Preoperative Breast MRI: Barking up the Wrong Endpoints

We should all agree that when it comes to invasive breast cancer, preoperative breast magnetic resonance imaging (MRI) is not going to alter survival. If it did, the underlying biology of breast cancer that allows conservation surgery in the first place would be undermined, if not negated entirely. Even Dr Bernard Fisher weighed in recently on this controversy to remind us that if variations in local treatment do not substantially alter survival, then preoperative MRI is unlikely to have any impact. That said, it should also be obvious that preoperative mammography does not alter survival either, and for the very same reason. No form of preoperative breast imaging should be expected to alter survival. The straightforward intent of imaging is to provide a roadmap for the surgeon. So, in a recent article by Houssami and colleagues, “An individual person data meta-analysis of preoperative magnetic resonance imaging and breast cancer,” we have another masterful display of sophisticated, labor-intensive statistics that go the extra mile, only to fall prey to ignoratio elenchi, a logical fallacy wherein the argument is entirely valid but does not address the issue in question. Stated alternatively, it is barking up the wrong endpoint. So, let’s move on from survival, be it disease free or not. What about local recurrence with and without preoperative MRI? First of all, our use of the term “roadmap” doesn’t necessarily mandate, or equate to, improved local control as defined by fewer in-breast recurrences. After all, where is the evidence that preoperative mammography for palpable cancers lowers local recurrence rates? Mammography, often considered part of the proven package, was not a requirement for enrollment in the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-06 trial and has never had its feet held to the flames of evidence-based medicine as has MRI. Even Fisher attributes today’s lower in-breast recurrence rates to improved systemic therapies, not preoperative mammography. When we began our breast MRI program in 2003d jointly directed by a breast surgeon (A.B.H.) and a breast radiologist (R.G.S.)dwe outlined 8 endpoints that we intended to monitor in real time to quantify potential benefit. Survival was not even on the list. As for in-breast recurrence rates, given the low numbers already being reported at the time, an improvement was considered possible but listed with a question mark. We later dropped this endpoint entirely, now supported by this article from Houssami and colleagues, in favor of other outcomes of surgical efficiency, at least 2 of which (discussed below) have never been studied, rendering the outcries to “stop performing preoperative MRI” decidedly premature. Our goal was simple, or so it seemed: to meet the needs of the newly diagnosed breast cancer patient in a single operation. This approach was to be distinguished from piecemeal oncology where, remarkably, it was widely agreed that 2, 3, or even 4 operations to remove a breast cancer was acceptable, perhaps even laudable, in the name of conservation, even though failures ended in mastectomy anyway. Given the 90% sensitivity of high spatial resolution MRI and the 40% sensitivity of mammography in head-to-head comparisons, breast MRI seemed the logical choice to achieve this goal, although several other options are available and promising, including molecular imaging and contrast-enhanced tomosynthesis mammography. The goal of “single surgery” could be accomplished in several ways: performing fewer re-excisions of the index lesion; addressing isolated multicentric foci with double lumpectomy or mastectomy; identifying diffuse multi-quadrant disease that usually requires mastectomy even after neoadjuvant therapy; and addressing synchronous contralateral cancers up front, rather than 6-24 months later. Reductionist critics have dissected these potential benefits, pointing out the small probabilities for each. When considered individually, these single-digit benefits are lost in a statistical morass. But a roadmap is intended to be used up front, before the journey begins. If breast MRI is performed shortly after a diagnosis of malignancy, the prospective, potential benefits are additive and must be considered as a group. Instead, “divide and conquer” has been the motto for critics, even if it means piling up negative outcomes for endpoints that are predictably unaffected by imaging. Will the American College of Surgeons Oncology Group (ACOSOG) Z11101/American College of Radiology Imaging Network (ACRIN) 6694 trial of preoperative MRI offer us answers? Ignoring the unshakable inclusion of survivalbased endpoints, perhaps there will be local benefits, given the high-quality MRI proposed. But we doubt it, given that the trial is focused exclusively on triple-negative and human epidermal growth factor receptor 2epositive tumors. Tumor biology in these cases translates into higher local recurrence rates and lower survival rates, ostensibly the reason for such exclusivity in trial design. Yet, once again, these are the 2 endpoints least likely to benefit from any form of preoperative imaging. The higher recurrence rates in these particular tumors are likely due to their inherent biology, imparting treatment resistance, even when very small amounts of tumor are left behind. The potential benefit of preoperative imaging cannot impact inherent tumor biology,