers to the questions of effectiveness and safety. In fact, the study team should be applauded for going to significant lengths to ensure rigor in study conduct: 1) Clear inclusion and exclusion criteria were put forth and the central study center reviewed each patient's SRS treatment plan to ensure protocol compliance; 2) patients wore large hats to cover SRS pin marks or ATL craniotomy scars and therefore maintain masking; 3) seizure outcomes were ascertained by blinded adjudicators; 4) adequate length of follow-up facilitated the evaluation of both short and long-term therapeutic implications; and 5) relevant quality of life measures and cognitive outcomes were evaluated to provide a comprehensive and informative assessments of outcomes. The design was perfect. The questions were meaningful. The investment was major: 14 epilepsy surgery centers in the United States alone, 3 continents with sites in the U.K. and India, 453 patients screened. So, what is the answer? We don't know for sure.

We don't have definitive answers because the required sample size could not be achieved, despite all the heroic measures mentioned earlier. ROSE needed 234 cases to show noninferiority: instead, it enrolled 58. Within the limitations of underpowered analyses, the signals suggest that ATL is actually superior (78% seizure free with ATL at 3 years versus 52% with SRS), with similar risk of complications and similar memory and language outcomes. The authors pointedly observe that this lack of memory preservation with the less invasive SRS may indicate that “changes in [verbal memory] are not caused by the surgical path taken to reach the medial temporal lobe, but rather by the effects of removal or damaging the medial structures.” In other words, surgical treatment of the drug-resistant epilepsy requires a surgical “hole” large enough to interrupt the epileptic network, effectively then large enough to cause some interruption to the normal cognitive networks, regardless of whether that “hole” was achieved via resective surgery, or SRS, or other methods.

One deeper implication from ROSE extends beyond the specific surgical question being posed to impact epilepsy surgery clinical trials more generally: Should we revisit the dogma that randomized clinical trials are truly the best approach to advancing our knowledge in the context of epilepsy surgery? In ROSE, at least half the patients who were eligible declined to participate. Odds are that most of those patients had some procedure to treat their epilepsy (either ATL or SRS). We did not learn from their experience because they were not randomized. They were not randomized because they did not want to be randomized. When presented with two vastly divergent options (craniotomy and resection: traditional approach, known outcomes vs radiosurgery: innovative approach, less invasive, but also less certainty about long-term outcomes), a stark choice automatically presents itself. Let's imagine ourselves in these patients’ shoes for a moment. After having dealt with drug-resistant epilepsy for years, often decades, we would most certainly lean toward one or the other treatment approach, but very unlikely be equally comfortable with both. This is the crux of the matter: being “equally comfortable with both” is equipoise from the patient's perspective. When that equipoise is lacking and the alternatives are so starkly different, and when a patient isn't equally comfortable with having ATL versus SRS, or having surgery now versus a couple of years from now, or taking medications versus having surgery, the patient will choose the approach he or she is most comfortable with, and the randomized clinical trial will not enroll. This problem is not unique to epilepsy surgery, but is rather faced by other neurosurgical disciplines, leading to a push toward alternative approaches, such as carefully designed observational studies and national registries “with a purpose” (1). As an epilepsy community, maybe we stick with randomized clinical trials as our sole source of absolute truth, and maybe we decide that we should be designing our clinical trials differently. But at least let's give this challenge some more thought, with an open mind—after all, the most elegantly designed randomized clinical trial will serve its purpose when completed, and only when completed.